Effects of sample handling and storage on quantitative lipid analysis in human serum

Abstract

There is sparse information about specific storage and handling protocols that minimize analytical error and variability in samples evaluated by targeted metabolomics. Variance components that affect quantitative lipid analysis in a set of human serum samples were determined. The effects of freeze-thaw, extraction state, storage temperature, and freeze-thaw prior to density-based lipoprotein fractionation were quantified. The quantification of high abundance metabolites, representing the biologically relevant lipid species in humans, was highly repeatable (with coefficients of variation as low as 0.01 and 0.02) and largely unaffected by 1-3 freeze-thaw cycles (with 0-8% of metabolites affected in each lipid class). Extraction state had effects on total lipid class amounts, including decreased diacylglycerol and increased phosphatidylethanolamine in thawed compared with frozen samples. The effects of storage temperature over 1 week were minimal, with 0-4% of metabolites affected by storage at 4 degrees C, -20 degrees C, or -80 degrees C in most lipid classes, and 19% of metabolites in diacylglycerol affected by storage at -20 degrees C. Freezing prior to lipoprotein fractionation by density ultracentrifugation decreased HDL free cholesterol by 37% and VLDL free fatty acid by 36%, and increased LDL cholesterol ester by 35% compared with fresh samples. These findings suggest that density-based fractionation should preferably be undertaken in fresh serum samples because up to 37% variability in HDL and LDL cholesterol could result from a single freeze-thaw cycle. Conversely, quantitative lipid analysis within unfractionated serum is minimally affected even with repeated freeze-thaw cycles.

Fig. 1

Coefficients of variation (CV)s for metabolites measured in fresh serum. The CVs were calculated for each metabolite measured (e.g. each fatty acid within each lipid class) in 5 replicates of the same fresh serum sample from Experiment 1. The CVs were plotted in decreasing order for all metabolites. Low abundance metabolites (e.g. DG20:4n3, PE14:1n5) had higher CVs whereas high abundance metabolites (e.g. TG18:1n9, CE16:0) had very low CVs as low as 0.01 and 0.02. This illustrates the high reproducibility of the analytical method for metabolites of biological significance in humans