Time-resolved cardiovascular risk after immune checkpoint inhibitors: A large real-world cancer cohort study.

12023 Background: With expanding use of immune checkpoint inhibitors (ICIs), immune-related toxicities increasingly influence outcomes beyond tumor control. Cardiovascular events are among the most consequential yet potentially preventable complications. Given the low incidence of major adverse cardiovascular events (MACE) and their relatively narrow post-initiation window, reliable time-resolved risk estimation requires very large cohorts, which remain scarce, limiting evidence to guide surveillance and prevention. Methods: We conducted a pan-cancer retrospective cohort study (N = 269,257; mean follow-up 4.5 years) using a large real-world cancer registry from Shandong Provincial Center for Disease Control and Prevention. Patients receiving ICIs after their broad clinical implementation were compared with patients treated in the pre-ICI era without immunotherapy, a design chosen to mitigate indication-related differences observed in the contemporary setting. Propensity scores derived from elastic net regularization were used to construct overlap weights for balancing covariates across cohorts. We evaluated MACE, AMI (acute myocardial infarction), and stroke across four prespecified intervals (0–90, 91–180, 181–270, and 271–365 days) using interval-stratified weighted Cox models. Absolute risk differences were further characterized using weighted incidence density differences and restricted mean survival time. Results: All covariates achieved excellent balance (maximum |SMD| = 0.004) after weighting. ICI exposure was associated with higher 1-year MACE risk (HR, 1.17; 95% CI, 1.12–1.23; P < 0.001), corresponding to a weighted incidence density difference of 10.90 per 1,000 person-years (95% CI 8.73–13.06). The largest absolute risk increase was observed during 0–90 days after ICI initiation(weighted incidence density difference, 17.83 per 1000 person-years; 95% CI, 12.43 to 23.24).The risk of AMI peaked early, with a statistically significant association observed only during 0–90 days (HR, 1.86; 95% CI, 1.54–2.24; P < 0.001). In contrast, stroke risk peaked during days 91–180 (HR, 1.27; 95% CI, 1.11–1.44; P < 0.001) and remained significantly elevated through 181–270 days (HR, 1.25; 95% CI, 1.08–1.44; P = 0.003). Stroke accounted for 51% (95% CI: 19% to 83%), 63% (95% CI: 37% to 89%), 66% (95% CI: 43% to 90%), and 67% (95% CI: 45% to 88%) of the total weighted incidence density difference for MACE during days 0–90, 0–180, 0–270, and 0–365, respectively. Conclusions: ICI exposure was associated with temporally heterogeneous cardiovascular risk, with AMI concentrated in the first 90 days and stroke risk remaining elevated thereafter. These results suggest a time-tailored approach to cardio-oncology care, characterized by intensified early monitoring for AMI and sustained attention to stroke prevention in subsequent months.