Benchmarking human tumor slices from gastrointestinal malignancies for precision cancer therapy

Patient-derived tumor slices (PDS) have been used to evaluate drug sensitivity in multiple human tumor types. Despite this, standardized parameters to establish PDS as a precision medicine tool are lacking. Here we benchmarked proliferation rates and standard deviations from 40 gastrointestinal cancer specimens, defining normal biological variation. Here the mean standard deviation rates in pancreatic duct adenocarcinoma (PDAC), appendiceal cancer, and colorectal adenocarcinomas were 10%, 6%, and 15%, respectively. Using these data, we established threshold criteria set at a z-score of 3 standard deviations in order to distinguish meaningful drug sensitivity to the pan-RAS(On) inhibitor RMC-6236, from that of normal biological variation. This framework enabled classification of PDAC and appendiceal PDS cultures as sensitive or resistant. To benchmark the predictive value of slice assays, we compared RAS-inhibitor activity across matched patient-derived organoids (PDOs) and xenografts (PDXs), and further determined that slice-based assay sensitivity predicted sensitivity and survival outcomes in matching orthogonal tumor models.