Abstract 5194: A Mouse-Human co-clinical trial with patient-derived xenograft (PDX) models demonstrates a predictive signature for dovitinib (TKI258), an FGFR and VEGFR inhibitor,in lung squamous cell carcinomalung cancer (LSCC)

Abstract

Abstract Background: We conducted a co-clinical trial with PDX models in conjunction with a phase II clinical trial with dovitinib in patients with fibroblast growth factor receptor 1 (FGFR1) amplified LSCC in order to identify a predictive biomarker for dovitinib, a FGFR inhibitor. Methods: PDX models were established using tumor samples obtained from LSCC patients enrolled in the trial. We conducted an in vivo efficacy test with dovitinib in PDX models and comprehensive molecular profiling by whole exome sequencing (WES), array comparative genomic hybridization (aCGH), and microarray-based gene expression. Results: The histological and genomic characteristics of xenograft tumors (F2) resembled case-matched original tumors (F0). Dovitinib treatment resulted in tumor shrinkage in PDXL01, but not in PDXL02-04, which corresponded to responses in LSCC patients enrolled in the clinical trial. The patient, from whom PDXL01 was derived, showed a partial response to dovitinib with progression-free survival of 6 months, whereas the patient, from whom PDXL04 was derived had rapid disease progression within 2 months of treatment. Mutation profiles of F0 and F2 were largely concordant with each other in at least 70% of somatic mutations. The genome-wide copy number alterations were also largely concordant between F0 and F2, suggesting that PDX tumors can successfully represent the genomic features of the tumors from LSCC patients. Notably, mutation in FGFR 1-3 genes was not observed. FGFR1 amplification largely due to arm-level 8p gain was consistently observed, regardless of sensitivity to dovitinib. This finding suggests that FGFR1 amplification may not be a predictor for dovitinib sensitivity. Finally, we compared gene expression between a responder and non-responders. In a Heatmap analysis of the top 50 significantly up- and down-regulated genes, FGFR gene signature including FGF3 and FGF19 was significantly up-regulated in the responder. Gene set enrichment analysis (GSEA) identified that gene sets, such as FGFR ligand binding and activation and SHC-mediated cascade pathway, were significantly enriched in the responder, highlighting FGFR pathway activation as a key molecular determinant for sensitivity to dovitinib. To identify predictive gene signatures from independent datasets, we further performed GSEA in dovitinib-sensitive lung cancer cell lines compared to resistant ones as available in the CCLE (Cancer Cell Line Encyclopedia) database. Among the four functional modules (FGFR signaling, Immune, Cell cycle, and RNA) that we identified, the FGFR signaling module performed well in the prediction of dovitinib sensitivity of PDXL01-04 Conclusions: FGFR gene expression signatures are predictors for response to dovitinib in LSCC. Citation Format: Hye Ryun Kim, Han Na Kang, Sung Moo Kim, Hwan Kim, Kyoung-Ho Pyo, Myung-Ju Ahn, Tae Min Kim, Byoung Chul Cho. A Mouse-Human co-clinical trial with patient-derived xenograft (PDX) models demonstrates a predictive signature for dovitinib (TKI258), an FGFR and VEGFR inhibitor,in lung squamous cell carcinomalung cancer (LSCC). [abstract]. In: Proceedings of the 107th Annual Meeting of the American Association for Cancer Research; 2016 Apr 16-20; New Orleans, LA. Philadelphia (PA): AACR; Cancer Res 2016;76(14 Suppl):Abstract nr 5194.