The WNT5A/ROR2 signaling pathway in pancreatic ductal adenocarcinoma (PDAC)

WNT5A/ROR2 signaling pathway has been shown to be involved in many human cancers. Its role in pancreatic ductal adenocarcinoma (PDAC) has not been claried yet. The aim of this study was to determine the prognostic value of WNT5A-expression in conjunction with the ROR2-expression in the same tumor tissues of PDAC patients. We retrospectively analyzed the expression of WNT5A and ROR2 in 117 paran-embedded PDAC specimens following surgical pancreatic resection by immunohistochemistry. The prognostic value of WNT5A and ROR2 was assessed using Kaplan-Meier survival curves and multivariate COX regression-models. is highly dependent on the different molecular receptor-ligand tissue contexts.


Introduction
Pancreatic ductal adenocarcinoma (PDAC) belongs to the most fatal malignancies counting as the seventh leading cause of cancer-related death worldwide while still demonstrating the poorest prognosis of all solid tumors [1] . Although radical surgical R0resection, in combination with adjuvant and / or neoadjuvant treatment, remains the most promising chance of cure, high recurrence rates have been recorded [2] , while unresectable locally advanced and disseminated disease display an unfavorable prognosis, with 5-year survival less than 9%, due to aggressive tumor biology and poor medical treatment e cacy and effectiveness [3] . Alternative therapeutic modalities, such as molecular therapy, which targets speci c biological markers that signi cantly in uence the lymph node metastasis and tumor progression, may add further therapeutic or prognostic bene ts. Identi cation of such novel targets or prognostic markers requires understanding of the molecular pathways underlying the PDAC tumor biology, which still remains largely unclear.
The embryological WNT-signaling pathway is important for development and tissue homeostasis, while its dysregulation has been associated with tumorigenesis [4] . WNT-signaling is either canonical or noncanonical, based on the role played by the effector protein β-catenin [5] . In the canonical WNT-signaling, WNT-signals activate β-catenin mediated transcription of target genes that promote proliferation and differentiation [6] . In contrast, two distinct, non-canonical WNT-signaling pathways, the planar cell polarity (PCP) / convergent extension (CE) pathway, and the Ca 2+ pathway (summarized in [7] ), are activated by non-transforming WNT family members and regulate cell migration and polarity, independently of βcatenin [7] .
WNT5A is classi ed as a non-transforming WNT family member which activates non-canonical WNTsignaling by interfering with receptor tyrosine kinase-like orphan receptors (ROR) [8] . ROR2 is a member of the ROR-family of receptor tyrosine kinases, acts as a receptor or co-receptor for WNT5A, and has been shown to display essential functions in developmental morphogenesis [9] . Activation of WNT5A/ROR2 signaling is reported to inhibit the canonical WNT-signaling in tumor cells, suggesting that WNT5A is a tumor suppressor [10] . In fact, absence of WNT5A-expression is associated with increased invasion and tumor progression of many carcinomas [11,12] . However, WNT5A is speculated to also have oncogenic potential based on reports demonstrating signi cant WNT5A-overexpression in melanoma, lung and breast cancers [13][14][15] . Meanwhile, ROR2-receptor demonstrates oncogenic properties regardless the presence of a ligand [16] . Therefore, the exact function of WNT5A appears highly dependent on the receptor-context and its signi cance in tumor biology needs to be examined in conjunction with the ROR2-expression.
PDAC is characterized by dysregulation of the canonical WNT/β-catenin signaling pathway, which has been associated with tumor initiation, progression and aggressive tumor biology [17][18][19] . The canonical WNT-signaling is reported activated in up to 65% of PanINs and increased even more frequently in invasive PDAC [20][21][22] . In contrast, only limited and independent reports exist so far investigating the role of non-canonical WNT5A and its ROR2-receptor in PDAC respectively. WNT5A has been shown to mediate the migration and invasion of pancreatic cancer cells via epithelial-to-mesenchymal transition (EMT) [23,24] . On the other hand, high ROR2-expression in tumor cytoplasm or stromal cells in PDAC has been signi cantly associated with malignant attributes and poor patients' survival [25] .

Page 4/28
The already existing reports on WNT5A-mediated aggressive tumor phenotype and ROR2-associated poor prognosis in PDAC, stress that downstream activation of WNT5A/ROR2 signaling pathway may signi cantly contribute to tumor biology. However, the prognostic value of WNT5A-expression in conjunction with the ROR2-expression in PDAC has not been investigated yet, and their true clinical signi cance in combination remains still ambiguous. Thus, the goal of this study was to de ne the expression patterns and the clinical signi cance of both, WNT5A and ROR2, in the same pancreatic cancer tissues in an effort to show their true prognostic value and to clarify their possible biological implications in PDAC.

Materials & Methods
Patients and tissue samples Formalin-xed, para n-embedded PDAC samples following pancreatic resection in curative intent and matched, tumor-adjacent normal pancreatic specimens were collected from 117 patients with PDAC, who underwent surgery at the University Hospital of Leipzig, Germany, from 2001 to 2013. No patient received chemotherapy or radiotherapy prior to the operation. Clinical data were obtained by medical records, and follow-up survival data were gained through telephone investigation. The data included epidemiological data, such as gender and age, as well as histopathological analysis, as tumor differentiation (grading), and the pTNM-classi cation, lymphatic invasion (L), vascular invasion (V), perineural invasion (Pn), residual tumor (R), and UICC-Union for International Cancer Control Stage. Pathological tumor stage (pTNM) and histological differentiation were classi ed by GI experienced pathologist (C.W., K.S.) according to the 7 th Edition of TNM staging in pancreatic cancer [26] . Data regarding adjuvant chemotherapy and distant or local relapse were also collected by follow-up. Clinical data were collected prospectively and retrospectively analyzed. This study was approved by the Ethic's Committee of the University Hospital of Leipzig, Germany (reference number: 234/14-ek).

Immunohistochemistry
Para n-embedded tissue samples were cut into 5 µm slices and dried at least 6 hours at 37° C, depara nized and rehydrated. Endogenous peroxidase activity was blocked by 3 % hydrogen peroxide at 4° C and afterwards, non-speci c bindings were blocked with 5 % normal-goat serum. The slides were incubated with the primary antibody against ROR2 (1:100, PA5-14727, Thermo Fisher Scienti c, Darmstadt, Germany) and WNT5A (1:200, MA5 15511, Thermo Fisher Scienti c, Darmstadt, Germany) overnight at 4° C. Horseradish-peroxidase-conjugated A ni-Pure goat anti-rabbit antibody IgG (Dianova, Hamburg, Germany) for ROR2 and anti-mouse antibody IgG (Dianova, Hamburg, Germany) for WNT5A were then applied. Immunoreactivity was visualized with diaminobenzidine (Sigma, Taufkirchen, Germany). Positive control for ROR2 was a renal cell carcinoma sample and for WNT5A a lymph node. Negative control was fatty tissue for both (Suppl. Figure 1A). H&E staining facilitated the recognition of the tumor region, stroma and adjacent normal tissue in all tissue samples (Suppl. Figure 1B).

Evaluation of immunohistochemistry reaction
The slides were evaluated by three independent reviewers (L.R., O.L., G.W.). The evaluation of expression of ROR2 and WNT5A was based on the percentage of positive cells [0 % (no positive cells), 35 %, 65 % and 100 % (all cells positive)] and their intensity [0 (no staining), 1 (weak intensity), 2 (moderate intensity) and 3 (strong intensity)]. For each tissue sample, the percentage of cells stained at certain intensity was determined and multiplied by the intensity score to generate an intensity percentage score, which ranged from 0 to 300, as described previously [25] . Expression in the tumor, stroma and normal tissue were analyzed. The cut-off value for distinguishing positive and negative expression was set at 65 based on consensus of the three independent experts.

Statistical Analysis
Biomedical statistician performed the statistical review of the study. A χ2-test was used to test the correlation of ROR2-and WNT5A-expression with clinicopathological variables. Overall survival (OS) was calculated from day of rst surgery until death or end of follow up using the Kaplan-Meier method, and the log-rank test was used for survival analysis. Cox-regression model was used to perform multivariate analysis. Correlation between ROR2 and WNT5A expression was tested using spearman correlation. For all tests, the signi cance level for statistical analyses was set at p<0.05. All data were analyzed using IBM © SPSS © Statistics (Version 24.0.0, IBM © , Armonk, USA).

Patient Cohort
In total, tissue samples were collected from 117 patients with PDAC, who underwent pancreatic resection at the University Hospital of Leipzig, Germany, from 2001 to 2013. At the time of surgery, patients' ages ranged from 40 to 84 years, with a median age of 66 years. There were 45 females (38.5%) and 72 males (61.5%). None of the patients had received neoadjuvant treatment prior to surgery. The surgical procedures included Kausch-Wipple operations in 53 patients, pylorus-preserving pancreaticoduodenectomy (PPPD) in 50 patients, left pancreatectomy in 12 patients, and total pancreatectomy in 2 patients. In 9 patients with intraoperatively detected oligometastasis (M1), the operation was still performed along with removal of the oligometastatic site due to younger age (5 patients with singular liver metastasis and 4 patients with interaortocaval lymph node metastasis). According to the decision of the interdisciplinary tumor board, based on the nal pathology report, 50 (43%) patients received adjuvant chemotherapy. The median follow-up reached 100 months (8 years). The median overall survival was 19.5 months with a 95% con dence interval [CI] of 14.4 -24.6 months ( Figure 1). Patients who received adjuvant chemotherapy demonstrated an improved overall survival. The 5-years survival rate following curative resection reached 16%. These ndings correlate with the statistics reported in patients able to undergo a successful curative resection, in which median survival time ranges from 12-19 months, and the 5-year survival rate is 15-20 % [27] . Patients' characteristics are summarized in Table 1.

Expression of ROR2 and WNT5A in PDAC
To analyse the expression of ROR2 and WNT5A in PDAC, we performed immunohistochemistry in all 117 tumor samples along with the adjacent normal tissue, if available (n=90 in ROR2 and n=88 in WNT5A). Positive staining of ROR2 and WNT5A were primarily localized in the cytoplasm of tumor cells, in the stroma and in normal pancreatic tissue. For statistical analysis, we dichotomized the expression of both, ROR2 and WNT5A, at a cut-off value of 65 based on tabular analysis. Tissues with protein staining of 65 or less were considered negative, otherwise they were labelled as positive. High ROR2-expression was found in 65.8 % (77/117) of tumors, in 28.2 % (33/117) in tumor-stroma, and in 71.1 % (65/90) of normal pancreatic tissue. The expression in tumor cells was signi cantly higher than in stromal tissue (p<0.001).

Association between ROR2-and WNT5A-expression patterns with clinicopathological parameters
To further elucidate the concomitant role of ROR2 and WNT5A in PDAC, we analysed the associations between their expression patterns and major clinicopathological parameters. As shown in Tables 2 and 3, no signi cant associations were detected between the positive ROR2-and WNT5A-expression in all three investigated tissue types (tumor, stroma and normal tissue) with any clinical parameter, respectively.

Survival analysis
Univariate analysis showed differences for overall survival between the following groups: differentiation grade (G), tumor depth of invasion (pT), regional lymph node metastasis (pN), lymphatic (L) and vessel invasion (V) along with recurrence and metastasis (Table 4). No signi cant association with overall survival was found for high expression in tumor, stroma and normal pancreatic tissue of either ROR2 ( Figure 3A) or WNT5A ( Figure 3B). Multivariate analysis identi ed the regional lymph node invasion and the grading to be independent prognostic factors of survival. Patients with lymph node invasion showed a Hazard Ratio (HR) of 3.00, thus a 3 times increased risk to cancer related death than patients without lymph node invasion. Poorly differentiated tumors (G3) showed a 1.6 higher risk to cancer related death than tumors of G1 or G2 grading. ROR2-and WNT5A-levels of expression could not be identi ed as independent prognostic parameters. Further subgroup analysis of possible combinations of ROR2-and WNT5A-expression patterns also failed to demonstrate signi cant differences in overall survival ( Figure  4A). Finally, in order to evaluate whether ROR2-and WNT5A-expression patterns may in uence response to adjuvant chemotherapy, we performed subgroup analyses only for patients who had received adjuvant chemotherapy (n=50). In this subgroup analysis, both ROR2-and WNT5A-expression failed to demonstrate any prognostic value in the setting of adjuvant chemotherapy ( Figure 4B & 4C).

Discussion
To our knowledge, this is the rst study to analyze simultaneously the prognostic value of ROR2-and WNT5A-expression patterns in PDAC patients, in an effort to reveal the true role of the non-canonical ROR2/WNT5A signaling pathway in this malignancy. We demonstrated variable expression patterns for ROR2 and WNT5A, in tumor, stroma and normal pancreatic tissues, which, however, failed to demonstrate any type of correlation to each other. High WNT5A-expression was found in 76.9% of PDAC tissues. This is in accordance with the previous study displaying a positive WNT5A-expression in 81.3% of pancreatic cancer tissues [24] . On the other hand, the positive expression rate of ROR2 in our study was higher, up to 65% of the examined PDCA tissues, compared to the previous reported rate of 45.7% [25] . Noteworthy, we demonstrated elevated expression rates for ROR2 and WNT5A also in the adjacent normal pancreatic tissue, up to 71% and 83% respectively. Regarding their clinical signi cance, expression patterns of ROR2 and WNT5A in our analysis failed to correlate with clinicopathological parameters or to demonstrate a prognostic signi cance in survival, even in the subgroup analysis. In contrast, regional lymph node invasion and differentiation levels were de ned as independent prognostic factors of survival.
ROR2 is a co-receptor for WNT5A and exerts its function predominantly via the non-canonical WNTsignaling [9] . On the other hand, WNT5A functions highly dependent on the receptor context, and its role in tumor biology is highly in uenced by the presence of the ROR2-receptor. Therefore, the prognostic signi cance of ROR2 should be evaluated only in conjunction with WNT5A and vice versa. The already reported high expression rates of WNT5A and ROR2 in PDAC along with their association with aggressive tumor phenotype and poor prognosis has been shown in independent studies, respectively [24,25] . These ndings undelay the activation of the WNT5A/ROR2 signaling pathway, which may contribute to the PDAC tumor promotion. However, WNT5A-mediated activation of the ROR2-receptor is claimed to mediate inhibition of the canonical Wnt/β-catenin signalling [28] , whose activation has also been strongly associated with PDAC carcinogenesis and tumor progression [18,22] . Therefore, expression patterns indicative of WNT5A/ROR2 signaling activation, such as double positive cancers (ROR2+/WNT5A+) should be expected to be associated with better prognosis, rather than poorer as reported previously. In an effort to clarify this uncertain background, our study aimed to evaluate the ROR2-and WNT5Aexpressions in combination in the same PDAC tissues and to facilitate a more con dent conclusion on their prognostic value. Although high expression rates of ROR2 and WNT5A were demonstrated, in 65% and 76.9% of the PDAC tissues respectively, our study failed to con rm any association with clinicopathological parameters or to demonstrate a prognostic impact for the expression of both.
Previous reports support an oncogenic role for ROR2 and WNT5A positive expression in PDAC [24,25] , which however could not be con rmed further in our study. The lack of prognostic signi cance can be attributed to the complexity of the ROR2 and WNT5A functions, which are highly dependent on the receptor-ligand context and vary under different molecular conditions. Even the study of Bo et. al [24] , which supported that WNT5A-upregulation promotes EMT and metastasis in pancreatic cancer via activation of the canonical WNT/β-catenin signalling and independently of ROR2 receptor, could not nally demonstrate a prognostic signi cance for WNT5A-expression. On the other hand, Huang and colleagues recently supported that high ROR2-expression is associated with poor prognosis without, however, to evaluate WNT5A-expression patterns or other downstream molecules of the ROR2 receptor in their cohort [25] . Regarding other malignancies, several reports have attempted to investigate the prognostic values of ROR2 and WNT5A in combination, supporting the viewpoint that double positive cancers (ROR2+/WNT5A+) might have signi cantly worse survival than those with double negative cancers (ROR2-/WNT5A-) [29][30][31][32][33][34] , while other studies suggest independent roles for WNT5A as tumorsuppressor and for ROR2-receptor as mediator of WNT5A-independent tumor-promotion [35,36,10] . This variety in WNT5A and ROR2 molecular functions in regard to tumor progression, surely re ects the differences in the receptor-ligand context or cell context of cancer cells even in PDAC and can in part explain the lack of signi cance, even in the subgroup analysis.
An interesting nding in our study is the high rates of positive ROR2-and WNT5A-expression in the normal pancreatic tissues. In adult tissues, expression levels of WNT5A and ROR2 are in general decreased [37] . Thus, the majority of oncological reports tend to speculate the potential of ROR2 and WNT5A as therapeutic targets based on their differential expressions within human malignancies and on the low to absent expression in normal tissues [38] . However, the expression in normal tissues is often neglected in reporting. Therefore, the positive expression of WNT5A and ROR2 in the adjacent normal pancreatic tissue as found here, signi es their role in pancreatic tissue physiology and in ammation. The roles of WNT5A-ROR2 axis in normal cells has been demonstrated, involving diverse cellular functions, including cell polarization, migration and stemness [37] . Such prominent cellular processes in the adjacent to PDAC normal pancreatic tissue may dictate the tumor microenvironment in favour of tumor progression and merit further investigation. This study has some limitations. Due to the retrospective character of the study, the current ndings may not be applicable to the general population. Larger prospective studies are required to verify or to exclude the prognostic value of WNT5A and ROR2 in PDAC. In addition, we utilized immunohistochemistry (IHC), which only assesses protein expression of WNT5A and ROR2 in pancreatic tissues and as such provide rather limited information on the functional level without allowing conclusions on their molecular interactions. Lastly, additional methods are needed to con rm the ROR2 and WNT5A expression levels in PDAC cells, since IHC data are semiquantitative.
In conclusion, in this study, we examine for the rst time the synergistic effect of ROR2 and WNT5A expression patterns on the prognosis of patients with PDAC. We report variable expression patterns for ROR2 and WNT5A in PDAC, stroma and adjacent normal pancreatic tissues, suggesting a role for noncanonical WNT5A/ROR2 signaling pathway in the pancreatic tissue physiology and PDAC tumor biology.
However, both markers failed to demonstrate prognostic signi cance, either alone or in the subgroup analysis, indicating the complexity of their role in PDAC, which appears highly dependent on the different molecular receptor-ligand tissue contexts. The correlation between ROR2 and WNT5A in PDAC merits further evaluation in larger sample size studies along with in vitro studies in order to clarify the mechanism of interaction in tumor progression and pancreatic pathophysiology. Abbreviations PDAC = pancreatic ductal adenocarcinoma, ROR = receptor tyrosine kinase-like orphan receptors,

Declarations
Ethics approval and consent to participate: This study was approved by the Ethic's Committee of the University Hospital of Leipzig, Germany (reference number: 234/14-ek).

Consent for publication: not applicable
Availability of data and materials: The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.
Competing interests: The authors declare that they have no competing interests Funding: This work was supported by the Clinician Scientist Scholarship program of the Faculty of Medicine, University of Leipzig, Germany for Orestis Lyros.
Authors' contributions: Remtisch L performed the experiements and analysis of data, Lyros O concepted, designed and coordinated the study, performed analysis and interpretation of data and wrote the paper; Wiltberger G, Katrin Schierle K. and Thieme R performed analysis and interpretation of data; Moulla Y, Boris Jansen-Winkeln B, Wittekind C and Gockel I performed interpretation of data and critically revised the manuscript.