Lipoprotein(a), ApoB/ApoA-I ratio, and atherogenic index of plasma in premature coronary artery disease: nonlinear associations and incremental discriminative value

Lipoprotein(a) [Lp(a)], the apolipoprotein B/apolipoprotein A-I (ApoB/ApoA-I) ratio, and the atherogenic index of plasma (AIP) reflect complementary aspects of atherogenic lipid-related risk, but their nonlinear associations with premature coronary artery disease (PCAD), component-level interpretation, and relation to coronary atherosclerotic severity remain incompletely characterized. In this single-center retrospective case-control study, after application of the prespecified inclusion criteria and clinical exclusion criteria, 625 symptomatic patients undergoing coronary angiography were considered potentially eligible. Two additional analytic eligibility requirements were then applied: availability of the relevant lipid measurements from our hospital laboratory and absence of documented pre-admission lipid-lowering medication use. Of these patients, 111 were excluded because the relevant lipid measurements had been performed outside our hospital and 56 because of documented pre-admission lipid-lowering medication use. The final cohort comprised 458 participants, including 233 patients with PCAD and 225 symptomatic angiographic controls with < 50% stenosis in all major coronary arteries. After mutual adjustment for the three markers and adjustment for age, sex, body mass index, smoking, hypertension, and diabetes, Lp(a), the ApoB/ApoA-I ratio, and AIP all remained associated with PCAD status (all P < 0.001), with significant nonlinearity. No material multicollinearity was identified; VIFs for Lp(a), the ApoB/ApoA-I ratio, and AIP were 1.11, 1.16, and 1.11, respectively. Adding the three markers to the clinical-plus-LDL-C model increased the AUC from 0.775 to 0.919 (ΔAUC = 0.145, 95% CI 0.105–0.184, P < 0.001), with a bootstrap-corrected AUC of 0.910. In component analysis, the clinical + LDL-C+ApoB+ApoA-I model had a higher AUC than the clinical + LDL-C+ratio model (0.924 vs. 0.891; ΔAUC = 0.033, 95% CI 0.010–0.056, P = 0.005). Among patients with PCAD, each 0.1-unit increase in AIP was associated with a 9.4% higher adjusted geometric mean Gensini score (95% CI 6.7%-12.3%, P < 0.001), whereas Lp(a) and the ApoB/ApoA-I ratio were not significant in the same adjusted model. In this angiography-selected case-control cohort, the three prespecified lipid-related markers showed nonlinear associations with PCAD and provided additional in-sample discriminative information beyond the selected available clinical covariates and LDL-C. The ApoB/ApoA-I ratio should not be interpreted as providing discrimination superior to its individual components. AIP was also associated with coronary atherosclerotic severity after multivariable adjustment. These findings require external validation and should not be interpreted as an absolute-risk prediction model for the general population.