LDL-C equation disagreement as a threshold-integrity flag for orthogonal lipid confirmation.

Abstract

Background: Multiple low-density lipoprotein cholesterol (LDL-C) equations can be applied to every standard lipid panel. Disagreement across a treatment threshold may indicate unstable classification.

Objective: To test LDL-C equation disagreement as a calculation-only threshold-integrity flag and as a trigger for orthogonal lipid confirmation.

Methods: We analyzed decontaminated All of Us lipid panels with same-day standard lipids and direct homogeneous LDL-C, recognizing that direct LDL-C is a clinical reference assay rather than beta-quantification. Friedewald, Sampson/National Institutes of Health (NIH), and Martin-Hopkins were evaluated at LDL-C thresholds of 70, 100, and 130 mg/dL; 55 mg/dL was added in the temporal audit. The 3-equation disagreement flag was compared against triglyceride (TG)-based reflex rules. Medical Information Mart for Intensive Care IV (MIMIC-IV) was used for external stress testing.

Results: Equation agreement identified a high-stability majority: in the primary All of Us cohort, equations agreed for 86% to 92% of panels with 92% to 95% threshold accuracy. In the temporal audit, future clean panels showed 3-equation disagreement in 10.2%, 11.4%, 9.2%, and 5.8% of rows at LDL-C thresholds of 55, 70, 100, and 130 mg/dL. Disagreement-subgroup accuracy was lower for all 3 equations. The 3-equation straddle outperformed TG ≥200 mg/dL for detecting apparent at-goal discordance at all 4 thresholds. Sparse apolipoprotein B (ApoB) analyses supported ApoB as an independent witness.

Conclusion: LDL-C equation disagreement is a threshold-integrity flag, not proof that any equation is wrong. When equations disagree near a treatment goal, clinicians should avoid treating the LDL-C threshold result as unqualified and consider orthogonal confirmation, preferably ApoB, when particle burden is the clinical question.