Age- and Sex-Adjusted Myocardial Flow Reserve Percentiles for Personalized Cardiovascular Risk Assessment

用于个体化心血管风险评估的年龄和性别校正心肌血流储备百分位数

阅读:1

Abstract

INTRODUCTION: Positron emission tomography (PET) myocardial flow reserve (MFR) is a robust indicator of coronary vascular health and a strong predictor of cardiovascular risk. Clinical guidelines typically use fixed MFR thresholds (e.g., <2.0) to stratify risk, yet this approach overlooks individual variation, particularly by age and sex. We aimed to establish age- and sex-adjusted MFR percentiles and to evaluate their prognostic and predictive performance for cardiovascular risk assessment, in comparison with conventional fixed-threshold MFR approach. METHODS: Using data from the REFINE PET registry (24,820 patients from 12 sites), we measured PET MFR and derived age- and sex-adjusted MFR reference percentiles using quantile regression in patients without known coronary artery disease. All patients were categorized into percentile-based quartile groups. The primary outcome for prognostic and prediction analyses was major adverse cardiovascular events (MACE), defined as all-cause mortality, myocardial infarction, or heart-failure hospitalization. Time-to-event associations were evaluated using covariate-adjusted survival models, with cumulative incidence and hazard ratios (HR) estimated at 1 and 5 years in the derivation dataset, an independent but similar validation dataset A, and a high-risk validation dataset B. Predictive performance for MACE was assessed using discrimination, calibration, and reclassification metrics, comparing percentile-based models with models using a fixed MFR threshold (<2.0). RESULTS: Among participants (mean age 66.5 years; median follow-up 3.6 years), age- and sex-adjusted MFR quartile groups were strong independent predictors of MACE, with adjusted HR increasing stepwise across quartile groups at both early and later follow-up. At 1 year, HR (95% CI) comparing the lowest to the highest quartile group were 4.06 (3.41-4.82) in the derivation cohort, 3.31 (2.32-4.71) in validation cohort A, and 2.35 (2.05-2.70) in validation cohort B. At 5 years, the corresponding HR were 2.18 (1.86-2.56), 1.77 (1.31-2.40), and 1.59 (1.36-1.86). Percentile-based models demonstrated consistently higher discrimination, better calibration, and greater net reclassification for MACE at both time points compared with fixed-threshold MFR models. Although 67.2% of patients had preserved MFR (>2.0), cardiovascular risk increased steadily across MFR percentiles even within this range.Several limitations should be considered. First, the study population may not represent the broader, non-referral population or specialized groups such as cardiac transplant patients. Second, although missing data was minimal overall, information on abnormal renal function was missing for a substantial proportion of participants and therefore could not be fully adjusted for in the multivariable models. Third, perfusion and flow measurements were fully automatically processed using standard quantitative software, which may differ from semi-automatic measurements, though the increasing adoption of AI-based tools and validated software is likely to standardize automated processing soon. Finally, the proportion of non-White participants was 15-20%, which is lower than that of the overall U.S. population, but may be appropriate given that the study included patients from North and Central America and Europe. CONCLUSION: Age- and sex-adjusted MFR percentiles provide a reliable, clinically actionable measure of vascular health, improving cardiovascular risk stratification by better capturing age- and sex-related heterogeneity in vascular risk. Compared with traditional fixed-threshold approaches, MFR percentiles demonstrate improved predictive performance for cardiovascular risk assessment across diverse patient populations.

特别声明

1、本页面内容包含部分的内容是基于公开信息的合理引用;引用内容仅为补充信息,不代表本站立场。

2、若认为本页面引用内容涉及侵权,请及时与本站联系,我们将第一时间处理。

3、其他媒体/个人如需使用本页面原创内容,需注明“来源:[生知库]”并获得授权;使用引用内容的,需自行联系原作者获得许可。

4、投稿及合作请联系:info@biocloudy.com。