Lipid concentrations in standard reference material (SRM) 1950: results from an interlaboratory comparison exercise for lipidomics

Abstract

The continued growth of the lipidomics research community, combined with a concomitant increase in the number of lipidomic applications, has culminated in an emerging need for the harmonization and standardization of lipidomics measurement. Harmonization and standardization of lipidomic measurement is a considerable undertaking, owing to the vast structural diversity and complexity of lipids, which also subsequently coincides with the use of a broad range of qualitative and quantitative measurement strategies employed by the lipidomics community. The lipidomics community needs to address the variability present in current lipidomics measurement before harmonization and standardization can begin to occur. Accordingly, this work encompasses the first community-supported harmonization effort via an interlaboratory comparison exercise, focused on ascertaining sources of lipidomic measurement variability and/or agreement, while also highlighting measurement challenges in regards to lipid quantitation.The main objectives of the interlaboratory comparison exercise were to 1) generate consensus estimates in nmol/mL for those lipids routinely measured by participants, 2) determine the extent of agreement present within the community using current quantitation lipidomics workflows, 3) and identify those lipids or lipid classes that require more attention. The basic framework of the National Institute of Standards and Technology (NIST) interlaboratory comparison for lipidomics was to distribute one vial of Standard Reference Material (SRM) 1950 -Metabolites in Frozen Human Plasma to each participating laboratory, and to encourage each participant to employ the analytical methodologies that they typically use to quantify lipids in their laboratory. SRM 1950 was chosen as the vehicle for the comparison exercise as it has been previously recognized and promoted as an appropriate reference material for metabolomics (1)(2)(3)(4)(5). In addition, SRM 1950 was constructed to approximate "normal" blood plasma indicative of the United States population (see http://srm1950.nist.gov/). Invitations were sent to a cohort of laboratories that were representative of the diverse cross-section of lipid measurement methodologies present within the lipidomics community. Consensus estimates (at sum composition level), with corresponding uncertainties, were generated for those lipids measured by at least five laboratories. Additional analyses were performed to further assess the collective submitted data, including coefficient of dispersion (COD) for each consensus estimate and zetascores (ζ-scores). COD values were used to evaluate the quality or "usefulness" of the consensus estimates. ζ-scores were used to determine the relative measurement agreement amongst the consensus estimates by lipid species and lipid class.The final consensus estimates and associated uncertainties generated from this exercise hold considerable potential for the lipidomics community, both to serve as inter-and intralaboratory benchmarks but also to initiate follow-up...

Development, characterization and comparisons of targeted and non-targeted metabolomics methods

“…For lipids measured by QqQ FI , NIST reference values were unavailable. A comparison of lipid data for SRM 1950 revealed that concentrations for up to 33 lipids measured in the present work were within 50–150% of those reported by either Quehenberger et al or Bowden et al [ 29 , 30 ]. For the remaining lipids measured in our study, concentrations were either outside this range or were not reported.…”

“…It is germane to note that some inconsistencies exist between previously published data for lipid concentrations in SRM 1950. For example, Quehenberger et al reported 16.2 μM of SM(d18:0/C18:0) while Bowden et al reported 2.0 μM [ 29 , 30 ]. Clearly, further work is necessary to develop reliable reference concentrations for lipids in SRM 1950.…”

Simultaneous non-polar and polar lipid analysis by on-line combination of HILIC, RP and high resolution MS

“…Lipid concentrations determined in the latter sample type were in the nmol to µmol range for 10 7 cells (Table 1) and compared well to previous studies on Pichia pastoris or other yeasts. 34,[65][66][67][68] Trueness bias of analysis was assessed in the SRM 1950 using the recently introduced accuracy assessment tool LipidQC, 69 which followed the publication of a NIST interlaboratory comparison report on high molecular lipids 70,71 (ESI Fig. S4 †).…”

Harmonized procedures lead to comparable quantification of total oxylipins across laboratories

“…There is only a very small number of interlaboratory comparisons that investigate the comparability of lipidomic approaches (25)(26)(27)(28) and even less studies that explicitly address the precision, reproducibility and comparability of oxylipin analysis using a standardized and harmonized protocol. The few existing studies that address these points therefore show poor comparability (26,38,39).…”

“…While Quehenberger et al quantified the concentrations for these oxylipins at 11.9, 4.22 and 0.8 pmol/mL. Thus, the deviations in the quantified oxylipins 5-, 12-and 15-HETE are respectively 14, 61 and 199%(26,38,43).Although a relative quantification based on a reference standard is often used, this approach is limited and cannot replace a true quantification of concentrations i.e. mol/amount per volume or gram sample.…”