Gynnild Mari Nordbø, Holtrop Joris, Hageman Steven H J, Vinje Victoria, Dorresteijn Jannick A N, Visseren Frank L J, Holte Espen, Dalen Håvard, Wethal Torgeir, Omland Torbjørn
Department of Circulation and Medical Imaging, Faculty of Medicine and Health Science, NTNU - Norwegian University of Science and Technology, Trondheim, Norway.
Clinic of Cardiology, St. Olavs Hospital, Trondheim, Norway.
Eur J Prev Cardiol. 2025 Jun 23. doi: 10.1093/eurjpc/zwaf356.
The 2022 ESC Cardio-oncology guidelines recommend cardiovascular disease (CVD) risk stratification for cancer patients and suggest using SCORE2 and SCORE2-OP. However, these models have not been validated or specifically adapted for cancer populations.
Refinement of SCORE2 and SCORE2-OP models to accurately predict 10-year fatal and non-fatal CVD risk in cancer patients.
We included 1,622 patients from the HUNT3 study (2006-2008) who were diagnosed with cancer within 4 years after their enrollment and followed until 2023 linked to national registries. The primary outcome was a composite of myocardial infarction (MI), stroke, or CVD mortality. Model performance was assessed using Harrel's C-statistic and calibration curves. Both models were recalibrated by applying a multiplicative adjustment factor based on expected-observed (E/O) ratios.
The most prevalent cancers were gastrointestinal (23%), prostate (17%), and breast (14%). Mean age was 65.2 years, 52% were female. During a median follow-up of 8.8 years [interquartile range 1.9-12.6], 252 CVD events (39% MI, 36% stroke, 25% CVD deaths) and 708 non-CVD deaths occurred. SCORE2 initially underestimated CVD risk (E/O ratio for men and women: 0.91 and 0.63, respectively) but showed adequate agreement after recalibration. C-statistics for SCORE2 was 0.693 (95% confidence interval (CI) 0.643-0.743), and 0.730 (95% CI 0.676-0.784) after excluding those not surviving the first 2 years. For SCORE2-OP, the C-statistics were 0.586 (95% CI 0.529-0.643) and 0.648 (95% CI 0.577-0.720).
SCORE2 underestimated CVD risk in cancer patients. After recalibration, the model may serve as a valuable tool for risk stratification in cancer patients.
2022年欧洲心脏病学会心脏肿瘤学指南建议对癌症患者进行心血管疾病（CVD）风险分层，并建议使用SCORE2和SCORE2-OP。然而，这些模型尚未在癌症人群中得到验证或专门适配。
完善SCORE2和SCORE2-OP模型，以准确预测癌症患者10年致命和非致命CVD风险。
我们纳入了HUNT3研究（2006 - 2008年）中的1622名患者，这些患者在入组后4年内被诊断患有癌症，并与国家登记处进行关联随访至2023年。主要结局是心肌梗死（MI）、中风或CVD死亡的复合结局。使用Harrel's C统计量和校准曲线评估模型性能。通过应用基于预期 - 观察（E/O）比率的乘法调整因子对两个模型进行重新校准。
最常见的癌症是胃肠道癌症（23%）、前列腺癌（17%）和乳腺癌（14%）。平均年龄为65.2岁，52%为女性。在中位随访8.8年[四分位间距1.9 - 12.6]期间，发生了252例CVD事件（39%为MI，36%为中风，25%为CVD死亡）和708例非CVD死亡。SCORE2最初低估了CVD风险（男性和女性的E/O比率分别为0.91和0.63），但重新校准后显示出足够的一致性。SCORE2的C统计量为- 0.693（95%置信区间（CI）0.643 - 0.743），排除前两年内未存活的患者后为0.730（95% CI 0.676 - 0.784）。对于SCORE2-OP，C统计量分别为0.586（95% CI 0.529 - 0.643）和0.648（95% CI 0.577 - 0.720）。
SCORE2低估了癌症患者的CVD风险。重新校准后，该模型可作为癌症患者风险分层的有价值工具。
