Inflammation
Lp(a) raises post-PCI cardiovascular risk mainly when pericoronary inflammation is high, machine-learning study (J Clin Lipidol 2026)
Original title: Lipoprotein(a) and coronary inflammation: Joint predictors of residual risk in post-percutaneous coronary intervention coronary artery disease
Study of patients with coronary artery disease undergoing percutaneous coronary intervention (PCI), using coronary CT angiography to quantify pericoronary fat attenuation index (FAI) as a marker of coronary inflammation, and machine learning (SHAP) to rank its predictive features, testing interaction with Lp(a) (30 mg/dL cut-off) for major adverse cardiovascular events (MACE). Over a median 3-year follow-up, 181 patients (9.0%) had MACE. High Lp(a) (HR 1.29, 95% CI 1.05-1.76) and high FAI around the left anterior descending artery (FAI-LAD, HR 2.06, 95% CI 1.51-2.81) each independently predicted MACE. High Lp(a) markedly increased MACE risk in the high FAI-LAD group (HR 2.69, 95% CI 1.75-4.10) but not in the low-FAI-LAD group (HR 1.10, 95% CI 0.65-1.87). Combining Lp(a) and FAI-LAD improved prognostic accuracy (AUC 0.713) over conventional risk factors (AUC 0.599) or either marker alone, with interactions also seen by age and diabetes.
Original abstract
Background: The interaction of elevated lipoprotein(a) [Lp(a)] with local coronary inflammation remains unclear.
Objective: We investigated whether pericoronary inflammation modulates Lp(a)'s prognostic impact on major adverse cardiovascular events (MACE) after percutaneous coronary intervention (PCI).
Methods: Patients with coronary artery disease undergoing PCI were included with coronary computed tomography angiography within 3 months before coronary angiography. Pericoronary fat attenuation index (FAI) was quantified for pericoronary inflammation evaluation. Four machine learning models with SHapley Additive exPlanations (SHAP) ranked FAI features. Lp(a) was categorized using a 30 mg/dL cut-off per Chinese lipid guidelines. Associations with MACE (all-cause mortality, nonfatal myocardial infarction, unplanned revascularization, and stroke) were assessed via multivariable Cox regression, restricted cubic splines (RCS), and subgroups.
Results: Over a median 3-year follow-up, 181 (9.0%) patients experienced MACE. High Lp(a) and elevated FAI independently predicted MACE (hazard ratio [HR] 1.29, 95% CI 1.05-1.76 for high Lp(a); HR 2.06, 95% CI 1.51-2.81 for high FAI-Left anterior descending artery [LAD]). Extreme gradient boosting (area under the curve [AUC] 0.936) ranked FAI-LAD highest (SHAP 1.57). RCS revealed nonlinear associations, with MACE risk escalating sharply above FAI-LAD -77.00 HU. High Lp(a) markedly increased MACE risk in the high FAI-LAD group (HR 2.69, 95% CI 1.75-4.10) but the risk increase was attenuated in the low-FAI-LAD group (HR 1.10, 95% CI 0.65-1.87). Combining Lp(a) and FAI-LAD improved prognostic accuracy (AUC 0.713) over conventional risk factors (AUC 0.599) or either alone (DeLong P < .05). Subgroup analyses showed interactions with age and diabetes (P < .05).
Conclusion: Lp(a) elevates post-PCI risk mainly in high pericoronary inflammation settings. Integrating Lp(a) and FAI refines risk stratification, supporting targeted Lp(a)-lowering in inflamed subgroups.
diabetesinflammationplaque imagingrisk prediction
Summary written by lp-a.org from the published abstract; figures as published. Page updated 17 August 2026. Methods.