Purpose: Ovarian cancer is the seventh most frequent cause of cancer death in women with 192.000 new cases a year worldwide. Despite treatment improvements and the high initial response to chemotherapy (80%), it is still impossible to predict which patient will progress during or recur after chemotherapy. This prediction is essential since patients that are (intrinsic) resistant might benefit from a different type of treatment. The aim of this study is to find a gene signature that can predict the type of response to platinum-based chemotherapy in ovarian cancer. Experimental design: The study was performed on fresh-frozen primary ovarian adenocarcinoma specimens (n=96). Fourteen of the 96 patients did not respond whereas 82 responded to platinum-based chemotherapy. A discovery set of 24 specimens of which five did not respond and 19 responded to platinum-based chemotherapy, was profiled with 18K cDNA microarrays. The confirmation was done with quantitative RT-PCR. Results:Sixty-nine genes were found to be significantly differentially expressed between the nonresponders and the responders in the discovery set. Of these, 60 genes were known genes and were involved in regulation of transcription (22%), apoptosis (18%), cell adhesion (17%), cell cycle regulation (7%) and immune or inflammatory response (6%). After reduction and confirmation, six genes discriminated the nonresponders from the responders in 92 tumors when measured with qRT-PCR. Step-down logistic regression demonstrated that a three-gene set was sufficient and was the strongest response predictor. The three-gene set correctly classified 80% of all tumors (n=92) with a sensitivity of 93% and a specificity of 78% (p<0.0001). Multivariate analysis including patient and tumor characteristics (age, residual tumor after surgery, histology and grade) demonstrated that the three-gene set is independent for the prediction of response. Within the patients with advanced FIGO stage (n=75), a multivariate analysis of the three gene-signature revealed that the three gene-signature is also a strong predictor of response within these patients. Conclusions: This study revealed a 69-gene signature that classifies the tumors according to their resistance. After gene reduction a predictive three-gene signature was generated that outperforms patient and tumor characteristics. Patients that show (intrinsic) resistance to platinum-based chemotherapy could benefit from other therapies. To further confirm whether this three gene-signature can identify these patients, a larger independent prospective multicenter study should be done.

[Proc Amer Assoc Cancer Res, Volume 46, 2005]