External validation of the Systematic COronary Risk Evaluation 2 (SCORE2) model in cancer patients

Abstract Background Guidelines recommend cardiovascular disease (CVD) risk stratification before cancer treatment to better identify patients for intensified follow-up or cardioprotective treatment. Only a limited number of CVD risk stratification tools exists for cancer patients which at present is not widely used nor extensively validated. The ESC Guidelines on Cardio-oncology 2022 suggest using traditional CVD models like the SCORE2 (Systematic Coronary Risk Estimation 2). However, external validation in cancer populations is lacking. Purpose Our aim was to study the prevalence of CVD risk factors and incidence of cardiovascular events in a cancer cohort and externally validate the Systematic COronary Risk Evaluation 2 (SCORE2) model in this context. Methods We included 1058 patients from a large-scale populational study without diabetes or established CVD aged 40 to 69 years and diagnosed with cancer between 2006 and 2012. Participants were assessed for cardiovascular risk factors at baseline and followed until Sept 2023. The Norwegian Cancer Registry provided patient specific information about cancer characteristics and therapy. The primary outcome was the combination of myocardial infarction (MI), stroke or CVD mortality, which was retrieved from the local hospitals’ registries and the national causes of death registry. Performance was assessed using smooth calibration curves of predicted and observed risk and Harrel’s C-statistic for discrimination, both corrected for competing risks. SCORE2 was recalibrated by adapting the model with a single multiplicative constant based on the expected-to-observed ratio (E/O-ratio). Results The most common cancer types were gastrointestinal cancer (19%), male genital including prostate cancer (19%) and breast cancer (18%). Mean (SD) age was 58.6 (7.6) years, 54% were female. Prevalence of cardiovascular risk factors are shown in Table 1. In total, 121 (11%) individuals had a CVD event (MIs 53 (44%), strokes 46 (38%) and CVD deaths 22 (18%)) during a median (interquartile range) follow-up of 11.1 (3.4 to 13.1) years, and 360 (34%) died due to non-CVD causes. Calibration plots showed adequate agreement between estimated and observed 10-year prognosis after recalibration (C-statistics 0.685, 95% CI 0.643 to 0.728), Figure 1. E/O-ratio for men was 0.65 and for women 0.63, indicating underestimation of the observed CVD incidence. After recalibration, the median estimated 10-year risk was 8.1% (IQR 5 to 11), 36% had >10% 10-year risk and 9% had >15% 10-year risk. Conclusion SCORE2 underestimated CVD risk in a general cancer population but showed adequate agreement after recalibration and C-statistics aligned with the original external validation cohorts. SCORE2 is useful also in cardio-oncology setting (after recalibration) to identify individuals at higher risk of developing CVD.