Reliability of different formulas for estimating plasma Apolipoprotein B levels in a large cohort of South European individuals.

Background And Aims: Direct measurement of apolipoprotein B (ApoB) is not always standardized and is relatively expensive, making it unavailable in several low-income settings. To address this issue, several formulas have been developed to estimate ApoB levels. Therefore, our study aims to compare the reliability of 23 formulas for estimating ApoB levels in a large cohort of South-European individuals.

Methods: We retrospectively assessed 4.577 clinical records in which ApoB measurements were obtained using the same standardized method. Overall concordance was defined as the proportion of cases where the directly measured ApoB level fell within the same category as the estimated ApoB level, based on ApoB quartiles (<80 mg/dL, 80-94 mg/dL, 95-114 mg/dL, and ≥115 mg/dL). In addition, overall concordance was assessed for different lipoprotein(a) (Lp(a)) and non-high density lipoprotein cholesterol (non-HDL-C) sub-levels. Ordinary least squares linear regression analyses were performed to compare estimated and measured ApoB values. Residual error plots were generated to visualize the difference between each estimation method and the actual ApoB measurements, stratified by Lp(a) and non-HDL-C levels.

Results: Plasma ApoB levels were best predicted by a non-HDL-C based formula and a formula using Friedewald's low-density lipoprotein cholesterol (LDL-C), regardless of ApoB plasma levels. Non-HDL-C levels did not significantly affect the concordance between measured and estimated ApoB across the different formulas, except at low non-HDL-C levels. Similarly, Lp(a) levels did not significantly impact concordance. However, the highest concordance level was 41 %.

Conclusion: Some simple formulas based on low-cost and widely available parameters can estimate ApoB levels independently of ApoB, non-HDL-C, and Lp(a) plasma levels. This approach may be particularly useful for estimating ApoB levels in low-resource settings.