Introduction: The Friedewald equation (F-LDL-C) and the Martin-Hopkins algorithm (MH-LDL-C) estimate direct LDL-C from a standard lipid panel. Discordant LDL-C estimates by the two methods may carry significant clinical implications. We evaluated the clinical variables associated with discordant LDL-C estimates and the association of discordance with risk of incident atherosclerotic cardiovascular disease (ASCVD) in the Dallas Heart Study (DHS), a multi-ethnic, population based prospective cohort. Methods: We estimated F-LDL-C and MH-LDL-C in 2824 DHS participants (42% male; mean age 43.5 years) with TG ≤ 400 mg/dL, who were not on baseline lipid lowering therapy and were free of prior ASCVD. We divided the cohort into quintiles of LDL-C discordance (MH-LDL-C minus F-LDL-C, in mg/dL) and assessed associations with ASCVD risk factors. We evaluated associations between discordance and incident ASCVD by sequentially adjusted Cox regression models, and we generated restricted cubic spline plots of discordance and hazard for ASCVD. Results: There were 228 ASCVD events over a median of 12.3 years. Clinical characteristics across discordance quintiles are shown in the Table . After adjustment for traditional ASCVD risk factors, there was a linear association between higher LDL-C discordance and increased risk of ASCVD events ( Figure ) with the highest hazard in Quintile 5 (HR 1.5, 95% CI 1.1 - 2.0). Conclusions: Discordant LDL-C estimates were largely associated with male sex, White and Hispanic races, and characteristics of the metabolic syndrome. Individuals in the highest quintile of discordant LDL-C estimates, with MH-LDL-C > F-LDL-C, had greater risk for incident ASCVD.
Read full abstract