ESC Congress 2026 Abstract · Oct 26, 2026
Integrated analysis of lipid and inflammatory residual cardiovascular risk components for incident stroke: a cohort study with machine learning
By Wang J
Background: Residual cardiovascular risk components have emerged recently as promising biomarkers for cardiovascular risk assessment. The atherogenic index of plasma (AIP), calculated as log10[triglyceride (mg/dL)/high-density lipoprotein (mg/dL)], and the high-sensitivity C-reactive protein (hsCRP) level are predictors of stroke, but the understanding of their combined effects on incident stroke is limited. This study aimed to investigate the association between the combined evaluation of the AIP and hsCRP level and incident stroke among middle-aged and elderly adults. Methods: This study included 9,012 respondents aged ≥45 years without stroke at baseline from the China Health and Retirement Longitudinal Study (CHARLS, 2011–2018). The primary outcome was new stroke events. Multivariable logistic regression was performed to evaluate the associations of the AIP, hsCRP level, and the combination of the AIP and hsCRP level with incident stroke. Two machine learning methods (Boruta algorithms and Extreme Gradient Boosting (XGBoost)) were employed to select the most critical variables for model 3, which is distinct from selection methods based only on clinical experience. Restricted cubic spline (RCS) analysis was applied to separately investigate the relationships between the AIP and hsCRP level and stroke risk, and subgroup analyses assessed these associations among different subgroups. Results: In total, 755 (8.38%) patients developed stroke during the seven-year period. High AIP or hsCRP levels were independently associated with stroke. Compared with the lowest AIP and hsCRP quartiles, the highest AIP quartile (odds ratio (OR) 1.57; 95% CI: 1.25–1.99) and highest hsCRP quartile (OR 1.44; 1.16–1.81), respectively, were significantly associated with incident stroke, with a linear dose–response relationship according to the RCS analysis. Joint effects analysis revealed that participants with concurrently high AIP and hsCRP levels had higher risks of incident stroke than did those with elevated levels of either marker alone (OR 1.69, 1.36–2.11). Subgroup and sensitivity analyses indicated that the results were robust. Conclusions: This cohort study suggests that combining the AIP and hsCRP level, two residual cardiovascular risk components, can better detect incident stroke than either marker alone does. Our findings underscore the importance of integrating both the AIP and hsCRP level for comprehensive stroke risk assessment.Graphical abstract imageFor image description, please refer to the figure legend and surrounding text.