Prognostic Value of Human Kallikrein 10 Expression in Epithelial Ovarian Carcinoma

hK10,² also known as normal epithelial cell-specific 1 protein, is a serine protease, which is secreted by breast and other epithelial cells (1). hK10 is encoded by a gene designated as KLK10(2), which maps on chromosome 19q13.3–4, spans about 5.5 kb of genomic sequence, and contains 5 coding exons and 1 untranslated exon (3). KLK10 is expressed in many tissues, including ovary, breast, prostate, colon, and testis, with the highest expression found in ovary. The physiological function of hK10 is still not clear, but it has been suggested that hK10 is a tumor suppressor, based on its down-regulation in breast and prostate cancer cell lines (1). This hypothesis is additionally supported by its ability to suppress tumor formation in nude mice (4).

hK10 is a new member of the human kallikrein family, comprised of 15 serine proteases, including hK1 (pancreatic/renal kallikrein), hK2 (glandular kallikrein 2), hK3 (prostate-specific antigen), hK4, and so on, up to hK15 (5). These serine proteases participate in various physiological processes, including release of vasoactive kinins, cleavage of seminogelins, activation of other enzymes, and so forth (5, 6). Some kallikreins are valuable disease biomarkers. The most well-known kallikrein is prostate-specific antigen, which is widely used as a biomarker for prostate cancer (7). Recently, other kallikreins have emerged as new biomarkers for various diseases (5). For example, hK2 is now being investigated as a new prostate cancer biomarker (8). hK6 was found to have value in the diagnosis of Alzheimer’s disease (9) and hK4, hK5, and hK8 have been examined as ovarian cancer prognostic markers (10, 11, 12). Although the expression of hK10 has been reported to be down-regulated in breast and prostate cancer cell lines, and hK10 appears to function as a tumor suppressor (1, 4), its role in the pathogenesis of human disease has not been studied. To investigate the potential value of hK10 as a disease biomarker, we have recently developed a highly sensitive and specific immunoassay for hK10 (13). With this assay, we were able to quantify hK10 protein in many tissues and biological fluids. We found that hK10 is abundantly expressed in the epithelial cells of ovarian tissue (13). The aim of this study is to investigate whether hK10 protein levels in ovarian tumor extracts correlate with clinical outcomes.

One hundred eighty-two patients with primary ovarian cancer were included in this study. These patients underwent surgery for ovarian cancer at the Department of Gynecology, University of Turin. Ages of these patients ranged from 25 to 82 with a median of 59 years. Clinical and pathological information were documented at the time of surgery, including stage, grade, histological types, residual tumor, debulking success, menopausal status, and response to chemotherapy. The staging of the tumors was according to the International Federation of Gynecologists and Obstetricians criteria. The classification of the histological types was based on the WHO and International Federation of Gynecologists and Obstetricians recommendations. Most of the tumors (81) included in this study were of serous papillary histological type, whereas the remaining tumors were endometrioid (30), undifferentiated (27), mucinous (11), clear cell (13), mullerian (12), sarcoma (2), other nonepithelial (4), and unknown (2). The size of the residual tumors ranged from 0 to 9 cm, with a median of 1 cm.

Follow-up information (median follow-up period of 62 months) includes survival status (alive or deceased) and disease status (disease-free or recurrence) and was available for 163 patients. Among these patients, 58 died and 85 relapsed.

Tumor tissues were frozen in liquid nitrogen immediately after surgery and stored at −80°C until extraction. Frozen tissue (200 mg) was first pulverized on dry ice to a fine powder. One ml of extraction buffer [50 mm Tris (pH 8.0), 150 mm NaCl, 5 mm EDTA, 10 g/liter of NP-40 surfactant, 1 mm phenylmethylsulfonyl fluoride, 1 g/liter of aprotinin, and 1 g/liter of leupeptin] was then added to the tissue powders and incubated on ice for 30 min with repeated shaking and vortexing every 10 min. Finally, the mixtures were centrifuged at 14,000 rpm at 4°C for 30 min, and the supernatants (cytosolic extracts) were collected. All of the tissue cytosolic extracts were stored at −80°C until they were analyzed. Protein concentration of the cytosolic extracts was determined with the bicinchoninic acid method with albumin as standard (Pierce Chemical Co., Rockford, IL).

The concentration of hK10 in the cytosolic extracts was quantified with a highly sensitive and specific noncompetitive immunoassay for hK10. This assay was described and evaluated in detail elsewhere (13). The assay incorporates both mouse and rabbit anti-hK10 antisera. In brief, mouse anti-hK10 polyclonal antiserum was captured with sheep antimouse IgG, Fc fragment-specific antibody (Jackson Immunoresearch, West Grove, PA) on 96-well polystyrene microtiter plates. Cytosolic extracts (25 μl) and 100 μl of BSA (60 g/liter) were then added into each well, incubated for 1 h with gentle shaking, and washed. Rabbit anti-hK10 polyclonal antiserum was subsequently applied, incubated, and washed. Finally, alkaline phosphatase-conjugated goat antirabbit IgG (Jackson Immunoresearch) was added, incubated, and washed as before. To detect the signal, time-resolved fluorometry was used. The detection range of this assay is 0.05–20 μg/liter. All of the tumor extracts were measured in duplicate, and hK10 concentrations in μg/liter were converted to ng of hK10/mg of total protein to adjust for the amount of tumor tissue extracted.

A rabbit polyclonal antibody was raised against hK10 full-size recombinant protein produced in yeast cells (13). Immunohistochemical staining for hK10 was performed according to a standard immunoperoxidase method. Briefly, paraffin-embedded tissue sections (4 μm) were fixed and dewaxed. Endogenous peroxidase activity was blocked with 3% aqueous hydrogen peroxide for 15 min. Sections were then treated with 0.4% pepsin at pH 2.0 for 5 min at 42°C and blocked with 20% protein blocker (Signet Labs) for 10 min. The primary antibody was then added at 1:400 dilution for 1 h at room temperature. After washing, we added biotinylated secondary antibody (Signet), diluted 4-fold in antibody dilution buffer (DAKO). We then added the streptavidin horseradish peroxidase complex for 30 min at room temperature. Detection was achieved with amino ethyl carbazole for 5–10 min. After counterstaining with hematoxylin, the slides were mounted with coverslips.

Statistical analysis was performed with SPSS software (SPSS Inc., Richmond, CA). To analyze data, patients were divided into different groups according to clinical and pathological parameters. Because the distribution of hK10 protein concentration in the ovarian tumor cytosols was not Gaussian, the differences between groups were determined by the nonparametric Mann-Whitney U test, and the analysis of differences among more than two groups was performed with the Kruskal-Wallis test in which hK10 was considered as a continuous variable. hK10 values were also classified into two categories (hK10-high and hK10-low groups), and their relationships to various clinicopathological variables were analyzed with the χ² test and the Fisher’s exact test (where applicable). The impact of hK10 on patient survival (PFS and OS) was assessed with the HR (a relative risk for relapse or death) that was calculated with the univariate and multivariate Cox proportional hazards regression model (14). In the multivariate analysis, the clinical and pathological variables that may affect survival, including stage of disease, tumor grade, residual tumor, histological type, and age were adjusted. Kaplan-Meier PFS and OS curves (15) were constructed to demonstrate the survival differences between the hK10-high and hK10-low patients. The log rank test (16) was used to examine the significance of the differences among the survival curves. Furthermore, the patients were divided into different subgroups based on disease stage, tumor grade, and debulking success. The survival analysis was then repeated separately for each subgroup of patients.

hK10 concentration in ovarian tumor cytosols from 182 patients ranged from 0 to 84 ng/mg of total protein. The mean was 7.1 ng/mg with a median of 2.6 ng/mg. The frequency distribution curve is shown in Fig. 1. hK10 concentration was highly elevated in ovarian tumor cytosols compared with the cytosols prepared from normal ovarian tissues or tissues with benign ovarian disease (Fig. 2). The mean, SE, and range of values, were: normal ovarian tissues, 0.27 ± 0.06, 0–0.62; benign ovarian disease, 0.61 ± 0.33, 0–3; ovarian cancer, 7.1 ± 0.7, 0–84. The optimal hK10 cutoff value for additional analyses was selected by the χ² test based on the ability of this value to predict the OS of the study population. A value of 1.35 ng/mg of total protein was found to be the optimal cutoff (χ² = 6.3; P = 0.012) and represents the 33^rd percentile. Tumors were then dichotomously categorized as hK10-high and hK10-low (Fig. 1).

The distributions of various clinicopathological variables among hK10-high and hK10-low patients are summarized in Table 1. The relationships between hK10 and these variables were examined with χ² test and Fisher’s exact test. No relationship was observed between hK10 status and tumor grade, menopause status, and response to chemotherapy. However, hK10-high patients more frequently had advanced disease stage (stage II-IV), serous histological type, larger residual tumor (>1 cm), and suboptimal debulking (all P < 0.05). With Mann-Whitney U test, it was also demonstrated that hK10 concentration was significantly higher in cytosols from stage III/IV ovarian cancer than those from stage I/II, indicating that high hK10 concentration is associated with advanced disease stage (Fig. 3).

The results of survival analysis are presented in Table 2. In univariate analysis, hK10-high patients had significantly increased risk for relapse (HR = 1.93) and death (HR = 2.42; P < 0.05). When hK10 was considered as a continuous variable, a similar result was also observed. Kaplan-Meier survival curves demonstrated survival differences between hK10-high and hK10-low patients. As Fig. 4 shows, the probabilities for PFS and OS are lower in hK10-high patients than in hK10-low patients. In multivariate analysis, PFS and OS in hK10-high patients were no different from the survival rates of hK10-low patients (Table 2). Stage of disease, grade, and residual tumor size were identified to have prognostic significance in univariate analysis; however, in multivariate analysis, only stage and residual tumor remained significant.

The patients were divided into different subgroups based on disease stage, tumor grade, and debulking success. Univariate and multivariate survival analyses were then performed. The results are shown in Table 3. Univariate analysis has shown that among patients who have stage III/IV disease, hK10-high patients are ∼2-fold more likely to relapse and die than hK10-low patients. This survival difference remained significant even after the data were subjected to multivariate analysis. However, this was not observed among patients who have stage I/II disease. hK10 status has no effect on relapse and survival between subgroups of patients who have different tumor grade or debulking success. When Kaplan-Meier survival curves were constructed, they showed similar results (Fig. 5).

In Fig. 6 we present the immunohistochemical localization of hK10 protein in a few representative specimens from ovarian cancer patients.

In this study, we found that hK10 protein is expressed in the epithelial cells of the ovary and that its expression is dramatically elevated in cancerous versus normal tissues. This elevation was more strongly associated with late disease stage, serous histological type, suboptimal debulking, and large residual tumor. For patients who have stage III/IV ovarian cancer, hK10 overexpression is an independent prognostic indicator, which correlates with poor PFS and OS.

Ovarian cancer is the most lethal gynecological malignancy (17). This is attributable to the fact that at early stage, this disease is occult and asymptomatic. By the time diagnosis is made, more than half of the patients have stage III/IV disease. Furthermore, this disease has a tendency to relapse (18). Some improvement in the OS has been observed in the past decade because of better therapeutic strategies (19). The clinical outcome of ovarian cancer varies from patient to patient. Biomarkers that can predict disease outcome could help tailor different therapeutic strategies to meet individual needs. Our results indicate that hK10 is one such prognostic biomarker. Combined with other prognostic indicators, this new biomarker may contribute to patient subclassification for the purpose of individualizing more effective treatments to such subgroups.

Advanced disease stage, serous histological type, and large residual tumor are known indicators for aggressiveness and poor outcome in ovarian cancer (20). In our study, we also observed the same association (Table 2) and have shown that hK10 expression correlates with these clinicopathological features. The biological rationale underlying the overexpression of hK10 and its association with aggressiveness in ovarian cancer is not clear. It is known that the aggressiveness of a tumor largely depends on its ability to invade adjacent tissues and to metastasize to distant sites. During the process of cancer invasion and migration, natural barriers such as interstitial connective tissues and basement membranes have to be degraded. Proteases are widely believed to be involved in these processes (21, 22). Therefore, the amount of proteases released by the primary tumor may reflect the ability of a tumor to spread. Overexpression of a number of other proteases has been reported to be associated with poor outcome in many cancers, such as urokinase plasminogen activator (23, 24), cathepsin D (25), and matrix metalloproteinase (26). In stage III/IV ovarian cancer, the tumor cells have already spread beyond the ovaries. We speculate that hK10 may participate in a cascade reaction, which catalyzes the breakdown of extracellular barriers, and, thus, overexpression of hK10 may facilitate ovarian tumor migration. However, this hypothesis needs experimental verification.

Early detection of ovarian cancer is hampered by the lack of a highly sensitive and specific biomarker. Currently, CA 125 is one widely used serum marker for ovarian cancer, but it is not specific or sensitive enough for diagnosis (27, 28, 29). Other, newly introduced serum markers (30) such as inhibin (31, 32) and OVX1 (33) have shown some promise but have not gained wide acceptance. Recently, we found that serum hK10 concentration (34) are significantly higher in ovarian cancer patients compared with normal individuals. Overexpression of hK10 in ovarian tumors may account for its elevation in serum. If this correlation is established, serum hK10 concentration may also have prognostic value.

In summary, this is the first report describing that hK10 is an independent prognostic biomarker for late-stage ovarian cancer. additional basic and clinical studies are warranted to help understand the role of hK10 in ovarian cancer pathogenesis and progression.