Comparing Patient-Derived Xenograft and Computational Response Prediction for Targeted Therapy in Patients of Early-Stage Large Cell Lung Cancer.

PURPOSE Targeted therapy (TT) provides highly effective cancer treatment for appropriately selected individuals. A major challenge of TT is to select patients who would benefit most. EXPERIMENTAL DESIGN The study uses cancer material from 25 patients primarily diagnosed with non-small cell lung cancer (NSCLC). Patient-derived xenografts (PDXs) are treated with cetuximab and erlotinib. Treatment response is measured by tumor shrinkage comparing tumor volume at day 25 (V25) with tumor volume at baseline (V0). Shrinkage below 40% is considered as treatment response: V25/V0 < 0.4 (<40%). Furthermore, RNA-seq data from each tumor sample are used to predict tumor response to either treatment using an in silico molecular signaling map (MSM) approach. RESULTS PDX response was 40% (10/25; 95% CI [21.13%, 61.34%]) under cetuximab and 20% (5/25; 95% CI [6.83%, 40.70%]) under erlotinib. MSM predicted response was 48% (12/25; 95% CI [27.8%, 68.7%]) under cetuximab and 40% (10/25; 95% CI [21.13%, 61.34%]) under erlotinib. Agreement between PDX and MSM response prediction is substantial under cetuximab and erlotinib: 84% (21/25, P = 0.001) and 80% (20/25, P = 0.003). A total of 5 from the 25 patients have been treated with cetuximab showing a clinical response identical to both predictions. CONCLUSIONS For NSCLC patients, this proof-of-concept study shows a considerable agreement in response prediction from MSM and PDX approaches, but MSM saves time and laboratory resources. Our result indicates the potential of MSM-based approach for clinical decision making when selecting cancer TTs. Clin Cancer Res; 22(9); 2167-76. ©2015 AACR.