Comparison of the accuracy of the Friedewald, Martin, and Sampson formulas to estimate very low levels of low-density lipoprotein cholesterol.

Introduction: The Martin (MF) and Sampson (SF) formulas have shown greater accuracy for low-density lipoprotein cholesterol (LDL-C) < 70 mg/dL compared to the Friedewald formula (FF); however, some disagreement is maintained. Non-high-density lipoprotein cholesterol (non-HDL-C) and apolipoprotein B (ApoB) are alternatives to assessing cardiovascular risk in patients with very low LDL-C. The objective was to evaluate the accuracy of FF, MF, and SF formulas to estimate LDL-C < 70 mg/dL vs. directly measured LDL-C (LDLd-C) and to compare non-HDL-C and Apo-B levels between the groups of patients with concordant vs. discordant LDL-C.

Material and methods: This was a prospective clinical study with measurements of lipid profile and LDLd-C in 214 patients with triglycerides < 400 mg/dL. For each formula, the estimated LDL-C was compared with the LDLd-C, and the correlation, the median difference, and the discordance rate were evaluated. Non-HDL-C and Apo-B levels were compared between the groups with concordant and discordant LDL-C.

Results: The estimated LDL-C was < 70 mg/dL in 130 (60.7%) patients by FF, 109 (50.9%) by MF, and 113 (52.8%) by SF. The strongest correlation was found between LDLd-C and Sampson estimated LDL-C (LDLs-C) (R2 = 0.778), followed by Friedewald-estimated LDL-C (LDLf-C) (R2 = 0.680) and Martin estimated LDL-C (LDLm-C) (R2 = 0.652). Estimated LDL-C < 70 mg/dL was lower than LDLd-C, with the largest median absolute difference (25-75th) of -15 (-19 to -10) with FF. For estimated LDL-C < 70 mg/dL, the discordant rate was 43.8%, 38.1%, and 35.1%, reaching for 62.3%, 50.9%, and 50% when LDL-C < 55 mg/dL by FF, SF, and MF, respectively. Patients in the discordant group presented significantly higher levels of non-HDL-C and ApoB for all 3 formulas (p < 0.001).

Conclusion: FF was the most inaccurate formula to estimate very low LDL-C. Despite MF and SF showing better results, their frequency in underestimating LDL-C was still considerable. In patients with falsely low estimated LDL-C, apoB and non-HDL-C were significantly higher, reflecting its true high atherogenic burden.