Incremental value of cardiac biomarkers and cancer-related factors beyond SCORE2 for prediction of cardiovascular events

Background Cancer therapy-related cardiovascular (CV) disease is an increasing cause of morbidity and mortality among cancer survivors. Current guidelines recommend CV risk prediction tools such as SCORE2; however, these do not account for cancer-specific factors or cardiac biomarkers. Whether incorporating cardio-oncology predictors improves risk stratification remains uncertain. Purpose To evaluate whether cardiac biomarkers and cancer-related variables provide incremental predictive value beyond SCORE2 in a population-based cohort, and to explore performance in cancer survivors. Methods Participants without prior cardiovascular disease (CVD) were identified from a population-based study (2017–2019) and linked to national cancer and cardiovascular registries with follow-up through 2023. The primary endpoint was CVD death, myocardial infarction, or stroke, with non-CVD death treated as a competing event. The original SCORE2 linear predictor was calculated for all participants. Associations between cancer-related variables and incident CVD were assessed using Fine and Gray competing risk models including the SCORE2 linear predictor as an offset term. Subdistribution hazard ratios (sHRs) were estimated. Model updating was performed by incorporating additional predictors based on estimated sHRs. Discrimination was assessed by C-statistics, and incremental changes were evaluated after adding individual predictors. Results Among 27,450 participants, 607 CVD events and 304 non-CVD deaths occurred over 5.7 years (IQR 5.3–6.0). In total, 1,503 were cancer survivors. After accounting for SCORE2, prior cancer was associated with lower subdistribution hazard for CVD (sHR 0.66, 95% CI 0.46–0.95), consistent with higher competing non-CVD mortality. Cancer type, time since diagnosis, and treatment exposure (previous chemotherapy, anthracyclines, radiation) were not independently associated with CVD. Elevated cTnI (sHR 1.33, 95% CI 1.23–1.44) and NT–proBNP (sHR 1.75, 95% CI 1.64–1.87) were associated with higher CVD risk. SCORE2 demonstrated a C-index of 0.748 (95% CI 0.729–0.766). Addition of NT-proBNP increased discrimination to 0.780 (95% CI 0.762–0.798), whereas addition of cTnI increased the C-index to 0.763 (95% CI 0.745–0.781). Cancer-related variables did not improve discrimination. In cancer survivors, SCORE2 demonstrated a C-index of 0.789 (95% CI 0.716–0.863). Addition of cTnI numerically increased the C-index to 0.802 (95% CI 0.732–0.871); however, estimates were imprecise due to limited events. Conclusions Cardiac biomarkers improved risk discrimination beyond SCORE2, whereas cancer-related variables did not provide incremental predictive value. Findings were directionally similar in cancer survivors. These results support biomarker-guided refinement of established risk models as a pragmatic strategy in cardio-oncology and underscore the importance of accounting for competing non-cardiovascular mortality.