生物学年龄在社区人群队列中预测心血管病风险的应用

Application of biological age for cardiovascular risk prediction in a community-based Chinese cohort

摘要

目的：在大样本社区人群队列中独立验证 Light BioAge 模型计算的简易生物学年龄预测全因死亡的区分度，并探索其与时序年龄的差值和心血管病发病及死亡结局事件的关联，以及生物学年龄对心血管病风险预测的实际应用价值。方法：研究对象为中国勤州电子健康档案研究（CHinese Electronic health Records Research in Yinzhou，CHERRY）队列中 40~79 岁基线无心血管病史的人群，利用 C 统计量（Harrell＇s concordance index，C-index）评估仅通过空腹血糖、血清肌酐和超敏 C 反应蛋白三个临床常用指标构建的 Light BioAge 模型在全人群及不同性别亚组中预测全因死亡的外部验证区分度，采用 Cox 比例风险模型分析年龄差值与首次发生的心血管病发病和心血管病死亡复合终点的关联，调整时序年龄、性别、教育水平、居住地、吸烟状态、体重指数、收缩压、总胆固醇及高密度脂蛋白胆固醇等影响因素后计算风险比（hazard ratio，HR）及其 95% 置信区间（confidence interval，CI）。采用限制性立方样条回归方法进一步分析年龄差值与心血管病结局事件的潜在非线性关联。在2019年世界卫生组织（World Health Organization，WHO）开发的心血管病风险预测简易模型中采用生物学年龄替换时序年龄，评估对 10年心血管病风险预测区分度和校准度的影响。结果：共纳入 226406 名研究对象，人群的基线平均年龄为 55.0 岁，53.2% 为女性。在中位 7.39 年（累计 1562141 人年）的随访期间，共有 11703 人（7.49/1000 人年）发生死亡，9815 人（6.30/1000 人年）发生心血管病发病和死亡复合结局事件。Light BioAge 模型计算的生物学年龄及年龄差值的中位数分别为49.31和-5.19，提示其整体上低于时序年龄；虽然 Light BioAge 模型在全人群中预测全因死亡的区分度较好（C 统计量： 0.742,95\%CI:0.738∼0.746），但在男性中的区分度低于女性[C 统计量分别为 0.722（95%CI:0.714∼0.730）和 $0.755\text{ }(95\mathrm{\%}\mathit{C}\mathit{I}:0.749\sim 0.761)\text{ }$]。调整其他影响因素后，心血管病结局事件的风险呈现随年龄差值增大而增高的趋势（P_trend <0.001）；年龄差值处于上四分位数组与下四分位数组相比，男性心血管病结局事件的风险增加了 $21\mathrm{\%}\text{ }(\mathit{H}\mathit{R}=\mathrm{1.21,95}\mathrm{\%}\mathit{C}\mathit{I}:1.12\sim 1.30)\text{ }$ ，女性增加了 $27\mathrm{\%}\text{ }(\mathit{H}\mathit{R}=\mathrm{1.27,95}\mathrm{\%}\mathit{C}\mathit{I}:1.17\sim 1.38)\text{ }$ 。限制性立方样条回归的结果进一步显示，总人群的心血管病结局事件的风险随年龄差值的增加而升高，且男性、女性亚组中并未发现明显的阈值效应（非线性关联 P>0.05）。在 WHO 模型中采用生物学年龄替换时序年龄后，预测心血管病风险的区分度无改善，但校准度有明显提升，尤其是在女性人群中从时序年龄模型整体风险高估的20.5%[预测/观察比（expected/observed ratio， EOR$)\text{ }=1.205,\text{ }95\mathrm{\%}\mathit{C}\mathit{I}:1.167\sim 1.246$]改善为生物学年龄替换后的整体风险仅略微低估 $2.1\mathrm{\%}\text{ }(\mathrm{E}\mathrm{O}\mathrm{R}=\mathrm{0.979,95}\mathrm{\%}\mathit{C}\mathit{I}:0.948\sim 1.012)\text{ }$ 。结论：Light BioAge 模型构建的简易生物学年龄在大样本社区人群中预测全因死亡的区分度较好，生物学年龄和时序年龄的差值增大将增加心血管病结局事件的风险，在 WHO 心血管病风险预测模型中用生物学年龄替换时序年龄后，模型的校准度明显改善，特别是在女性人群。

Abstract

Objective：To independently evaluate the discrimination of the light version of biological age（Light BioAge）model for predicting all-cause mortality，to explore the association of the difference on Light BioAge and chronological age（AgeDiff）with the composite outcomes of cardiovascular disease （CVD），and to assess the performance of CVD risk prediction using the Light BioAge instead of chrono- logical age in a large Chinese population-based cohort. Methods：Participants aged 40-79 years without a history of CVD at baseline were drawn from the CHinese Electronic health Records Research in Yinzhou （CHERRY）study.Harrell＇s concordance index（C-index）was employed to assess the discrimination of Light BioAge in predicting all-cause mortality across the overall population and sex-specific subgroups. Cox proportional hazards models were used to assess the association between AgeDiff and the composite outcome of CVD onset and death，adjusting for chronological age，sex，education，region，smoking sta- tus，body mass index，systolic blood pressure，total cholesterol，and high-density lipoprotein cholesterol. Hazard ratios（HR）and 95％confidence intervals（CI）were calculated.Restricted cubic spline（RCS） regression was used to further analyze the potential nonlinear association between AgeDiff and CVD out- comes.Light BioAge was introduced to replace chronological age in the World Health Organization （WHO）non-laboratory CVD risk model to evaluate the discrimination and calibration in predicting 10- year CVD risk. Results：A total of 226406 adults were included，with a mean age of 55.0 years at base- line，53.2％of whom were women.During a median follow-up of 7.39 years（cumulative 1562141 person-years）， 11703 deaths（7.49 per 1000 person-years）and 9815 CVD events（6.30 per 1000 person-years）occurred.The median Light BioAge and AgeDiff were 49.31 and-5.19 years，respec- tively，suggesting an underestimation of chronological age.Although the Light BioAge model demonstrated good discrimination for predicting all-cause mortality in the overall population（C-index：0.742，95％CI： 0.738-0.746），discrimination was lower in men（0.722,95\%CI:0.714-0.730）than in women （0.755,95\%CI:0.749-0.761）.After adjusting for confounders，the risk of CVD events showed an elevated trend with increasing AgeDiff（P_trend <0.001）.Compared with the lowest quartile of AgeDiff， the risk of CVD events in the highest quartile increased by 21％in men（HR=1.21,95\%CI:1.12- 1.30) and 27% in women（HR=1.27,95\%CI:1.17-1.38）.RCS regression further indicated that in the overall population，the risk of CVD events increased with AgeDiff.Besides，no significant thre- shold effect was observed in sex-specific subgroups（P for non-linearity >0.05）.Replacing chronologi- cal age with the Light BioAge in the WHO model did not improve discrimination；however，it significantly enhanced calibration.Calibration improvement was especially evident in women：while chronological age overestimated risk by 20.5％[expected/observed ratio $\left(\mathrm{E}\mathrm{O}\mathrm{R}\right)=1.20\mathrm{5,95}\mathrm{\%}\mathit{C}\mathit{I}:1.167-1.246$），the Light BioAge reduced this to a marginal 2.1％underestimation $(\mathrm{E}\mathrm{O}\mathrm{R}=\mathrm{0.979,95}\mathrm{\%}\mathit{C}\mathit{I}:0.948-1.012)$. Conclusion：The discrimination of the Light BioAge in predicting all-cause mortality seems good，and a wider AgeDiff indicates higher cardiovascular risk in this large population-based Chinese cohort.Repla- cing chronological age with biological age in the WHO non-laboratory model significantly improved cali- bration for women.

关键词

生物学年龄 / 心血管病 / 队列研究 / 风险评估 / 性别差异

Key words

Biological age / Cardiovascular disease / Cohort study / Risk assessment / Sex disparity

参考文献

[1] van Daalen KR, Zhang D, Kaptoge S, et al. Risk estimation for the primary prevention of cardiovascular disease:Considerations for appropriate risk prediction model selection[J]. Lancet Glob Health, 2024, 12(8):e1343-e1358.

[2] Ferrucci L, Gonzalez-Freire M, Fabbri E, et al. Measuring bio- logical aging in humans:A quest[J]. Aging Cell, 2020, 19(2): e13080.

[3] Jiang M, Tian S, Liu S, et al. Accelerated biological aging ele- vates the risk of cardiometabolic multimorbidity and mortality[J]. Nat Cardiovasc Res, 2024, 3(3):332-342.

[4] Moqri M, Herzog C, Poganik JR, et al. Biomarkers of aging for the identification and evaluation of longevity interventions[J]. Cell, 2023, 186(18):3758-3775.

[5] Hamczyk MR, Nevado RM, Barettino A, et al. Biological versus chronological aging[J]. J Am Coll Cardiol, 2020, 75 (8): 919 - 930.

[6] Pedretti RFE, Asteggiano R, Gevaert AB, et al. Cardiovascular risk factors management in older adults:A clinical consensus state- ment from the European Association of Preventive Cardiology of the ESC and the ESC Council for Cardiology Practice[J]. Eur J Prev Cardiol,2025:zwaf175.

[7] Joyce BT, Gao T, Zheng Y, et al. Epigenetic age acceleration re- flects long-term cardiovascular health[J]. Circ Res, 2021, 129 (8):770-781.

[8] Cui X, Yi C, Zhang H, et al. DNA methylation algorithms of aging and incident cardiovascular disease:A prospective cohort study[J]. J Am Heart Assoc, 2025, 14(23):e045545.

[9] Macdonald-Dunlop E, Taba N, Klarić L, et al. A catalogue of omics biological ageing clocks reveals substantial commonality and associations with disease risk[J]. Aging, 2022, 14(2): 623 - 659.

[10] Rutledge J, Oh H, Wyss-Coray T. Measuring biological age using omics data[J]. Nat Rev Genet, 2022, 23(12):715-727.

[11] Khan SS, Greenland P, Hayman LL, et al. Criteria to assess the predictive and clinical utility of novel models,biomarkers,and tools for risk of cardiovascular disease:A scientific statement from the American Heart Association[J]. Circulation, 2026, 153 (11):e953-e970.

[12] Hao M, Zhang H, Wu J, et al. Gompertz law-based biological age (GOLD BioAge):A simple and practical measurement of biologi- cal ageing to capture morbidity and mortality risks[J]. Adv Sci, 2025, 12(32):e01765.

[13] Lin H, Tang X, Shen P, et al. Using big data to improve cardio- vascular care and outcomes in China:A protocol for the CHinese Electronic health Records Research in Yinzhou(CHERRY)Study[J]. BMJ Open, 2018, 8 (2):e019698.

[14] WHO CVD Risk Chart Working Group.World Health Organization cardiovascular disease risk charts:Revised models to estimate risk in 21 global regions[J]. Lancet Glob Health, 2019, 7 (10): e1332-e1345.

[15] Harrell FE. Regression modeling strategies:With applications to linear models,logistic and ordinal regression,and survival analysis[M]. Cham,Switzerland: Springer International Publishing, 2015:26-28.

[16] Fransquet PD, Wrigglesworth J, Woods RL, et al. The epigenetic clock as a predictor of disease and mortality risk:A systematic re- view and meta-analysis[J]. Clin Epigenetics, 2019, 11(1): 62.

[17] Liu Z, Kuo PL, Horvath S, et al. A new aging measure captures morbidity and mortality risk across diverse subpopulations from NHANES IV:A cohort study[J]. PLoS Med, 2018, 15 (12): e1002718.

[18] Lin S, Liu M, Lu X, et al. Association between biological aging and coronary heart disease:A secondary analysis of two observa- tional studies[J]. iScience, 2025, 29(1): 114478.

基金资助

国家科技创新2030四大慢病重大专项(2024ZD0527406)

国家自然科学基金(82373662)

北京市自然科学基金(IS24047)