Although a number of treatments are available for pancreatic cancer, these are often administered in prescribed sequences, which may not represent the optimal therapeutic strategy. Technologies enabling multiple regimens to be evaluated simultaneously to identify beneficial therapies are needed. Drug screening in patient-derived xenografts (PDXs) is a potential solution. We examined whether pancreatic PDXs capture patient responses to different drugs and report performance metrics highlighting clinical utility.
Pancreatic tumors from 94 patients were engrafted into immunodeficient mice to generate PDX models. Of these, 19 models were sequenced to identify key genomic alterations with therapeutic implications. Sensitivity to different therapeutics was evaluated and effects on tumor growth aligned to clinical RECIST criteria. A total of 16 PDX screening outcomes were correlated with individual clinical responses and statistical parameters such as sensitivity, specificity, and predictive values calculated.
Of the 94 implanted pancreatic tumors, 82 have completed the implantation process, with 72 (88%) successfully engrafting. Sequencing revealed alterations in 451 common genes, including those informing treatment choices such as EGFR, KRAS, and BRCA 1/2. PDX models from 39 patients were screened in 144 drug tests employing 56 FDA-approved therapies and 9 experimental agents. In 14/16 (88%) cases with available data, a correlation between clinical and PDX outcomes was noted and from this cohort we calculated positive and negative predictive values of 82% and 100% respectively.
Using a small cohort, we showed drug responses in pancreatic PDX model correlate with clinical outcomes to the same therapy. Application of such models to guide treatment decisions for pancreatic cancer patients may help lead to better outcomes. Moreover, given the clinical relevance of these models, they could also be deployed as real-time patient surrogates during drug development and clinical trials, permitting real-time analysis of treatment responses and identification of biomarkers that predict different therapeutic outcomes.
Clinical trial identification
Legal entity responsible for the study
A. Davies, D. Ciznadija: Employee of Champions Oncology Stockholder in Champions Oncology. M. Hidalgo, J. Stebbing: Advisory board for Champions Oncology. A. Katz: Employee of Champions Oncology Stockholder of Champions Oncology. D. Sidransky: Chairman of Board for Champions Oncology Stockholder in Champions Oncology.