Abstract
BACKGROUND: Chemotherapy selection traditionally relies on tumor tissue of origin. However, since genetic alterations drive tumor behavior, it remains unclear if mutations can better predict response. We hypothesized that genetic aberrations might influence chemotherapy outcomes more than tissue origin. METHODS: We retrospectively analyzed 15,474 Japanese patients with solid tumors who underwent comprehensive genomic profiling (CGP) and received cytotoxic chemotherapy. Genetic alterations and tumor origin were evaluated for objective response rate (ORR) and time to next treatment (TNT). Gene mutations were assessed across five chemotherapy classes: platinum-based, alkylating agents, antimetabolites, microtubule inhibitors, and topoisomerase inhibitors. RESULTS: Genomic alteration data alone did not surpass organ-based models in predicting response. For platinum-based agents, the gene-only model had an AUC of 0.575 versus 0.604 for the organ-only model. A combined gene-organ model yielded an AUC of 0.618 (P < 0.01). Certain gene-organ interactions were associated with improved outcomes. For example, APC-mutated colorectal cancer showed higher ORR and prolonged TNT (hazard ratio, 0.82; 95% CI, 0.73-0.92; P < 0.001) for platinum-based drugs. CONCLUSIONS: While genetic alterations alone did not outperform tumor origin as a predictor, incorporating both may improve exploratory predictions of chemotherapy response. These exploratory findings require prospective validation before any clinical application.