Low-Density Lipoprotein Subfractionation: What Is the Role of the Lab When Reporting Discrepant Results?

Abstract

Abstract Plasma concentrations of low-density-lipoprotein cholesterol (LDL-C) are directly associated with risk for coronary artery disease (CAD). Multisociety guidelines define LDL-C>160mg/dL as a risk factor for CAD and LDL-C>190mg/dL as an indication for lipid lowering medication, regardless of other clinical factors. Subfractionation of LDL according to size (LDL-s) enables differentiation between two LDL phenotypes: large-buoyant LDL and small-dense LDL. The small-dense LDL phenotype reportedly conveys increased risk for CAD. Major societies do not recommend LDL subfractions be used for clinical decision making and most payers do not cover LDL subfraction testing. Despite these restrictions, LDL subfraction is routinely requested by clinicians. Nuclear magnetic resonance (NMR) spectroscopy measures LDL-C and LDL-s. Following inquiries regarding interpretation of conflicting LDL-C and LDL-s results, we investigated associations between LDL-C and LDL-s measured by NMR in order to determine how often they provide contradicting or additive information. Verification of NMR LDL-C accuracy was confirmed by ß-quantification in a subset of patient samples (n=250). The average bias was -4.5mg/dL and the correlation coefficient was 0.92. High-risk was defined as LDL-C>160mg/dL or LDL-s<20.5 nm (small-dense LDL); and low-risk was defined as LDL-C<70mg/dL or LDL-s>20.5nm (large-buoyant LDL). In 26,710 clinical NMR analyses, the median LDL-C was 94.0mg/dL (range:5-436mg/dL) with median LDL-s of 20.8 nm (range:19.4–23.0nm). LDL-s moderately correlated with LDL-C (Ï#129;=0.51;p<0.01). Small-dense-LDL was identified in only 18% (407/2,191) of patients with elevated LDL-C (>160mg/dL) and was more common (73.2% of 6,093) in patients with low LDL-C (<70mg/dL;p<0.001). Associations with CAD were investigated among patients without cholesterol-lowering medication treatment referred for angiography (n=356). CAD (defined as stenosis >50% in one or more coronary artery) was diagnosed in 14% (1/7) of subjects with low LDL-C (<70mg/dL) compared to 59% (47/80) of subjects with elevated LDL-C (p=0.01). When stratifying by LDL-s, CAD was diagnosed in 50% (57/115) of subjects with small-dense LDL compared to 43% (104/241) of subjects with large-buoyant LDL (p=0.2). Small-dense LDL was identified in only 33% (26/80) of cases with elevated LDL-C. Limiting to subjects with elevated LDL-C, CAD was diagnosed in 50% (13/26) of subjects with concordant (high-risk) small-dense LDL compared to 61% (33/54) of subjects with discordant (low-risk) large-buoyant LDL (LDL-s>20.5nm) (p=0.3). Our data confirm that LDL-s subfraction measured by NMR is reported discordantly in most cases when LDL-C is unequivocally high or low. Furthermore, CAD diagnosis was significantly associated with LDL-C, but not with LDL-s. Our data also show that in discrepant samples, elevated LDL-C correlates better with disease state compared to LDL-s. Therefore, LDL-s should not be used to justify treatment decisions in patients with elevated LDL-C. Laboratories should consider carefully whether or not to report LDL-s when it is known that misleading and discordant values will be reported in a majority of cases.