Assessment of three equations to calculate plasma LDL cholesterol concentration in fasting and non-fasting hypertriglyceridemic patients.

Abstract

Low-density lipoprotein cholesterol (LDL-C) concentration was calculated for many years using the Friedewald equation, but those from Sampson and extended-Martin-Hopkins perform differently. Their accuracy in fasting hypertriglyceridemia and non-fasting state were compared and the clinical impact of implementing these equations on risk classification and on the setting of lipid treatment goals was assessed. Seven thousand six standard lipid profiles and LDL-C concentrations measured after ultracentrifugation (uLDL-C) were retrospectively included. uLDL-C were compared to calculated LDL-C in terms of correlation, root mean square error, residual error, mean absolute deviations and cardiovascular stratification. In fasting state (n=5,826), Sampson equation was the most accurate, exhibited the highest percentage of residual error lower than 0.13 mmol/L (67 vs. 57 % and 63 % using Friedewald, or extended-Martin-Hopkins equations respectively) and the lowest misclassification rate. However, the superiority of this equation was less pronounced when triglyceride concentration (TG) <4.5 mmol/L were considered. In post-prandial state (n=1,180), extended-Martin-Hopkins was the most accurate equation, exhibited the highest percentage of residual error lower than 0.13 mmol/L (73 vs. 39 % and 57 % using Friedewald and Sampson equation respectively). Overall, the negative bias with Sampson equation may lead to undertreatment. Conversely, a positive bias was observed with extended Martin-Hopkins. None of the equations tested are accurate when TG>4.52 mmol/L. When TG<4.52 mmol/L both Sampson and Martin-Hopkins equations performed better than Friedewald. The switch to one or the other should take in account their limitations, their ease of implementation into the lab software and the proportion of non-fasting patients.