E2F5 status significantly improves malignancy diagnosis of epithelial ovarian cancer

Abstract

Background: Ovarian epithelial cancer (OEC) usually presents in the later stages of the disease. Factors, especially those associated with cell-cycle genes, affecting the genesis and tumour progression for ovarian cancer are largely unknown. We hypothesized that over-expressed transcription factors (TFs), as well as those that are driving the expression of the OEC over-expressed genes, could be the key for OEC genesis and potentially useful tissue and serum markers for malignancy associated with OEC.

Methods: Using a combination of computational (selection of candidate TF markers and malignancy prediction) and experimental approaches (tissue microarray and western blotting on patient samples) we identified and evaluated E2F5 transcription factor involved in cell proliferation, as a promising candidate regulatory target in early stage disease. Our hypothesis was supported by our tissue array experiments that showed E2F5 expression only in OEC samples but not in normal and benign tissues, and by significantly positively biased expression in serum samples done using western blotting studies.

Results: Analysis of clinical cases shows that of the E2F5 status is characteristic for a different population group than one covered by CA125, a conventional OEC biomarker. E2F5 used in different combinations with CA125 for distinguishing malignant cyst from benign cyst shows that the presence of CA125 or E2F5 increases sensitivity of OEC detection to 97.9% (an increase from 87.5% if only CA125 is used) and, more importantly, the presence of both CA125 and E2F5 increases specificity of OEC to 72.5% (an increase from 55% if only CA125 is used). This significantly improved accuracy suggests possibility of an improved diagnostics of OEC. Furthermore, detection of malignancy status in 86 cases (38 benign, 48 early and late OEC) shows that the use of E2F5 status in combination with other clinical characteristics allows for an improved detection of malignant cases with sensitivity, specificity, F-measure and accuracy of 97.92%, 97.37%, 97.92% and 97.67%, respectively.

Conclusions: Overall, our findings, in addition to opening a realistic possibility for improved OEC diagnosis, provide an indirect evidence that a cell-cycle regulatory protein E2F5 might play a significant role in OEC pathogenesis.

Figure 1

Genes putatively controlled by tentative promoter model of keratin group. Summary of the main finding in analysis of 19 genes over-expressed in ovarian cancer.

Figure 2

Differential expression pattern of E2F5 in ovarian tumour specimens examined by immunohistochemistry. A, benign serous cystadenoma, B, epithelial inclusion cyst, C, cystic serous carcinoma, and D, endometrioid adenocarcinoma samples. The low power photomicrograph in each section shows the full tissue microarray "punch" (x100), while the high power photomicrograph shows detail of the antibody expression (x600). There is no E2F5 expression in normal and benign tissue (A, B), but significant expression in ovarian epithelial cancer (C, D).

Figure 3

Validation of potential target (AREB6, PAX8, ELF3) using western blots (A) and expression of E2F5 in serum from normal, benign and, malignant patient samples (B). (A) Target validation using western blots of AREB6, PAX8, ELF3, in serum of healthy volunteers (N) (n = 2), patients with benign ovarian cysts (B1, B2) (n = 2), and patients with late- (L1, L2) (n = 2) and early-stage (E1-E3) (n = 3) serous adenocarcinoma. AREB6 was overexpressed in late stage disease, but was not discriminatory for early cancer. PAX8 and ELF3 were equally expressed in all serum samples. (B) Expression of E2F5 in serum from normal (n = 56), benign (n = 40) and, malignant (n = 48) serum obtained from patients with cancer of the ovary.