Clinical Relevance of E2F Family Members in Ovarian Cancer—An Evaluation in a Training Set of 77 Patients

Epithelial ovarian cancer is the leading cause of death from gynecologic malignancies. The major obstacle to more successful treatment of ovarian cancer is the rapid development of resistance to platinum during treatment, and indeed, up to 75% of initial platinum responders relapse within the first 2 years (1). In addition, a small fraction of carcinomas of the ovary are primarily refractory to platinum compounds. This intrinsic resistance that translates to highly malignant, fast-growing tumors exhibiting cross-resistance to most other available cytostatic drugs further deteriorates clinical outcome. All this implies that the overall prognosis of the disease is primarily determined by the point at which platinum resistance occurs.

Most of the reports addressing this issue presume that platinum resistance, either intrinsic or acquired, is multifactorial in origin. Nonetheless, identifying key genes and pathways crucially involved in that phenomenon should be an essential goal in ovarian cancer research, firstly to avoid administering these compounds to patients whose tumors have little chance of responding and secondly to recruit new targets for therapeutic reversal of platinum resistance.

A recent report showing that E2F activity contributes substantially to cis-platinum-induced cell death in vitro by modulating deoxynucleoside triphosphate synthesis (2) prompted us to do a detailed investigation of the E2F pathway in ovarian carcinomas.

E2F function is composed of a family of proteins acting as heterodimeric complexes, which are formed by one member of the E2F family of transcription factors and one member of their coactivators from the family of DP proteins. Regulation of E2F transcription factors is mediated by the “pocket proteins” Rb, p107, and p130. Based on sequence homology, eight members of the E2F family of transcription factors (E2F1 to E2F8) have already been identified, which can be subdivided into two functionally different groups (3–8). E2F1, E2F2, and E2F3a act predominantly as activating transcription factors by targeting proliferation-promoting genes. Moreover, under certain circumstances, including malignant transformation, E2F1 and probably also E2F3a can also be involved in apoptosis by targeting key genes in the p53 and p73 pathway (9, 10). In contrast, E2F3b, E2F4, and E2F5 act predominantly as transcriptional repressors by heterodimeric arrangement with “pocket proteins” and histone deacetylase, which subsequently causes chromatin remodeling (11). The dual mode of action suggests that, at least in normal cellular physiology, transcriptional activity of the E2F family is governed by an autoregulative control through “intrafamilial” cross-talk. On the other hand, the functional role of the newly discovered E2F6, E2F7, and E2F8 is not yet fully understood; they are currently deemed to act predominantly as transcriptional repressors (7, 12, 13).

Mutations in upstream components of the Rb pathway were found to be associated with tumor formation (8, 14). However, there is a growing body of evidence showing that deregulation of E2F transcription factors is also causatively involved in carcinogenesis, and several members of the E2F family are associated with clinical outcome in various cancer entities (15–20). Most of these studies focus on the prognostic value of the proliferation-promoting E2F1 and E2F3. To our knowledge, no data are available on the clinical relevance of deregulated E2F family members in ovarian cancer or on the putative role of these transcription factors in platinum resistance.

We here report on in vivo expression of all known E2F family members and their relevance for disease-free and overall survival as evaluated in a training set of 77 patients. Furthermore, the significance of the various E2F family members in predicting platinum sensitivity was assessed.

Patients. Tissue samples from patients with invasive ovarian cancer were collected during primary debulking surgery at the Department of Obstetrics and Gynecology (Innsbruck Medical University, Innsbruck, Austria; n = 77) between 1998 and 2004. Specimens for this training set were selected on a random basis using the registration code from the gynecopathologic unit: every third ovarian cancer specimen, as numbered by the registration code, from a total of 232 patients undergoing primary surgery for ovarian cancer at our department during the study period was included in the investigation. If a particular registration code number did not contain enough tissue, the specimen from the subsequent registration code number was used.

Only epithelial ovarian cancers were included in this study and tumors with borderline malignancy were excluded. Ovarian tissue samples obtained from postmenopausal patients during surgery for other than inflammatory or malignant conditions served as control (n = 8). All tissue samples were used for research in compliance with the patient and approved by the local Institutional Review Board.

RNA extraction and reverse transcription reaction. Total RNA was isolated from patient samples using the guanidium thiocyanate-phenol-chloroform method according to the manufacturer's protocol (RNAgents Total RNA Isolation System, Promega, Madison, WI). Integrity was evaluated by assessing the 18S and 28S rRNA bands in 1% ethidium bromide–stained agarose gels. To remove any contaminating genomic DNA, DNase treatment of typically 4 μg total RNA was done according to the manufacturer's protocol (Roche, Basel, Switzerland).

Reverse transcription of typically 2 μg total RNA was done in a final volume of 25 μL containing 1× reverse transcription buffer [50 mmol/L Tris-HCl (pH 8.3), 75 mmol/L KCl, 5 mmol/L MgCl₂], 40 units of rRNAsin RNase Inhibitor (Promega), 10 mmol/L DTT, 250 nmol/L random hexamers, and 200 units of Moloney murine leukemia virus reverse transcriptase (Invitrogen, Carlsbad, CA). Incubation periods were 10 minutes at 25°C and 50 minutes at 37°C followed by heating at 70°C for 15 minutes to inactivate the reverse transcriptase enzyme.

Primers and probes. Specific primers and probes for E2F1 to E2F8 and for the TATA box-binding protein (a component of the DNA-binding protein complex transcription factor IID as an endogenous RNA control) were determined with the computer program “Primer Express” (Applied Biosystems, Foster City, CA). To prevent amplification of contaminating genomic DNA, the probe was placed at a junction between two exons. Sequences of primers and probes are shown in Table 1.

Real-time PCR amplification. Real-time PCRs were done using an ABI Prism 7900 Detection System (Applied Biosystems) in a total volume of 25 μL reaction mixture containing 5 μL of each appropriately diluted reverse transcriptase sample (standard curve points and test samples), 12.5 μL Taqman Universal PCR Master Mix (Applied Biosystems), 900 nmol/L of each primer, and 250 nmol/L of the probe. Cycling conditions were an initial step at 50°C for 2 minutes, a denaturing step at 95°C for 10 minutes, and 45 cycles at 95°C for 15 seconds and 65°C for 1 minute. Real-time PCR efficiencies were acquired by amplifying serially diluted cDNA isolated from the ovarian cancer cell line HTB-77. Real-time PCR assays were conducted in triplicate, and the mean value was used for calculation. Levels of E2F transcripts detected in patient samples were normalized to the TATA box-binding protein.

Clinicopathologic characteristics. All patients (n = 77) were monitored on the outpatient follow-up program of the Department of Obstetrics and Gynecology (Innsbruck Medical University), and the median observation period of included patients was 54.6 (24.12-85.08) months. Clinicopathologic characteristics are summarized in Table 2. Although tissue samples were randomly selected, mucinous ovarian cancers were overrepresented with 42.9% in the investigated training set. This is not in accordance with the general incidence of this histologic subtype, which was 28.6% for the period study samples were collected. Primary debulking surgery was done in all but one patient, who received carboplatin-based chemotherapy due to impaired performance status. With the exception of six patients who presented in Fédération Internationale des Gynaecologistes et Obstétristes (FIGO) stage Ia and Ib with low-grade tumors, all patients received six cycles of a platinum-based chemotherapy (92.2%; n = 71). During follow-up, intrinsic platinum resistance, defined as progression under first-line therapy, was observed in 12 patients (15.6%), whereas platinum resistance, defined as relapse within 6 months after end of first-line therapy, occurred in 11 patients (14.3%). All other patients, including those showing no recurrence (32; 41.6%) and those with relapse beyond 6 months after completion of first-line therapy (16; 20.8%), were defined as platinum sensitive. Altogether, 40 of 77 patients (51.9%) died during follow-up.

Statistical analysis. Differences in expression level of E2F transcripts between normal and malignant tissues were assessed with the Mann-Whitney U test. Because E2F expression showed neither a clear negative value nor a biphasic distribution and no clinically relevant cutoff levels have thus far been defined for E2F expression in ovarian cancer, we had to determine the optimal cutoff level in our training set of 77 patients. For this purpose, each percentile from 10th to 80th was calculated for E2F1 to E2F8 expression and used as cutoff point. At each of these cutoff values, the data were divided into two groups for low and high expression, and difference in overall survival was assessed with the log-rank test. The greatest difference between the two groups was defined as the optimal clinical cutoff value (Table 3).

Differences between groups characterized by low or high E2F expression in clinicopathologic characteristics and platinum resistance were evaluated by the Mann-Whitney U test. Survival analyses were done for disease-free and overall survival with the Kaplan-Meier method, and differences between groups were determined with the log-rank test. To assess the independence of the predictive value of E2F expression, the Cox proportional hazards model with a stepwise backward method was used with adjustment for confounding variables. The final model included age at diagnosis, histopathologic grading, histologic subtype, FIGO stage, residual disease after primary debulking, and levels of E2F expression. To evaluate a possible influence of the cross-talk between proliferation-promoting and proliferation-inhibiting E2F transcription factors on prognosis, ratios of E2F1 or E2F2 to E2F4 or E2F7 were calculated and survival was analyzed with the Kaplan-Meier and the Cox proportional hazards model.

Statistical significance was defined as P < 0.05. Statistical Package for the Social Sciences for Windows 12.0 software (SPSS, Inc., Chicago, IL) was used for all analyses.

Expression of E2F family of transcription factors in ovarian cancer. Quantitative expression of transcript levels of all known E2F family members was estimated in 77 ovarian cancer samples and 8 control samples (Table 4). Expression of all but E2F6 was found to be superior in malignant tissue compared with controls. The most prominent levels were detected for E2F2 and E2F8 (3.2404- and 9.6272-fold that of controls, respectively), whereas E2F2 and E2F8 expression levels were close to the detection limit in the control samples.

E2F expression was shown to correlate with tumor grading and residual disease after primary debulking surgery. High expression levels of the proliferation-promoting transcription factors E2F1 and E2F2 were mainly associated with grade 3 tumors (P = 0.001 and 0.01, respectively). This association with tumor grading was also true for the transcription factor E2F8, which is regarded as a cell cycle–inhibiting member of the E2F family (P = 0.026). Whereas no significant changes in expression levels of E2F1, E2F2, and E2F8 were observed between groups with no residual disease and residual tumor nodules up to 2 cm in diameter after primary surgery, high E2F1, E2F2, and E2F8 expression levels were significantly elevated in patients with residual disease >2 cm in diameter (E2F1, P = 0.018; E2F2, P = 0.006; E2F8, P = 0.036). Moreover, mucinous carcinomas were positively associated with E2F2 transcript levels (P = 0.014) compared with cancers of the serous and other cell types.

Prognostic relevance of E2F transcription factors in ovarian cancer. We hypothesized that best survival is associated with low expression levels of the proliferation-promoting E2F transcription factors and probably also with high expression of several inhibitory E2F transcription factors. Indeed, patients with high expression levels of the proliferation-promoting E2F1 and E2F2 showed significantly shorter disease-free and overall survival. Whereas median survival of patients assigned to the group with low E2F1 or E2F2 expression was not reached for disease-free and overall survival, median time to progression was 24 months (P = 0.039) and median overall survival was 45 months (P = 0.047) for patients with high E2F1 expression. For high E2F2 expression, the corresponding survival data gave a median time to progression of 23 months (P = 0.009) and a median overall survival of 41 months (P = 0.006; Fig. 1A and B). Of special note is the fact that disease-free and overall survival of ovarian cancer patients were influenced in an opposite way by expression of the inhibitory transcription factor E2F4 (Fig. 1C and D). Furthermore, patients with high E2F7 transcript levels also showed significantly longer time to progression (P = 0.048) and overall survival (P = 0.042). In contrast, low E2F8 expression yielded longer overall survival (P = 0.039), whereas disease-free survival was not significantly affected (data not shown).

Assessment of multivariate Cox regression with a stepwise backward method showed expression of various E2F transcription factors to be independent predictors for disease-free and overall survival. Whereas high E2F1 and E2F2 expressions were independently associated with poor disease-free and overall survival, high E2F4 expression was seen to be an independent prognosticator for favorable overall survival and high expression of E2F7 predicted longer disease-free survival (Table 5A). From the other clinicopathologic variables included in the stepwise regression analysis, only age at diagnosis and residual disease after debulking surgery retained independent prognostic significance for disease-free survival as well as for overall survival.

Clinical relevance of the E2F family considering the cross-talk between cell cycle–promoting and cell cycle–inhibiting factors. To estimate the prognostic relevance of E2F cross-talk between proliferation-promoting members (E2F1 and E2F2) and proliferation-inhibiting factors (E2F4 and E2F7), ratios between the respective proponents were calculated and survival analyses were done. As expected, ratios calculated for high expression levels of proliferation-promoting transcription factors and for low expression of the inhibitory E2F members were associated with poor disease-free and overall survival in univariate survival analysis. However, the highest statistical significance was reached when E2F1 or E2F2 was related to E2F4. The corresponding survival curves for E2F2 to E2F4 ratio are shown in Fig. 2A and B. In the multivariate analysis, including the above-mentioned clinicopathologic variables, E2F2 to E2F4 and E2F1 to E2F7 ratios were shown to be independent prognostic factors for disease-free survival and E2F1 to E2F4 and E2F2 to E2F7 ratios for overall survival (Table 5B). It is noteworthy that the E2F2 to E2F4 ratio was not significantly associated with overall survival but yielded the highest hazard ratio (HR) when prognostic relevance for disease-free survival was assessed (HR, 6.494; P = 0.002).

E2F expression reflecting platinum resistance. With regard to the hypothesis that E2F transcription factors are involved in mechanisms mediating platinum resistance, we subdivided our collective into three groups (i.e., patients refractory to platinum, patients resistant to platinum, and patients with platinum-sensitive tumors). Patients refractory or resistant to platinum showed no statistically significant difference in expression of E2F (data not shown).

On the other hand, the inhibitory transcription factors E2F4 and especially E2F7 were expressed at significantly lower levels in tumors resistant or refractory to platinum compared with patients who remained free of disease at 5 years after diagnosis or those who relapsed beyond 6 months after completing first-line therapy (considered platinum sensitive; P = 0.012 and P = 0.009, respectively). Results are shown in Table 6.

Assessment of the relationship between E2F1 or E2F2 and E2F7 expression revealed a significantly lower E2F1 to E2F7 or E2F2 to E2F7 ratio in platinum-refractory or platinum-resistant tumors (P = 0.001 and 0.007, respectively), indicating that the mentioned ratios, especially that of E2F1 to E2F7, could be a useful tool for predicting platinum sensitivity of a tumor. Platinum sensitivity was predicted in tumors exhibiting E2F1 to E2F7 ratios below a cutoff value of 2.08 with a positive predictive value of 82% and a negative predictive value of 86%.

Mutations in the Rb-E2F cascade are found in a wide range of various tumor entities (8, 14). Whereas most of these alterations affect Rb or upstream regulators of the E2F transcription factors, there is growing evidence that deregulation of the E2F family itself is crucially involved in carcinogenesis. However, most of the studies done thus far focused on the deregulation of proliferation-promoting members of the E2F family, especially E2F1 and E2F3 (15–20). In consequence, prognostic relevance of deregulation of these cell cycle–promoting transcription factors has been described in various tumor entities. The authors (15–20) argue that overexpression of a proliferation-promoting E2F transcription factor could contribute to a significant growth advantage of tumor cells and might therefore contribute to poor survival.

On the other hand, it is conceivable that tumor growth could also be enhanced by down-regulation of inhibitory E2F transcription factors, such as E2F4 or E2F7, and that such an imbalance could possibly cause a switch from cellular quiescence to enhanced proliferation. To our knowledge, no data are available on the role of E2F in the tumor biology and clinical outcome in ovarian cancer.

We therefore investigated the expression of all currently known E2F family members in ovarian cancer to study the clinical relevance of proliferation-promoting as well as proliferation-inhibiting E2F transcription factors.

In fact, in ovarian cancer, the proliferation-promoting E2F1 and especially the E2F2 transcription factors were overexpressed when compared with healthy control tissue. Nonetheless, we cannot completely rule out that a higher stroma to epithelial cell ratio is responsible for the low E2F expression levels detected in control tissue. On the other hand, overexpression of proliferation-promoting E2F transcription factors in ovarian cancers is corroborated by our recent in vitro findings, showing that E2F1 and predominantly E2F2 were significantly up-regulated in ovarian cancer cell lines compared with normal human peritoneal mesothelial cells (21). Furthermore, E2F1 and E2F2 correlated significantly with grade 3 tumors and residual disease >2 cm in diameter after primary debulking surgery, suggesting a pivotal role of these proliferation-promoting E2F transcription factors in the biology of fast-growing ovarian carcinomas. However, in addition to its proliferation-promoting properties, the physiologic function of E2F1 also comprises induction of apoptosis by targeting key genes of the p53 and p73 pathways (6, 22, 23). In ovarian cancer, the proliferation-promoting activity of E2F1 probably predominates over its proapoptotic properties. Because E2F1, E2F2, E2F3, and E2F7 were shown to be E2F target genes themselves (5, 24–26), it could also be hypothesized that overexpression of E2F1 subsequently induces E2F2 up-regulation and thus causes an excess of proliferation-promoting stimuli in cancer cells.

Accordingly, the present investigation shows both transcription factors to be independent prognostic factors for disease-free and overall survival. E2F2 expression showed the highest HR (3.737) for recurrence, thus showing its pivotal role in fast-growing tumors that tend to relapse early. When the relationship between proliferation-promoting E2F1 or E2F2 and proliferation-inhibitory factors E2F4 or E2F7 was assessed, E2F1 to E2F7 and E2F2 to E2F4 ratios were shown to independently predict outcome for disease-free survival. High E2F2 to E2F4 ratios were clearly associated with poor disease-free survival, although E2F4 expression itself was not found to be of independent prognostic relevance for disease-free survival. This finding tempted us to speculate that cross-talk between primarily E2F2 and E2F4 is crucially involved in processes causing highly aggressive tumors, and therefore, calculation of E2F2 to E2F4 ratios is helpful in predicting disease-free survival in ovarian cancer.

In the context of the potential relevance of E2F2 to E2F4 cross-talk, the herein revealed significant higher expression of E2F2 in mucinous ovarian carcinomas needs to be particularly emphasized. As gene expression profile of epithelial mucinous ovarian cancers was found to be very similar to that of intestinal carcinomas (27) and hereditary nonpolyposis colorectal cancers are frequently associated with mutations of the E2F4 gene (28), it is tempting to speculate that alterations in the E2F4 gene could also play a significant role in mucinous cancers of the ovary, and through the lack of the inhibitory property of mutated E2F4 on E2F2 transcription, overexpression of E2F2 especially in this histologic subtype could be explained. This hypothesis is furthermore supported by our recent data obtained in IFN-γ-treated ovarian cancer cell lines showing a direct inhibitory E2F4-pocket protein effect on the E2F1 and especially E2F2 promoters (21). Moreover, E2F4 was reported to act as a critical regulator of the E2F1 gene in the Burkitt's lymphoma cell line Daudi treated with IFN-α (29).

It is noteworthy that E2F8, which was shown to act synergistically with E2F7 as a repressor of E2F target genes in primary mouse embryonic fibroblasts and human diploid fibroblasts (12, 13), was seen to be expressed at high levels in ovarian cancers, whereas expression in control samples was near the detection limit. However, the role of E2F8 is not fully understood and E2F8 could possibly act in a proliferation-promoting manner under certain conditions, such as cancer. Although E2F8 and E2F7 are thought to act synergistically by inhibiting cell cycle progression (7, 12, 13), our clinical observations clearly argue that, at least in ovarian cancer, E2F8 may have a different function than does E2F7. In univariate survival analyses, high expression levels of E2F8 are associated with poor overall survival, whereas high E2F7 expression improves clinical outcome. These findings tend to underscore the hypothesis that E2F8 undergoes autoregulative up-regulation due to E2F1-mediated proliferation enhancement rather than that of autonomous deregulation of this factor in ovarian cancer.

Contrary to our assumption, the inhibitory transcription factors E2F4, E2F5, and E2F7 were not generally down-regulated in ovarian cancer or associated with low-grade tumors but were shown to be overexpressed in the investigated samples. This could also argue for an autoregulatory overexpression of inhibitory transcription factors as a consequence of proliferation-promoting stimuli and might indicate the importance of cross-talk between differently acting E2F family members in ovarian cancer. On the other hand, our data generated on the clinical effect of ratios between cell cycle–promoting and cell cycle–inhibiting E2F family members clearly indicate that, in a subgroup of ovarian carcinomas exhibiting a high malignant phenotype (e.g., tumors with high E2F2 to E2F4 ratios), this autoregulative cross-talk seems to be seriously disconnected. Accordingly, high E2F4 expression was independently associated with improved overall survival and high E2F7 expression was shown to be an independent prognostic factor for favorable disease-free survival. The latter finding prompted us to speculate that predominantly low E2F7 expression could possibly be associated with early recurrence and platinum resistance.

In clinical practice, patients showing progression under treatment are considered to be platinum refractory and those relapsing within 6 months after first-line therapy are regarded as platinum resistant. In our collective, we were not able to distinguish E2F7 expression between patients refractory and those considered resistant to platinum-based therapy. However, patients refractory or resistant to platinum showed significantly lower expression levels of E2F7 compared with patients whose tumors were platinum sensitive. Furthermore, cross-talk between E2F1 or E2F2 and E2F7 turned out to more consistently predict platinum sensitivity in ovarian carcinomas, where the ratio of E2F1 to E2F7 emerged as the best discriminator. These results strongly suggest that especially E2F7 expression plays a key role in mechanisms leading to platinum resistance in ovarian cancer cells. As development of platinum resistance is the major obstacle to efficient treatment of ovarian cancers with a more favorable clinical outcome, we are currently investigating the effect of E2F7 together with E2F1 or E2F2 in ovarian cancer cell lines after stepwise induction of cisplatin resistance.

In conclusion, our data generated from a training set of 77 ovarian cancer patients show cross-talk between both proliferation-promoting and proliferation-inhibiting E2F transcription factors to be clinically relevant in predicting survival. Especially, the relationship between E2F2 and E2F4 turned out to be of major prognostic value. Moreover, down-regulation of the inhibitory transcription factor E2F7 seems to be involved in the development of platinum resistance and could serve as a putative predictor, especially when the E2F1 to E2F7 ratio is considered. Further studies in independent and larger sets of ovarian cancer patients are warranted to confirm the herein postulated clinical effect of the E2F family members, especially of E2F7, as a predictor of platinum resistance/refractoriness in this tumor entity.

We thank Julia Rössler and Martina Fleischer for technical assistance.