**Cross-Species Proteomics Identifies CAPG and SBP1 as Crucial Invasiveness Biomarkers in Rat and Human Malignant Mesothelioma**

**Joëlle S. Nader <sup>1</sup> , Alice Boissard <sup>2</sup> , Cécile Henry <sup>2</sup> , Isabelle Valo <sup>2</sup> , Véronique Verrièle <sup>2</sup> , Marc Grégoire <sup>1</sup> , Olivier Coqueret <sup>3</sup> , Catherine Guette <sup>2</sup> and Daniel L. Pouliquen 3,\***


Received: 16 July 2020; Accepted: 23 August 2020; Published: 27 August 2020

**Abstract:** Malignant mesothelioma (MM) still represents a devastating disease that is often detected too late, while the current effect of therapies on patient outcomes remains unsatisfactory. Invasiveness biomarkers may contribute to improving early diagnosis, prognosis, and treatment for patients, a task that could benefit from the development of high-throughput proteomics. To limit potential sources of bias when identifying such biomarkers, we conducted cross-species proteomic analyzes on three different MM sources. Data were collected firstly from two human MM cell lines, secondly from rat MM tumors of increasing invasiveness grown in immunocompetent rats and human MM tumors grown in immunodeficient mice, and thirdly from paraffin-embedded sections of patient MM tumors of the epithelioid and sarcomatoid subtypes. Our investigations identified three major invasiveness biomarkers common to the three tumor sources, CAPG, FABP4, and LAMB2, and an additional set of 25 candidate biomarkers shared by rat and patient tumors. Comparing the data to proteomic analyzes of preneoplastic and neoplastic rat mesothelial cell lines revealed the additional role of SBP1 in the carcinogenic process. These observations could provide new opportunities to identify highly vulnerable MM patients with poor survival outcomes, thereby improving the success of current and future therapeutic strategies.

**Keywords:** malignant mesothelioma; biomarkers; proteomics; macrophage-capping protein; fatty acid-binding protein; laminin subunit beta-2; selenium-binding protein 1; carcinogenesis

#### **1. Introduction**

The management of malignant mesothelioma (MM) remains a challenge today given its complex biology and aggressiveness, and the absence of specific early symptoms [1]. The effect of current and new therapies on overall survival also remains very modest [2], prompting the need to search for biomarkers that could improve early diagnosis, prognosis, and treatment [3]. Sequential Window Acquisition of all Theoretical Mass Spectra (SWATH-MS) has recently emerged as a promising new tool in cancer proteomics, making it possible to identify biomarkers of increasing stages of invasiveness in MM experimental models, for example [4].

Proteomic analyzes of MM have already provided lists of putative cancer biomarkers, although significant differences are observed between primary and commercial MM cell lines [5],

for example, emphasizing the need to use best-suited preclinical cellular models [6]. Moreover, long-established human cell lines [7], some genetically engineered mouse models [8], and subcutaneous xenograft models of human tumors [9,10] often fail to predict drug effects in clinical practice. Therefore, to recapitulate the spectrum of tumor heterogeneity seen in patients, and limit the impact of differences in stromal conditions observed between patient and cancer models, cross-species proteomic analyzes are suggested to improve preclinical evaluation [11].

Remembering the importance of potential sources of bias when identifying biomarkers with potential application in oncology [12,13], to determine which invasiveness biomarkers identified in MM experimental models evolved similarly in human MM, we compared lists of proteins of interest from three biological sources. Data were collected firstly from two MM cell lines, secondly from rat MM tumors grown in syngeneic immunocompetent animals and human MM tumors grown in immunodeficient mice, and thirdly from paraffin-embedded sections of patient MM tumors of the epithelioid and sarcomatoid subtypes. The results identified one main biomarker, CAPG, associated with invasiveness and common to all three categories of tumors and human cell lines. Moreover, two other biomarkers were common to the three tumor sources, while 25 other candidates of interest were shared by rat and patient MM tumors. Finally, comparing these data with proteomic analyzes of a large collection of preneoplastic and neoplastic rat mesothelial cell lines revealed the additional role of SBP1 in the carcinogenic process.

#### **2. Results**

#### *2.1. Characterization of Cell Lines and MM Tumors*

The four rat MM tumor models shared a sarcomatoid morphology of tumor cells but differed in their infiltrative potential. The M5-T2 tumor is noninvasive, with tumor cell development restricted to the omentum without liver capsular breakthrough (Figure 1A, top left). The F4-T2 tumor is moderately invasive with a regular tumor front (Figure 1A, top right). The F5-T1 and M5-T1 tumors are both characterized by deep infiltration of the liver with irregular tumor fronts, however their respective tumor cells differ in their levels of atypia (Figure 1A, bottom). The highly invasive nature of the M5-T1 tumor is also revealed by necrosis of the liver parenchyma and the presence of apoptotic hepatocytes at the tumor front (Figure 1A, bottom right), associated with the specificities of its proteome [4]. The mean time required for macroscopic tumor development following the injection of 3–5 × 10<sup>6</sup> cells i.p. into syngeneic rats also differs among the four models: five weeks for M5-T2, four weeks for F4-T2, and three and a half weeks for F5-T1 and M5-T1.

The tumor rate development of the two models of human MM xenografts grown in NOD SCID mice (mice homozygous for the severe combined immune deficiency spontaneous mutation Prkdcscid , characterized by an absence of functional T cells and B cells) also differed markedly, with six weeks for MM34 versus three and a half weeks for MM163. These differences were also confirmed at microscopic level, as MM163 was characterized by tumor cells with heterogeneous nuclei in size and shape, prominent nucleoli, the presence of mitotic figures, and frequent atypia (Figure 1B, right) compared with MM 34 (Figure 1B, left).

The two sarcomatoid MM tumors from patients (SMM-1 and S-MM-2) were characterized by abundant intercellular collagen deposition, the presence of spindle-shaped tumor cells with oval nuclei, considerable heterogeneity in cell dimensions, and frequent atypia (Figure 1C, right column). The two epithelioid MM tumors (EMM-1 and EMM-2) contained tumor cells with abundant eosinophilic cytoplasm, round nuclei, and mild nuclear atypia (Figure 1C, left column).

One of the most frequent genomic alterations found in MM concerned *CDKN2A*, observed in the different histologic types [14]. Analysis of mRNA levels of this gene by qRT PCR in cell lines from the two species has previously revealed a comparable decreased relative expression in human pleural MM cell lines relative to normal mesothelial cells, and in rat MM cell lines relative to preneoplastic mesothelial cell lines [15]. Additionally, *Cdkn2a* relative expression was even more decreased in the three invasive MM cell lines (F4-T2: 2.54; F5-T1: 2.10; and M5-T1: 0.79) relative to the non-invasive M5-T2 cell line (4.97) [15]. The bi-allelic deletion of the *CDKN2A* gene, further confirmed in a list of MM human cell lines including the least invasive MM34 (Meso 34), was found to be strongly associated with overexpression of *IL34* and weakly with mutations of the *NF2* gene (with no association with other genetic alterations in *BAP1*, *LATS2* or *TP53* genes) [16]. MM163 (Meso 163) differed from MM34 by a homozygous deletion of the *IFNB1* gene (located in the same 9p21.3 chromosome region as *CDKN2A*) that encodes IFN-β [17]. A transcriptomic analysis of the group of human MM cell lines sharing the same features as MM163, comparing cells exposed to measles virus with untreated cells, revealed these cells were characterized by a weak IFN-I response, some canonical pathways involved in antigen presentation and cytotoxic T lymphocyte-mediated apoptosis of target cells being particularly hit [17].

**Figure 1.** Histological features of the three sources of malignant mesothelioma MM tumors. High magnification views, hematoxylin-phloxine-saffron (HPS) staining (×800, scale bars represent 25 µm), and general views in inserts (×25, scale bars represent 1 mm) with open red arrows indicating the location of magnifications. (**A**) Rat MM tumors of the four experimental models (the names of the corresponding cell lines are indicated on the external side of the photographs). These representative

tumor (T) histological sections included liver tissue (L) and tumor cells exhibiting increasing levels of invasiveness. (**B**) Xenografts of human MM tumors grown in NOD SCID mice, with the corresponding cell line names indicated on the external side of the photographs. (Om) = omentum, (G) = gut, (S) = spleen. The large open arrow shows a mitotic figure. (**C**), Human MM tumors from patients. EMM-1 and EMM-2 (left column) = epithelioid histotype, SMM-1 and SMM-2 (right column) = sarcomatoid histotype.

#### *2.2. Main Biomarkers Sharing the Same Evolution in the Three Sources of MM Tumors*

SWATH-MS data on increased MM tumor invasiveness were collected from (1) comparison of the three invasive rat MM tumors (F4-T2, F5-T1, M5-T1) versus the noninvasive one, M5-T2; (2) comparison of Meso 163 xenografts versus Meso 34 human MM tumors grown in immunodeficient mice; and (3) comparison of human MM tumors from patients, sarcomatoid versus epithelioid subtypes. The main findings are summarized in Table 1. The number of proteins with a fold change > 1.5 and statistical *p*-value < 0.05 estimated by MarkerView was 433, 133, and 191 in each experiment, respectively. Volcano plots for comparisons (1) (2) and (3) are provided in Figure 2A–C, respectively. Comparing these lists, represented by the green, brown, and orange circles, respectively (illustrated in Figure 2D), led us to identify a first pattern of common changes observed in the three experiments and shared by the macrophage-capping protein (encoded by *CAPG*), the fatty acid-binding protein, adipocyte (encoded by *FABP4*), and the laminin subunit beta-2 (encoded by *LAMB2*). These proteins are involved in actin filament finding, lipid transport (fatty acid binding) and extracellular matrix constitution (cell adhesion), respectively. Additional consideration of the comparison of Meso 163 versus Meso 34 human MM cell lines revealed that CAPG was the only biomarker exhibiting similar changes (a strong tendency was also observed for LAMB2), while there were no significant changes for FABP4 (Table 1 and Figure 3). No additional change was observed in the comparison of invasive versus noninvasive rat MM cell lines for the three proteins.


**Table 1.** Summary of proteomics biomarkers ofMM invasiveness and carcinogenesis. Abundance changes: + *p* < 0.05; - ns (*p* > 0.09); (+) tendency (0.05 < *p* < 0.09).

A second pattern of common changes was represented by proteins sharing the same differences between rat and human MM but showing no significant changes in human MM xenografts. Three proteins were involved, poly [ADP-ribose] polymerase 1 (encoded by *PARP1*), vesicle-fusing ATPase (encoded by *NSF*), and inosine-5′ -monophosphate dehydrogenase (encoded by *IMPDH2*). These proteins are involved in DNA repair, vesicle-mediated transport (Golgi) and de novo synthesis of guanine nucleotides, respectively. This situation confirms that transplantable tumors established subcutaneously in immunodeficient mice are less relevant in terms of stromal/vascular interactions than orthotopic models of tumors in syngeneic animals [6]. However, these limitations were counterbalanced by the observation of tendencies toward an increase in MM163 vs. MM34 xenografts, while significant differences were also found between corresponding human cell lines (Table 1 and Figure 3).

**Figure 2.** *Cont.*

**Figure 2.** Volcano plots and schematic representation of the comparative proteomic analyzes. (**A**), Volcano plot of the comparison of the three invasive rat MM tumors (F4-T2, F5-T1, and M5-T1) versus the noninvasive one, (M5-T2). (**B**), Volcano plot of the comparison of Meso 163 versus Meso 34 xenografts of human MM tumors grown in immunodeficient mice. (**C**), Volcano plot of the comparison of human MM tumors from patients, sarcomatoid versus epithelioid subtypes. The locations of CAPG (in red), FABP4 and LAMB2 (in blue) are indicated in the three volcano plots. (**D**), Schematic representation of the comparative proteomic analyzes. The three different sources of MM tumors are illustrated by the green (Rat MM), brown (xenografts of human MM grown in NOD SCID mice) and orange (human MM from patient tumor samples) circles. The green circle represents the 433 proteins showing significant changes in abundance (*p* < 0.05) between the three invasive rat MM tumors versus the noninvasive one. The brown circle illustrates the 133 proteins showing significant changes in abundance (*p* < 0.05) between Meso 163 (MM163) and Meso 34 (MM34) xenografts. The orange circle represents the 191 proteins affected by significant changes in abundance (*p* < 0.05) between the two sarcomatoid versus the two epithelioid MM tumors from patients. Genes coding for proteins exhibiting common significant changes are given for *homo sapiens* in italics (increase in red, decrease in blue).

**Figure 3.** Common biomarkers of MM invasiveness. Proteins showing comparable abundance changes in MM from the three sources and between human mesothelioma cell lines. Increase and decrease are indicated by red and blue bars, respectively (with *p* values). Blank bars reflect the absence of significant changes (*p* > 0.09), while light red or blue bars indicate tendencies (0.05 < *p* < 0.09).

#### *2.3. Additional Biomarkers of Interest Common to Rat and Human MM*

Several additional conclusions were drawn from the common changes in abundance limited to rat and patient MM tumors. Firstly, in the 3 versus 1 comparative analysis (Figure 2D), among the 18 proteins exhibiting a common increase, annexin A5 (encoded by *ANXA5*), involved in the blood coagulation cascade (anticoagulant), was the only one showing the same pattern of changes in human MM cell lines (Figure 4). Moreover, three more proteins revealed the same tendency, cytochrome c oxidase subunit 2 (encoded by *MT-CO2*), vesicle-trafficking protein SC22b (encoded by *SEC22B*), and fibronectin (encoded by *FN1*) (Table 1 and Figure 4). These proteins are involved in electron transport (respiratory chain), vesicle-mediated transport (membrane), and extracellular matrix structural composition (cell adhesion and motility), respectively. Finally, among the seven other proteins exhibiting a decreased abundance, two presented the same pattern, selenium-binding protein 1 (encoded by *SELENBP1*), an oxidoreductase also involved in protein transport, and synaptic vesicle membrane protein VAT-1 homolog (encoded by *VAT1*), which negatively regulates mitochondrial fusion (Table 1 and Figure 4).

The rest of the proteins listed in the 3 versus 1 comparison involved candidate biomarkers for which the difference in abundance between cells was insignificant (*p* > 0.090). By order of magnitude, proteins showing increased abundance with invasiveness included Ras-related protein Rab-31 (encoded by *RAB31*), Ragulator complex protein LAMTOR1 (encoded by *LAMTOR1*), isoform LCRMP-4 of dihydropyrimidinase-related protein 3 (encoded by *DPYSL3*), leucine-rich repeat-containing protein 59 (encoded by *LRRC59*), isoform 2 of tropomyosin alpha-3 chain (encoded by *TPM3*), endoplasmic reticulum resident protein 29 (encoded by *ERP29*), PRA1 family protein 3 (encoded by *ARL6IP5*), ferritin light chain (encoded by *FTL*), isocitrate dehydrogenase [NAD] subunit alpha, mitochondrial (encoded by *IDH3A*), V-type proton ATPase subunit B, brain isoform (encoded by *ATP6V1B2*), and 40S ribosomal protein S18 (encoded by *RPS18*) (Table 1 and Figure 5). Finally, proteins exhibiting a common decrease in both rat and human MM from patients were EH domain-containing protein 2 (encoded by *EHD2*), septin-7 (encoded by *SEPTIN7*), serum albumin (encoded by *ALB*), and two subunits of hemoglobin (encoded by *HBA1* and *HBB*) (Table 1 and Figure 5).

**Figure 4.** *Cont.*

**Figure 4.** Main invasiveness biomarkers in human vs. rat MM, and cell lines. Proteins showing comparable abundance changes in MM from rat models and patients, and human mesothelioma cell lines. Increase and decrease are indicated by red and blue bars, respectively (with *p* values). Blank bars reflect the absence of significant changes (*p* > 0.09), while light red bars indicate tendencies (0.05 < *p* < 0.09).

#### *2.4. Candidate Biomarkers Common to Xenografts and Rat or Patient MM*

Compared with the previous situation (3 versus 1), the numbers of common proteins found in conditions 2 versus 1, and 3 versus 2, were significantly reduced (Figure 2D). Among these lists, the parallel increase in prohibitin (encoded by *PHB*), and decrease in peroxiredoxin-6 (encoded by *PRDX6*) and ezrin (encoded by *EZR*) have previously been reported to be linked to the acquisition of invasive properties in rat MM models [4]. Moreover, these lists contain several candidate invasiveness biomarkers common to MM and other cancer types and reported in the literature, including gelsolin (encoded by *GSN*), profiling-1 (encoded by *PFN1*), glutathione-S-transferase P (encoded by *GSTP1*), keratin, type I cytoskeletal 10 (encoded by *KRT10*), and serpin H1 (encoded by *SERPINH1*) [13].

#### *2.5. Abundance Changes during Rat MM Carcinogenesis*

We next investigated whether some of the 28 candidate biomarkers (the 18 increased and 7 decreased proteins listed in the 3 versus 1 comparison, plus CAPG, FABP4, and LAMB2) common to the rat and human MM (Figure 2D) exhibited additional abundance changes during the carcinogenesis process. For that purpose, we first examined the SWATH-MS proteomic data of the whole biocollection of rat mesothelial cell lines, looking in particular at the list of 674 proteins differentiating preneoplastic cell lines with sarcomatoid versus epithelioid morphology [18]. In a second step, we compared this list to another list of 192 proteins discriminating the two subgroups of preneoplastic cell lines with sarcomatoid morphology PNsarc2 vs. PNsarc1, which differ in their relative expression of *Hif1a* [18]. Finally, comparing the 94 proteins exhibiting significant abundance changes in the two previous situations with the 28 candidate biomarkers described above (see Figure 2D and Sections 2.2 and 2.3), led to six proteins common to the four proteomic analyzes (Figure 6A). The absence of FABP4 in this list (the protein was not detected in cells) suggests a location in the stroma.

**Figure 5.** *Cont.*

**Figure 5.** Additional invasiveness biomarkers in MM from patient and rat models. Proteins showing comparable abundance changes only in patient MM vs. rat models. Increase and decrease are indicated by red and blue bars, respectively (with *p* values). Blank bars reflect the absence of significant changes (*p* > 0.09). For clarity, data on the beta subunit of hemoglobin (encoded by *HBB*) have been excluded as they were similar to those observed for the alpha subunit (encoded by *HBA*).

**Figure 6.** *Cont.*

**Figure 6.** Biomarkers of human vs. rat MM and rat mesothelial cell carcinogenesis. (**A**), Diagram of the methodology used to identify biomarkers showing additional changes during the course of rat mesothelial cell carcinogenesis. For both CAPG (**B**) and RAB31 (**C**), a common rise in abundance was specifically observed in PNsarc2 vs. PNsarc1 and between the whole groups of preneoplastic cell lines with sarcomatoid vs. epithelioid morphology. (**D**), Evolution of abundance changes for SBP1. (**E**), Evolution of abundance changes for LAMB2.

Interestingly, among these six proteins, selenium-binding protein 1 (SBP1, encoded by *Selenbp1*) was the only one exhibiting a continuous decrease from the different subgroups of preneoplastic cell lines with epithelioid morphology to PNsarc1 and PNsarc2, including a final additional decrease in neoplastic cell lines (Table 1 and Figure 6D). Conversely, for CAPG and RAB31, protein abundances in neoplastic cells differed significantly from only one of the two groups of preneoplastic cell lines (Table 1 and Figure 6B,C). For comparison, proteomic data for LAMB2 revealed the absence of significant changes within the different groups and subgroups of preneoplastic cell lines, while there was a dramatic decrease in all neoplastic cell lines (Table 1 and Figure 6E). For fibronectin, the evolution of abundance showed a progressive rise within the first four subgroups of preneoplastic cell lines but as above discrimination with neoplastic cells was incomplete (Table 1 and Figure S1). Finally, for TPM3 and VAT1, no clear evolution was observed within the different groups and subgroups of preneoplastic cells in comparison with neoplastic cells (Table 1 and Figure S1).

#### **3. Discussion**

This study investigated the proteomic changes associated with MM invasiveness that were common to experimental and human cell lines or tumor models generated in the F344 rat, human tumor xenografts, and tumor specimens from patients. Our investigations identified three major invasiveness biomarkers not documented so far in integrative molecular studies characterizing MM [14], common to the three tumor sources, CAPG, FABP4, and LAMB2, and an additional set of candidate biomarkers shared by rat and patient tumors. Among these, SBP1 appeared to play an additional crucial role in the carcinogenic process of mesothelial cells.

CAPG, together with ANXA5 and FABP4, was previously found within a group of biomarkers differentiating invasive from noninvasive MM rat tumor models, their abundance being very significantly increased and decreased, respectively [4]. This actin filament end-capping protein was initially reported to be increased in the transformation of human breast cancer cells into a highly metastatic variant [19]. Herein, we confirm that CAPG is also increased in human MM cell lines, human MM tumor models, and patient MM. Interestingly, our observations are consistent with several previous reports showing this protein's overexpression in different cancer types. Its role in promoting the invasiveness of cholangiocarcinoma and hepatocellular carcinoma has been established by Morofuji et al. [20], and Kimura et al. [21], respectively. Its involvement in migration and invasiveness has been documented for ovarian carcinoma by Glaser et al. [22], and for breast cancer by Davalieva et al. [23] and Huang et al. [24]. Its upregulation in clinical high-grade glioblastoma has also been reported by Xing and Zeng [25], while the correlation of its expression level with shorter survival time was demonstrated by Fu et al. [26]. Moreover, the link between its abundance and occurrence of lymph node metastasis has also been documented for three different types of cancer [20,27,28], as well as its association with the prediction of response to treatment [20,29].

FABP4 (also called A-FABP or aP2) is 1 of 10 members of a family of proteins involved in intracellular fatty acid transport and lipid trafficking regulation in cells, which show different tissue-specific expression patterns [30]. Its previously mentioned adipokine function regulating macrophage and adipocyte interactions during inflammation [31] may be consistent with the absence of significant differences observed in our study between mesothelial and MM cell lines. We previously reported that the extent of the decrease was related to increasing invasiveness in rat MM [4]. Interestingly, our observations also agree with the findings of Mathis et al. showing that FABP4 loss was associated with high stage/grade and the presence of metastatic lymph nodes in invasive bladder cancer [32]. Zhong et al. have also demonstrated that similar observations are made in hepatocellular carcinoma, with the protein's overexpression leading to tumor growth inhibition in vivo [33]. A second common protein exhibiting a decreased abundance in all tumor sources was laminin subunit beta-2 (LAMB2). This protein belongs to a family of 16 laminin isoforms, which combine with subunits of collagen IV to build the basement membranes surrounding blood vessels, lymphatics, nerves, and muscle cells. Hewitt et al. initially reported that within carcinomas, vascular basement membrane staining for the subunit beta-2 is clearly weaker relative to normal tissues, probably due to their incomplete maturation [34]. This observation was further confirmed by immunohistochemistry by Mustafa et al. when studying angiogenesis in glioblastoma [35]. The fascinating aspects of their structural diversity have been emphasized by Hohenester and Yurchenco [36], raising crucial questions on the challenge that studying their complex interactions in vivo presents.

The first of an additional subset of common biomarkers of interest, which differed from the previous three by the absence of significant changes in xenografts (only a tendency), was represented by PARP1. The recent development of PARP1 inhibitors for the treatment of cancers presenting compromised HR repair has led to interesting findings on biomarkers associated with their clinical use against MM [37]. Moreover, Gaetani et al. revealed the relationship between PARP1 and miR-126 regulation in the context of asbestos-induced malignancy [38]. Regarding NSF, changes have not yet been documented in the context of cancer invasiveness; however. our data suggest that the increase commonly observed is related to the reassembly pathway of Golgi cisternae at the end of mitosis [39]. Finally, our results are consistent with the recent finding by Kofuji et al. that overexpression of the rate-limiting enzyme for de novo guanine nucleotide biosynthesis, IMDH2, relative to primary glia, promotes glioblastoma tumorigenesis [40]. Among the other biomarkers for which no changes were observed in xenografts, the most significant differences in abundance were found for annexin A5. The potential of the smallest member of the annexin family as a predictive biomarker for tumor development, metastasis, and invasion has already been reviewed [41], with it also being involved in cell membrane repair [42]. Our results are consistent with reports of its overexpression in several other cancer types, including renal cell carcinoma [43], colon cancer [44], and hepatocarcinoma [45,46]. Other highly significant changes mainly involve two proteins, COX2 for increase and VAT1 for decrease. Cytochrome c oxidase dysfunction has already been demonstrated to be related to the Warburg effect in invasive cancers [47]. The involvement of VAT1, a largely uncharacterized enzyme,

has also been reported in the regulation of cancer cell motility and its interaction with Talin-1, a key cytoskeletal protein [48].

Two other proteins caught our attention among the second additional subset of common biomarkers of interest, EHD2 and RAB31, characterized by highly significant changes in abundance in both rat and patientMM, but not in human cell lines or in xenografts. The level of the first protein, which belongs to the EHD family associated with plasma membrane, has been reported to be reduced in human esophageal squamous carcinoma in comparison with adjacent normal tissues, linked to increased motility of the tumor cells [49]. Subsequently, a decreased expression was also observed, correlated with histological grade, in an immunohistochemical study of 96 human breast carcinoma samples, leading Shi et al. to suggest that this protein inhibits metastasis by regulating EMT [50]. The second protein, which belongs to the small GTPase family Rab and to the Rab5 subfamily, presents an estrogen receptor-responsive element in its promoter region which can be dysregulated in breast cancer cells, the consequences of this key finding in cancer research having been reviewed by Chua and Tang [51].

Both CAPG and RAB31 shared a similar pattern of changes during the course of rat mesothelial cell carcinogenesis. However, these changes were only observed in the first two subgroups of preneoplastic cell lines with sarcomatoid morphology, suggesting a link to increased *Hifa* expression [18]. The pattern of changes observed for SBP1 markedly contrasted with these situations as decreases in abundance were observed at three main stages of the carcinogenic process. Firstly, the decrease observed between PNep and PNint was concomitant with the first dramatic decrease in the expression of *Cdh1* and *Il10*, and parallel increase in the expression of *Acta 2*, *Tgfb1* [15]. Secondly, the new decrease observed between PNsarc1 and PNsarc2, and continuous decrease from PNep to PNsarc2, confirm the existence of links to both the level of expression of *Hifa* [18] and EMT process [15]. Thirdly, the decrease observed between preneoplastic cell lines with both epithelioid and sarcomatoid morphologies and neoplastic cell lines leads to the conclusion that SBP1 presents additional interest as a biomarker of neoplastic transformation. Finally, the decrease in SBP1 also observed in association with increased invasiveness in human cell lines, rat and patient MM tumors tends to confirm the protein's crucial role. The downregulation of another selenium-containing protein was earlier reported by Apostolou et al., suggesting that selenium could be useful as a chemopreventive agent in individuals at high risk of MM due to asbestos exposure [52]. Interestingly, Rundlöf et al. found differential expression within isoforms of the selenoenzyme thioredoxin reductase 1 (TrxR1) in MM cell lines, with the sarcomatoid phenotype showing the lower total TrxR1 mRNA level [53]. The mechanisms by which dietary selenium may affect MM tumor progression have only been partly explored, mostly in cell lines, pointing to the crucial role of redox metabolism [54]. Although it is well established that low levels of SBP1 are frequently associated with poor clinical outcome [55], the complexity of selenium metabolism has highlighted the fact that among selenocysteine-containing proteins that are members of the glutathione peroxidase family, SBP1 is the only one for which no catalytic function has been assigned [56]. Therefore, many aspects of this research field require further investigation. To give just a few more very recent examples of the protein's importance, Lee et al. have suggested that hepatitis B virus-X-expressing cells, which show markedly decreased *SELENBP1* expression, might be one factor in the development of hepatocellular carcinoma caused by HBV infection [57]. Wang et al. have also reported this protein's novel function in transcriptionally modulating p21 expression through a p53-independent mechanism, with a resulting impact on the G0/G<sup>1</sup> phase cell cycle arrest in bladder cancer [58].

#### **4. Materials and Methods**

#### *4.1. Study Approval*

The human studies were conducted according to the ethical guidelines of the Declaration of Helsinki. The paraffin-embedded human MM tumor pieces were prepared from samples of the Tumor Bank of the Reims University Hospital Biological Resource, Collection No. DC-2008-374, declared to the Ministry of Health according to French law, for the use of tissue samples for research. The two

human cell lines MM34 (Meso 34) and MM163 (Meso 163) were established from pleural effusions of patients with suspected pleural MM [59], according to the ethics committee approval (Comité de Protection des Personnes Ouest IV-Nantes, dossier n◦ DC-2011-1399). The animal studies were carried out in agreement with European Union guidelines for the care and use of laboratory animals in research protocols (Agreement #01257.03). All experiments were approved by the ethics committee for animal experiments of the Pays de la Loire Region, France (CEEA.2011.38 and CEEA.2013.7.).

#### *4.2. Rat and Human Cell Lines, and Tumor Samples*

The 27 cell lines of the rat biocollection were grown in RPMI 1640 medium, supplemented with 10% heat-inactivated fetal calf serum, 2 mM L-glutamine, 100 U/mL penicillin, and 100 µg/mL streptomycin (all reagents from Gibco Life Technologies Limited, Paisley, UK) at 37 ◦C in a 5% CO<sup>2</sup> atmosphere. Cells were collected from preconfluent 75 cm<sup>2</sup> flasks and cell pellets of 2 × 10<sup>6</sup> cells were used for SWATH-MS proteomic analysis after washing in PBS buffer. The four rat neoplastic cell lines (M5-T2, F4-T2, F5-T1, and M5-T1) were injected into syngeneic rats, and tumors collected and fixed as previously described [4]. The two human cell lines Meso 34 and Meso 163 were established from pleural effusions of patients with suspected pleural MM, aseptically collected by thoracocentesis as previously described [56], and cultivated as rat cell lines. Meso 34 and Meso 163 xenografted tumors were collected and fixed after injection of the corresponding cell lines into the peritoneal cavity of two groups of five immunodeficient NOD SCID mice. For patient tumors, four pieces of paraffin-embedded pleural MM tumor pieces collected from four different patients were obtained from the Tumor Bank of the Reims University Hospital Biological Resource. They represented two tumors of the sarcomatoid subtype (S-MM1 and S-MM2) versus two tumors of the epithelioid subtype (E-MM1 and E-MM2).

#### *4.3. SWATH-MS Analysis*

The spectral libraries, DDA experiments, peptide identification, and peak extraction of the SWATH data were performed as previously described [4], using either Spectronaut software (v 8.0, Biognosys, Schlieren, Switzerland) or the SWATH micro app embedded in PeakView (v 2.0, AB Sciex Pte. Ltd., Framingham, MA, USA). Sections of the tumors, stained with hematoxylin-phloxine-saffron (HPS), were first examined to select areas of interest, then removed with a scalpel. Five 20 µm thick sections of the samples were used, and the areas of interest collected in a microtube. Samples were deparaffinized, and then cell pellets and dried deparaffinized tumor samples treated as previously described [4]. After centrifugation, salts were removed using OASIS® HLB extraction cartridges (Waters SAS., St Quentin-en-Yvelines, 78, France), and the samples dried under SpeedVac. Peptide concentrations of the samples were determined using the Micro BCATM protein assay kit (Thermo Fisher Scientific, St Herblain, 44, France).

Five micrograms of each sample were analyzed with a SWATH-MS acquisition method. The method of acquisition, peak extraction of the SWATH data, calibration of the retention time of extracted peptide peaks and quantification followed the procedure already described in [4]. For statistical analysis of the SWATH data set, the peak extraction output data matrix from PeakView was imported into MarkerView (v 2, Sciex, Framingham, MA, USA) for data normalization and relative protein quantification. Proteins with a fold change >1.5 and statistical *p*-value < 0.05 estimated by MarkerView were declared differentially expressed under different conditions.

#### **5. Conclusions**

This study pointed to some proteins of interest that exhibited the same patterns of quantitative changes in different situations, and for which the relationship with tumor invasiveness has already been reported in the literature for other cancer types. Although this study was limited by the small number of samples, an interesting point was the similarity of observations made on malignant mesothelioma cells and tumors from different sources and from two different species. Extending these studies to a larger number of samples would be the logical next step, which may later contribute to improving current therapies for patients with the worst survival outcomes. Another interesting prospect is related to the questions raised by the additional involvement of the selenium-binding protein 1 in the carcinogenic process, a point that would present a good basis for further basic research in cancerology, and probably also for improving early MM diagnosis.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2072-6694/12/9/2430/s1, Figure S1: Additional biomarkers of human vs. rat MM and rat mesothelial cell carcinogenesis.

**Author Contributions:** Conceptualization of experiments (rat experimental cell lines, rat MM tumor models, human MM cell lines, and MM34 and MM163 xenografts in NOD SCID mice), J.S.N. and D.L.P. Preparation of samples from patient tumors and histological examination, I.V. and V.V. Preparation of samples for proteomic analysis, D.L.P., A.B. and C.H. Relative quantification by SWATH-MS acquisition and statistical analysis, validation, C.G. Formal analysis, D.L.P. Funding acquisition, M.G., D.L.P., C.G. and O.C. Original draft preparation, D.L.P. All authors have read and agreed to the published version of the manuscript.

**Funding:** This study was conducted with the support of the Ligue contre le Cancer (Ligue inter-régionale du Grand-Ouest, comités 16, 29, 44, 72), the Fondation pour la Recherche Médicale (FRM), and the "Comité Féminin 49 Octobre Rose".

**Acknowledgments:** The authors are indebted to Philippe Birembaut, CHU de Reims, Hôpital Maison Blanche, Laboratoire de Pathologie, F-51092 Reims, France, for his invaluable help in providing human MM tumor samples from patients from the Tumor Bank of the Reims University Hospital Biological Resource, Collection No. DC-2008-374.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

## *Article* **Malignant Pleural Mesothelioma Interactome with 364 Novel Protein-Protein Interactions**

**Kalyani B. Karunakaran <sup>1</sup> , Naveena Yanamala <sup>2</sup> , Gregory Boyce <sup>2</sup> , Michael J. Becich <sup>3</sup> and Madhavi K. Ganapathiraju 3,4,\***


**Simple Summary:** Internal organs like the heart and lungs, and body cavities like the thoracic and abdominal cavities, are covered by a thin, slippery layer called the mesothelium. Malignant pleural mesothelioma (MPM) is an aggressive cancer of the lining of the lung, where genetics and asbestos exposure play a role. It is not diagnosable until it becomes invasive, offering only a short survival time to the patient. To help understand the role of the genes that relate to this disease most of which are poorly understood, we constructed the 'MPM interactome', including in it the protein-protein interactions that we predicted computationally and those that are previously known in the literature. Five novel protein-protein interactions (PPIs) were tested and validated experimentally. 85.65% of the interactome is supported by genetic variant, transcriptomic, and proteomic evidence. Comparative transcriptome analysis revealed 5 repurposable drugs targeting the interactome proteins. We make the interactome available on a freely accessible web application, Wiki-MPM.

**Abstract:** Malignant pleural mesothelioma (MPM) is an aggressive cancer affecting the outer lining of the lung, with a median survival of less than one year. We constructed an 'MPM interactome' with over 300 computationally predicted protein-protein interactions (PPIs) and over 2400 known PPIs of 62 literature-curated genes whose activity affects MPM. Known PPIs of the 62 MPM associated genes were derived from Biological General Repository for Interaction Datasets (BioGRID) and Human Protein Reference Database (HPRD). Novel PPIs were predicted by applying the HiPPIP algorithm, which computes features of protein pairs such as cellular localization, molecular function, biological process membership, genomic location of the gene, and gene expression in microarray experiments, and classifies the pairwise features as interacting or non-interacting based on a random forest model. We validated five novel predicted PPIs experimentally. The interactome is significantly enriched with genes differentially ex-pressed in MPM tumors compared with normal pleura and with other thoracic tumors, genes whose high expression has been correlated with unfavorable prognosis in lung cancer, genes differentially expressed on crocidolite exposure, and exosome-derived proteins identified from malignant mesothelioma cell lines. 28 of the interactors of MPM proteins are targets of 147 U.S. Food and Drug Administration (FDA)-approved drugs. By comparing disease-associated versus drug-induced differential expression profiles, we identified five potentially repurposable drugs, namely cabazitaxel, primaquine, pyrimethamine, trimethoprim and gliclazide. Preclinical studies may be con-ducted in vitro to validate these computational results. Interactome analysis of disease-associated genes is a powerful approach with high translational impact. It shows how MPMassociated genes identified by various high throughput studies are functionally linked, leading to clinically translatable results such as repurposed drugs. The PPIs are made available on a webserver with interactive user interface, visualization and advanced search capabilities.

**Citation:** Karunakaran, K.B.; Yanamala, N.; Boyce, G.; Becich, M.J.; Ganapathiraju, M.K. Malignant Pleural Mesothelioma Interactome with 364 Novel Protein-Protein Interactions. *Cancers* **2021**, *13*, 1660. https://doi.org/10.3390/ cancers13071660

Academic Editors: Daniel L. Pouliquen and Joanna Kopecka

Received: 28 February 2021 Accepted: 22 March 2021 Published: 1 April 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

**Keywords:** malignant mesothelioma; protein-protein interactions; systems biology; network analysis; drug repurposing

#### **1. Introduction**

Internal organs such as heart and lung, and body cavities such as thoracic and abdominal cavities, are covered by a thin slippery layer of cells called the "mesothelium". This protective layer prevents organ adhesion and plays a number of important roles in inflammation and tissue repair [1]. The mesothelia that line the heart, lung and abdominal cavity are called pericardium, pleura and peritoneum, respectively. Mesothelioma is the cancer that originates from this lining (described in detail in a recent review article [2]). Most types of mesothelioma metastasize to different locations in the body [3]. Pleural mesotheliomas account for ~90% of malignant mesotheliomas and have a short median survival, of less than 1 year [4].

Malignant pleural mesothelioma (MPM) is associated with exposure to asbestos; it has a long latency period after exposure and is conclusively diagnosable only after reaching the invasive phase [3]. It tends to cluster in families and occurs only in a small fraction of the population exposed to asbestos, suggesting the involvement of a genetic component [5]. These factors necessitate expeditious discovery of genetic predispositions, molecular mechanisms and therapeutics for the disease.

The molecular mechanisms of disease are often revealed by the protein-protein interactions (PPIs) of disease-associated genes. For example, the involvement of transcriptional deregulation in MPM pathogenesis was identified through mutations detected in *BAP1* and its interactions with proteins such as *HCF1*, *ASXL1*, *ASXL2*, *ANKRD1*, *FOXK1* and *FOXK2* [6]. PPI of *BAP1* with *BRCA1* was central to understanding the role of *BAP1* in growth-control pathways and cancer; *BAP1* was suggested to play a role in *BRCA1* stabilization [7,8]. Studies on *BAP1* and *BRCA1* later led to clinical trials of the drug vinorelbine as a second line therapy for MPM patients, and the drug was shown to have rare or moderate effects in MPM patients [9,10]. *BAP1* expression was shown to be necessary for vinorelbine activity; 40% of MPM patients in a study showed low *BRCA1* expression and vinorelbine resistance [11–13]. Further, 60% of the disease-associated missense mutations perturb PPIs in human genetic disorders [14].

Despite their importance, only about 10–15% of expected PPIs in the human protein interactome are currently known; for nearly half of the human proteins, not even a single PPI is currently known [15]. Due to the sheer number of PPIs remaining to be discovered in the human interactome, it becomes imperative that biological discovery be accelerated by computational and high-throughput biotechnological methods. We developed a computational model, called HiPPIP (high-precision protein-protein interaction prediction) that is deemed accurate by computational evaluations and experimental validations of 18 predicted PPIs, where all the tested pairs were shown to be true PPIs ([16,17] and current work, and other unpublished works). HiPPIP computes features of protein pairs such as cellular localization, molecular function, biological process membership, genomic location of the gene, and gene expression in microarray experiments, and classifies the pairwise features as interacting or non-interacting based on a random forest model [16]. Though each of the features by itself is not an indicator of an interaction, a machine learning model was able to use the combined features to make predictions with high precision. The threshold of HiPPIP to classify a protein-pair as "a PPI" was set high in such a way that it yields very high-precision predictions, even if low recall. Novel PPIs predicted using this model are making translational impact. For example, they highlighted the role of cilia and mitochondria in congenital heart disease [18,19], that oligoadenylate synthetase-like protein (*OASL*) activates host response during viral infections through RIG-I signaling via its PPI with retinoic acid-inducible gene I (*RIG-I*) [17], and led to the identification of drugs

potentially repurposable for schizophrenia [20], one of which is currently under clinical trials.

In this work, we studied MPM-associated genes and their PPIs assembled with HiPPIP and analyzed the MPM interactome to draw translatable results. We demonstrate the various ways in which systems-level analysis of this interactome could lead to biologically insightful and clinically translatable results. We made the interactome available to the cancer biology research community on a webserver with comprehensive annotations, so as to accelerate biomedical research on MPM.

#### **2. Results**

We collected 62 MPM-associated genes from the Ingenuity Pathway Analysis (IPA) suite, which will be referred to as 'MPM genes' here; these genes have been reported to affect MPM through gene expression changes or genetic variants, or by being targeted by drugs clinically active against MPM (see details in Data File S1) [21]. Previously known PPIs of the 62 MPM genes were collected from Human Protein Reference Database (HPRD), version 9 [22] and Biological General Repository for Interaction Datasets (BioGRID) version 4.3.194 [23]. Novel (hitherto unknown) PPIs were predicted with HiPPIP, a computational model. We discovered 364 novel PPIs of MPM genes (Table 1), which are deemed highly accurate according to prior evaluation of the HiPPIP model including experimental validations [16]. The MPM interactome thus assembled has 2459 known PPIs and 364 novel PPIs among the 62 MPM-associated genes and 1911 interactors (Figure 1 and Data File S2). Nearly half of the MPM genes had 10 or less known PPIs each, and about 130 novel PPIs have been predicted for these (Figure 2). HiPPIP predicted 920 PPIs of which 556 PPIs were previously known, leaving 364 PPIs to be considered as novel PPIs of the MPM genes. There were an additional 1903 PPIs that are known and not predicted by HiPPIP. This is as expected because the HiPPIP prediction threshold has been fixed to achieve *high precision* by compromising *recall*, which is required for adoption into biology; in other words, it is set to predict only a few PPIs out of the hundreds of thousands of unknown PPIs, but those that are predicted will be highly accurate. It has to be noted that neither PPI prediction nor high throughput PPI screening can be performed with high-precision *and* high-recall. Co-immunoprecipitation (Co-IP) based methods show high-precision and extremely-low recall (detecting only one PPI at a time), whereas multi-screen high-quality yeast 2-hybrid methods show high-precision with low recall (detecting a few tens of thousands of PPIs). Thus, HiPPIP is on par with other methods in terms of precision and the number of new PPIs detected. 18 novel PPIs predicted by HiPPIP were validated to be true (validations have been reported in [16,17], the current work and other unpublished works); the experiments were carried out by diverse research labs.

**Table 1.** Novel Interactors of each of the malignant pleural mesothelioma (MPM) Genes: Number of known (K) and computationally predicted novel (N) protein-protein interactions (PPIs) and lists the novel interactors. Bold genes in the 4th column are Novel Interactors that were experimentally validated in the current study.



**Table 1.** *Cont*.


**Table 1.** *Cont*.

**Figure 1.** Malignant pleural mesothelioma (MPM) Protein-Protein Interactome: Network view of the MPM interactome is shown as a graph, where genes are shown as nodes and protein-protein interactions (PPIs) as edges connecting the nodes. MPM-associated genes are shown as dark blue square-shaped nodes, novel interactors and known interactors as red and light blue colored circular nodes respectively. Red edges are the novel interactions, whereas blue edges are known interactions.

**Figure 2.** Number of protein-protein interactions (PPIs) in the malignant pleural mesothelioma (MPM) Interactome: The 62 MPM genes are shown along the X-axis, arranged in ascending order of their number of known PPIs. Blue line, right-side axis: Number of known PPIs is shown. Red bars, left-side axis: Number of novel PPIs.

#### *2.1. Experimental Validation of Selected Protein-Protein Interactions (PPIs)*

We carried out experimental validations of five predicted PPIs chosen for their biological relevance and proximity to MPM genes, namely, *BAP1*-*PARP3*, *KDR*-*ALB*, *PDGFRA*-*ALB*, *CUTA*-*HMGB1* and *CUTA*-*CLPS*. They were validated using protein pull-down followed by protein identification using mass spectrometry (Table S1) or size-based protein detection assay (Figure 3). Each bait protein was also paired with a random prey protein serving as control (specifically, *BAP1*-phospholambin, *ALB*-*FGFR2* and *CUTA*-*FGFR2*). All predicted PPIs were validated to be true, while control pairs tested negative. In addition to these five, another PPI from the MPM interactome, namely *HMGB1*-*FLT1* was validated in our prior work through co-immunoprecipitation [16]. Three novel PPIs, namely *HLA-DQA1*— *HLA-DQB1*, *FGFR2*—*FGF2* and *CDKN2A*—*CDKN2B*, that we reported in the preprint of this work [24], have since been reported as known PPIs in a recent version of BioGRID (downloaded February 2021); these three are treated as known PPIs in the remaining description.

**Figure 3.** Validation of predicted *ALB* interactions and *CUTA* interactions using Wes™ Simple Western total protein detection assay: Pseudo-gel or virtual-blot like image of the validated interactions of *ALB* (lanes 1–2) and *CUTA* (lanes 4, 7) along with negative control (lane 3). In addition to the final pull-down samples, wash and/or flow through after binding 'bait' and 'prey' proteins for the *CUTA* interactions are also shown (lanes 5, 6, 8 and 9). The electro-pherogram image of Simple Western results using Total protein size-based assay. (**A**) *ALB* interactions with true positives *KDR*/*VEGFR2*, *PDGFRA* and false positive *FGFR2*. (**B**) *CUTA* interactions with *HMGB1*. (**C**) *CUTA* interactions with *CLPS*. An overlay of the electro-pherogram of the wash from *HMGB1* after *CUTA* binding is also shown in (**C**) for comparison.

→

→ →

#### *2.2. Functional Interactions of Malignant Pleural Mesothelioma (MPM) Genes with Predicted Novel Interactors*

We used ReactomeFIViz [25], a Cytoscape plugin, to extract known functional interactions between MPM-associated genes and their novel interactors. Seven novel PPIs had such functional interactions, namely (MPM genes are shown in bold), *PDGFRB*-*RAPGEF1* ('*part of the same complex*', '*bound by the same set of ligands*'), *SP1*→*HNRNPA1* ('*expression regulation*'), *HLA-DQA1*→*HLA-DPB1*, *HLA-DQA2*→*HLA-DQA1* ('*part of the same complex*', '*catalysis*'), *CTLA4*-*CD28*, *PDGFRB*-*PLAUR* ('*bound by the same set of ligands*') and *FGFR2*- *MDM2* ('*ubiquitination*').

#### *2.3. Web Server*

We made the MPM interactome available on a webserver called *Wiki-MPM* (http: //severus.dbmi.pitt.edu/wiki-MPM). It has advanced-search capabilities, and presents comprehensive annotations, namely Gene Ontology, diseases, drugs and pathways, of the two proteins of each PPI side-by-side. Here, a user can query for results such as "PPIs where one protein is involved in mesothelioma and the other is involved in immunity", and then see the results with the functional details of the two proteins side-by-side. The PPIs and their annotations also get indexed in major search engines like Google and Bing; thus a user searching for '*KDR* and response to starvation' would find the PPIs *KDR*-*CAV1* and *KDR*-*ALB*, where the interactors are each involved in 'response to starvation'. Querying by biomedical associations is a unique feature which we developed in Wiki-Pi that presents known interactions of all human proteins [26]. Wiki-MPM is a specialized version for disseminating the MPM interactome with its novel PPIs, visualizations and browse features. The novel PPIs have a potential to accelerate biomedical discovery in mesothelioma and making them available on this web server brings them to the biologists in an easily-discoverable and usable manner. Wiki-MPM will be integrated into the National Mesothelioma Virtual Bank [27,28], and will be available to the mesothelioma research community as part of our translational support of cancer research.

#### *2.4. Pathway Analysis*

We compiled the list of pathways that any of the proteins of MPM interactome are associated with, using Ingenuity Pathway Analysis suite [29]. Top 30 pathways by statistical significance of association are shown in Figure 4A. A number of pathways such as *NFκB signaling*, *PI3/AKT signaling*, *VEGF signaling* and *natural killer cell signaling,* are highly relevant to mesothelioma etiology. They are found to be connected to MPM genes through novel PPIs that were previously unknown. For example, the PI3K/AKT signaling pathway regulating the cell cycle is aberrantly active in MPM, and the mesothelioma gene *FGFR1* is connected to this pathway via its novel predicted PPIs with *EIF4EBP1* and *PRP2CB* (Figure 4B) [30]. Statistical significance of association to the interactome, and various MPM genes and novel interactors belonging to these pathways are shown in Table 2 and Data File S3. A cancer biologist may utilize the Supplementary Data (Data Files S2 and S3) to study novel PPIs that connect MPM genes to a pathway that they are interested in studying.

**Table 2.** Pathways that are relevant to the pathophysiology and genetics of malignant pleural mesothelioma: Pathway analysis revealed that molecular mechanisms underlying various types of cancers, axonal guidance signaling, PI3/AKT signaling, VEGF signaling, natural killer cell signaling and inflammation signaling pathways may be pertinent to the development of MPM. A list of all the associated pathways is shown in Data File S3.


*κ*

**Figure 4.** Pathways associated with malignant pleural mesothelioma (MPM) interactome: (**A**) Number of genes from MPM interactome associated with various pathways are shown. Top 30 pathways based on significance of association with the interactome are shown. (**B**) PI3K/AKT Signaling Pathway: Dark blue nodes are MPM genes, light blue nodes are known interactors and red nodes are novel interactors. Nodes with purple labels are genes involved in the PI3K/AKT signaling pathway.

#### *2.5. Potentially Repurposable Drugs*

We previously identified drugs potentially repurposable for schizophrenia through interactome analysis, and one of them is currently in clinical trials (ClinicalTrials.gov Identifier: NCT03794076) and another clinical trial has been funded and is yet to start [20]. Following this methodology, we constructed the MPM drug-protein interactome that shows the drugs that target any protein in the MPM interactome. This analysis has been carried out on an earlier version of BioGRID (3.4.159), which had fewer known PPIs, as reported in the preprint version of the paper [24], and has not been recomputed with the latest version of BioGRID unlike the other analyses presented here. There are 513 unique drugs that target 206 of these proteins (of which 28 are novel interactors that are targeted by 147 drugs) (Figure 5 and Data File S4). We adopted the established approach of comparing druginduced versus disease-associated differential expression using the BaseSpace correlation software (previously called NextBio) [31,32], to identify five drugs that could be potentially repurposable for MPM (Table 3; the table also shows corresponding information for two known MPM drugs). These are: *cabazitaxel*, used in the treatment of refractory prostate cancer; *primaquine* and *pyrimethamine*, two anti-parasitic drugs; *trimethoprim*, an antibiotic; and *gliclazide*, an anti-diabetic drug (See Appendix A, titled 'Repurposable Drugs for Treatment of Malignant Pleural Mesothelioma'). The drugs were selected based on whether they induced a differential expression (DE) in genes that showed a negative correlation with lung cancer associated DE, and affected genes of high DE in MPM tumors/cell lines (GSE51024 [33] and GSE2549 [34]), or underwent prior clinical testing in lung cancer. Lung cancers share common pathways with mesothelioma initiated on asbestos exposure. Therefore, drugs targeting lung cancers can potentially be used in MPM [35]. Table 3 shows pharmacokinetic details of the drugs as reported in Drug Bank [36].

**Figure 5.** Malignant pleural mesothelioma (MPM) Drug-Protein Interactome: The network shows the drugs (green color nodes) that target the proteins in the MPM interactome. Larger green nodes correspond to drugs that target the anatomic category 'antineoplastic and immunomodulating agents'. The color legend for genes (proteins) is as shown in Figure 1, with MPM genes in dark blue, their known interactors in light blue and novel interactors in red.

**Table 3.** Pharmacokinetic details of known mesothelioma drugs and the drugs that are presented as candidates for repurposing. Known mesothelioma drugs are shown in bold italics. Score corresponds to scaled correlation score with lung cancer expression studies from BaseSpace (NextBio) analysis.



**Table 3.** *Cont*.

Although in each case, there would be some genes that are differentially expressed in the same direction for both the drug and the disorder (for e.g., both the drug and the disease cause some genes to overexpress), the overall effect on the entire transcriptome has an anti-correlation. A correlation score is generated based on the strength of the overlap between the drug and the disease datasets. Statistical criteria such as correction for multiple hypothesis testing are applied and the correlated datasets are then ranked by statistical significance. A numerical score of 100 is assigned to the most significant result, and the scores of the other results are normalized with respect to this top-ranked result. We excluded drugs with unacceptable toxicity (e.g., minocycline) or unsuitable pharmacokinetics. The final list comprised 15 drugs, out of which 10 have already been tested against mesothelioma in clinical trials/animal models, and several of them were found to display clinical activity [37–53] (Table S2). Gemcitabine and pemetrexed are being used as first-line therapy for mesothelioma, in combination with cisplatin [45,53]. Ipilimumab has been identified to be a potential second-line or third-line therapy in combination with nivolumab [47]. Ixabepilone stabilizes cancer progression for up to 28 months [49]. Zoledronate, which showed modest activity in MPM, induced apoptosis and S-phase arrest in human mesothelioma cells and inhibited tumor growth in an orthotopic animal model [54,55]. Sirolimus/cisplatin increased cell death and decreased cell proliferation in MPM cell lines [56]. α-Tocopheryl succinate increased the survival of orthotopic animal models of malignant peritoneal mesothelioma [57]. Pre-clinical testing of vitamin E and its analogs are in progress [58,59].

Primaquine targets *KRT7*, a novel interactor of *KRT5*, whose high expression has been correlated with tumour aggressiveness and drug resistance in malignant mesothelioma [60–62]. Primaquine may be re-purposed for MPM treatment at least as an adjunctive drug with pemetrexed, the drug currently used for first-line therapy. Primaquine enhanced the sensitivity of the multi-drug resistant cell line KBV20C to cancer drugs [63]. Gliclazide is an anti-diabetic drug inhibiting *VEGFA* [64], a known interactor of *KDR*, and is significantly upregulated in MPM tumour (Log2FC = 1.83, *p*-value = 0.0018). Glicazide inhibits VEGFmediated neovascularization [64]. High levels of VEGF have been correlated with both asbestos exposure in MPM and advanced cancer [65,66]. Glibenclamide, a drug with a similar mechanism of action as that of glicazide, increases caspase activity in MPM cell lines and primary cultures, leading to apoptosis mediated by *TRAIL* (TNF-related apoptosis inducing ligand) [67].

Eliminating those drugs which are being/have already been tested in mesothelioma with varying results, we arrived at a list of five potentially repurposable drugs in the descending order of negative correlation scores: pyrimethamine, cabazitaxel, primaquine, trimethoprim and gliclazide (Table 3). Cabazitaxel targets the MPM genes, *TUBB1* and

*TUBA4A*, and was effective in treating non-small cell lung cancer (NSCLC) that was resistant to docetaxel, a drug that targets *TUBB1* along with other known interactors of MPM genes [37]. Pyrimethamine and trimethoprim target the MPM gene *TYMS* involved in folate metabolism, which was found to be differentially expressed in MPM tumors (GSE51024 [33]) (log2FC = 1.82, *p*-value = 4.10 × 10−17). MPM tumors have been shown to be responsive to anti-folates [68].

#### *2.6. Analysis with Other High-Throughput Data*

This section describes the overlap of the MPM interactome with various types of MPMrelated biological evidence. 1690 (85.65%) proteins in the interactome were supported by genetic variant, transcriptomic, and proteomic evidence, and are listed in Data File S5. Table 4 shows 48 novel interactors that had three or more pieces of biological evidence.

**Table 4.** Novel interactors in the malignant pleural mesothelioma (MPM) interactome with biological evidences related to MPM. The table shows the following data in columns labeled A to F. (A) 48 novel interactors of MPM associated genes that have been linked to four or more biological evidences related to MPM, namely, **B1**: high or medium gene expression in lungs, **B2**: differential gene expression in MPM tumor versus other thoracic tumors, **B3**: differential gene expression in MPM tumor versus normal adjacent pleural tissue, **B4**: differential gene expression in MPM tumors of epithelioid, biphasic and sarcomatoid types, **B5**: differential gene methylation in MPM, **B6:** gene expression correlated with unfavorable lung cancer prognosis, **B7**: differential gene expression on exposure to asbestos or asbestos-like particles, **C**: isolation as exosome-derived proteins from malignant mesothelioma cell lines, **D**: differential protein abundance levels in epithelioid and sarcomatoid types of malignant mesothelioma, and **E**: genetic variants in MPM. Last column, **F**, gives the total number of sources of evidences for each gene. The complete list of biological evidence for all the genes in the interactome can be found in Data File S5.



**Table 4.** *Cont*.

We compiled the list of genes harboring MPM-associated genetic variants from Bueno et al. [5], and compared this list with all the genes in the MPM interactome (i.e., MPM-associated genes, their known and novel interactors) to identify overlaps. 275 genes in the MPM interactome harbored either germline mutations, or somatic single nucleotide variants (SNVs) or indels (insertions or deletions) (Figure 6, Table 4 and Data File S5) associated with MPM tumors. Of these 275 genes, 37 were novel interactors of MPM genes. *MGMT* carried germline mutations while the following carried somatic mutations: *ASTN2*, *BARX1*, *BRD2*, *CALML5*, *CAPRIN1*, *CLK1*, *CPS1*, *DPYD*, *EIF3H*, *EPB41L3*, *GMPS*, *GPR12*, *ITGAM*, *KIAA1524*, *KMT2D*, *KRT4*, *MGAT4A*, *NBR2*, *NDUFV2*, *NFIB*, *NFX1*, *NUDC*, *PLCL1*, *PRDM2*, *PRKAG1*, *PRMT1*, *PTPRT*, *PTRH2*, *RBBP6*, *SGK3*, *SLC20A1*, *SMCHD1*, *SPOCK1*, *TMPRSS15*, *TNC* and *XPO4*. Fourteen of these interact with MPM genes that also harbored a genetic variant (MPM genes are shown in bold): *CDKN2A*-*NFX1*, *FLT1*-*LATS2*, *TUBA3C*-*XPO4*, *PDGFRA*-*SPOCK1*, *TYMS*-*SMCHD1*, *TYMS*-*EPB41L3*, *GART*-*TMPRSS15*, *TYMS*-*NDUFV2*, *TYMS*-*ITGAM*, *RRM2*-*BARX1*, *RRM2*-*MGAT4A* and *ATIC*-*CPS1*, *ATIC*-*KIAA1524* and *POLE*-*NOS1*.

**Figure 6.** Genes with biological evidences in the malignant pleural mesothelioma (MPM) Protein-Protein Interactome: On the interactome network shown in Figure 1, various biological evidences of relation to malignant pleural mesothelioma (MPM) are shown as node border colors. Genes with variants associated with MPM have orange borders, genes with MPM/lung cancer/asbestos exposure-associated gene/protein expression changes have light green-colored borders and genes with black border have both genetic variants and gene/protein expression changes associated with them. The gene expression-associated features include differential expression in MPM tumors versus normal adjacent pleura, MPM tumors versus other thoracic tumors, differential gene methylation (affecting gene expression) in MPM tumors, gene expression correlated with unfavorable lung cancer prognosis, differential gene expression on exposure to asbestos or asbestos-like particles and high/medium expression in normal lungs. The protein expression-associated features include isolation as exosome-derived proteins from malignant mesothelioma cell lines and differential protein abundance levels in epithelioid and sarcomatoid types of malignant mesothelioma. The complete list of genes in the interactome and their corresponding evidence can be found in Data File S5.

We collected the methylation profile of pleural mesothelioma [69], and found 8 novel interactors to be hypomethylated in pleural mesothelioma versus non-tumor pleural tissue, namely, *ACVR1B*, *IL6*, *MGMT*, *NRG1*, *OAT*, *PHLDA2*, *PLAUR* and *TNC* (Table S3). Some of them have little or no expression in lung tissue but are overexpressed in MPM. *PLAUR* is a prognostic biomarker of MPM [70]. Similarly, *FGFR1* and its novel interactor *NRG1* had elevated mRNA expression in H2722 mesothelioma cell lines and in MPM tissue, both contributing to increased cell growth under tumorigenic conditions [71,72]. *TNC*, involved in invasive growth, is a prognostic biomarker overexpressed in MPM, having low expression in normal lung tissues [73,74]. Thus, these novel interactors, which are not normally expressed in lung tissue, may be hypomethylated in MPM leading to their overexpression, contributing to MPM etiology.

Three hundred and ninety three (393) genes in the MPM interactome were also differentially expressed in mesothelioma tumors versus normal pleural tissue adjacent to tumor (GSE12345 [75]) (*p*-value of overlap = 9.525 × 10−19, odds ratio = 1.51). 52 out of the 314 novel interactors in the interactome were differentially expressed in this dataset (*p*-value = 0.046, odds ratio = 1.26). 938 genes, including 132 novel interactors, in the interactome were found to be differentially expressed in MPM tumors of epithelioid, biphasic and sarcomatoid types versus paired normal tissues (GSE51024 [33]) (*p*-value of overlap = 1.415 × 10−18, odds ratio = 1.24). Genes with fold-change >2 or <<sup>1</sup> <sup>2</sup> were considered as overexpressed and underexpressed, respectively, at a *p*-value < 0.05. Similarly, 744 genes in the MPM interactome were differentially expressed in MPM tumors versus other thoracic cancers such as thymoma and thyroid cancer (GSE42977 [76]) (*p*-value = 3.04 × 10−<sup>41</sup> , odds ratio = 1.53). 112 out of the 314 novel interactors in the interactome were differentially expressed in this dataset (*p*-value = 7.77 × 10−<sup>6</sup> , odds ratio = 1.45). This shows that the MPM interactome is enriched with genes whose expression helps in distinguishing MPM from other thoracic tumors and also with genes differentially expressed in mesothelioma tumors versus normal pleural tissue (Figure 6 and Data File S5). From RNA-seq data in GTEx, we found that 1311 genes, including 189 novel interactors, in the interactome have high/medium expression in normal lung tissue (median transcripts-per-million (TPM) > 9) (Figure 6 and Data File S5) [77].

A recent study had examined the gene expression profiles from the lungs of mice exposed to asbestos fibers (crocidolite and tremolite), an asbestiform fiber (erionite) and a mineral fiber (wollastonite) [78]. Crocidolite, tremolite and erionite are capable of inducing lung cancer and mesothelioma in humans and animal models [78]. On the other hand, wollastonite is a low pathogenicity fiber that shows no association with the incidence of lung cancer and mesothelioma in humans, or carcinogenesis in animal models [79]. The MPM interactome showed significant enrichment with all the 4 fibers (Figure 6 and Data File S5). The highest statistical significance was shown for the human orthologs of the mouse genes that were differentially expressed upon crocidolite exposure (199 genes, *p*-value = 1.16 × 10−<sup>18</sup> , odds ratio = 1.88). This was followed by tremolite (47 genes, *p*-value = 2.445 × 10−<sup>5</sup> , odds ratio = 1.87), wollastonite (16 genes, *p*-value = 0.0037, odds ratio = 2.09) and erionite (10 genes, *p*-value = 0.025, odds ratio = 2.01). Altogether, 245 genes in the interactome, including 29 novel interactors, have transcriptomic evidence with respect to exposure to asbestos or asbestos-like fibers. These novel interactors are: *ALB*, *B4GALT4*, *CAPN2*, *CDC40*, *DES*, *FMO1*, *FMR1*, *GML*, *GRIA1*, *HMG20B*, *HNRNPA1*, *ITSN2*, *LARP4*, *LPIN1*, *MGAT4A*, *NEK7*, *NFIB*, *NRG1*, *OCRL*, *PAX6*, *PDCD4*, *PITX3*, *PTRH2*, *REG3G*, *TAF1B*, *THOC1*, *TMED1*, *TNC* and *XPO4*.

From data in Pathology Atlas, we found that high expression of 73 genes, including that of 10 novel interactors, in the interactome has been positively correlated with unfavorable prognosis for lung cancer (*p*-value = 1.72 × 10−<sup>9</sup> , odds ratio = 2.05) [80]. These novel interactors are: *SPOCK1*, *SLC7A5*, *SCARB1*, *PLIN3*, *PLAUR*, *PIEZO1*, *KRT6A*, *GJB3*, *B3GNT3* and *ARL2BP*. We predicted *ARL2BP* to interact with *FLT1*, a VEGF receptor expressed in MPM cells. VEGF level in MPM patients is a biomarker for unfavorable prognosis, and lung cancer tumors expressing *FLT1* have been associated with poor prognosis [81,82].

Exosomes are extracellular vesicles secreted into the tumor microenvironment. They facilitate immunoregulation and metastasis by shuttling cellular cargo and directing intercellular communication. In a proteomic profiling study, 2176 proteins were identified in exosomes of at least one of the four human malignant mesothelioma cell lines (JO38, JU77, OLD1612 and LO68) [83]. 324 proteins in the MPM interactome appeared among these exosome-derived proteins (*p*-value = 8.86 × 10−10, odds ratio = 1.36), out of which 47 were novel interactors. Six hundred and thirty one (631) exosome-derived proteins were identified in all four malignant mesothelioma cell lines. Out of these, 127 occurred in the MPM interactome (*p*-value = 4.54 × 10−12, odds ratio = 1.84), out of which 15 were novel interactors (*PRKAG1*, *HNRNPA1*, *HNRNPH1*, *SORD*, *RNH1*, *RAN*, *PYGL*, *SLC7A5*, *RPS20*, *PARP4*, *YBX1*, *DCTN1*, *TUFM*, *EXOC4* and *GNPDA1*). In the following novel PPIs, both proteins

involved in the interaction appeared among exosome-derived proteins (MPM gene in the interaction is shown in bold): *TUBB3*-*SLC7A5*, *HSP90AB1*-*PROS1*, *HSP90AB1*-*GNPDA1*, *TUBB4A*-*PLIN3*, *LYN*-*ARFGEF1*, *HSP90AA1*-*PHLDA2*, *HSP90AA1*-*TCIRG1*, *TUBG1*-*PHB*, *GART*-*NMI*, *SRC*-*CUL4B* and *ATIC*-*CPS1*.

We computed the overlap of the interactome with 142 proteins that showed significant differences in abundance levels between epithelioid and sarcomatoid types of diffuse malignant mesothelioma [84]. In that study, a Fourier transform infrared (FTIR) imaging approach was employed to identify pathologic regions from diffuse malignant mesothelioma tissue samples [84]. These pathologic regions were then harvested using laser capture microdissection for proteomic analysis. 32 proteins in the interactome were more abundant in either epithelioid or sarcomatoid subtypes (*p*-value = 5.16 × 10−<sup>5</sup> , odds ratio = 2.06), including six novel interactors (*p*-value = 0.038, odds ratio = 2.43). The novel interactors *KRT78*, *NDUFV2*, *PRMT1*, *RAN* and *RNH1*—predicted to interact with the MPM genes *KRT72*, *TYMS*, *PDPN*, *POLE* and *RRM1*, respectively—had higher abundance in epithelioid samples, whereas *IGHA2*—predicted to interact with *HSP90AA1*—had higher abundance in sarcomatoid samples. The predicted interactions of these protein biomarkers with MPMassociated genes provide a mechanistic basis for experimental dissection of their ability to act as factors differentiating epithelioid tumors from sarcomatoid tumors (and vice versa).

#### **3. Discussion**

Currently, mesothelioma biologists only study a handful of genes, such as *BAP1*, *CDKN2A* and *NF2*. To shed light onto the other MPM-associated genes, whose functions remain poorly characterized, we assembled the 'MPM interactome' with ~2400 previously known PPIs and 364 computationally predicted PPIs (five of which have been validated in this work), which along with their biological annotations are being made available to researchers. We demonstrate the power of interactome-scale analyses to generate biologically insightful and clinically translatable results. The interactome has highly significant overlaps with MPM-associated genetic variants, genes differentially expressed or methylated in MPM or upon asbestos exposure, genes whose expression has been correlated with lung cancer prognosis, and with exosome-derived proteins in malignant mesothelioma cell lines. The interactome was enriched in cancer-related pathways. We extended the MPM interactome to include the drugs that target any of its proteins and analyzed it to identify a shortlist of 5 drugs that can potentially be repurposed for MPM—an example of a clinically translatable result.

We validated in vitro five novel PPIs in the interactome, namely, *BAP1*-*PARP3*, *ALB*-*KDR*, *ALB*-*PDGFRA*, *CUTA*-*HMGB1* and *CUTA*-*CLPS*. Literature evidence shows that these PPIs may be viable candidates for further experimentation in MPM cell lines or animal models. We hypothesize that the *BAP1*-*PARP3* interaction may enhance cancer growth in MPM. *BAP1* is a tumor suppressor protein playing a role in cell cycle progression, repair of DNA breaks, chromatin remodeling, and gene expression regulation; variants in *BAP1* have been implicated in hereditary and sporadic mesothelioma [85]. *PARP3* is involved in DNA repair, regulation of apoptosis, and maintenance of genomic stability and telomere integrity [86]. Interaction of *BAP1* with *BRCA1* has been shown to inhibit breast cancer growth [7]. In the absence of *BRCA1* activity or with a perturbation in its interaction with *BAP1*, cancerous growth is enhanced [87]. Loss of *BRCA1* protein expression has been noted in MPM [12]. In this scenario, it is possible that the novel interaction of *BAP1* with *PARP3* in cancerous cells may be promoting cancerous growth, possibly through regulation of DNA repair and apoptosis. *BAP1* and *PARP3* were found to be moderately overexpressed in sarcomatoid MPM tumors compared with normal pleural tissue (log2FC = 0.575, *p*-value = 0.028, and log2FC = 0.695, *p*-value = 0.0212, respectively) (GSE42977 [76]). Perturbation of the interaction of *BAP1* with *PARP3*, using *PARP3* inhibitors, may then suppress cancerous growth, at least in sarcomatoid MPM. Several studies and clinical trials [87], have shown that PARP inhibitors influence cancers in which mutations in *BRCA1* or *BRCA2* are observed, which led us to assume that the

cancerous growth-inhibiting interaction of *BAP1* with *BRCA1* may already be perturbed in this case, and that PARP inhibitors may actually be blocking the novel interaction of *BAP1* with *PARP3* which enhances cancer growth. It has been pointed out that upon inhibiting PARP activity, cancerous cells that lack *BRCA1* or *BRCA2* activity may undergo cell cycle arrest and apoptosis, possibly due to an accumulation of chromatid aberrations and an inability to perform DNA repair in the absence of BRCA [7,87]. Thus, we suspect that the novel interaction of *BAP1* and *PARP3* may also be perturbed by PARP inhibitors, leading to inhibition of cancer growth.

Low levels of *ALB* have been correlated with poor prognosis in MPM patients [88]. The two MPM genes, *KDR* and *PDGFRA*, that *ALB* is predicted to interact with, are members of the PI3K/AKT pathway which has been shown to be aberrantly active in mesothelioma [89]. High expression of *CUTA* has been correlated with favorable prognosis in lung cancer (Pathology Atlas). It was found to be overexpressed in MPM tumors versus normal pleura (log2FC = 0.871, *p*-value = 0.0039) (GSE2549 [34]). *CLPS* inhibits metastasis of the melanoma cell line, B16F10, to lungs by blocking the signaling pathway involving β1 integrin, *FAK* and paxillin [90]. *CLPS* has a novel interaction with *NEDD9*, which has been shown to mediate β1 integrin signaling and promote metastasis of non-small lung cancer cells [91]. *CD26*, a cancer stem cell marker of malignant mesothelioma, has been shown to associate with the integrin α5β1 (or *ITGA5*, a novel interactor of the MPM gene, *FGFR2*) and promote cell migration and invasion in mesothelioma cells [91]. Another cancer stem cell marker of malignant mesothelioma, *CD9*, inhibits this metastatic effect mediated by *CD26*. Depletion of *CD26* and *CD9* was shown to respectively lead to decreased and increased expression of *NEDD9* and *FAK* in mesothelioma cells lines, hinting at the involvement of *NEDD9* in mesothelioma tumor invasiveness [91]. *NEDD9* has a known interaction with *LYN*, an MPM gene, shown to play a negative role in the regulation of integrin signaling in neutrophils [92]. *CUTA* has a novel interaction with *HMGB1*, which has been shown to activate the integrin αMβ2 (or *ITGAM*, a novel interactor of the MPM gene, *TYMS*) and the cell adhesion and migratory function of neutrophils mediated by αMβ2 [93]. *HMGB1* also has a novel interaction with the MPM gene, *FLT1*, shown to be involved in the migration of multiple myeloma cells by associating with β1 integrin, and mediating PKC activation [94].

A recent bioinformatics study identified the genes differentially expressed in epithelioid MPM tissues versus normal pleural tissues (GSE42977 [76]), and extracted the known PPIs interconnecting these genes from the STRING database [95]. They identified 10 hub genes from this network and shortlisted 31 drugs targeting the proteins in the network based on scores from the Drug-Gene Interaction Database (DGIdb). The DGIdb score takes into account the literature evidence for a particular drug-protein interaction, the number of proteins in the network that interact with the given drug, and the ratio of the average number of known protein interactors for all drugs compared to the number of known protein interactors for the given drug. *CDK1*, which is one of the hub genes identified in their study, is a known interactor of three MPM-associated genes, namely, *LYN*, *SP1* and *RRM2*, and we showed that it has association to MPM in three omics datasets: high expression correlated with unfavorable lung cancer prognosis, differential expression in MPM tumors versus adjacent pleural tissue, and isolation as an exosome-derived protein in malignant mesothelioma cell lines. Our work overall presents a more comprehensive study in terms of a larger number of MPM genes analyzed, which were compiled from multiple sources by IPA, and analysis of a larger number of MPM associated omics data sets, and presents transcriptomic-driven shortlisting of repurposable drugs for which additional evidence is presented from clinical trial data, literature, and differential expression of target genes in MPM datasets.

Our study provides an integrative and mechanistic framework for functional translation of mesothelioma-related multi-omics data. The novelty of our work stems from two key factors: (a) we present computationally predicted PPIs of high precision, which link MPM-related genes from disparate genetic-variant / transcriptomic/proteomic studies in hitherto unknown ways within the functional landscape of the interactome, and (b) the

richly annotated MPM interactome is made available on a webserver to facilitate analysis by biologists and computational systems biologists. Our approach has some limitations. The drug-associated expression profiles analyzed in this study were induced in a diverse set of cell lines rather than in mesothelioma cell lines. The effect of the proposed drugs should be examined in MPM cell lines or animal models. We reported the overlap of mouse genes differentially expressed upon asbestos exposure [78] with corresponding human orthologs in the interactome. Mouse models have been routinely used to study pathologic changes associated with asbestos exposure, including gene expression, and these findings have been extrapolated to human diseases such as mesothelioma [96–99]. Nevertheless, our results should be interpreted with caution. It is not possible to draw direct transcriptomic/proteomic/phenotypic equivalences between mice and humans, unless these levels are comprehensively characterized in both the species, and a clear equivalence of factors defining a condition such as asbestos exposure is demonstrated in both the species [100]. Next, it is beyond the scope of our expertise to validate the large number of computationally predicted PPIs in a tissue or cell line of interest. However, we demonstrated the validity of computational predictions on a small number of PPIs on purified proteins with appropriate controls. The computational model has also been validated through additional experiments previously; some of the novel PPIs predicted previously by our method have translated into results of biomedical significance [17–19].

#### **4. Methods**

#### *4.1. Data Collection*

A search using the keyword "malignant pleural mesothelioma" on IPA (Ingenuity Pathway Analysis) retrieved genes causally related to the disease. IPA retrieves genes from the Ingenuity Knowledge Base which has ~5 million experimental findings expert-curated from biomedical literature or incorporated from other databases [29].

#### *4.2. High-Precision Protein-Protein Interaction Prediction (HiPPIP) Model*

PPIs were predicted by computing features of protein pairs, namely, cellular localization, molecular function and biological process membership, genomic location of the gene, gene expression from microarray experiments, protein domains and tissue membership of proteins, as described in Thahir et al. [101], and developing a random forest model to classify the pairwise features as interacting or non-interacting. A random forest with 30 trees was trained using the feature offering maximum information gain out of four random features to split each node; minimum number of samples in each leaf node was set to be 10. The random forest outputs a continuous valued score in the range of [0,1]. The threshold to assign a final label was varied over the range of the score for positive class (i.e., 0 to 1) to find the precision and recall combinations that are observed.

#### *4.3. Evaluation of PPI Prediction Model*

Evaluations on a held-out test data showed a precision of 97.5% and a recall of 5% at a threshold of 0.75 on the output score. Next, we created ranked lists for each of the hub genes (i.e., genes that had >50 known PPIs), where we considered all pairs that received a score >0.5 to be novel interactions. The predicted interactions of each of the hub genes are arranged in descending order of the prediction score, and precision versus recall is computed by varying the threshold of predicted score from 1 to 0. Next, by scanning these ranked lists from top to bottom, the number of true positives versus false positives was computed.

#### *4.4. Novel PPIs in the MPM Interactome*

Each MPM gene, say Z, is paired with each of the other human genes (G1, G<sup>2</sup> . . . GN), and each pair is evaluated with the HiPPIP model. The predicted interactions of each of the MPM genes (namely, the pairs whose score is >0.5) were extracted. These PPIs, combined

with the previously known PPIs of MPM genes collectively form the 'MPM interactome'. Interactome figures were created using Cytoscape [102].

Note that 0.5 is the threshold chosen not because it is the midpoint between the two classes, but because the evaluations with hub proteins showed that the pairs that received a score >0.5 are highly confident to be interacting pairs. This was further validated through experiments for a few novel PPIs above this score.

#### *4.5. Previously Known PPIs in the MPM Interactome*

Previously known PPIs of the 62 MPM genes were collected from Human Protein Reference Database (HPRD) version 9 [22] and Biological General Repository for Interaction Datasets (BioGRID) version 4.3.194 [23]. The data behind our web-server will be updated once in a year with recent versions of BioGRID, and if novel PPIs are shown validated by such updates to known PPIs, the information will be posted on the web-server.

#### *4.6. In Vitro Pull-Down Assays*

An initial screening to find physical interactions was performed using an in vitro pulldown assay for some of the predicted novel PPIs. This technique utilizes a His/biotin tagfused protein immobilized on an affinity column as the bait protein and a passing-through solution containing the 'prey' protein that binds to the 'bait' protein. The subsequent elution will pull down both the target (prey) and tagged-protein (bait) for further analysis by immunoblotting to confirm the predicted interactions. The pull-down assays were conducted using the Pull-Down PolyHis Protein:Protein Interaction Kit (Pierce™, Rockford, IL, USA) according to the manufacturer's instructions.

#### *4.7. Protein Identification Methods*

Peptide sequencing experiments were performed using an EASY-nLC 1000 coupled to a Q Exactive Orbitrap Mass Spectrometer (Thermo Scientific, San Jose, CA, USA) operating in positive ion mode. An EasySpray C18 column (2 µm particle size, 75 µm diameter by 15 cm length) was loaded with 500 ng of protein digest in 22 µL of solvent A (water, 0.1% formic acid) at a pressure of 800 bar. Separations were performed using a linear gradient ramping from 5% solvent B (75% acetonitrile, 25% water, 0.1% formic acid) to 30% solvent B over 120 min, flowing at 300 nL/min.

The mass spectrometer was operated in data-dependent acquisition mode. Precursor scans were acquired at 70,000 resolution over 300–1750 m/z mass range (3e6 AGC target, 20 ms maximum injection time). Tandem MS spectra were acquired using HCD of the top 10 most abundant precursor ions at 17,500 resolution (NCE 28, 1e5 AGC target, 60 ms maximum injection time, 2.0 *m/z* isolation window). Charge states 1, 6–8 and higher were excluded for fragmentation and dynamic exclusion was set to 20.0 s.

Mass spectra were searched for peptide identifications using Proteome Discoverer 2.1 (Thermo Scientific, Waltham, MA, USA) using the Sequest HT and MSAmanda algorithms, peptide spectral matches were validated using Percolator (target FDR 1%). Initial searches were performed against the complete UniProt database (downloaded 19 March 2018). Peptide matches were restricted to 10 ppm MS1 tolerance, 20 mmu MS2 tolerance, and 2 missed tryptic cleavages. Fixed modifications were limited to cysteine carbamidomethylation, and dynamic modifications were methionine oxidation and protein N-terminal acetylation. Peptide and protein grouping and results validation was performed using Scaffold 4.8.4 (Proteome Software, Portland, OR, USA) along with the X! Tandem algorithm against the previously described database. Proteins were filtered using a 99% FDR threshold.

#### *4.8. Ingenuity Pathway Analysis*

Pathway associations of genes in the MPM interactome were computed using Ingenuity Pathway Analysis (IPA). Statistical significance of the overlaps between genes in the MPM interactome and pathways in the Ingenuity Knowledge Base (IKB) was computed with Fisher's exact test based on hypergeometric distribution. In this method, *p*-value is

computed from the probability of k successes in n draws (without replacement) from a finite population of size N containing exactly k objects with an interesting feature, where N = total number of genes associated with pathways in IKB, K = number of genes associated with a particular pathway in IKB, n = number of genes in the MPM interactome and k = K ∩ n. This value was further adjusted for multiple hypothesis correction using the Benjamini-Hochberg procedure.

#### *4.9. Analysis of Differential Gene Expression in Pleural Mesothelioma Tumors and Lungs of Asbestos-Exposed Mice Versus Normal Tissue in Lungs*

The overlap of the MPM interactome with genes differentially expressed in pleural mesothelioma tumors compared with normal pleural tissue adjacent to mesothelioma was computed using the dataset GSE12345 [75]. Genes differentially expressed in the lungs of mice exposed to crocidolite and erionite fibers were obtained from the dataset GSE100900 [78]. Genes with fold change >2 or <sup>1</sup> <sup>2</sup> were considered as significantly overexpressed and underexpressed respectively at *p*-value < 0.05.

#### *4.10. Analysis of DNA Methylation in MPM Tumors*

The dataset GSE16559 [69] was used to analyze the methylation profile of pleural mesotheliomas. In this study, genes found to be differentially methylated in mesothelioma were identified from a set of 773 cancer-related genes associated with 1413 autosomal CpG loci. Methylation values (M-values) were computed as M = log2 (β (1−β)) for both control (non-tumor pleural tissue) and test (pleural mesothelioma) cases, where β is the ratio of methylated probe intensity and overall intensity. Difference between M-values of test and control cases was then computed, and genes with M-value > 1 and M-value < 1 were considered to be hypermethylated and hypomethylated respectively at *p*-value < 0.05.

#### *4.11. Correlating Expression of MPM Genes with Lung Cancer Prognosis*

Data for correlation of gene expression and fraction of patient population surviving after treatment for lung cancer was taken from the Pathology Atlas [80]. Genes with logrank *p*-value < 0.001 were considered to be prognostic. Unfavorable prognosis indicates positive correlation of high gene expression with reduced patient survival.

#### *4.12. Identification of Repurposable Drugs in the MPM Drug-Protein Interactome*

Negative correlation between lung cancer and drugs were studied using the BaseSpace correlation software, which uses a non-parametric rank-based approach to compute the extent of enrichment of a particular set of genes (or 'bioset') in another set of genes [31]. Readers may refer to Appendix A, titled 'Repurposable Drugs for Treatment of Malignant Pleural Mesothelioma (MPM)' for more details on the methodology used to identify repurposable drugs.

#### **5. Conclusions**

Biomedical discovery in the field of MPM research has to be accelerated to fuel clinically translatable results due to an urgent need to diagnose MPM preemptively, prevent its post-treatment recurrence, and curb its predicted increase in incidence in western and economically emerging nations [103]. In this study, we presented the MPM interactome as a valuable resource for mesothelioma biologists. We demonstrated its biological validity through comparison with MPM-related multi-omics data, which served to contextualize the novel PPIs within the mesothelioma landscape. Making novel MPM PPIs available freely on a webserver will catalyze investigations into these by cancer biologists and may lead to biologically or clinically translatable results. The MPM interactome with diseaseassociated proteins and their interacting partners will help biologists, bioinformaticians and clinicians to piece together an integrated view on how MPM-associated genes from various studies are functionally linked. Biological insights from this 'systems-level' view will help generate testable hypotheses and clinically translatable results. Future work

will focus on expanding this interactome by including interactions from additional PPI repositories, other mesothelioma types and mesothelioma datasets.

**Supplementary Materials:** The following are available online at https://www.mdpi.com/article/10 .3390/cancers13071660/s1: Table S1: Identification of protein interactors using liquid chromatography– mass spectrometry (LC-MS), Table S2: Overlaps between drugs tested in NSCLC and drugs occurring in the MPM drug-protein interactome, that were negatively correlated with lung cancer expression studies, Table S3: Some novel interactors which are hypomethylated and their MPM genes, Data File S1: List of MPM genes and their corresponding biological evidences extracted from IPA suite, Data File S2: List of genes from the MPM interactome with their labels (MPM genes, known interactors and novel interactors), Data File S3: List of all the pathways associated with at least one of the MPM genes, Data File S4: List of all the drugs that target any of the genes from the MPM interactome, and Data File S5: Master table of all biological evidences (genetic variant, transcriptomic and proteomic evidence) for each of the MPM interactome genes discussed in the paper.

**Author Contributions:** In sequence of work: M.K.G. conceptualized and supervised the study and carried out interactome construction and analysis of pathway and drug associations. K.B.K. carried out studies of the overlap of the interactome with various high-throughput data, literature-based evidence gathering, and identification of repurposable drugs. Experimental validations were carried out by N.Y. and G.B. Written description of methods of experimental validation were provided by N.Y. and G.B. Manuscript has been written by K.B.K. and edited by M.K.G., M.J.B. provided consultation and valuable feedback on the manuscript. Manuscript has been read and approved by all authors. All authors have read and agreed to the published version of the manuscript.

**Funding:** This work has been funded by U24OH009077 (Becich) from the Center for Disease Control (CDC), National Institute of Occupational Safety and Health (NIOSH) and R01MH094564 (Ganapathiraju) from National Institute of Mental Health (NIMH), of National Institutes of Health (NIH), USA. The content is solely the responsibility of the authors and does not necessarily represent the official views of the CDC, NIOSH or NIMH, NIH, USA.

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** On journal website and at http://severus.dbmi.pitt.edu/wiki-MPM.

**Acknowledgments:** We thank David Boone (Department of Biomedical Informatics), J. Richard Chaillet (Office of Research Health Sciences) and Adrian Lee (Department of Pharmacology and Chemical Biology) of University of Pittsburgh for detailed and valuable feedback on the manuscript. We thank the team of National Mesothelioma Virtual Bank, particularly Waqas Amin and Jonathan Silverstein (University of Pittsburgh), Harvey Pass (New York University Langone Medical Center) and Carmelo Gaudioso (Roswell Park Comprehensive Cancer Center) for valuable discussions. M.K.G. and K.B.K. thank N. Balakrishnan (Indian Institute of Science) for valuable technical feedback. M.K.G thanks Sai Supreetha Varanasi for system administration assistance in hosting the website.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **Appendix A Repurposable Drugs for Treatment of Malignant Pleural Mesothelioma (MPM)**

We present here five drugs (*cabazitaxel*, *pyrimethamine*, *trimethoprim*, *primaquine* and *glicazide*) that could potentially be repurposed for the treatment of malignant pleural mesothelioma (MPM). These drugs were shortlisted through three types of analysis: (A) considering those that were already tested in non-small cell lung cancer (NSCLC), (B) gene expression analysis of drugs that target MPM genes or novel interactors in the MPM interactome, or (C) gene expression analysis of drugs that target known interactors in the malignant pleural mesothelioma (MPM) interactome. Drugs were selected based on whether they were already tested against lung cancer in clinical trials and/or showed overall negative correlation with lung cancer expression studies, because both mesothelioma and lung cancers have been shown to share common pathways that are initiated on exposure to asbestos fibres in mesothelial cells and lung epithelial cells respectively [35]. Another criterion used was whether the genes targeted by the drugs showed high differential expression in MPM tumours/cell lines. The details of these methods and observations are presented below.

#### *Appendix A.1 Repurposable Drugs Already Tested in Non-Small Cell Lung Cancer*

Nine overlapping drugs were found between drugs tested in NSCLC and drugs occurring in the MPM drug-protein interactome, that were negatively correlated with lung cancer expression studies, namely, cabazitaxel, dasatinib, docetaxel, gemcitabine, ipilimumab, ixabepilone, minocycline, pazopanib and pemetrexed. Minocycline was eliminated due to its toxicity. All of the remaining eight drugs were found to be effective in treatment of NSCLC (Table S2). Out of these eight drugs, cabazitaxel was the only drug that was not tested for treatment of mesothelioma. The fact that the other seven drugs were already tested against mesothelioma in clinical trials demonstrates the validity of our approach. It was interesting to note that drugs that targeted known interactors in addition to some MPM genes were found to have either no effect or limited clinical activity in mesothelioma, for e.g., dasatinib, docetaxel and pazopanib. On the other hand, drugs that targeted only MPM genes were found to be effective in treatment of mesothelioma or were capable of preventing disease progression, for e.g., gemcitabine, ipilimumab, ixabepilone and pemetrexed. This raises the suspicion that drugs that do not act on "off-target" genes (known interactors, in this case) may be more effective. In this respect, cabazitaxel, which targets the MPM genes *TUBB1* and *TUBA4A*, may be a suitable candidate for mesothelioma. Cabazitaxel was found to be effective in treatment of NSCLC resistant to docetaxel, a drug that targets *TUBB1* and other known interactors [37].

#### *Appendix A.2 Repurposable Drugs Targeting MPM Genes and Novel Interactors*

The MPM genes that were most differentially expressed with high significance in MPM tumors (GSE51024 [33]) were *TYMS* (log2FC = 1.82, *p*-value = 4.10 × 10−17) and *DHFR* (log2FC = 0.89, *p*-value = 1.20 × 10−14), and the drugs that target these genes (also having negative correlation with lung cancer expression) were pyrimethamine and trimethoprim. The first line drug currently used to treat mesothelioma is premetrexed, which targets proteins in the folate metabolic pathway, namely, *DHFR*, *TYMS* and *GART* [104]. Since MPM tumors have been shown to be responsive to anti-folates [68], both pyrimethamine (which targets only *DHFR*) and trimethoprim (which targets both *DHFR* and *TYMS*), seem to be interesting candidates. Pyrimethamine, an anti-parasitic drug commonly used to treat toxoplasmosis and cystoisosporiasis, has shown anti-tumor activity in metastatic melanoma cells and in murine models of breast cancer [105,106]. Trimethoprim, an antibacterial drug commonly used in the treatment of urinary bladder and respiratory tract infections, is also used to treat bacterial infections in cancer patients [107,108].

Keratin proteins form important components of the cell cytoskeleton, called intermediate filaments, in epithelial cells, and are commonly used as diagnostic markers in cancer [60]. In addition to their role as cancer markers, their involvement in cellular functions such as cell motility, proliferation, cell polarity, protein synthesis, membrane trafficking and most importantly, tumour invasion and metastasis make them attractive as candidates for drug development [60]. *KRT7* is a keratin primarily expressed in mesothelial cells, apart from cells lining ducts and the intestine [60]. In a patient with malignant mesothelioma of the epithelioid type (which spreads to mediastinum and breast), *KRT7* was found to be significantly overexpressed when she developed resistance to pemetrexed/carboplatin, provided as a second line therapy [61]. The cancer cells showed a drastic increase in their immunoreactivity to CK7, the protein encoded by *KRT7* [61]. At the last stage of cancer progression (which was followed by her death), the patient showed dyspnoea (difficulty in breathing), increased tumour volume and pleural fluid [61]. In another case, *KRT7* was found be significantly overexpressed in an aggressive state of MPM, prior to treatment [61]. Two-thirds of malignant mesothelioma cases have been reported to be K7+/K20<sup>−</sup> (positive for expression of *KRT7* and negative for expression of *KRT20*) [60]. Expression of

*KRT7* in three histological types of mesothelioma, namely, epithelioid, sarcomatid and biphasic, has been used to distinguish them from synovial sarcoma that metastasizes to the lungs and pleura [62]. *KRT7* has been identified as marker of circulating tumour cells in lung cancer [109]. *KRT7* was also found to be significantly upregulated in MPM tumours (log2FC = 3.80, *p*-value = 0.0002), and in cell line models of MPM (log2FC = 2.266, *p*-value = 0.029) (GSE2549 [34]). Positive expression of *KRT7* was noted in various types of non-small cell lung cancers, including large cell neuroendocrine carcinoma and lung adenocarcinoma [110,111]. In the MPM interactome, *KRT7* was predicted to interact with *KRT5*, an MPM gene that serves as a marker for malignant mesothelioma, along with vimentin, and is specifically used to distinguish pleural mesothelioma of the epithelioid type from pulmonary adenocarcinoma and non-pulmonary adenocarcinoma metastasizing to pleura [60,112]. *KRT7* is a target of primaquine, an-antimalarial agent known to destroy the malarial parasites, *Plasmodium vivax* and *Plasmodium ovale*, inside the liver [113,114]. The exact mechanism of action has not been elucidated for this drug. However, in independent studies, primaquine has been shown to bind to keratin in a concentration-dependent manner, and also mediate strong membrane perturbations in cell membrane models [113,115]. Since high expression of *KRT7* has been shown to be related to tumour aggressiveness and drug resistance in malignant mesothelioma, and its high expression was also noted in MPM tumours and cell lines, primaquine may be re-purposed for treatment of MPM at least as an adjunctive drug with pemetrexed, the drug currently used for first line therapy. It is interesting to note that primaquine enhanced the sensitivity of KBV20C cells to cancer drugs, namely, vinblastine, vinorelbine, paclitaxel, docetaxel, vincristine and halaven [63]. KVB20C is a multi-drug resistant cell line derived from oral squamous carcinoma. Primaquine compounds (substituted quinolines) have also been shown to exert anti-tumor activity in breast cancer cells [116].

#### *Appendix A.3 Repurposable Drugs Targeting Known Interactors*

Out of the four drugs targeting known interactors in the MPM interactome and showing negative correlation with lung cancer associated gene expression, three drugs were already known to exhibit anti-tumour activity in pre-clinical models of mesothelioma, namely, zoledronate, sirolimus and the vitamin E analog, alpha-tocopheryl succinate, which shows the validity of our approach. Zoledronate, which showed modest activity in MPM, induced apoptosis and S-phase arrest in human mesothelioma cells and inhibited tumor growth in the pleural cavity of an orthotopic animal model [54,55]. Sirolimus/cisplatin increased cell death and decreased cell proliferation in cell lines of MPM [56]. Alpha-tocopheryl succinate increased survival of orthotopic animal models of malignant peritoneal mesothelioma [57]. Zoledronate has demonstrated modest clinical activity in patients with advanced MPM [54]. Sirolimus has not been tested against MPM in clinical trials, but everolimus, a drug derived from sirolimus sharing similar properties with it, has shown only limited clinical activity in MPM, and further testing as a single-agent was not advised based on the results from this study [117]. Both vitamin E and its analog, alpha-tocopheryl succinate have not been tested against MPM in clinical trials. However, testing of vitamin E and its analogs are being carried out in various pre-clinical settings [58,59]. Hence, it was the drug gliclazide that emerged as a potentially repurposable drug, untested against MPM.

Gliclazide, an anti-diabetic drug, inhibits *VEGFA*, which has been shown to be significantly upregulated (Log2FC = 1.83, *p*-value = 0.0018) in MPM tumour (GSE2549 [34]). This drug inhibits VEGF expression induced by advanced glycation end products in bovine reticular endothelial cells, and VEGF expression induced by ischemia in retinal tissue of mice [64,118]. In the latter case, glicazide also inhibits neovascularization, a process known to be mediated by VEGF. VEGF has been identified as a prognostic marker for MPM. High levels of VEGF have been correlated with both asbestos exposure in MPM, and an advanced stage of the disease [65,66]. It is interesting to note that glibenclamide, a drug whose mechanism of action is similar to that of glicazide, has been shown to increase caspase activity in MPM cell lines and primary cultures, leading to apoptosis mediated by

TNF-related apoptosis inducing ligand (*TRAIL*) [67]. Hence, glicazide may be repurposed to inhibit neovascularization and perhaps enhance apoptosis in MPM.

#### **References**


## *Review* **Biomarkers for Malignant Pleural Mesothelioma—A Novel View on Inflammation**

**Melanie Vogl , Anna Rosenmayr, Tomas Bohanes, Axel Scheed, Milos Brndiar, Elisabeth Stubenberger and Bahil Ghanim \***

> Department of General and Thoracic Surgery, Karl Landsteiner University of Health Sciences, University Hospital Krems, 3500 Krems an der Donau, Austria; melanie.vogl@krems.lknoe.at (M.V.); arosenmayr@gmail.com (A.R.); bohanest@gmail.com (T.B.); axel.scheed@krems.lknoe.at (A.S.); milos.brndiar@krems.lknoe.at (M.B.); elisabeth.stubenberger@krems.lknoe.at (E.S.)

**\*** Correspondence: bahil.ghanim@kl.ac.at; Tel.: +43-2732-9004-4294

**Simple Summary:** In view of the recent advances in immunoncology, we want to reevaluate and summarize the role of the immune system in malignant pleural mesothelioma (MPM). MPM is an aggressive disease with limited treatment options and devastating prognosis. Exposure to asbestos and chronic inflammation have long been acknowledged as main risk factors. In this review, we summarize the current knowledge about local and systemic inflammation promoting pathogenesis and progression of MPM. We focus on the prognostic and predictive value of infiltrating immune cells within the tumor and its microenvironment as local inflammation on the one hand and systemic inflammatory parameters on the other. We found that suppression of the specific and activation of the unspecific immune system are essential drivers of MPM, resulting in poor patient outcome. Numerous local and systemic inflammatory parameters are promising potential biomarkers for MPM, worth further research.

**Abstract:** Malignant pleural mesothelioma (MPM) is an aggressive disease with limited treatment response and devastating prognosis. Exposure to asbestos and chronic inflammation are acknowledged as main risk factors. Since immune therapy evolved as a promising novel treatment modality, we want to reevaluate and summarize the role of the inflammatory system in MPM. This review focuses on local tumor associated inflammation on the one hand and systemic inflammatory markers, and their impact on MPM outcome, on the other hand. Identification of new biomarkers helps to select optimal patient tailored therapy, avoid ineffective treatment with its related side effects and consequently improves patient's outcome in this rare disease. Additionally, a better understanding of the tumor promoting and tumor suppressing inflammatory processes, influencing MPM pathogenesis and progression, might also reveal possible new targets for MPM treatment. After reviewing the currently available literature and according to our own research, it is concluded that the suppression of the specific immune system and the activation of its innate counterpart are crucial drivers of MPM aggressiveness translating to poor patient outcome.

**Keywords:** malignant pleural mesothelioma; inflammation; infiltrating immune cells; prognostic biomarker; predictive biomarker; immune therapy

#### **1. Introduction**

Malignant pleural mesothelioma (MPM) is an aggressive neoplasm of mesothelial origin. Patients face a devastating prognosis of 12 months median survival only after diagnosis [1]. Despite recent—and, in part, promising—developments regarding both systemic therapy and cytoreductive surgery, MPM remains a clinical challenge, especially when it comes to treatment allocation [2,3]. Furthermore, the optimal (multimodal) treatment regimens still remain to be defined from the available arsenal of immune therapy, surgery, radiation and systemic treatment [4].

**Citation:** Vogl, M.; Rosenmayr, A.; Bohanes, T.; Scheed, A.; Brndiar, M.; Stubenberger, E.; Ghanim, B. Biomarkers for Malignant Pleural Mesothelioma—A Novel View on Inflammation. *Cancers* **2021**, *13*, 658. https://doi.org/10.3390/ cancers13040658

Academic Editors: Daniel L. Pouliquen and Joanna Kopecka Received: 16 December 2020 Accepted: 2 February 2021 Published: 6 February 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

The pathogenesis of MPM was already associated with chronic inflammation, induced by asbestos exposure, sixty years ago by Wagner et al. [5]. Asbestos remains the main risk factor for developing this rare disease with a latency period of up to 40 years from time of exposure to diagnosis [1,6]. When inhaled, the long and thin asbestos fibers penetrate the lung parenchyma and deposit in the pleura, causing irritation and chronic inflammation. Consequently, the activation of surrounding immune cells leads to the secretion of cytokines, formation of reactive oxygen and nitrogen species, tumor necrosis factor α (TNF-α) release and nuclear factor 'kappa-light-chain-enhancer' of activated Bcells (NFκ-B) expression, in the end resulting in the accumulation of DNA damage and thus malignant evolution as reviewed before [6,7].

The activated immune system—especially with regard to its innate blood derived components—proved to be associated with worse patient outcome, late stage of disease, high Ki67 expression and poor treatment response in MPM as shown before by the authors and other research groups [8–13]. Not only for MPM but generally in oncology, the tumor promoting role of the immune system has been increasingly recognized as reflected in the latest version of the hallmarks of cancer by Hanahan and Weinberg [14]. Most recently, the immune system also evolved as a promising treatment target and modern immune therapy revealed as effective treatment modality in many solid tumors including MPM [15–19].

In light of the past and recent insights regarding the role of inflammation in the development and progression of MPM, inflammatory parameters are currently considered promising biomarkers [20]. In this review, we provide an overview about up to date knowledge of local inflammation in MPM and its involved immune cells as well as the tumor induced systemic inflammatory response. Special focus lies on the use of local and systemic inflammatory parameters as biomarkers for prognostic and predictive purposes in hope to facilitate and optimize treatment decisions and highlight new therapeutic targets for the future management of MPM. Predictive biomarkers might help to answer these crucial questions and are therefore desperately needed [21]. Despite the urgent need, to date there are no biomarkers recommended for MPM in daily practice in the current European guidelines since most studies failed to show sufficient reproducibility, sensitivity and specificity to justify the use of any suggested diagnostic biomarker so far. Unfortunately, the same holds true when it comes to prognostic, predictive or follow up biomarkers and thus further research is requested to better personalize treatment for MPM patients [22].

For this review we performed literature research in PubMed including English literature only. The following search terms were used: mesothelioma combined with prognostic and predictive biomarker, inflammation, inflammatory markers, C-reactive protein, fibrinogen, neutrophil to lymphocyte ratio, monocyte to lymphocyte ratio, thrombocyte to lymphocyte ratio, neutrophils, leukocytes, monocytes, albumin, Glasgow prognostic score, IL-6, ferritin, tumor microenvironment, tumor infiltrating lymphocytes, tumor associated macrophages/monocytes, PD-L1 and PD1, CTLA-4, immune therapy, and complement system. Since mesothelioma is a very rare disease and research regarding inflammatory biomarkers is limited, we included all available studies regarding biomarkers and only excluded case reports

Literature from the very early days of mesothelioma research ranging back to Wagner et al. from 1960 were included [5] as well as the most recent MPM literature from the beginning of 2021, resulting in 194 included references.

#### **2. Findings**

#### *2.1. The Role of Local Inflammation in MPM*

Several studies proved that (pre)malignant cells of various origins induce an inflammatory response with a paradox tumor promoting effect [23]. Local inflammation and immune cell infiltration within the tumor nests as well as the surrounding tumor microenvironment (TME) strongly influence the development and progression of numerous malignant diseases [23,24] including MPM as reviewed by Hendry et al. [25].

On the other hand, the immune system and its cellular components also play a protective role, especially with regard to acquired immunity as Leigh et al. observed already in 1986 correlating high lymphoid infiltration in mesothelioma specimens to a better prognosis [26]. In the past, the role of different infiltrating immune cells within MPM and the stroma has, therefore, become of increasing research interest since the immune system seems to be characterized here by a—not yet fully understood—duality [12]. Our adaptive immune system is protective against cancer development and spread [27], but it is also well documented that the immune system plays a crucial tumor promoting role in eventually all steps of malignant evolution by contributing to carcinogenesis, proliferation, angiogenesis, local infiltration and finally metastatic progression as reviewed by Coussens and Werb [28].

Very heterogenic immune cell infiltration in MPM tumor specimens and its TME has been described [29–33], with most studies reporting a predominant infiltration of tumorassociated macrophages (TAM) and tumor infiltrating lymphocytes (TIL), in particular CD4+ and CD8+ T-lymphocytes as reviewed by Chu et al. [34]. These cells are assumed to be the key players in the tumor associated immunoreaction. However, also rarer detectable myeloid derived suppressor cells (MDSC) [35,36], natural killer (NK) cells [28,31,32] and regulatory T cells (Treg) [29,36,37] have been studied before. These different immune cells infiltrating the tumor tissue but also contributing to the TME will be summarized in the following paragraphs as well as in Table 1 with regard to their role on MPM outcome and treatment response.


**Table 1.** Potential local inflammatory biomarkers.

TIL tumor infiltrating lymphocyte, M2 macrophage subtype 2, Treg regulatory T cell, FGF fibroblast growth factor, TGF-β transforming growth factor β, COX-2 cyclooxygenase 2, TAM tumor associated macrophages, IL-34 interleukin 34, M-CSF macrophage colony stimulating factor, NK cells natural killer cells, PD-L1 Programmed cell death ligand 1, HR hazard ratio, N.R. not reported, N.S. not significant, Prog prognostic biomarker, Pred predictive biomarker, R retrospective, P prospective.\* measured in pleural effusion.

#### *2.2. Tumor Infiltrating Lymphocytes (TIL)*

TIL comprise T- and B-lymphocytes that have left the blood stream and infiltrated the tumor itself as well as the tumor stroma. Invading CD4+ T cells and proinflammatory cytokines prime CD8+ T cells to become effector cytotoxic T-lymphocytes (CTL), which then play a key role in eliminating cancer cells as reviewed before [48]. During tumor progression, cancer cells can avoid this effect by overexpression of programmed death ligand 1 (PD-L1) and cytotoxic T-lymphocyte antigen 4 (CTLA-4) (compare corresponding subchapter). Simultaneously, TIL release cytokines, thereby influencing various other immune cells, including the differentiation of TAM towards the immune suppressive type 2 macrophages (M2). This mechanism represents a negative feedback loop to avoid an over activated immune response. However, both aforementioned—in principal protective immunosuppressive mechanisms (PD-L1/CLA-4 on the intercellular signaling level and type 2 macrophage differentiation on the cellular level) might lead to tumor immune evasion and thus uncontrolled tumor growth and progression [28,49,50].

For MPM, a predominant infiltration of CD8+ and CD4+ T-lymphocytes has been described by various researchers [30–32,51], but also the role of B lymphocytes [29] and Treg [51] is under investigation as described below.

The influence of B lymphocytes, key players in adaptive humoral immunity, is not fully understood and controversial results have been published so far for mesothelioma. Several studies reported low numbers of infiltrating B lymphocytes as reviewed by Minnema-Luiting [29]. Nevertheless, others discovered high CD 20+ B lymphocytes infiltration as well as the ratio CD163+ macrophages/CD20+ B lymphocytes as independent prognostic factors indicating better prognosis [37].

The prognostic value of CD8+ T lymphocytes, likewise part of the adaptive immunity, is better investigated and thus better understood for different MPM patient populations:

A number of studies investigated tumor samples of patients receiving trimodal therapy including cytoreductive surgery. Some reported an independent favorable prognostic value of high levels of CD8+ TIL [31,32], others found the ratio M2 count/CD8+ TIL count independently indicating negative prognosis [37], also suggesting that patients with low M2 and high CD8+ count have better outcomes. One other study found a correlation of local tumor overgrowth and low levels of CD8+ TIL in surgical patients [49] suggesting again an association with worse prognosis when the adapted immune system is underrepresented in the tumor compared to its innate counterpart.

On the contrary, Pasello et al. found high levels of CD8+ TIL in treatment naïve patients correlating not only with poor prognosis and aggressiveness of the tumor, but also a predictive value of high CD8+ TIL count for low response to chemotherapy. However, high levels of CD8 + TIL correlated additionally with high PD-L1 expression, which the authors speculate to be causal for the observed poor prognosis [38].

Additionally, high CD4+ T cell count in the tumor correlated with better outcome. Yamada et al. showed a tendency for better survival if CD4 + TIL and NK levels were high but did not reach level significance [32]. Marcq et al. compared treatment of naïve patients with those pretreated with chemotherapy and found high count of CD4+ TIL in the TME to be an independent positive prognostic marker for both therapy subgroups [30].

Treg, the immunosuppressive subset of CD4+ T cells, physiologically regulates immune tolerance, but also plays a major role in tumor development. Whereas only scarcely present in healthy tissue, a strong infiltration of Treg has been shown for many tumor entities [50] including MPM [51].

First data suggested that Treg and their deactivation via depletion of the surface marker CD25+ influenced survival in a murine model in a positive way [51]. Additionally, it was hypothesized that response to chemotherapy might be influenced by T effector cells and Treg [30].

While only a few studies analyzed the prognostic and predictive potential of Treg count in MPM, others investigated the cytokines responsible for Treg recruitment and activation, such as transforming growth factor (TGF-β) [6,52–54] and cyclooxygenase-2 (COX-2)/prostaglandin E2 (PGE2) [39–42,55,56], which are released by cancer cells directly or indirectly by cancer associated fibroblasts [50].

#### *2.3. Cancer-Associated Fibroblasts (CAF)*

CAF are abundantly present in numerous tumor entities and play a key role in the immunosuppressive effect of TME via cross-talk with Treg. High numbers of CAF are hence often associated with tumor promotion and poor prognosis. In turn, Treg stimulate

resident fibroblasts to differentiate into CAF, which emphasizes the tight cross-talk between Tregs and CAFs [50].

A number of studies confirmed high numbers of CAF in TME from MPM samples [29]. As mentioned above, the major cytokines mediating CAF and Treg function have been under thorough investigation. Latest studies suggested a correlation between fibroblast growth factor (FGF) overexpression and high numbers of CAF with tumor aggressiveness and worse prognosis; however, the prognostic or predictive value is currently unknown and further research is obligatory [57,58]. Schelch et al. analyzed the role of different FGFs and their receptors in MPM in vitro and in vivo and proved that the FGF axis promotes cell proliferation, epithelial to mesenchymal transition, migration, invasion and clinical tumor aggressiveness. Inhibition of FGF receptor not only showed anti-proliferative effects itself but also a synergism with radiation and cisplatin and might, therefore, serve as a novel therapeutic target in MPM [58–60]. Furthermore, Li et al. also proved that FGF-2 besides platelet derived growth factor (PDGF) and hepatocyte growth factor (HGF)—is expressed by MPM. In addition, this study showed that MPM cell lines stimulate fibroblast motility and growth on the one hand and fibroblasts vice versa stimulate MPM growth and motility by HGF on the other hand, indicating an important cross-talk and tumor promoting symbiosis of CAFs and MPM cells [57].

#### *2.4. Transforming Growth Factor-β (TGF-β)*

TGF-β is known as an important inducer of CAFs and thus supporter of the immunosuppressive TME. Besides this tumor promoting characteristic, TGF-β can directly induce proliferation and epithelial to mesenchymal transition. In addition, TGF-β expression was associated with resistance to immune therapy as summarized recently [61]. With regard to MPM, TGF-β and its subtype activin A have been shown to be overexpressed in MPM cells with tumorigenic effects and thus inhibition or silencing was suggested as possible therapeutic target—first clinical results, however, were unsatisfactory with regard to fresolimumab, a TGF-β targeting antibody [6,53,54]. In addition, activin A blood levels were increased in MPM patients when compared to healthy controls and high activin A levels correlated with advanced tumor stage, high tumor volume and histological subtype translating to poor patient survival [54].

#### *2.5. COX-2*

Overexpression of COX-2 is detectable in various tumor cells and mostly associated with worse prognosis [62,63]. Nuvoli et al. reviewed the tumor promoting effects of proinflammatory prostaglandins, synthesized by COX-2 in general and for MPM in particular [55]. COX-2 overexpression was also found in MPM specimens [56]. Although some authors described controversial results regarding the prognostic value of COX-2, the majority reports a negative prognostic value of high tumor COX-2 expression [39–42].

In addition, the therapeutic effect of COX-2-inhibitors such as celecoxib has already been studied in other cancer types extensively [63]. However, COX-2 is not a routine target in modern oncology due to controversial results, e.g., for colorectal [64–66] and lung cancer [67,68]. For MPM on the contrary COX-2-inhibitors achieved promising results in vitro [35,57,69,70] and in vivo [36,71]. Unfortunately, neither NSAIDs nor COX-2 inhibitors prevented MPM development in an asbestos exposed risk group and in murine models [69]. The currently ongoing phase III randomized INFINITE trial assesses the effect of systemic celecoxib and chemotherapy combined with intrapleural INF-α (NCT03710876) and might answer the question whether COX-2 is an eligible treatment target in MPM.

#### *2.6. M2 Macrophages*

Under normal conditions, macrophages of the subtype M1 are part of the early inflammatory response enhancing the immune reaction while the immunosuppressive M2 macrophages limit a possible inflammatory overreaction [28,49,50]. A large proportion

of M2 of total TAM consequently enforces tumor-promoting and immunosuppressive conditions and has been shown to indicate poor survival for different malignancies [70].

Additionally, for MPM specimens, various studies described strong infiltration of macrophages, predominantly of immunosuppressive M2, as prognostic marker [46,50,72]. High count of infiltrating M2 not only correlated significantly with poor prognosis [73] and increased proliferation rate but also reduced response to chemotherapy [74]. Others found no correlation between prognosis and absolute count of TAM or M2 but reported that high percentage of M2 of total TAM correlates significantly with local overgrowth [49] and negative prognosis [44].

In conclusion, current scarce data indicate that tumor infiltrating M2 might have prognostic and predictive potential. Interestingly, there is abundant research on M2 promoting cytokines and their impact on prognosis and treatment response.

Hematopoietic cytokines, including granulocyte macrophage colony stimulating factor (GM-CSF), were shown to be released by MPM cells especially when exposed to inflammatory cytokines but also autonomously [72,75] promoting the release of monocytes to the peripheral blood.

Additionally, cytokines promoting the M2 differentiation, namely IL-34 [45], macrophage colony stimulating factor (M-CSF) [74] and C-C motif chemokine ligand 2 (CCL2) [76,77] have been found to be elevated in tumor specimens or pleural effusion of MPM patients. High pleural levels of IL-34 correlated with worse prognosis [45], as well as high serum M-CSF, the latter also with response to chemotherapy [78]. Furthermore CCL2, a proinflammatory chemokine for monocyte recruitment, has been investigated over the past years. MPM patients showed significant higher serum levels of CCL2 than an asbestos exposed cohort without evidence of disease [79]. Similar results have been published by Gueugnon et al. as well as Blanquart et al. who found significant higher concentrations of CCL2 in pleural effusion of MPM patients compared to benign effusion or metastatic adenocarcinomas [76,77]. CCL2 released by MPM cells directly plays an important role in monocyte recruitment. CCL2 inhibition is also a potential treatment target and currently under investigation [50,78,80].

#### *2.7. Myeloid-Derived Suppressor Cells (MDSCs)*

The immune-suppressive MDSC are immature myeloid cells stimulated by tumorderived cytokines. Abundantly detectable in MPM TME [35,36] they activate tumorpromoting Treg and inhibit tumor-suppressing CD4+ and CD8+ T cells [36]. A negative prognostic potential of MDSC can, therefore, be assumed; however, to our knowledge, no data regarding the prognostic or predictive value is currently available for MPM.

#### *2.8. Natural Killer Cells and Dendritic Cells (DC)*

The majority of studies reported low proportion of DC and NK in MPM specimens [28,31,32]. Yamada et al. additionally investigated the prognostic potential of NK infiltration and found no correlation with outcome [32]. Hegmans et al. confirmed a weak infiltration of DC, although they found a strong infiltration by NK. As possible explanation for low DC numbers they suggest the high levels of Interleukin-6 (IL-6) produced by MPM cells, since IL-6 inhibits the differentiation of progenitor cells to DC [51].

Summarizing these findings, DC and NK—both part of the innate immune system are currently suspected to play a subordinate role in MPM and are, therefore, underrepresented in medical literature when compared to the aforementioned more prominent cellular players in the tumor and its TME.

#### *2.9. Programmed Death Ligand 1 (PD-L1) and Cytotoxic T-Lymphocyte Antigen 4 (CTLA-4)*

PD-L1 is expressed on the surface of various tumor cells and has the ability to bind to PD-1 receptors of CD4+ and CD8+ T Cells thus altering proliferation and cytokine production, leading to T cell inactivation and apoptosis of these important cellular players of adaptive immunity. As reviewed before by Zielinski et al. both PD-1/PD-L1 and CTLA-4 act as similar pathways downregulating lymphocyte response and accordingly adaptive

immunity [80]. This tumor immune evasion results in progression and poor prognosis of various solid tumors [81]. With regard to thoracic oncology, the PD-L1 axis and its prognostic role were already analyzed in malignant pleural effusion [82] and stage IV lung cancer [83,84]. Furthermore, PD-L1 expression showed also prognostic potential in MPM as summarized in a recent meta-analysis [85].

PD-L1 positivity in MPM tumor cells was reported at heterogeneous levels ranging from 16 to 68% [29,47,86]. According to Marcq et al., PD-L1 and PD-1 are decreased after chemotherapy [30]. However, for peritoneal mesothelioma, controversial effects of chemotherapy on PD-L1 expression were reported [87]. PD-1 expression on TIL was furthermore described as a negative predictive factor for response to chemotherapy [30]. Additionally, PD-L1 was found to correlate with the sarcomatoid and biphasic histology of MPM [86]. Our study group recently showed that tumor PD-L1 expression is not only prognostic in an international cohort suffering from malignant pleural effusion (in part also caused by MPM) but was significantly interacting with CRP, thus suggesting that the prognostic values of both markers influence each other. This observation translated to the poorest survival in the patient group characterized by high CRP in the patient blood and high PD-L1 expression in the corresponding tumor specimen [82]. Inaguma et al. demonstrated the independent prognostic impact of PD-L1 and activated leukocyte cell adhesion molecule (ALCAM, CD166) in MPM. Expression of both led to the shortest overall survival (OS). Additionally, a significant association between PD-L1 and ALCAM was drawn [46]. Similar prognostic results have been shown previously [43,47].

To reverse the tumor promoting effect of a downregulated specific immune system, checkpoint inhibitors like humanized monoclonal antibodies against PD-1 or PD-L1 have been developed. Immune evasion can be stopped to increase tumor defense [16,81]. The therapeutic benefit of targeting PD-1 with pembrolizumab [16] or nivolumab [17], and PD-L1 with avelumab [18], in pre-treated MPM patients with PD-L1 positive tumors was demonstrated.

The aforementioned other—by malignant disease misused—pathway of adapted immunity downregulation, namely CLTA-4 has also been investigated and proved to be an interesting treatment target to reactivate the immune system against MPM progression as reviewed before [88]. More recently, the combination of nivolumab with ipilimumab was approved by the FDA for unresectable MPM as first line therapy according to promising results documented during the CHECKMATE 743 trial, indicating that the combination of PD-1 and CLTA-4 targeting immune therapy is effective in MPM [19].

Finally, soluble PD-L1 (sPD-L1) from the sera of patients before and during immune therapy was suggested as a predictive biomarker, indicating poor treatment response when elevated before and during immune therapy. Additionally, sPD-L1 levels were also correlating to the inflammatory parameters NLR, neutrophil count and CRP, blood parameters that will be described later on in more detail [15]. Most recently, the role of sPD-L1 was also investigated in pleural effusions [89]. Both serum as well as pleural effusion derived PD-L1 status might represent an easily available method for clinical monitoring of the treatment target during immune therapy in the future.

Although there is great hope for a more personalized immune therapy, the exact background of the heterogeneity in PD-L1 expression and in treatment response is not yet fully understood. The interplay of tumor immunology, immunotherapy and somatic mutations is currently intensively researched [87,90–92]. Yang et al. recently reviewed the complex interactions of molecular characteristics of MPM cells and TME with histological subtype and genomic mutations [93] underlining the need for a deeper understanding of the pathobiological processes in MPM in order to optimize personalized biomarker-guided immunotherapy.

The complex interactions of tumor infiltrating immune cells with MPM cells as well as the resulting systemic inflammatory processes—which will be discussed in the next chapter—are also graphically shown in Figure 1.

κ *α* α **Figure 1.** Interaction of local and systemic immune response in malignant pleural mesothelioma. *AB* antibody*, B* Blymphocyte, *C4d* circulating complement component 4d, *CAF* cancer associated fibroblast, *CCSF* C-C motif chemokine ligand 2, *COX-2* cyclooxygenase-2, *CRP* C-reactive protein, *CTL* cytotoxic T-lymphocyte*, CTLA4* cytotoxic T-Lymphocyte Antigen 4, *FGF* fibroblast growth factor, *GM-CSF* granulocyte macrophage colony stimulating factor, *HGF* hepatocyte growth factor, *IL* interleukin, *Lymph* lymphocyte, *M2* M2-macrophage, *M-CSF* macrophage colony stimulating factor, *Mono* monocyte, *MPM* malignant pleural mesothelioma, NFκB nuclear factor kappa-light-chain-enhancer B, *Neutro* neutrocyte, *PD-1* programmed cell death protein 1, *PD-L1* programmed cell death ligand 1, *PDGF* platelet derived growth factor, *PGE2* prostaglandin E2, RNS reactive nitrogen species, *ROS* reactive oxygen species, *sPD-L1* soluble programmed cell death ligand 1, *T* T-lymphocyte, *TGF-ß* transforming growth factor-ß, *TIL* tumor-infiltrating lymphocyte, *TNF-α* tumor necrosis factor α, *Treg* regulatory T-cell, *VEGF* vascular endothelial growth factor.

In summary, the role of local inflammation and the components of TME in MPM have been investigated by various researchers. We encountered promising data regarding the prognostic potential of the different tumor infiltrating immune cells and also first results for predictive potential of some of these biomarkers. However, most interestingly, we noticed that generally low numbers of specific immune cells as well as high numbers of unspecific immune cells seem to be unfavorable, suggesting a controversial impact of the innate and adaptive immune cells on local tumor progression.

#### *2.10. The Role of Systemic Inflammatory Response in MPM*

Systemic inflammation is becoming an increasingly acknowledged factor in the development and progression of different solid tumors, including MPM. Consequently, peripheral blood derived inflammatory markers, which are determined routinely in daily practice for almost all patients, have been extensively examined regarding their applicability as biomarkers in MPM as reviewed before [12].

Since systemic inflammatory parameters can indicate inflammatory and infectious processes in the patient's body as well as malignancy, they are highly unspecific for diagnostic or screening purposes. However, after exclusion of acute inflammation or infection, some of the established and widely available inflammatory markers have been identified as prognostic or predictive markers in various solid tumors [94–99] including MPM [100–102].

As mentioned before, the current European guidelines for MPM management do not recommend any prognostic biomarkers for clinical use [4]. Nevertheless, two prognostic scores have been developed that are widely accepted and well established, namely the EORTC score (European Organization for Research and Treatment of Cancer) [103] and the CALGB score (Cancer and Leukemia Group B) [104]. Both scores have been validated for MPM by different researchers and proved their reproducibility [105–108]. Interestingly, these two scores not only integrate clinical, pathological and epidemiological factors, but also acknowledge systemic inflammation as tumor aggressiveness criteria by including the blood characteristics leukocytosis, thrombocytosis and elevated C-reactive protein (CRP) as negative prognostic factors [103,104]. The following paragraphs discuss the current literature on systemic inflammatory markers in MPM as also summarized in Table 2.


**Table 2.** Potential Systemic Inflammatory Biomarkers.

BC white blood cell, M-CSF macrophage colony stimulating factor, NLR neutrophil-to-lymphocyte ratio, LMR lympho-cyte-to-monocyte ratio, PLR platelet-to-lymphocyte ratio, IL-6 interleukin 6, CRP C-reactive protein, CAP CRP-to-Albumin ratio, mGPS modified Glasgow prognostic score, C4d Circulating complement component 4d, sPD-L1 soluble programmed cell death ligand 1, HR hazard ratio, N.R. not reported, N.S. not significant, Prog prognostic bi-omarker, Pred predictive biomarker, R retrospective, P prospective, \* most frequently used cut-off value.

Leukocytosis, a well-known biomarker of acute inflammation, has been widely investigated as biomarker for MPM and a number of studies reported a negative prognostic value of elevated pretreatment white blood cell count after uni- [103,111,112,116] and multivariate [102,110,115] survival analyses. Absolute lymphocyte count, as sign of an activated specific immune system, was studied by fewer researchers as single biomarker, but an association with poor prognosis and reduced clinical response to chemotherapy has been reported so far [109]. However, the role of lymphocyte count on MPM outcome has been investigated more intensively with regard to different ratios, especially the neutrophil to lymphocyte ratio (NLR) which will be later described more in detail.

Monocyte count on the contrary has been studied more extensively as single prognostic marker in MPM. Burt et al. found an independent negative prognostic value of pretreatment monocytosis for patients undergoing cytoreductive surgery [73] and Zhang et al. and Tanrikulu et al. confirmed these findings for patients receiving different therapies [100,110]. Monocytes, as part of the unspecific immune system, are the procurer cells of tissue specific macrophages [73] including TAM who play an important role in the TME and thus contribute to local immune modulation as mentioned before.

Interestingly, neutrophil count, likewise representing the unspecific immune response, is rarely reported as single blood marker. Few studies describe controversial results, reporting adverse prognostic value of high neutrophil count in univariate analysis [100] or no correlation with prognosis [109,110]. Nevertheless, the neutrophil count compared to the lymphocyte count is more intensively studied when it comes to the NLR.

Thrombocytosis, a known unspecific systemic phenomenon in response to inflammation [126], has long been suggested as independent prognostic factor. Already in 1989, first data suggested an independent negative prognostic value of high platelet count [111], which has been confirmed by others in the following decades [109–111,116,117]. Other studies could not validate the prognostic value at all [112], or found, instead of platelet count, the platelet to lymphocyte ratio (PLR) to be prognostic, as explained more in detail below [102–114].

Next to single blood parameters, special focus has lately been laid on ratios between some blood markers, such as the neutrophil to lymphocyte ratio (NLR), lymphocyte to monocyte ratio (LMR) or platelet to lymphocyte ratio (PLR). These markers are easily accessible and calculated from routine blood cell count and reflect the relation between specific and unspecific systemic immune response.

With increasing knowledge of the role of specific and unspecific immune response in cancer, these ratios have become of rising interest as possible biomarkers in numerous malignancies with promising prognostic potential [94,113–119].

As for other solid tumors [94,113–120], a negative prognostic value of high NLR has been shown for MPM in numerous studies analyzing cohorts of patients receiving different therapy concepts [102,103,112,119,121] including systemic therapy [121]. Furthermore, two studies found in a subgroup analysis that normalization of pretreatment elevated NLR under chemotherapy was predictive for better OS [106,121]. Additionally, for surgical patients undergoing cytoreductive surgery, high NLR was found to correlate with worse prognosis [122].

Low LMR, displaying a domination of unspecific monocytes, has been found to be a negative prognostic marker for numerous malignancies as reviewed by Gu et al. [123]. For MPM, comparable results have been published [102,114,122] showing that low LMR is associated with adverse prognosis in line with the reported negative prognostic value of elevated monocyte count as mentioned before. Of note, Yin et al. published comparable results for peritoneal mesothelioma [124].

Furthermore, high PLR has been studied and reported as a negative prognostic marker for multiple malignancies [119,125,127–129], also including MPM. As already indicated above, of the four named studies with no correlation between platelet count and survival, three, however, did find PLR to be associated with poor prognosis after univariate analyses [102,103,114]. Thus, one might speculate that even if absolute platelet count alone is not prognostic, a relative increase in platelets compared to low lymphocytes might be.

#### *2.11. Acute-Phase Proteins*

Already in 1998, Nakano et al. observed significantly elevated serum levels of some acute phase response proteins (APP) and cytokines—namely fibrinogen, IL-6, alpha1-acid glycoprotein and CRP levels—in MPM patients compared to patients with adenocarcinoma of the lung. They also reported significantly higher levels of IL-6 in the pleural fluid of MPM patients and concluded that the pleural IL-6, when entering systemic circulation, enhances the systemic acute phase response (APR) [130].

The APR, as part of the unspecific immune response, is the physiological and biochemical systemic reaction to inflammation, infection, tissue damage due to, for example burn injuries or trauma and malignancies. The process is mediated by proinflammatory cytokines, causing fever, leukocytosis and the release of APP. Gabay et al. provide a detailed list of well-known APP, some of which have been under investigation with regard to applicability as inflammatory biomarkers in MPM—particularly IL-6, CRP, fibrinogen [126].

#### *2.12. Interleukin 6*

The proinflammatory cytokine IL-6 is released by various immune cells triggered by IL-1β and TNF, but also produced by tumor cells directly as also proven for MPM with tumor-promoting effects [75,131]. The (patho)physiological functions of IL-6 are reviewed by Hunter in general [132] and by Abdul Rahim et al. for mesothelioma in particular—emphasizing the promoting effect of IL-6 on cell proliferation, angiogenesis via stimulation of VEGF expression, resistance to chemotherapy and physical symptoms negatively influencing wellbeing of the patient [133].

In contrast to other malignancies [134–144] current data does not support the prognostic or predictive value of IL-6 serum concentration for MPM [130]. However, IL-6 levels have been reported to correlate significantly with other markers [130,133] of verified prognostic impact for MPM such as VEGF [145–147], thrombocytosis [107,111,142] and CRP levels [11,116,142]. Adachi et al. found that IL-6 encouraged cell proliferation as autocrine growth factor and the expression of VEGF [148] and accordingly investigated an IL-6 inhibitor as VEGF targeting therapeutic approach in a subsequent study [149].

Antiangiogenic therapeutic approaches have been widely investigated as reviewed recently by Novak et al. [150]. Thus, the clinical use of the VEGF inhibitor bevacizumab is now also regarded as promising improvement of the almost 20 year old standard chemotherapy regimen published by Vogelzang et al. [151] according to the promising results of the MAPS trial [152].

From the current point of view, IL-6 does not seem to be applicable as prognostic or predictive marker for MPM; however, it can be assumed that it plays a major role in promotion of systemic inflammation with release of other proinflammatory cytokines as already suggested two decades ago by Nakano et al. [130].

#### *2.13. C-Reactive Protein (CRP)*

CRP, first described in 1930, is one of the earliest discovered and most established acute-phase response proteins [153]. The CRP synthesis in hepatocytes is mainly stimulated by IL-6, IL-1β and tumor necrosis factor α (TNF-α) [126]. Clinical use for inflammation and treatment response is currently well established since elevated CRP levels correlate with the course of chronic and acute infections but also inflammatory (autoimmune) disorders, general tissue injury and various malignancies [154,155]. Lately, elevated serum CRP levels were found to be associated with poor prognosis for multiple tumor entities [82,94,95,97,156–158]. Consequently, this potentially interesting biomarker has also been investigated in MPM. Elevated CRP levels were reported to be associated with shorter survival—regardless of different applied treatment modalities [103,116,123], specifically for patients receiving systemic treatment [112], as well as patients undergoing trimodal

therapy including surgery [11]. Some groups even described a level dependent negative prognostic potential of pretreatment CRP serum concentration [112,159].

Furthermore, the predictive potential of CRP for MPM has been explored by the study group of the authors before. It was proven that of patients undergoing aggressive multimodality treatment including cytoreductive surgery only those with normal pretreatment CRP levels benefit from this type of therapy. Thus, patients with normal CRP values before therapy receiving multimodality therapy survived 36 months in median. In contrast to these findings, patients with elevated pretreatment CRP only had 10 months overall survival despite multimodality therapy indicating, that indeed this subgroup of MPM is of distinct treatment responsiveness [11]. Of note, Kao et al. additionally described a correlation between elevated inflammatory markers—specifically elevated CRP and NLR and advanced clinical symptoms such as fatigue and anorexia in course of an engraved systemic inflammatory response and consequently compromised health-related quality of life [160].

#### *2.14. Fibrinogen*

Fibrinogen, a well-known clotting factor, is also an important positive acute phase protein. Its synthesis is increasing significantly when stimulated by proinflammatory cytokines, mainly IL-6 [161]. Fibrinogen as biomarker has been investigated for several diseases including chronic obstructive pulmonary disease and coronary heart disease [161,162]. Additionally, a negative prognostic value of high pretreatment fibrinogen has been found for numerous tumor entities [98,114,163–167]. So far, only the previous study of the authors reported not only a prognostic but also predictive value for pretreatment fibrinogen in MPM. Patients with high fibrinogen plasma levels were shown to have significantly shorter OS. Additionally, of patients receiving trimodal treatment with cytoreductive surgery, those with high pretreatment fibrinogen did not benefit from multimodality treatment [10].

#### *2.15. Albumin—A Negative Acute Phase Protein*

Serum albumin not only reflects nutritional status but also inflammatory response as negative acute phase protein mediated by cytokines including IL-6, IL-1β and TNF-α [168]. Hypoalbuminemia has long been acknowledged to impair wound healing and outcome after interventions and surgeries [169–173]. In addition, it was described to indicate short survival in different malignancies [168]. For MPM, hypoalbuminemia has been associated with poor survival for patients receiving different treatment modalities [102,103,114], but also selectively for chemotherapy patients [174] and surgical patients [175].

In a classification and regression tree analysis, Brims et al. found the best survival for patients with no weight loss, a high hemoglobin level and a high serum albumin level [176]. Harris et al. validated these findings for surgical patients undergoing cytoreductive surgery [175].

Hypoalbuminemia and elevated CRP concentration have been integrated in a systemic inflammation based prognostic score, the so-called modified Glasgow Prognostic Score (mGPS). Its prognostic value has also been confirmed for multiple cancer types as reviewed in detail by McMillan [177] and has been acknowledged for mesothelioma in univariate analysis [101].

Furthermore, the prognostic value of elevated CRP/Albumin ratio (CAR), reflecting increased CRP values and decreased albumin concentration as indicator of poor nutritional and activated acute phase response, has been widely explored. Elevated CAR has been shown to predict poor outcome in acute diseases including sepsis [178,179] as well as in various malignant diseases [96,99,180–183]. Takamori et al. investigated CAR for MPM patients and found a high CAR to be independently prognostic [184]. Otoshi et al. confirmed these results for inoperable MPM patients [102] whereas Tanrikulu et al. could not reproduce these results [100].

#### *2.16. Ferritin*

The positive APP ferritin is up-regulated under inflammatory or infectious conditions to reduce the iron accessibility of pathogenic organisms [185,186]. For numerous malignancies elevated serum ferritin concentrations have been reported as well, in part with prognostic impact [187]. Healthy human cells, foreign organisms but also cancer cells depend on iron supply for a number of cellular metabolic processes. The role of iron metabolism and its regulation—partially by cells of the TME—have been reviewed excellently by Hsu et al. for cancer in general [187] and by Toyokuni et al. for MPM in particular, especially in context of asbestos-induced oxidative stress [188]. MPM has been associated with elevated ferritin serum levels [189–191], but to our knowledge the prognostic or predictive impact of ferritin has not been investigated so far. However, correlations of ferritin with TAM and modulated lymphocyte function has been suggested [186,187] so in context of the APR as well as the importance of iron metabolism in MPM the study of ferritin as biomarker might reveal interesting new results. Of note, also reduction of iron storage was suggested as possible treatment target after promising preclinical results from a rat model [191,192].

#### *2.17. The Complement System*

With regard to the innate immune system and its systemic circulating compartments, the complement component 4d (C4d) was also found to be of prognostic relevance in MPM patients. High plasma C4d levels were associated with high tumor volume, worse response to induction therapy, high acute phase response proteins and shorter survival after multivariate analyses as reported by Klikovits et al. [8]. Furthermore, Agnostinis et al. investigated the role of complement protein C1q in MPM. It was shown that C1q did not activate the classic complement pathway in MPM as one might expect, but instead bound to hyaluronic acid and thereby induced cell adhesion and proliferation of mesothelioma cells. Interestingly, the activation of the classic complement pathway was abandoned by hyaluronic acid [193]. These findings are in line with Klikovits et al., where high C4d (as downstream target of C1q during the classic complement pathway) was not correlating with high C1q [8]. Thus, the activation and exact role of the complement system and its subunits is yet not fully understood and might be of future interest in MPM.

Taken together, many common systemic inflammatory parameters have been studied regarding their prognostic potential for MPM and some additionally for their predictive impact. It is remarkable—compatible with our conclusions on local inflammation—that high unspecific inflammatory markers seem to be adverse whereas high specific inflammatory markers appear to be beneficial reflecting the tumor-promoting influence of the innate immune system and the tumor-suppressing impact of the adaptive immune system, respectively.

#### **3. Conclusions**

While preparing the present review and summarizing the established as well as most recent knowledge, it became fairly clear that a large amount of research considering this topic has been performed within the past few decades. Despite the fact that a lot of data is based on retrospective studies—which is most likely explainable by the rare incidence of MPM—high quality research supports the important role of inflammation in MPM. Not only in the setting of pathogenesis, tumor promotion, poor prognosis or treatment response inflammatory processes play a decisive role but inflammation and immune response are also under investigation as promising treatment targets. Most markers and key findings were not only published once but have been validated in the past, thus resulting in several inflammatory related biomarkers characterized by reproducibility and accordingly reliability. Furthermore, comparable results have been documented in other malignancies thus indicating, that some of the above mentioned findings have a universal impact in (immune-) oncology.

In the clinical management of MPM, physicians are confronted with multiple—yet not fully standardized—treatment modalities on the one hand, opposed to poor outcome and treatment resistance on the other hand. Consequently, MPM in general is the ideal candidate for biomarker research especially when it comes to treatment guiding predictive parameters [22].

The immune system plays a key role in MPM, since this rare disease has already been associated with inflammation several decades ago [5]. This theory is supported by many more recent studies summarized in this review. During the past few decades and especially within the last years, it was shown that an upregulated unspecific immune response on the one hand translates to poor outcome. On the one hand, a downregulated specific immune system results in tumor progression, tumor immune evasion and finally poor prognosis, regardless if investigating the inflammatory status in patient blood, pleural effusion, tumor tissue or its associated TME. Thus, the tumor suppressive characteristics of the specific immune system get obvious when the—through MPM suppressed—specific immune response gets reactivated by immune therapy resulting in prolonged survival.

This above-described duality of the immune system in MPM has been analyzed and described before by Linton et al. [12]. However, today we would reply to the question "Inflammation in malignant mesothelioma—friend or foe?" that it is both friend and foe. More precisely, and to simplify the key message from this review, the specific immune system is our friend and its unspecific counterpart the foe which is also reflected in the prognostic value of the corresponding biomarkers—both on a local as well as systemic level.

Further research on the immune system in MPM might help in treating this therapy refractory disease and reveal modern insights in the complex interaction of our immune system with the tumor thus resulting in a better biological understanding, new treatment approaches and finally improved clinical management and patient outcome. There is still need for future studies to gain detailed knowledge about this topic and, thus, we look forward to learning more about the interaction of our immune system with malignant disease in general and MPM in particular.

**Author Contributions:** All authors contributed to the conceptualization of this review, M.V., B.G. and A.R. conducted the literature research and wrote the manuscript; T.B., A.S., M.B. and E.S. revised the original draft critically; M.V. was responsible for visualization of data in tables and the figure. All authors gave final approval of the version to be published and agreed to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by Open Access Funding by Karl Landsteiner University of Health Sciences, Krems, Austria.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** No new data were created or analyzed in this study. Data sharing is not applicable to this article.

**Acknowledgments:** This work was generously supported by Open Access Publishing Fund of Karl Landsteiner University of Health Sciences, Krems, Austria.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


## *Review* **Pathological Characterization of Tumor Immune Microenvironment (TIME) in Malignant Pleural Mesothelioma**

**Francesca Napoli <sup>1</sup> , Angela Listì 2 , Vanessa Zambelli <sup>1</sup> , Gianluca Witel <sup>3</sup> , Paolo Bironzo 1,2, Mauro Papotti 1,4 , Marco Volante <sup>1</sup> , Giorgio Scagliotti 1,2,† and Luisella Righi 1,\* ,†**


**Simple Summary:** Tumor immune microenvironment is an important structural component of malignant pleural mesothelioma that contributes to disease growth support and progression. Its study and pathological characterization are important tools to find new biomarkers for advanced therapeutic strategies.

**Abstract:** Malignant pleural mesothelioma (MPM) is a rare and highly aggressive disease that arises from pleural mesothelial cells, characterized by a median survival of approximately 13–15 months after diagnosis. The primary cause of this disease is asbestos exposure and the main issues associated with it are late diagnosis and lack of effective therapies. Asbestos-induced cellular damage is associated with the generation of an inflammatory microenvironment that influences and supports tumor growth, possibly in association with patients' genetic predisposition and tumor genomic profile. The chronic inflammatory response to asbestos fibers leads to a unique tumor immune microenvironment (TIME) composed of a heterogeneous mixture of stromal, endothelial, and immune cells, and relative composition and interaction among them is suggested to bear prognostic and therapeutic implications. TIME in MPM is known to be constituted by immunosuppressive cells, such as type 2 tumor-associated macrophages and T regulatory lymphocytes, plus the expression of several immunosuppressive factors, such as tumor-associated PD-L1. Several studies in recent years have contributed to achieve a greater understanding of the pathogenetic mechanisms in tumor development and pathobiology of TIME, that opens the way to new therapeutic strategies. The study of TIME is fundamental in identifying appropriate prognostic and predictive tissue biomarkers. In the present review, we summarize the current knowledge about the pathological characterization of TIME in MPM.

**Keywords:** mesothelioma; tumor microenvironment; tumor-associated macrophages; dendritic cells; immunohistochemistry

#### **1. Introduction**

Malignant pleural mesothelioma (MPM) is a rare and highly aggressive disease arising from pleural mesothelial cells. The recognized risk factors of MPM are asbestos, radiation exposure, genetic mutations, and the exposition to Simian Virus 40, but asbestos is certainly the most relevant and most well-known cause [1]. The overall prognosis of advanced stage MPM is poor, with a median survival of less than 15 months [2]. MPM consists of three histological variants: epithelioid (~60% of mesotheliomas), sarcomatoid, characterized

**Citation:** Napoli, F.; Listì, A.; Zambelli, V.; Witel, G.; Bironzo, P.; Papotti, M.; Volante, M.; Scagliotti, G.; Righi, L. Pathological Characterization of Tumor Immune Microenvironment (TIME) in Malignant Pleural Mesothelioma. *Cancers* **2021**, *13*, 2564. https:// doi.org/10.3390/cancers13112564

Academic Editors: Daniel L. Pouliquen and Joanna Kopecka

Received: 24 April 2021 Accepted: 19 May 2021 Published: 24 May 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

by spindle cell morphology (~20% of mesotheliomas), and biphasic, which presents both epithelioid and sarcomatoid features (~20% of mesotheliomas) [3]. Diagnosis of MPM relies on an integration of clinical, radiological, and pathological findings, with histological examination being the mainstay for diagnosis and prognostication [4,5]. Since MPM is diagnosed in advanced stage in the majority of cases, the standard of care consists in systemic chemotherapy. However, the standard combination of cisplatin and pemetrexed chemotherapy agents [6] prolongs the median survival time by approximately 3 months only [7]. In the last years, genetic studies on MPM reported a low prevalence of oncogene driver mutations and low tumor mutational burden, but frequent copy-number losses and recurrent somatic mutations in oncosuppressor genes such as BAP1, NF2, and CDKN2A [8–13]. Unfortunately, no targeted therapies exploiting these alterations have emerged.

The etiopathogenetic evolution of MPM is mostly due to the generation of a tumor immune microenvironment (TIME) as a consequence of asbestos-induced damage, that may support tumor growth, possibly in association to genetic predisposition [14,15]. Over time, chronic inflammation determines an increased production of free radicals and reactive oxygen species by inflammatory cells and/or an alteration of immunocompetent cells, resulting in a reduction of tumor immunity [16].

The unique role of TIME in MPM development and progression still needs an accurate characterization in terms of infiltrating cell types, expression of co-inhibitory molecules, and activation of immune pathways (e.g., INFγ). As histological examination remains the gold standard in the diagnosis of MPM, the characterization of TIME could be crucial to visualize all cellular components and achieve a better understanding of the disease. Despite the different biological and clinical features between pleural and peritoneal mesothelioma [17], the presence of tertiary lymphoid structures (TLS) as a component of the host immune response was highlighted in epithelioid peritoneal mesothelioma (EMPM), as well. However, no association between TLS-EMPM and different oncological outcomes was found, thus suggesting that TLS would reflect an indirect mechanism of therapy resistance to drugs in EMPM as in its pleural counterpart [18].

Given the role of TIME in MPM, the use of immune checkpoint inhibition treatment has the rationale to provide new potential therapeutic opportunities. Indeed, the combination of monoclonal antibodies directed against programmed cell death protein 1 (PD-1) and cytotoxic T-lymphocyte-associated protein 4 (CTLA-4) recently showed its superiority over platinum-pemetrexed chemotherapy in a phase 3 trial [19]. Notably, a greater benefit was observed in biphasic/sarcomatoid MPM. Moreover, single-agent anti-PD-1 therapy demonstrated to significantly increase survival as compared to best supportive care in platinum pre-treated patients [20].

Another novel potential treatment in MPM is cell therapy. Clinical trials using CAR-T cells in MPM have shown that this potential therapy is relatively safe, but efficacy remains modest, likely due to the strong immunosuppressive conditions in MPM microenvironment [21,22]. Furthermore, preclinical studies are ongoing in a bimodal treatment approach consisting of dendritic cell (DC) vaccination to prime tumor-specific T cells, a strategy to reprogram the desmoplastic microenvironment in mesothelioma and pancreatic cancer [23].

The most adequate tissue specimens for MPM pathological characterization derive from video-assisted thoracoscopic (VATS) biopsies or pleurectomies, which are the recommended samples for complete histological diagnosis [2,3,7]. The availability of large amounts of tissue allows both the definition of histological tumor features and immune cells' spatial distribution on routine hematoxylin-eosin slide. On these specimens, the cheapest and fastest tool used for pathological characterization studies is immunohistochemistry (IHC), that allows to visualize both tumor cells and microenvironment components, according to their immunophenotype and biomarker expression (Figure 1). Despite its advantages, a limitation of chromogen-based IHC analysis is the impossibility of using more than one or two markers per slide. Novel and innovative multiplex immunophenotyping techniques are in development to deeply analyze as a whole both the spatial distribution and immunophenotypic interaction of each single cell subtype [24–26].

**Figure 1.** Pathological characterization of TIME in MPM. Histological appearance of MPM, epithelioid type (**a**), Ematoxilin & eosin (100×); reticulin stain showing connective tissue around neoplastic cells (**b**), (100×); SMA IHC stain showing scattered fibrocytes (**c**), (100×); CD3 IHC stain highlighting T lymphocytes (**d**), (100×); CD4 IHC stain showing scattered T cells (**e**), (100×); CD8 IHC stain showing moderate T lymphocyte infiltrate (**f**), (100×); CD20 IHC stain showing a small aggregate of B lymphocytes (**g**), (100×); CD68 IHC stain showing diffuse macrophage infiltration (**h**), (100×); CD163 IHC stain showing activated TAMs (**i**), (100×); PD-L1 IHC stain showing small aggregates of positive tumor cells (**j**), (100×); VISTA IHC stain showing moderate expression in immune cells (**k**), (100×); STING IHC stain showing diffuse immune cell positivity (**l**), (100×). Notes: TIME: tumor immune microenvironment; MPM: malignant pleural mesothelioma; SMA: smooth muscle actin; IHC: immunohistochemistry; PD-L1: programmed death ligand 1; VISTA: V-domain Ig-containing suppressor of T-cell activation; STING: STimulator of Interferon Genes.

Given the need to explore TIME in its components and constituents, in this review, we summarize the current data on TIME pathological characterization and biomarker identification in MPM.

#### **2. The Tumor Immune Microenvironment**

TIME is a complex and heterogeneous mixture of stromal, endothelial, and immune cells admixed in a connective matrix and its composition differs among individuals and histological types. In fact, studies suggest that TIME profoundly differs between epithelioid and non-epithelioid pleural mesotheliomas: the former typically have an immuneactivated TIME with greater proportion of plasmacytoid dendritic cells (DC), CD20+ B cells, CD4+ helper T cells, and exhausted CD8+ tumor-infiltrating lymphocytes (TILs), whereas non-epithelioid mesotheliomas have a TIME with a larger proportion of macrophages, regulatory T cells, mesothelioma stem cells, and neutrophils [27].

In past years, the prognostic and predictive role of TIME in MPM was investigated mainly on small and heterogeneous series, with no conclusive data due to difficulties in MPM microenvironment characterization [28,29]. Moreover, qualitative and quantitative changes in tumor/stroma ratio may produce a dramatic rewiring in the MPM-infiltrating immune cell subsets [30].

Increasing evidence suggests that analysis of gene expression or copy numbers in cancer samples helps to understand immune cell infiltration into the tumor ME. Yoshihara et al., by means of transcription profiling, have developed the ESTIMATE algorithm (Estimation of STromal and Immune cells in MAlignant Tumor tissues using Expression data) to analyze the stromal and immune infiltration associated to tumor cellularity in cancer samples [31]. Using gene expression data, a 'stromal signature', that describes the presence of stroma in tumor tissue, and an 'immune signature', that represents the infiltration of immune cells, were identified. Recently, the ESTIMATE algorithm was applied to MPM samples and the involvement of 14 immune/stromal-related genes was found to have significant prognostic potential. In silico analyses revealed that all these genes are involved in immune responses and may predict the survival of patients with MPM, playing also a role as biomarkers of the sensitivity to immunotherapy [32].

Additionally, Lee and coworkers, using mass spectrometry and comprehensive analysis of intra-tumoral immune system, described a distinct immunogenic TIME signature which was associated with favorable OS and response to checkpoint blockade [33]. The importance of understanding TIME of different MPM histotypes in relation to hypofractionated radiation therapy response was recently demonstrated as well [27].

#### *2.1. Extracellular Matrix and Stroma Components*

In MPM, the intra-tumoral stroma is not merely a scaffold but also promotes tumor growth, invasion, and protection from an anti-tumor immune response.

Several studies reported that many genes involved in extracellular matrix (ECM) production and remodeling are upregulated in MPM, especially in the biphasic [34] and sarcomatoid [35] variants. Furthermore, increased expression of these ECM-related genes is associated with "immune desert" tumor regions, characterized by a poor lymphocytic infiltrate, suggesting that MPM-altered stroma might act as a barrier to the immune response [36]. Very recent studies that analyze public mRNA-sequencing datasets through bioinformatic analyses have identified several differentially expressed genes (DEGs) in MPM. In these studies, genes specifically associated to the ECM component, structural constituents, organization, and receptor interaction were found overexpressed. These genes resulted in being involved in different protein–protein interaction (PPI) networks, gene ontology (GO), biological processes (BC), and molecular functions [37,38], and were also validated in MPM cell line models [39].

ECM components such as collagen, laminin, fibronectin, and integrins can be produced by mesothelioma cells that can also promote, under the influence of various growth factors, the synthesis of matrix metalloproteases (MMP), favoring ECM remodeling and tumor cell invasion [40].

In vitro studies demonstrated that different histotypes are characterized by specific ECM profiles, and that these differences determine a varying ability of MPM cells to spread and migrate towards ECM substrates [41,42]. In particular, characterization of cell culture conditions showed that 3D growth of malignant cells was enhanced in the presence of their own ECM, while invasion was stimulated by fibronectin in epithelioid and biphasic MPM histotypes, while homologous cell-derived ECM stimulated invasion in the most aggressive (sarcomatoid) form of MPM.

Furthermore, inhibition of collagen production delays MPM tumor growth [43]. Morphometric and immunohistochemical analysis of tumor collagen V (Col V), along with the quantitative inverse relationship between Col V and CD8+ T lymphocytes, demonstrated that high levels of Col V and low CD8+ T lymphocytes confer an immune-privileged TIME for tumor invasion and poor patients' prognosis [44,45].

The architecture of connective tissue in MPM per se, highlighted by silver-based reticulin staining (Figure 1b), has been recently proposed to distinguish the transitional variant of MPM, showing intermediate features between epithelioid and sarcomatoid histotypes, and bearing a specific prognosis [46]; in fact, a delicate reticulin pattern around single cells is indicative of this transitional type, as compared to a rough pattern banding individual cells in the sarcomatoid, and a large cluster pattern in the epithelioid type [47].

#### Cancer-Associated Fibroblasts (CAFs)

Tumor stroma is mostly composed by both fibrocytes with small spindle-shaped nuclei, derived from macrophages or dendritic cells, and activated fibroblasts (or cancerassociated fibroblasts, CAFs) that are identified by alpha smooth muscle actin (SMA) (Figure 1c) [48,49].

In recent years, fibroblast growth factor receptor (FGFR) signaling has been recognized as increasingly important, both in cancer pathogenesis and as a potential therapeutic target [50]. There are strong preclinical data suggesting that FGF is important in MPM as well. In MPM cell lines, FGFR1 and FGFR2 are co-expressed and their expression is strongly associated with sensitivity to FGFR-active tyrosine kinase inhibitors [51]. Inhibiting FGF autocrine signaling using an FGF-ligand trap reduces proliferation in MPM cell lines and reduces tumor growth in xenografts [52]. Unfortunately, the phase II clinical trial with a FGFR 1–3 inhibitor did not demonstrate efficacy in patients with MPM, who had progressed after first-line treatment with platinum-based chemotherapy [53].

CAFs have been shown to exert pro-tumorigenic effects by secreting several growth factors that promote cancer cell proliferation and invasion [54]. Literature data reported that TGFβ, IL-6, and CCL2, synthetized by CAFs [55], were detected in pleural effusions of MPM patients, where they seem to contribute to the recruitment and differentiation of immunosuppressive cells [56,57].

Our group identified Caveolin 1 (CAV1)-positive CAFs in a subgroup of epithelioid MPM with poorer prognosis [58]. CAV1 acts as a multifunctional scaffolding protein with multiple binding partners and is associated with cell surface caveolae in the regulation of lipid raft domains, but it is also involved in cancer growth and progression, modulating tissue responses through architectural regulation of the microenvironment. Recently, caveolae and their components emerged as integrators of different cell functions, mechanotransduction, and ECM–cell interactions [59]. Furthermore, in vitro studies on quantitative proteomic profiling revealed that CAV1 is required for exosomal sorting of ECM protein cargo subsets and for fibroblast-derived exosomes to efficiently deposit ECM and promote tumor invasion of breast cancer cells [60].

Furthermore, connective tissue growth factor (CTGF), a pro-tumorigenic CAF marker [49], is more expressed in sarcomatoid than in epithelioid MPM [61], and it is produced by both MPM cells and fibroblasts, and promotes the invasion of MPM cells in vitro [62]. Ohara's group has demonstrated that a CTGF-specific monoclonal antibody (FG-3019, pamrevlumab) could inhibit mesothelioma cell growth in vitro [63]. Based on these data, it was suggested that the use of FG-3019, currently under clinical trials for idiopathic pulmonary fibrosis [64] and pancreatic ductal adenocarcinoma [65], could be a therapeutic option for MPM. This is supported by preclinical data including a strong in vivo cancer

growth inhibition observed in melanoma and pancreatic cancer with the use of the same anti-CTGF monoclonal antibody [66,67].

#### *2.2. Inflammatory Cellular Component of TIME*

#### 2.2.1. Tumor-Associated Macrophages

Macrophages are specialized phagocytic cells that play a dual role in cancer depending on their differentiation. Tumor-associated macrophages (TAMs) derive from circulating monocytic precursors and are the major component of MPM TIME (Figure 1h,i). They are divided into two classes: classically activated (M1) macrophages, which have proinflammatory, tissue destructive, and anti-tumor activity, and alternatively activated (M2) macrophages, which have pro-tumorigenic properties [68]. M2 macrophages are the ones mostly present in MPM and their differentiation is regulated by interleukins, such as IL-4, IL-13, and IL-10, produced by tumor-infiltrating lymphocytes (TILs) [69].

Asbestos phagocytosis by macrophages triggers the formation of the inflammasome complex and promotes secretion of IL-1β [70,71]. Additionally, IL-1β/IL-1 receptor (IL-1R) signaling was reported to contribute to the oncogenesis of asbestos-induced mesothelioma [72]. These studies highlight the important role of the inflammasome in MPM development. The phagocytosed asbestos fibers remain undegraded and induce apoptosis of macrophages [73]. Undegraded asbestos fibers then undergo phagocytosis by nearby macrophages. Thus, asbestos is not completely removed and constitutively activates the inflammasome in macrophages. Moreover, it was reported that high mobility group box 1 (HMGB1) protein is abundantly secreted by MPM cells and serum levels of HMGB1 are associated with poor prognosis in MPM patients [74,75]. HMGB1 is one of the damageassociated molecular pattern proteins and promotes pro-IL-1β production functioning as an agonist of Toll-like receptor 4 (TLR4) [76]. Both HMGB1 derived from MPM cells and asbestos-activated inflammasome in TAMs induce IL-1β production, resulting in enhanced aggressiveness of MPM [77].

The tissue localization of M2 macrophages has been investigated in different immunohistochemical studies. Marcq and coworkers demonstrated that the number of stromal CD68+ macrophages found in MPM specimens was positively correlated to the number of stromal Tregs, suggesting a direct action of macrophages on stimulating and recruiting CD4+ immunosuppressive cells [78]. Burt et al. found that the absolute number of CD68+ macrophages was associated with worse prognosis in non-epithelioid MPM [79]. Finally, Cornelissen and coworkers reported that patients who develop recurrence after radiation treatment have a higher M2/total TAM ratio and lower CD8+ cell count at diagnosis, compared to patients who did not develop this outgrowth [80].

#### 2.2.2. T Cells and Natural Killer Cells

The CD3+ T-lymphocytes are the second most common immune cell type in MPM (Figure 1d–f) TIME and constitute, on average, 20–42% of the immune cell infiltrate [68,81]. T helper CD4+ cells play an important role in the generation of T cell-mediated antitumor response via activation of antigen-presenting cells (APCs), which stimulate CD8+ cytotoxic TILs and natural killer (NK) cells. The latter are lymphoid cells of the innate immune system with strong immunostimulatory functions and cytotoxic capacity [82].

A recent study by Alay and coworkers, performing an integrative transcriptome analysis on a publicly available dataset of 516 MPMs, revealed a clinically relevant immunebased classification based on CD4+ T-helper 2 (TH2) and CD8+ cytotoxic T cells, that were found to be consistently associated with better overall survival [83].

CD8+, CD4+, and CD4+/FoxP3+ T-cells are present in the majority of patients [84], but the number of T-reg cells in pleural effusions of MPM patients is lower than in other solid tumors [85], confirming the presence of an immunosuppressive milieu in MPM tumoral mass, rather than in pleural effusion [86]. The positive effect of CD4+ tumor-infiltrating lymphocytes (TILs) on prognosis has been previously suggested for epithelioid [78,87–89], but remains controversial in sarcomatoid MPM [81,88]. On the other hand, low CD8+ and high FoxP3+ TILs counts were shown to correlate with a high risk of both death and recurrence, regardless of the presence of a sarcomatoid component [87–90].

Ujiie and coworkers demonstrated the prognostic role of CD8+ and CD20+ expressing lymphocytes in 230 epithelioid mesothelioma patients [29]. In particular, they found that rather than the single type of infiltrating cells, the combination of high M2-polarized TAMs (CD163<sup>+</sup> ) with low CD8+ T cells, and low M2-polarized TAMs (CD163<sup>+</sup> ) with high CD20+ B cells, were independent markers of worse and better overall survival, respectively. These data were confirmed by Pasello et al., except for the fact that CD8+ T-lymphocytes were found in MPM samples showing aggressive features (sarcomatoid/biphasic histology, higher necrosis, and proliferation index), when associated with higher CD68+ macrophages and PD-L1 expression [90].

In a study by our group, Salaroglio et al. [91] performed a simultaneous comprehensive analysis of the immune infiltrate in pleural fluid and fresh pleural biopsy tissues aiming to identify an immune phenotype with diagnostic and prognostic value in MPM patients. It was confirmed that CD8+ TILs in pleural effusion have no prognostic significance, while intratumor immune infiltrate is more effective in predicting the patient's outcome. The same result was obtained by Chee et al., who state that high proportions of FoxP3+ T cells are associated with a poor prognosis in epithelioid and sarcomatoid tumors [88].

Moreover, Fusco et al. found an increased presence of stromal CD4+ T and CD19+ B lymphocytes with a positive correlation between each other, possibly indicating a positive feedback loop between these two lineages [92].

Our group also characterized TIME in MPM by immunohistochemistry, as a validation step of gene expression profiling. In MPM cases with higher expression of T-cell lineage genes, T-effector genes, and T-regulatory genes, we observed a high expression of CD3+ T-infiltrating lymphocytes, with a similar amount of CD4+ and CD8+ T-cells. On the contrary, high amounts of CD20+ B lymphocytes, with follicular chronic inflammation as a morphological hallmark, were observed in the group that showed higher relative expression of B cell and lower expression of T cell genes [36].

#### 2.2.3. Myeloid-Derived Suppressor Cells

Myeloid-derived suppressor cells (MDSC) are myeloid cells with suppressive activity on innate and adaptive immune cells that have been described to inactivate immune response against the tumor in cancer patients [93]. Based on their surface markers, MDSC can be subdivided in granulocytic MDSC (GR-MDSC), which express granulocytic markers like CD66b and/or CD15, and monocytic MDSC (MO-MDSC), which express the monocytic antigen CD14 [91]. The main mechanisms by which MDSC exert their suppressive activity on other immune cells are the depletion of arginine and tryptophan by expression of effector enzymes arginase I (Arg I), inducible NO-synthase (iNOS), and indolamin-2,3-dioxygenase (IDO), as well as by production of reactive oxygen species (ROS) [93].

In mice, MDSCs are characterized by IL-4 expression [94]. Burt et al. found IL-4R to be highly expressed on the surface of human MPM tumor cells: IL-4R was present in 97% of epithelial and 95% of non-epithelial tumors. Only a scattered and small fraction of stromal cells stained positive for IL-4R, and conversely, IL-4R-positive macrophages were predominantly found in the stroma [95]. Myeloid CD33+ cells were found to represent approximately 42% of CD45+ immune cells: 0.6–31% of these myeloid cells were typed as MDSCs [96].

In their study, Salaroglio et al. reported that GR- and MO-MDSCs abrogated proliferation and cytotoxic activity of autologous TILs and of TILs derived from patients with pleuritis, suggesting an important role of MDSCs in immunosuppression mediation. Moreover, the intratumor-infiltrating MDSCs, but not the MDSCs of pleural fluid, resulted significantly associated with poorer PFS and OS [91].

Furthermore, it was recently reported that MPM TIME is enriched in infiltrating granulocytes, which inhibit T-cell proliferation and activation. Immunohistochemistry and transcriptomic analysis revealed that a majority of MPMs express GM-CSF, and that high GM-CSF expression correlates with clinical progression. Blockade of GM-CSF with neutralizing antibodies or ROS inhibition restores T-cell proliferation, suggesting that targeting GM-CSF could be of therapeutic benefit in MPM patients [97].

#### 2.2.4. Dendritic Cells

Dendritic cells (DCs) are powerful antigen-presenting cells with key roles in the initiation and regulation of immune responses. DCs are unique in their ability to activate naïve T cells and initiate primary immune responses in lymph nodes, and they also play a central role in reactivating memory T-cell responses in the lungs. DC-derived signals regulate both the degree of T-cell activation and the nature of immune response (e.g., T helper (Th) 1, Th2, Th17, B-cell help) [98]. Several DC subpopulations have been defined: DCs are broadly divided into myeloid dendritic cells (mDCs), usually referred to as conventional dendritic cells (cDCs), and plasmacytoid dendritic cells (pDCs). In human lungs, cDCs form dense networks throughout the epithelium of large conducting airways, bronchioles, alveoli, and interstitial space, and they express CD141, CD1c, and the C-type lectin domain family 9 member A (CLEC9A) [99,100]. pDCs are best characterized by their ability to synthesize great amounts of IFN. They are relatively inefficient at presenting antigens to T cells and seem to play an important role in tolerance induction, probably via induction of regulatory T cells. In humans, pDCs are identified by surface markers such as CD303 (a C-type lectin), CD304 or neuropilin-1, Ig-like transcript 7, and IL-3 receptor-a chain [101]. Under normal conditions, activated pDCs exhibit robust IFN-α production and promote both innate and adaptive immune responses. In several cancer models [102], including MPM [103], pDCs demonstrate an impaired response to T activation, decreased or absent IFN-α production, and contribute in establishing an immunosuppressive TIME and a reduced ability to generate effective anti-mesothelioma T cell responses. On the other hand, a comprehensive proteomic analysis on 12 surgically resected MPMs highlighted a correlation between the presence of activated pDCs (CD40+ and CD86+) and tumors having a good TIME signature as well as a favorable response to immune checkpoint therapy [33]. Finally, evidence to date suggests that CD40+ DC activation is a critical and nonredundant mechanism to convert "cold" tumors (i.e., lacking a T cell tumor infiltrate) into "hot" ones (i.e., having a prominent T cell tumor infiltrate), sensitizing them to checkpoint inhibition therapy [104,105].

#### 2.2.5. B Lymphocytes

B lymphocytes contribute to humoral immunity as they can differentiate into antibodysecreting plasma cells. Additionally, B cells can stimulate T cells or serve as APCs. In MPM, B lymphocyte infiltrate is associated with better patient survival [29,90]. Generally, B cell infiltrate in mesothelioma is scant [55,89].

As mentioned above, in our study on immune gene expression profiling in MPM, the subgroup with downregulated T-cell effector and upregulated B-cell genes failed to show correlation with increased expression of genes associated with antigen presentation, thus we concluded that these B cells may be part of the adaptive cytotoxic response [36].

#### *2.3. PD-L1 and Other Immune Checkpoints*

The programmed cell death pathway (PD-1/PD-L1) plays a critical role in tumor immune escape control. PD-1 is mainly expressed on activated CD4/CD8 T cells and B cells [96]. PD-L1, the ligand of PD-1, is not only expressed in immune cells, but also in others, including cancer cells, helping immune evasion by interacting with PD-1 on T-cells [106]. The interaction between tumor PD-L1 and PD-1 on T cells results in the inhibition of T cell activation and proliferation, as well as immune evasion by PD-L1 expressing tumors [107].

PD-L1 immunohistochemical expression in tumor tissue has been widely accepted as a predictive biomarker [108], because of its association with increased efficacy of immune checkpoint inhibitors (ICIs) in several malignancies [109]. Immunotherapy based on

monoclonal antibodies against PD-1 and PD-L1 has also been tested for MPMs in clinical trials (Figure 1j). Several nonrandomized phase I/II trials, testing single-agent ICI, showed variable antitumor activity (9–29%) and median progression-free survival ranging from 2.8 to 6.2 months [110]. Preliminary results from phase II clinical trials combining inhibitors of cytotoxic T-lymphocyte-associated antigen 4 (CTLA4) and anti-PD1/PD-L1, such as ipilimumab, nivolumab, tremelimumab, or durvalumab, showed promising results but significant toxicity [111]. In those clinical trials, PD-L1 expression showed limited value in predicting benefit from ICIs, and PD-L1 expression analysis currently has no role as a clinical predictive biomarker in MPM. Moreover, the prognostic value of PD-L1 expression in MPM is controversial. In a recent meta-analysis, Jin et al. reported that PD-L1 overexpression significantly correlated with poor overall survival, irrespective of the sample size of the series, treatment method, or PD-L1 cut-off value. Furthermore, overexpression of PD-L1 was associated with sarcomatoid and biphasic histology [112].

The above-mentioned integrative transcriptome analysis of MPM [83] revealed a clinically relevant immune-based classification of the same, identifying three immune groups (IG1–IG3) that represent different immune infiltration patterns and are associated with distinct survival outcomes. The group with the shortest overall survival (IG1) represented more than 50% of cases, whereas the IG3 group, having the best prognosis, accounted for 8.5% of cases only. Interestingly, while most immune checkpoint markers correlated with the different immune groups, CD276 (B7-H3) showed an opposite expression pattern, decreasing from IG1 to IG3. CD276 is a member of the B7 family of immunoregulatory proteins and is overexpressed in several tumor types. It has been shown that CD276 can promote tumor proliferation, angiogenesis, and metastasis, and is associated with shorter survival time in multiple tumor types [113]. A recent study reported a wide immunohistochemical expression of B7-H3 in MPM and demonstrated that PD-L1 and B7-H3 were significantly co-expressed in tumor cells of the non-epithelioid histotype [114]. Similarly, CD44 is the only T-cell exhaustion marker that showed negative correlation with the immune groups [83]. This marker has been associated with metastasis and low survival rates in multiple cancer types [115]. In MPM, CD44 has been shown to promote invasiveness when interacting with hyaluronan [116,117].

V-domain Ig-containing suppressor of T-cell activation (VISTA) is another immune checkpoint that inhibits anti-tumor immune responses (Figure 1k). In a TCGA-based study, VISTA gene expression was reported to be higher in MPM than in all other cancer types. This was particularly observed in the epithelioid subtype and strongly correlated with mesothelin expression [11]. Moreover, VISTA was recently described as a new potential target for mesothelioma immunotherapy. Muller et al. investigated the tissue expression of VISTA and PD-L1 in a large cohort of MPMs. They found frequent expression of VISTA and infrequent expression of PD-L1 (88% and 33% of epithelioid, 90% and 43% of biphasic, and 42% and 75% of sarcomatoid) with favorable and unfavorable survival correlations, respectively [118].

In this context, the expression of STimulator of Interferon Genes (STING) protein is described as having a crucial role in identifying "inflamed" or "hot" tumors that could be successfully treated with immunotherapy (Figure 1l). STING absence implies a tumor environment with no activation of the INFγ pathway, which is a known parameter of response to ICIs [119]. Moreover, it has been reported that targeting DNA damage response promotes anti-tumor immunity through STING-mediated T-cell activation in small-cell lung cancer [120].

#### **3. Angiogenesis**

The prognosis of MPM is best explained by a continuous model, which shows specific expression patterns of genes involved in angiogenesis and immune response [121]. Asbestos fibers have a direct effect on mesothelial cells, causing the release, together with inflammatory cytokines, of vascular endothelial growth factor (VEGF), which attracts leukocytes [122]. VEGF signaling is crucial in MPM pathophysiology [123], regulating blood

vessel function, inducing tumor cell growth, and suppressing immune activation [124]. VEGF also acts as a powerful mitogen for mesothelial cells themselves. Indeed, MPM cell lines secrete VEGF-A and VEGF-C, as well as expressing both VEGF receptors Flt-1 (VEGF-R1) and KDR (VEGFR-2) [125,126]. Thus, VEGF signaling can induce MPM cell growth in an autocrine fashion. This may explain why mesothelioma cells show striking sensitivity to anti-VEGF agents, in addition to the more canonical role of such agents in inhibiting neo-angiogenesis. Moreover, MPM has been shown to produce the highest levels of VEGF among solid tumors [127].

Other growth factors can also regulate migration, survival, and differentiation of endothelial cells, contributing to neoangiogenesis, such as TGFb, EGF, angiogenin, IL-8, and platelet-derived growth factor (PDGF) [128]. All this evidence provides the rationale for the development of VEGF and angiogenesis inhibitors as a therapeutic strategy in MPM [129].

Although there has been over a decade of intense investigation, there are still no validated biomarkers of angiogenesis able to predict the efficacy of anti-angiogenic agents both in MPM and in other cancers [130]. The complementary LUME-Meso biomarker study has reviewed the plasma levels of 58 angiogenic factors and single-nucleotide polymorphisms (SNPs) in genes for VEGFR1 (FLT1), and VEGFR3 (FLT4) and mesothelin (MSLN), and assessed micro-vessel density via CD31 immunohistochemistry on archival biopsy samples. Although PFS and OS benefits were observed in patients with low plasma endoglin and homozygous VEGFR1/3 genotypes, no biomarkers showed any significant and conclusive association with antiangiogenic efficacy [131].

Recently, Chia and coworkers evaluated VEGF, PDGF, FGFR, and CD31 by immunohistochemistry in tissue microarrays from 329 patients who underwent surgical resection or biopsy for MPM. They found that high CD31 density and high PDGF expression levels were associated with poor prognosis in the epithelioid MPM group [132].

#### **4. Conclusions**

TIME is a challenging component with an emerging pathogenic, immunomodulatory, and growth-promoting role in MPM. Given the relatively low mutational burden of MPM, biological events other than genetics may be critical determinants of MPM growth and aggressiveness and could influence cells' immune-escape.

A greater understanding of infiltrating immune cells, their role and function, and the presence of ligand or modulatory marker expression will give a wider and better structured picture of the tumor–immune cell interplay (Figure 2).

A precise pathological and immuno-phenotypical characterization of TIME, in terms of extracellular matrix profiles, subtypes of immune-infiltrating cells, expression of coinhibitory molecules, and activation of immune pathways could provide important knowledge for translational pathology studies. Practical identification of specific biomarkers that could influence the host immunity has to be performed and would represent a major advance for clinical translation of neoantigen-directed immunotherapies, paving the way to understand how to personalize future therapeutic approaches in MPM patients.

**Figure 2.** Graphical representation of tumor immune microenvironment interactions in MPM.

**Author Contributions:** Conceptualization, L.R. and F.N.; methodology, F.N.; software, F.N.; validation, F.N., A.L. and V.Z.; formal analysis, F.N. and G.W.; investigation, F.N.; resources, L.R., M.P., M.V., P.B. and G.S.; data curation, F.N. and L.R.; writing—original draft preparation, F.N.; writing—review and editing, G.W., L.R., M.P., P.B., M.V. and G.S.; visualization, L.R.; supervision, M.V., M.P. and G.S.; project administration, L.R.; funding acquisition, G.S. and P.B. All authors have read and agreed to the published version of the manuscript.

**Funding:** The research plan has received funding from AIRC under IG 2019-ID. 23760 project—to G.S., and from University of Turin, Ricerca Locale (Ex 60%) Funding 2019, to P.B.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


## *Article* **Identification of Redox-Sensitive Transcription Factors as Markers of Malignant Pleural Mesothelioma**

**Martina Schiavello <sup>1</sup> , Elena Gazzano 2,3 , Loredana Bergandi <sup>4</sup> , Francesca Silvagno <sup>4</sup> , Roberta Libener <sup>5</sup> , Chiara Riganti 3,4 and Elisabetta Aldieri 3,4,\***


**Simple Summary:** Malignant pleural mesothelioma is a lung tumor associated with asbestos exposure, with a poor prognosis, and a difficult pharmacological approach. Asbestos exposure is very toxic for the lungs, which counteract this toxic effect by activating some antioxidant defense proteins. When these proteins are more active that in normal conditions, as in several cancers, these tumors become able to survive and resist to stress or chemotherapy. In our laboratory, we collected cellular samples of mesothelioma and non-transformed mesothelium from Hospital's Biobank and we evaluated these proteins. Our results demonstrated these proteins are upregulated in mesothelioma cells and not in nontransformed mesothelium. This event could be associated to toxic effects evoked by asbestos exposure, highlighting the need in the future to monitor asbestos-exposed people by measuring biomarkers identified, in the attempt to identify them as possible predictive markers and potential pharmacological targets addressed to improve mesothelioma prognosis.

**Abstract:** Although asbestos has been banned in most countries around the world, malignant pleural mesothelioma (MPM) is a current problem. MPM is an aggressive tumor with a poor prognosis, so it is crucial to identify new markers in the preventive field. Asbestos exposure induces oxidative stress and its carcinogenesis has been linked to a strong oxidative damage, event counteracted by antioxidant systems at the pulmonary level. The present study has been focused on some redoxsensitive transcription factors that regulate cellular antioxidant defense and are overexpressed in many tumors, such as Nrf2 (Nuclear factor erythroid 2-related factor 2), Ref-1 (Redox effector factor 1), and FOXM1 (Forkhead box protein M1). The research was performed in human mesothelial and MPM cells. Our results have clearly demonstrated an overexpression of Nrf2, Ref-1, and FOXM1 in mesothelioma towards mesothelium, and a consequent activation of downstream genes controlled by these factors, which in turn regulates antioxidant defense. This event is mediated by oxidative free radicals produced when mesothelial cells are exposed to asbestos fibers. We observed an increased expression of Nrf2, Ref-1, and FOXM1 towards untreated cells, confirming asbestos as the mediator of oxidative stress evoked at the mesothelium level. These factors can therefore be considered predictive biomarkers of MPM and potential pharmacological targets in the treatment of this aggressive cancer.

**Keywords:** malignant pleural mesothelioma; mesothelium; oxidative stress; redox-sensitive factors; asbestos; biomarkers

**Citation:** Schiavello, M.; Gazzano, E.; Bergandi, L.; Silvagno, F.; Libener, R.; Riganti, C.; Aldieri, E. Identification of Redox-Sensitive Transcription Factors as Markers of Malignant Pleural Mesothelioma. *Cancers* **2021**, *13*, 1138. https://doi.org/10.3390/ cancers13051138

Academic Editor: Daniel L. Pouliquen

Received: 17 February 2021 Accepted: 3 March 2021 Published: 7 March 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

#### **1. Introduction**

Exposure to asbestos has been clearly associated to the development of lung diseases, among which the most serious is the Malignant Pleural Mesothelioma (MPM), a tumor that originates from the pleura, with an increased incidence throughout the world due to the long latency period, and the direct correlation between asbestos exposure and MPM development is unequivocal [1]. Histologically, three main subtypes of MPM can be distinguished: epithelioid (60–80%), sarcomatoid (<10%), and biphasic or mixed (10–15%) [2]. Although this is a rather rare neoplasm, the incidence is expected to grow over the next few years with a peak between 2020 and 2030 [3], mainly due to the extensive exposure to asbestos fibers in the past years [3]. Most patients are diagnosed at an advanced stage of the disease [4], and for this reason the MPM needs a timely diagnosis and an improvement in the prognosis.

Numerous studies have been focused on trying to clarify the molecular mechanisms underlying the carcinogenesis induced by asbestos, however, some aspects still need to be defined [5]. It is known that asbestos causes chronic inflammation and induces a strong oxidative damage mediated by an increased production of Reactive Oxygen Species (ROS), free radicals that have been shown to be carcinogenetic mediators, by causing DNA mutations and inducing tumor cell proliferation [6]. Several studies have shown that ROS are important second messengers in mediating the toxicity of asbestos [6], especially at the level of the pulmonary mesothelium [7]. Thus, ROS production can modulate different redox-sensitive signal pathways by different transcription factors, in the attempt to counteract the oxidative damage [8]. Among these, a role in carcinogenesis has been shown to be linked to the following redox-sensitive transcription factors: Nuclear factor erythroid 2—related factor 2 (Nrf2 o NFE2L2)/Kelch-like protein ECH-associated protein 1 (KEAP-1) [9], Apurinic-apyrimidinic endonuclease 1 (APE-1)/Redox effector factor 1 (Ref-1) [10] and Forkhead box protein M1 (FOXM1) [11].

The need of these factors in survival of tumor cells, strongly suggests a fundamental role of their activation in carcinogenesis [9–11]. Cancer cells become able to survive against oxidative stress by activating these factors constitutively in different types of tumors (lung, pancreas, breast) [12–14], with increased aggressiveness and resistance to chemotherapy [15], thus up-regulating pro-survival antioxidant responses.

Nrf2 is a redox-sensitive factor belonging to the subfamily cap'n'collar (CNC), containing seven conserved domains (Neh1-7), the latter being involved in the regulation of its stability and transcriptional activity [16]. The intracellular regulator of Nrf2 is KEAP-1, containing 27 cysteines sensitive to oxidative stress: under basal conditions, KEAP-1 degrades Nrf2 by promoting its ubiquitination via proteasome [17]. It has been shown that cancer cells are able to survive against oxidative stress by activating Nrf2 constitutively, and in this way upregulating the antioxidant response in different types of tumors (lung, pancreas, breast, and endometrium), with increased tumor aggression and resistance to chemotherapy [18,19]. Particularly in lung cancer, inactivating somatic mutations on KEAP-1 cysteine residues have been observed, resulting in constitutive activation of Nrf2 [20]. Elevated levels of ROS, by acting on cysteine residues, cause a conformational change of KEAP-1 with the dissociation of the Nrf2/KEAP-1 complex and consequent nuclear translocation of Nrf2, which in turn activates genes that regulate the antioxidant response, such as Mn-Superoxide Dismutase (SOD2) and catalase (CAT), and upregulating the expression of phase II detoxification (glutathione S-transferase, GST) and antioxidant (heme oxygenase 1, HO-1) enzymes [18,19], thus playing a central role in cellular antioxidant defense [20]. Moreover, ROS increase induces the phosphorylation of Nrf2 at the N-terminal region, resulting in a further detachment from KEAP-1 and translocation of the transcription factor from the cytoplasm into the nucleus [21]. Nrf2 is active against oxidative stress when phosphorylated by different kinases, such as MAPK (Mitogen-activated protein kinase)/Erk (Extracellular signal-regulated kinase), PKC (Protein kinase C), and PI3K (Phosphoinositide 3-kinase) at the level of serine and threonine residues, by breaking the binding with the KEAP-1 inhibitor and thus translocating into the nucleus [21].

APE-1/Ref-1 is a multifunctional enzyme involved, respectively, in DNA repair and cellular redox regulation. The two main activities are encoded by two distinct regions of the protein: N-terminal region controls the redox function and C-terminal region checks the DNA repair [10]. Redox-sensitive factor Ref-1, when activated, induces in turn various transcription factors, among which the Nuclear Factor kappa B (NF-kB), the Activator Protein-1 (AP-1) [10], both involved in redox cellular control, and the Hypoxia-Inducible Factor 1 α (HIF-1α), and modulates some tumor suppressors, such as p53 and PTEN (Phosphatase and tensin homolog) [22]. It is known that DNA oxidative damage accelerates cancer development: ROS has been shown to activate the overexpression of Ref-1 with consequent increase in endonuclease activity [22]. As Nrf2, Ref-1 results to be overexpressed in various types of tumors, with increased resistance to antineoplastic therapies [23]: some studies showed an increased expression of Ref-1 in non-small cell lung cancer (NSCLC) with consequent resistance to cisplatin treatment [23], and in knock-down mice there is a significant improvement against the cytotoxic response to drugs [24].

FOXM1 is a transcription factor of the Forkhead box (FoxO) protein superfamily [25]. Unlike FoxO transcription factors, which are activated in quiescent cells and inhibit cell proliferation, FOXM1 is only expressed in proliferating cells and has critical functions in cell-cycle progression [25,26]. Expression of FOXM1 is induced by increased oncogenic stress requiring ROS, and the upregulated FOXM1 counteracts elevated intracellular ROS levels by stimulating the expression of antioxidant enzyme genes to protect tumor cells from oxidative stress [27], such as those involved in the antioxidant system. It has been demonstrated that elevated FOXM1 downregulates ROS levels by stimulating the expression of ROS scavenger genes, such as *SOD2* and *CAT* [27]. As Nrf2 and Ref-1, FOXM1 is overexpressed in different human cancers [28], particularly in lung cancers, and resulted activated by oncogenic pathways, such as those mediated by the axis Ras/MAPK/Erk [26]: induction of FOXM1 by oncogenic Ras requires ROS increase [27], so stimulating FOXM1 nuclear translocation via MAPK/Erk and thus promoting the transcriptional activity of FOXM1 [29].

In this context, our study has been addressed to clarify the correlation between oxidative stress, asbestos and the development of mesothelioma, going to investigate the involvement of all these factors associated to the antioxidant response at a diagnostic and therapeutic level. However, although there are some evidence in literature that demonstrate the overexpression of Nrf2, Ref-1, and FOXM1 in MPM, a close correlation between the pro-oxidant effects exerted by asbestos and these factors, in association to the development of mesothelioma, has not yet been clearly demonstrated. Actually, speaking of asbestos, it should be noted that asbestos includes six different types of fibers [30], among which the most pathogenic in inducing MPM are the iron-containing fibers crocidolite and amosite [31], in particular the crocidolite asbestos (used in this work) has been demonstrated to be the most carcinogenic asbestos fiber [31]. Recent evidence of activation of Nrf2, caused by exposure to asbestos, is reported in murine peritoneal macrophages, in which the use of Nrf2 inhibitory molecules showed an increased apoptosis of tumor cells [32], while other studies in human mesothelioma cell lines showed the involvement of the antioxidant role of Nrf2 in resistance to chemotherapy [33] or in improving therapeutic approach against MPM [34]. Moreover, a proteomic analysis identified Nrf2 as one of the proteins more expressed on biphasic MPM [35] and experiments in human mesothelioma MSTO-211H cells demonstrated Nrf2 overexpression via ROS induction [36], although not in association with asbestos exposure. Concerning Ref-1, Flaherty et al. [37] demonstrated an increased Ref-1 activity after crocidolite asbestos incubation in human alveolar macrophages, as already previously shown in rat pleural mesothelial cells by Fung et al. [38], but, until now, no clear evidence has been associated to MPM. Finally, in recent literature, the role of FOXM1 in association to MPM, particularly by considering the emerging role of FOXM1 as hallmark in many tumors is emerging [28], has been studied. Cunniff et al. [39,40] demonstrated a link between FOXM1 expression and the mitochondrial oxidant metabolism in mesothelioma cell lines, Mizuno et al. [41] showed a direct

regulation of FOXM1 transcription in mesothelioma cells by YAP (Yes-associated protein) oncogenic protein, and Romagnoli et al. [42] identified, by gene expression analysis, FOXM1 as a potential target for novel therapies against mesothelioma. Nevertheless, until now, no link has been shown to correlate FOXM1 overexpression to primary asbestos exposure.

In literature, the characterization of new markers, potentially useful in the diagnosis and therapy of asbestos-related diseases, is becoming increasingly important. In recent years, some molecules such as Mesothelin [5] and BAP1 (BRCA1 associated protein-1) [43] have had special relevance and now are used in MPM diagnosis. Moreover, also the High Mobility Group Box 1 (HMGB1), mediator of pulmonary inflammation, has been detected at high level in the serum of patients exposed to asbestos compared to those not exposed [4,5]. Notably, by examining The Cancer Genome Atlas (TCGA) and Genomic Data Commons (GDC) datasets concerning MPM patients analyzed and eventual Nrf2, Ref-1, and FOXM1 prognostic values, the results showed, out of 87 MPM samples analyzed, that none of the three proposed transcription factors have been analyzed up to now, although in lung cancer they have already been identified and quite associated with a worse prognosis. However, markers as Mesothelin or BAP1 are not able to provide an early diagnosis of MPM. We therefore evaluated the possible involvement of the above mentioned redox-sensitive transcription factors in MPM development in correlation to crocidolite asbestos exposure, analyzing the expression of these factors in human mesothelial and mesothelioma cells, notably the last ones derived from asbestos exposed MPM patients. This is a crucial point aimed to identify these redox-sensitive transcription factors as predictive markers for this aggressive cancer.

#### **2. Results**

#### *2.1. Nrf2, Ref-1, and FOXM1 Are Overexpressed in MPM Cells*

We evaluated the expression of Nrf2, Ref-1, and FOXM1 in human mesothelial cells (HMC) and MPM cells. Our results showed clearly an increased basal expression of the redox-sensitive transcription factors in all three histological types of MPM, epithelioid (EMM), sarcomatoid (SMM), and biphasic (BMM) forms, towards HMC (Figure 1A,B). As documented in literature, we used NSCLC cells (A549) as positive control of the basal overexpression of these factors in lung tumor cells.

#### *2.2. Nrf2 Phosphorylation in MPM Cells Mediates its Nuclear Translocation*

ROS increase induces the phosphorylation of Nrf2 in the N-terminal region [21]. We evaluated the presence of the phosphorylated form of Nrf2 (p-Nrf2) in nuclear extracts of HMC and MPM (EMM, SMM, BMM) cells, and in A549 cell line, used as positive control of basal Nrf2 phosphorylation. As shown in Figure 2A,B, the presence of the phosphorylated form of Nrf2 in all histological types of MPM cells unless the mesothelium demonstrated the activation of Nrf2 via its phosphorylation, as the mechanism which drives and activates Nrf2.

#### *2.3. Increased Antioxidant Target Genes Induced by by Nrf2, Ref-1, and FOXM1 in MPM Cells*

Nrf2 activation drives the transcription and induction of some target genes involved in the antioxidant response, some of these already associated to asbestos exposure [44]. We demonstrated an increased expression of SOD2, GST, CAT, and HO-1 proteins in MPM cells towards HMC, as shown in Figure 3A,B.

**Figure 1.** Nrf2, Ref-1, and FOXM1 overexpression in MPM cells. (**A**) Western blot analysis of Nrf2, Ref-1, FOXM1, and TBP proteins on nuclear extracts of HMC, EMM, SMM, BMM, and A549 cells. (**B**) Densitometric analysis of the expression levels of Nrf2 (*n* = 3, \* *p* < 0.001), Ref-1 (*n* = 3, \* *p* < 0.001) and FOXM1 (*n* = 3, \* *p* < 0.001).

**Figure 2.** Phospho-Nrf2 overexpression in MPM cells. (**A**) Western Blot analysis of phosphorylated Nrf2 (p-Nrf2) and TBP proteins on nuclear extracts of HMC, EMM, SMM, BMM, and A549 cells. (**B**) Densitometric analysis of the relative expression of p-Nrf2 (*n* = 3, \* *p* < 0.001).

**Figure 3.** Expression of antioxidant genes induced by Nrf2 and FOXM1 in MPM cells. (**A**) Western Blot of SOD2, GST, CAT, HO-1, and Tubulin proteins in HMC, EMM, SMM, and BMM cells. (**B**) Densitometric analysis of the relative expression of SOD2, GST, CAT, and HO-1 (*n* = 3, \* *p* < 0.001).

As Nrf2, also FOXM1 activated the antioxidant proteins SOD2 and CAT in MPM cells towards HMC (Figure 3A,B), so counteracting oxidative stress in tumor cells.

Ref-1, when activated, still controls some target genes involved in the antioxidant response, such as NF-kB. Our results demonstrated an increased nuclear accumulation of p50 active subunit of NF-kB in MPM cells towards HMC (Figure 4A,B). Among Ref-1 related controlled genes, the tumor suppressors p53 and PTEN are crucial in cancer suppression when expressed at nuclear level. So, in our experimental models, both p53 and PTEN are significantly expressed in the cytosol of MPM cells in comparison to HMC (Figure 4C,D), thus both not working as tumor suppressors at nuclear level.

At the same time, we evaluated p53 and PTEN at nuclear level: the results evidentiated a partially not so significative downregulation of PTEN and p53 proteins in MPM cells towards HMC (Figure S1), although both resulted partially decreased in MPM cells.

#### *2.4. Phosphorylation of Erk Mediates Nrf2 Phosphorylation and FOXM1 Overexpression*

Nrf2 phosphorylation has been demonstrated to be mediated by different kinases, among which the MAPK/Erk pathway is one of the main involved [21]. Besides, ERK phosphorylation has been widely documented in mesothelial cells exposed to crocidolite asbestos and in MPM cells [45]. Our results show an increased active phosphorylated form of Erk (p-Erk) in all three histological types of MPM cells and not in HMC (Figure 5A,B).

**Figure 4.** Expression of genes induced by Ref-1 in MPM cells. (**A**) Western Blot of nuclear p50 active subunit of NF-kB and TBP protein in HMC, EMM, SMM, and BMM cells, and (**B**) the relative densitometric analysis (*n* = 3, \* *p* < 0.001). (**C**) Western Blot of cytosolic p53, PTEN and Tubulin proteins in HMC, EMM, SMM, and BMM cells, and (**D**) the relative densitometric analysis (*n* = 3, \* *p* < 0.001).

**Figure 5.** Erk phosphorylation mediates Nrf2 and FOXM1 activation. (**A**) Western Blot of phosho-Erk (p-Erk), Erk (1,2) and Tubulin proteins in HMC, EMM, SMM, and BMM cells. (**B**) Densitometric analysis of the relative expression of p-Erk versus Erk (*n* = 3, \* *p* < 0.001).

Several mechanisms have been proposed to explain the activity of FOXM1 in cancer progression, including the activation of this factor by several oncogenic protein and signaling pathways, such as Ras and MAPK/Erk [29]. As for Nrf2, our results demonstrated an overexpression of the p-Erk in MPM cells (Figure 5A,B) and not in mesothelial cells.

#### *2.5. Increased Expression of Nrf2, Ref-1, and FOXM1 after Crocidolite Asbestos Exposure in Mesothelial Cells*

Crocidolite asbestos (the most carcinogenic variant of asbestos fibers) exposure, as well known in literature, is strictly associated to the development of cellular oxidative stress, induced both by fibers themselves and generated by pulmonary cells, particularly at the mesothelium level, in response to asbestos exposure [46].

We already demonstrated that in HMC incubated with crocidolite asbestos fibers there is a strong induction of an oxidative stress, via a significant increase in ROS production, event completely reverted by antioxidants co-incubation [47]. In our experimental model, as expected, HMC incubated with crocidolite asbestos showed an increased significant expression of Nrf2, Ref-1 and FOXM1 compared to untreated cells, in a dose dependent manner (Figure 6A,B).

**Figure 6.** Increased expression of Nrf2, Ref-1 and FOXM1 after crocidolite asbestos exposure. (**A**) Western Blot of nuclear extracts of Nrf2, Ref-1 and FOXM1 from HMC untreated (-) or treated (+) for 24 h with crocidolite (Croc) asbestos (Croc 1: 1 µg/cm<sup>2</sup> ; Croc 2: 5 µg/cm<sup>2</sup> Croc 3: 10 µg/cm<sup>2</sup> ; Croc 4: 25 µg/cm<sup>2</sup> ). (**B**) Densitometric analysis of the relative expression of Nrf2 (*n* = 3, \* *p* < 0.001), Ref-1 (*n* = 3, \* *p* < 0.001) and FOXM1 (*n* = 3, \* *p* < 0.001), respectively.

To confirm our results, we also performed some experiments by incubating HMC with an inert, nonpathogenic monodispersed synthetic amorphous silica, made up of spheres (MSS): results demonstrated clearly that Nrf2, Ref-1 and FOXM1 are overexpressed only when incubated with crocidolite asbestos and not after MSS exposure (Figure S2).

Furthermore, to correlate Nrf2, Ref-1 and FOXM1 overexpression, evoked by asbestos exposure, to MPM development, we measured the basal ROS level in HMC and MPM cells. The results (Figure S3) showed a significant lower level of ROS in MPM cells than in HMC, thus confirming that the hyper-activation of these redox-sensitive transcription factors in MPM is crucial in mediating MPM development and promoting mesothelioma resistance against oxidative stress.

#### **3. Discussion**

Malignant mesothelioma is a tumor with a poor prognosis and, to date, the only therapeutic approach remains surgical excision and chemotherapy, although the latter is not so effective, and the survival is low. There is therefore growing interest in identifying more precise and unequivocal methods of investigation and treatment. Above all, the attempt is addressed, on the one hand, to clarify the bio-molecular mechanisms underlying the neoplastic transformation of the mesothelium after asbestos exposure and, on the other hand, to identify new and more specific predictive and diagnostic markers for this aggressive tumor.

Some mechanisms have been clarified with reference to the toxicity of asbestos at the pulmonary level. In particular, both cytotoxicity and genotoxicity have been widely associated with an increased oxidative stress, mediated by the production of ROS, induced by fibers themselves or as a response from the lung to asbestos [48]. Consequently, this increased ROS production at cellular level represents one of the causes underlying the known toxic effects exerted by asbestos in the lung, particularly at mesothelial level, which seek to counteract oxidative stress by inducing antioxidant cellular defense.

In our cellular mesothelial and MPM models, we evaluate three redox-sensitive factors that recently have been demonstrated to be overexpressed in different tumors and strictly involved in antioxidant defense, Nrf2, Ref-1, and FOXM1 [19,22,26]. In comparison to not transformed HMC, Nrf2, Ref-1, and FOXM1 resulted overexpressed in MPM, and this overexpression was confirmed also in NSCLC pulmonary carcinoma (A549 cells). The results obtained clearly show the overexpression of Nrf2, Ref-1, and FOXM1 in all histologic types of MPM cells (epithelioid, sarcomatous, and biphasic) but not in the not transformed mesothelium. Particularly, Nrf2 translocates into the nucleus when phosphorylated by different kinases, such as MAPK/Erk [21]. We have demonstrated clearly the phosphorylation of Erk in MPM cells but not in HMC, thus proposing this molecular mechanism in mediating Nrf2 phosphorylation and activation.

Asbestos fibers exposure induces a strong oxidative stress. Previous results in our lab demonstrated that crocidolite asbestos increased ROS production in HMC, event completely reverted by antioxidants co-incubation [47]. These results have been confirmed in our experimental models, in which HMC cells exposed to crocidolite asbestos showed an increased and significantly activation of Nrf2, Ref-1, and FOXM1, in a dose-dependent manner, in HMC exposed to crocidolite asbestos, consistently with a high ROS production, thus confirming the response to oxidative stress induced by asbestos at the mesothelium level, which could drive MPM development.

Confirming our data, linearity was observed concerning Nrf2 in results proposed by other research groups on immortalized cell lines of mesothelioma, which showed an increased expression of this factor [32,36]. In some tumors, such as lung cancer, Nrf2 is found to be constitutively expressed primarily for mutations affecting the KEAP-1 suppressor [20]. So, in our MPM models, the expression of Nrf2, in mesothelioma, remains to be confirmed if it is associated with possible mutations of KEAP-1. As demonstrated, Nrf2 controls the transcription of many genes involved in the antioxidant response and in cellular ROS detoxification [18,19], by upregulating enzymes such as SOD2, GST, CAT, and HO-1, which, when overexpressed, protect cells to oxidative damage. We demonstrated clearly, in our experimental model, a significant overexpression of SOD2, GST, CAT, and HO-1 in MPM cells towards HMC, thus confirming the increase in antioxidant defense mediated by Nrf2 and a consequent alteration of redox balance, so increasing the survival of cancer cells. In the context of MPM therefore, in which there is a prolonged exposure to asbestos related oxidative stress induction, other studies have shown that an aberrant increase in the antioxidant systems, mediated by Nrf2 overexpression, may have a role in promoting tumorigenicity and chemoresistance [49], supporting the importance of this factor as a possible pharmacological target in many types of cancer [19].

Ref-1 still counteracts oxidative stress by activating a series of related factors [10], such as NF-kB. We demonstrated the p50 active subunit of NF-kB is overexpressed in MPM cells,

thus enhancing antioxidant system against oxidative stress. This NF-kB upregulation in turn regulates p53 and PTEN oncosuppressors. In our cellular models, p53 and PTEN were overexpressed into the cytosol, but not in the nucleus, thus avoiding their role as tumor suppressors. Although p53 is considered a "guardian of the cell cycle" and is changed in many tumors, in the results obtained there is a confirmation of this event in MPM. However, from the literature, it emerges that the p53 mutation is present, although rare, in mesothelioma [50,51], and a similar point of view concerns PTEN, which has still not been well clarified in MPM [51], but it has been already demonstrated to be inactive in many tumors. However, previous studies have clarified that PTEN expression is not related to a better prognosis in patients with mesothelioma and its expression decreases with chemotherapeutic treatments [52].

FOXM1 mediates antioxidant defense via a dual mechanism. It can modulate the transcription of some genes involved in redox regulation, such as *SOD2* and *CAT* [27], via its induction by the active phosphorylated form of Erk, which in turn could be regulated by ROS increase [26,29]. As for Nrf2, our results demonstrated an overexpression of SOD2 and CAT proteins in MPM cells and not in the mesothelium, thus confirm also for this factor its strong involvement in MPM resistance against oxidative stress and its overexpression in cancer cells. It has been shown FOXM1 nuclear translocation is mediated by MAPK/Erk [29]. As for Nrf2, we demonstrated the mechanism of FOXM1 activation is mediated by Erk phosphorylation, which resulted upregulated in MPM cells and not in HMC. Therefore, from these data, it can be highlighted that there is the same mechanism underlying the activation of Nrf2 and FOXM1, mediated by Erk, and in this way it is possible to elicit a possible synergy or crosstalk between these two factors.

Mutagenesis, a phenomenon initiator of carcinogenesis, reflects DNA damage, which, in cells exposed to asbestos, is mediated by ROS. Therefore, the activation of Nrf2, Ref-1, and FOXM1 can be a key event in maintaining the right balance between apoptosis and carcinogenesis. Several studies have demonstrated the central role of Nrf2 signaling pathways in carcinogenesis and the potential benefit in inducing the inhibition of Nrf2 controlled enzymes [53]. Furthermore, MPM occurs following the accumulation of a series of acquired genetic events, which lead to the deactivation of tumor suppressor genes, by means of a complex cascade mechanism. Ref-1 is therefore necessary for cell survival, and its frequent overexpression in tumor cells strongly suggests a fundamental role of this protein in preventing apoptosis and in controlling cell proliferation. FOXM1, which is variously expressed in many tumors, controls not only the antioxidant defense, but it is widely involved in the control of cell cycle and proliferation [25,26], promoting neoplastic transformation, thus it is can also be rightly considered a possible mediator of MPM development after asbestos exposure.

Chronic oxidative stress and increased ROS production are present at the beginning of an inflammatory response of the mesothelium that involves still the High Mobility Group Box 1 (HMGB1). Until now, numerous studies have shown its relevance in the context of mesothelioma [5]. Our data confirmed an overexpression of this factor in our MPM models compared to the mesothelium (data not shown). This event can be associated to a crosstalk with Nrf2: ROS activates Nrf2 which consequently induces the transcription of antioxidant genes which in turn block the signaling pathway leading to HMGB1 activation. Therefore, the hyper-functioning antioxidant defenses are such that they cannot stem the emergence of the anti-inflammatory response triggered by HMGB1, exacerbating the molecular picture related to MPM. Moreover, redox-sensitive transcription factors, such as Nrf2, when overexpressed in cancer, contributed to contrast oxidative stress also when induced by chemotherapeutic agents [33,34], thus preserve tumor environment and contribute to make MPM resistant to therapeutic approach.

Redox-sensitive factors have long been studied in many tumors, since numerous studies report an important involvement of oxidative stress in neoplastic diseases. The cellular response to oxidative stress by these factors may therefore be representative of a key molecular mechanism related to the carcinogenic effects of asbestos, particularly

crocidolite asbestos, which could explain the attempt by the mesothelial cells to counteract both oxidative stress and induced ROS production. The mesothelium probably cannot cope with this situation, and for this reason these factors, once deregulated, can probably be the potential "initiators" of the neoplastic process in the development of MPM. A peculiar aspect of asbestos-induced carcinogenesis, however, is the latency time between exposure and clinical manifestation [1]. This aspect can play a double role: on the one hand it could be important in the context of a therapeutic intervention, on the other hand it can become a major obstacle in the use of a mouse model for the study over time of the effects of a continuous exposure to asbestos.

Although there are still many aspects to be clarified, the present study proposes Nrf2, Ref-1, and FOXM1 as potential predictive markers of MPM associated with the primary toxic effect evoked by asbestos fibers at mesothelial level. Since MPM has a poor prognosis and a low survival, it is very crucial to detect new prognostic markers and to propose the use of new pharmacological treatments in the attempt to prevent and counteract this serious disease. Moreover, this aspect is important because there are no currently biomarkers predictive of mesothelioma development in asbestos-exposed people, so these potential predictive biomarkers and possible pharmacological targets are crucial in the fight against MPM, particularly important when foreseeing the growing increase in MPM in the next years.

#### **4. Materials and Methods**

#### *4.1. Chemicals*

Electrophoresis reagents were obtained from Bio-Rad Laboratories (Hercules, CA, USA). The protease inhibitor cocktail set III was obtained from Millipore (Billerica, MA, USA). Unless specified otherwise, all reagents were purchased from Sigma Chemicals Co. (St. Louis, MO, USA).

#### *4.2. Cells*

Primary human mesothelial cells (HMC) were isolated from three patients with pleural fluid secondary to congestive heart failure, with no history of a malignant disease, as detailed previously [54]. In total, nine primary human MPM samples (3 epithelioid MPM, 3 biphasic MPM, 3 sarcomatous MPM) were obtained from diagnostic thoracoscopies (see Table S1). MPM cells were obtained after written informed consent from the Biologic Bank of Malignant Mesothelioma, SS. Antonio e Biagio Hospital (Alessandria, Italy). MPM samples, identified with an Unknown Patient Number (UPN), were used within passage 6. The Ethical Committee of Biological Bank of Mesothelioma, S. Antonio e Biagio Hospital, Alessandria, Italy approved the study (#9/11/2011). HMC and MPM cells were grown in Ham's F10 nutrient mixture medium, supplemented with 10% *v*/*v* fetal bovine serum (FBS, Invitrogen Life Technologies, Carlsbad, CA, USA) and 1% *v*/*v* penicillin-streptomycin (Sigma Chemical Co). Cells were checked for Mycoplasma spp. contamination by PCR every three weeks and contaminated cells were discharged. The mesothelial origin of the isolated cells was confirmed by positive immunostaining, as detailed previously [55], and authenticated by the STR analysis method. Cells were used until passage 6.

The NSCLC cells (A549) were provided by the "Bruno Umbertini" experimental zooprophylactic institute (Brescia, Italy). Cells were grown in RPMI-1640, supplemented with 10% *v*/*v* FBS, and 1% of penicillin and streptomycin.

The plasticware for cell culture was provided by Falcon (Becton Dickinson, Franklin Lakes, NJ, USA).

#### *4.3. Asbestos Fibers*

Crocidolite fibers (from Union for International Cancer Control, UICC) were sonicated (Labsonic sonicator, Hielscher, Teltow, Germany, 100 W, 10 s) before incubation with cell cultures, to dissociate fibers bundles, and allow a better suspension and diffusion of fibers in the culture medium. Crocidolite fibers (at concentrations of 1–5-10–25 µg/cm<sup>2</sup> ) were incubated for 24 h in HMC.

#### *4.4. Western Blot Analysis*

Cytosolic and nuclear extracts were obtained using an Active Motif nuclear extraction kit (Active Motif, La Hulpe, Belgium) according to the manufacturer's instructions. The protein content in the cells was detected using a bicinchoninic acid assay (BCA) kit (Sigma Chemical Co., Saint Louis, MO, USA). Cytosolic and nuclear extracts were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE), transferred to polyvinylidene difluoride (PVDF) membrane sheets (Immobilon-P, Millipore, Billerica, MA) and probed with the required antibody diluted in 0.1% PBS-Tween with 5% nonfat dry milk. After 1 h of incubation, the membranes were washed with 0.1% PBS-Tween and then incubated for 1 h with peroxidase-conjugated sheep anti-mouse or sheep anti-rabbit IgG antibody (Amersham International, Little Chalfont, UK) diluted 1:3000 in 0.1% PBS-Tween with 5% nonfat dry milk. The membranes were washed again with 0.1% PBS-Tween, and proteins were detected by enhanced chemiluminescence (Perkin Elmer, Waltham, MA, USA). Ultrapure water (Millipore, Billerica, MA, USA) was used for all experiments.

Antibodies against Nrf2 and phospho-Nrf2 were purchased from Abcam (Cambridge, UK). Antibodies against Ref-1, FOXM1, p53, PTEN, SOD2, GST, HO-1 tubulin, and TATAbinding protein (TBP) were all provided by Santa Cruz Biotechnology, Inc. (Santa Cruz, CA, USA). The anti-Erk and anti-phospho Erk antibodies were provided by Millipore (Billerica, MA, USA). The anti-p50 antibody was provided by Sigma Chemical Co (St. Louis, MO, USA). Tubulin and TBP were used as loading controls for the cytosol and the nucleus, respectively. Band density was calculated using ImageJ software (http://www.rsb.info.nih. gov.bibliopass.unito.it/ij/, accessed date: 17 February 2021).

#### *4.5. Statistical Analysis*

The results were analyzed by a one-way analysis of variance (ANOVA) and Tukey's test, using GraphPad Prism software (v6.01, San Diego, CA, USA). *p* < 0.05 was considered significant. All data in the text and figures are provided as means ± SD.

#### **5. Conclusions**

Nrf2, Ref-1, and FOXM1 are upregulated in MPM and not in non-transformed mesothelium, presumably as consequence of the toxic effect evoked by asbestos fibers at the mesothelium level. These factors can therefore be considered potential candidates as predictive markers of the development of MPM, particularly important considering asbestosrelated damages that predispose to mesothelioma development.

In conclusion, our results and proposed considerations lay and broaden the foundations for future studies in the context of MPM, a tumor that continues to be a public health problem.

**Supplementary Materials:** The following are available online at https://www.mdpi.com/2072-669 4/13/5/1138/s1, Figure S1: Nuclear expression of PTEN and p53 proteins induced by Ref-1 in MPM cells, Figure S2: Expression of Nrf2, Ref-1 and FOXM1 in HMC, Figure S3: Intracellular ROS levels in all three histological types of MPM, epithelioid (EMM), sarcomatoid (SMM) and biphasic (BMM) forms, towards HMC and Table S1: analysis data on MPM cells obtained from total 9 MPM patients, 3 for each histotype (epithelioid, biphasic, sarcomatous), of the Biological Bank of Mesothelioma (AO Nazionale di Alessandria, Italy).

**Author Contributions:** Conceptualization, E.A.; methodology, M.S. and F.S.; software, L.B.; validation, M.S. and E.G.; formal analysis, M.S. and L.B.; investigation, E.A. and M.S.; resources, R.L.; data curation, M.S. and L.B.; Writing—Original draft preparation, E.A. and M.S.; Writing—Review and editing, E.A. and C.R.; visualization, M.S. and C.R.; supervision, E.A.; project administration, E.A.; funding acquisition, E.A. All authors have read and agreed to the published version of the manuscript. **Funding:** This work was funded by Italian Ministry of University and Research (EX60% Funding 2019 to E.A.), grant "ALDE\_RILO\_19\_01".

**Institutional Review Board Statement:** The study was conducted according to the guidelines of the Declaration of Helsinki and approved by the Ethical Committee of Biological Bank of Mesothelioma, SS. Antonio e Biagio Hospital, Alessandria, Italy (#9/11/2011).

**Informed Consent Statement:** Informed consent was obtained from all subjects involved in the study.

**Data Availability Statement:** All data generated or analyzed during this study are included in this published article.

**Acknowledgments:** We acknowledge Costanzo Costamagna, Department of Oncology, University of Torino, for the technical support.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


## *Article* **Digital Gene Expression Analysis of Epithelioid and Sarcomatoid Mesothelioma Reveals Differences in Immunogenicity**

**Luka Brcic 1,† , Alexander Mathilakathu 2,†, Robert F. H. Walter <sup>2</sup> , Michael Wessolly <sup>2</sup> , Elena Mairinger <sup>2</sup> , Hendrik Beckert <sup>3</sup> , Daniel Kreidt <sup>2</sup> , Julia Steinborn <sup>2</sup> , Thomas Hager <sup>2</sup> , Daniel C. Christoph <sup>4</sup> , Jens Kollmeier <sup>5</sup> , Thomas Mairinger <sup>6</sup> , Jeremias Wohlschlaeger <sup>7</sup> , Kurt Werner Schmid <sup>2</sup> , Sabrina Borchert 2,‡ and Fabian D. Mairinger 2,\* ,‡**


**Simple Summary:** Malignant pleural mesothelioma (MPM) is a rare, biologically extremely aggressive tumor with an infaust prognosis. In this retrospective study, we aimed to assess the role of tumor-infiltrating immune cells and their activity in the respective histologic subtypes. We confirmed a substantial difference between epithelioid and sarcomatoid mesothelioma regarding the host's anti-cancer immune reaction. Whereas antigen processing and presentation to resident cytotoxic T cells as well as phagocytosis is highly affected in sarcomatoid mesothelioma, cell–cell interaction via cytokines seems to be of greater importance in epithelioid cases. Our work reveals the specific role of the immune system within the different histologic subtypes of MPM, providing a more detailed background of their immunogenic potential. This is of great interest regarding therapeutic strategies addressing immunotherapy in mesothelioma.

**Abstract:** Malignant pleural mesothelioma (MPM) is an aggressive malignancy associated with asbestos exposure. Median survival ranges from 14 to 20 months after initial diagnosis. As of November 2020, the FDA approved a combination of immune checkpoint inhibitors after promising intermediate results. Nonetheless, responses remain unsatisfying. Adequate patient stratification to improve response rates is still lacking. This retrospective study analyzed formalin fixed paraffin embedded specimens from a cohort of 22 MPM. Twelve of those samples showed sarcomatoid, ten epithelioid differentiation. Complete follow-up, including radiological assessment of response by modRECIST and time to death, was available with reported deaths of all patients. RNA of all samples was isolated and subjected to digital gene expression pattern analysis. Our study revealed a notable difference between epithelioid and sarcomatoid mesothelioma, showing differential gene expression for 304/698 expressed genes. Whereas antigen processing and presentation to resident cytotoxic T

**Citation:** Brcic, L.; Mathilakathu, A.; Walter, R.F.H.; Wessolly, M.; Mairinger, E.; Beckert, H.; Kreidt, D.; Steinborn, J.; Hager, T.; Christoph, D.C.; et al. Digital Gene Expression Analysis of Epithelioid and Sarcomatoid Mesothelioma Reveals Differences in Immunogenicity. *Cancers* **2021**, *13*, 1761. https:// doi.org/10.3390/cancers13081761

Academic Editors: Daniel L. Pouliquen and Joanna Kopecka

Received: 24 February 2021 Accepted: 2 April 2021 Published: 7 April 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

cells as well as phagocytosis is highly affected in sarcomatoid mesothelioma, cell–cell interaction via cytokines seems to be of greater importance in epithelioid cases. Our work reveals the specific role of the immune system within the different histologic subtypes of MPM, providing a more detailed background of their immunogenic potential. This is of great interest regarding therapeutic strategies including immunotherapy in mesothelioma.

**Keywords:** pleural mesothelioma; gene expression; immunogenicity; sarcomatoid; epithelioid

#### **1. Introduction**

Malignant pleural mesothelioma (MPM) is a rare type of cancer that is heavily associated with asbestos exposure [1,2]. This malignancy originates from the pleural mesothelium and is associated with a bad prognosis. Median survival times range from 14–20 months after initial diagnosis [3–5]. Generally, MPM can be differentiated into three major histologic subtypes, epithelioid (EMM), sarcomatoid (SMM), and biphasic (BMM). EMM accounts for up to 80% of all MPM cases [6]. It has also a more favorable outcome compared with the SMM or BMM, especially when surgery is applied [7]. Though it needs to be noted that epithelioid morphology can differ greatly [6,8], thereby also impacting clinical outcome [9–11]. The sarcomatoid subtype is the least prevalent subtype of mesothelioma (<10%) [8]. SMM is considered to be more aggressive in a clinical setting with a higher tendency of distant metastasis [6,12]. The BMM has a mixed composition of both epithelioid and sarcomatoid histology [8]. It is currently discussed whether a proportion of specific histology in biphasic MPM has a prognostic value [13,14].

As distinct biomarkers are lacking [15], early detection is often impeded, thereby worsening patients' outcome. Unfortunately, only a small fraction of patients is suitable for pleurectomy [16], while most patients are treated with a cisplatin/pemetrexed combination. The treatment may prolong overall survival by 3 months [5]. Meanwhile, patients undergoing palliative care including palliative chemotherapy may have an overall survival of 9 months. Immune checkpoint inhibitors are also used as a treatment option in MPM. These inhibitors target negative regulatory immune checkpoints on immune cells, thereby enhancing a prevalent immune response against the tumor. Single agents (pembrolizumab, a PD-1 inhibitor) have shown increased response rates; however, they have failed to show benefits for progression-free (PFS) or overall survival (OS) [17]. Despite this setback, the Checkmate 743 study revealed a four-month OS benefit (mOS, HR: 0.74, CI: 0.60–0.91, *p*-value: 0.0020) and increased two-year survival rate (41% vs. 27%), when comparing immune checkpoint doublet therapy (ipilimumab and nivolumab) with standard systematic chemotherapy [4]. Nonetheless, responses remain unsatisfying with only marginal improvements compared to the best supportive care [18]. With immune therapy now in the focus of current mesothelioma treatment, a deeper knowledge of the tumor's immunogenic potential may help to improve patient selection for this form of therapy.

Though the immune system is widely recognized for its anti-tumor activity, it plays a dual role in MPM and may also support tumor survival and progression. Inhaled microfibres, which are released during processing, corrosion, and weathering of asbestos, often reside in pleural tissue. Unfortunately, macrophages are unable to decompose them [3]. Over time, the persistent fibers damage adjacent cells, leading to necrosis and potentially triggering an immune response. The resulting chronic inflammatory reaction can induce tumor mutagenesis via release of reactive oxygen species (ROS) [19]. These macrophages, together with other various not-tumor-derived cell types essential for MPM development, constitute the so-called tumor microenvironment (TME) [20]. Three important immune cell types, known to infiltrate MPM, are tumor-associated macrophages (TAMs), T-lymphocytes, and myeloid-derived suppressor cells (MDSCs) [20]. TAMs are generally considered to be the largest subset of cells infiltrating MPM (up to 42%) [21,22]. Non-tissue resident macrophages are attracted to the tumor site via expression of the chemokine CCL2 [23].

Once within the tumor, growth factors expressed by the tumor (M-CSF, IL-34, TGF-b, and IL-10) induce an immunosuppressive macrophage phenotype (M2 macrophages) [23–25]. From a clinical perspective, the immune suppressive effects of macrophages are associated with poor prognosis and resistance to standard chemotherapy [23]. Some studies suggested macrophage-based biomarkers to estimate prognosis and outcome in EMM [26–28]. Despite next-generation sequencing studies identifying few mutations resulting in presented neoepitopes and increased immunogenicity [29], T-lymphocytes are the second biggest fraction of the immune cell infiltrate (20–42%), closely following TAMs [27,30,31]. It is speculated that the neoepitope load is higher than suggested, as chromosomal rearrangements can not be detected by targeted amplicon-based NGS, which are often present in MPM [32]. The infiltrating lymphocytes are mostly CD8-positive cytotoxic T lymphocytes (CTL), as well as CD4 and FoxP3 positive regulatory T cells (Tregs) [22,31]. Strikingly, based on pleural effusions of MPM, regulatory T-cells are less common when compared to other tumor entities [25]. Though high infiltration rates and activity of CTL are observed in MPM [25,33], they display signs of anergy or exhaustion [34]. MDSCs are the smallest fraction of the immune cell infiltrate (up to 9%) [30,35]. These cells are predominately associated with suppression of T-cells via releasing of ROS and PD-L1 expression [35–37]. Furthermore, a higher concentration of MDSCs can be linked to poor prognosis in EMM [27,38]. Based on these findings one can conclude that the majority of acting immune cells at the tumor site are either ineffective or are reprogrammed to support tumor growth and progression. Unfortunately, most studies did not distinguish between EMM and SMM when analyzing tumor immune infiltration or are only based on limited numbers of SMM samples. A recent study showed the infiltration of CD8+ T cells as being twice as high in SMM than in EMM but included only six SMM [39].

The above-mentioned points highlight the importance of the immune system for MPM development and progression and raise the question of how different immunogenicity contributes to the different outcomes between EMM and SMM. Deepening the understanding of the biological background of immune escape mechanisms in those histologic subtypes might carry the potential for new therapeutic approaches and improved clinical management of patients in the future.

#### **2. Materials and Methods**

#### *2.1. Patient Cohort and Experimental Design*

This retrospective study was performed on therapy-naïve, formalin-fixed paraffinembedded samples of 22 patients with MPM treated at the West German Cancer Centre or the West German Lung Centre (Essen, Germany) between 2006 and 2009 and the Helios Klinikum Emil von Behring (Berlin, Germany) between 2002 and 2009. Twelve of those were diagnosed as SMM and 10 as EMM. The diagnosis was confirmed by two experienced pathologists (JWO, KWS), based on the current WHO classification [40]. Patients were staged according to the 2017 UICC/AJCC staging [41]. Inclusion criteria were the availability of sufficient tumor material and a complete set of clinical data concerning follow-up and treatment. All patients received platinum-based chemotherapy. The radiologic response rate was assessed by modified Response Evaluation Criteria in Solid Tumours (modRECIST) [42]. Surveillance for this study was stopped on August 31, 2014. Complete follow-up was available for all patients with reported deaths of all patients. Clinicopathological data of the study cohort are summarized in Table 1.


**Table 1.** Clinicopathological data of the study cohort.

Legend: EMM—epithelioid malignant mesothelioma, SMM—sarcomatoid malignant mesothelioma.

#### *2.2. RNA Isolation and Integrity Assessment*

RNA was purified from 20 µm thick FFPE sections, using the Maxwell RSC RNA FFPE Kit supplied by Promega. Obtained RNA was eluted in 50 µL RNase-free water and stored at −80◦C. Before the assessment, RNA concentration was determined via Qubit Fluorometric Quantification (Thermo Fisher Science, Waltham, MA, USA) undergoing manufacturer's instructions for the RNA broad range assay kit. Ultimately, 200 ng of each sample was processed.

#### *2.3. Digital Gene Expression Analysis*

For evaluation of the RNA expression pattern, the commercially available NanoString PanCancer Immune Profiling Panel including 770 immune-related as well as 30 reference genes was used. All code sets along with experiment reagents were designed and synthesized by NanoString Technologies (Seattle, WA, USA). The post-hybridization processing was performed using the nCounter MAX/FLEX System (NanoString) and cartridges were scanned on the Digital Analyzer (NanoString). Samples were analyzed on the NanoString nCounter PrepStation, using the high-sensitivity program, and cartridges were read at maximum sensitivity (555 FOV).

#### *2.4. NanoString Data Processing*

NanoString data processing was performed with the R statistical programming environment (v4.0.2) using NanoStringNorm [36] and NAPPA package, respectively. Considering the counts obtained for positive control probe sets, raw NanoString counts for each gene were subjected to a technical factorial normalization, carried out by subtracting the mean counts plus two-times standard deviation from the CodeSet inherent negative controls. Afterward, a biological normalization using the geometric mean of all reference genes was carried out. To overcome basal noise, all counts with *p* > 0.05 after one-sided *t*-test versus negative controls plus 2× standard deviations were interpreted as not expressed.

#### *2.5. Statistical Analysis*

Statistical analysis was carried out using the R statistical programming environment V 4.0.2. Prior to exploratory data analysis, the Shapiro–Wilks-test was applied to test

for normal distribution of each dataset for ordinal and metric variables. The resulting dichotomous variables underwent either the Wilcoxon Mann–Whitney rank sum test (nonparametric) or the two-sided student's *t*-test (parametric). For comparison of ordinal variables and factors with more than two groups, either the Kruskal–Wallis test (nonparametric) or ANOVA (parametric) were used to detect group differences.

Double dichotomous contingency tables were analyzed using Fisher's exact test. To test the dependency of ranked parameters with more than two groups the Pearson's Chisquared test was used. Correlations between metrics were tested applying Spearman's rank correlation test as well as Pearson's product-moment correlation testing for linearity.

Basic quality control of run data was performed by mean-vs-variances plotting to find outliers in target or sample level. True differences were calculated by correlation matrices analysis. Pathway analysis is based on the KEGG database and was performed using the "pathview" package of R. Differences were specified by −log2 fold changes between means (if parametric) or medians (if non-parametric) of compared groups. Significant pathway associations were identified by gene set enrichment analysis using the WEB-based GEne SeT AnaLysis Toolkit (WebGestalt) [43–45]. Each run was executed with 1000 permutations. Finally, all associations were ranked according to the false discovery rate (*p* < 0.05).

Due to the multiple statistical tests, the *p*-values were adjusted by using the false discovery rate (FDR). The level of statistical significance was defined as *p* ≤ 0.05 after adjustment.

#### **3. Results**

#### *3.1. Gene Expression Pattern of Immune-Related Genes*

Overall, 304 out of 698 (43.6%) significantly expressed immune-related genes show differential expression between EMM and SMM, indicating an overall difference in interaction with the host's immune system. In particular, 90 of those 304 genes (29.6%) show expression only or in a much stronger manner in SMM compared to EMM cases, whereas 214 targets (70.4%) present with overexpression in EMM. In ranked order, ABCB1, SYCP1 und IFNA7 show most differences between both subtypes, with solid expression levels (between about 500 counts for SYCP1ˆ and up to nearly 3000 counts for IFNA7) in EMM but an absence of expression in SMM, whereas MAPK8, AXL und UBC show gene expression predominantly in sarcomatoid cases.

No differences in infiltration density of CD8+ CTL could be observed (FDR adj. *p* = 0.901). Of note, CD4+ T-cells, as well as CD68+ macrophages, were enriched in the SMM. CD20+ B cells tend to be denser in EMM than in SMM, but the overall expression of MS4A1 (CD20) is only slightly above background (20 vs. 100 counts in median) and the association did not reach statistical significance after adjustment (*p*-value: 0.050; FDR adj. *p*-value: 0.094).

An overview of all differences in gene expression pattern between the two histologic subtypes is illustrated in Figure 1, an overview of all *p*-values and statistical parameters can be found in Table S1.

#### *3.2. Gene Set Enrichment Analysis (GSEA)*

To identify biological background mechanisms (pathways and biological functions/ categories) behind the different expression patterns regarding immune-related genes in EMM and SMM, a Gene Set Enrichment Analysis (GSEA) was performed (Figure 2).

In the SMM mainly the pathways for phagosome, antigen processing and presentation, lysosome, autoimmune thyroid disease, viral myocarditis, Fc gamma R-mediated phagocytosis, Eppstein–Barr virus infection, endocytosis, focal adhesion, and proteoglycans in cancer show the strongest enrichment. On the other hand, cytokine–cytokine receptor interaction, salmonella infection, inflammatory mediator regulation of TRP channels, adrenergic signaling in cardiomyocytes, amoebiasis, African trypanosomiasis, parathyroid hormone synthesis, secretion and action, NF-kappa B signaling pathway, inflammatory bowel disease, and Kaposi sarcoma-associated herpesvirus infection are identified as enriched and thereby potentially activated in EMM.

**Figure 1.** Volcano plot illustrating the differential expression between EMM and SMM. 90 of 304 differentially expressed genes (29.6%) show expression only or in a much stronger manner in SMM (right side) compared to EMM cases, whereas 214 targets (70.4%) present with overexpression in EMM (left side). Red dots indicate highly significant and green dots significant association identified by explorative data analysis using either Wilcoxon Mann–Whitney rank sum test (nonparametric) or the two-sided student's *t*-test (parametric).

**Figure 2.** Gene set enrichment analysis of differential expressed genes between EMM and SMMpresenting an overview of gene sets enriched in SMM (right side, blue bars) and EMM (left side, yellow bars). In the SMM the pathways for phagosome, Fc gamma R-mediated phagocytosis, antigen processing and presentation and proteoglycans in cancer show enrichment. Cytokine–cytokine receptor interaction is enriched and thereby potentially activated in EMM.

Details of the GSEA, including normalized enrichment score, the *p*-value of enrichment, exact targets included in the gene sets, and those differentially regulated, can be found in Table S2.

The main altered/influenced pathways are described in particular in the following paragraphs:

#### 3.2.1. Phagocytosis and Antigen Presentation

All phagocytosis- and antigen-presentation associated signaling pathways, including phagosome (Figure S1), antigen processing and presentation (Figure S2), lysosome, Fc gamma R-mediated phagocytosis (Figure S3), and endocytosis are strongly enriched in SMM. For direct phagocytosis, this includes important factors involved in the phagolysosome, like LAMP or cathepsin β, antigen processing and cross-presentation, like TAP1/2

or MHC I/II molecules, or the cytochrome b558 mediated activation of NADPHoxidase, with strong overexpression of gp91 and p40phox. Furthermore, strong expression levels of most phagocytosis-promoting receptors, including Fc receptors, complement receptors, integrins, toll-like receptors, C-lectin receptors as well as Scavenger receptors, could be shown. Accumulation of CD45 positive cells, as activators of T cell response, expression of the Fcγ receptors FcγIIA and B, and downstream signaling via Src and Syk could be verified. Besides antigen processing via autophagy, the "classic" proteasome-associated mechanism for antigen processing and presentation via TAP1/2, TAPBP, and MHC1 binding showed strong activation on all levels of the MHC I pathway for antigen presentation to CD8+ CTL and KIR+ NK cells. Furthermore, the MHC II pathway, important for antigen presentation to CD4+ helper T-cells via MHC II, is overexpressed in total, including but not limited to Ii, MHC2, SLIP, CTSB/L/S, CLIP, and HLA-DM.

#### 3.2.2. Cell–Cell Interaction and Communication within the Tumor Microenvironment

MPM subtypes show a clear difference in the communication networks used between the tumor cells and/or different immune cell types. This spans biological mechanisms and pathways from cytokine–cytokine receptor interactions over cell–cell interaction via proteoglycans up to differences in focal adhesion (Figures S4 and S5). This could be shown by highly increased expression levels of hyaluronan (HA, including CD44, CD44v3), heparan sulfate proteoglycans (HSPGs, including the integrins α2β1, avβ3 or α5β1 and fibronectin) as well as chondroitin/dermatan sulfate proteoglycans (CSPG/DSPG, including TLR2 and TLR4) (Figure S4).

For cell communication via cytokines, especially γ-chain utilizing class I helical cytokine receptors (IL2RA, IL2RG, IL4R, IL15RA, IL21R, IL7R) and IL4-like receptors (IL3RA, CSF2RB, IL13RA1), significantly elevated gene expression in SMM compared to EMM was shown. In EMM samples, an enrichment of IL6/12-like (IL6R, IL11RA, IL12RB2) and IL1-like receptors (IL1R2, IL1RL2, Il18R1, ST2) could be observed.

On the side of chemokine secretion, markable differences in CXC subfamily member expression was observed, whereas those binding CXCR1 (CXCL1, CXCL5, CXCL6) and CXCR2 (CXCL2, CXCL3, CXCL7) are expressed in EMM and those binding CXCR3 (CXCL9, CXCL10, CXCL11) or CXCR5 (CXCL13) are expressed in SMM (Figure S5).

#### **4. Discussion**

For a long time, tumors have been widely underestimated in their complexity, viewed as a clustering of cancer cells on their own, and not considered in terms of the importance of extracellular signaling and complex interactions in the TME. Since then, extensive research has been conducted on the topic of tumor-associated immune events, revealing their enormous influence on tumor progression. In this study, we have approached MPM as a cancer entity with an especially heterogenous TME, whose composition might also be of prognostic value [46]. Our data analysis revealed numerous factors and pathways involved in the cell cycle progression, presumably acting in a synergistic effect and offering an explanation for the progression of MPM despite therapy.

#### *4.1. Phagocytosis*

Despite the understanding of the decisive role the phagosome pathway plays in cancer, it has not yet been described for MPM. GSEA in our study revealed the following phagocytotic pathways being affected with high significance: phagosome, Fc gamma R-mediated phagocytosis, lysosome, and endocytosis. As the phagosome pathway showed the highest enrichment (2.5), we focused on differences between gene expression of selected SMM and EMM genes in this pathway (Figures S1 and S3). The phagosome pathway is mainly involved in the response of the innate immune defense and includes endocytosis, phagocytosis, phagosome maturation, and the development of the lysosome [47]. Phagocytes (macrophages, granulocytes, or dendritic cells) use their plasma membrane to engulf a large particle (e.g., apoptotic cell or microbes) [47]. Tumor cells are also engulfed by phago-

cytes. The ensuing early endosome fuses with the lysosome into a late endosome, then diffused through the membrane of the phagolysosome. Cathepsins are key acid hydrolases within the lysosome. They are associated with the processes of the lysosome, including the process of antigen presentation [48]. Cathepsins represent the principal effectors of protein catabolism and autophagy and support the increased metabolic needs of proliferating cancer cells [48]. In this study, cathepsin was overexpressed in SMM. Overexpression of cathepsin is associated with poor prognosis [48,49]. LAMPs were also overexpressed in SMM. This family of glycosylated proteins is involved in supporting tumor growth and metastatic spread [50].

Toll-like receptors (TLRs) are involved in the response of the innate immunity, but can also organize several downstream signaling pathways leading to the formation or suppression of cancer cells [51]. Once synthesized, they are translocated to the Golgi complex and subsequently delivered to the plasma or endosomes [51]. Overexpression of TLRs has been reported for several cancers like prostate cancer, neuroblastoma, lung cancer, and ovarian cancer. While in some studies overexpression of TLRs has been associated with more aggressive forms of, e.g., squamous cell carcinoma [52], other studies revealed high expression being indicative of longer survival rates [53]. In our study, in contrast to SMM with increased expression of TLR2 and TLR4, EMM exhibited overexpression of TLR6. TLR6 is suggested to have an anticancer function, as described in the literature for colon cancer [54]. TLR2 and TLR4 have been associated with gastric cancer [55].

The TAP transporter and MHC class I and II molecules are involved in the process of antigen processing and cross-presentation. These are overexpressed in phagocytes of SMM. As these molecules are also involved in antigen processing and presentation, this finding is further discussed in Section 3.2.

#### *4.2. Antigen Processing and Presentation*

Modern immunotherapeutic approaches have already been investigated in clinical trials in MPM [56–58]. One possible explanation for different responses might be in the processing and presentation of tumor-specific epitopes [59,60] important for the activation of tumor-specific T-cells [61]. A complex intracellular pathway is involved in processing these antigenic peptides (Figure S2). It starts with the polyubiquitination of the protein, which is then degraded by the proteasome. We have previously demonstrated strong 20S proteasome expression in MPM [62]. Its function is to remove misfolded/dysfunctional proteins, but high expression might lead to an "overheated" proteasome with deficient antigen processing capabilities. This could explain why the high expression of proteasomal components is associated with worse outcomes in MPM [62]. Translocation of small fragments processed by the proteasome into the endoplasmatic reticulum is performed via the TAP-transporter, a homodimer composed of TAP1 and TAP2 [63]. These peptide fragments bind the HLA class I molecule, and the whole complex is transported to the cell surface where it is recognized by CTL [61]. Classically, three genes (HLA-A, HLA-B, HLA-C) with an ample number of alleles code for the HLA class I molecule, but inferior genes are also known [64]. In the present study, we demonstrated a markable upregulation of gene expression levels of the above-mentioned components in SMM. Elevated CD68 expression levels (higher amount of macrophages) increased the activation of antigenpresentation-associated pathways in macrophages and dendritic cells with simultaneously even levels of CD8+ CTL, and no signs of direct anti-cancer immune aggression (like an expression of perforin or granzymes), implies altered processing of tumor neoantigens. This results in a "last-ditch attempt" of antigen-presenting cells to stimulate cytotoxic lymphocytes and NK cells. Deficiencies of the antigen presentation resulting in immune evasion from CTL are well described in different tumors [65,66]. These include the deficiency of HLA/MHC class I molecules due to point mutations or large deletions, but also mutations in HLA/MHC class I subunits, like β-2 microglobulin [56]. Furthermore, tumors might be capable of regulating HLA/MHC class I expression on an epigenetic level via DNA hypermethylation [67]. Johnsen et al. observed the development of large and persistent

tumors through TAP1-negative parental transformed murine fibroblast cell line. In the case of tumor progression, TAP1-negative cells have been reported to be selection-wise favored over TAP1-positive cells [68]. Already in 1993, Restifo et al. suggested a possible tumor escape mechanism through deficient antigen presentation and processing based on finding of low mRNA levels for LMP-2 and LMP-7 (proteasome subunits) and TAP1 and TAP2 in small lung cell carcinomas [69]. Additional escape mechanisms involving TAP-mutations and cofactors that interact with TAP have been described [63]. The missing potency of cytotoxic T lymphocytes activity against the tumor cells by altered antigen processing and presentation could explain the inhomogenous response rates in the Checkmate 743.

#### *4.3. Proteoglycans in Cancer*

In recent decades, extracellular matrix (ECM) and TME have been recognized as major factors of tumor development and progression. In ECM, many different proteins and molecules are regulating different processes important for carcinogenesis. One of the key players in ECM is fibronectin (FN), which was found to be overexpressed in SMM in this study. FN is a glycoprotein with a central role in tumor cell proliferation, angiogenesis, invasion, and metastasis development, but also in processes involved in tumor evasion of the immune system (for review see [70]). Furthermore, its overexpression in SMM is not surprising, since FN is an important mesenchymal marker, and when found in epithelial malignancies is used as a sign of epithelia-mesenchymal transition (EMT) [71]. Its activation of TGF-β induces a partial EMT phenotype, usually at the invasive front of the epithelial tumors [72]. We have also found increased expression of integrin receptors α5 β1, α2 β1 and αv β3 in SMM in our cohort. Integrins are cell adhesion receptors, and the main receptor for ECM proteins and FN, and therefore also involved in many pro-tumor activities like tumor cell proliferation, metastasis, tumor angiogenesis. Binding between FN and integrins is further enhanced by integrin clustering and interacting with urokinase plasminogen activator receptor (uPAR), also overexpressed in SMM [73,74].

Another overexpressed protein in SMM was CD44. CD44 is a transmembrane glycoprotein and primary receptor through which hyaluronan (HA) activates different intracellular pathways resulting in tumor cell growth, migration, invasion, and angiogenesis [75,76]. HA, the only proteoglycan which is not covalently attached to protein core is related to poor prognosis in breast, colon, and ovarian carcinoma [77], and its presence in tumor stroma is an indication of the more aggressive tumor [78–80]. It has been shown that HA in MPM is overexpressed in intracellular, but also in pleural, fluid [81]. Hanagiri et al. demonstrated that the interaction of HA with CD44 is important for the proliferation and migration of tumor cells in MPM [82]. Interestingly, overexpression of CD44 was not observed in the EMM group.

As previously mentioned, we have also found overexpression of TLR 2 and TLR4, which are receptors for decorin, proteoglycan important for growth control, usually with binding and inactivation of TGF-β [83–85], inhibition of angiogenesis, and inducing of apoptosis through EGFR down-regulation [86]. It has been shown that decorin, through TLR2 and TLR4, induces proinflammatory tumor suppressor programmed cell death 4 (PDCD4), whose degradation is further prevented through the TGF-β1 blockade [87].

Thrombospondin-1, overexpressed in SMM, is a very controversial ECM protein involved in cell survival, migration, invasion, angiogenesis, and inflammation. However, its role is not straightforward and depends on tumor and ECM type. It is regarded as an anti-angiogenic factor, but some studies have reported its angiogenic activity as well [88]. It was described as a pro-adhesive protein but can also decrease the adhesion of tumor cells and promote invasion and metastases [89,90].

Very similar is the role of lumican, keratan sulfate, in cancer. Its expression is correlated with poor outcome in lung carcinoma, and in colorectal carcinoma, but is a favorable prognostic factor for osteosarcoma and melanoma [91–93]. It is known that lumican induces FAS by binding FAS ligands and in this way plays a role in the initiation of apoptosis and suppresses cell proliferation [94–96]. FAS is highly expressed in our EMM cohort. At the same time, TGF-β2, which is involved in growth suppression and cell adhesion in osteosarcoma [97], and is negatively regulated by lumican, has been highly expressed in SMM.

#### *4.4. Secretion of Cytokines and Communication with the Immune System*

To establish themselves and progress properly, it is inevitable for cancer cells to shape their local microenvironment to their benefit. This goal is achieved through continuous inflammatory reactions and heavy modulations of the immune response [98]. With cytokines and out of those especially chemokines being essential mediators for such a process, changes in their expression patterns are of great interest if we are to develop a deeper understanding of MPMs acquired TME. Various ligands, as well as receptors within the CC chemokine subfamily, were overexpressed in both MPM subtypes. This upregulation might support the flourishment of MPM since these chemokines have already been considered to play a vital role in tumor genesis, while their overexpression also appears to modulate the hosts' immune response against cancer cells [99]. We found a more distinguishable expression pattern regarding the CXC chemokine family. The EMM cases overexpress ligands for CXCR1/2, whereas the sarcomatoid subtype appears to stimulate the CXCR3-pathway with CXCL 9–13. Especially, the activation patterns measured in the EMM stick out, as the CXCR1/2 pathways are thought to contribute massively to the development of, among others, prostate, lung, colorectal, and breast cancer, as well as inflammatory diseases such as COPD and asthma [100,101]. Furthermore, malignancies appear to increase their therapy resistance by overexpression of these receptors and their ligands. In fact, the CXCR1/2 axis has been unrevealed as a potential therapeutic target in malignant melanoma, with pathway-inhibition significantly improving sensitivity for chemotherapy in otherwise resistant melanoma cells in vitro [102], while also decreasing progression and metastasis even in advanced disease [103].

Interleukins are considered to play a key role in MPM development. It was shown that asbestos-exposed knockout mice bearing modified inflammasomes, resulting in a diminished IL-1β release, had a significantly reduced incidence of MPM and later disease onset compared to their wild-type counterparts [104]. Furthermore, IL-6 is thought to not only essentially contribute to MPMs asbestos-related development, but also to impede effective chemotherapy and inducing angiogenesis by increasing VEGF expression [105,106]. In our study, the SMM demonstrated a surprisingly broad spectrum of elevated receptor expressions throughout interleukin 2-, as well as interleukin 4-like receptors. Interestingly, both subtypes, epithelioid via receptor-, sarcomatoid via ligand-upregulation, heavily stimulate the IL-6R pathway.

Especially the recruitment of TAMs has already been considered as a promising therapeutic target in MPM [107]. This hypothesis is further substantiated by Blondy et al., who discovered that MPM cells are directly involved in the recruitment of immunosuppressive macrophages by stimulation of the M-CSF/IL-34/CSF-1R pathway [108]. This perfectly fits the above-mentioned narrative since we were also able to demonstrate an elevated expression of mentioned pathways in our GSEA. Moreover, particularly the SMM upregulates the production of TNF- related TWEAK and TRAIL, as well as TGF-β related ligands TGFB-1 and -2. While the role of TNF has already been established in various malignant processes [109], TGF-ß has even been unraveled as an essential factor in MPM genesis [110,111].

An interesting thought occurred while regarding our expression patterns in the light of modern therapeutic approaches. In a recent study, Horn et al. demonstrated improved immune response and prognostically favorable TME remodeling of breast and lung cancer in a murine model after simultaneous inhibition of the CXCR1/2 and TGF-ß pathway during PDL-1 therapy [112]. As PDL-1 treatment in combination with a cisplatin-pemetrexed based chemotherapy [4,113] has yielded relatively promising results in MPM therapy so far, and with us showing increased activation of the corresponding pathways, transferring this experimental approach to the MPM might be important for future multimodal treatment.

Our study has several technical and biological limitations. As the present study is only based on gene expression data, the final proof of differences in the composition and quantity of the infiltrating immune cells, and chemokine secretion described above is lacking. Furthermore, the relatively small sample sizes of SMM and EMM reduce study strength, as variability, especially between samples of different ethnical origins, may be underestimated. Furthermore, it would be of great interest to analyze the expression of genes involved in innate and acquired immunity in normal mesothelium and compare these findings with EMM and SMM. It is known that normal mesothelial cells form a protective barrier, and are involved in antigen presentation, inflammation, and cell adhesion [114,115]. However, normal pleural tissue from healthy patients can only rarely be provided, which makes it more difficult to characterize a "normal" state and define EMM and SMM-specific features.

#### **5. Conclusions**

Immune evasion as a hallmark of cancer and both in EMM and SMM can be a problematic issue for therapeutic intervention. Our work reveals the specific gene expression pattern of genes involved in immunological and inflammatory processes within the different histologic subtypes of MPM, providing a more detailed background of their immunogenic potential and demonstrating their distinct pattern of immunogenicity. Those differences comprise genes associated with antigen processing and presentation to resident cytotoxic T cells as well as phagocytosis, but also cell–cell communication via the cytokine system. Knowledge about underlying biological processes has the potential to pave the ground for patient stratification for modern therapeutic approaches such as immune-checkpoint blockades and will be the key for improved clinical management of patients with MPM.

**Supplementary Materials:** The following are available online at https://www.mdpi.com/article/ 10.3390/cancers13081761/s1, Figure S1: Gene set enrichment analysis of differential expressed genes between EMM and SMM involved in phagosome pathway, Figure S2: Gene set enrichment analysis of differential expressed genes between EMM and SMM involved in antigen processing and presentation pathway, Figure S3: Gene set enrichment analysis of differential expressed genes between EMM and SMM involved in Fc gamma R-mediated phagocytosis, Figure S4: Gene set enrichment analysis of differential expressed genes for proteoglycans, Figure S5: Gene set enrichment analysis of differential expressed genes between EMM and SMM involved in cytokine-cytokine receptor interaction, Table S1: Overview of *p*-values and 95% CI of associations to histological subtype calculated for all genes, Table S2: Results of the gene set enrichment analysis between EMM and SMM.

**Author Contributions:** Conceptualization, L.B., S.B. and F.D.M.; Methodology, F.D.M. and R.F.H.W.; Software, M.W., F.D.M.; Validation, R.F.H.W., S.B. and F.D.M.; Formal analysis, R.F.H.W., M.W., and F.D.M.; Investigation, A.M., S.B., J.W. and F.D.M.; Resources, F.D.M., L.B. and K.W.S.; Data curation, A.M., L.B., M.W., E.M., H.B., D.K., J.S., T.H., D.C.C., J.K., T.M. and F.D.M.; Writing—original draft preparation, R.F.H.W., L.B., A.M., M.W. and F.D.M.; Writing—review and editing, R.F.H.W., L.B., A.M., M.W., E.M., A.M., H.B., D.K., L.B., J.S., T.H., D.C.C., J.K., T.M., J.W., S.B., K.W.S., F.D.M.; Visualization, M.W., and F.D.M.; Supervision, F.D.M.; Project administration, L.B. and F.D.M.; Funding acquisition, F.D.M. and K.W.S. (institutional). All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Institutional Review Board Statement:** The study was conducted according to the guidelines of the Declaration of Helsinki and approved by the Ethics Committee of University Hospital Essen (protocol code 14-5775-BO).

**Informed Consent Statement:** As the majority of patients were deceased at the time of data collection and collection of follow-up data, a written informed consent has not been obtained from them. All patient data have been anonymized prior to analysis. The ethics committee of the University Hospital Essen waived the necessity for a written informed consent when they approved the present study.

**Data Availability Statement:** All data are available from the author directly.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


## *Article* **Response to Immune Checkpoint Inhibitor Therapy in Patients with Unresectable Recurrent Malignant Pleural Mesothelioma Shown by FDG-PET and CT**

**Kazuhiro Kitajima 1,\*, Mitsunari Maruyama <sup>1</sup> , Hiroyuki Yokoyama <sup>1</sup> , Toshiyuki Minami <sup>2</sup> , Takashi Yokoi <sup>2</sup> , Akifumi Nakamura <sup>3</sup> , Masaki Hashimoto <sup>3</sup> , Nobuyuki Kondo <sup>3</sup> , Kozo Kuribayashi <sup>2</sup> , Takashi Kijima <sup>2</sup> , Seiki Hasegawa <sup>3</sup> and Koichiro Yamakado <sup>1</sup>**


**Simple Summary:** This is the first known study to compare three FDG-PET/CT criteria (EORTC, PERCIST, imPERCIST) with CT criteria (combined modified RECIST and RECIST 1.1) used to evaluate tumor response to ICI therapy in patients with recurrent MPM as well as prediction of prognosis. All of the FDG-PET/CT and CT criteria analyzed were found to be accurate for both evaluation of tumor response and prediction of progression free survival in the present cohort. In comparison with CT, all three FDG-PET/CT criteria judged a greater percentage of patients (16.7%) as CR, while two (EORTC, PERCIST) judged a greater percentage (10–13.3%) as PD.

**Abstract:** Background: To compare three FDG-PET criteria (EORTC, PERCIST, imPERCIST) with CT criteria (combined modified RECIST and RECIST 1.1) for response evaluation and prognosis prediction in patients with recurrent MPM treated with ICI monotherapy. Methods: Thirty MPM patients underwent FDG-PET/CT and contrast-enhanced CT at the baseline and during nivolumab therapy (median 10 cycles). Therapeutic response was evaluated according to EORTC, PERCIST, imPERCIST, and CT criteria. PFS and OS were examined using log-rank and Cox methods. Results: CMR/PMR/SMD/PMD numbered 5/3/4/18 for EORTC, 5/1/7/17 for PERCIST, and 5/3/9/13 for imPERCIST. With CT, CR/PR/SD/PD numbered 0/6/10/14. There was high concordance between EORTC and PERCIST (κ = 0.911), and PERCIST and imPERCIST (κ = 0.826), while that between EORTC and imPERCIST (κ = 0.746) was substantial, and between CT and the three PET criteria moderate (κ = 0.516–0.544). After median 14.9 months, 26 patients showed progression and nine died. According to both PET and CT findings, patients with no progression (CMR/PMR/SMD or CR/PR/SD) showed significantly longer PFS and somewhat longer OS than PMD and PD patients (EORTC *p* = 0.0004 and *p* = 0.055, respectively; PERCIST *p* = 0.0003 and *p* = 0.052; imPERCIST *p* < 0.0001 and *p* = 0.089; CT criteria *p* = 0.0015 and *p* = 0.056). Conclusions: Both FDG-PET and CT criteria are accurate for response evaluation of ICI therapy and prediction of MPM prognosis. In comparison with CT, all three FDG-PET/CT criteria judged a greater percentage of patients (16.7%) as CMR, while two (EORTC, PERCIST) judged a greater percentage (10–13.3%) as PMD. For predicting PFS, the three FDG-PET criteria were superior to the CT criteria, and imPERCIST demonstrated the highest rate of accurate prediction.

**Keywords:** mesothelioma; immunotherapy; therapy response; survival; FDG; PET-CT

**Citation:** Kitajima, K.; Maruyama, M.; Yokoyama, H.; Minami, T.; Yokoi, T.; Nakamura, A.; Hashimoto, M.; Kondo, N.; Kuribayashi, K.; Kijima, T.; et al. Response to Immune Checkpoint Inhibitor Therapy in Patients with Unresectable Recurrent Malignant Pleural Mesothelioma Shown by FDG-PET and CT. *Cancers* **2021**, *13*, 1098. https://doi.org/ 10.3390/cancers13051098

Academic Editors: Daniel L. Pouliquen and Joanna Kopecka

Received: 3 February 2021 Accepted: 1 March 2021 Published: 4 March 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

#### **1. Introduction**

Individuals affected by malignant pleural mesothelioma (MPM), a rare type of aggressive malignancy, have a poor prognosis. Platinum-based chemotherapy has been commonly used as the standard first-line treatment in unresectable MPM cases, though few other treatment options are available for those not showing response. However, a paradigm shift has occurred in recent years because of development of immune checkpoint inhibitors (ICIs), and several groups have reported survival benefits for patients with recurrent MPM [1–5]. Those include a single-arm phase II study conducted in Japan (MERIT study) that examined nivolumab (anti-PD-1 monoclonal antibody) monotherapy for efficacy and safety in 34 MPM patients with a history of chemotherapy, with their findings leading to approval of nivolumab for unresectable recurrent MPM treatment in Japan [3].

A crucial factor for effective cancer treatment management is adequate assessment of systemic treatment response, with efficient monitoring of responsiveness to systemic therapy by the tumor vital for moderating the high risk of mortality and also cytotoxic effects associated with systemic therapeutic regimens. Classic methods have been developed for examining patients undergoing cytotoxic chemotherapy and given molecular targeted agents are used for evaluation of treatment response, such as the Response Evaluation Criteria in Solid Tumors version 1.1 (RECIST 1.1) [6] for computed tomography (CT), and the European Organization for Research and Treatment of Cancer (EORTC) criteria [7] and Positron Emission Tomography Response Criteria in Solid Tumors (PERCIST) [8] for [ <sup>18</sup>F]fluorodeoxyglucose positron emission tomography/computed tomography (FDG-PET/CT), as those treatments can directly result in reduced tumor cell viability. However, immunotherapy differs from classical cytotoxic drugs in regard to the action mechanism, as that mechanism of the former is based on stimulation of host immune response against cancer cells, possibly resulting in inflammation development at the tumor site, leading to a subsequent antitumor response [9].

ICI therapeutic efficacy is difficult to assess and the role of FDG-PET has not yet been established. An increase in FDG uptake or appearance of new lesions following therapy may represent infiltration of cancer foci by host immune cells (pseudo-progression) rather than true tumor progression, thus making evaluation of treatment response using FDG-PET/CT results challenging. As a result, another group recently proposed immunotherapymodified PERCIST (imPERCIST) findings for this evaluation, in which new lesions are not considered to define progressive metabolic disease (PMD) during the early period of assessment (2–4 cycles) of ICI response in metastatic melanoma patients [10].

No other known studies have examined or compared use of FDG-PET/CT and CT for determining MPM patient response to ICI therapy. The present retrospective investigation compared three functional FDG-PET criteria (EORTC, PERCIST, imPERCIST) with morphological CT criteria (combined modified RECIST [11] and RECIST 1.1 [6]) to evaluate response to treatment and predict prognosis in patients with recurrent MPM undergoing nivolumab monotherapy treatment.

#### **2. Materials and Methods**

#### *2.1. Patients*

Approval from a local review board was received for this retrospective study, and the requirement for patient-informed consent was waived. A search of our database was used to obtain the records of patients with unresectable recurrent MPM and treated with nivolumab monotherapy between June 2018 and December 2019. For the present analysis, a total of 30 (mean 68.1 ± 7.2 years old, range 46–77 years) who underwent FDG-PET/CT and contrast-enhanced CT examinations at our institution at the baseline and during nivolumab monotherapy (after 4–6 cycles in 3, 7–9 in 9, 10–12 in 9, 13–15 in 4, 16–18 in 3, 19–21 in 2; median 10 cycles) for treatment response evaluation were included. Baseline FDG-PET/CT and baseline contrast-enhanced CT examinations were conducted at a median 1.0 months (1.0–2.2 months) and 1.4 months (0.7–2.3 months), respectively, before

initiation of nivolumab therapy. The interval of FDG-PET/CT and contrast-enhanced CT was less than two weeks at the baseline and during nivolumab therapy in every patient. Table 1 shows patient and tumor characteristics. CT, FDG-PET/CT, and brain magnetic resonance imaging (MRI) results were used for diagnosis of disease recurrence, metastasis, and progression during the follow-up period. When disease progression or recurrence was suspected on the physical findings, CT or FDG-PET/CT was undertaken for the evaluating the whole-body state, and the brain MRI was carried out for the screening of the brain. In some patients without suspected progression or recurrence, those imaging examinations were undertaken every 6–12 months for surveillance.


**Table 1.** Study population characteristics.

Data are presented as numbers.

Intravenous nivolumab was given at 3 mg/kg every two weeks until apparent disease progression or unacceptable toxicity was observed, or the patient or attending physician decided to discontinue treatment. Of the 30 enrolled patients, treatment-related adverse events were noted in nine (30.0%) (rash in two, hypothyroidism in two, interstitial lung disease in one, increased lipase level in one, diarrhea in one, hypoadrenocorticism in one, fatigue in one). After discontinuing nivolumab treatment, alternative treatment (cisplatin/carboplatin and pemetrexed, pemetrexed, or irinotecan and gemcitabine) was tried.

#### *2.2. FDG-PET/CT*

Four different PET/CT scanners installed at our institution (Gemini GXL16, Gemini TF64, Ingenuity TF: Philips Medical Systems, Eindhoven, The Netherlands; Discovery IQ: GE Healthcare, Waukesha, WI, USA) were used for performing the FDG-PET/CT examinations. Each patient was instructed to fast for five hours before the examination, and blood glucose was measured immediately prior to FDG injection (4.0 MBq/kg body weight for GXL16, 3.0 MBq/kg for TF64, 3.7 MBq/kg body weight for Ingenuity TF and Discovery IQ), with all in the present cohort showing a level lower than 160 mg/dL. Approximately 60 min after the injection, static emission images were obtained. For attenuation correction and anatomic localization, helical CT scan images from the top of

the head to mid-thigh were obtained with the following parameters: tube voltage 120 kV (all four scanners), effective tube current auto-mA up to 120 mA (GXL16), 100 mA (TF64), 155 mA (Ingenuity TF), or 15–390 mA (Smart mA: noise index 25) (Discovery IQ), gantry rotation speed 0.5 s, detector configuration 16 × 1.5 mm (GXL16), 64 × 0.625 mm (TF64 and Ingenuity TF), or 16 × 1.25 mm (Discovery IQ), slice thickness 2 mm, and a transverse field of view 600 mm (GXL16, TF64, Ingenuity TF) or 700 mm (Discovery IQ). Immediately after completion of the CT examination, PET imaging was performed from the head to mid-thigh for 90 s (GXL16, TF64, Ingenuity TF) or 180 s (Discovery IQ) per bed position in three-dimensional mode. The patient was allowed to breathe normally during PET scanning. For the GXL16, attenuation-corrected PET images were reconstructed with a line-of-response row-action maximum likelihood algorithm, while for the TF64 and Ingenuity an ordered-subset expectation maximization (OSEM) iterative reconstruction algorithm (33 subsets, three iterations) was used, and Q.Clear (block sequential regularized expectation maximization (BSREM)) (β = 400) was utilized for the Discovery IQ.

#### *2.3. Contrast-Enhanced CT*

To obtain pre-contrast and contrast-enhanced CT images of the neck, chest, abdomen, and pelvis, a 128-detector row CT (SOMATOM Definition AS: Siemens Healthcare, Erlangen, Germany) was used at 120 kV, with an effective mA of 220 (CAREDose4D), beam pitch of 0.6, collimation of 1.2 × 32 mm, and B31 + medium smooth + image reconstruction. Details regarding the contrast-enhanced CT procedures have been previously presented. Briefly, blood creatinine level determined prior to the examination was ≤1.5 mg/dL in all of the patients. Iodinated contrast material (Iopamiron Inj, Syringe, Bayer Schering Pharma, Berlin, Germany) containing 300 mg of iodine per ml at a dose of 600 mg of iodine per kg of body weight was intravenously administered using a power injector, with scanning started at 120 s after the injection.

#### *2.4. Image Analysis*

A board-certified nuclear medicine expert with 12 years of oncologic FDG-PET/CT experience and without knowledge of the other imaging results, or clinical or histopathologic data for the present patients, retrospectively reviewed the FDG-PET/CT images. To assist the attending clinician with treatment response monitoring, the GI-PET software package (AZE Co., Ltd., Tokyo, Japan), which can harmonize standardized uptake values (SUVs) obtained with different PET/CT systems using phantom data [12], was employed. Maximum SUV (SUVmax) was defined as the maximum concentration in the target lesion (injected dose/body weight). For calculating SUVpeak, a 1.2-cm diameter volume region of interest (ROI) placed on the hottest site of the tumor was used, then normalized to SUV corrected for lean body mass (SULpeak) (SUVpeak × [lean body mass]/[total body mass]).

A board-certified radiologist with 12 years of experience with CT retrospectively evaluated the contrast-enhanced CT images and made determinations, in the absence of knowledge of the other imaging results or clinical data for the present patients. Coronal, axial, and sagittal section images were viewed and analyzed, with appropriate winding applied.

#### *2.5. EORTC*

Using the EORTC guidelines [7], complete resolution of FDG uptake within the tumor volume indistinguishable from surrounding normal tissue was determined as complete metabolic response (CMR), while PMD was the classification for appearance of new FDG uptake in another region in the second FDG-PET/CT scan. The EORTC recommends defining regions of high FDG uptake that represent a viable tumor by use of pre-treatment scan findings and also utilization of the same ROI volumes in subsequent scanning examinations positioned as close to the original tumor as possible, as well as determination of maximal tumor ROI count per pixel per second calibrated as MBq/L [7]. The number of lesions to be measured is not recommended by the EORTC, thus up to five with the highest level of FDG uptake and up to two per organ, with same lesions measured

in subsequent follow-up scan imaging results, were the parameters used in the present study [13]. The values for all five targets used for SUVmax measurement were summed for each scan, resulting in ΣSUVmax. Percentage changes from baseline to second summed SUVmax were calculated, with a reduction of ≥25% in summed SUVmax value defined as partial metabolic response (PMR). PMD was classified as an increase in tumor summed SUVmax value ≥25% within the ROI defined based on the baseline scan, while stable metabolic disease (SMD) was defined as an increase in the summed SUVmax value of <25% or a decrease <25%.

#### *2.6. PERCIST*

For therapeutic response determination according to PERCIST [8], SUL values were calculated using a 1.2-cm diameter volume ROI placed on the target lesion, and SUL values were calculated. Additionally, the SULpeak value of the tumor was determined and noted if it was 1.5 times or more greater than that of the liver SUL (mean ± 2 standard deviations) in a 3-cm diameter spherical ROI on the normal right lobe. When complete resolution of FDG uptake within the target lesion was lower than mean liver activity and indistinguishable from the background blood-pool level, CMR was the classification. For cases with metabolically active lesions noted in follow-up scan findings, the SULpeak of up to five lesions at the baseline and follow-up examinations was summed (maximum two per organ) [8]. The hottest lesions in each scan were selected; thus, the target lesions detected in follow-up imaging were not necessarily the same as those in the baseline images. When the SULpeak sum was decreased by ≥30%, tumor response for that case was classified as PMR. Conversely, an increase in SULpeak sum ≥30% or appearance of new hypermetabolic lesions or ≥75% increase in total lesion glycolysis (TLG) in follow-up FDG PET/CT scan imaging was defined as PMD. Any cases not defined as CMR, PMR, or PMD were classified as SMD.

#### *2.7. imPERCIST*

imPERCIST was performed in the same manner as used for PERCIST, though appearance of new lesions alone did not result in a classification of PMD [10], as that was defined only by increase in sum of SULpeaks of ≥30%. New lesions were included in the SULpeak sum when a higher uptake level than the existing target lesions was shown or when fewer than five target lesions were detected in baseline scan results.

#### *2.8. Combined Modified RECIST and RECIST 1.1*

Pleural tumor thickness perpendicular to the chest wall or mediastinum was measured at two different points at three different levels for evaluations with modified RECIST [11]. For assessing the morphological response of nonplural lesions, RECIST 1.1 was used [6]. The target lesion was defined as a well-defined soft tissue lesion with the longest axis for the lymph node ≥ 1 cm and the shortest axis ≥1.5 cm, and the greatest sum of the diameter of five target lesions, maximum two lesions per organ, and used for evaluation. Sclerotic or lytic/sclerotic (mixed type) bone metastasis was considered to be a non-measurable lesion. With both modified RECIST and RECIST 1.1, a decrease ≥30% in largest diameter sum was considered to be partial response (PR), while progressive disease (PD) was determined in cases with an increase ≥20%. Stable disease (SD) was considered to be any change between PR and PD of <−30% to <+20%; complete response (CR) was determined in cases with disappearance of nonplural target lesions and lymph nodes in the shortest axis <1 cm, and PD when there was appearance of a new lesion. In a comparison of mRECIST and RECIST 1.1 results, the worst objective response was chosen as the final classification shown by CT.

#### *2.9. Statistical Analysis*

Cohen's κ coefficient was used to examine concordance between criteria methods was assessed using [14], with a slight (κ < 0.21), fair (κ = 0.21–0.40), moderate (κ = 0.41–0.60), substantial (κ = 0.61–0.80), or nearly perfect (κ > 0.80) level of agreement noted. Progressionfree survival (PFS) was defined based on time elapsed from start of nivolumab therapy to date of disease progression shown in radiological and/or clinical examination results, or death from any cause. Any patient with no evidence of progressive disease was censored at the date of the last follow-up examination. Time from start of nivolumab therapy until death from any cause was used to determine overall survival (OS). Patients living at the final follow-up examination were censored, and classified as alive with disease or no evidence of progression. The Kaplan–Meier method was used to generate actuarial survival curves, with a log-rank test employed to examine differences between groups. Statistical analyses were performed with the SAS software package, version 9.3 (SAS Institute Inc., Cary, NC, USA), with *p* values < 0.05 considered to be significant.

#### **3. Results**

#### *3.1. Treatment Response Assessment*

Using EORTC criteria with FDG-PET/CT findings resulted in CMR being noted in five patients (16.7%), PMR in three (10.0%), SMD in four (13.3%), and PMD in 18 (60.0%), while use of PERCIST with FDG-PET/CT findings showed CMR in five (16.7%), PMR in one (3.3%), SMD in seven (23.3%), and PMD in 17 (56.7%) patients, respectively, and use of imPERCIST with FDG-PET/CT findings showed CMR in five (16.7%), PMR in three (10.0%), SMD in nine (30.0%), and PMD in 13 (43.3%) patients, respectively. When the combination of modified RECIST and RECIST 1.1 with CT was used, no patients (0%) had CR, six (20.0%) had PR, 10 (33.3%) had SD, and 14 (46.7%) had PD. Figures 1 and 2 present data of two representative cases.

Prior to nivolumab treatment, FDG-PET/CT examinations showed only pleural lesions in 25 patients, while two had pleural and nodal lesions, one had only nodal lesions, one had pleural and lung lesions, and one had pleural, nodal, and peritoneal lesions. Tiny nodal or peritoneal lesions were not detected with contrast-enhanced CT in two patients before starting nivolumab treatment, though those are not included as target lesions in the RECIST criteria due to their small size. The second FDG-PET/CT examination detected new lesions in eight patients (lung metastasis in two; pleural lesions in one; lymph node metastasis in one; bone metastasis in one; small intestine metastasis in one; lymph node and peritoneal dissemination in one; lymph node, peritoneal, bone, and muscle metastasis in one). Of those eight cases with new lesions revealed in the second FDG-PET/CT examination, the CT reader was unable to detect new lesions in three (bone metastasis in one; small intestine metastasis in one; lymph node, peritoneal, bone, and muscle metastasis in one).

#### *3.2. Treatment Response Assessment Comparisons among Criteria Methods*

Twenty-seven (90%) of the cases demonstrated concordance between the EORTC criteria and PERCIST response classifications, while discordance was noted in three (10.0%), with nearly perfect agreement (κ = 0.911) for response classification between them (Table 2). As for EORTC and imPERCIST, concordance between them was seen in 23 (76.7%) cases and discordance was noted in seven (23.3%), with substantial agreement (κ = 0.746) for response classification found between them (Table 3). Furthermore, in 26 (86.7%) cases, concordance between PERCIST and imPERCIST was seen, and discordance was noted in four (13.3%), with nearly perfect agreement (κ = 0.826) for response classification found between them (Table 3). Four PMD patients defined by PERCIST were classified as SMD (two patients) and PMR (two patients) based on imPERCIST due to the definition of the latter.

+ + + **Figure 1.** 61 year-old woman with left epithelioid malignant pleural mesothelioma who previously received neoadjuvant chemotherapy (pemetrexed + cisplatin), pleurectomy, and decortication surgery (pT3N1M0), then six cycles of chemotherapy (pemetrexed + cisplatin) after the operation, followed by 10 cycles of second-line therapy (irinotecan + gemcitabine) and then nivolumab as third-line chemotherapy. (**a**) Pre-nivolumab treatment FDG-PET/CT shows several areas of strong FDG uptake related to a pleural lesion (curved arrow) and mediastinal lymph nodal lesion (arrow). (**b**) Pre-nivolumab treatment contrast-enhanced CT shows mass-forming thickness of pleura lesion (curved arrow) and mediastinal lymph nodal lesion (arrow). (**c**) During-treatment FDG-PET/CT after 13 cycles of nivolumab shows FDG uptake disappearance in both pleural (curved arrow) and nodal (arrow) lesions. (**d**) During-treatment contrast-enhanced CT after 13 cycles of nivolumab shows remarkable improvements of both pleural (curved arrow) and nodal (arrow) lesions. EORTC, PERCIST, and imPERCIST indicated CMR. Interpretation of combined modified RECIST and RECIST 1.1 indicated a classification of PR, with the sum pleural lesion size decreasing by 45.5% and the sum mediastinal node size decreasing by 78.3%. The patient continued with 29 more cycles of nivolumab and was alive without progression at 15.1 months after nivolumab initiation.

+ **Figure 2.** 74 year-old man with right epithelioid malignant pleural mesothelioma (cT2N0M0), who previously received six cycles of first-line chemotherapy (pemetrexed + cisplatin) and then 12 cycles of nivolumab as second-line chemotherapy. (**a**) Pre-nivolumab treatment FDG-PET/CT shows multiple areas of strong FDG uptake in areas of right pleural lesions (arrows). (**b**) Pre-nivolumab treatment contrast-enhanced CT shows mass-forming thickness of right pleura (arrows). (**c**) Post-treatment FDG-PET/CT after 12 cycles of nivolumab shows remarkable progression of multiple pleural lesions (arrows) and appearance of new pleural lesions. (**d**) Post-treatment contrast-enhanced CT after 12 cycles of nivolumab shows remarkable progression of pleural lesions (arrows). EORTC, PERCIST, imPERCIST, and CT criteria (modified RECIST and RECIST 1.1) indicated PMD or PD due to remarkable progression and appearance of new lesions. In FDG-PET/CT results, the SULpeak sum of the five highest level pleural lesions was increased by 98.6%. In CT findings, the sum size of six pleural lesions perpendicular to the chest wall was increased by 40.3%. According to the second (**c**) FDG-PET/CT and (**d**) contrast-enhanced CT result, the patient started another chemotherapy series (irinotecan + gemcitabine), though was alive at 13.9 months after initiation of nivolumab.


**Table 2.** Comparison of treatment response assessments in EORTC criteria and PERCIST.

Data are presented as numbers. Abbreviations: EORTC: European Organization for Research and Treatment of Cancer, PERCIST: Positron Emission Tomography Response Criteria in Solid Tumors, PMD: progressive metabolic disease, SMD: stable metabolic disease, PMR: partial metabolic response, CMR: complete metabolic response.


**Table 3.** Comparison of treatment response assessments in imPERCIST and two other PET citeria (EORTC criteria and PERCIST).

Data are presented as numbers. Abbreviations: EORTC: European Organization for Research and Treatment of Cancer, PERCIST: Positron Emission Tomography Response Criteria in Solid Tumors, imPERCIST: immunotherapy-modified Positron Emission Tomography Response Criteria in Solid Tumors, PMD: progressive metabolic disease, SMD: stable metabolic disease, PMR: partial metabolic response, CMR: complete metabolic response.

Finally, in 18 (60.0%) cases concordance was noted between the CT criteria (combined modified RECIST and RECIST 1.1) and three PET response classifications (EORTC, PER-CIST, imPERCIST), while discordance was noted in 12 (40.0%), with moderate agreement (κ = 0.516 between CT criteria and EORTC, κ = 0.529 between CT criteria and PERCIST, κ = 0.544 between CT criteria and imPERCIST) noted between them for response classification (Table 4). Five (16.7%) of the present 30 patients were classified as CMR based on the EORTC, PERCIST, and imPERCIST criteria, which was not demonstrated by CT criteria (combined modified RECIST and RECIST 1.1).

**Table 4.** Comparison of treatment response assessments in CT criteria (combined modified RECIST and RECIST1.1) and three PET criteria (EORTC criteria, PERCIST, imPERCIST).


Data are presented as numbers. Abbreviations: EORTC: European Organization for Research and Treatment of Cancer, PERCIST: Positron Emission Tomography Response Criteria in Solid Tumors, imPERCIST: immunotherapy-modified Positron Emission Tomography Response Criteria in Solid Tumors, PMD: progressive metabolic disease, SMD: stable metabolic disease, PMR: partial metabolic response, CMR: complete metabolic response, PD: progressive disease, SD: stable disease, PR: partial response, CR: complete response.

#### *3.3. Progression Free Survivals (PFS)*

Twenty-six (86.7%) of the 30 patients had progressive disease noted after a median period of 8.0 months (3.3–22.4 months). Both PET (EORTC, PERCIST, imPERCIST) and CT (combined modified RECIST and RECIST 1.1) criteria indicated a significantly longer PFS in patients with no progression (CMR/PMR/SMD, CR/PR/SD) as compared to those with PMD or PD (EORTC: *p* = 0.0004, PERCIST: *p* = 0.0003, imPERCIST: *p* < 0.0001, combined modified RECIST and RECIST 1.1: *p* = 0.0015) (Figure 3). Similarly, responders (CMR/PMR) based on PET criteria (EORTC, PERCIST, imPERCIST) showed significantly longer PFS than non-responders (SMD/PMD) (EORTC: *p* = 0.0064, PERCIST: *p* = 0.0007, imPERCIST: *p* = 0.0005), whereas use of CT criteria (combined modified RECIST and RECIST 1.1) showed that responders (CR/PR) had a tendency for longer PFS as compared to non-responders (SD/PD), though the difference was not significant (*p* = 0.074) (Figure 4).

**Figure 3.** Progression-free survival (PFS) of malignant pleural mesothelioma patients treated by nivolumab therapy, with and without progression. (**a**) EORTC demonstrated that patients with no progression (CMR/PMR/SMD) showed significantly longer PFS than those with PMD (*p* = 0.0004). (**b**) PERCIST demonstrated that patients with no progression (CMR/PMR/SMD) showed significantly longer PFS than those with PMD (*p* = 0.0003). (**c**) imPERCIST demonstrated that patients with no progression (CMR/PMR/SMD) showed significantly longer PFS than those with PMD (*p* < 0.0001). (**d**) Combined modified RECIST and RECIST 1.1 demonstrated that patients with no progression (CR/PR/SD) showed significantly longer PFS than those with PD (*p* = 0.0015).

**Figure 4.** Progression-free survival (PFS) of malignant pleural mesothelioma patients treated by nivolumab therapy, with and without response. (**a**) EORTC demonstrated that responders (CMR/PMR) showed significantly longer PFS than non-responders (SMD/PMD) (*p* = 0.0064). (**b**) PERCIST demonstrated that responders (CMR/PMR) showed significantly longer PFS than non-responders (SMD/PMD) (*p* = 0.0007). (**c**) imPERCIST demonstrated that responders (CMR/PMR) showed significantly longer PFS than non-responders (SMD/PMD) (*p* = 0.0005). (**d**) Combined modified RECIST and RECIST 1.1 demonstrated that responders (CR/PR) tended to show longer PFS than non-responders (SD/PD), without a significant difference (*p* = 0.074).

#### *3.4. Overall Survival (OS)*

Nine (30.0%) of the 30 patients died from MPM after a median 14.9 months (5.8–25.6 months). Both PET (EORTC, PERCIST, imPERCIST) and CT (combined modified RECIST and RECIST 1.1) criteria indicated that patients without progression (CMR/PMR/SMD, CR/PR/SD) had a tendency for longer OS as compared to patients with PMD or PD (EORTC: *p* = 0.055, PERCIST: *p* = 0.052, imPERCIST: *p* = 0.089, combined modified RECIST and RECIST 1.1: *p* = 0.056), though the difference was not significant (Figure 5). Similarly, according to both PET (EORTC, PERCIST, imPERCIST) and CT (combined modified RECIST and RECIST 1.1) criteria, responders (CMR/PMR, CR/PR) showed longer OS than non-responders (SMD/PMD, SD/PD) (EORTC: *p* = 0.055, PERCIST: *p* = 0.052, imPERCIST: *p* = 0.053) without a significant difference, whereas CT criteria (combined mRECIST and RECIST 1.1) indicated that OS values for responders (CR/PR) and non-responders (SD/PD) were not different (*p* = 0.87) (Figure 6).

**Figure 5.** Overall survival (OS) of malignant pleural mesothelioma patients treated by nivolumab therapy, with and without progression. (**a**) EORTC demonstrated that patients with no progression (CMR/PMR/SMD) tended to show longer OS than those with PMD, without a significant difference (*p* = 0.055). (**b**) PERCIST demonstrated that patients with no progression (CMR/PMR/SMD) tended to show longer OS than those with PMD, without a significant difference (*p* = 0.052). (**c**) imPERCIST demonstrated that patients with no progression (CMR/PMR/SMD) tended to show longer OS than those with PMD, without a significant difference (*p* = 0.089). (**d**) Combined modified RECIST and RECIST 1.1 demonstrated that patients with no progression (CR/PR/SD) tended to show longer OS than those without PD, without a significant difference (*p* = 0.056).

**Figure 6.** Overall survival (OS) of malignant pleural mesothelioma patients treated by nivolumab therapy, with and without response. (**a**) EORTC demonstrated that responders (CMR/PMR) tended to show longer OS than non-responders (SMD/PMD), without a significant difference (*p* = 0.055). (**b**) PERCIST demonstrated that responders (CMR/PMR) tended to show longer OS than non-responders (SMD/PMD), without a significant difference (*p* = 0.052). (**c**) imPERCIST demonstrated that responders (CMR/PMR) tended to show longer OS than non-responders (SMD/PMD), without a significant difference (*p* = 0.053). (**d**) Combined modified RECIST and RECIST 1.1 demonstrated no significant difference for OS between responders (CR/PR) and non-responders (SD/PD) (*p* = 0.87).

#### **4. Discussion**

This is the first known study to compare three FDG-PET/CT criteria (EORTC, PER-CIST, imPERCIST) with CT criteria (combined modified RECIST and RECIST 1.1) used to evaluate tumor response to ICI therapy in patients with recurrent MPM as well as prediction of prognosis. All of the FDG-PET/CT and CT criteria analyzed were found to be accurate for both evaluation of tumor response and prediction of PFS in the present cohort, though the FDG-PET/CT criteria showed a slight superiority. FDG-PET/CT is known as an accurate tool for evaluating tumor viability, and the results are useful for clear diagnosis of CMR when a residual tumor does not have abnormal FDG uptake during or after treatment. We noted that the EORTC, PERCIST, and imPERCIST criteria classified five (16.7%) of the present 30 patients as CMR, which was not obtained with use of the contrast-enhanced CT criteria (combined modified RECIST and RECIST 1.1). Additionally, FDG-PET/CT findings are known to be accurate for detecting bone/muscle and tiny lymph node metastasis, as well as very small dissemination in a second FDG-PET/CT examination. This study found that the EORTC and PERCIST criteria were able to classify four and three more patients (10–13.3%) as PMD in comparison to contrast-enhanced CT results with use of the combined modified RECIST and RECIST 1.1 criteria. The number of PMD cases determined by imPERCIST was lower than that by the EORTC and PERCIST criteria, due to the imPERCIST definition (new lesions do not result in PMD and are included in the sum of SULpeak if they showed a higher uptake level than existing target lesions).

In summary, all three FDG-PET/CT criteria clearly judged more patients (16.7%) as CMR and two of those, EORTC and PERCIST, were able to judge more patients (10–13.3%) as PMD in comparison with CT criteria. For predicting PFS, the three FDG-PET criteria

were superior to the CT criteria and imPERCIST demonstrated the highest rate of accurate prediction. It is considered that FDG-PET/CT might be a powerful tool for late (≥4 cycles) response assessment when evaluating ICI therapy and able to identify MPM patients who can most benefit from that. If MPM patients undergoing nivolumab were judged as non-PMD, nivolumab is continued. Unfortunately, MPM patients undergoing nivolumab were judged as PMD, alternative treatment (cisplatin/carboplatin and pemetrexed, pemetrexed, or irinotecan and gemcitabine) is tried in order to improve patient outcome.

Tumor infiltration by immune cells can delay tumor shrinkage or even cause a temporary increase in size (pseudoprogression), making assessment of tumor response to ICI treatment challenging. Although several criteria have been proposed for use with CT findings to determine response to that treatment, such as immune-related response criteria (irRC) [15], immune-related RECIST (irRECIST) [9], and immune RECIST (iRECIST) [16], as well as for use with FDG-PET results, including PET/CT criteria for early prediction of Response to Immune checkpoint inhibitor Therapy (PECRIT) [17], PET Response Evaluation Criteria for Immunotherapy (PERCIMT) [18], imPERCIST [10], and immune PERCIST (iPERCIST) [19], an optimal evaluation method has yet to be determined. Although pseudoprogression must be considered in the early phase following initiation of ICI treatment, that was not observed in any of the present 30 patients, probably due to late (≥4 cycles) response assessment.

There have been several articles demonstrating the usefulness of FDG-PET/CT for assessing the ICI therapeutic response, especially early response (2~4 cycles of ICI) in metastatic melanoma patients [10,17,18,20]. Cho et al. [17] analyzed PECRIT, which includes change in lesion size combined with change in FDG avidity shown by FDG-PET/CT after one cycle of ICI monotherapy (ipilimumab, nivolumab, or BMS-936559), in a study of 20 advanced melanoma patients. They found that criteria including SD shown by RECIST 1.1 and an SULpeak increase >15.5% in the hottest lesion shown by FDG-PET/CT were accurate for predicting treatment response after four months, with values for sensitivity, specificity, and accuracy of 100%, 93%, and 95%, respectively. In another study, PERCIMT, which uses absolute number of new lesions rather than changes in metabolic parameters (i.e., SUV) shown by FDG-PET/CT, was introduced by Anwar et al. [18] to evaluate 41 patients with metastatic melanoma after four cycles of ipilimumab. Those criteria, which include four or more new lesions <1 cm in functional diameter, were found to be accurate for clinical benefit prediction, with a sensitivity of 84% and specificity of 100%. Ito et al. [10] originally presented imPERCIST, in which the appearance of new lesions is not used to define PMD. Those authors noted that an increase in SULpeak sum of ≥30% in up to five measured lesions in FDG-PET/CT results accurately reflected PMD after 2–4 cycles of ipilimumab treatment in 60 metastatic melanoma patients. Although the significant and apparent superiority of FDG-PET/CT was not observed in our series, the potential reason may be biological difference between malignant melanoma and MPM, late (≥4 cycles) response assessment, or small sample size. With iPERCIST, Goldfarb et al. [19] introduced two new categories used for response to PMD, unconfirmed (UPMD) and confirmed (CPMD). Results of 28 non-small cell lung cancer patients who were receiving nivolumab were analyzed and indicated that any metabolic progression observed at eight weeks (after four cycles) should be confirmed by another FDG-PET/CT examination performed four weeks later, while they also noted that iPERCIST was useful for differentiation of responders from non-responders and OS prediction (*p* = 0.0003).

The present study has some limitations, including its retrospective nature, performance at a single center, and small sample size. Thus, generalization of the findings is limited and statistical errors are possible. To clarify the roles of FDG-PET/CT and CT for decision making, as well as predicting long-term outcomes in clinical settings a prospective multicenter trial with a larger cohort will be necessary. Additionally, the enrolled cohort was heterogeneous, as patients who underwent nivolumab monotherapy and received the second FDG-PET/CT examination after from four to 21 cycles were included; thus, confounding factors were likely introduced. The impact of PET/CT is primarily early

within the course of treatment, because metabolic changes proceed volumetric changes [20]. This cannot be demonstrated in this study due to the very large and relatively late variation of the time points for the follow-up study. We are planning a prospective study to clarify both early and late response evaluation with less variation of the time to second and third FDG-PET/CT examinations from ICI treatment start, using three times of FDG-PET/CT examinations in MPM patients receiving ICI treatment Although we used four different PET/CT scanners, we harmonized PET quantitative values by a software, which can harmonize SUVs obtained with different PET/CT systems using phantom data [12]. Finally, irRC, irRECIST, iRECIST, and iPERCIST were not evaluated, because regular and follow-up CT and FDG-PET/CT examinations were not performed in every case.

#### **5. Conclusions**

In conclusion, results obtained with the use of three FDG-PET/CT (EORTC, PERCIST, and imPERCIST) and one CT (combined modified RECIST and RECIST 1.1) criteria were found useful to evaluate tumor response to ICI therapy as well as prediction of progression in recurrent MPM patients. In comparison with CT, all three FDG-PET/CT criteria judged a greater percentage of patients (16.7%) as CMR, while two (EORTC, PERCIST) judged a greater percentage (10–13.3%) as PMD. For predicting PFS, the three FDG-PET criteria were superior to the CT criteria, and imPERCIST demonstrated the highest rate of accurate prediction. Further validation in a prospective study with a larger cohort is warranted.

**Author Contributions:** Conceptualization: K.K. (Kazuhiro Kitajima), H.Y., and K.Y.; methodology: K.K. (Kazuhiro Kitajima) and M.M.; recruiting of patients: T.M., T.Y., K.K. (Kazuhiro Kitajima), T.K., A.N., M.H., N.K., and S.H.; statistical analysis: H.Y.; resources: K.K. (Kazuhiro Kitajima); data curation: K.K. (Kazuhiro Kitajima) and K.K. (Kozo Kuribayashi); writing—original draft preparation: K.K. (Kazuhiro Kitajima); writing—review and editing: T.M., K.K. (Kozo Kuribayashi), and K.Y.; supervision: K.Y. and T.M.; project administration: K.K. (Kazuhiro Kitajima); All authors have read and agreed to the published version of the manuscript.

**Funding:** This work was supported by a JSPS KAKENHI grant (No. 19K08187).

**Institutional Review Board Statement:** The study was conducted according to the guidelines of the Declaration of Helsinki, and approved by the Institutional Review Board of Hyogo College of Hospital (protocol code 3315 and date of approval 2019/12/26).

**Informed Consent Statement:** Patient consent was waived due to the retrospective study by IRB.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


## *Article* **VATS Pleurectomy Decortication Is a Reasonable Alternative for Higher Risk Patients in the Management of Malignant Pleural Mesothelioma: An Analysis of Short-Term Outcomes**

**Dong-Seok Lee \* , Andrea Carollo, Naomi Alpert, Emanuela Taioli and Raja Flores**

Thoracic Surgery Department, Icahn School of Medicine at Mount Sinai, Mount Sinai Health System, New York, NY 10029, USA; Andrea.Carollo@mountsinai.org (A.C.); Naomi.Alpert@mountsinai.org (N.A.); Emanuela.Taioli@mountsinai.org (E.T.); Raja.Flores@mountsinai.org (R.F.)

**\*** Correspondence: Dong-Seok.Lee@mountsinai.org

**Simple Summary:** Malignant pleural mesothelioma (MPM) is an aggressive malignancy that drastically affects a patient's quality of life. Surgery typically entails radical resection with or without the removal of the underlying lung. In an era where minimally invasive surgery is sought after, MPM remains an anomaly. The purpose of this study is to assess the feasibility of minimally invasive surgery as an alternative to more radical surgery in MPM. We examined short-term outcomes between the radical approaches and minimally invasive surgery and minimally invasive surgery had improved outcomes. Minimally invasive surgery can be considered in patients with MPM.

**Abstract:** Surgery is a mainstay of treatment allowing for debulking of tumor and expansion of the lung for improvement in median survival and quality of life for patients with malignant pleural mesothelioma (MPM). Although optimal surgical technique remains open for debate—extrapleural pneumonectomy (EPP) vs. pleurectomy/decortication (P/D)—minimally invasive surgery (VATS-P/D) remains underutilized in the management of MPM. We examined whether VATS-P/D is a feasible alternative to EPP and P/D. We evaluated the New York Statewide Planning and Research Cooperative System (SPARCS) from 2007–2017 to assess the short-term complications of EPP vs. P/D, including a subanalysis of open P/D vs. VATS-P/D. There were 331 patients with open surgery; 269 with P/D and 62 with EPP. There were 384 patients with P/D; 269 were open and 115 VATS. Rates of any complication were similar between EPP and P/D patients, but EPP had significantly higher rates of cardiovascular complications. After adjusting for confounders, those with a VATS approach were less likely to have any complication, compared to an open approach and significantly less likely to have a pulmonary complication. VATS-P/D remains a viable alternative to radical surgery in MPM patients allowing for improved short-term outcomes.

**Keywords:** malignant mesothelioma; VATS; extrapleural pneumonectomy; pleurectomy decortication

#### **1. Introduction**

Malignant pleural mesothelioma (MPM) is a rare but aggressive cancer with an overall poor prognosis. Treatment frequently involves multimodal therapy, of which surgical resection remains an essential component, significantly improving median survival compared to patients who do not undergo surgery [1]. However, there remains debate about the optimal surgical technique. Extrapleural pneumonectomy (EPP) theoretically offers the better chance at complete resection and was considered the standard. However, lung-sparing pleurectomy/decortication (P/D) has become more common, as research has indicated decreased perioperative morbidity and mortality and similar survival compared to EPP [2–6]. In addition, quality of life appears better as physical and social function and global health measures are better at 12 months with P/D over EPP [7,8].

**Citation:** Lee, D.-S.; Carollo, A.; Alpert, N.; Taioli, E.; Flores, R. VATS Pleurectomy Decortication Is a Reasonable Alternative for Higher Risk Patients in the Management of Malignant Pleural Mesothelioma: An Analysis of Short-Term Outcomes. *Cancers* **2021**, *13*, 1068. https:// doi.org/10.3390/cancers13051068

Academic Editors: Daniel L. Pouliquen and Joanna Kopecka

Received: 20 January 2021 Accepted: 23 February 2021 Published: 3 March 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

Despite the increasing utilization of minimally invasive techniques in many oncologic surgical procedures, MPM-directed surgeries have historically been performed as open procedures. Although minimally invasive lung surgery has improved short-term outcomes with equivalent long-term survival compared to open surgery [9,10], its use in MPM is more challenging. Video-assisted thoracoscopic surgery (VATS) has been primarily focused on diagnosis or palliation of symptoms. Although there is extensive literature comparing outcomes of EPP to P/D, there is a paucity of data examining outcomes of minimally invasive surgery for MPM. Our group had previously reported improved shortterm outcomes for patients with P/D compared to EPP using New York State hospital discharge data [3]. The aims of this study were to utilize the same large database to provide updated results of our prior study, with an added focus on comparing a minimally invasive approach to open surgery.

#### **2. Materials and Methods**

#### *2.1. Data Source and Sample Selection*

This analysis used the New York Statewide Planning and Research Cooperative System (SPARCS) from 2007–2017. SPARCS includes all hospital discharges in the state, and has information on patient demographics, diagnoses, procedures, admission and discharge type. This research was approved by the Mount Sinai Institutional Review Board (IRB# 18-00947, FWA #00005656).

There were 4,959,270 patients at least 50 years old, with a patient identifier who had an inpatient discharge between 1 January 2007 and 31 December 2017. Those with an admission accompanied by a diagnosis of pleural mesothelioma (*n* = 2169) and who had either EPP or P/D (See Supplementary Materials Table S1 for ICD-9 and ICD-10 diagnosis and procedures codes) were included (*n* = 589) for analysis. For patients with multiple mesothelioma-related surgeries, the first surgery was chosen. Patients where the surgical approach (open or minimally invasive) was unknown were excluded, as were the few who were coded as having minimally invasive EPP (nexcl = 143). The initial analysis was limited to patients with an open EPP or P/D surgery (*n* = 331), while a secondary analysis compared surgical approach among those with P/D (*n* = 384) (Figure 1).

**Figure 1.** Patient Selection.

#### *2.2. Predictors and Outcomes*

The primary predictors of interest were the type of surgery and surgical approach. Outcomes of interest were short-term complications after surgery. In-hospital complications were defined based on diagnosis codes that were not present at the time of admission

(Supplementary Materials Table S2), and were categorized as cardiovascular, pulmonary, infectious or intraoperative complications. Patient comorbidities were defined using the algorithm described by Elixhauser, et al. [11], and a count of non-cancer-related comorbidities was created. Other covariates of interest included age, gender, race (Non-Hispanic White (NHW) vs. Hispanic or Non-White), primary insurance payer (government vs. non-government), type of admission to the hospital (urgent/emergency vs. elective), and the year of surgery.

#### *2.3. Statistical Analysis*

Patients were compared across surgical type on all variables, using t-tests for continuous variables, and χ 2 -tests for categorical variables. Univariate and multivariable logistic regressions were used to model the independent associations between covariates and type of surgery, using Odds Ratios (ORs) and 95% Confidence Intervals (CI). Multivariable logistic regression models were also used to assess the association of surgical type with having complications (any, cardiovascular, or pulmonary), adjusting for possible confounders. Supraventricular arrhythmia was examined individually as a subset of cardiovascular complications. As there were a very small number of infectious and intraoperative complications, these were individually assessed only at the univariate level. Multivariable models were adjusted for age, gender, race/ethnicity, admission type, insurance, number of comorbidities, and year of surgery, to account for changes over time. Outcomes were also assessed using an optimal propensity matching analysis, with a maximum difference of 0.01, matching on all variables.

Analyses were repeated on the subset of patients with P/D, in order to compare outcomes in patients with minimally invasive and open approaches. All analyses were conducted using SAS software, v 9.4 (SAS Institute, Cary, NC, USA).

#### **3. Results**

#### *3.1. Extrapleural Pneumonectomy vs. Pleurectomy Decortication*

There were 331 patients with open surgery; 269 (81.3%) with P/D and 62 (18.7%) with EPP. EPP patients were significantly younger (mean age: 64.6 vs. 69.1 years, *p* < 0.0001), more likely to have non-government insurance coverage (61.3% vs. 44.6%, *p* = 0.0217), and had fewer comorbidities (29.0% vs. 55.4% with ≥2 comorbidities; *p* = 0.0002). EPP patients also more frequently had elective admissions (*p* = 0.0552) (Table 1).

**Table 1.** Demographics of the sample, according to surgery type.



**Table 1.** *Cont.*

Abbreviations: P/D, Pleurectomy Decortication; EPP, Extrapleural pneumonectomy. \* Exact cell sizes masked to protect against identification of patients.

After adjustment, those with EPP were significantly younger (ORadj: 0.91, 95% CI: 0.86–0.96) and significantly less likely to have an urgent or emergency surgery (ORadj: 0.21, 95% CI: 0.05–0.97). There was no significant difference in gender, race/ethnicity, type of insurance, or number of comorbidities (Table 2).

**Table 2.** Independent Factors Associated with Receipt of EPP vs. P/D (*n* = 326).


\* Adjusted for all variables listed and year of surgery.

At the univariate level, rates of any complication were similar between EPP and P/D patients (43.5% for EPP vs. 42.0% for P/D; *p* = 0.8248), but EPP had significantly higher rates of cardiovascular complications (32.3% vs. 13.4%; *p* = 0.0004) supraventricular arrhythmia (27.4% vs. 10.0%; *p* = 0.0003), and lower rates of pulmonary complications (21.0% vs. 34.2%; *p* = 0.0439) (Table 1).

In the multivariable analysis, those with EPP were significantly more likely to have any complication (ORadj: 2.12, 95% CI: 1.08–4.18), as well as have cardiovascular complications (ORadj: 5.00, 95% CI: 2.23–11.24), and supraventricular arrhythmia specifically (ORadj: 6.63, 95% CI: 2.64–16.64). There was no significant difference in the odds of a pulmonary complication (Table 3).

**Table 3.** Odds of Complications in EPP vs. P/D patients, multivariable and propensity-matched analyses.


Abbreviations: EPP, extrapleural pneumonectomy; P/D, Pleurectomy Decortication. \* Adjusted for/propensity matched on age, gender, race/ethnicity, admission type, insurance, number of comorbidities, and year of surgery. Adjusted models were not conducted for infection or intraoperative complication due to an insufficient number of outcomes.

After propensity matching, there were 50 EPP and 50 P/D patients, who were well matched on all covariates (range of *p*-values: 0.5637 to 1). Although not statistically significant, patients with EPP continued to have more cardiovascular complications in general (OR: 2.60, 95% CI: 0.93–7.29), and specifically supraventricular arrhythmia (OR: 2.75, 95% CI: 0.88–8.64) (Table 3).

#### *3.2. Minimally Invasive vs. Open P/D*

There were 384 patients with P/D; 269 (70.1%) with an open surgical approach, and 115 (29.9%) with a minimally invasive approach. Patients with a minimally invasive surgical approach were significantly older (mean age: 71.8 vs. 69.1 years; *p* = 0.0132) and more likely to have an urgent/emergency admission (47.0% vs. 11.2%; *p* < 0.0001). They were also less often NHW (*p* = 0.0524) (Table 4).

**Table 4.** Demographics of the sample according to surgical approach among P/D patients.



**Table 4.** *Cont.*

Abbreviations: P/D, Pleurectomy Decortication \* Exact cell sizes masked to protect against identification of patients. Percentages and *p*-values are presented for non-missing values.

After adjustment, those with a minimally invasive approach remained significantly older (ORadj: 1.05, 95% CI: 1.01–1.08) and more likely to have an urgent/emergency admission (ORadj: 7.18, 95% CI: 4.07–12.64), compared to those with an open approach (Table 5).

**Table 5.** Independent Factors Associated with Receipt of Minimally Invasive vs. Open Surgery (*n* = 378).


\* Adjusted for all variables listed and year of surgery.

After adjusting for confounders, those with a minimally invasive approach were less likely to have any complication, compared to those with an open approach (ORadj: 0.58, 95% CI: 0.34–1.01) and significantly less likely to have a pulmonary complication (ORadj: 0.55, 95% CI: 0.31–0.99) (Table 6).

**Table 6.** Odds of complications in minimally invasive vs. open P/D patients, multivariable and propensity-matched analyses.


Abbreviations: P/D, Pleurectomy Decortication. \* adjusted for/propensity matched on age, gender, race/ethnicity, admission type, insurance, number of comorbidities, and year of surgery. Adjusted models were not conducted for infection or intraoperative complication, due to an insufficient number of outcomes.

The propensity-matched analysis was well balanced on all covariates (range of *p*-values: 0.3980–1) with 75 patients per group. Although not significant, results were similar, with minimally invasive surgery having lower risk of overall complications (OR: 0.70, 95% CI: 0.37–1.32) and pulmonary complications (OR: 0.65, 95% CI: 0.30–1.38) (Table 6).

#### **4. Discussion**

Our study utilized the New York SPARCS database in order to compare perioperative morbidity with EPP, P/D, and VATS-P/D for MPM. Complications examined were cardiovascular, pulmonary, infectious, and intraoperative complications. The majority of complications were either cardiovascular or pulmonary. Perioperative mortality was not included in the present analysis due to limited observations. Generally, the more radical resection was associated with younger age, elective procedure, and increased incidence of complications. EPP patients were more likely to have cardiovascular complications, primarily supraventricular arrhythmias, than P/D patients on multivariable analysis and propensity matching. On the other hand, cardiovascular complications were similar in open and minimally invasive P/D patients but open patients were more prone to pulmonary complications on multivariable analysis and propensity matching.

The goal of oncologic surgery with curative intent is removal of all macroscopic and, if possible, microscopic disease. This is challenging in MPM as it is an insidious diffuse disease throughout the pleura and often requires radical resection. Therefore, the mainstay of surgical treatment for MPM includes extrapleural pneumonectomy and pleurectomy/decortication. A number of studies have been performed showing that EPP and P/D confer similar overall survival but that the short-term mortality and morbidity associated with EPP is greater than P/D [2–6]. Less radical and more minimally invasive surgery has primarily been limited to diagnostic biopsy or symptom management with talc pleurodesis or indwelling pleural catheters. VATS- P/D has not achieved widespread use in the management of MPM, as it is primarily considered to be a palliative surgical option [12] as opposed to a potentially curative one. The goal of VATS- P/D is the debulking of enough pleural disease and decortication of the underlying trapped lung in order to obliterate the pleural space to allow pleural apposition.

The only randomized control trial to date, MesoVATS, compared VATS partial pleurectomy (VATS-PP) to talc pleurodesis [13]. The primary endpoint was overall survival at

12 months and no significant difference was noted between the two groups. Although VATS-PP had a non-significant trend towards increased morbidity, the authors noted a 70% resolution of pleural effusion with VATS-PP compared to 77% resolution with talc pleurodesis but significantly improved quality of life scores at 6 and 12 months for the VATS-PP group. A follow-up study, currently in progress, aims to address VATS-PP against the use of indwelling pleural catheters for patients with MPM and trapped lung [14].

In addition to providing a palliative benefit, VATS-P/D appears to confer a survival benefit as cytoreduction and post-resection tumor volume may play a role in long-term outcomes [15,16]. It is unclear how it compares with more radical surgery. A previously published single institutional study looking at VATS P/D showed a modest non-significant improvement in survival with VATS versus EPP (14 months vs. 11.5 months). They also noted symptomatic improvement in the majority of patients and statistically significant advantage in 30-day mortality versus EPP [17].

Our study is the first to utilize a large population-based database in order to assess short-term outcomes in EPP, P/D, and minimally invasive P/D. However, it is not without its limitations. Despite the extensive size of the dataset, MPM remains an uncommon disease such that it accounts for a very small percentage of admissions, and thus, numbers remain relatively small. There may be selection bias in regards to surgical technique due to both surgeon preference and elective versus emergent presentation. Confounders that are unable to be addressed include information that could not be ascertained from the database, such as tumor grade, oncologic stage, long-term outcomes, surgeon experience, and potential use of induction therapy. However, this analysis includes a greater number of patients than would be available from a single-center study.

In confirmation of our previous analysis, P/D was associated with improved shortterm outcomes compared to EPP and likely explains the shift from equivalent amounts of EPP and P/D performed (46.6% EPP, 53.4% P/D) from 1995–2012 [3] to predominantly P/D (81.3% P/D, 18.7% EPP) performed for the treatment of MPM from 2007–2017. Despite the increasing age of patients with less radical surgery, VATS P/D patients exhibited improved short-term outcomes, when controlling for this difference. Further investigation in regards to long-term survival with VATS P/D in comparison to EPP and P/D is needed.

#### **5. Conclusions**

Malignant pleural mesothelioma remains a challenging cancer to treat. Surgical options range from the more radical curative techniques such as EPP and P/D to the less invasive palliative VATS P/D. Patients who undergo VATS P/D have better short-term outcomes compared to those who undergo curative attempts at surgery. Therefore, VATS P/D should be considered in the armamentarium of treatment for MPM, especially in older and frailer patients who may not tolerate more radical surgery.

**Supplementary Materials:** The following are available online at https://www.mdpi.com/2072-669 4/13/5/1068/s1, Table S1: Diagnosis and Procedure Codes to Identify Surgical Pleural Mesothelioma Patients, Table S2: Diagnosis Codes to Identify Complications.

**Author Contributions:** Conceptualization, D.-S.L., A.C., N.A.; Methodology, N.A.; Formal Analysis, N.A.; Resources, R.F., E.T.; Data Curation, N.A.; Writing—Original Draft Preparation, D.-S.L.; A.C.; Writing—Review & Editing, D.-S.L., N.A.; Supervision, E.T., R.F. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Institutional Review Board Statement:** This research was approved by the Mount Sinai Institutional Review Board (IRB# 18-00947, FWA #00005656).

**Informed Consent Statement:** Patient consent was waived due to utilization of a retrospective population-based database where researchers had no method of contact for included subjects.

**Data Availability Statement:** No new data was generated by the authors of this study. The data used and analyzed during the current study are available from the New York State Department of Health.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


## *Article* **Radical Hemithoracic Radiotherapy Induces Systemic Metabolomics Changes That Are Associated with the Clinical Outcome of Malignant Pleural Mesothelioma Patients**

**Emanuela Di Gregorio <sup>1</sup> , Gianmaria Miolo <sup>2</sup> , Asia Saorin <sup>1</sup> , Elena Muraro <sup>1</sup> , Michela Cangemi <sup>1</sup> , Alberto Revelant <sup>3</sup> , Emilio Minatel <sup>3</sup> , Marco Trovò 4 , Agostino Steffan <sup>1</sup> and Giuseppe Corona 1,\***



**Citation:** Di Gregorio, E.; Miolo, G.; Saorin, A.; Muraro, E.; Cangemi, M.; Revelant, A.; Minatel, E.; Trovò, M.; Steffan, A.; Corona, G. Radical Hemithoracic Radiotherapy Induces Systemic Metabolomics Changes That Are Associated with the Clinical Outcome of Malignant Pleural Mesothelioma Patients. *Cancers* **2021**, *13*, 508. https://doi.org/10.3390/ cancers13030508

Academic Editors: Daniel L. Pouliquen and Joanna Kopecka Received: 23 December 2020 Accepted: 25 January 2021 Published: 29 January 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

**Simple Summary:** Radical hemithoracic radiotherapy represents a promising new advance in the field of radiation oncology and encouraging results have been achieved in the treatment of malignant pleural mesothelioma patients. This study showed that this radiotherapy modality produces significant changes in serum metabolomics profile mainly affecting arginine and polyamine biosynthesis pathways. Interestingly, individual metabolomics alterations were found associated with the clinical overall survival outcome of the radiotherapy treatment. These results highlight metabolomics profile analysis as a powerful prognostic tool useful to better understand the mechanisms underlying the interpatients variability and to identify patients who may receive the best benefit from this specific radiotherapy treatment.

**Abstract:** Radical hemithoracic radiotherapy (RHRT) represents an advanced therapeutic option able to improve overall survival of malignant pleural mesothelioma patients. This study aims to investigate the systemic effects of this radiotherapy modality on the serum metabolome and their potential implications in determining the individual clinical outcome. Nineteen patients undergoing RHRT at the dose of 50 Gy in 25 fractions were enrolled. Serum targeted metabolomics profiles were investigated at baseline and the end of radiotherapy by liquid chromatography and tandem mass spectrometry. Univariate and multivariate OPLS-DA analyses were applied to study the serum metabolomics changes induced by RHRT while PLS regression analysis to evaluate the association between such changes and overall survival. RHRT was found to affect almost all investigated metabolites classes, in particular, the amino acids citrulline and taurine, the C14, C18:1 and C18:2 acyl-carnitines as well as the unsaturated long chain phosphatidylcholines PC ae 42:5, PC ae 44:5 and PC ae 44:6 were significantly decreased. The enrichment analysis showed arginine metabolism and the polyamine biosynthesis as the most perturbed pathways. Moreover, specific metabolic changes encompassing the amino acids and acyl-carnitines resulted in association with the clinical outcome accounting for about 60% of the interpatients overall survival variability. This study highlighted that RHRT can induce profound systemic metabolic effects some of which may have a significant prognostic value. The integration of metabolomics in the clinical assessment of the malignant pleural mesothelioma could be useful to better identify the patients who can achieve the best benefit from the RHRT treatment.

**Keywords:** metabolomics; mesothelioma; radiotherapy; biomarkers; cancers

#### **1. Introduction**

Malignant pleural mesothelioma (MPM) is a rare primary carcinoma originating from the pleural cavity, strongly linked to asbestos exposure [1]. The long latency after exposure and its characteristics of invasiveness and high aggressiveness contribute to make MPM a silent and invariable fatal disease with a median survival of less than 1 year when untreated [2]. The trimodal therapeutic approach that combines surgery, chemotherapy, and sequential radiotherapy (RT) represents the mainstream of current therapeutic protocols for MPM [3,4]. Over the last decades, RT technology has evolved [5], and the intensity-modulated radiation therapy (IMRT) has become one of the most interesting advance allowing the delivery of highly conformal radiation doses to the whole hemithorax limiting the normal tissue exposure. In MPM patients, this new RT modality referred as radical hemithoracic radiotherapy (RHRT) is delivered with a curative intent. However, despite its potential, its wide application is still debated especially for its possible severe toxicity [6], even if recent clinical investigations have shown encouraging results in enhancing patients' survival with acceptable toxicity [7–9]. Despite the relevant overall survival gain, the clinical outcome of the RHRT was very heterogeneous among patients and there is an urgent need for prognostic biomarkers to guide clinical decision-making and to tailor the RHRT treatment. The knowledge of the molecular mechanisms involved in tumour and normal tissue response to RT has retained an important footstep to improve the efficacy of the treatment through the identification of specific molecular signatures useful to recognize patients who may achieve the best benefit from RT. In order to get more insight into the role of RHRT in the treatment of MPM patients, we investigated the host response to this specific treatment evaluating the systemic metabolic changes by the application of metabolomics and searched for potential new prognostic biomarkers.

Metabolomics is a rapidly advancing field that aims to characterize the concentration changes of all metabolites (<1 KDa) present in biological fluids or tissues [10]. The metabolomics profile describes the biochemical events occurring in an organism and reflects the complex interactions among age, sex, gene transcription, protein expression, physio-pathological conditions, and environmental effects as well as chemical or physical interventions such as the RT [11,12]. The radiation treatments may induce whole-body responses that can be mirrored and observed at the blood metabolome level. Hence, the blood metabolites composition represents a hypothetical source of biomarkers and the understanding of how metabolites and their concentrations change under RT interventions may allow the discovery of potential biomarkers for RT efficacy and toxicity. The effect of anticancer drug treatments on local and systemic metabolism have been widely investigated in different cancer types by the metabolomics tool [13–16]. Nevertheless, only a few broad-based metabolomics studies have been so far reported about the effects of RT on the host system [17–23] and none in the specific MPM field.

In attempt to fill this gap, this study aims to investigate the RHRT effects on the systemic metabolism by the analysis of changes in serum metabolomics profiles consequent to the treatment. The investigation provides new insights on the host biochemical alterations induced by the RHRT treatment and on their potential role in determining the individual clinical outcome. The results of this explorative translational investigation indicate that RHRT can produce profound effects on the serum metabolomics profile engaging amino acids and lipids metabolic pathways that could be relevant to establish the effective clinical benefit of the treatment.

#### **2. Results**

#### *2.1. Demographic and Clinical Baseline Patients' Characteristics*

This translational study investigated 19 nonmetastatic MPM patients who underwent RHRT treatment consisting of 50 Gy in 25 fractions with a simultaneous integrated boost of 60 Gy in residual active disease. The clinical and demographic characteristics of the 19 MPM patients are reported in Table 1. The median age of the patients was 70 years (range: 33–79) with a great prevalence of male patients (89%). At baseline, 31% of patients presented adequate clinical conditions reporting an ECOG performance status (PS) score of 0, the majority of patients (53%) presented a PS score of 1 and only 16% had a PS score of 2. At the diagnosis, 95% of the MPM tumours had an epithelioid origin, while only 5% showed a biphasic histotype. Stage I–II characterized 47% of tumours, while the remaining 53% were classified as stage III–IV. The majority of the patients underwent previously nonradical surgical intervention for diagnostic purposes as biopsy (63%), and lung-sparing surgery pleurectomy/decortication (26%) or decortication (11%) leaving gross residual disease. All patients received systemic pharmacological treatment based on the pemetrexed and cisplatin chemotherapy. The RHRT was administered 4–6 weeks from the chemotherapy treatment.


**Table 1.** Clinical characteristics of 19 malignant pleural mesothelioma (MPM) patients.

\* Evaluated by ECOG, Eastern Cooperative Oncology Group.

The baseline characteristics of a reference group consisted of 15 MPM patients treated with standard palliative local RT (LRT). They are reported in Table S2. This reference group was characterized by superimposable demographic and clinical characteristics and underwent sparing surgery and chemotherapy treatment analogously to the RHRT group.

#### *2.2. RHRT Effect on Serum Metabolome*

The study of baseline and post-RHRT serum metabolomics profiles aimed to investigate the complex biochemical effects that RT may induce in the host. A targeted serum profile of 188 metabolites covering wide biochemical metabolic pathways was considered for this investigation (Table S1). Twenty-seven metabolites showed concentrations lower than the limit of detection and were excluded from further statistical analyses. Exploratory data analysis performed using principal component analysis (PCA) (Figure S1) did not detect any outliers. The metabolomics profile at baseline and after RHRT resulted homogeneous without any clusters of patients associated with the different diagnostic intervention as well as patients' outliers. However, when the metabolomics profile at baseline and post-RHRT where compared, the PCA model explained only 22% of total variance and did not allow to characterize differences, supporting the application of supervised orthogonal partial least squares discriminant analysis (OPLS-DA) approach. Multivariate OPLS-DA model clearly differentiated the baseline serum metabolomic profiles from those post-RHRT with a significant discrimination power (Figure 1a) (*p* = 0.007, CV-ANOVA). OPLS-DA model showed good performance when internal leave-one-out cross-validation (LOOCV) was assessed (R<sup>2</sup> = 0.77, Q<sup>2</sup> = 0.54) without any potential risk of over-fitting verified by

permutation test (Figure 1b). The extent of the RHRT effects on the serum metabolomics profile was estimated by determining the relative percentage of variation (∆%) of serum concentrations of the metabolites with VIP > 1 that most contribute to the OPLS-DA model (Figure 1c). After RHRT, the 52% of investigated metabolites showed a serum variation of ≥10%; among them, 14 were upregulated, while 69 were downregulated. All these changes were not found associated with the gross residual disease indicating that they could not be attributed to the tumour extent but likely to host metabolic response.

The wide decrease in serum metabolites concentrations after RHRT is clearly indicated by the heat map for the selected statistically significant metabolites (Figure 1d). Only a small set of metabolites, mainly belonging to the amino acids class, significantly increased after irradiation. It included the aromatic AA phenylalanine and tryptophan, the branched amino acids (BCAA) valine and leucine and alpha aminoadipic acid, methionine and carnitine. Conversely, almost all the phospholipids, encompassing PC, lysoPC and SM, significantly decreased after RHRT. Among amino acids, citrulline resulted the most altered metabolite (*p* = 9 × 10−<sup>6</sup> , *q* = 0.001), followed by taurine (*p* = 3 × 10−<sup>5</sup> , *q* = 0.002) and the acyl-carnitines C14 (*p* = 0.002, *q* = 0.04), C18:1 (*p* = 0.002, *q* = 0.05) and C18:2 (*p* = 2.8 × 10−<sup>4</sup> , *q* = 0.02), while for phospholipids, the significant changes regarded the PC ae C42:5 (*p* = 0.001, *q* = 0.03), PC ae C44:5 (*p* = 0.001, *q* = 0.03) and PC ae C44:6 (*p* = 3.7 × 10−<sup>4</sup> , *q* = 0.02) derivatives (Table 2). The individual concentration variations of such metabolites are shown in Figure S3 where it is possible to appreciate the homogeneous decreasing trend for each patient as a consequence of the RHRT treatment.

#### *2.3. Metabolic Patterns Influenced by RHRT*

All the metabolites significantly altered after RHRT were considered for the metabolic set enrichment analysis addressed to elucidate the biochemical pathways most influenced by RHRT. The Over Representation Analysis (ORA) indicates that polyamines biosynthesis, urea cycle as well as arginine and proline metabolism were the pathways more significantly perturbed by RHRT (Figure 2a). The polyamines biosynthesis pathway resulted downregulated, indeed the serum concentrations of putrescine, spermidine and spermine were significantly lower in post-RHRT serum samples (Figure 2b). Polyamines biosynthesis is linked to the urea cycle through ornithine whose serum level was found 12.3% lower post-RHRT (*p* = 0.02, *q* = 0.113). This latter amino acid is also the precursor of both citrulline and arginine. However, while arginine concentration remained constant and independent from the RHRT, citrulline underwent a dramatic drop (33%) (*p* = 9.0 × 10−<sup>6</sup> , *q* = 0.001). Such citrulline depletion was found highly correlated with that of ornithine (*r* = 0.72, *p* = 0.0005) but it was not associated with the common precursor glutamine, which did not undergo significant variations after RHRT. Analogously, proline, a further ornithine precursor, resulted in 19.2% lower post-RHRT compared with its baseline level (*p* = 5.3 × 10−<sup>3</sup> , *q* = 0.078).

**Figure 1.** Orthogonal partial least squares discriminant analysis (OPLS-DA) score plot discriminated serum metabolomics profiles (*n* = 19) at baseline (T<sup>0</sup> , blue) and post-radical hemithoracic radiotherapy (RHRT) (T<sup>1</sup> , green) (**a**). Internal validation by permutation test showed R<sup>2</sup> (green) and Q<sup>2</sup> (blue) values from the permuted models (bottom left) significantly lower than the corresponding original model (top right) (**b**). Percentage variations of metabolites altered by RHRT (**c**). Heat map plot of the significantly changed serum metabolites between T<sup>0</sup> samples (left) and T<sup>1</sup> samples (right) ranked by *t*-test. Metabolites significantly decreased were in green, while metabolites significantly increased were in red. The brightness of the colour corresponded to the magnitude of the difference with the mean value (**d**).


**Table 2.** Metabolites significantly altered as effect of radical hemithoracic radiotherapy (RHRT) in 19 MPM patients.

 <sup>−</sup> − <sup>−</sup> − − − − − − Fold change, metabolite concentration after RHRT divided by baseline concentration; Cit, citrulline; Orn, ornithine; Pro, proline; Phe, phenylalanine; Val, valine; alpha-AAA, alpha-amino adipic acid; Trp, tryptophan; Cn:z, acylcarnitine, *n* = number of carbons, *z* = number of unsaturations; lysoPC Cn:z, lysophosphatidylcholine; PC Cn:z, phosphatidylcholine; SM Cn:z, sphingomyelins; SM OH Cn:z, hydroxylated sphingomyelins. In bold are metabolites with *q*-values < 0.05. <sup>−</sup> , down-regulated; <sup>−</sup> , up-regulated.

  <sup>−</sup>

−

−

−

−

<sup>−</sup>

<sup>−</sup>

<sup>−</sup>

−

−

−

−

−

−

−

−

−

−

−

−

−

−

−

<sup>−</sup>

<sup>−</sup>

<sup>−</sup>

<sup>−</sup>

<sup>−</sup>

−

<sup>−</sup>

<sup>−</sup>

<sup>−</sup>

<sup>−</sup>

<sup>−</sup>

<sup>−</sup>

<sup>−</sup>

<sup>−</sup>

<sup>−</sup>

<sup>−</sup>

−

<sup>−</sup>

<sup>−</sup>

<sup>−</sup>

<sup>−</sup>

<sup>−</sup>

<sup>−</sup>

<sup>−</sup>

<sup>−</sup>

<sup>−</sup>

<sup>−</sup>

<sup>−</sup>

<sup>−</sup>

<sup>−</sup>

<sup>−</sup>

<sup>−</sup>

<sup>−</sup>

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

<sup>−</sup>

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

<sup>−</sup>

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

<sup>−</sup>

 

 

 

<sup>−</sup>

<sup>−</sup>

−

−

**Figure 2.** Over Representation Analysis plot from enrichment analysis. Bars represent matched pathways coloured according to their significance values, with gradations from yellow (low significance) to red (high significance) (**a**). Metabolic pathways altered as effect of RHRT and relative metabolites concentrations prior- (T<sup>0</sup> ) and post-RHRT (T<sup>1</sup> ) in 19 malignant pleural mesothelioma (MPM) patients. *p*-values derive from the Student's *t*-test, \*\*\* *p* < 0.001, \*\* *p* < 0.01, \* *p* < 0.05 (**b**).

#### *2.4. Serum Metabolome Variations as Function of Radiation Dose*

The amino acids class resulted mostly influenced by the RHRT since 14 out of 36 quantified amino acids and derivatives were significantly altered. Conversely, such widespread effect did not occur in the reference group subjected to palliative LRT at the dose of 21 Gy in 3 fractions. In this latter group, none of the metabolites included in the metabolomics profile analysis underwent significant variations, except citrulline that showed a slight (<10%) decrease (*p* = 0.04, *q* = 0.72) (Figure S4a). The mean fold changes of the metabolites belonging to the amino acids class in the two investigated groups are displayed in Figure S4b where it is evident that the RHRT produced higher variations on serum amino acid metabolites as compared with the LRT. The mean absolute variation of amino acids derivatives was 11.8% (range: 0.44–31.76%) and 4.0% (range: 0.26–11.8%) for RHRT and LRT treatments, respectively.

#### *2.5. RHRT Metabolomics Alterations and Clinical Outcome*

The clinical outcome, expressed as median overall survival (OS), was 24 months (95% CI, 17–43 months) for the patients who underwent RHRT. At last follow-up, before the metabolomics data analysis, all investigated patients succumbed to the MPM disease. Their overall OS outcome was not found associated with age, tumour stage or performance status. Conversely, OS was significantly correlated with the serum metabolomics variations induced by RHRT. When partial least square (PLS) analysis was applied, the regression

model showed a relevant association between the metabolites' fold changes and the OS, which explained about 60% of interpatients OS variability (Figure 3a). The metabolites that most contributed to the model, as emerged by the loading plot (Figure 3b) and by Pearson correlation analysis were: asymmetric dimethyl-arginine (ADMA), threonine, symmetric dimethyl-arginine (SDMA), putrescine, serine, asparagine and the acyl-carnitines C2, C10:1, C16:2 and C18:1 whose serum variations were positively correlated with OS (Figure 3c).

**Figure 3.** Partial least square (PLS) score plot for the first two latent variables t(1) and u(1), in which each point represents one patient, plotted as scores (or coefficients) from the metabolomics fold changes data (X block) vs. the score from the overall survival (OS) (Y block). Colour gradations from blue to red represents increasing values of OS (**a**). PLS loading plot. Each point is a metabolite plotted as the coefficient from PLS LV1 (first latent variable) vs. the coefficient from LV2 (second latent variable). Metabolites in the top right (highest positive coefficient) or in the bottom left (lowest negative coefficient) have a strong correlation with the OS (**b**). Metabolites fold changes most correlated with OS by Pearson correlation analysis (**c**).

When the patients' population was stratified according to OS quartiles, the mean foldchanges of these specific metabolites were 0.77 ± 0.07 (range: 0.64–0.90) for the patients' group with OS < 16.9 months (Q1), 1.01 ± 0.14 (range: 0.81–1.26) and 1.23 ± 0.17 (range: 0.98–1.47) for those with OS between 16.9 and 28.8 months (IQ) and > 28.8 months (Q4), respectively (Figure S5). Interestingly, when the analysis was focused on the entire class of amino acids, the overall variations for the long survival patients (16.9–43.17 months) resulted 22% higher than that observed for short survival patients (OS < 16.9) (Figure 4).

**Figure 4.** Amino acids mean overall variations of long survival patients normalized to those of short survival patients. Long survival patients belong to IQ (interquartile) and Q4 OS groups; short survival patients belong to the Q1 OS group. Amino acids statistically significant (*p* < 0.05) are highlighted in grey \*\*\* *p* < 0.001; \*\* *p* < 0.01, \* *p* < 0.05.

#### **3. Discussion**

The application of the ionizing radiation to tumour and surrounding normal tissues elicits complex responses that, behind the DNA cytotoxic activity, disrupt tumour metabolic processes and influence the overall host biochemistry [24].

The present study highlighted that in MPM patients the RHRT treatment is able to produce remarkable systemic metabolomics alterations involving a large set of biochemical pathways as a result of its activity on both normal and tumour tissues. This RT modality was found to produce an overall serum decrease in almost all investigated metabolites. Such depleting effect was commonly observed also for other RT treatment and cancer types suggesting that these serum metabolome drop phenomena could represent a distinctive tract of the host systemic metabolic response to irradiation [18,20–23,25]. Among the metabolites found altered after RHRT, only a small set that included valine and leucine was upregulated. These BCAAs are involved in protein metabolism, energy production and in various biosynthetic pathways which are all overactivated in tumour cells [26,27]. The high BCAA catabolism reported in tumour tissues has been found associated with a systemic deprivation of these amino acids [27–29]. In this context, the post-RHRT increase in valine and leucine may suggest a reduced tumour demand of BCAA as analogously observed in breast cancer patients where a specific raise of serum isoleucine, leucine and valine to normal range was reported after RT treatment [17]. Beyond BCAA, phenylalanine was another essential amino acid significantly increased post-RHRT. This amino acid increases during inflammation conditions [30,31] and its serum levels were found correlated with those of immune activation markers such as neopterin and isoprostane-8 [32,33]. The radiation exposure is known to promote oxidative stress leading to an acute inflammatory status [34,35] and mounting evidence indicates as the RT itself could also stimulate the immune system [36–39] that may be indirectly mirrored by the increase in serum phenylalanine observed after RHRT.

Notably, RHRT was found to influence extensively the lipid metabolism, and in particular, choline-containing phospholipids such as PC, lysoPC and SM derivatives, which underwent a significant serum concentration drop likely associated with the elevated lipids membrane turnover consequent to radiation tissue damage. Thus, this effect may not be specific for RHRT treatment but rather a common trait of the radiation exposure, since a wide lipids drop was also reported in other metabolomics studies regarding different RT treatments [18,19,22,40]. Phospholipids have not only a structural role in the cellular

membrane but they also act as signal-transducer metabolites in different cellular pathways including apoptosis [41–43]. In this context, their downregulation, consequent to the RHRT treatment, may contribute to disrupt tumour signalling pathways and synergize with the radiation cellular killing effect [44]. Beyond phospholipids, the acyl-carnitines metabolite class was also found perturbed by RHRT. These lipids derivatives play a critical role in energy production working as a shuttle of fatty acids into the mitochondria, where they undergo β-oxidation for ATP production. In the investigated series of patients, the concomitant increase in carnitine precursor and the overall decrease in acyl-carnitines derivatives may suggest an alteration in their synthesis likely due to low availability of free fatty acids or acetyl-CoA intermediates that may be diverted to restore the phospholipids pool. Despite the broad perturbation of lipid metabolism, only amino acids-related pathways emerged from the set enrichment analysis. These involved the polyamine biosynthesis, urea cycle, and the arginine and proline metabolism that are strictly interconnected to each other sharing arginine as central metabolite. This latter is synthesized in the kidney and, besides its involvement in the urea cycle for ammonia detoxification, it is the substrate for other essential cellular metabolic pathways such as nitric oxide (NO) production [45]. The main endogenous source of arginine is citrulline that was the metabolite subjected to the highest decrease after RHRT. This nonproteogenic amino acid is synthesized in the small intestine from glutamine and ornithine precursors, but it can be also produced in other tissues as a recycled product of the NO synthesis [46]. The citrulline depletion after RHRT might be attributed to the reduction in its enterocytes biosynthesis as a side effect of the high-dose radiations that partially reach the hemithorax surrounding organs. Indeed, citrulline is a well-known biomarker of intestinal failure and a decrease in its blood concentration was registered in inflammatory bowel diseases [47] as well as in patients who received chemotherapy [14,48–50] or RT [51,52]. The systemic loss of citrulline does not seem to affect the arginine synthesis, since its level was unvaried by RHRT, suggesting that the host metabolism maintains a systemic reservoir of such semiessential amino acid at the expense of citrulline.

The RHRT effect is not limited only to arginine pathways but encompasses the whole class of amino acids likely consequent to the high dose of RHRT. Indeed, the patients treated with a low palliative dose of LRT did not show such significant alterations in the observed time-frame compared with those who received the high dose of RHRT. However, a modest but significant citrulline serum shortage was detected also in the LRT group where the involvement of the intestine was negligible suggesting other citrulline fates. Interestingly, citrulline has been revealed to exhibit antioxidant properties working as a suicidal radicalscavenger [53,54], thus in both RHRT and LRT groups, it may be oxidized by the reactive oxygen species (ROS) produced over the RT treatments. In addition, the radiolysis of the protein lysine-residuals leads to the release of alpha-aminoadipic acid [55,56] that was significantly high in the serum post-RHRT. Radiation oxidative injuries could be also suppressed by the sulphur amino acid taurine that works as antioxidant by reinforcing the endogenous radical-scavenger cellular systems [57–60] as recently demonstrated in lung tissues animal model [61]. Therefore, the taurine serum exhaustion after RHRT may indicate an increased tissue up-take of this amino acid as a consequence of the oxidative stress induced by the treatment.

Ornithine, another amino acid belonging to the arginine metabolism, was found significantly affected by RHRT treatment. Its serum decrease was found significantly correlated with that of citrulline being ornithine, together with glutamine, the principal precursor of citrulline [46]. Ornithine can be synthetized also from proline [62], which significantly decreased post-RHRT, suggesting that the radiation treatment can influence the whole ornithine biosynthetic pathways. The ornithine shortage may have relevant consequences because it represents an important precursor of putrescine, spermidine and spermine, collectively called polyamines. These cationic amino acid derivatives play a key role in cell proliferation [63] and the pharmacological inhibition of their synthesis has been demonstrated to induce tumour growth suppression in xenograft models of MPM [64]. In the context of this study, the significant downregulation of the polyamines, due to the low availability of their common precursor ornithine, may be translated into a potential inhibition of tumour growth with beneficial effect on MPM disease control.

Taken all together, these serum metabolomics changes seem to reflect the overall host metabolic mobilization not only to deal with the radiotoxic effects but also to indirectly control the MPM tumour growth. In agreement with this hypothesis, the individual metabolic response to RHRT could have implications in determining the patients' clinical outcome. Indeed, a significant association between the metabolic profile variations and the OS was found for the amino acids and acyl-carnitines derivatives such as ADMA, threonine, SDMA, putrescine, asparagine and serine as well as the acylcarnitines C2, C10:1, C16:2 and C18:1. The patients' groups with medium and high OS showed a higher increase in these metabolites after RHRT indicating that a greater metabolic response to RHRT may yield a better outcome. This observation can be extended to the whole amino acids metabolism further supporting that in long survival patients, the RHRT stimulates a highly dynamic metabolic response likely associated with a superior individual biochemical resilience. Such metabolic activity can be attributed to their high biological reserves availability that allows not only to better contrast the stress but also to integrate the potential stimulating effects of RHRT.

The low sample size of the present investigation does not allow to properly validate the results, and longitudinal studies with a larger cohort of patients are needed before the discovered systemic metabolomics signatures may find definitive clinical applications. Further investigations have to include time-series analyses along the RT treatment and the patients' follow-up to distinguish the acute and long-term effects of RHRT on patients' metabolomics profiles as well as to better identify the most powerful prognostic biomarkers. Moreover, the extension of the coverage of the metabolome considering other biological matrix such urine would allow having a full view of the metabolic response to RHRT.

#### **4. Materials and Methods**

#### *4.1. Patients' Population*

This metabolomics study enrolled 34 patients from 2014 to 2018 with histologically confirmed MPM referred to Centro di Riferimento Oncologico of Aviano, Italy, for RT after nonradical surgery and systemic chemotherapy. All patients were enrolled within an ongoing randomized phase III study addressed to assess the OS advantages of RHRT over palliative LRT treatment. A test group of 19 patients received RHRT while a reference group of 15 patients underwent standard palliative LTR. The RHRT treatment was delivered with curative intent by IMRT technique to the hemithorax at the pleural surface level from the lung apex to the upper abdomen at the dose of 50 Gy in 25 fractions. The dose was delivered so that 95% of the planned target volume (PTV) was covered by 95% of the prescription dose. Tumour sites with high-fluorodeoxyglucose avidity received simultaneous integrated boosts of 60 Gy. The LRT for the reference group of patients was delivered at 21 Gy in 3 fractions at the thoracotomy scar level. All patients belonging to the test and the reference groups had good respiratory function and normal baseline renal, hepatic and bone marrow functions. The investigation was carried out in accordance with the principles of the Declaration of Helsinki and with approval from Ethics Committee of Centro di Riferimento Oncologico di Aviano (Clinical Trial code ID: CRO-2013–38). All subjects gave written informed consent.

#### *4.2. Sample Collection*

Overnight fasting sample (5 mL) was collected from peripheral venous blood at the baseline and at the end of RHRT and LRT treatments. The blood was allowed to clot for 30 min at room temperature and then centrifuged at room temperature for 15 min at 2100 rpm. Serum samples were immediately stored at −80 ◦C until metabolomics analysis.

#### *4.3. Study Design*

The study aims to explore the systemic metabolomics effects induced by RHRT in a group of 19 MPM patients. For each enrolled patient, serum targeted metabolomics profile was investigated both at the baseline, before the delivery of the RHRT and at the end of the daily 50 Gy/25 fractions administration. The significant metabolomics changes induced by such RT modality were analysed by univariate and multivariate analysis. The serum metabolomics alterations consequent to RHRT were compared with those of a reference group who received LRT to better distinguish the metabolic pathways specifically induced by RHRT.

#### *4.4. Targeted Serum Metabolomics Profile Analysis*

Metabolomics analysis of serum samples was performed using the Biocrates Absolute-IDQ P180 kit (Life Science AG, Innsbruck, Austria) targeted to 188 metabolites belonging to the following classes: amino acids (*n* = 21), biogenic amines and polyamines (*n* = 19), acylcarnitines (*n* = 40), lysophosphatidylcholines (*n* = 15), phosphatidylcholines (*n* = 77), sphingolipids (*n* = 15) and hexoses (*n* = 1). The list of all measured metabolites is reported in Table S1. Sample preparation was carried out following the manufacturer's instructions. Briefly, after thawing, 10 µL of serum was transferred into a filter on the upper well of a 96-well sandwich plate. A mixture of internal standards labelled isotopically with deuterium, <sup>13</sup>C or <sup>15</sup>N was already present in each well. Nitrogen steam drying of filters was followed by derivatization of amino acids with 5% phenyl isothiocyanate (PTC) and a second drying step. Metabolites were then extracted with 500 µL of 5 mM ammonium acetate in methanol and the extraction solution was filtered and diluted with MS running solvent for the analysis.

The instrumentation consisted of a LC ultimate 3000 (Thermo Fisher Scientific, Milan, Italy) coupled with a 4000 QTAP (AB Sciex Framingham, MA, USA) mass spectrometer. Flow injection analysis coupled with tandem mass spectrometry (FIA-MS/MS) was used for the analysis of carnitine, acylcarnitines, lipids and hexoses, while liquid chromatography with tandem mass spectrometry (LC-MS/MS) was used for amino acids and biogenic amines PTC-derivatives separated in a ZORBAX SB 100 × 2.1 mm column (Agilent, Santa Clara, CA, USA). The triple quadruple operated in multiple reaction monitoring, neutral loss and precursor ion scan modes in positive and negative polarity. The MS/MS signals were integrated by Analyst 1.6.1 software (AB Sciex, Framingham, MA, USA) and quantified using a calibration curve according to the manufacturer's instructions. Quality controls (QCs) at three concentration levels, low (QC1), medium (QC2) and high (QC3), were used to evaluate the performance of the analytical assay using the MetIQ software. Metabolites with serum concentration under the limit of detection were excluded for the statistical analysis.

#### *4.5. Statistical Data Analysis*

Quantitative metabolomics data were preprocessed by log transformation and autoscaling normalization. Unsupervised multivariate PCA of serum metabolomics data was applied to identify outliers. Supervised OPLS-DA was used to classify the metabolomics dataset and build a model able to differentiate serum metabolomics profiles at baseline (T0) and post-RHRT (T1). The OPLS-DA was validated to exclude data over-fitting using LOOCV by evaluation of the goodness of fit (R<sup>2</sup> ) and predictive ability (Q<sup>2</sup> ) values and by random permutation test to verify the true predictive ability of the model. Analysis of variance of cross-validated predictive residuals (CV-ANOVA) was computed for assessing model reliability. Variable Importance in Projection (VIP) that ranks the metabolites contribution in the OPLS-DA model and paired univariate Student's *t*-test were used to identify metabolites whose concentrations differed significantly between T<sup>0</sup> and T1. Multiple testing false discovery rate (FDR) correction was performed according to Benjamini–Hochberg method and a *q* < 0.05 was considered statistically significant unless otherwise specified.

Metabolite Set Enrichment Analysis was applied to detect the relevant metabolic pathways significantly altered by RHRT. All the metabolites selected by VIP > 1 and *p* < 0.05 were imported and matched in HMDB, PUBCHEM, SMPDB and KEGG databases, thus categorized according to SMPDB library. The most meaningful biochemical patterns altered by RHRT were inferred by the ORA plot where the metabolic pathways were ranked according to their significance values.

Association between the serum metabolomics fold-change (ratio T1/T0) and the OS of the patients, calculated from the date of first radiation fraction administration to the death date, was investigated by multivariate PLS analysis between two groups of variables. The Y variable (OS) was predicted using a few linear combinations of X variables (metabolites foldchanges) called latent variables (LVs). The extent of the correlation was evaluated by the regression coefficient of the PLS model (R<sup>2</sup> ), while the validation of PLS regression model was performed by LOOCV and permutation test. The metabolites that best contributed to the PLS model were identified and selected by the loading plot for the component w\*c [1] >0.15 and <−0.15. Data analysis and the statistical evaluations were carried out using SIMCA (Umetrics, v. 14.1) software, GraphPad Prism 7 and MetaboAnalyst v. 4.0 [65].

#### **5. Conclusions**

The results of this first exploratory study support the integration of metabolomics for the clinical evaluation of MPM patients. The metabolomics investigation may contribute to better understand the mechanisms underlying the interpatients OS variability to RHRT treatment and to recognize frail patients as a function of their specific metabolic phenotypes. Further validation of such powerful diagnostic tool could effectively improve the selection of the patients who could not receive clinical benefit from the RHRT treatment moving toward alternative personalized treatments.

**Supplementary Materials:** The following are available online at https://www.mdpi.com/2072 -6694/13/3/508/s1, Figure S1: Principal component analysis (PCA) of metabolomics profiles of 19 malignant pleural mesothelioma (MPM) patients before (T<sup>0</sup> ) and after (T<sup>1</sup> ) radical hemithoracic radiotherapy (RHRT), Figure S2: Metabolites ranked according to their Variable Importance in Projection (VIP) scores in the orthogonal partial least squares discriminant analysis (OPLS-DA) model, Figure S3: Serum concentrations at baseline (T<sup>0</sup> ) and post-RHRT (T<sup>1</sup> ) for metabolites resulted significantly altered in 19 MPM patients, Figure S4: Citrulline serum concentrations at baseline (T<sup>0</sup> ) and post-RHRT (T<sup>1</sup> ) in MPM patients underwent local radiotherapy (LRT) and RHRT, Figure S5: Fold changes of serum amino acids and derivatives expressed in MPM patients under RHRT and LRT treatments, Table S1: Metabolites included in the targeted metabolomics analysis, Table S2: Clinical characteristics of LRT group's patients.

**Author Contributions:** Conceptualization, G.C.; methodology, E.D.G., G.C.; E.M. (Elena Muraro), A.S. (Asia Saorin), A.R., E.M. (Emilio Minatel) and M.C.; formal analysis, E.D.G., G.C., G.M., and A.R.; investigation, E.D.G., A.S. (Asia Saorin), E.M. (Elena Muraro), A.R., E.M. (Emilio Minatel), M.C., and M.T.; data curation, A.R., G.C., and E.D.G.; writing original draft preparation, E.D.G., G.M., and G.C.; supervision, G.C. and A.S. (Agostino Steffan); visualization, E.D.G., G.C.; G.M., E.M. (Elena Muraro), A.S. (Asia Saorin), A.R., E.M. (Emilio Minatel), M.C., M.T., and A.S. (Agostino Steffan); project administration, G.C.; funding acquisition, G.C. and A.S. (Agostino Steffan). All authors have read and agreed to the published version of the manuscript.

**Funding:** This work was supported by the Italian Ministry of Health (Ricerca Corrente).

**Institutional Review Board Statement:** The study was conducted according to the guidelines of the Declaration of Helsinki and approved by the Ethics Committee of Centro di Riferimento Oncologico di Aviano (Clinical Trial code ID: CRO-2013–38, December 2013).

**Informed Consent Statement:** Informed consent was obtained from all subjects involved in the study.

**Data Availability Statement:** The data presented in this study are available on request from the corresponding author.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


*Article*
