ImportanceTo date, the clinical benefit and utility of implementing a DPYD/UGT1A1 pharmacogenetic-informed therapy with fluoropyrimidines and/or irinotecan have not been prospectively investigated.ObjectiveTo examine clinically relevant toxic effects, hospitalizations, and related costs while preserving treatment intensity and efficacy outcomes in patients with gastrointestinal cancer.Design, Setting, and ParticipantsThis nonprespecified secondary analysis stems from Pre-Emptive Pharmacogenomic Testing for Preventing Adverse Drug Reactions (PREPARE), a multicenter, controlled, open, block-randomized, crossover implementation trial conducted from March 7, 2017, to June 30, 2020, and includes data from Italy according to a sequential study design. The study population included 563 patients (intervention, 252; control [standard of care], 311) with gastrointestinal cancer (age â„18 years) who were eligible for fluoropyrimidine and/or irinotecan treatment. Data analysis for the present study was performed from May 27 to October 10, 2024.InterventionsParticipants with actionable variants (DPYD*2A, DPYD*13, .DPYD c.2846A>T, and DPYD c.1236G>A for fluoropyrimidines, and UGT1A1*28, UGT1A1*6, and UGT1A1*27 for irinotecan) received drug or dose adjustments based on Dutch Pharmacogenetics Working Group recommendations.Main Outcomes and MeasuresThe primary outcome was clinically relevant toxic effects (National Cancer Institute Common Terminology Criteria for Adverse Events grade â„4 hematologic, grade â„3 nonhematologic, or causing hospitalization, fluoropyrimidines and/or irinotecan causally related). Secondary outcomes included hospitalization rates, toxic effect management costs, intensity of treatment, quality-adjusted life-years, and 3-year overall survival.ResultsOverall, 1232 patients were enrolled in Italy, with 563 included in this analysis (317 [56.3%] men; median age, 68.0 [IQR, 60.0-75.0] years). In the intervention arm, carriers of any actionable genotype exhibited a 90% lower risk of clinically relevant toxic effects compared with the control arm (odds ratio, 0.1; 95% CI, 0.0-0.8; Pâ=â.04). They also presented higher toxic effect management costs per patient ($4159; 95% CI, $1510-$6810) compared with patients in the intervention arm ($26; 95% CI, 0-$312) (Pâ=â.004) and a higher rate of hospitalization (34.8% vs 11.8%; Pâ=â.12). The differences were not significant among all patients. Three-year overall survival did not differ significantly between arms, while quality-adjusted life-years significantly improved in the intervention arm. The pharmacogenetics-informed approach did not manifest a detrimental effect on treatment intensity in actionable genotype carriers.Conclusions and RelevanceIn this secondary analysis of PREPARE, pretreatment application of DPYD- and UGT1A1-guided treatment appeared to increase safety and reduce hospitalizations and related costs in patients with gastrointestinal cancer. Clinical benefit did not appear to be affected.Trial RegistrationClinicalTrials.gov Identifier: NCT03093818