Abstract 379: Predictive profile of pathological response in advanced ovarian carcinoma derived from angiogenesis-related genes

Background: Studies have consistently shown that the volume of residual disease remaining after cytoreductive surgery inversely correlates with survival. Women with optimally resected tumor have, on average, a 20-month improvement in median survival compared to those with suboptimal resection. It is difficult to predict which patients can be optimally debulked. CA 125 level and scoring imaging techniques have been used, however they were not enough accurate to obviate the initial surgical exploration. Several genes involved in angiogenesis proved to have prognosis capacity in advanced ovarian carcinoma. The aim of this study is to find model derived from angiogenesis genes to predict pathological response to treatment. Material and Methods: 34 patients with III/IV FIGO stage ovarian cancer that underwent surgical cytoreduction and received a carboplatin plus paclitaxel chemotherapy regime were included. A second look laparotomy was performed in all the patients. Optimal debulking was defined as ≤1 cm (diameter) residual disease. RNAs were collected from formalin-fixed paraffin-embedded advanced ovarian carcinoma samples. Expression levels of 82 angiogenesis related genes were measured using quantitative real time polymerase chain reaction (qRT-PCR). A logistic regression method was used to build multiple models based on the significant genes in the univariate analysis. The accuracy of the models was evaluated using Receiver Operating Characteristic (ROC) curves. The Akaike Information Criterion based selection was used to find the most accurate one. And Leave-one-out Cross Validation (LOOCV) method was applied to avoid overoptimistic predictions. Results: The median age at diagnosis was 50 years (range, 35 to 79 years). All patients had advanced disease (FIGO stages III/IV). Most of them had FIGO stage III (27, 79.4%), grade 3 tumors (19, 55.9%), and serous histology (23, 67.6%). 20 patients (50.8%) achieved complete response after standard treatment. It was found an independent model able to predict any degree of response to therapy comprising 5 genes with an Area Under the Curve (AUC) of 0.950 (p Conclusions: It is feasible to create a predictive profile for pathological response in AOC derived from genes involved in ovarian angiogenesis. Our profile could be used as predictive tool to enable clinicians to identify those high risk patients who will potentially benefit from alternative drug combinations. Nevertheless, the predictive value of our qRT-PCR based angiogenesis-related gene expression profile should be further evaluated in prospective studies of patients with AOC. Citation Format: {Authors}. {Abstract title} [abstract]. In: Proceedings of the 101st Annual Meeting of the American Association for Cancer Research; 2010 Apr 17-21; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2010;70(8 Suppl):Abstract nr 379.