Integrated next-generation sequencing and patient-derived xenografts to personalized cancer treatment.

Abstract

3068 Background: The knowledge of actionable somatic genomic alterations present in each tumor is making possible the era of personalized cancer treatment. Methods: Using massively parallel sequencing we performed whole exome sequencing analysis of tumor and matched normal blood samples of 8 patients (2 pancreatic adenocarcinoma, 1 neuroendocrine tumor, 1 glioblastoma, 1 uveal melanoma, 1 colon cancer) to identify putatively actionable tumor specific genomic alterations. We used 2 in silico methods (Polyphen and SIFT) to estimate the functional significance of a given confirmed mutation. Primary xenografts (PDX), generated by direct engraftment of tumor samples from the patients into immunocompromised mice, were used as an in vivo platform that provided the opportunity to test proposed personalized medicine strategies. Results: At this time exome sequencing analyses have been performed for 5 patients (1 patient died prematurely, 1 tumor sample was insufficient, 1 result is pending). More than 30 million bases of target DNA were analyzed in the tumor and normal samples in every case, with at least 70 distinct reads at each base. Tumor specific mutations (Muts) and copy number variations (CNVs) were identified: 5 Muts in the neuroendocrine tumor; 62Muts/6CNVs and 38Muts/10CNVs in the pancreatic tumors; 63Muts/23CNVs in the glioblastoma; 5 Muts in the melanoma. All samples profiled contained actionable alterations with the most relevant mutations affecting NF1, PTPN11, EPHA3, CDKN2A, FAS (glioblastoma); PI3KCA, ARID1A, ARID1B, DDR2, SMAD4, TP53, KRAS, PTCHD3 (pancreatic); CREB3L3, ITPR2 (neuroendocrine); GNA11, TAOK3 (melanoma). PDX from the pancreatic cancer patient was treated with a PI3K inhibitor and dasatinib, reported to be effective in discodin domain receptor 2 (DDR2) mutant cancer with no effect. Accordingly treatment of that patient with dasatinib was not effective and the level of this mutation in the tumor was observed to be low. Conclusions: Detection of actionable tumor-specific genomic alterations in the clinical setting is feasible. In silico methods and primary xenografts can help in the challenge of linking confirmed mutations to protein function and ultimately to clinical utility.