Abstract A018: Patient-derived xenograft (PDX) models of NSCLC reflect clinical drug responses and predict effective treatments for patients

Background: Non-small cell lung cancer (NSCLC) is a leading cause of cancer-related mortality and prognosis remains poor despite the availability of numerous therapies. Integration of drug screening and sequencing in PDX models may allow for improved understanding of mechanisms of resistance (de novo and acquired), identification of biomarkers, and optimization of therapeutic strategies for NSCLC patients. In this study, we evaluated the response of NSCLC PDX models to multiple therapies and correlated responses to known clinical outcomes and molecular characteristics. Methods: PDX models were developed from 86 patients with NSCLC and evaluated by next-generation sequencing for genomic alterations (mutations, amplifications/deletions, fusions, and gene expression changes). Models were screened against different therapies including first-line platinum and nonplatinum doublets and triplets, second-line single agent docetaxel and pemetrexed (second-line therapies), and EGFR-targeted inhibitors. Tumor regression (TR) values and RECIST criteria were determined and correlated with known literature-based response rates (RR) as well as individual patient outcomes. Results: Eighty-eight PDX models from 86 patients were interrogated. To date, 63 (72%) models have been sequenced and 39 screened against standard-of-care therapies. There was robust concordance in mutational and allelic frequency profiles between patient tumors and corresponding PDX models. Based on PDX tumor growth, regression (CR/PR RECIST) was observed for at least one regimen in 50% (12/24) of models screened against first-line therapies and 25% (2/8) screened against second-line docetaxel or pemetrexed. No tumor regression was observed in any NSCLC model treated with the EGFR inhibitor, erlotinib (0%, 0/15), despite EGFR amplification and the absence of KRAS mutations. Nine PDX responses could be correlated to clinical outcomes, with 89% (8/9) accurately reflecting patient responses to the same treatment. Conclusion: Our study demonstrated the strong alignment between PDX model response to standard-of-care therapies and patient clinical outcomes, which highlights the potential application of PDX models for translational modeling and utilizing cohorts of PDX models for clinical trial simulation. The responses of these models to different lines of therapy reflected corresponding patient outcomes both at an individual and population level. Comprehensive sequencing (WES and RNA) and standard-of-care drug testing of these PDX models is planned and could allow a deeper understanding of such mechanisms. In this context, application of PDX models to drug development and stratification of clinical trial patients for treatment will continue to evolve. Citation Format: Daniel Ciznadija, Igor Astaturov, Haiying Cheng, Nir Peled, Jennifer Jaskowiak, Angela Davies, David Sidransky. Patient-derived xenograft (PDX) models of NSCLC reflect clinical drug responses and predict effective treatments for patients [abstract]. In: Proceedings of the AACR-NCI-EORTC International Conference: Molecular Targets and Cancer Therapeutics; 2017 Oct 26-30; Philadelphia, PA. Philadelphia (PA): AACR; Mol Cancer Ther 2018;17(1 Suppl):Abstract nr A018.