Testing
Lp(a) above 58 mg/dL overestimates LDL-C by 10% or more across all major formulas, study of 3,923 (Clin Lab 2026)
Original title: Impact of Lipoprotein (a) on the Quantification of LDL-Cholesterol
Comparison of beta-quantification LDL-C (reference method) with Friedewald, Martin/Hopkins and Sampson formula-based LDL-C in 3923 datasets from two cohorts, testing the distorting effect of Lp(a) cholesterol content. Above 10 mg/dL Lp(a), the effect on beta-quantification LDL-C became significant and grew with higher Lp(a), proportionally larger at lower LDL-C values; above 58 mg/dL Lp(a), true LDL-C was overestimated by 10% or more, a clinically relevant threshold. Similar overestimation occurred with the Friedewald, Martin/Hopkins and Sampson formulas, with Friedewald showing the smallest deviation regardless of Lp(a) correction. Artificial intelligence models could not flag elevated Lp(a) from the conventional lipid profile alone. The authors recommend Lp(a)-corrected LDL-C (crude LDL-C minus Lp(a) times 0.23 plus 1.00) when Lp(a) is high, LDL-C is low, or LDL-C-lowering treatment underperforms expectations.
Original abstract
Background: Accurate LDL-C measurement is essential for cardiovascular risk management. The established methods to determine LDL-C also include Lp(a)-C and potentially distort the actual LDL-C value. The need for Lp(a)-adjusted LDL-C remains debated. Our study aimed to evaluate the impact of Lp(a) on the determination of LDL-C.
Methods: We included 3,923 datasets from two cohorts. LDL-C was determined by beta-quantification (LDL-CUC), the reference method recommended by the Lipid Research Clinics, and according to Friedewald (LDL-CFW), Martin/Hopkins (LDL-CMH), and Sampson (LDL-CSN). Correction of LDL-C was performed as follows: corrected LDL-C* = crude LDL-C - (Lp(a) x 0.23 + 1.00). Passing-Bablok regression and Spearman correlation were used for intermethod comparisons.
Results: Above 10 mg/dL Lp(a) had a significant effect on LDL-CUC. The effect increased with increasing concentrations of Lp(a) levels and, in relative terms, was most pronounced at lower LDL-C values. For Lp(a) > 58 mg/dL, the actual LDL-CUC was overestimated by ≥ 10%, which was considered clinically relevant. Similar overestimations were observed with the Friedewald, Martin/Hopkins, and Sampson formulas, with Friedewald showing the smallest deviation from LDL-C regardless of Lp(a)-correction. Artificial intelligence models showed that it was not possible to raise the suspicion of elevated Lp(a) from the conventional lipid profile.
Conclusions: The influence of Lp(a) on the determination of LDL-C may lead to clinically significant overestimations of the actual LDL-C. Therefore, we recommend using Lp(a)-corrected LDL-C when 1) the Lp(a) concentration is high, 2) the LDL-C concentration is low, and 3) the LDL-C-lowering treatment is less effective than expected.
Summary written by lp-a.org from the published abstract; figures as published. Page updated 17 August 2026. Methods.