Epithelial ovarian cancers usually spread whole intraabdominal cavity with peritoneal dissemination, and most of the cases with large amount of ascites are diagnosed at Stage III or more. The current standard treatment for advanced epithelial ovarian cancer is the maximal primary cytoreductive surgery followed by combination chemotherapy with paclitaxel and carboplatin. An alternative to primary cytoreduction in patients with unresectable tumors or poor performance status is the use of chemotherapy in the neoadjuvant setting, which potentially harbors the benefits of earlier treatment start and lower invasiveness at debulking surgery.1–6 Although epithelial ovarian cancer is relatively considered to be chemosensitive disease, patients showing no response to neoadjuvant chemotherapy seem to have poorer prognosis compared with patients who achieved primary cytoreduction, therefore, judgment of chemosensitiveness is a major clinical issue especially in neoadjuvant setting.
As prognostic markers relating to tumor cell biology of epithelial ovarian cancer, prior studies have reported important roles of cell cycle regulators,7 of which, G1-S phase transition is a critical step in cell cycle progression. The p16INK4a-cyclin D1/CDK4-pRb pathway (Rb pathway) works at the G1-S checkpoint. Inactivation of the cyclin-dependent kinase inhibitors including p16INK4a, which has been originally discovered as a tumor suppressor gene by gene deletion in several cancers,8, 9 including ovarian cancer,10 renders the cell susceptible to uncontrolled proliferation signals. Although p16 protein loss has been detected in several cancer types, human papillomavirus-induced carcinomas of the uterine cervix show overexpression of p16, which also marks functional inactivation of RB pathway. p16 overexpression is readily detectable by immunohistochemical analysis and commonly used in the study of human papillomavirus-induced tumor of uterine cervix. In ovarian cancer, expression of p16INK4a has also been investigated in various studies using immunohistochemistry with conflicting results of the prognostic significance of differential expression levels of this protein.11–14 However, a recent study for a larger number of cases demonstrated that p16INK4a negative tumors had significantly worse prognosis in advanced ovarian cancers, showing an independent prognostic significance of p16INK4a expression.15
Although various molecular markers for chemosensitivity have also been suggested to be helpful in predicting response to chemotherapy, limitations of previously reported approaches included the need for adequate tissue samples requiring a certain invasive procedure, hopefully being avoided especially in patients who were considered to be the candidate of neoadjuvant chemotherapy because of poor performance status. Because clinical trial of neoadjuvant chemotherapy in prospective setting showed positive cytology of ascites combined with image scan and serum tumor marker could confirm the diagnosis of advanced epithelial ovarian cancer for potential indication of neoadjuvant chemotherapy,5 use of ascites samples to predict chemotherpeutic response, which can be obtained in less invasive manner, seems to be promising method in those settings. In this study, we investigated p16INK4a expression as a novel prognostic marker for advanced ovarian cancer by immunocytochemistry of ascites using liquid-based cytology samples providing complementary molecular techniques16, 17 and explored the possibility to predict the chemotherapeutic response and prognosis.
Material and methods
Of Stage III or Stage IV ovarian cancer patients at Saitama Medical University International Medical Center and National Defense Medical College Hospital (Saitama, Japan), the following patients were included in this study: (i) patients who have ascites, which was positive for cytologic diagnosis of adenocarcinoma; (ii) patients who harbored measurable tumors before chemotherapy with or without primary debulking surgery and (iii) patients who agreed to participate in the current study with written informed consent. The patients were treated with chemotherapy using paclitaxel (175 mg/m2) and carboplatin (AUC = 6) after primary debulking surgery or as neoadjuvant chemotherapy, which were determined clinically according to their performance status or operability. After 2 or 3 cycles of chemotherapy the tumor responses were evaluated using computed tomography or magnetic resonance imaging and the patients were divided into the following 3 groups: (i) partial response (PR) group; (ii) stable disease (SD) group; and (iii) progressive disease (PD) group. A total of 37 patients were included in the study. Patients whose tissue specimens were also available at the same time when ascites was obtained from primary debulking surgery were typed according to the WHO criteria and classified as either serous, mucinous, endometrioid, clear cell, whereas patients who received neoadjuvant chemotherapy were typed as adenocarcinoma, not otherwise specified. The study protocol was approved by each institutional review board.
All samples of ascites were submitted for routine cytologic examination in the usual manner. Specimens were spun down and the supernatant fluid was discarded. The pellet was mixed to form a homogeneous cell slurry with an 50% cell-crit and 5-10 drops of which were added to 10 mL of CytoRich RED Preservative Fluid (TriPath Imaging, Inc., Burlington, NC) and mixed to an even cell suspension. The specimen was processed using the Cytyc ThinPrep processor (Cytyc Corporation, Boxborough, MA) and stained using the Pap staining technique for routine screening. Specimens in which clusters of neoplastic cells suggesting adenocarcinoma were recognized were diagnosed as positive for adenocarcinoma.
To evaluate p16 expression, at least 2 ThinPrep slides were made from each case. One was treated with p16 antibody and the other was used as a negative control. Immunostaining for p16INK4a primary antibody was performed using p16INK4a Research Kit (DakoCytomation, Glostrup, Denmark) according to the manufacture's instructions. Cervical cancer specimens were included as positive controls. Ascites sample was considered as immunopositive for p16INK4a when at least 10 neoplastic cells constituting more than 1 cluster were stained, whatever the intensity, provided at least 100 neoplastic cells could be counted in the whole specimen. Staining patterns of nuclear only, cytoplasmic only and combined nuclear/cytoplasmic were accepted as specific. Slides were examined by 2 independent observers with no knowledge of clinical data. Conflicting results were reviewed until final agreement was achieved.
Cell lines and culture conditions
KF28 is a single-cell clone of the human ovarian carcinoma cell line KF. KFr13Tx is a paclitaxel-resistant subline derived from cisplatin-resistant KFr13 cells which derived from KF28 cells as described previously.18, 19 These cell lines were grown as monolayer cultures in RPMI-1640 (Immuno-Biological Laboratories Co., Ltd, Gunma, Japan) medium supplemented with 10% fetal bovine serum (Invitrogen Japan K.K., Tokyo, Japan), 2 mM glutamine, 100 units penicillin/mL and 100 μg streptomycin/mL (Invitrogen, Japan K.K.) in a humidified atmosphere of 5% CO2 at 37°C and routinely tested for mycoplasma infection.
Western blotting analysis
Protein concentrations were determined by Bradford assay (Bio-Rad Laboratories, Hercules, CA) and equal amounts of whole cell extracts or nuclear and cytoplasmic protein fractions were separated on a 10% SDS-polyacrylamide gel before electrotransfer onto a polyvinylidene difluoride membrane (Hybond P, GE Healthcare UK Ltd., England, UK). Nonspecific binding sites were blocked by overnight incubation with 5% dried skimmed milk in Tris-buffered saline (TBS, 130 mM Nacl, 20 mM Tris, pH 7.6). Primary antibodies to p16INK4a (DakoCytomation) and Lamin B1 (Santa Cruz Biotechnology, Inc, CA) were used at 1:1,000, whereas the antibody to β-actin (Abcam, Cambridge, UK) was diluted 1:100,000. Primary antibodies were detected using horseradish peroxidase linked anti-mouse or anti-rabbit conjugates as appropriate (Dako Cytomation) and visualized using the ECL detection system (GE Healthcare UK Ltd, England, UK). Densitometric quantification of each band was normalized in comparison with the relevant expression of β-actin or Lamin B1 in the reprobed blot.
Chi-square test was used to measure the correlation between p16INK4a immunopositivity and chemotherapeutic responses. Overall survival was defined as the time from the start of the treatment until death or end of the follow-up period. Kaplan-Meier plots and the log-rank test were used to analyze the association of p16INK4a expression with survival. A p value below 0.05 was considered to be statistically significant.
The characteristics of patients included in the study are detailed in Table I. Thirty-seven ovarian cancer patients, including 27 Stage III patients and 10 Stage IV patients, were enrolled. They consisted of 18 serous adenocarcinomas, 2 mucinous adenocarcinomas, 3 endometrioid adenocarcinomas, 3 clear cell adenocarcinomas, which were diagnosed at primary debulking surgery and 11 adenocarcinomas, not otherwise specified, which were diagnosed by cytology of ascites before neoadjuvant chemotherapy.
Figure 1 shows representative immunocytochemical stainings of p16INK4a immunopositive (a,b) and immunonegative (c,d) cytology of ascites. By combination with Pap staining the diagnosis of adenocarcinoma was not complicated in these cases as many 3-dimensional clusters with acinar or papillary groups of cells were observed. Tumor cells characteristically showed eccentrically placed nuclei, touching the cellular borders and occasionally vacuolated cytoplasm. As for nucleus, irregular shape, high nuclear/cytoplasmic ratios and irregularly distributed chromatin were also distinctive. In most immunopositive cases, immunoreactivity was diffusely observed in cytoplasm, although sometimes it was both nuclear and cytoplasmic.
Chemotherapeutic response and p16INK4a immunopositivity
Table II summarizes chemotherapeutic response and p16 immunopositivity. Twenty-one of 21 (100%) responders (PR) and 3 of 6 SD and 3 of 10 PD showed p16INK4a immunopositivity (p < 0.001). The rates of p16INK4a immunopositive cases were also significantly different between PR and SD/PD (p < 0.001) as well as PR/SD and PD (p = 0.001).
Table II. Chemotherapeutic Response and p16INK4a Immunopositivity in Cytology of Acites
Response rates to chemotherapy and p16INK4a immunopositivity according to histological subtypes
Figure 2 shows representative immunohistochemical stainings of p16INK4a immunopositive histological specimens from patients with serous adenocarcinomas. Strong staining islands of ovarian cancer cells and in contrast, absent stromal cell staining was observed in immunopositive cases. Adjacent table demonstrates the summary of response rate to chemotherapy and p16 immunopositivity according to histological subtypes. Immunopositive cases were frequently observed in serous adenocarciomas (17 of 18, 94%). Response rates and p16 immunopositivity showed similar trend in each histological subtype.
Subcellular distribution of p16INK4a immunopositivity
In cytology of ascites, cytoplasm was diffusely immunostained in p16INK4a immunopositive cases (n = 27), 12 of which showed stronger nuclear immunostaining shown in Figures 3a and 3b. Overall survival were analyzed between immunopositive and immunonegative cases (Fig. 3c, p = 0.0006) as well as between cases with and without stronger nuclear immunostaining (Fig. 3d), which demonstrated superior overall survival in cases with stronger nuclear immunopositivity.
Localization of p16INK4a expression in ovarian cancer cell lines
To further investigate our present findings, ovarian cancer cell lines were used for in vitro analysis. KF28, parent chemosensitive cell line and KFr13Tx, chemoresistant cell line, were analyzed by immunocytochemistry for p16INK4a expression. Chemosensitive KF28 cells showed stronger nuclear staining (Fig. 4a) compared with chemoresistant KFr13Tx, which showed stronger cytoplasmic staining (Fig. 4b). These findings were also confirmed by western blotting, which demonstrated remarkable differential p16INK4a expression in nuclear fraction and cytoplasmic fraction (Fig. 4d), although expressions in whole cell protein were comparable (Fig. 4c).
In this study, we described the expression of p16INK4a, detected by immunocytochemistry in ascites of advanced ovarian cancers before first-line chemotherapy. We demonstrated p16INK4a expression was correlated with chemotherapeutic response and prognosis. We previously reported deletion of the gene is an indicator for poor chemotherapy response and adverse prognosis in advanced ovarian cancer.20 Conflicting results of the prognostic significance of p16INK4a expression in ovarian cancer have been reported in the past. In the large retrospective study the prognostic significance of this marker was confirmed for uniformly treated group of patients with advanced-staged ovarian cancer,15 which was relatively close to our patient population. The present study also contains only patients with malignant ascites who could not achieve optimal cytoreduction and received first-line chemotherapy, thus chemotherapeutic response was considered to be the most important prognostic factor in these patients. Indeed, p16INK4a expression was strongly associated with chemotherapeutic response, which in turn seemed to demonstrate distinctive overall survival. Decreased expression of p16INK4a could possibly lead to intensified proliferation of neoplastic cells and chemotherpeutic drug resistance.
For histological subtypes, p16INK4a immunopositivity was considerably frequent in patients with serous adenocarcinomas similar to the past reports.21, 22 Some reports demonstrated that loss of p16 expression was detected mainly in mucinous and endometrioid types.11, 23 Because of patients eligibility including only advanced-stage patients with massive ascites, serous adenocarcinoma was the most representative histological subtype in the present study and the differential expression pattern according to histological subtypes could not be drawn from our results.
In the present report, it is noteworthy that cases with stronger nuclear immunoreactivity of p16INK4a showed especially better overall survival. In in vitro analysis using representative ovarian cancer cell lines with differential chemosensitivity, cellular localization of p16INK4a was clearly different between parent chemosensitive cells and chemoresistant cells. The importance of cellular localization of p16 was reported in the evolution of colorectal carcinoma, demonstrating that carcinomas showed more cytoplasmic overexpression for p16 than adenomas and that normal epithelium showed more nuclear overexpression than adenomas and carcinomas.24 In contrast, in the study of endometrial cancer, it was reported that the cytoplasmic staining of p16INK4a was observed in normal and malignant endometrial tissues, whereas nuclear staining was observed only in endometrial carcinomas.25 Another study demonstrated that both nuclear and cytoplasmic p16 formed complexes with cdks although, cytoplasmic p16 was present in 2 forms, whereas nuclear p16 was in 1 form,26 indicating different unknown functional mechanism. The translocation of intracellular p16 and chemosensitivity needs to be further investigated, loss of p16 expression in the nucleus could suggest its functional loss in cell cycle regulation and the cytoplasmic expression of p16 may have different functional consequences from its status in nucleus.
Immunocytochemistry has presented an important impact on cytological diagnosis of body cavity fluids.27 Liquid-based cytology technique, by which additional multiple slides can be easily produced from a single specimen, seemed to be suitable especially in immunocytochemistry. The use of the ThinPrep processor has also resulted in increased cellularity and marked reduction of debris and erythrocytes.28, 29 The assessment of immunocytochemistry in cytology of ascites used in this study has also been applied in some previous studies, showing the impact of biomarkers to be examined in the diagnosis of malignant diseases30, 31; however, the development of reproducible definition of cutoff values seems to be important assignment to achieve comparability of results to be widely used.
This study is limited by relatively small number of cases, and selection bias cannot be completely ruled out. Further analysis by a large prospective study is needed to confirm our findings. However, the results of our study suggest the assessment of p16INK4a expression level in cytology of ascites could be a less invasive and convenient predictive method in advanced-staged ovarian cancer especially in consideration of neoadjuvant chemotherapy.
In conclusion, this study demonstrates that immunocytochemistry for p16INK4a expression in cytology of ascites is useful in prediction of the primary response to chemotherapy and the disease outcome. We recommend a large multicentre prospective study to confirm the clinical significance of p16INK4a in cytology of ascites in advanced epithelial ovarian cancer be performed.