Abstract IA017: Single mouse testing (SMT): An alternative approach to encompass clinical heterogeneity of pediatric cancers

Abstract Patient derived xenografts (PDXs) have become the predominant models for testing novel therapeutics, however the failure rate for new cancer therapeutics remains high around 95%. Several publications have analyzed at what stage failure occurs, some due to unrecognized toxicity in preclinical models, but predominantly failure occurs due to lack of efficacy. For many cancer drugs failure to elicit adequate therapeutic activity may be a consequence of how we do preclinical studies. Through the Pediatric Preclinical Testing Program (PPTP) its successor PPT/Consortium we conducted over 2100 tumor/drug studies that included 68 drugs and up to 83 PDX models of childhood cancer. These studies used conventional testing approaches with 10 mice per treatment group for solid tumors and 8 mice/group for leukemia models. Mi imal statistical differences between control and treatment groups was defined as a P value of 0.05. However, such statistical difference does not imply biologically meaningful activity. Limitations of conventional testing include 1) a failure to recapitulate the genomic/epigenomic heterogeneity of a human cancer type, 2) use of activity criterial in preclinical studies that are less stringent than in clinical trials and 3) differences in species tolerance to drugs has not been considered critical for accurate translation. Here we consider the first two limitations. We analyzed 2106 tumor model drug studies. These studies we used 10 mice/group (olid tumors) or 8 mice/group (leukemias), that allowed inclusion of 6-8 models representing a pediatiric cancer type (neuroblastoma, osteosarcom ALL etc.) and asked a simple question: would 1 mouse per group, chosen at random, predict the group median response? The result showed that in 78% of drug studies the single mouse accurately predicted the group response. If some latitude was allowed (+/- 1 response criteria (PR vs CR) etc.) the accuracy was 95%. Further, if we used the same approach to objective responses the single mouse accurately predicted group response for 67 of 68 drugs tested over a range of antitumor activities (0-100%). We have validated the SMT results in prospective testing with different childhood cancer models (solid tumors and leukemias) using PDX models largely not included in the initial analysis, with similar concordance beteeen SMT and conventional testing (CT). Based on these data we consider that SMT allows for inclusion of a greater number of PDX models that more accurately simulate the genetic/epigenetic heterogeneity of pediatric cancers and potentially allows identification of 1) truly acive drugs and 2) biomarkers of response. The SMT approach is now being used by the NCI supported Preclinical Pediatric In Vivo Testing (PIVOT) program. Citation Format: Peter J Houghton, Samson Ghilu, Andrew Robles, Raushan T Kurmasheva, Richard B Lock. Single mouse testing (SMT): An alternative approach to encompass clinical heterogeneity of pediatric cancers [abstract]. In: Proceedings of the AACR Special Conference in Cancer Research: Functional and Genomic Precision Medicine in Cancer: Different Perspectives, Common Goals; 2025 Mar 11-13; Boston, MA. Philadelphia (PA): AACR; Cancer Res 2025;85(5 Suppl):Abstract nr IA017.