Association of apolipoprotein B and excess apolipoprotein B with cardiovascular risk in type 2 diabetes: a prospective cohort study of the UK Biobank

BACKGROUND: Residual cardiovascular risk persists in type 2 diabetes mellitus (T2DM) despite intensive risk-factor management. Apolipoprotein B (apoB) and excess apoB are potentially promising biomarkers for identifying residual cardiovascular risk. We assessed apoB and excess apoB in T2DM for incremental prediction of atherosclerotic cardiovascular disease (ASCVD) risk. METHODS: This prospective cohort included 11,918 UK Biobank participants (mean age 59.7 ± 6.6 years; 61% male) with T2DM and no ASCVD at baseline. Excess apoB was defined as the observed minus predicted apoB, where the predicted value was derived using a linear regression model of apoB on low-density lipoprotein cholesterol (LDL-C) fitted in a statin-naïve reference subset with triglycerides ≤ 1.0 mmol/L. The primary endpoint was incident ASCVD. Secondary endpoints included major adverse cardiovascular events (MACE) and all-cause mortality. Hazard ratios (HRs) and 95% confidence intervals (CIs) were estimated using multivariable Cox models. Nonlinearity was assessed using restricted cubic splines. Incremental improvements were quantified using the C-index, net reclassification improvement (NRI). RESULTS: During a median 185.3-month follow-up, 2,548 ASCVD and 1,205 MACE events occurred. ApoB was linearly related to ASCVD and MACE, while excess apoB showed J-shaped associations with a nadir near - 7.5 mg/dL for ASCVD. Both apoB and excess apoB showed positive associations with ASCVD across ascending percentile categories. Versus < 50th percentile, HRs (95% CIs) for ASCVD in higher apoB categories (50-<75th, 75-<90th, ≥ 90th) were 1.31 (1.16-1.49), 1.51 (1.25-1.81), and 1.47 (1.10-1.95); corresponding HRs (95% CIs) for excess apoB were 1.50 (1.36-1.66), 1.45 (1.29-1.63), and 1.53 (1.33-1.76), respectively. Similar but weaker risk gradients were observed for MACE. Neither apoB nor excess apoB was associated with all-cause mortality. Excess apoB yielded greater prediction improvement than apoB (ΔC-index: 0.009 vs. 0.002; NRI: 0.270 vs. 0.101) and better stratified risk in statin users and those with LDL-C ≤ 100 mg/dL (P for interaction < 0.05). CONCLUSIONS: In T2DM, apoB is independently associated with ASCVD but adds limited discrimination over conventional lipids. Excess apoB yielded improved discrimination and reclassification, and may serve as a complementary ASCVD risk marker, particularly in statin-treated settings. However, its clinical application requires external validation and standardization.