Beschreibung:
<jats:title>Abstract</jats:title>
<jats:p>Systemic therapy of ovarian cancer can include chemotherapy and targeted therapy. Prognostic biomarkers are capable to predict survival and predictive biomarkers are capable to predict therapy response. To date, multiple online tools were established to identify prognostic biomarkers, but no platform is yet available for predictive biomarkers. In this study, we describe the first release of an online available tool capable to validate gene expression based predictive biomarkers using transcriptomic data of a large set of ovarian cancer patients.</jats:p>
<jats:p>Published gene expression data of 35 independent datasets was integrated with treatment data into a unified database. The classification is based on either author-reported pathological complete response (n=1,022) or relapse-free survival status at six months (n=1,347) or relapse-free survival status at twelve months (n=1,282). Treatment data includes chemotherapy (platin, taxol, docetaxel, paclitaxel, gemcitabine, topotecan) and targeted therapy (avastin). The transcriptomic database includes 54,675 probe sets corresponding to 20,089 distinctive genes. Finally, we performed a sample collection at the National Institute of Oncology (NIO cohort), and used these patient samples to validate the top six genes in ovarian cancer patients. Gene expression and therapy response were compared using receiver operating characteristics and Mann-Whitney tests.</jats:p>
<jats:p>In the validation of the tool we focused on paclitaxel-resistance associated genes. We selected the top genes after running the analysis across all samples and validated these by PCR in the NIO cohort of patients (n=80). The best performing paclitaxel-resistance biomarker candidates were ARAF (AUC=0.743, p=4.1E-09), PPCS (AUC=0.733, p=4.2E-07), GNL2 (AUC=0.718, p=2.7E-08), TFE3 (AUC=0.718, p=4.8E-06), PDXK (AUC=0.717, p=1.1E-05) and TOP1 (AUC=0.716, p=4.6E-05).</jats:p>
<jats:p>The analysis pipeline enables to validate and rank predictive biomarker nominees. By analyzing the candidate genes in a large set of independent patients, we can select the most reliable candidate and abolish those which are most likely to fail in a clinical setting. The registration-free interface of the online analysis platform is accessible at www.rocplot.org/ovar.</jats:p>
<jats:p>Citation Format: Balazs Györffy, Janos Fekete, Agnes Osz, Imre Pete, Marcell Szasz. An online tool to validate predictive biomarkers of therapy response using transcriptomic data of 3,651 ovarian cancer patients [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2019; 2019 Mar 29-Apr 3; Atlanta, GA. Philadelphia (PA): AACR; Cancer Res 2019;79(13 Suppl):Abstract nr 2484.</jats:p>