Quantitative assessment and clinical relevance of kallikrein-related peptidase 5 mRNA expression in advanced high-grade serous ovarian cancer

In women, ovarian cancer (OC) is the fifth common cause of cancer-related death [1]. In 2018, over 22,000 new OC cases and about 14,000 deaths due to OC are estimated [2]. Because 75% of OCs are detected at late stage (FIGO stage III and IV) [3], the 5-year survival of patients is less than 47.4% [2]. Late detection of OC is due to the fact that early stage OC frequently does not cause any obvious or specific symptoms [4]. Hence, novel tumor biomarkers for early detection, diagnosis and prognosis of OC are strongly required.

The human kallikrein-related peptidase gene family (KLK1–15) is located within a single cluster in the chromosomal region 19q13.4. All KLK peptidases belong to the chymotrypsin (S1) family of secreted serine proteases [5]. In the past years, numerous reports indicated that KLKs are aberrantly expressed in malignancies of the breast, ovary, prostate, bladder, colon, stomach, lung, and brain [6‐8]. Moreover, accumulating reports demonstrated that KLKs, when overexpressed in malignant tissues, are also detectable in serum and in effusion fluids, consistent with the fact that KLKs are secreted by epithelial/glandular cells [9, 10]. Thus, KLKs may serve as biomarkers in detection of primary cancer, clinical diagnosis, and prediction of the clinical outcome [5].

KLK5 was originally identified as the stratum corneum tryptic enzyme in the normal human epidermis, where it is involved in degradation of corneodesmosomes in the outermost layer of human skin, resulting in skin desquamation [11]. Previous studies have shown that KLK5 mRNA and protein are differentially expressed especially in prostate, ovarian, breast, and testicular tumors [12‐17]. In previous OC-related studies it has been described that KLK5 expression is correlated with progressive disease as well as higher tumor grade [12, 18]. Moreover, overexpression of KLK5 at mRNA and protein level was found to be associated with poor prognosis of OC patients both with respect to progression-free (PFS) and overall survival (OS) [19, 20]. Using OVMZ-6 OC cells simultaneously overexpressing KLK4–7, we observed that KLK5 in concert with the other three KLKs leads in vitro to an induction of TGFß-1 signaling, increased invasion, and chemoresistance as well as enhanced tumor burden in vivo [21‐24].

It is noteworthy, that in most of the previous OC-related studies the patient cohorts encompassed low- and high-grade tumors as well as different histological types (serous, mucinous, endometroid, clear cell) of OC (see e.g. [12, 19]). Hence, to further specify the clinical value of KLK5, we have now analyzed KLK5 mRNA expression and its association with patients’ outcome in a defined OC subgroup, advanced high-grade serous ovarian cancer (HGSOC).

So far, several studies have been published dealing with the prognostic relevance of KLK5 in OC. In a PCR-based study, Kim and co-workers [12] reported that higher KLK5 mRNA levels are associated not only with advanced stage and higher grade, but also with failure of response to chemotherapy. Furthermore, higher KLK5 mRNA levels were linked to unfavorable PFS and OS. Similar results were observed when KLK5 antigen levels, determined by ELISA, were compared in extracts of epithelial ovarian tumor tissue and of low malignant potential (LMP) tumors. There, KLK5 antigen levels were significantly higher in ovarian tumor tissue than in LMP tumors [19]. Furthermore, elevated KLK5 levels were associated with later stage and higher grade as well as shorter PFS and OS. In another ELISA-based study, high KLK5 levels were linked to unfavorable PFS in univariate analysis [30]. Interestingly, larger KLK5 level differentials (between primary tumor and omentum metastasis), as measured by ELISA, were also associated with shorter PFS [31], which may indicate a tumor-supporting role of KLK5 also during metastasis and not only in primary tumor growth. Together with KLK6 and KLK7, KLK5 belongs to the highly expressed KLK genes in OC [32, 33]. Therefore, it is not surprising that KLK5 antigen can also be measured in body fluids (serum; ascites fluid) of patients suffering from OC. In fact, higher KLK5 protein levels were measured in serum samples of patients afflicted with OC compared to healthy controls, benign and/or LMP tumors [20, 34]. Furthermore, elevated KLK5 antigen levels were observed in advanced versus early stages of OC [34]. In both, uni- and multivariate analyses, elevated KLK5 serum levels have been linked to both shorter PFS and OS [35], while in another study a statistically significant association between high KLK5 serum levels with shorter PFS was found in univariate analysis [20]. Taken together, there are data indicating an association between increased KLK5 expression in OC patients and poor prognosis. In fact, there is only one study pointing to a tumor-suppressive function of KLK5 in tumor tissue. There, expression of KLK5 was analyzed by immunohistochemistry in tissue samples of patients suffering from advanced OC. Interestingly, an increased expression of KLK5 in tumor-associated stromal cells, but not in tumor cells, was found to be related to a favorable clinical outcome [25].

In most of the above described clinical studies, OC patients belonging to several subtypes were included, like high and low grade serous, mucinous, clear cell and endometrioid ovarian cancer. However, these subtypes are very different with regard to (i) the origin of the tumors, (ii) molecular characteristics, and (iii) course of the disease. [1, 36]. In the study presented here, we analyzed the clinical relevance of KLK5 mRNA expression in a cohort including exclusively patients suffering from high-grade serous ovarian cancer (HGSOC) stage FIGO III/IV. This histological subtype represents about 70% of all OC cases. It should be noted, however, that excluding all other OC subtypes does result in a rather small cohort size (n = 138).

In line with previous findings that enhanced KLK5 expression is correlated to a higher grade and advanced stage in OC patients, we observed robust KLK5 mRNA expression in the majority of the analyzed cases. Furthermore, mRNA levels were found to correlate with KLK5 protein antigen levels (r_s = 0.683, p < 0.001) indicating that in case of KLK5 expression there is no major post-transcriptional regulation and, thus, mRNA and protein data are comparable.

Our investigation of possible associations between KLK5 mRNA expression and clinical pathological parameters demonstrated that in advanced HGSOC there is a significant association between elevated KLK5 mRNA expression and residual tumor mass after surgery. A significantly increased proportion of tumors displaying elevated KLK5 mRNA expression (p = 0.041) was observed in the patient group with post-operative residual tumor (44%, 29/66), compared to the tumor-free group (27%, 19/70). This suggests an unfavorable role of KLK5 expression with respect to therapy success and therefore prognosis of OC patients. Indeed, in the study presented here, we could demonstrate that an elevated expression of KLK5 mRNA is significantly associated with shortened progression-free survival in univariate analysis (p = 0.047). This result could be confirmed by analysis of Affymetrix-based mRNA data gathered from patients suffering from HGSOC and made accessible for the research community [29]. In multivariable analysis, KLK5 mRNA expression lost significance (p = 0.095), which may be either due to the relative low numbers of included cases or its association with the strong clinical factor ‘residual tumor mass’, which remains as an independent factor (p < 0.001). Interestingly, in the study by Kim and co-workers [12], KLK5 mRNA expression, although significantly associated with both PFS and OS in univariate analysis, only showed independent prognostic value in the subset of tumors with lower grade disease (grades I and II), but not in high-grade tumors.

Neither in our patient cohort nor in the publicly available data set, we observed a statistically significant association of KLK5 mRNA expression with OS. This is in line with several other reports [20, 30, 31], which observed a statistically significant association of KLK5 expression with PFS only. However, other groups have reported that KLK5 expression is linked to both PFS and OS [12, 19]. Whether these differences are due to the rather varying composition of the different cohorts concerning histological subtypes and low-grade versus high-grade tumors can presently not be answered.

Under physiological conditions, one of the major functions of KLK5 - together with KLK7 - is to mediate turnover and desquamation of the skin via degradation of cell-cell and cell-matrix adhesion molecules. Thus, KLK5 overexpression might contribute to unfavorable prognosis in OC patients via supporting tumor cell shedding from the primary tumor as well as cleavage of extracellular matrix proteins during metastasis. In fact, KLK5 efficiently degrades a variety of extracellular matrix (ECM) proteins including fibronectin, laminin and collagen I, II, III and IV [37]. Furthermore, in breast cancer cells, KLK5 overexpression was shown to result in down-regulation of a multitude of miRNAs and up-regulation of another set of mRNAs, finally affecting miRNA networks involved in post-transcriptional gene regulation of ECM molecules and cell adhesion pathways [38]. Thus, the link between increased KLK5 levels with worse outcome indicates KLK5 to be a potential target for therapy.

Moreover, KLK5 targets a broad range of substrates such as insulin-like growth factor binding proteins, transforming growth factor (TGF-β), and protease-activated receptors (PAR), which upon (in-)activation by KLK5 modulate important tumor-associated signaling pathways [39, 40]. In a secretome and degradome profiling study, co-overexpression of KLK4–7 resulted in distinct more than two-fold changes in relative protein abundances as compared to KLK4–7-deficient ovarian cancer cells. Many of the identified differentially expressed proteins are involved in cell-cell communication, including TGF-ß [24]. Furthermore, KLK4–7 regulate gene expression of other cancer-related factors: simultaneous overexpression of these four KLKs lead, e.g., to a distinct up-regulation of moesin and keratin 19 mRNA and protein expression, while keratin 7 is strongly down-regulated [41]. In fact, under physiological conditions, several members of the KLK family are known to interact with each other in different tissues [42]. Analysis of co-over- and under-expression of KLKs in OC and the impact on prognosis should be undertaken in the future, as it may shed more light on the pathobiology of the KLK network in OC. KLK5 may also affect the extracellular proteolytic network in the tumor cell microenvironment by activating the zymogen forms of other tumor-associated proteases including pro-urokinase-type plasminogen activator (pro-uPA) and pro-KLK11 [43]. It is of note that in other cancer cell types functional analyses have demonstrated an association of KLK5 with invasiveness and cell-cell cohesion. In bladder carcinoma cells, siRNA-mediated inhibition of KLK5 expression led to a significant reduction of invasion in Matrigel-based assays [44], whereas in oral squamous cell carcinoma cells silencing of KLK5 enforced cell-cell adhesion that promoted loss of junctional integrity and, hence, metastasis [45].

In summary, the study presented here demonstrates that, in univariate analysis, elevated expression of KLK5 mRNA is significantly related with shortened PFS of patients afflicted with advanced HGSOC. This association was subsequently validated in silico using data from The Cancer Genome Atlas. These results, thus, indicate that KLK5 expression can be considered a prognostic biomarker for PFS in advanced HGSOC. Together with previous findings of clinical, biochemical, and cell biological studies pointing to a tumor-supporting role of KLK5, this kallikrein-related peptidase may be regarded as a novel target for therapy of OC and presumably of other cancers as well.