ABSTRACT: Trifluridine/tipiracil increases overall survival (OS) in patients with refractory, metastatic colorectal cancer (mCRC). A post hoc exploratory analysis of the RECOURSE randomized clinical trial (RCT) established two categories, a good prognosis corresponding to subjects having a low tumor burden and indolent disease. Other models in refractory mCRC are the FAS-CORRECT and Colon Life nomogram. The main objective was to externally validate the prognostic factors of the RECOURSE and FAS-CORRECT trials, and the Colon Life nomogram in a multicenter, real-world series of mCRC treated in 3rd and successive lines with trifluridine/tipiracil. The secondary aim was to develop an OS predictive model, TAS-RECOSMO. Between 2016 and 2019, 244 patients were recruited. Median OS was 8.15 vs 8.12 months for the poor (85% of the subjects) and good (15%) prognosis groups from the RESOURCE trial, respectively, log-rank p = 0.9. The most common grade 3-4 toxicities were neutropenia (17%), asthenia (6%), and anemia (5%). The AFT lognormal model TAS-RECOSMO included six variables: ECOG-PS, KRAS/NRAS/BRAF mutation status, time between diagnosis of metastasis and beginning of trifluridine/tipiracil, NLR, CEA, and alkaline phosphatase. The model's bootstrapped bias-corrected c-index was 0.682 (95% CI, 0.636-0.722). The factors from the Colon Life model, FAS-CORRECT, and RECOURSE displayed a c-index of 0.690, 0.630, and 0.507, respectively. TAS-RECOSMO, FAS-CORRECT, and the Colon Life nomogram appear to predict OS in patients with refractory mCCR who begin trifluridine/tipiracil treatment in the real world. The prognostic groups of the RECOURCE RCT were unable to capture the situation of real-world subjects treated with trifluridine/tipiracil in this series.
