Nuclear Magnetic Resonance (NMR)–Based Serum Metabolic Profiles: Impact of Fasting Status

Abstract

Abstract Introduction Measurement of lipoprotein subclass concentration (-c), particle number (-p), and size (-s) by nuclear magnetic resonance (NMR) has gained traction recently in the clinical laboratory due to associations between smaller lipid particle sizes and atherogenic risk, especially for LDL-p. Since patients may not comply with the clinician’s recommendation of fasting before sample collection, our objective was to evaluate the impact of fasting status on NMR-based lipid and amino acid determinations. Methods Fifty self-reported healthy participants were recruited using an internal review board-approved protocol: males (n = 25; 41.3 ± 10.5 years; BMI 27.7 ± 4.6 kg/m2) and females (n = 25; 35.6 ± 9.1 years; BMI 25.3 ± 5.4 kg/m2). Blood was collected after overnight fast and 4 hours after a high-fat meal. Samples were analyzed using the AXINON lipoFIT by NMR assay (Numares) with an Avance III NMR spectrometer and Ascend 600 MHz magnet (Bruker). This assay included measurements of triglycerides, total cholesterol, IDL-c, and LDL, HDL, VLDL concentrations; particle numbers; and sizes. Additionally, NMR-based glucose, alanine, valine, leucine, and isoleucine concentrations were acquired. Conventional lipid parameters were obtained using cobas 8000 chemistry system (Roche). Fasting and postprandial results were compared for significant differences. Results As expected, mean triglycerides increased after the meal (53%, P < .0001). Increases were also observed for VLDL-c (22%, P < .0001) and large VLDL-p (56%, P < .0001), as well as LDL-s, HDL-s, and VLDL-s (all <3.5%, P < .005). In contrast, LDL-p, small LDL-p, large LDL-p, HDL-p, small HDL-p, glucose, total cholesterol, and LDL, HDL, and IDL cholesterols demonstrated at most an 11% decrease (all P < .002). Valine, leucine, and isoleucine also increased after the meal (31%-46%, P < .0001 for all). Differences in individual lipids or amino acids were not sufficient to distinguish between fasting vs postprandial states, although incorporation of valine into a ratio with VLDL-c and LDL cholesterol (Valine*VLDL-c/LDL cholesterol) provided adequate differentiation (ROC analysis: AUC 0.91 [SD 0.03, P < .0001]). A cutoff of 75 for the ratio had a sensitivity of 88.9% (95% CI, 76.0%-96.3%) and specificity of 85.4% (95% CI, 72.2%-93.9%). Interference with NMR results occurred for three participants’ postprandial samples. These individuals had significantly higher fasting triglycerides, LDL-p, small LDL-p, large VLDL-p, VLDL-s, VLDL-c, IDL-c and lower LDL-s, and HDL-s compared to other participants. Conclusions The clinical impact on results from postprandial samples warrants further evaluation prior to accepting nonfasting samples for NMR analysis. Considering a subpar compliance with fasting, algorithms to differentiate fasting and postprandial specimens may be useful in identifying suboptimal specimens.