Evaluation of twelve formulas for LDL-C estimation in a large, blinded, random Italian population

Background and aims: Low-density lipoprotein-cholesterol (LDL-C) is the major determinant of cardiovascular disease (CVD) burden. Being the direct assays time consuming, expensive, not fully standardized and not worldwide available, indirect formulas represent the most used laboratory estimation of LDL-C. In this study we analyzed the accuracy of twelve formulas for LDL-C estimation in an Italian population of 114,774 individuals.

Methods: All lipid samples were analyzed using direct homogeneous assay. The population was divided into various subgroups based on triglycerides and directly dosed LDL-C (D-LDL) levels. Twelve formulas (Friedewald, DeLong, Hata, Hattori, Puavillai, Anandaraja, Ahmadi, Chen, Vujovic, de Cordova, Martin, and Sampson) were compared in terms of their mean absolute deviations and the correlation and concordance of their estimated LDL-C with the respective D-LDL values.

Results: LCL-C measured by Friedewald formula and direct assay differed by more than 9 mg/dL. For D-LDL>115 mg/dl, we observed a concordance rate of only 55% between Friedewald and the respective D-LDL values. For TG<250 mg/dl, the proportion of reclassification between the different formulas and D-LDL was 14.1% with Vujovic, 14.4% Sampson, 15.9% DeLong, 16.5% Puavilai, 19.9% Martin, 21.9% Friedewald, 23.5% Chen, 29% Anandaraja, 31.1% Ahmadi, 31.5% Hata, 33.2% Hattori, and 44.4% with De Cordova formula.

Conclusions: Our study compared for the first time 12 different LDL-C formulas on a Southern European population of more than 100,000 people. 'Several formulas showed better accuracy compared to Friedewald. Sampson, Martin and Vujovic resulted the most accurate formulas.