Abstract 2598: Evaluation of antitumor efficacy of Trabectedin in patient derived tumor xenografts in vitro and in vivo, and determination of a predictive gene signature

Abstract

Abstract Trabectedin is a marine derived antitumoral agent, originally isolated from Ecteinascidia turbinata. It acts by binding to the minor groove of DNA interfering with cell division, transcription, and the DNA repair machinery. Trabectedin was approved by the EMEA as second line therapy for the treatment of advanced soft tissue sarcoma and for ovarian cancer in combination with Doxil. Further Phase II trials with Trabectedin in breast and prostate cancer are underway. We characterized Trabectedin for antitumor efficacy and selectivity in patient-derived tumor xenografts to identify target tumor types for further clinical studies. The compound was tested in 67 tumor xenografts of 15 histo types using an ex vivo clonogenic assay. Pronounced concentration dependent antitumor activity (mean IC70 = 1.3 nM) and selectivity was observed, with sensitive tumor models being on average about 7-fold more sensitive than the average of all tumors tested. Trabectedin was also given to tumor-bearing nude mice at 0.2 and 0.15 mg/kg/d iv once weekly for 3 weeks and showed substantial inhibition of tumor growth at a dose level of 0.2 mg/kg/d in tumors of lung, colon, and breast. The activity data of Trabectedin against 67 tumors in the clonogenic assay were used for further bioinformatic analysis. Subsets of tumors and their corresponding data were randomly split into a training set (n=44) and an independent validation set (n=23). By matching in vitro antitumor efficacy data (IC70) of the tumors with the corresponding gene expression profiles (determined by Affymetrix HG-U133 Plus 2.0 gene chip array), a signature of 19 gene transcripts being specific for the responsiveness towards Trabectedin was determined. The classification border for activity of Trabectedin was IC70 = 0.5 nM. The signature was validated by leave-one-out-cross-validation (LOOCV) on the training set, and prediction of tumors of the independent validation set. In the LOOCV sensitivity or resistance of tumors was predicted correctly in 13/18 (72%) and 26/26 tumors (100%), in the validation set in 4/7 (57%) and 13/16 tumors (81%), respectively. In the validation set, the predicted responders showed a 4.5-fold lower median IC70 compared to the predicted non-responders (p= 0.02). Moreover, the signature was used to predict responsiveness of 173 tumors of the Oncotest xenograft collection with unknown sensitivity to Trabectedin. In this set of tumors, the signature identified sarcoma, leukemia/lymphoma, as well as ovarian, head and neck, small cell lung, mammary and bladder cancer as Trabectedin sensitive tumor types. Experimental testing of the predicted tumors so far confirmed the predictions in 11/13 cases (85%). This study shows the feasibility of combining experimental testing and virtual prediction to identify additional tumor types as candidates for further preclinical and clinical investigations. 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 2598.