**From Interconnection between Genes and Microenvironment to Novel Immunotherapeutic Approaches in Upper Gastro-Intestinal Cancers—A Multidisciplinary Perspective**

**Giulia Accordino 1, Sara Lettieri 1, Chandra Bortolotto 2, Silvia Benvenuti 3, Anna Gallotti 2, Elisabetta Gattoni 4, Francesco Agustoni 5, Emma Pozzi 5, Pietro Rinaldi 6, Cristiano Primiceri 6, Patrizia Morbini 7, Andrea Lancia <sup>8</sup> and Giulia Maria Stella 1,\***


Received: 7 July 2020; Accepted: 25 July 2020; Published: 29 July 2020

**Abstract:** Despite the progress during the last decade, patients with advanced gastric and esophageal cancers still have poor prognosis. Finding optimal therapeutic strategies represents an unmet need in this field. Several prognostic and predictive factors have been evaluated and may guide clinicians in choosing a tailored treatment. Data from large studies investigating the role of immunotherapy in gastrointestinal cancers are promising but further investigations are necessary to better select those patients who can mostly benefit from these novel therapies. This review will focus on the treatment of metastatic esophageal and gastric cancer. We will review the standard of care and the role of novel therapies such as immunotherapies and CAR-T. Moreover, we will focus on the analysis of potential predictive biomarkers such as Modify as: Microsatellite Instability (MSI) and PD-L1, which may lead to treatment personalization and improved treatment outcomes. A multidisciplinary point of view is mandatory to generate an integrated approach to properly exploit these novel antiproliferative agents.

**Keywords:** immunotherapy; genetics; gastric cancer; esophageal cancer; multidisciplinary

#### **1. Introduction**

The definition of upper gastrointestinal (GI) cancers essentially refers to gastric and esophageal tumors. The latter, including both squamous cell carcinoma and adenocarcinoma histologies represents the nineth cause of cancer death worldwide and nearly 40% of patients present with metastatic disease at diagnosis [1–3]. The median 5-year survival rate is 47% in case of early stage diseases whereas it decreases to 25% in locally advanced and to 5% for metastatic disease, respectively. Gastric cancer (GC) remains one of the most common and deadly cancers worldwide. Over one million cases of GCs are diagnosed every year around the world. It is the 5th most diagnosed cancer in the world [4]. The epidemiology of stomach cancer harbors substantial geographical heterogeneity and the 5-year survival rate is around 20%, with peaks of about 65% in Japan and 71.5% in South Korea, due high number of diagnosis in early stage disease revealed by massive population screening programs [5]. The geographic variations are mainly related to differences in environmental factors such as dietary patterns and salt intake, the prevalence of *Helicobacter pylorii* (*H.P*) infection and the virulence of strains, as well as host factors [6]. Overall patients affected by resectable cancer can undergo surgery and perioperative therapy with potentially curative purposes. However, most of GC diagnoses are performed in stage III or IV disease and patients are candidates only to palliative chemotherapy. In metastatic diseases, 5-year survival rate is poor with a median overall survival (OS) lower than 12 months [7]. Genomic and proteomic expression profiles of oncogenic signaling pathways unveiled different molecular subtypes of gastric and gastro-esophageal cancers, characterized by specifically targetable markers [8–10]. The most relevant example regards the HER2 inhibitor trastuzumab, a monoclonal antibody that binds to the extracellular domain of the receptor, which is now approved in United States and Europe as the first-line treatment in combination with conventional chemotherapy for HER2-overexpressing locally advanced or metastatic GCs (about 20% of cases [11]) leading to increased overall response rates and survivals [12]. Nevertheless, the introduction of targeted molecules does not result in increased outcome rates and most phase III clinical trials evaluating molecularly designed agents in GC have failed [13]. In this complex landscape, growing evidence supports the routine use of immunotherapy with checkpoint inhibitors in the treatment of upper GI cancers, although their effective role is, still, poorly understood. The main reason is due to the lack of knowledge on how the genetic asset cooperates with the surrounding stroma giving rise to the highly malignant phenotype which defines these tumors. Here we summarize the already available data on the use of checkpoint inhibitors and discuss more recent findings regarding the use of modern immunotherapy, including adoptive cell therapy and vaccines, alone or in combination with conventional drugs. A deep understating of the complex interaction between tumor microenvironment and genetic heterogeneity in this group of tumors, fully requires a multidisciplinary approach that will allow effective and significant clinical results.

#### **2. How to Diagnose and Stage Upper GI Cancers**

Primary esophageal cancer (EC) constitutes the majority (more than 95%) of all esophageal malignancies. The two main pathologic subtypes of esophageal cancer are squamous cell (ESCC) carcinoma and adenocarcinoma (EAC). The latter can mimic metastases or direct extension from tumors of lung or breast. Adenocarcinomas (AC) represent more than 90% of gastric cancers; considering tumor localization they are subdivided into true gastric AC and GEJ-AC. Growing evidence documents a shift in the anatomical distribution of gastric cancer, which increasingly originates from the proximal stomach near the junction with the esophagus and in parallel an increase of EAC affecting the lower esophagus [14]. Thus, a significant uncertainty might regard the identification of the primary organ site of adenocarcinomatous transformation. Immunohistochemistry (IHC) is helpful in defining pathologic entities in case of undifferentiated cancers from the upper GI tract [15,16]. The most common secondary malignant lesions are associated to localization of lymphoma and sarcoma; metastatic masses arousing in the esophagus are rare [17]. Table 1 summarizes the main morphological and IHC features of primary upper GI cancers. In the case of esophageal adenocarcinoma lesion, differential diagnosis

to establish the putative primary origin takes into consideration the lung, in which cells frequently express the thyroid transcription factor 1 (TTF-1), and breast adenocarcinomas, which are generally positive for estrogen receptor (ER), mammaglobin, gross cystic fluid protein and GATA3. On the other hand, ESCCs carry some of the same features of small cells carcinomas which develop in other organs, particularly in the lung and which differentially express common neuroendocrine markers, namely synaptophysin, chromogranin A and CD56/NCAM. Notably, the TTF-1 expression can be found in a proportion of ESCCs as well; thus it cannot specifically indicate the lung only as site of primary tumor growth.

**Table 1.** A summary of the morphologic and immunohistochemical profiles of upper GI tumors [18–20]. The most common morphologic and immunohistochemical (IHC) traits distinctive of main neoplastic lesions affecting the upper GI tract.


SCC = squamous cell cancer ER = estrogen receptor TTF-1 = thyroid transcriptional factor-1 NSE = neuron specific enolase.

Comprehensive description of epidemiologic, clinic and pathologic features of upper GI cancers goes beyond the scope of this work and is already available in many published review papers. All the data summarized in Table 2. Once diagnosis of esophageal/gastric cancer is accurately confirmed, disease pathologic classification and staging are required to address patients to the better therapeutic approach. Siewert classification is a widely used anatomic classification of adenocarcinoma of GEJ and it is based on tumor location with respect to the gastric cardia. Three types are described: Siewert type I tumors are adenocarcinomas of distal esophagus, Siewert type II tumors are adenocarcinomas of gastric cardias and Siewert type III tumors correspond to sub-cardial adenocarcinomas of proximal stomach infiltrating the GEJ. The most recent WHO histopathological classification (WHO Classification of tumors: Digestive system tumors 2019) modified the conventional Lauren's criteria distinguishing gastric cancer into diffuse and intestinal types: diffuse type was reclassified as "poorly cohesive, including signet ring histotype," while intestinal type was split into architectural types papillary and tubular [21,22]. Previous gastrectomy is a known risk factor for the onset of gastric cancer. The so-called Gastric Stump Cancer (GSC), which occurs in the gastric remnant at least 5 years after the surgery for peptic ulcer, identifies a separate subtype of GC (1.1/7% of diagnosis) which mainly affects men [23–25]. The TNM classification represents the most used staging system for upper GI tumors. Details regarding upper GI staging and classification are available as Supplementary Material as Table S1–S7 [26–30].


**Table 2.** Clinical, anatomic, pathologic and imaging characteristics of upper GI cancers. Main clinically relevant features of esophageal and gastric cancers derived from already available literature data [31–46].

#### **3. Main Mutational Patterns and Regulatory Networks**

#### *3.1. Oesophageal Cancer*

The genomic landscape of ESCC and EAC have been extensively studied through next generation sequencing (NGS) and computational approaches, even though the understanding of the complex network of its driver genes is far to be fully understood. ESCC and EAC display distinct sets of driver genes, mutational signatures and prognostic biomarkers.

Esophageal squamous cell carcinomas resemble squamous carcinomas of other organs more than they did esophageal adenocarcinomas. The work conducted by Cancer Genome Atlas Research Network revealed that ESCC showed frequent genomic amplifications of *CCND1* and *SOX2* and/or *TP63* genes, whereas *ERBB2*, *VEGFA* and *GATA4* and *GATA6* were more commonly amplified in adenocarcinomas [47]. Inactivation of the tumor suppressor *NOTCH1* gene has been reported in ESCC but not in EAC [48]. Interestingly inactivating mutations clustered in defined geographic areas, being more frequent in those ECSSs which affect North American patients than in those aroused in Chinese population. Moreover, germline mutations in the RHBDF2 gene (17q25) which cause tylosis (focal non-epidermolytic palmoplantar keratoderma) have been reported to be markers of genetic familial susceptibility for the early onset of ESSC [49–51].

On the other hand, EAC derives from progressive accumulation of multiple genetic abnormalities and aneuploidy. Comparative analysis show that most mutations found in EAC could be already detected in the matched BE which, - at least under genetic profile - identifies an early phase of malignant transformation [52]. Mutations in the *PIK3CA* oncogene and in the *CTNNB1* gene that encodes for β-catenin are known to occur in BE and changes in several tumor suppressor genes involved in chromatin remodeling, such as *ARID1A* and *SMARCA4* as well as in *TP53* and *SMAD4* are usually found in tissues with high-grade dysplasia and EAC. Oncogene amplification is typically a late event in EAC progression. Coherently, genomes of BE tissues are relatively stable compared to those of invasive tumors, in which almost 40% of the genome is non-diploid. The only common copy number alteration found in BE is 9p loss of heterozygosity (*CDKN2A*) [53,54]. Advanced tumors have an increased copy numbers of several oncogenes (*GATA4*, *KLF5*, *MYB*, *PRKCI*, *CCND1*, *FGF3*, *FGF4*, *FGF19 and VEGFA*) and loss of common fragile sites (*FHIT*, *WWOX*, *PDE4D*, *PTPRD and PARK2*) [55,56]. In conclusion, EACs emerge rather than from the gradual accumulation of tumor-suppressor alterations, from a straighter pathway driven by mutations in *TP-53* gene and subsequent acquisition of oncogene amplifications [57]. In this perspective, EACs strongly resemble the chromosomally unstable variant of gastric adenocarcinoma, suggesting that these cancers could be considered as a single disease entity. However, some molecular features, including DNA hypermethylation, occur disproportionally in esophageal adenocarcinomas. Epigenetic modifications are known to contribute significantly to the pathogenesis of the disease and specific methylation signatures are known to be associated to tumor progression processes and thus emerge as novel actionable markers. Among them, the methylation classifier which encompasses the *TRIM15*, *TACC3*, *SHANK2*, *MCC* and *CDKN2A* gene silencing is differentially reported in the progression from BE to transformed areas and not in normal mucosa [58].

#### *3.2. Gastric Cancer*

#### Genetic Features

Gastric cancer is characterized by an extreme molecular heterogeneity, which is defined by the occurrence of multiple genetic and epigenetic alteration in each disease stage. It should be underlined that 3–15% of all diagnosis refer to familial and hereditary gastric cancers, among which hereditary diffuse gastric cancer (HDGC), gastric adenocarcinoma and proximal polyposis of the stomach (GAPPS) and familial intestinal gastric cancer (FIGC). One third of HDGC is attributed to hereditary *CDH1* mutations [59–61]. Other hereditary syndromes, such as Lynch syndrome, familial adenomatous polyposis (FAP), Li-Fraumeni, Muir-Torre and Peutz-Jeghers syndromes can occur with gastric involvement as well [62–64]. In case of genetic diagnosis, prophylactic gastrectomy might be suggested [65,66]. Within respect to the non-hereditary forms of GC, recent molecular profiling studies have allowed the shift from the conventional histological classification systems to four molecularly-based classification groups: (i) EBV-positive cancers (9–10% of gastric AC) harboring high frequency of *PI3KCA* gene changes (80%), high levels of DNA hypermethylation, mutations in *PTEN*, *SMADA*, *CDKN2A*, *ARIDA* (55%) and *BCOR* (23%) and increased copies of *JAK2*, *ERBB2*, *PD-L1* and *PD-1* genes, (ii) microsatellite unstable (MSI) tumors, accounting for 22% of diagnosis, which mainly arise in women and older patients and frequently carry hypermethylation MLH1 promoter in association with recurrent mutations in the *PIK3CA*, *ERBB3*, *ERBB2* and *EGFR* genes [67–69], (iii) genomically stable (GS) tumors (20% of cases, mainly diffuse-type AC) which mostly affect younger subjects and are enriched with recurrent *CDH1* (37%), *RHOA* (15%) and inactivating *ARID1A* gene

changes. Fusions involving the RHO-family GTPase-activating proteins CLDN18 and ARHGAP26, have been reported as well [70]; (iv) chromosomal unstable (CIN) subtypes [71] which account for 50% of GCs and harbor extensive aneuploidy, *TP53* mutations (71%) and increased copy number of several genes encoding for receptor tyrosine kinases and their downstream effectors as *EGFR*, *ERBB2*, *ERBB3*, *VEGFA*, *FGFR2*, *MET*, *NRAS*/*KRAS*, *JAK2*, *CD274*, *PDCD1LG2* and *PIK3CA* [72]. Overall tyrosine kinase receptors (TKR) are among the most frequently altered oncogenes in GC and identify actionable therapeutic targets. A recent genomic study of gastric cancers identified somatic copy number alterations of seven oncogenes involved in tyrosine kinase/MAP-kinase pathways: *KRAS*, *EGFR*, *HER2*, *FGFR1*, *FGFR2*, *MET* and *IGF1R* [73].

#### *3.3. Targeted-Based Therapeutic Strategies*

Although a deep analysis of genetic basis of targeted therapy in GCs falls beyond the scope of this review, some relevant issues are discussed due not only to their clinically relevant role but mainly to their interaction with microenvironment and immune response. A first example regards the blockade of HER2 signaling which has significantly improved the outlook for esophagogastric cancer patients and has allowed the approval of trastuzumab in HER2-positive metastatic gastric/gastroesophageal junction cancers, as first line approach in combination with cisplatin and a fluoropyrimidine (capecitabine or 5-fluorouracil) [12,74]. HER2 is activated most frequently by increased gene copy number, whereas somatic mutations rarely occur [75]. *HER2* gene amplification in GC is associated with higher invasive and proliferative tumor cell capacity [76]. HER2-overexpression is associated with male gender, intestinal type and well/moderate cell differentiation [77]. Anti EGFR antibodies, cetuximab and panitumumab, combined with chemotherapy did not show benefit in overall survival in first line treatment in metastatic gastric patients, as reported in two phase III trials, EXPANDED and REAL 3 [78,79]. The angiogenesis is another target in the therapeutic strategy against some solid tumors like breast, colon and lung cancer and in some instances; it resulted as good target of therapy. In upper GI cancer Bevacizumab with chemotherapy obtained in one phase III trial better overall response rate but failed to gain benefit in overall survival (OS), the primary endpoint of that study [80]. Ramucirumab also, another antiangiogenic drug, combined with chemotherapy compared with chemotherapy alone in phase III trial in metastatic GC patient chemotherapy naive, showed better progression free survival (PFS) (5.7 vs. 5.4 months) in absence of significant OS improvement [81]. In metastatic patients progressed after platinum-based chemotherapy, ramucirumab plus paclitaxel gained benefit in OS, as reported in RAINBOW, phase III trial [82]. Also used as single agent, in a phase III double blind study, in metastatic patient progressed after standard first line chemotherapy, ramucirumab obtained benefit in OS and good tolerability (REGARD STUDY) [83]. Inappropriate activation of MET signaling occurs in a several cancer types, including gastric cancer and promotes tumor cell growth, survival, migration and invasion, namely the Invasive growth genetic program which is involved tumor spreading and metastatic growth [84,85]. Amplification/overexpression of the HGF-receptor MET rather than mutated gene can activate receptor tyrosine kinase [86,87]. MET overexpression/amplification is more common in intestinal-type GC and reported in diffuse GC [88]. Notably, a cross talk between amplified MET and EGFR, HER2 and HER3 has been described and can establish a signaling network, allowing constitutive PI3K/AKT cascade activation [89]. This observation suggests robust rationale for combinatorial therapeutic approaches against MET and EGF receptor family, at least in metastatic GCs [90,91]. DNA repair *BRCA1*/*2* genes mutations are implicated in defective DNA repair processes and are known to be associated to the susceptibility towards hereditary breast and ovarian cancers and can occur in other sporadic cancers, among which gastric cancers. *BRCA1*/*2* mutations are found in 15% of GCs and are associated with poor patient survival [92]. Overall, *BRCAness*—the phenotypic condition that characterizes some cancers with carry defective caretaker gene functions—is associated to high sensitivity to the antitumor agents which cause double strand breaks of DNA, such as platinum [93]. However this condition suggests that GCs might benefit from either platinum therapy or poly (ADP-ribose) polymerase (PARP) inhibitors,

a family of nuclear proteins with enzymatic, scaffolding properties and recruiting ability for DNA repair proteins and have been already tested in gastric cancer. However, the first results with PARP inhibitors did not provide encouraging results in metastatic gastric cancer according to a phase 3 study (GOLD), in which olaparib was used in combination with paclitaxel, since the study did not meet its first endpoint—defined by increase in overall survival—there being some advantage in those cases featuring low expression of ATM telangiectasia mutated) protein measured by IHC [94]. These results confirm that even in a biomarker-enriched population, GC results in a variety and unpredictable pattern of responses in absence of frankly evident drivers.

#### *3.4. miRNAs as Actionable Biomarkers*

Strong evidence suggests that alteration in micro-RNA (miRNA) expression acts as important hallmark of cancer [95–97]. Expression profiles of miRNAs can distinguish esophageal tumor histology and can discriminate between normal tissue and the transformed one. Moreover microRNA expression might identify patients with BE at high risk for progression to adenocarcinoma [98–100]. MiRNA signatures have been investigated in GC for both diagnostic, prognostic purposes as well as to differentiate histologic subtypes and other gastrointestinal cancers [101–109]. Thus, miRNA signatures might act as diagnostic and prognostic in upper GI tumors, biomarkers as summarized in Table 3.

**Table 3.** MiRNA expression in esophageal and gastric cancers. In each case, expression has been associated with its functional (diagnostic and/or prognostic) value based on literature reports (PubMed search according to the following keywords: esophageal/gastric cancer & miRNA).


**Table 3.** *Cont.*


**Table 3.** *Cont.*


#### **4. Tumor Inflammatory and Immune Microenvironment**

The concept of targeted cancer approach has been centered on the neoplastic cells. This paradigm has been now shifted to a more comprehensive understanding of molecular machinery of cancer development which points out the complex interaction between malignant cells and tumor surrounding stroma, which is essential to support each steps of malignant progression [110,111]. This issue is mainly relevant in upper GI cancers in which, on one hand, detection of actionable genetic drivers is rarely reported and on the other, environmental exposure is known to induce inflammatory responses, which ultimately leads to constitutive activation of cellular pro-proliferative and pro-survival signals.

#### *4.1. Cancer-Related Immunogenic Cascades*

Esophageal cancer cells are considered to display high immunogenicity and can induce massive antitumor immune responses already in the early disease stages. Moreover, all the main cancer-associated risk factors, namely smoking and alcohol, are associated with chronic irritation of the esophageal epithelium and to tissue damage mediated by the consequent production of reactive-oxygen-species (ROS) [112,113]. In addition, changes in the microbiome defined by a relevant decrease of Gram-positive bacteria, are associated to both esophagitis and BE [114] and promote production of lipopolysaccharide (LPS) which in turn, induces inflammation via Toll-like receptor 4 and NF-κB activation. Overall, chronic inflammation activates several cancer-associated signaling pathways [115]. Among them, interleukin 6 (IL-6)/signal transducer and activator of transcription 3 (STAT3) cascades are known to play a relevant pathogenic role in EC. Many different cell types, monocytes, fibroblasts and endothelial cells that reside around the tumor mass produce IL-6. Moreover, EC cells produced both IL-6 and its receptor (IL-6R), thus suggesting that an autocrine/paracrine loop might cooperate in tumor progression and invasion [116,117]. The overexpression of NF-κB (nuclear factor kappa-light-chain-enhancer of activated B cells) transcription factor defines a second mechanism that is known to modulate EC-surrounding microenvironment. Notably, NF-κB is emerging as a potentially effective target since it is involved in regulating cellular apoptosis and angiogenesis [118]. Its main downstream transducers are interleukin-1β (IL-1β) and interleukin-8 (IL-8). The latter, also known as CXCL-8 (C-X-C Motif Chemokine Ligand 8), acts as neutrophil chemotactic molecule and is implicated in the progression of several cancer types, among which EC [119]. Similarly, the activation of IL-1β is associated to tumor growth, chemoresistance and poor patient prognosis [120]. STAT3 and NF-κB converge on several transducers: among them, prostaglandin E, produced by cyclooxygenase-2 (COX-2), which is active in promoting upper GI cancer-related inflammatory reactions and, ultimately, in inducing chemo-resistance [121]. Chronic inflammation is also involved in attenuating anti-tumor immunity, which is orchestrated by several cell populations such as myeloid-derived suppressor cells (MDSCs) and regulatory T cells (Tregs). Expansion of MDSC or immature myeloid cells is modulated by inflammatory mediators (IL-1β, IL-6 and PGE2) [122] and growth factors (i.e., VEGF). These cells can directly inhibit T-cell activation and natural killers (NKs) cytotoxicity, while induce Tregs [123]. The latter are also directly recruited by EC cells through CCL-17 (C-C Motif Chemokine Ligand 1) and CCL-22 (C-C Motif Chemokine Ligand 22) production and by macrophages via the CC-R4 (C-C chemokine receptor 4 receptor) [124]. Moreover, they can derive from the conversion of Th-17 cells when stimulated by TGF-β and IL-6 [125]. Other immune cells, such as tumor-associated macrophages

(TAM), display more pro-tumorigenic functions, such as induction of angiogenesis and promotion of malignant cell invasive capacity. TAM expansion with M2 polarization can occur in the presence of Th2-related cytokines. Furthermore, TAMs and malignant cells both express immune checkpoint molecules as PD-L1/2 that can inhibit T cell activation. Indeed, high PD-1/2 expression in EC [126] has been correlated with decreased CD8+ T cell infiltration [127]. The other checkpoint molecule CTLA-4 most often acts as inhibitory receptor on immune cells; however, its expression has been also reported directly in tumor cells [128]. In EAC patients, the upregulation of Th-2 associated cytokine (e.g., IL-4 and IL-13) promotes M2-differentiated macrophage infiltration. In ESCC patients, increased secretion of tumor-derived macrophage chemoattractant protein-1 (MCP-1) results in TAM infiltration [128]. In addition to the above described cells, which overall feature immunosuppressive behavior, another type of stromal element, the cancer associated fibroblasts (CAFs) negatively modulate antitumor immunity in various cancer types, among which EC [129]. CAFs—in EC—can trigger the expression of fibroblast activation protein (FAP) and, in turn, induces the secretion of IL-6 and CCL2 [130] which are involved in creating an immune-suppressive tumor stroma, mainly characterized by M2 polarization of activated macrophages [131].

Esophageal cancer and gastric cancers are known to carry many common molecular features, which are, more frequently, shared by EACs and intestinal type of gastric tumors [132]. They derived from the inflammation-metaplasia cascade that occurs in the esophageal epithelium in OAC and in the gastric epithelium in intestinal-type GC. Barrett esophagus and OAC may thus originate from a unique gastric stem fraction, originated from the cardia. Within respect to GC, by matching two key elements, which define the tumor-associated immune milieu, namely the tumor-infiltrating lymphocytes (TILs) and the PD-L1 expression level, four different neoplastic subgroups, emerge with specific and prognostic score. The type I (TILs+ PD-L1+) is defined by adaptive immune resistance, quite opposite the type II (TILs− PD-L1−)is characterized by immune neutrality; type III (TILs− PD-L1+), shows intrinsic induction whereas in type IV (TILs+ PD-L1−) other suppressors might have a role in initiating immune tolerance. [133]. Overall, high expression of PD-L1 associated to CD8+, CD3+ and CD57+ TILs and low densities of FOXP3+ TILs represent favorable prognostic factors [133]. As reported above, an increase in the M2 macrophage component predicts poor prognosis, except for signet ring cell carcinoma and mucinous adenocarcinoma in which it has been associated to a favorable outcome [134].

#### *4.2. The Role of Extracellular Vesicles*

Extracellular vesicles (EV) cooperate in modulating the crosstalk between GC cells and surrounding stroma. They are secreted by several cell types and released to the extracellular space; based on their size they are defined as: exosomes (30–100 nm in diameter), microvesicles (MVs, 100–1000 nm in diameter) and apoptotic bodies (1000–5000 nm in diameter). The smallest type, the exosomes, are nano-sized vesicles, which are enveloped by a lipid bilayer and are, then, secreted from the plasma membrane into the extracellular space. They play an important role in GC onset and progression [14] mainly through overexpression of multiple proteins, miRNAs and LncRNAs [135,136]. Interestingly it has been documented that they actively promote distant growth of neoplastic clones. In detail, Zhang et al. showed that epidermal growth factor receptor (EGFR)-containing exosomes secreted by GC cells can be delivered into the liver, where they were ingested by liver stromal cells. Here, EGFR, by inhibiting miR-26a/b expression, activated hepatocyte growth factor (HGF) [137]. The latter, through a quite paracrine loop, bound its receptor MET expressed on the migrated cancer cells thus triggering the MET-driven invasive growth process [138].

#### *4.3. Modulation of Tumor Microenvironment by Ionizing Radiation*

In this complex context, the role of ionizing radiation and its interaction with TME emerges as relevant, both locally and under the perspective of its potential abscopal effect. Indeed radiotherapy (RT) represents one of the main treatment strategies in the therapeutic management of oncologic

patients, among which those affected by upper GI cancers. Although primarily addressed to kill cancer cells, ionizing radiation also regulates the expression of the different immune cells normally recruited at the periphery of the tumor [139]. Such interactions are likely to impact on tumor growth/dissemination and the capability of a systemic treatment to be particularly efficacious in tumor control. More specifically, radiation, which can be delivered in different doses and treatment fractions, can from one side act as an in situ vaccine leading to immunogenic tumoral cell death; this event is responsible for the release of specific tumor associated-antigen and other molecules (DAMPS) which activates antigen-presenting cells (APC) which ultimately lead to CD8+ Cells activation. Besides, radiation can not only stimulate intratumoral infiltration of macrophages but can also lead to an overexpression of both FGF2 and its receptors (FGFRs). This signaling pathway can switch macrophage phenotype from M1 to M2, which is typically associated to resistance to radiation [140]. Moreover, in EC and GC, increased PD-L1 expression levels have been associated to worse response to ionizing radiation, at least in neoadjuvant setting. The mechanistic explanation of this finding has been related to the overexpression of PD-L1. The latter is directly promoted by the interferon-gamma produced by the CD8+ lymphocytes, through the Janus kinases-Signal Transducer and Activator of Transcription proteins (JAK-STAT) pathway. Notably, high PD-L1 expression has been associated with the induction of the epithelial-to-mesenchymal transition phenomenon is required to tumor distant spreading [141].

#### **5. Immunotherapies: Novel Insights and Advances**

Systemic treatment of advanced upper GI cancers encompasses combination of multiple lines of chemotherapy, in absence of standard of care regimens. Combinatorial schedules include platinum and fluoropyrimidine doublets, cisplatin/5-fluorouracil (5-FU) or cisplatin/capecitabine. Trastuzumab is associated in HER2-positive cases. Other molecules, such as irinotecan and taxanes, can be associated with fluoropyrimidines and/or platinum or monoclonal antibodies as ramucirumab (a fully humanized molecule directed against vascular endothelial growth factor receptor 2-VEGFR2) or used as monotherapy for unfit patients (for detail see https://www.nccn.org, [142–146]).

#### *5.1. Immune Checkpoint Inhibition in Clinical Trials*

Immunotherapy with immune checkpoint inhibitors (ICIs) has led to a deep change in therapeutic paradigms of advanced tumors, including that of upper GI cancers. However, until now no validated role for immunotherapy has been approved. About 50% of these tumors express PD-L1 but unlike NSCLC or melanoma, this expression occurs predominantly in the peri-tumor inflammatory stroma while it is minimal on cancer cells [147]. Thus, the specific localization, affects, on one hand, PD-L1 expression as validated biomarker, whereas, on the other, is coherent to the poor responses to ICIs that typically characterize these cancers. Similarly, CTL-4 is considerably expressed in GCs (about 80% of cases but it mostly regards immune stroma cells. Detailed lists of studies evaluating immune checkpoint inhibitors are easily available in literature [148,149]: Table 4 summarizes the first and key clinical trials which evaluate the role of most known ICIs (nivolumab and pembrolizumab) in therapeutic intervention against upper GI tumors. The results from ATTRACTION family trials provide robust evidence for the use of nivolumab in case of first line chemotherapy failure. Overall, they led to the approval of nivolumab as therapeutic option in PD-1-unselected metastatic/recurrent gastric cancer in Asian population (Japan, Taiwan and Korea) [150,151]. The PD-binding monoclonal antibody pembrolizumab has been reported, by the KEYNOTE series trials, to add an advantage in patient outcome when used in advanced disease, mainly in those tumors which overexpress PD-L1. However, in the KEYNOTE061 trial [152], pembrolizumab failed to provide a survival benefit over paclitaxel in advanced GC patients who had progressed after first line treatment with standard chemo. More recently, the novel anti PD-1 antibody, toripalimab, has demonstrated a safe profile and promising antitumor activity in patients with advanced GC alone or in combination with conventional chemotherapy schedules. In this context, the high tumor mutational burden (TMB) emerged as powerful predictive marker of overall survival (OS) [153]. The phase III, randomized

JAVELIN Gastric 300 trial has been the first comparing avelumab, an anti-PD-L1 antibody, with chemotherapy in the third-line setting in advanced GC/GEJ cancers. Avelumab failed in improving OS but demonstrated an anti-tumor activity comparable to that of chemotherapy with a more advantageous safety profile [154]. The combination of two different ICIs (anti PD-1 and PD-L1 or anti CTL4) have shown, until now, controversial results. The association of tremelimumab (anti CTLA-4) to durvalumab (anti PD-L1) did not add significant advantages in chemo-refractory GC and GJE cancers [155], whereas the combination of nivolumab and ipilimumab demonstrated more favorable safety and efficacy profiles [156]. Although further investigations are required, combinatorial approaches are now under investigation even in adjuvant settings [157,158]. Among phase III studies, the KEYNOTE-585 trial (NCT03221426) is evaluating perioperative administration of pembrolizumab plus chemotherapy [159] and the Asian ATTRACTION-05 trial (NCT03006705) is comparing S-1/CAPOX (capecitabine plus oxaliplatin) plus nivolumab vs. S-1/CAPOX plus placebo as postoperative approach. Furthermore, two randomized phase II trials are currently ongoing: the DANTE trial (NCT 03421288) evaluating peri-operative use of atezolizumab (anti-PD-L1 antibody) combined with FLOT (docetaxel, oxaliplatin, leucovorin and 5-fluorouracil) [160] and the IMAGINE NCT04062656 randomized, four-arm, chemotherapy-controlled modular trial in subjects with histologically confirmed, resectable GC or GEJ adenocarcinoma. An increase to 35% is estimated to be clinically relevant when patients are treated with either nivolumab in combination with chemotherapy or nivolumab and another immuno-oncology agent (e.g., ipilimumab or relatlimab) [161,162]. Great interest is now addressed to the combination of ICIs and targeted molecules, which seems to be promising although findings are still afar to be conclusive. The combination of durvalumab, targeting PD-L1 olaparib, a PARP (poly ADP ribose polymerase) inhibitor, seemed to be well tolerated in absence of serious adverse event as demonstrated by the phase II MEDIOLA basket trial, which included advanced GCs [163]. Similar results, in terms of safety and clinical activity, have been obtained by adding durvalumab to targeted VEGFR2 inhibitor ramucirumab [164].


**Table 4.** Main clinical trials evaluating ICI in upper GI cancers [165–172].


**Table 4.** *Cont.*

Overall, there is an extreme heterogeneity regarding the efficacy of immune checkpoint inhibition in upper GI cancers. It should be noted that published data are highly heterogeneous within respect to disease stage, treatment schedules, different methods of evaluation of PD-L1 expression levels (tumor proportion score (TPS), combined positive score (CPS), different antibodies used for PD-L1 immunostaining, heterogeneity of considered cells (tumor, stroma and immune cells), cut-offs for positivity (1–50%). Moreover, results could be biased by the fact most of the trials have been conducted only in Asia. The findings also reflect the heterogeneity of the patients enrolled in the trials, which led to controversial results concerning the prognostic implications of PD-L1-expressing tumors. From published data however, several issues deserve main special attention. A relevant example regards a meta-analysis of 15 studies (the vast majority enrolling Asian patients) performed by Gu et al. Overall the authors analyzed 3291 patients and a tremendous heterogeneity in PD-L1 IHC positive expression was reported (from 14.3% to 69.4%) mainly as a consequence of the cut-off values used in different studies (ranging between >1% and >50%) [173].

#### *5.2. Tumor Mutational Burden as Actionable Targets*

As above mentioned, tumor mutational burden (TMB) behaves as effective biomarker for response to anti-PD-L1 treatment in diverse tumor types and in chemo-refractory GCs [157,174]. Accurate TMB measure requires next generation sequencing techniques (NGS), thus surrogate markers are under investigation for routine sample management. Among them, the TGFB family members (TGFB1, TGFB2 and TGFB3) are active transducers in the epithelial-to-mesenchymal transition (EMT) process. Overexpression of TGFB2 has been reported to be positively associated with EMT status and negatively with TMB levels in GC. It affects TMB levels by regulating the DNA damage repair pathways and immune infiltrates, thus suggesting that detection of TGFB2 expression may predict response to ICIs in GC patients [175]. Furthermore, immune checkpoint inhibitors have been used to treat advanced GCs carrying high-frequency microsatellite instability (MSI-H) or mismatch repair defects (dMMR). Microsatellites are short tandem repeats of DNA, which are mostly located in the non-coding genetic or near the chromosome telomeres and their instability defines hyper-mutable phenotype likely caused by defects in mismatch repair (MMR) [176]. The presence of defective MMR genes, which affect about 17–21% of GCs [177] increases the occurrence of somatic mutation to a mean value of more than 1780 compared to 70 changes that can be found in non-defective lesions [178]. It might predict response to anti PD-L1 agents since the occurrence of genetic changes can potentially allow to encode for not-self immunogenic neoantigens.

#### *5.3. Active Immunization Strategies*

The above-described results provide a solid rationale for identifying GC patients who may benefit from ICI therapy based on specific tumor genetic asset [179]. In addition, more recent progresses have been reached in the field of tumor immunotherapy. During the past decade, the definition of a strategy to molecularly identify tumor antigens (TA) recognized by immune cells in patients with cancer lead to dramatic progress in tumor immunology. Active immunization is based on the use of an immunogen to generate a host response aimed at eliminating malignant clones in a controlled way. Several strategies have been developed.

#### 5.3.1. Adoption of Cytokines

A first approach regards the adoption of cytokines (e.g., IFN-γ, IL-10, IL-2) as relevant component of immune response. Indeed, they can directly act on immune cells and modulating their proliferation and signaling against cancer cells. It is well known that cytokines, such as IL-10, are mainly released because of HP-associated chronic inflammation which is implicated in upper GI cancer onset and progression [180]. In this perspective, several ongoing trials are under investigation with both diagnostic and therapeutic purposes. The NCT00197470 study is focused on evaluating the association of the host genetics with the susceptibility to various gastroduodenal disorders, including HP-associated gastric cancer in Japanese population. The study aims at identifying polymorphisms in the IL-1, tumor necrosis factor-alpha (TNF-α) and IL-10 coding genes to clarify the association between those changes and cancer risks to early locate those individuals at higher risks for gastrointestinal malignancies development. Another strategy that is now active in solid cancers among which upper GI tumors regards early detection of cancer recurrence by monitoring changes in a panel of circulating inflammatory cytokines (IL-1, 6, 8, 10, 12 and TNF-α) before and after chemo-radiation (NCT00502502). The phase II randomized clinical trial NCT03554395 compares activated CIK (cytokine induced killer cells) armed with anti-CD3-MUC1 bispecific antibody for advanced GCs to evaluate its safety and clinical efficacy. Another ongoing trial (NCT01783951) has been designed with a similar goal, namely, to evaluate the antitumor effect and safety of activated dendritic cell CIKs (DC-CIK) plus S-1-based chemotherapy for advanced gastric cancer. Interestingly, it has been reported that PD-L1 in human GC inhibits cells to cancer progression and improves cytotoxic sensitivity of cancer cells to CIK therapy [181].

#### 5.3.2. Cancer Vaccines

A second promising strategy is related to cancer vaccination. Cancer is a disease of genes and the occurrence of somatic mutations in oncogenes and tumor suppressor genes drives malignant transformation. However, the accumulation of passenger and driver genetic changes generate cancer-specific neoepitopes that are recognized by autologous T cells as not-self: these molecules on the surface of cancer cells identify ideal targets for vaccines [182]. Great interest in addressed towards clinical development of such therapeutic approach. Well known cancer peptides/proteins recognized by CD8+ and CD4+ lymphocytes are, for instance, melanoma-associated antigen (MAGE-3) [183] and HER-2/neu [184]. Several studies are ongoing. The NCT02276300 study is phase I clinical trial which investigates vaccination against HER2-derived peptide in advanced breast and gastric cancer. BVAC-B is immunotherapeutic vaccine, which uses B cell and monocytes as antigen presenting cell and is under investigation in patients with progressive or recurrent HER2/neu positive GCs (NCT03425773 study). The NCT00023634 trial has been designed to determine toxicity of EGFRvIII peptide vaccine with sargramostim (GM-CSF) or keyhole limpet hemocyanin (KLH) as adjuvant approach in patients carrying EGFRvIII-expressing upper GI cancer. Although not fully documented in upper GI cancers, the variant III of the EGFR receptor seems to behave as oncogene in several solid tumors [185]. Another vaccination strategy aims at using epitope peptide restricted to HLA-A\*0201 and a first I trial has confirmed the feasibility and safety of this approach [186]. Subsequent phase II trial is ongoing (NCT00681252). Vaccination using survivin epitope peptide might induce cytotoxic T lymphocytes (CTL) from peripheral blood mononuclear cells of healthy donors. It exhibited specific lysis against HLA-A2 matched tumor cells in vitro and in primary cell cultures derived from GC patients [187]. Vaccination with autologous tumor-derived heat shock proteins (HSPs) is another novel promising approach in GC. The HSP gp96-peptide complexes, as chaperone, can specifically interact with antigen-presenting cells (APCs) and induce their activation. This process allows the secretion of several cytokines and chemokines which, in turn, promote CD4+ and CD8+ T-cell antitumor immune response [188]. This approach resulted safe and advantageous in neoadjuvant settings when combined with conventional chemotherapy in patients affected by les aggressive diseases [189]. Some trials have investigated the use of vaccines against dendritic cells (DCs), which infiltrate tumor stroma. Importantly, the DC density predicts GC prognosis, being higher levels associated to improved OS [190]. An ongoing trial (NCT03185429) aims at learning about the safety and tolerance of autologous TSA-DC cell and evaluates the efficacy and feasibility of the cell therapy compared to standard regimens. Preclinical [191,192] and clinical studies [193,194] have demonstrated that DCs transfected with stabilized mRNA coding for tumor-associated antigen/whole tumor RNA can generate potent anticancer immune responses. In theory, RNA-based vaccines present some potential benefits if compared to classical vaccination approaches: (i) they are pharmaceutically safer, since they cannot integrate with DNA and seem to be active in absence of serious adverse event; (ii) they can target multiple tumor-associated epitopes; (iii) they are not MHC-restricted. However, their clinical application has been limited, until now, by difficulties in obtaining stable and efficient mRNA delivery and a technical improvement is required before fully reaching the clinical scenario [195,196]. More integrated strategies encompass combination of vaccines with standard chemotherapy, which aims at exploiting the above-mentioned potentiality of chemotherapy to upregulate tumor immunogenicity. Notably, a preliminary treatment with conventional chemotherapeutic agents can promote ICI sensitivity, as widely demonstrated in NSCLC [197] and in BRCA1-deficient triple-negative breast cancer models [198]. In adjuvant setting in GC, several combinatorial trials are ongoing. The combination of an adjuvant bacille Calmette-Guérin (BCG) vaccine with chemotherapy can improve OS when compared to chemotherapy alone [199]. Similar results have been obtained with vaccination with gastrin-17 diphtheria toxoid (G17DT)-targeting gastrin peptide combined with chemotherapy [200]. Chemotherapy treatment can sensitize to vaccine against tyrosine kinase receptors, as well. For instance, vaccination using peptides derived from human VEGFR 1 and 2 combined with standard chemotherapy can significantly increase the OS of patients carrying advanced GCs [201]. Preliminary results from vaccination with IMU-13, a structure made of three individual B-cell epitope peptide sequences selected from HER2/neu receptor, plus chemotherapy vs. chemotherapy alone is ongoing on upper GI cancer patients [202]. Finally, attempts of combinations of different novel immunotherapeutic strategies are under investigation. In vitro and in vivo strategies have been adopted to enhance immune response to a low immunogenic tumor cell line obtained from a spontaneous gastric tumor of a CEA424-SV40 large T antigen (CEA424-SV40 TAg) transgenic mouse. In detail, lymphodepletion has been obtained by treating animals with cyclophosphamide and then

reconstructed by using syngeneic spleen cells. Subsequently mice underwent effective vaccination with a whole tumor cell vaccine combined with GM-CSF. However, recurrence of Tregs should reduce efficacy of this kind of vaccine in long-term perspective [203].

#### *5.4. Passive Immunization Strategies*

Passive immunization is—by definition—induced artificially when antibodies are given as a therapy to a nonimmune individual. Within respect to cancer, this concept refers to the administration of active humoral immunity in the form of pre-formed antibodies or effector lymphocytes against neoplastic clones. Several approaches are under investigation.

#### Adoptive Cell Therapy

The most promising approach of passive immunization regards adoptive cell therapy. The latter provides T cells isolated from a patient, manipulated and expanded in vitro and then re-infused into the patient itself [204]. Adoptive cell therapy (ACT) using TILs refers to the passive transfer into a patient of antitumor T lymphocytes which can virtually destroy the tumor mass. Similarly, to active immunization contexts, concomitant treatment with chemotherapy can increase ACT efficacy in GCs. To sustain this hypothesis, it has been shown that oxaliplatin, by stimulating high-mobility group box 1 (HMGB1) protein to induce anti-cancerous T lymphocytes, can promote immune-mediated apoptosis of cancer cells [205]. Several in vitro and in vivo studies on drug-resistant GCs, demonstrated that the combination of alkylating-like agents with CIK cells induces the release of a high amount of cytokines. It seemed that the T lymphocyte reduction obtained by chemotherapy, can improve the efficacy of ACT therapy by stimulating the persistence of endogenous T cells in circulation, in parallel with a reduction of immune reactions in non-transformed organs. However, these encouraging results were associated with the occurrence of severe infectious adverse events and this point seriously limited the clinical development of this strategy. A more promising type of adoptive T cell immunotherapy is related the use of chimeric antigen receptor (CAR) T cells. The latter are synthetic receptors that can re-program T cells. Their signaling domain enables the CAR T cells to activate effector functions and expand upon recognition of antigens on cancer cells [206]. Results from preclinical studies of the clinical use of CAR T cells against upper GI cancers are encouraging, although this approach requires complex technologies. Moreover, an important issue is the identification of the surface antigen. Targeting therapeutic tumor markers, such as HER2, CEA and DF2, have been carried out in basic and clinical studies. The recently designed bispecific T-cell engagers (BiTEs) identify a class of artificial bi-specific antibodies that are made of two single-chain variable fragments (scFv): the first specific for a T-cell (typically CD3) molecule and the second specific for a tumor-related antigen. The novel secretable BiTE, αHER2/CD3, consists of HER2-specific scFv 4D5, CD3-specific scFv OKT3 and flexible linkers can specifically target HER2+ tumor cells, such as those found in gastric cancer and CD3+ human T cells [207]. Folate receptor 1 (FOLR1), also known as folate receptor alpha and folate binding protein, is a glycosylphosphatidylinositol-linked protein is frequently overexpressed on the GC cell surface and it cannot be found in health areas [208]. Both FOLR1-CAR KHYG-1, a natural killer cell line and FOLR1-CAR T cells has been demonstrated to recognized FOLR1-expressing GC cells in a MHC-independent manner: this fact promotes the release of several cytokines and induce cancer cell apoptosis [209]. PSCA, formerly named as prostate stem cell antigen, is a glycosylphosphatidylinositol (GPI)-anchored cell surface protein belonging to the Thy-1/Ly-6 family. Notably, anti-PSCA CAR-T cells exert strong anti-tumor cytotoxicity in vitro and can impair tumor dissemination in in vivo animal models [210]. Interestingly, CAR T cell approach has been exploited also against mesothelin, that is expressed in GC tissue, both in vitro and in vivo with favorable results defined by strong cytotoxicity and significant regression of GC subcutaneous masses [211]. Within respect to esophageal cancer, EphA2 (erythropoietin-producing hepatocellular receptor A2), which is one of the Ephrin receptor family, is a frequently overexpressed surface antigen. CAR-T cells designed against EphA2 induce the secretion of many cytokines and display a dose-dependent capacity of cancer cell death in vitro [212]. Moreover, it is well known that PD-1 can trigger or inhibit signals, which play a main role in the tumor environment, through combining with PD-L1. This combination can not only block the activation of T cells by blocking the first and second T cell signal but can also assist regulatory T cells (Tregs) to play an inhibitory function and induce helper T cells (Ths) convert to Tregs. The widespread presence of immune checkpoints in a variety of solid tumors, among which upper GI cancers, may be one of the main reasons for the poor effect of CAR-T technology in solid tumors. Recent indications show that bi-specific Trop2/PD-L1 CAR-T cells have the high therapeutic potential against GC [213]. Several clinical studies are ongoing. Among them, the combined phase I and II NCT03706326 trial in advanced EC, aims at assessing the safety and efficacy exploiting combination of immune checkpoint blockade and CAR T cells. In detail, the study evaluates and compares the effects of anti- MUC1 CAR T cells alone, anti- MUC1 CAR T combining PD-1 knockout engineered T cells and PD-1 knockout engineered T cells. The efficacy of this approach is now under investigation also in several trials in gastric cancer patients (NCT02862028, NCT03615313 and NCT03182803).

#### **6. Conclusions and Remarks**

Although a relevant number of genomic alterations are known to be active in upper GI cancers, few actionable targets can be effectively exploited for diagnostic and therapeutic purposes. Growing evidence suggests that immunotherapy could play a relevant therapeutic role alone and in combination with chemo-radiotherapy and other systemic therapies. Viral infection, mutational burden and MSI status are specific players into constant interconnection between tumor and microenvironment, which modulates response to immune checkpoint inhibitors. The therapeutic landscape is rapidly evolving due to constant refinement and validation of molecular biomarkers. The unique context-related malignant behavior that characterizes upper GI cancers drives responses to novel immune and cell therapies. It remains to be clarified if the genetic and immunological heterogeneity may be somehow related to the different anatomic districts that globally defines the upper GI tract. A deep understating of these processes is challenging and requires a multidisciplinary approach. This will lead—in the near future—to more durable clinical responses in a perspective of full treatment personalization.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2072-6694/12/8/2105/s1, Table S1: TNM staging for oesophageal cancer, Table S2: TNM stage grouping for oesophageal cancer, Table S3: Post-neoadjuvant pathologic stage grouping, Table S4: Tumor cell grading for oesophageal cancer, Table S5: TNM staging for oesophageal cancer, Table S6: Pathologic TNM staging (A) and grouping (B) for gastric cancer, Table S7: Pathologic staging following neoadjuvant therapy in gastric cancer.

**Author Contributions:** All the authors contribute to manuscript concept and design; G.A., S.L., C.B. and G.M.S. drafted the manuscript; G.M.S. supervised the work. All authors have read and agreed to the published version of the manuscript.

**Funding:** Ricerca Corrente-IRCCS Fondazione Policlinico San Matteo to G.M.S.

**Acknowledgments:** G.M.S. would like to thank the invaluable support received from Benedetta Marchelli and Elena Morganti.

**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* **Preservation of Organ Function in Locally Advanced Non-Metastatic Gastrointestinal Stromal Tumors (GIST) of the Stomach by Neoadjuvant Imatinib Therapy**

**Nikolaos Vassos 1,2, Jens Jakob 3, Georg Kähler 2, Peter Reichardt 4, Alexander Marx 5, Antonia Dimitrakopoulou-Strauss 6, Nils Rathmann 7, Eva Wardelmann <sup>8</sup> and Peter Hohenberger 1,\***


**Simple Summary:** This study reports a single-center analysis of 55 patients with primary, locally advanced gastric GIST treated with imatinib mesylate (IM) preoperatively for a median of 10 months. The therapy yielded shrinkage of median tumor size from 113 mm to 62 mm. This facilitated 50 patients to undergo significantly less-extensive surgical procedures and resulted in a stomach preservation rate of 96%. The rate of R0 resections was 94% and was followed by a mean recurrencefree-survival time of 132 months with the median not reached. The approach was successful even for patients starting IM during an episode of upper gastrointestinal bleeding. Neoadjuvant IM therapy for locally advanced, non-metastatic gastrointestinal stromal tumors (GIST) of the stomach may play an important role in preserving organ function which might be important for IM plasma levels in an adjuvant or metastatic setting.

**Abstract:** Background: Neoadjuvant imatinib mesylate (IM) for advanced, non-metastatic gastrointestinal stromal tumors (GIST) of stomach is recommended to downsize the tumor prompting less-extensive operations and preservation of organ function. Methods: We analyzed the clinicalhistopathological profile and oncological outcome of 55 patients (median age 58.2 years; range, 30–86 years) with biopsy-proven, cM0, gastric GIST who underwent IM therapy followed by surgery with a median follow-up of 82 months. Results: Initial median tumor size was 113 mm (range, 65–330 mm) and 10 patients started with acute upper GI bleeding. After a median 10 months (range, 2–21 months) of treatment, tumor size had shrunk to 62 mm (range, 22–200 mm). According to Response Evaluation Criteria In Solid Tumors version 1.0 and version 1.1 (RECIST 1.1), 39 (75%) patients had partial response and 14 patients had stable disease, with no progressive disease. At plateau response, 50 patients underwent surgery with an R0 resection rate of 94% and pathological complete response in 24%. In 12 cases (24%), downstaging allowed laparoscopic resection. The mean recurrence-free survival (RFS) was 123 months (95%CI; 99–147) and the estimated 5-year RFS was 84%. Conclusions: Neoadjuvant IM allowed stomach preservation in 96% of our patients with excellent long-term RFS, even when starting treatment during an episode of upper GI bleeding.

**Citation:** Vassos, N.; Jakob, J.; Kähler, G.; Reichardt, P.; Marx, A.; Dimitrakopoulou-Strauss, A.; Rathmann, N.; Wardelmann, E.; Hohenberger, P. Preservation of Organ Function in Locally Advanced Non-Metastatic Gastrointestinal Stromal Tumors (GIST) of the Stomach by Neoadjuvant Imatinib Therapy. *Cancers* **2021**, *13*, 586. https://doi.org/10.3390/cancers13040586

Academic Editor: Carmine Pinto Received: 6 January 2021 Accepted: 28 January 2021 Published: 3 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/).

Preservation of the stomach provides the physiological basis for the use of oral IM in the adjuvant or metastatic setting.

**Keywords:** gastrointestinal stromal tumor; GIST; stomach; neoadjuvant therapy; imatinib; organ preservation

#### **1. Introduction**

Gastrointestinal stromal tumors (GISTs) are the most common mesenchymal tumors of the gastrointestinal tract, arising mostly in the stomach [1,2]. GISTs exhibit a broad spectrum of clinical behavior [2–4] and are characteristically driven by activating mutations of KIT- or platelet-derived growth factor receptor-a (PDGFR-a) gene in approximately 85–90% of cases [5,6]. Surgery was the mainstay of curative treatment of GIST [7,8]. Since 2001, the natural history of GISTs has been dramatically altered through the use of imatinib mesylate (IM), a receptor tyrosine kinase inhibitor of KIT [9–11]. Imatinib is approved for treatment of metastatic or unresectable GISTs and for adjuvant therapy after R0 resection of GIST with significant risk of metastatic spread [11,12].

Although the majority of GISTs are resectable at presentation, a significant number of GISTs are either locally advanced, requiring challenging and complex operations which can lead to postoperative morbidity [3], or the tumors also might present as primarily not resectable with clear margins; however, debulking procedures are difficult to perform due to the high vascularity of growing GIST lesions [13]. In GIST of the stomach, adopting the standard therapy of epithelial gastric cancer, i.e., total gastrectomy may produce a conflict with adjuvant therapy as imatinib plasma levels are significantly below the therapeutic threshold [14].

The use of IM in the neoadjuvant setting can play an important role by downsizing the tumor, in this way decreasing the extent of resection (i.e., organ-preserving operation) [15–25]. Particularly in cases of gastric GISTs, neoadjuvant IM therapy may also convert surgical procedures from an open to a laparoscopic approach.

The purpose of this study was to evaluate the clinic-pathological profile and the surgical and oncological outcomes of patients with gastric GIST who underwent a neoadjuvant IM therapy followed by surgery from a prospectively kept database. We particularly were interested in analyzing a subgroup of patients who had started neoadjuvant therapy in the clinical setting of acute upper GI bleeding from imatinib-sensitive gastric GIST.

#### **2. Material and Methods**

#### *2.1. Patient Selection*

From November 2002 to December 2019, 989 patients with histologically proven GIST were treated by one therapeutic surgical team (PH). Of these, 476 patients had the primary GIST originating from the stomach(Figure 1). In addition to a prospective phase II neoadjuvant study (NCT00112632) [23], we subjected patients who had locally advanced, histopathologically proven gastric GISTs to the similar protocol. The indication was given when patients would have required extensive surgeries (total gastrectomy or multivisceral resection) for curative treatment or when the tumors were ill-located (e.g., GIST at the esophagogastric junction) requiring an abdomino-thoracic approach [26]. Patients with metastatic disease at time of diagnosis or patients treated because of local recurrence of gastric GIST were not included in this analysis.

**Figure 1.** CONSORT statement.

#### *2.2. Clinical Condition*

Only seven patients (12.7%) were asymptomatic and the tumor was detected incidentally (Figure 2). Seven patients had already undergone an exploration (exploratory laparotomy (*n* = 3) or diagnostic laparoscopy (*n* = 4)) at another hospital declaring inoperability or tumor resection with only multivisceral procedure and therefore had been referred to our institution.

It is of note that in 10 patients (18.2%), the GIST was diagnosed due to an acute upper GI bleeding. When there was suspicion on endoscopy and abdominal CT scan, we immediately started with imatinib after endoscopical control of the bleeding and tumor biopsy. Patients who suffered from subacute melena or occult fecal bleeding prompting the diagnosis of GIST were subsumed in the subgroup of tumor-specific symptoms.



**Figure 2.** Demographic, clinical and histopathological data.

#### *2.3. Imatinib Mesylate Therapy and Response Assessment*

The treatment plan of each patient was managed by a multidisciplinary GIST team consisting of surgical oncologists, medical oncologists, pathologists and radiologists. Before treatment started, tumor biopsy was obtained and all patients had confirmed diagnosis of GIST. Mutation analysis was always carried-out when enough tissue material was available. Risk stratification into very low, low, intermediate and high risk followed the NIH-Fletcher criteria for GIST risk assessment [1].

Imatinib was given orally at 400 mg per day, as a single daily dosing; one patient with KIT exon 9 mutation received 800 mg of imatinib. The duration of IM therapy was intended to last 6 months or as long as the tumor was still shrinking in size.

Response to neoadjuvant IM therapy was evaluated 1 month after the treatment start and then every 3 months either with positron emission computed tomography (PET-CT), dual-energy computed tomography (DE-CT) or contrast-enhanced magnetic resonance

imaging (CE-MRI). PET imaging was only used to make sure that the tumor would respond to imatinib. Thus, it was of importance in the earlier patients when the results of mutational analysis took longer than it does today. Response was determined according to the Response Evaluation Criteria In Solid Tumors version 1.0 and version 1.1 (RECIST 1.1) as a complete response (CR), partial response (PR), stable disease (SD) or progressive disease (PD) [27].

#### *2.4. Conduct of Surgery*

Based on the imaging data, removal of the residual tumor was indicated when the maximum therapeutic response was reached (no further reduction in tumor diameters in consecutive imagings of 3 months) or when no further influence on the resectional strategy was expected. A margin of safety of 1 cm was considered enough to spare organ function [28]. The type of surgery was classified according to EORTC STBSG classification: local excision (wedge resection), limited resection (partial resection of the stomach), typical organ resection (total gastrectomy) and multivisceral resection (including adjacent organs) or other (with verbal specification) [29]. Postoperative complications were classified using the Clavien–Dindo classification [30].

The extent of tumor regression was measured at the resection specimen. Complete remission was defined as 100% necrosis (complete absence of viable tumor cells), a near total remission was defined as 95–99% necrosis, subtotal 90–95%, partial remission with 50–90% necrosis and stable disease with <50% necrosis. Resection margin status was defined as R0, R1 and R2 [31].

#### *2.5. Postoperative Drug Therapy*

There was no stringent policy regarding drug treatment after residual tumor resection. We did not continue imatinib in patients with >95% regression of the tumor and discussed further drug therapy on an individual basis.

#### *2.6. Follow-Up*

Postoperative follow-up consisted of a physical examination and acquisition of DE-CT at 3-month intervals for the first 2 years, every 6 months for the next 3 years and yearly thereafter for the next 5 years. Recurrence was defined as recurrent disease in the region of the previously located tumor. Metastasis was defined as disease in distant sites, predominantly liver and peritoneum. All patients were followed-up for a median of 82 months (range, 3–182 months) and were last updated in June 2020.

#### *2.7. Statistical Analysis*

The statistical analysis of the prospectively maintained database was performed with SPSS (version 21). Survival outcomes in terms of RFS was analyzed. RFS was calculated from the date of surgical resection to the date of clinical or radiological evidence of disease relapse, last follow-up or death, whichever occurred first. RFS percentages and treatment effect comparisons were obtained from the Kaplan–Meier method [32] and log-rank test [33]. Date is given as median with range or mean +/− standard deviation. Correlative analytics were obtained by Pearson and Spearman rank-coefficient tests. Differences were considered statistically significant when *p* ≤ 0.05.

#### **3. Results**

#### *3.1. Demographic and Clinicopathological Data*

Fifty-five patients (22f, 33m) with a median age of 58.2 years, (range, 30–86 years) were included in this study (Table 1). Detailed demographic, clinical and histopathological data are listed in Table 1. The median tumor size before start of imatinib was 113 mm (range, 65–330 mm). Mitotic index could be determined in 41 patients (75%); in the remaining patients the size of the biopsy did not allow us to count enough high-power fields (HPFs). According to National Institutes of Health (NIH) consensus [1], the risk classification was

"high risk" in 26 patients (48%), "intermediate risk" in 15 patients (28%) and "low risk" in 13 patients (24%).


**Table 1.** Analysis of Response Data.

#### *3.2. Mutational Data*

Mutation analysis was performed in 49 patients and showed exon 11 mutation in 47 patients with the majority of the mutations consisting of deletions involving codons 557 and 558 (*n* = 15) or point mutation (*n* = 12) (Table 1). In one patient, a SDHB mutation was determined after 2 months of imatinib and the patient was operated on immediately. In six patients mutation analysis could not be carried out from the biopsy and even at the resection specimen it was not feasible due to complete tumor necrosis and significant tumor shrinkage. In another patient who had significant tumor shrinkage, a K642E mutation at KIT exon 13 was found with NGS-sequencing from the residual tumor mass.

#### *3.3. Imatinib Therapy and Clinical Response*

The median time of preoperative imatinib therapy was 10 months (range, 2–21 months) and 52 patients (94.5%) completed the expected treatment duration. Among the patients who did not complete therapy there was the patient with the SDHB mutation whose treatment had to be stopped and a male patient of 69 years (with a history of vascular occlusive disease) who died from a cerebral insult 3 weeks after start of imatinib. Another patient developed perforation of the tumor, located at the posterior part of the stomach, 8 weeks after start of treatment due to extensive tumor regression. He underwent emergency subtotal gastrectomy, partial resection of the diaphragm and splenectomy. Another patient experienced grade 3 skin toxicity which could be resolved by switching therapy to nilotinib. Grade 2 side effects included skin toxicity (*n* = 1) and depression (*n* = 2).

The median tumor size prior to surgery shrank to 62 mm (range, 22–200 mm). According to RECIST 1.1, 39 patients had a PR (75%) while 14 patients had SD (25%) and no patient had progressive disease, see Table 1.

All 10 patients who started neoadjuvant therapy with acute upper GI bleeding experienced no further episode of bleeding and completed their drug schedule until surgery.

#### *3.4. Surgical Data*

Fifty patients underwent surgery after achieving the plateau response. Except for the patient who experienced tumor perforation, all others were operated on at the intended date. Two patients refused surgery. Of these two, one patient preferred to continue with imatinib while another one committed suicide.

At laparotomy, in two patients previously undetected peritoneal metastases were found at the bursa omentalis and the omentum. In 48 patients, total gastrectomy or abdomino-thoracic resection could be avoided, resulting in a stomach preservation rate of 96% (see Figures 3 and 4). The details of the surgical procedures are illustrated in Figure 4 (e.g., treatment option before IM vs. surgical procedure after IM). In 12 cases (24%), the downstaging of the tumor allowed a laparoscopic procedure instead of open laparotomy.

**Figure 3.** 76 year old female, gastrointestinal stromal tumors (GIST) with broad contact to the greater curvature, scheduled for total gastrectomy, left hemicolectomy and left-sided pancreatico-/splenectomy (**A**) prior to and (**B**) after 10 months of imatinib therapy.

> All patients having undergone prior exploratory laparotomy and declared inoperable (*n* = 3) could be resected with clear margins by multivisceral (MVR) total gastrectomy, total gastrectomy alone and subtotal gastrectomy in one patient each. In the four patients who had prior diagnostic laparoscopy (MVR only), segmental resection with Merendino reconstruction (*n* = 1), subtotal (*n* = 1) or segmental (*n* = 1) gastric resection, and MVR (*n* = 1) had to carried out for R0 tumor removal.

> Surgical complications were observed in seven patients (14%) and included postoperative pancreatic fistula (*n* = 2), surgical site hemorrhage (*n* = 2), prolonged pleural effusion (*n* = 2) and wound infection (*n* = 1). Of them, three patients required intervention (grade 3 Clavien–Dindo). None of these patients required reoperation and there was no postoperative mortality.

#### *3.5. Histopathological Data*

Except for the two patients with peritoneal metastases (R2 resection), only in one patient was an R1 resection stated by the pathologist, resulting in an R0 resection rate of 47/50 patients (94%).

The final histopathological report showed no residual viable tumor (pCR) in 12 cases (24%); for details see Table 1. Interestingly, in seven cases there was a less than 50% necrosis observed, despite the fact that even in this subgroup the median tumor size shrank from 84 mm (range, 65–122 mm) to 60 mm (range, 22–115 mm).

Only the mutational type correlates significantly (*p* = 0.037) with the extent of tumor regression. There was no significant relationship of RECIST classification to the difference in tumor size from prior to imatinib vs. post-imatinib (Table 1). The difference between tumor size prior to imatinib vs. at the resection specimen exceeded *p* = 0.05.




**Figure 4.** Comparison of scheduled surgery prior to imatinib vs. surgical procedure performed after neoadjuvant therapy.

#### *3.6. Recurrence-Free Survival*

Of the 50 patients operated, 34 (68%) are alive with no evidence of disease. The most common sites at detection of distant metastases were peritoneum (*n* = 7) and liver (*n* = 6), both combined (*n* = 2), while one patient developed a locoregional recurrence at the surgical site. Three patients have died from their disease, and another four patients from other causes.

Kaplan–Meier curve (Figure 5) demonstrates a mean recurrence-free survival of 123 months (95%CI 99–147 months) with the median not yet reached.

We evaluated prognostic factors with respect to the influence on RFS. However, no statistically significant result could be obtained from initial tumor size (*p* = 0.34), mutational type (*p* = 0.86) and mitotic count (*p* = 0.12). Furthermore, the most logical factor (extent of tumor necrosis) was not proven to be of significant influence (*p* = 0.33, all log-rank).

With respect to adjuvant imatinib therapy after residual tumor removal, the single patient who experienced tumor perforation continued with imatinib and is free from recurrence after 12 years. Another seven patients were recommended to continue with imatinib therapy in order to complete a 36 months total duration of neoadjuvant plus adjuvant drug therapy. Of these, three patients developed hepatic and/or peritoneal metastases and died from their disease after multiple lines of therapy. Another two patients completed 3 years with no evidence of recurrence, one patient chose to stick with the drug until now and one patient is still in the completion phase.

**Figure 5.** Recurrence-free survival.

#### **4. Discussion**

Surgery for primary GIST of the stomach is different from surgery for epithelial gastric cancer. Detailed lymphadenectomy, the mainstay to treat gastric carcinoma, is not required except in a subgroup of patients with Carney–Stratakis syndrome or SDHdeficient tumors presenting typically in young females [34]. On the other hand, GISTs originating from the muscularis layer of the intestine tend to grow luminally with potential acute bleeding or exophytically towards the surrounding organs. As GIST may be a fragile mass and often represents a highly vascularized lesion, larger gastric GIST may require more extended surgery with major morbidity and functional deficits due to the proximity to vital structures or the location in difficult sites (e.g., gastroesophageal junction). This may be the indication for imatinib therapy prior to surgery as recommended by several guidelines [35]. The neoadjuvant administration of IM turns out to be beneficial for patients with locally advanced or marginally resectable, non-metastatic GIST. Proper selection of candidates for neoadjuvant therapy is the prerequisite for successful therapy and requires tumor genotyping based on preoperative biopsy with the mutational spectrum not different from the metastatic situation [36]. In our study, all patients were either diagnosed with imatinib-sensitive mutations of *KIT* or we used 18F-FDG-PET to make sure the expected efficacy would really take place. This was particularly the case in the earlier patients when mutational analysis took more time than today [37,38].

The few formal trials on preoperative imatinib therapy often include both locally advanced and metastatic patients with GIST arising from the whole GI tract [15,21,24,36,39–42]. In the RTOG study [24] only 15 patients were truly treated under neoadjuvant conditions across all locations. Thus, our series comprises the largest patient cohort of locally advanced, nonmetastatic gastric GIST patients treated with neoadjuvant IM therapy followed by surgery. Large GISTs carry an increased risk of intraoperative tumor rupture and dissemination because of their fragility and hypervascularity which has a detrimental effect on disease-free status and overall survival [43–46]. Beyond the organ-saving approach through objective tumor downsizing, preoperative imatinib also improves the integrity of tumor capsule and decreases the risk of intraperitoneal bleeding/tumor perforation, leading to a very high rate of R0 resections [21,46]. We demonstrate a significant regression of median tumor size from 11.3 cm to 6.2 cm, which reflects the main advantage of imatinib as induction therapy in patients with locally advanced GIST. Particularly the subgroup of 10 patients (18.2%) with upper GI bleeding from the tumor profited from this approach. None of them had to be operated on prematurely due to a recurrent bleeding episode and the surgical tumor resection could be moved from an emergency procedure to an elective operation. No postoperative imatinib-related complications were observed.

The approach provided an excellent oncological long-term result. Based on the NIH consensus criteria [1], 48% of the patients could be classified as high risk for tumor recurrence. Even if one uses the contour maps by Joensuu et al. [47], providing a better assessment tool and eliminating the dichotomous threshold of 5 cm and 5 mitoses pro 50 HPF, the mean recurrence-free survival of more than 10 years looks very promising. The basis probably is laid by the fact that 44% of the resection specimen showed >95% necrosis and 94% of the patients have undergone R0 resection. This is due to patient selection with *KIT* exon 11 mutations almost exclusively. It is also known from treating metastatic patients that KIT deletions involving codons 557 and 558 respond very well to imatinib [48] (Figure 2). Our data are in line with a multicenter study including 161 patients with locally advanced non-metastatic GISTs pooled from 10 EORTC-STBSG sarcoma centers showing that >80% of the tumors responded to imatinib, facilitating R0 resection in >80% of the cases [24]. After a median 40 weeks of imatinib, the R0 resection rate was 83% and the 5-year DFS was 65% with median OS of 104 months [24].

Another recently published series on 150 patients with GIST treated on a neoadjuvant basis across all tumor sites reports an overall survival rate of 81% at 5 years [49]. The difference might be due to shorter treatment duration (median 7.1 months with a range starting at 0.2 months) and a clearly lower rate of partial tumor remissions of 40% which was 75% in our series. Furthermore, the resection margins with 63.3% R0 resections and 18% each of R1 and R2 resection are inferior to our study [49]. It has been noted from further studies that patients after R0 resection have a significantly lower risk of developing tumor progression compared to patients with R1/R2 resection (60% vs. 23.8%, *p* = 0.11, [22,40]).

The duration of neoadjuvant imatinib administration may be important to obtain adequate tumor response. An early compilation of case reports by Haller et al. [50] suggested that the longer the treatment the better the remission. We indicated surgery after having reached a plateau with no further tumor shrinkage and the risk of developing secondary resistance to therapy still remaing low [39]. At this time point, all our patients showed either PR or SD and no patient showed any progression during imatinib therapy. The rate of partial responses in our patients is higher compared with the phase II RTOG 0132 trial, in which 83% of patients had stable disease after 12 months of imatinib [15].

A strength of this study is that it demonstrates that laparoscopic procedures more and more can be successfully used in this setting of locally advanced GIST with median tumor size of more than 10 cm after downstaging with tyrosine kinase inhibitors. The study, however, also has limitations referring mainly to patient selection which is hardly avoidable. Patients can easily be convinced to swallow a pill per day and avoid total gastrectomy or multivisceral resection. A randomized trial does not look feasible at all under these circumstances and therefore the formal evidence of using neoadjuvant imatinib is not better than grade 2+ according to SIGN+1 [51].

Given the fact that in small tumor biopsies the number of mitoses could not be counted per 5 mm<sup>2</sup> or 50 HPF, the risk classification of patients according to NIH or Miettinen/Lasota is doubtful in some cases. This also influences the decision of whether or not to subject patients to postoperative adjuvant imatinib therapy. Using the extent of regression from the resection specimen in the seven mentioned patients does not allow us to draw conclusions. The willingness of the patients to continue with the drug also influenced the administration of adjuvant imatinib therapy. Several patients felt relief from the drug and the tumor after surgery and were not willing to continue. Our individualized approach does not allow us to draw further conclusions.

In gastric GIST a problem is the rather high rate of tumors without mutations in KIT or with mutations in PDGFRA. We tried to overcome this with 18F-FDG-PET scanning to eliminate patients who would not respond adequately. We also postponed patients with epitheloid GIST until mutation analysis had been performed, as this feature is often associated with PDGFRA mutations not being sensitive to imatinib.

#### **5. Conclusions**

In conclusion, neoadjuvant imatinib in our series of locally advanced gastric GIST proved to allow organ-sparing surgical procedures with a very high rate of R0 resections and excellent long-term recurrence-free survival. This holds true also for patients starting their treatment during an episode of upper gastrointestinal bleeding. Toxicity was mild and tolerable and in 96% of the patients major parts of the stomach could be preserved, maintaining the physiological basis for the use of oral tyrosine kinase inhibitors in the adjuvant or metastatic setting.

**Author Contributions:** Conceptualization, N.V., J.J., G.K. and P.H.; methodology, N.V. and P.H.; software, N.V. and P.H.; validation, P.H.; formal analysis, N.V. and P.H.; investigation, N.V., P.H. and G.K.; resources, G.K., A.M., N.R., E.W. and P.H.; data curation, N.V., A.M., A.D.-S., N.R., E.W. and P.H.; writing—original draft preparation, N.V. and P.H.; writing—review and editing, N.V., J.J., G.K., P.R., A.M., A.D.-S., N.R., E.W. and P.H.; supervision, P.H. 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 Medical Ethics Commission II, Medical Faculty Mannheim of the University of Heidelberg, Maybachstr. 14, 68169 Mannheim, Reference 2020-827R.

**Informed Consent Statement:** All patients consented to the use of their tissues and data for research purposes.

**Data Availability Statement:** The data that support the findings of this study are available from the corresponding author upon reasonable request.

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

#### **References**


MDPI St. Alban-Anlage 66 4052 Basel Switzerland Tel. +41 61 683 77 34 Fax +41 61 302 89 18 www.mdpi.com

*Cancers* Editorial Office E-mail: cancers@mdpi.com www.mdpi.com/journal/cancers

MDPI St. Alban-Anlage 66 4052 Basel Switzerland

Tel: +41 61 683 77 34 Fax: +41 61 302 89 18