Clinical relevance of the discordance between the Friedewald and Martin-Hopkins equations in the estimation of LDL-cholesterol in patients with acute coronary syndrome

Introduction Low-density lipoprotein cholesterol (LDL) levels are usually calculated using the Friedewald (F) equation, although the Martin-Hopkins (MH) equation has been proposed as a more accurate alternative. The difference between the application of one or the other could have clinical relevance in patients with very high cardiovascular risk, such as those admitted for acute coronary syndrome (ACS). Purpose Our aim was to assess the discordance between F and MH to calculate the estimated LDL lowering ratio (LR_LDL) needed to reach the target <55mg/dL in patients hospitalised with ACS and to compare the LR_LDL calculated from F and MH with a direct measurement (DM) of LDL in a subgroup of patients. Methods Observational, single-centre study of a cohort of 563 consecutive patients admitted for ACS with triglycerides <400 mg/dL. Lipid profile was determined on admission and LDL was calculated by using F and MH equations. In a subgroup of 84 patients, DM LDL (enzymatic colorimetry) was performed. The difference between LDL estimated by each method and 55 mg/dL was used to calculate LR_LDL. The difference in LR_LDL between methods was calculated and discordance was considered when it was >6% (LDL decrease expected when doubling the statin dose). Results In the overall population of 563 patients, the median LR_LDL according to F was 44.4% [29.5-57.0]; while by MH it was 46.9% [32.8-58.0], with a strong positive correlation (r=0.974; p<0.001). 68 cases (12.1%) were identified with discordance between F and MH (table), which was associated with low LDL and elevated triglyceride levels. In the 84 patients with DM, the median LR_LDL was 49.5% [33.9-58.3], with a strong positive correlation between LR_LDL obtained with DM and that calculated with both F and MH (0.961 and 0.975, respectively; p<0.001). However, there was discordance between LR_LDL obtained with DM and calculated with F and/or MH in 15 patients (17.8%), and it was associated with elevated triglyceride levels (OR 1.017; CI95% 1.009-1.026; p<0.001). As illustrated by the Bland Altman plots (figure), the discordance was due to underestimation of calculated LR_LDL in all but two cases. The discordance rate was higher with F (23.8%) than with MH (10.7%), p<0.001; and the magnitude of the difference in discordant cases was also higher with F than with MH (8.4% vs 2.5%, p<0.001), with respect to the LR_LDL obtained with DM. Conclusions Despite the good correlation in the estimation of LR_LDL using the F and MH equations, clinically relevant discordance exists in approximately one in eight patients admitted for ACS. In most cases of discordance, the MH equation is more accurate when compared to DM, especially in patients with elevated triglyceride levels. In these circumstances, in patients hospitalised with ACS, the use of MH to estimate LR_LDL would be particularly advisable.