*Article* **High Dual Expression of the Biomarkers CD44v6/**α**2**β**1 and CD44v6/PD-L1 Indicate Early Recurrence after Colorectal Hepatic Metastasectomy**

**Friederike Wrana <sup>1</sup> , Katharina Dötzer <sup>1</sup> , Martin Prüfer <sup>1</sup> , Jens Werner 1,2 and Barbara Mayer 1,2,\***


**Simple Summary:** Distant metastasis in colorectal cancer still correlates with poor prognosis, emphasizing the high need for new diagnostic and therapeutic strategies. In the present study, liver and lung metastases revealed profound differences in the expression pattern of metastasis-driving protein biomarkers. This suggests the adaption of the therapy to the biology of the metastatic organ site. High expression of the cell adhesion molecule CD44v6 and high dual expression of CD44v6, combined with the cell adhesion molecules integrin α2β1, as well as the checkpoint inhibitor molecule PD-L1, correlated significantly with early recurrence after hepatectomy, in a substantial number of liver metastatic patients. These findings suggest the need for the implementation of biological risk factors into clinical risk scores, aiming to make the prognosis of the individual patient more precise. Further, dual expression of protein biomarkers that are druggable, such as CD44v6/α2β1 and CD44v6/PD-L1, can identify high-risk patients for targeted therapy that might provide a survival benefit.

**Abstract:** Considering the biology of CRC, distant metastases might support the identification of high-risk patients for early recurrence and targeted therapy. Expression of a panel of druggable, metastasis-related biomarkers was immunohistochemically analyzed in 53 liver (LM) and 15 lung metastases (LuM) and correlated with survival. Differential expression between LM and LuM was observed for the growth factor receptors IGF1R (LuM 92.3% vs. LM 75.8%, *p* = 0.013), EGFR (LuM 68% vs. LM 41.5%, *p* = 0.004), the cell adhesion molecules CD44v6 (LuM 55.7% vs. LM 34.9%, *p* = 0.019) and α2β1 (LuM 88.3% vs. LM 58.5%, *p* = 0.001) and the check point molecule PD-L1 (LuM 6.1% vs. LM 3.3%, *p* = 0.005). Contrary, expression of HGFR, Hsp90, Muc1, Her2/neu, ERα and PR was comparable in LuM and LM. In the LM cohort (*n* = 52), a high CD44v6 expression was identified as an independent factor of poor prognosis (PFS: HR 2.37, 95% CI 1.18–4.78, *p* = 0.016). High co-expression of CD44v6/α2β1 (HR 4.14, 95% CI 1.65–10.38, *p* = 0.002) and CD44v6/PD-L1 (HR 2.88, 95% CI 1.21–6.85, *p* = 0.017) indicated early recurrence after hepatectomy, in a substantial number of patients (CD44v6/α2β1: 11 (21.15%) patients; CD44v6/PD-L1: 12 (23.1%) patients). Dual expression of druggable protein biomarkers may refine prognostic prediction and stratify high-risk patients for new therapeutic concepts, depending on the metastatic location.

**Keywords:** colorectal cancer; liver metastases; lung metastases; protein biomarker; dual expression; early recurrence; poor prognosis

#### **1. Introduction**

According to international guidelines [1–3], metastasectomy currently offers the best chance for long-term survival for selected colorectal cancer patients. Additional standard chemotherapy for patients with resectable liver metastases resulted in the prolongation of

**Citation:** Wrana, F.; Dötzer, K.; Prüfer, M.; Werner, J.; Mayer, B. High Dual Expression of the Biomarkers CD44v6/α2β1 and CD44v6/PD-L1 Indicate Early Recurrence after Colorectal Hepatic Metastasectomy. *Cancers* **2022**, *14*, 1939. https:// doi.org/10.3390/cancers14081939

Academic Editor: Stephane Dedieu

Received: 9 March 2022 Accepted: 7 April 2022 Published: 12 April 2022

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

**Copyright:** © 2022 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/).

disease-free survival (DFS) and progression-free survival (PFS) but revealed no significant improvement in overall survival (OS) [4,5]. In patients with resectable pulmonary metastases, the outcome of peri-operative chemotherapy is inconclusive [6,7]. However, despite curative-intent metastasectomy, more than half of the patients suffer recurrence [8,9]. This highlights the urgent need for the implementation of new strategies to identify high-risk patients suitable for personalized therapy, aiming to improve treatment outcome and survival [10].

Colorectal cancer preferentially metastasizes to the liver, followed by the lung and the peritoneum and, more rarely, in bone, ovary and the brain [11–13]. The metastatic pattern depends on the sidedness of the primary colorectal tumor. Elucidating the underlying mechanisms of the metastatic organotropism, profound molecular differences were observed between right-sided and left-sided CRC cancers. Similarly, the tumor microenvironment seems to have a deep impact on the metastatic site [14]. Indeed, for primary metastatic colorectal cancer, a growing body of molecular data is available, resulting in the continuous development of targeted therapies and improvement in survival [15,16].

Comparative analysis of primary CRC and corresponding metastatic sites revealed maintenance of the main driver mutations in both liver and lung metastases, some of which are approved for CRC therapy, such as RAS, BRAF and MSI [17–19]. In contrast, genomic [20–22], transcriptomic [23] and proteomic [24] profiling identified molecular differences between primary tumor, liver and lung metastases that might have potential therapeutic implications for specific metastatic sites. Moreover, distant metastases in different organs revealed discordant responses to standard chemotherapy [25], all together, supporting the concept of inter- and intratumor heterogeneity, which is one of the key factors in tumor progression, therapeutic resistance, and poor patient outcome.

In the present study, a panel of protein biomarkers was selected, which drive the complex metastatic process of primary colorectal cancer and lead to poor prognosis. In contrast, little information is available on the expression pattern of these prognostic factors in liver and lung metastases. The protein biomarker panel encompassed the growth factor receptors epidermal growth factor receptor (EGF-R) and hepatocyte growth factor receptor (HGF-R) [26], human epidermal growth factor receptor (Her2/neu) [27], insulin-like growth factor 1 receptor (IGF-1R) [28], estrogen receptor alpha (Erα) [29] and progesterone receptor (PR) [30], the cell adhesion molecules CD44v6 [31], Muc1 [32] and integrin α2β1 [33], the chaperone heat shock protein 90 (Hsp90) [34], and the immune checkpoint molecule programmed death ligand 1 (PD-L1) [35]. Interestingly, the protein biomarkers selected are drug targets, for which drugs are already approved or for which clinical trials are ongoing, in primary colorectal cancer or other cancer types. This could open up new options for second and further line treatments in colorectal cancer.

The present study aimed (1) to identify the phenotypic heterogeneity in tumor biology between colorectal liver and lung metastases and (2) to stratify patients with a high risk for early recurrence after hepatic metastasectomy.

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

#### *2.1. Patient Cohort*

The patient cohort consists of 68 patients with metastatic colorectal cancer, receiving metastasectomy with curative intent at the Department of General, Visceral, and Transplant Surgery, Ludwig-Maximilians-University, Munich, Germany. A liver metastasis (LM, *n* = 53) or a lung metastasis (LuM, *n* = 15) was analyzed from each patient. Doublecoded tissues and the corresponding data used in this study were provided by the Biobank of the Department of General, Visceral, and Transplant Surgery, Ludwig-Maximilians-University Munich, Munich, Germany. This Biobank operates under the administration of the Human Tissue and Cell Research (HTCR) Foundation. The framework of HTCR Foundation, which includes obtaining written informed consent from all donors, has been approved by the ethics commission of the Faculty of Medicine at the LMU (approval number 025-12) as well as the Bavarian State Medical Association (approval number 11142) in

Germany. All liver metastases were diagnosed as the first relapse of the individual patient. Lung metastases represented first (*n* = 3), second (*n* = 8) and later stage relapse (*n* = 4). Survival analysis was performed for 52 patients diagnosed with liver metastases. One patient was lost to follow up. Follow-up period of the patient cohort was from December 2010 until February 2018.

#### *2.2. Immunohistochemistry and Evaluation of Biomarker Expression*

Fresh tumor samples including adjacent benign reference tissue were collected according to international biobanking standards. After surgery the tumor samples were immediately snap frozen in liquid nitrogen. Serial cryosections (5 µm) were performed and air dried over night at room temperature. Sections were either fixed in acetone, or for the ERα und PR staining in formalin solution (10%). Immunohistochemistry was performed using the standard avidin-biotin-peroxidase complex method [36–38]. Briefly, unspecific Fc receptors were blocked with 10% AB-serum in D-PBS, pH 7.4 for 20 min. Endogenous biotin was blocked using the Avidin-/Biotin-blocking Kit for 15 min. The primary antibodies (Table 1) were incubated for one hour. Some antibodies were detected with the secondary biotinylated antibody (111-065-114; wc 7.0 µg/mL; JacksonImmunoResearch, West Grove, PA, USA for anti-rabbit and 315-065-048; wc 0.75 µg/mL; JacksonImmunoResearch for anti-mouse) for 30 min, followed by the peroxidase-conjugated streptavidin (016-030-084; wc 1.0 µg/mL; affymetrix eBiosciences, Santa Clara, CA, USA) for another 30 min. Other primary antibodies were detected with the amplification Kit ZytoChem Plus (HRP060; Zytomed Systems, Bargteheide, Germany) according to the instructions of the manufacturer (marked in Table 1 with Kit: +). For visualization of the antigen–antibody reaction all slides were developed in a 3-Amino-9-ethylcarbazole solution containing 35% hydrogen peroxide (AEC staining) for eight minutes in darkness. Counterstaining was performed with Mayer's hemalum solution. All incubation steps were performed in a humid chamber at room temperature. Specificity of the staining was controlled by the corresponding isotype controls (Table 1). Cancer cells were visualized by EpCAM and pan-cytokeratin expression.

For the evaluation of biomarker expression, the size of the measurement field was standardized using a normalized grid at 100× magnification (Olympus microscope BX50, Olympus, Hamburg, Germany). The biomarker-positive tumor area was determined in relation to the total tumor area. The percentage of biomarker-positive tumor cells was expressed by semiquantitative estimation in 10% increments. Staining results were evaluated by two independent observers (FW, BM). External monitoring was performed by local pathologists (Institute of Pathology, LMU Munich, Munich, Germany, T. Kirchner) and for Her2/neu expression by J. Rüschoff (Institute of Pathology Nordhessen, Kassel, Germany, Rüschoff) [39].

For some biomarkers standardized cut-off values are given, namely ERα and PR [40], Her2/neu [39,41], Muc1 [42,43], and PD-L1 [36,44]. In the absence of standardized cutoffs for other biomarkers, cut-offs were assessed using the biphasic distribution, which was statistically defined using the mean antigen expression in liver or lung metastases. Biomarker expression below the calculated cut-off was defined as low expression, and biomarker expression above the calculated cut-off was defined as high expression. The same cut-off values were used for single biomarker analysis and the evaluation of dual biomarker expression. In addition to the tumor tissue, antigen expression was evaluated on the adjacent benign liver and lung tissues.


**Table 1.** Antibody Panel for Immunophenotyping of Colorectal Liver and Lung Metastases.

#### *2.3. Statistical Analysis*

All statistical analyses were performed with IBM SPSS v. 23. Mean biomarker expression between liver and lung metastases was compared using the Mann–Whitney U-test. The prognostic impact of single and dual biomarker expression was evaluated using Kaplan–Meier analysis (log rank test, 'pairwise over strata') and multivariate Cox regression analysis (biomarker expression used as 'categorical covariate', 'First' as reference category). OS was defined as the time from metastasectomy until the last follow-up or death of the patient. PFS was defined as the time from metastasectomy until the next progression. A *p*-value of ≤0.05 was considered as significant.

#### **3. Results**

#### *3.1. Patient Characteristics*

In the present study, 53 liver metastases and 15 lung metastases surgically resected from colorectal cancer patients were analyzed. Men were more frequently affected than women (LM: ratio 1.79:1; LuM: ratio 4:1). Most (66.04%) liver metastases were detected at primary diagnosis (synchronous), whereas all lung metastases were documented at a later time (metachronous). Liver and lung metastases were diagnosed as single organ metastases. However, at the organ site, tumor disease was frequently extensive (number of nodules within the metastatic organ >1; LM: 64.15%, LuM: 53.33%; multilobular involvement; LM: 56.6%, LuM: 66.67%). Still, most patients were resected with curative intent (R0; LM: 73.58%, LuM: 80%). Further, 32 of 53 (60.38%) patients diagnosed with liver metastases received first-line chemotherapy (5-FU as single agent: 34.38%, oxaliplatin-based: 43.75%, irinotecanbased: 15.63%, others: 6.25%) and 23 of 53 (43.40%) received neoadjuvant chemotherapy before liver metastasectomy. Of these, 10 of 15 (66.67%) patients were treated with front line chemotherapy (5-FU as single agent: 10%, oxaliplatin-based: 80%, others: 10%) and 8 of 15 (53.33%) patients received neoadjuvant chemotherapy, right before surgery of the

lung metastasis studied. Complete treatment records were not available for all patients with lung metastases.

Patient characteristics are summarized in detail in Table 2.



*n*, number of patients; **R-status**, residual status after surgery; **\***, nodules within the metastatic organ; # , administered directly before metastasectomy.

Survival analysis was performed in the patient cohort with liver metastases but was omitted in patients with lung metastases because of small sample size. Patients diagnosed with multiple (>1) LM had a significantly shorter PFS compared to patients diagnosed with a single liver metastasis (multiple metastases, PFS: 6.5 months; single metastasis, PFS: 10 months; log-rank, *p* = 0.014). Patients with synchronous LM relapsed much faster compared to patients with metachronous LM (synchronous, PFS: 7 months; metachronous, PFS: 16 months; log rank, *p* = 0.001). None of the patient characteristics revealed an impact on OS.

#### *3.2. Differential Biomarker Expression in Colorectal Liver and Lung Metastases*

Liver and lung metastases were comparatively analyzed with a panel of metastasisrelated protein biomarkers. A differential expression pattern between liver and lung metastases was observed for the growth factor receptors IGF-1R (LuM 92.3% vs. LM 75.8%, *p* = 0.013) and EGF-R (LuM 68% vs. LM 41.5%, *p* = 0.004), showing a significantly higher fraction of positive cancer cells in the lung metastases, respectively. Similar results were obtained for the cell adhesion molecules CD44v6 (LuM 55.7% vs. LM 34.9%, *p* = 0.019) and integrin α2β1 (LuM 88.3% vs. LM 58.5%, *p* = 0.001), as well as for the check point molecule PD-L1 (LuM 6.1% vs. LM 3.3%, *p* = 0.005). In contrast, no significant difference was observed for the growth factor receptor HGF-R and the chaperon molecule Hsp90, both showing a high fraction of positive cancer cells in almost all distant metastases. Conversely, all but one metastatic lesion were found negative for the hormone receptors ERα and PR. One individual liver metastasis demonstrated 30% ERα positive cancer cells. Moreover, in colorectal liver and lung metastases, a minor fraction of the cancer cells were found positive for the cell adhesion molecule Muc1 and growth factor receptor Her2/neu. In fact, only one liver metastasis (60% Her2/neu positive cancer cells) qualified for anti-Her2/neu therapy. The number of biomarker-positive lesions and the means of biomarker expression are given in Table 3. The distribution of biomarker expression is shown for liver and lung metastases (Figure 1).

**Table 3.** Positivity and Distribution of Biomarkers in Liver and Lung Metastases.


*n*, number of patients; **n.t.**, not tested; **\***, calculation of the cut-offs is given in the Materials and Methods Section.


**Figure 1.** Biomarker Expression Pattern of Liver and Lung Metastases. horizontal bars, Means. Each dot represents a metastatic lesion; empty dots represent liver metastases (**LM**); filled dots represent lung metastases (**LuM**). **Figure 1.** Biomarker Expression Pattern of Liver and Lung Metastases. horizontal bars, Means. Each dot represents a metastatic lesion; empty dots represent liver metastases (**LM**); filled dots represent lung metastases (**LuM**).

Biomarker analysis showed most of the benign liver tissues positive for HGF-R, EGF-R, and Hsp90. IGF-1R and PD-L1 were detected in a fraction of benign liver samples (IGF-1R: 11 out of 52, 21.2%; PD-L1: 10 out of 52, 19.2%). Interestingly, benign liver tissue was negative for Muc1, CD44v6 and the integrin α2β1. In contrast, all biomarkers tested were detected on benign lung tissue, although the integrin α2β1 (10 out of 15, 66.6%) and Muc1 (8 out of 15, 53%) were observed on a reduced number of adjacent lung tissues. Data obtained in benign tissue samples are summarized in Table S1. Figure 2 demonstrates the significantly different staining patterns by each biomarker of liver and lung metastases.

#### *3.3. Prognostic Impact of Biomarker Expression in Colorectal Liver Metastases*

The prognostic impact of the biomarkers was analyzed in patients with liver metastases. CD44v6, but none of the other biomarkers tested, was identified as an indicator for early recurrence. Liver metastases with a high fraction (>30%, *n* = 22) of CD44v6+ tumor cells significantly correlated with a shorter (median 7.0 months) PFS compared to LM with a low CD44v6 expression (≤30% CD44v6+ cells, *n* = 30; median 15.5 months; log rank *p* = 0.01). Recurrent liver metastases with a high proportion of CD44v6+ cancer cells showed more frequent multi-organ metastases (6 out of 19, 31.58%), compared to liver metastases with a low proportion of CD44v6+ cancer cells (3 out of 22, 13.65%). Almost all multi-organ metastases involved liver and lung, regardless of the extent of CD44v6 expression. Cox regression analysis confirmed the independent prognostic impact of CD44v6 on PFS (Table 4). No significant correlation was found between CD44v6 expression in LM and OS.

**Table 4.** Multivariate Survival Analysis of CD44v6 Expression in Colorectal liver Metastases.


**HR**, Hazard ratio; *p*-value was calculated for progression free survival; **CI**, confidence interval; **\***, nodules within the metastatic organ.

#### *3.4. CD44v6-Related Dual Biomarker Expression in Colorectal Liver Metastases*

Co-expression analysis was performed on CD44v6 and the metastasis-related biomarkers. Univariate analysis identified three pairs of highly expressed biomarkers associated with short PFS. Patients with liver metastases with strong expression of CD44v6 and integrin α2β1 showed a shorter mean PFS (3 months) compared to the group with only high expression of CD44v6 (7 months) (Table 5, Figure 3). Multivariate Cox regression analysis identified the combination of a high CD44v6 and a high integrin α2β1 expression (HR: 4.135, 95% CI: 1.648–10.375, *p* = 0.002) and the combination of a high CD44v6 and a high PD-L1 expression (HR: 2.882, 95% CI: 1.213–6.848, *p* = 0.017), as independent prognostic factors for short progression-free survival (Table 6). High co-expression was detected in a substantial number of patients; i.e., CD44v6 high (>30% positive tumors cells) combined with integrin α2β1 high (>80% positive tumor cells) in 11 out of 52 (21.15%) patients, CD44v6 high combined with Hsp90 high (>70% positive tumor cells) in 14 out of 52 (26.92%) patients and CD44v6 high combined with PD-L1 high (>1% positive cells) in 12 out of 52 (23.1%) patients.

**Figure 2.** Immunohistochemical Staining of Different Biomarkers. Differential biomarker expression between liver (**1**) and lung (**2**) metastases demonstrated by immuno-histochemistry. (**A**), CD44v6; (**B**), α2β1; (**C**), PD-L1; (**D**), IGF-1R; (**E**), EGFR; **Tu**, tumor tissue; **BT**, Benign tissue. **Figure 2.** Immunohistochemical Staining of Different Biomarkers. Differential biomarker expression between liver (**1**) and lung (**2**) metastases demonstrated by immuno-histochemistry. (**A**), CD44v6; (**B**), α2β1; (**C**), PD-L1; (**D**), IGF-1R; (**E**), EGFR; **Tu**, tumor tissue; **BT**, Benign tissue.


**Table 5.** Univariate Survival Analysis of CD44v6-Related Dual Biomarker Expression in Colorectal Liver Metastases.

**PFS**, progression-free survival; cut-off values defining high and low for the individual biomarker are given in Table 3; **\***, calculation of the cut-offs is given in the Materials and Methods Section.

**Table 6.** Multivariate Survival Analysis of CD44v6-Related Dual Biomarker Expression in Colorectal Liver Metastases.


**HR**, Hazard ratio; **PFS**, progression free survival; **CI**, confidence interval; **\***, nodules within the metastatic organ; cut-off values defining high- and low-level expression for the individual biomarker are given in Table 3.

**Figure 3.** Kaplan–Meier Curves of CD44v6-Related Biomarker Expression in Colorectal Liver Metastases. **Blue lines**, low/low expression; **green lines**, high/high expression; **red lines**, high/low and low/high expression; log-rank *p*-values are given; cut-off values defining high- and low-level expression for the individual biomarker are given in Table 3. **Figure 3.** Kaplan–Meier Curves of CD44v6-Related Biomarker Expression in Colorectal Liver Metastases. **Blue lines**, low/low expression; **green lines**, high/high expression; **red lines**, high/low and low/high expression; log-rank *p*-values are given; cut-off values defining high- and low-level expression for the individual biomarker are given in Table 3.

#### **4. Discussion**

**Table 6.** Multivariate Survival Analysis of CD44v6-Related Dual Biomarker Expression in Colorectal Liver Metastases*.* **Variable Groups Cox Regression (PFS) HR** *p***-Value 95% CI** age (median in years) >64/≤64 1.561 0.256 0.724–3.366 number of metastases \* >1/≤1 1.398 0.358 0.684–2.855 type of metastases synchronous/metachronous 3.813 **0.008** 1.407–10.332 CD44v6/α2β1 expression high/high vs. low/low 4.135 **0.002** 1.648–10.375 high/low and low/high vs. low/low 1.784 0.145 0.819–3.886 age (median in years) >64/≤64 1.129 0.773 0.496–2.568 number of metastases >1/≤1 1.321 0.460 0.632–2.762 type of metastases synchronous/metachronous 3.345 **0.013** 1.289–8.680 CD44v6/Hsp90 expression high/high vs. low/low 2.039 0.085 0.906–4.586 high/low and low/high vs. low/low 1.412 0.443 0.585–3.404 age (median in years) >64/≤64 1.290 0.493 0.623–2.675 number of metastases >1/≤1 1.341 0.418 0.659–2.728 type of metastases synchronous/metachronous 4.154 **0.004** 1.584–10.893 CD44v6/PD-L1 expression high/high vs. low/low 2.882 **0.017** 1.213–6.848 high/low and low/high vs. low/low 0.872 0.723 0.409–1.860 It is well published that primary colorectal cancer differs in its biology, depending on sidedness [45]. This also includes treatment-relevant characteristics, such as the RAS [46,47], MSI [48] and BRAF status [49]. In the present study, biomarker heterogeneity was identified between colorectal liver and lung metastases, namely for the cell adhesion molecules α2β1, CD44v6, the growth factor receptors IGF-1R, EGF-R and the immune checkpoint biomarker PD-L1. These site-specific differences in biomarker expression might reflect the complex multifactorial interactions between disseminated cancer cells and the target organ microenvironment [50]. Cancer cells with a unique tumor biology are homing to metastatic niches with a microenvironment promoting colonization, survival, and proliferation [51,52]. Liver and lung metastases reveal biological differences; for example, in the cellular composition of the microenvironment [36,52–54], the ECM signature [52,55,56] and the secretome profile [52,57]. Quantitative differences in protein biomarker expression were found between liver and lung metastases, showing a significantly higher proportion of IGF-1R-, EGR-R-, CD44v6-, α2β1-, and PD-L1-positive cancer cells in the lung. This observation confirms published data, showing a higher frequency of genetic drivers, such as KRAS alterations and MET amplification in lung metastases [20,58]. At the same time, lung metastases exhibit an increased immunosuppressive microenvironment and prometastatic inflammation [36,59]. These findings suggest distinct colonization mechanisms, involving both specific cancer cells with a higher propensity to metastasize to the lung and a lung-specific environment that facilitates metastasis of specific cancer cells. Targeting metastasis-relevant biomarker expression will open up new therapeutic opportunities, adjusted to specific metastatic localizations. This is in deep contrast to the current guideline, which recommends the concept of treating distant metastasis with the same therapy, independent from the metastatic organ site.

The protein biomarker expression pattern in liver metastases was tested for prognostic relevance. A high (>30%) fraction of CD44v6+ liver metastatic cells was identified as an independent prognostic factor mediating short progression-free survival. This finding supports CD44v6 as a metastatic driver. Multiple underlying molecular mechanisms have been described for CD44v6-mediated progression in colorectal cancer. Examples are interactions with the extracellular matrix components osteopontin and hyaluronic acid and the binding of different cytokines, such as HGF, EGF and VEGF [31,60]. Co-expression analysis identified two new independent risk factors associated with poor prognosis of CRC patients with liver metastases. Most interesting, high dual expression of CD44v6 and integrin α2β1 represents an indicator of early recurrence, defined as tumor relapse within six months after liver resection for colorectal metastases [61,62]. Direct and extracellular matrix-mediated molecular crosstalk between CD44v6 and various integrins, including α2β1, was found to promote cancer cell proliferation and invasion, tumor angiogenesis and chemoresistance, all involved in a considerable shortening of progression-free survival compared to the single CD44v6 expression [63–65]. In addition, dual expression of CD44v6 and PD-L1, indicating the crosstalk between tumor cells and the tumor microenvironment, significantly correlated with short survival. The subset of CD44v6+ colorectal cancers simultaneously expressing PD-L1 might represent stem-like properties and contributes to immune evasion mediating poor prognosis [66,67]. Similarly, co-mutations in RAS, TP53 and SMAD4, as well as in APC and PIK3CA, resulted in a worse outcome after hepatectomy compared to single mutations [19]. Therefore, our findings support the strategy of combining prognostic protein biomarkers to render the prediction of outcome more precise [68,69]. Further, these new factors might be included in clinical risk scores, similar as reported for the KRAS status in the GAME score [70] and the KRAS/NRAS/BRAF status in the CERR score [71], which resulted in the refinement to predict recurrence after resection of CRC liver metastases. In contrast to some of the most investigated therapeutic biomarkers, namely BRAF, MSI-high, and Her2/neu, all detected in a very small patient cohort [19,20], dual expression of the druggable targets CD44v6/α2β1 and CD44v6/PD-L1 was identified in about 20% of the liver metastatic patients.

In addition, these novel findings might have an impact on the development of new therapeutic strategies for liver metastatic CRC patients. Currently, new anti-CD44v6 treatment strategies, such as half antibodies conjugated nanoparticles [72], peptides (NCT03009214) and CD44v6-specific CAR gene-engineered T cells (NCT04427449, [73]) are under investigation and might also become a treatment option for CRC patients with CD44v6-positive liver metastases. Combination of two biomarkers might help to stratify patients more precisely for targeted therapy compared to single biomarker expression. For example, Shek et al., 2021, reported that only a subgroup of PD-L1-positive mCRCs responded to checkpoint inhibitor therapy [74]. In addition, dual expression of druggable biomarkers will further promote the promising concept of multiple target inhibition, aiming to improve treatment outcome and reduce the risk of drug resistance. Recently, the combination of the BRAF inhibitor Encorafenib with the EGF-R inhibitor Cetuximab has been reported as the new standard for the treatment of metastatic BRAF-mutated colorectal cancer [75]. Currently, a number of clinical trials are ongoing in advanced colorectal cancer, simultaneously inhibiting different targets. This includes combination therapy of the EGF-R inhibitor Panitumumab with the multi-kinase inhibitor Cabozantinib [76]. Further, anti-PD-L1 checkpoint inhibitors have been combined with targeted therapies, aiming to improve the response to immunotherapy [77]. In the present study, dual expression of PD-L1 and CD44v6 was found to correlate with poor prognosis and might represent a new therapeutic option for combination therapy. The second interesting pair of therapeutic targets identified in the present study was the co-expression of CD44v6 and the integrin α2β1. Both cell adhesion molecules were found to mediate chemoresistance [65,78]. Simultaneous inhibition of both targets might result in the circumvention of chemoresistance and represent a new anti-metastatic strategy of targeted therapy. Consideration of metastasis-driving protein biomarkers that predict early recurrence after hepatectomy might play a critical role in the clinical management of patients diagnosed with liver metastases [79]. The findings in the present study need to be confirmed in a larger, prospective trial.

#### **5. Conclusions**

A differential expression pattern of the druggable protein biomarkers α2β1, CD44v6, IGF-1R, EGF-R and PD-L1 was identified between colorectal liver and lung metastases. High expression of CD44v6, CD44v6/α2β1, and CD44v6/PD-L1 correlated significantly with early recurrence after hepatic metastasectomy. Dual biomarker expression may render the prognostic prediction more precise and stratify high-risk patients for new therapeutic concepts, depending on the metastatic organ site.

**Supplementary Materials:** The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/cancers14081939/s1, Table S1: Positivity and Distribution of Biomarkers in Benign Tissue of Liver and Lung Metastases.

**Author Contributions:** Conceptualization, B.M.; methodology, K.D. and F.W.; validation, B.M. and K.D.; investigation, F.W. and B.M.; resources, B.M.; data curation, F.W. and M.P.; writing—original draft preparation, F.W. and B.M.; writing—review and editing, K.D., J.W. and M.P.; visualization, F.W.; supervision, B.M.; project administration, B.M.; funding acquisition, B.M. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by the German Federal Ministry of Education and Research, Leading Edge Cluster m4 (B.M.) under Grant FKZ 16EX1021N.

**Institutional Review Board Statement:** The study was conducted according to the guidelines of the Biobank of the Department of General, Visceral, and Transplant Surgery, Ludwig-Maximilians-University Munich, Munich, Germany. This Biobank operates under the administration of the Human Tissue and Cell Research (HTCR) Foundation. The framework of the HTCR Foundation has been approved by the ethics commission of the Faculty of Medicine at the LMU (approval number 025-12), as well as the Bavarian State Medical Association (approval number 11142) in Germany.

**Informed Consent Statement:** Informed consent from all patients was obtained by the HTCR Foundation.

**Data Availability Statement:** Data corresponding to the analyzed tissues were delivered in anonymized form by the HTCR Foundation.

**Acknowledgments:** We thank the staff members of the Biobank, especially Maresa Demmel, Nadine Gese, Ute Bossmanns and Beatrice Rauter for tissue sample/data organization, Michael Pohr for technical support and Robin v. Holzschuher for providing language help.

**Conflicts of Interest:** The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.

#### **References**


## *Article* **Discoidin Domain Receptor 1 Expression in Colon Cancer: Roles and Prognosis Impact**

**Kaouther Ben Arfi <sup>1</sup> , Christophe Schneider <sup>2</sup> , Amar Bennasroune <sup>2</sup> , Nicole Bouland 2,3, Aurore Wolak-Thierry <sup>4</sup> , Guillaume Collin <sup>5</sup> , Cuong Cao Le <sup>2</sup> , Kevin Toussaint <sup>2</sup> , Cathy Hachet <sup>2</sup> , Véronique Lehrter <sup>5</sup> , Stéphane Dedieu <sup>2</sup> , Olivier Bouché 5,6 , Hamid Morjani <sup>5</sup> , Camille Boulagnon-Rombi 1,2,3,† and Aline Appert-Collin 2,\* ,†**


**Simple Summary:** Colorectal cancer (CRC) is the third leading cause of cancer death in both sexes. Identification of the influencing factors and molecular mechanisms in CRC progression could improve patient survival. This study aimed first to characterize the expression of Discoidin Domain Receptor 1 (DDR1), a receptor tyrosine kinase for collagens in a large cohort of CRC patients, and second to establish in vitro whether DDR1 expression level is linked to CRC aggressiveness potential. Our immunohistochemical study indicated that DDR1 is highly expressed in colon cancer compared to normal colonic mucosa and its expression is associated with shorter event-free survival. In vitro, the invasive properties of several CRC cell lines seem to be correlated with the expression level of DDR1. Taken altogether, our results show that DDR1 is highly expressed in most colon adenocarcinomas and appears as an indicator of worse event free survival.

**Abstract:** Extracellular matrix components such as collagens are deposited within the tumor microenvironment at primary and metastatic sites and are recognized to be critical during tumor progression and metastasis development. This study aimed to evaluate the clinical and prognostic impact of Discoidin Domain Receptor 1 (DDR1) expression in colon cancers and its association with a particular molecular and/or morphological profile and to evaluate its potential role as a prognosis biomarker. Immunohistochemical expression of DDR1 was evaluated on 292 colonic adenocarcinomas. DDR1 was highly expressed in 240 (82.2%) adenocarcinomas. High DDR1 immunostaining score was significantly associated, on univariate analysis, with male sex, left tumor location, *BRAF* wild type status, *KRAS* mutated status, and Annexin A10 negativity. High DDR1 immunohistochemical expression was associated with shorter event free survival only. Laser capture microdissection analyses revealed that DDR1 mRNA expression was mainly attributable to adenocarcinoma compared to stromal cells. The impact of DDR1 expression on cell invasion was then evaluated by modified Boyden chamber assay using cell types with distinct mutational profiles. The invasion capacity of colon adenocarcinoma is supported by DDR1 expression. Thus, our results showed that DDR1 was highly expressed in most colon adenocarcinomas and appears as an indicator of worse event free survival.

**Citation:** Ben Arfi, K.; Schneider, C.; Bennasroune, A.; Bouland, N.; Wolak-Thierry, A.; Collin, G.; Le, C.C.; Toussaint, K.; Hachet, C.; Lehrter, V.; et al. Discoidin Domain Receptor 1 Expression in Colon Cancer: Roles and Prognosis Impact. *Cancers* **2022**, *14*, 928. https://doi.org/10.3390/ cancers14040928

Academic Editor: Isabelle Van Seuningen

Received: 29 December 2021 Accepted: 10 February 2022 Published: 13 February 2022

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

**Copyright:** © 2022 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:** colon cancer; discoidin domain receptor; prognosis; event free survival; survival

#### **1. Introduction**

Colorectal cancer (CRC) is ranked among the most common cancers in the world and is a significant public health issue in developed countries. Recent data indicated that CRC is the third most common cancer and the second leading cause of cancer death in both sexes [1]. The important mortality in CRC patients is highly correlated to its potential of metastasis reported in 50% of patients after surgery [2]. Indeed, about 39% of CRC patients are diagnosed at early stage with localized-stage disease. For these patients, the 5-year survival rate is 90%, while for the patients diagnosed with stage IV CRC, the survival declines to 12% [3].

However, at the same stage, all CRC do not have the same prognosis. Some parameters set by the tumor stage could refine the prognosis prediction and some histoprognosis factors have been identified: lymphovascular invasion, perineural invasion, tumor differentiation, or molecular profiles [2]. Treatment decisions could be influenced by these factors. In fact, many studies have been recently conducted to find new molecularly based prognostic markers, which are complementary to the data obtained by pathological diagnosis and therefore may increase the patient's survival. However, new biomarkers able to stratify the prognosis groups of patients and improve treatment strategies remain necessary. For this purpose, several studies investigate the signaling pathways that promote the metastatic process in CRC in order to identify new key players in this process that could constitute potential targets [4].

Receptor tyrosine kinases (RTKs) play an important role in several cellular processes in tumors including growth, migration, invasion, and the response to therapies [5]. For instance, the mitogen-activated protein kinase (MAPK) pathway and the phosphoinositide-3-kinase–protein kinase B/Akt (PI3K-PKB/Akt) pathway, two main intracellular pathways activated by the epidermal growth factor receptor (EGFR), were the most used therapeutic targets in metastatic colon cancer [6].

Discoidin domain receptors (DDRs) are collagen receptors with tyrosine kinase activity. The expression of the two members of this family, DDR1 and DDR2, is different: DDR1 is preferentially located in epithelial cells whereas DDR2 is expressed more importantly in connective tissues of the embryonic mesoderm [7]. Both DDR1 and DDR2 are activated by fibrillar collagens such as type I collagen [8]. Several studies have suggested a pivotal role of DDRs in tumor progression [9–11]. DDR1 expression appears to be increased in a variety of tumors and is correlated to poor prognosis [9–11]. Indeed, high level of DDR1 expression has been observed in several tumors such as prostate [12], lungs [13], breast [14], and ovary [15], suggesting a potential role of DDR1 in tumorigenesis and tumor progression. Moreover, experimental models have demonstrated that DDR1 plays an important role in cell proliferation and the metastasis process [16–19].

However, its role appeared to be tumor dependent. DDR1 overexpression was associated with advanced tumor stages in esophageal cancer [20], brain tumors [21] and with poor survival, in lung adenocarcinoma [22] and serous ovarian cancer [15].

In colon carcinoma, the role of DDR1 remains incompletely elucidated. The prognosis impact of DDR1 in CRC had not been studied much until now. High DDR1 expression seemed to be associated with poor overall survival [23–25]. Moreover, Sirvent and coworkers have shown that DDR1 plays a key role in the invasion potential of CRC [26]. The pharmacological inhibition of DDR1-BCR signaling axis using nilotinib has indeed been reported to decrease invasion and metastatic processes in CRC. These results suggest that DDR1 could represent a potential target in CRC treatment [26].

In the present study, we evaluated the expression of DDR1 in a cohort of CRC that, to our knowledge, is the largest set of CRC specimens studied for this receptor up to date. Specifically, we assessed the association between DDR1 expression and associated clinicopathological and molecular characteristics and its potential value as a prognosis marker. Finally, we examined in vitro the role of DDR1 in cell invasion in several CRC cell lines to establish whether DDR1 expression level is linked to CRC aggressiveness potential.

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

#### *2.1. Culture Cells*

HCT116, SW480, SW620 and HT-29 colorectal carcinoma cell lines were purchased from American Type Culture Collection (ATCC, Rockville, MD, USA). HT-29DDR1-GFP and HT-29GFP were obtained as previously described [27]. All cell lines were grown in Dulbecco's Modified Eagle Medium (DMEM) with high glucose (4.5 g/L) (Thermo Fisher Scientific, Villebon sur Yvette, France) containing 10% (*v*/*v*) fetal bovine serum (FBS) (Dutscher, Bernolsheim, France) and 1% penicillin-streptomycin (*v*/*v*, Invitrogen). Cells were regularly controlled for the absence of mycoplasma by PCR methods.

#### *2.2. RNA Isolation from Cell Culture*

Total RNA from cells was extracted as described previously [28] and single-stranded cDNA was synthesized from 250 ng total mRNAs using VERSO cDNA kit (Thermo Fisher Scientific, Waltham, MA, USA) according to the manufacturer's instructions. Determination of the mRNA of DDR1 was carried out by real-time PCR as described [27].

#### *2.3. Total Protein Extraction and Immunoblotting*

Seventy-two hours after seeding, cells were washed with ice-cold phosphate-buffered saline (PBS) and harvested in lysis buffer (10 mM Tris, 150 mM NaCl, 5 mM EDTA, 1% triton, protease inhibitors (Roche Diagnostics, Indianapolis, IN), and 5 mM Na orthovanadate). Cell lysates were then centrifugated at 14,000 *g* for 10 min at 4 ◦C. Protein concentration was quantified by BCA Protein Assay Kit (Thermo Fisher Scientific, Waltham, MA, USA). Proteins were separated on acrylamide gels and electroblotted onto nitrocellulose membranes (Amersham Biosciences, Little Chalfont, UK). The blots were incubated with primary antibodies (DDR1 (D1G6), GFP (D5.1) and GAPDH (14C10)) and corresponding peroxidase conjugated secondary antibody as previously indicated [27].

#### *2.4. Invasion Assay*

Cell invasion was evaluated using type-I collagen-coated 24-well cell culture inserts with an 8 µm pore size (Dustscher, Bernolsheim, France). The Boyden chambers were coated with 25 µg cm−<sup>2</sup> type-I collagen and then washed twice with PBS. A total of 5 × 104 cells were seeded into the upper chambers in a 200 µL DMEM culture medium, supplemented with 2% FBS, 1% penicillin–streptomycin. DMEM culture medium with 10% FBS and 1% penicillin–streptomycin was added in the lower chamber. After 24 h, the chambers were washed with PBS, fixed with methanol and stained with Di Aminido Phenyl lndol (DAPI, Santa Cruz Biotechnology). Cells remaining on the upper face of the membranes were suppressed by scraping, and those on the lower side were counted after being imaged on the EVOS® FL Auto Imaging System using a 40<sup>×</sup> objective (Thermofisher scientific, Waltham, MA, USA). Experiments were reproduced three times in triplicates.

Concerning experiments using Nilotinib and DDR1-IN-1 inhibitors, 7.5 <sup>×</sup> <sup>10</sup><sup>4</sup> cells were seeded into the upper chambers in a 200 µL DMEM culture medium, supplemented with 2% FBS, 1% penicillin–streptomycin in the presence or not of Nilotinib (100 nM, No.S1033) or DDR1-IN-1 (10 µM, No.S7498, Selleckchem). DMEM culture medium with 10% FBS and 1% penicillin–streptomycin was added in the lower chamber. After 24 h, the chambers were washed with PBS, fixed with methanol, and stained with crystal violet. Cells remaining on the upper face of the membranes were suppressed by scraping. Upon

solubilization in acetic acid (10%), the amount of dye on the filter was quantified by spectrophotometry at 560 nm.

#### *2.5. Patients*

Patients and selection were clarified in paper from Boulagnon-Rombi et al. [29].

The study was conducted on adult patients who underwent surgery for sporadic colon cancer in the Digestive Surgery Department of the University Hospital of Reims between September 2006 and December 2012. Patients with rectal cancer were excluded.

Clinical data including age at the time of surgery, sex, performance status, surgical circumstances (tumor perforation, occlusion), tumor location, synchronous or metachronous metastases, tumor recurrence, treatment, death and pathological and molecular data including adenocarcinoma type, grade, and pTNM stage were collected. Patients were classified as having a right colonic cancer if the primary tumor was located in the caecum, ascending colon, hepatic flexure or transverse colon, and left colonic cancer if the tumor site was within the splenic flexure, descending colon, sigmoid colon, or rectosigmoid junction.

#### *2.6. Pathology*

All colon adenocarcinomas were classified and subtyped according to The World Health Organization criteria [30] and staged according to the International Union Against Cancer 2009 guidelines [31]. Tumor budding was assessed on Hematoxylin- Eosin-Saffron slides and classified as low budding rate if less than 5 buds were present in the 0.785 mm<sup>2</sup> hot spot [32].

#### *2.7. Immunohistochemistry*

Tissue samples were analyzed via tissue microarrays (TMA). For each tumor, 3 cores were punched in the central part and 3 cores at the invasive front of the tumor from the same original formalin-fixed paraffin-embedded tumor block. The cores were 2 mm in diameter and were precisely arrayed into a recipient paraffin block using the MiniCore Tissue Arrayer (Excilone, Elancourt, France). Sections of 4-µm thickness were cut and mounted on SuperFrost Plus Gold adhesive slides (Thermofisher Scientific, Waltham, MA, USA). Immunohistochemistry (IHC) was performed using DDR1 (D1G6) XP® Rabbit mAb, rabbit Monoclonal antibody (1/100, Cell Signaling ref: #5583) after heat-induced epitope retrieval in citrate pH 6 buffer (95 ◦C, 40 min) and overnight antibody incubation at 4 ◦C and then visualized using 3-Amino-9-Ethylcarbazole (AEC).

#### *2.8. Scoring*

Immunostaining intensity (SI) was graded independently by two pathologists (CBR, KBBA).

Immunopositivity was defined as a brown cytoplasmic color in the tumor cells. Staining intensity was scored as follows: 0, negative staining signal in >50% of tumor cells; 1+, weak staining signal detected in >50% of tumor cells; 2+, moderate staining signal in >50% of tumor cells; 3+, strong staining signal in >50% of tumor cells (Figure 1). The staining intensity was then divided into score 0/1+ for low DDR1 expression or score 2+/3+ for high DDR1 expression as previously described [23].

#### *2.9. Molecular Analyses*

Tumor DNA was extracted and the mutation profile (*BRAF, KRAS,* and MSI status) of the samples was determined as described earlier [33].

#### *2.10. Laser Capture Microdissection*

Laser capture microdissection was performed on fresh frozen colon cancer specimens cut into 12 µm serial sections and mounted on PALM membrane slides (Zeiss, Oberkochen, Germany) as previously noticed [29].

DDR1 expression as previously described [23].

at 4 °C and then visualized using 3-Amino-9-Ethylcarbazole (AEC).

**Figure 1.** Representative images of DDR1 immunolabeling in colon adenocarcinoma. A. Strong and diffuse staining (red) in adenocarcinoma cells (arrow), (magnification ×10), scored 3+/high; B. Moderate and diffuse staining (red) in adenocarcinoma cells (arrow), (magnification ×20), scored 2+/high; C. Faint staining (red) in adenocarcinoma cells (arrow), (magnification ×10), scored 1+/low. Stromal cell highlighted by an asterisk (\*) was weak (**A**) or negatively (**B**,**C**) stained in all cases.

Patients and selection were clarified in paper from Boulagnon-Rombi et al. [29]. The study was conducted on adult patients who underwent surgery for sporadic colon cancer in the Digestive Surgery Department of the University Hospital of Reims between September 2006 and December 2012. Patients with rectal cancer were excluded.

Clinical data including age at the time of surgery, sex, performance status, surgical circumstances (tumor perforation, occlusion), tumor location, synchronous or metachronous metastases, tumor recurrence, treatment, death and pathological and molecular data including adenocarcinoma type, grade, and pTNM stage were collected. Patients were classified as having a right colonic cancer if the primary tumor was located in the caecum, ascending colon, hepatic flexure or transverse colon, and left colonic cancer if the tumor site was within the splenic flexure, descending colon, sigmoid colon, or rectosigmoid junc-

All colon adenocarcinomas were classified and subtyped according to The World Health Organization criteria [30] and staged according to the International Union Against Cancer 2009 guidelines [31]. Tumor budding was assessed on Hematoxylin- Eosin-Saffron slides and classified as low budding rate if less than 5 buds were present in the 0.785 mm2

Tissue samples were analyzed via tissue microarrays (TMA). For each tumor, 3 cores were punched in the central part and 3 cores at the invasive front of the tumor from the same original formalin-fixed paraffin-embedded tumor block. The cores were 2 mm in diameter and were precisely arrayed into a recipient paraffin block using the MiniCore Tissue Arrayer (Excilone, Elancourt, France). Sections of 4-μm thickness were cut and mounted on SuperFrost Plus Gold adhesive slides (Thermofisher Scientific, Waltham, MA, USA). Immunohistochemistry (IHC) was performed using DDR1 (D1G6) XP® Rabbit mAb, rabbit Monoclonal antibody (1/100, Cell Signaling ref: #5583) after heat-induced epitope retrieval in citrate pH 6 buffer (95 °C, 40 min) and overnight antibody incubation

Immunostaining intensity (SI) was graded independently by two pathologists (CBR,

Immunopositivity was defined as a brown cytoplasmic color in the tumor cells. Staining intensity was scored as follows: 0, negative staining signal in >50% of tumor cells; 1+, weak staining signal detected in >50% of tumor cells; 2+, moderate staining signal in >50% of tumor cells; 3+, strong staining signal in >50% of tumor cells (Figure 1). The staining intensity was then divided into score 0/1+ for low DDR1 expression or score 2+/3+ for high

RNA from tumor and stromal microdissected tissues were isolated and purified as indicated [29].

#### *2.11. DDR1 mRNA Expression*

*2.5. Patients* 

tion.

*2.6. Pathology* 

hot spot [32].

*2.8. Scoring* 

KBBA).

*2.7. Immunohistochemistry* 

Analysis of mRNA expression was performed as previously described [29]. Only RNAs with RQI values ≥5 were used for further analyses. Determination of the mRNA of DDR1 was carried out by real-time PCR as described [27].

#### *2.12. Data Mining and Bioinformatic Analyses*

Survival analyses were performed using publicly available data from TCGA, Martineau and SieberSmith gene expression dataset in the R2 microarray analysis and visualization platform (http://r2.amc.nl; last access date: 5 November 2021). The scan online algorithm was used to determine the cut-off values for separating high and low DDR1 expression groups.

#### *2.13. Statistical and Survival Analyses*

Statistical analyses and factors associated with immunohistochemical expression of DDR1 were clarified in paper from Boulagnon-Rombi et al. [29].

#### **3. Results**

#### *3.1. Association of DDR1 Immunohistochemical Expression with Clinico-Pathological Features*

The relationship between DDR1 expression and disease aggressiveness was investigated in a cohort of 292 colon cancer patients. The clinicopathological features are summarized in Table 1. The population consisted of 166 (57%) men and 126 (43%) women, whose mean age was 70.8 ± 10.8 years. Tumors were right-sided in 123 cases (42%), left-sided in 164 cases (56%), and multifocal in 5 cases (2%). The mean follow-up time was 43 months (±32 months).

Figure 1 illustrates representative IHC patterns of DDR1 expression. The immunostaining showed the localization of DDR1 mostly in the cytoplasm. The immunostaining intensity was strong in 144 (49.3%) samples, moderate in 96 (33%), and weak in 52 (17.8%), and no samples were found negative for DDR1 staining (score 0). DDR1 immunostaining was diffuse (>50% of positive tumor cells) in all cases. DDR1 immunolabeling in tumor stroma was weak or negative in all cases. For the statistical analysis, patients were divided into two groups: low expression of DDR1 for patients with immunostaining intensity scored 1 and high DDR1 expression for patients with immunostaining intensity scored 2 or 3. Thus, DDR1 expression by IHC was rated high in 240 (82.2%) cases. In case of samples presenting heterogeneity in immunostaining, the highest intensity was considered for scoring.


**Table 1.** Clinicopathological features of the cohort.

The relationship between DDR1 immunohistochemical expression and different clinicopathological and molecular characteristics was analyzed. Data are detailed in Table 2. In univariate analysis, a high DDR1 immunostaining score was significantly associated with

male sex (*p* = 0.0195), left tumor location (*p* = 0.0114), BRAF wild-type status (*p* < 0.0001), KRAS mutated status (*p* = 0.0041), and absence of expression of the serrated markers Annexin A10 (*p* = 0.0097). In multivariate analysis, high DDR1 immunostaining score was independently associated with BRAF wild-type status only (*p* < 0.0001).

**Table 2.** Relationship between DDR1 expression and clinical and molecular characteristics.



*Cancers* **2022**, *14*, x FOR PEER REVIEW 8 of 16

**Table 2.** *Cont*.

n.s: not significant; ‡ : khi-2; † : Fisher test; \*: Satterthwaite. Microsatellite status (MSS vs. MSI) 72 0.4009 0.2294

#### *3.2. Survival Analysis* DDR1 IHC tumor score

We next investigated the relation between DDR1 expression and prognosis. Univariate analysis demonstrated that age, tumor stage, vascular invasion, and metastasis were predictors of overall survival (OS) in our cohort (Table 3). (low vs. High) 281 0.5832 **0.0391** High DDR1 immunostaining was not correlated with overall survival in all stages (*p*

High DDR1 immunostaining was not correlated with overall survival in all stages (*p* = 0.5832, Figure 2A) nor in metastatic (stage IV) patients (*p* = 0.8376, data not shown). Regarding event-free survival (EFS), univariate analysis revealed that occlusion, stage, vascular invasion, lymphatic invasion, differentiation grade, RAS status, CIMP status, and the level of DDR1 immunostaining scores were associated with shorter EFS (Table 3). High DDR1 expression was a predictor of shorter EFS in the entire cohort (*p* = 0.0391, Figure 2B). Stage specific analyses showed that DDR1 was not a predictor of EFS in stage II (*p* = 0.1181, Figure 3A), stage III (*p* = 0.3389, Figure 3B) and in metastatic patients (*p* = 0.9102, Figure 3C). = 0.5832, Figure 2A) nor in metastatic (stage IV) patients (*p* = 0.8376, data not shown). Regarding event-free survival (EFS), univariate analysis revealed that occlusion, stage, vascular invasion, lymphatic invasion, differentiation grade, RAS status, CIMP status, and the level of DDR1 immunostaining scores were associated with shorter EFS (Table 3). High DDR1 expression was a predictor of shorter EFS in the entire cohort (*p* = 0.0391, Figure 2B). Stage specific analyses showed that DDR1 was not a predictor of EFS in stage II (*p* = 0.1181, Figure 3A), stage III (*p* = 0.3389, Figure 3B) and in metastatic patients (*p* = 0.9102, Figure 3C).

**Figure 2.** DDR1 value as a prognosis indicator in colon cancer patients from our cohort. Kaplan-Meier curves of overall survival (**A**) and event free-survival (**B**) probability for low (blue line) and high (red line) DDR1 immunohistochemical expression in adenocarcinoma cells from all tumor stages. **Figure 2.** DDR1 value as a prognosis indicator in colon cancer patients from our cohort. Kaplan-Meier curves of overall survival (**A**) and event free-survival (**B**) probability for low (blue line) and high (red line) DDR1 immunohistochemical expression in adenocarcinoma cells from all tumor stages.


**Table 3.** Analysis of factors associated with overall and event-free survival.

In our cohort DDR1 mRNA expression levels successfully evaluated in 66 patients were not correlated with OS (*p* = 0.86) nor EFS (*p* = 0.46), whatever the CCR stage (data not shown).

To corroborate our previous results, we next performed survival analyses in Sieber-Smith (*n* = 286), Martineau (*n* = 124) [34] and TCGA cohorts (*n* = 174) obtained from R2 database [35,36]. In these cohorts, DDR1 mRNA expression was not correlated with overall nor relapse free survival (Figure 4).

#### *3.3. DDR1 Is More Expressed in Tumor Cells Compared with Stromal Cells*

DDR1 mRNA expression has been determined by RT-qPCR on 65 colonic adenocarcinoma samples and 78 colonic mucosa samples. Surprisingly, data showed a significant decrease in DDR1 expression within tumor samples when compared with normal colon samples (Figure 5A). Due to the difference observed in DDR1 expression between stromal and malignant cells when evaluated by IHC analysis, we used Laser Capture Microdissection (LCM) to thereafter quantify DDR1 mRNA expression in tumoral and stromal areas of each sample as previously described [29]. LCM was performed on 25 colon adenocarcinoma samples and RT-qPCR revealed that DDR1 mRNA expression was higher in the tumoral area than in the stroma (Figure 5B).

**Figure 3.** Stage specific event free survival analysis in colon cancer patients from our cohort according to DDR1 immunohistochemical expression. Kaplan-Meier curves of event free-survival probability for low (blue line) and high (red line) DDR1 immunohistochemical expression in cells in stage II (**A**), stage III (**B**) and stage IV patients (**C**). **Figure 3.** Stage specific event free survival analysis in colon cancer patients from our cohort according to DDR1 immunohistochemical expression. Kaplan-Meier curves of event free-survival probability for low (blue line) and high (red line) DDR1 immunohistochemical expression in cells in stage II (**A**), stage III (**B**) and stage IV patients (**C**).

#### In our cohort DDR1 mRNA expression levels successfully evaluated in 66 patients were *3.4. DDR1 Mediates the Invasion of CRC Cells*

not correlated with OS (*p* = 0.86) nor EFS (*p* = 0.46), whatever the CCR stage (data not shown). To corroborate our previous results, we next performed survival analyses in Sieber-Smith (*n* = 286), Martineau (*n* = 124) [34] and TCGA cohorts (*n* = 174) obtained from R2 database [35,36]. In these cohorts, DDR1 mRNA expression was not correlated with overall nor relapse free survival (Figure 4). We then investigated the possible role of DDR1 in CRC aggressiveness in vitro. We used HCT116, HT-29, SW480, and SW620 cell lines, which express different levels of DDR1 expression, and analyzed their ability to invade type I collagen as one of the main extracellular matrix components. These cell lines harbor different *KRAS*/*BRAF* statuses and their main characteristics are summarized in Supplementary Figure S2. The level of DDR1 expression was analyzed by both RT-qPCR and immunoblotting (Figure 6A,B, uncropped western blot images in Supplementary Figure S1). Data showed that the expression of DDR1 at the mRNA and protein levels was higher in HCT116 cells than in the other cell lines. In order to investigate deeply the impact of DDR1 on invasive properties of CRC cells, we used HT-29 cells expressing DDR1 at a basal level (HT-29GFP) and overexpressing the receptor (HT-29DDR1-GFP). As shown in Supplementary Figure S2, HT-29DDR1-GFP expressed a high level of DDR1 when compared to wild-type HT-29 or HT-29GFP cells. The invasion potential of CRC cell lines was evaluated based on modified Boyden chamber assay using type I collagen coating. Data showed that HCT116 cells exhibited a higher invasion rate than SW480 and SW620 cells. When DDR1 was overexpressed in HT29 cells (HT-29DDR1-GFP), the invasion rate was significantly increased compared to the control (HT-29GFP) (Figure 6C). Interestingly, the invasion rate positively correlated with DDR1 expression level. To confirm the role of DDR1 in the invasion process of colorectal cells, nilotinib (100 nM) and DDR1-IN-1 (10 µM) have been used to inhibit specifically DDR1. As shown in Figure 6D, significant inhibition of cell invasiveness was observed when the cells were treated with nilotinib or DDR1-IN-1 compared with the control ones. Overall, these data suggest that DDR1 is involved in CRC invasion phenotype and could be associated in this way with the worse event free survival.

**Figure 4.** Survival analysis according to DDR1 mRNA expression profile in independent colorectal cancers patients' cohorts. Kaplan-Meier curves of overall survival (**A**,**B**) and relapse or progression free-survival (**C**,**D**) probability for low (red line) and high (blue line) DDR1 mRNA expression in all stages colorectal cancers patients and in stage IV (metastatic) patients (**C**,**D**). Survival analysis and Kaplan Meyier curves of the TCGA, Martineau and SieberSmith gene expression dataset were obtained from R2 platform (http://r2.amc.nl; last access date: 5 November 2021). All *p*-values were calculated using R2 online tools. **Figure 4.** Survival analysis according to DDR1 mRNA expression profile in independent colorectal cancers patients' cohorts. Kaplan-Meier curves of overall survival (**A**,**B**) and relapse or progression free-survival (**C**,**D**) probability for low (red line) and high (blue line) DDR1 mRNA expression in all stages colorectal cancers patients and in stage IV (metastatic) patients (**C**,**D**). Survival analysis and Kaplan Meyier curves of the TCGA, Martineau and SieberSmith gene expression dataset were obtained from R2 platform (http://r2.amc.nl; last access date: 5 November 2021). All *p*-values were calculated using R2 online tools. *Cancers* **2022**, *14*, x FOR PEER REVIEW 11 of 16 eas of each sample as previously described [29]. LCM was performed on 25 colon adenocarcinoma samples and RT-qPCR revealed that DDR1 mRNA expression was higher in

*3.3. DDR1 Is More Expressed in Tumor Cells Compared with Stromal Cells* 

the tumoral area than in the stroma (Figure 5B).

**Figure 5.** Comparison of DDR1 mRNA expression between tumor cells, normal colon and stromal cells. (**A**) Real-time PCR analysis of the DDR1 mRNA expression performed in colon adenocarcinoma and in normal colon mucosa fresh frozen samples. Values are represented as dCt normalized with RPL32. (**B**) Real-time PCR analysis of the DDR1 mRNA expression performed in adenocarcinoma cells and in stromal cells after laser capture microdissection. Values are represented as dCt normalized with RPL32. \* *p* < 0.05, \*\*\*\* *p* < 0.0001, Mann Whitney test. **Figure 5.** Comparison of DDR1 mRNA expression between tumor cells, normal colon and stromal cells. (**A**) Real-time PCR analysis of the DDR1 mRNA expression performed in colon adenocarcinoma and in normal colon mucosa fresh frozen samples. Values are represented as dCt normalized with RPL32. (**B**) Real-time PCR analysis of the DDR1 mRNA expression performed in adenocarcinoma cells and in stromal cells after laser capture microdissection. Values are represented as dCt normalized with RPL32. \* *p* < 0.05, \*\*\*\* *p* < 0.0001, Mann Whitney test.

We then investigated the possible role of DDR1 in CRC aggressiveness in vitro. We used HCT116, HT-29, SW480, and SW620 cell lines, which express different levels of

and their main characteristics are summarized in Supplementary Figure S2. The level of DDR1 expression was analyzed by both RT-qPCR and immunoblotting (Figure 6A,B, uncropped western blot images in Supplementary Figure S1). Data showed that the expression of DDR1 at the mRNA and protein levels was higher in HCT116 cells than in the other cell lines. In order to investigate deeply the impact of DDR1 on invasive properties of CRC cells, we used HT-29 cells expressing DDR1 at a basal level (HT-29GFP) and overexpressing the receptor (HT-29DDR1-GFP). As shown in Supplementary Figure S2, HT-29DDR1-GFP expressed a high level of DDR1 when compared to wild-type HT-29 or HT-29GFP cells. The invasion potential of CRC cell lines was evaluated based on modified Boyden chamber assay using type I collagen coating. Data showed that HCT116 cells exhibited a higher invasion rate than SW480 and SW620 cells. When DDR1 was overexpressed in HT29 cells (HT-29DDR1-GFP), the invasion rate was significantly increased compared to the control (HT-29GFP) (Figure 6C). Interestingly, the invasion rate positively correlated with DDR1 expression level. To confirm the role of DDR1 in the invasion process of colorectal cells, nilotinib (100 nM) and DDR1- IN-1 (10 μM) have been used to inhibit specifically DDR1. As shown in Figure 6D, significant inhibition of cell invasiveness was observed when the cells were treated with nilotinib or DDR1-IN-1 compared with the control ones. Overall, these data suggest that DDR1 is involved in CRC invasion phenotype and could be associated in this way with the worse

event free survival.

*3.4. DDR1 Mediates the Invasion of CRC Cells* 

**Figure 6.** Human CRC cell invasion is modulated by DDR1 expression. (**A**) The relative mRNA expression of DDR1 was assessed using RT-qPCR. Values in HCT-116, HT-29, SW480, and SW620 were normalized with both RPL32 and RS18 mRNA expression. (**B**) The expression of DDR1 and GAPDH was assessed by western blotting using anti-DDR1 and anti-GAPDH antibodies in HCT-116, HT-29, SW480, and SW620 cells. Quantitative analysis of DDR1 protein was obtained by densitometry: the amount of DDR1 was normalized to GAPDH expression level (bottom panel). (**C**) HCT-116, HT-29, SW480, and SW620 were seeded into the collagen type I coated chambers for 24 h. Cells were then fixed with methanol and stained with DAPI. Results are expressed as mean ± SD of three independent experiments. Statistical significance was analyzed by a one-way ANOVA test using Dunnett's multiple comparisons. \* *p* = 0.05, \*\* *p* = 0.01, \*\*\* *p* = 0.001 as compared to HCT-116 cells or HT-29 cells. Correlation between DDR1 expression and cell invasion (right panel). (**D**) HT-29, HT-29DDR1-GFP, and HT-29GFP cells were seeded into the collagen type I coated chambers for 24 h in absence or presence of nilotinib (100 nM) or DDR1-IN-1 (10 μM). Cells were then fixed with methanol and stained with crystal violet. Results are expressed as mean ± SD of three independent experiments. Statistical significance was analyzed by a one-way ANOVA test using Dunnett's multiple comparisons. \*\* *p* = 0.01, \*\*\* *p* = 0.001 as compared to HT-29, HT-29DDR1-GFP, or HT-29GFP cells. **Figure 6.** Human CRC cell invasion is modulated by DDR1 expression. (**A**) The relative mRNA expression of DDR1 was assessed using RT-qPCR. Values in HCT-116, HT-29, SW480, and SW620 were normalized with both RPL32 and RS18 mRNA expression. (**B**) The expression of DDR1 and GAPDH was assessed by western blotting using anti-DDR1 and anti-GAPDH antibodies in HCT-116, HT-29, SW480, and SW620 cells. Quantitative analysis of DDR1 protein was obtained by densitometry: the amount of DDR1 was normalized to GAPDH expression level (bottom panel). (**C**) HCT-116, HT-29, SW480, and SW620 were seeded into the collagen type I coated chambers for 24 h. Cells were then fixed with methanol and stained with DAPI. Results are expressed as mean ± SD of three independent experiments. Statistical significance was analyzed by a one-way ANOVA test using Dunnett's multiple comparisons. \* *p* = 0.05, \*\* *p* = 0.01, \*\*\* *p* = 0.001 as compared to HCT-116 cells or HT-29 cells. Correlation between DDR1 expression and cell invasion (right panel). (**D**) HT-29, HT-29DDR1-GFP, and HT-29GFP cells were seeded into the collagen type I coated chambers for 24 h in absence or presence of nilotinib (100 nM) or DDR1-IN-1 (10 µM). Cells were then fixed with methanol and stained with crystal violet. Results are expressed as mean ± SD of three independent experiments. Statistical significance was analyzed by a one-way ANOVA test using Dunnett's multiple comparisons. \*\* *p* = 0.01, \*\*\* *p* = 0.001 as compared to HT-29, HT-29DDR1-GFP, or HT-29GFP cells.

#### **4. Discussion**

Many cancers are characterized by dysregulated expression of one or more RTKs. Such alteration has functional consequences at the cellular level which directly impact tumor progression, especially cell invasion and metastasis. DDRs play a key role in tumor progression, in part by regulating the reciprocal interplay between cancer cells and stromal collagens [37]. One of their major roles in the literature is their involvement in tumor invasion and metastasis [38].

In this study, we investigated the expression of DDR1 using immunohistochemistry in colon adenocarcinoma and studied the link between DDR1 expression with clinicopathologic and molecular parameters, including overall and event-free survival. Because DDR1 seems to play a role in CRC cell invasion and metastasis [5], we also investigated the impact of DDR1 on invasion properties of CRC cell lines in vitro using type I collagen as a main extracellular matrix component.

In this work, we showed that DDR1 expression was higher in adenocarcinoma cells than in normal colonic epithelium. DDR1 was highly expressed in a large majority (82.2%) of colon cancers. These results corroborate previous data showing a high DDR1 overexpression in 94% of colon cancer samples [23] and in tumor tissues from patients with primary CRC and hepatic CRC metastasis [24].

Our results demonstrated in univariate analysis that the clinico-pathological and molecular characteristics associated with DDR1 expression in colon adenocarcinoma were: male sex, left colon tumor localization, *BRAF* wild-type status, and absence of the expression of the serrated marker Annexin A10.

To our knowledge, no study had investigated these clinico-pathological and molecular characteristics in association with DDR1 expression in colon adenocarcinoma, especially the potential association with the serrated pathway highlighted by its markers Annexin A10. The molecular profile associated with DDR1 high expression could be integrated into the CMS4 molecular subtype of colorectal cancer. These tumors are characterized by strong stromal infiltration and show clear upregulation of genes playing a role in epithelial mesenchymal transition and associated to transforming growth factor β (TGF β) signaling pathway, angiogenesis, matrix remodeling pathways, and the complementmediated inflammation. These CMS4 tumors presented worse overall survival and relapsefree survival [39]. Indeed, DDR1 mRNA expression was not associated with any CMS subtype [25]. Our bioinformatic analyses revealed that high DDR1 mRNA expression was independently associated with worse OS and PFS in stage IV patients. Moreover, any significant association between DDR1 mRNA expression and OS or EFS has been found in our cohort of patients undergoing surgery for colonic adenocarcinoma. However, divergent results showed that DDR1 high mRNA expression was associated with worse OS whatever the tumor stage [24].

The major limitations of our study were its retrospective and single-center design and that few patients had DDR1 mRNA expression data. However, our results were validated with bioinformatic analyses in three other patients' cohorts. In our patients' cohort, DDR1 immunohistochemical expression was only associated with worse EFS whatever the stage. DDR1 high protein expression was not associated with OS or with stage specific EFS. In a previous immunohistochemical study, high DDR1 immunoreactivity score was correlated with a shorter overall survival in a cohort of 100 patients with colorectal cancer [23]. In this study, EFS was not evaluated and stage specific survival analyses were not performed.

The molecular mechanisms underlying the roles of the DDRs in various steps of colon carcinoma progression are largely undefined. To fill this gap, we investigated the potential role of DDR1 in tumor cell invasion by using several colorectal cancer cell lines that differentially express DDR1. In addition, HT-29 cells overexpressing DDR1 were established and led to enhanced cell invasiveness. The data showed that the tumor cell invasion capacity is closely correlated to DDR1 expression. Moreover, specific pharmacological inhibition of DDR1 with nilotinib and DDR-IN-1 significantly reduced HT-29 cell invasion. These results ascertained previous reports indicating DDR1 pro-invasive role in

several tumor cell lines and DDR1 metastatic function in many cancers [12,17,19,40], and demonstrate the importance of DDR1 in invasive tumors. For instance, DDR1 expression is increased by the microRNA MiR-199a-5p and promotes invasion in CRC by activating epithelial-mesenchymal transition [41]. In human A375 melanoma, HT29 colon carcinoma and SK-HEP hepatoma cells, chemical inhibition or silencing of DDR1 reduces cell adhesion to collagen I and MMP-dependent invasion [42]. Recently, Romayor and coworkers have demonstrated that DDR1 expressed by tumor cells promotes hepatic cell ability to alter the ECM structure by regulating collagen and MMPs expression, thus suggesting an impact of DDR1 in the desmoplastic response of hepatic tumor microenvironment during CRC tumorigenesis [24].

In addition, it has been recently demonstrated that DDR1 can have a great impact on the invasion function of metastatic colon carcinoma [26]. Indeed, invasion and metastatic processes were decreased by DDR1-BCR signaling axis inhibition in vivo in colon carcinoma suggesting that DDR1 could be an effective therapeutic target in this cancer. The authors concluded that the inhibition of DDR1 kinase activity with nilotinib may be a therapeutic benefit in patients with advanced CRC [26].

In other cancers, DDR1 expression could also have a prognostic implication. Indeed, high expression of DDR1 has also been identified in 52.2% of hepatocellular carcinoma samples [43], 61.0% of non-small cell lung cancer [13], and 69% of serous ovarian cancer tissues [15]. Moreover, high DDR1 expression was more frequently expressed in invasive carcinoma than in bronchioloalveolar carcinoma in lung cancers and was associated with shorter overall survival in non-small cell lung carcinomas [22]. On the contrary, DDR1 was not associated with survival in prostate cancer [12] and low DDR1 expression was associated with triple negative subtype of breast cancer and with shorter survival in this cancer type [44].

Thus, the overexpression of DDR1 in these malignant diseases, particularly in colorectal cancer, supports the hypothesis that DDR1 upregulation is widespread in cancer and can play an important role in tumorigenesis and/or tumor invasion and metastasis.

#### **5. Conclusions**

In summary, DDR1 is highly expressed in colon cancer compared to normal colonic mucosa. This overexpression of DDR1 is found in a large majority of colon cancers, suggesting a role of DDR1 in colorectal carcinogenesis. Although DDR1 was associated with shorter EFS, its role as a prognosis marker remains uncertain. However, frequent high expression of DDR1 in colon cancer could be further explored as a potential therapeutic target in this indication.

**Supplementary Materials:** The following are available online at https://www.mdpi.com/article/10 .3390/cancers14040928/s1, Figure S1: (A) Summary table of the CRC cell lines characteristics. (B) DDR1 expression in HT-29, HT-29DDR1-GFP and HT-29GFP cell lines; Figure S2: Original, uncropped western blot scans.

**Author Contributions:** Conception or design of the work: A.A.-C., C.B.-R., H.M. Data collection: K.B.A., C.B.-R., N.B., A.A.-C., C.C.L., K.T., G.C., V.L., C.H. Data analysis and interpretation: K.B.A., C.B.-R., N.B., A.A.-C., C.C.L., C.S., A.W.-T. Drafting the article: K.B.A., C.B.-R., A.A.-C. Critical revision of the article: S.D., O.B., A.B., H.M. All authors have read and agreed to the published version of the manuscript.

**Funding:** This study was funded by grants from: 1. Ligue Nationale Contre le Cancer Conférence de Coordination Interrégionale du Grand Est, Appel d'Offre 2014 et 2015. Grant Numbers 61/2014/KM/ ML/Recherche and 2015/KM/NM/PFRS/232 (SD). 2. Centre Hospitalier Universitaire de Reims, Appel d'Offre Local 2014 et 2015. Grant Numbers AU14-03 and AOL 2015-11 (CBR). 3. Structure Fédérative de Recherche Cap Santé (SFR Cap Santé) (A.A.-C., C.B.-R., H.M.). All these study sponsors have no roles in the study design, in the collection, analysis, and interpretation of data.

**Institutional Review Board Statement:** The study was performed in accordance with the ethical standards laid down in the Declaration of Helsinki. Written patients' consent for biospecimen use was obtained in all cases. Approval for the study was previously obtained from the local Institutional Review Board and the Tissue Bank Management Board [29]. The ethical committee of the Plateforme des Centre de Ressources Biologiques de Champagne Ardennes, protocol code: AC-2019-3408.

**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.

**Acknowledgments:** We thank Frederic Saltel (INSERM U1053, Bordeaux) for the gift of the lentivirus encoding DDR1-GFP. We also thank Damien Rioult from the technical platform MOBICYTE of the University of Reims Champagne-Ardenne for their excellent technical assistance.

**Conflicts of Interest:** S.D. serves as Chair of the Scientific and Clinical Advisory Board and has equity interest in Apmonia Therapeutics (Reims, France), a biotechnology company developing anticancer strategy. The other authors declare that they have no further financial or other conflicts of interest in relation to this research and its publication.

## **References**

