This paper compared the metabolome of EDTA plasma from blood of fasting and non-fasting healthy donors that was stored at 4, 25, and 37°C for 5-210 min before plasma separation and analysis by untargeted liquid chromatography-mass spectrometry (LC-MS).
Prediction and evaluation of the effect of pre-centrifugation sample management on the measurable untargeted LC-MS plasma metabolome.
Zheng R, Brunius C, Shi L, Zafar H, Paulson L, Landberg R, Naluai ÅT
Purpose
Conclusion
Precentrifugation storage temperature was accurately predicted in plasma specimens based on 137 metabolite features; most of the misclassified specimens were stored for ≤90 min. In the generalized linear models (GLM), 7 of the 13 metabolites affected by storage temperature were not affected by storage duration). The authors report that a lower percentage of metabolites had an intensity drift >10% but more had an intensity drift >30% and 60% when precentrifugation storage was at 4°C compared to storage to 25°C or 37°C. The metabolite features identified as being affected by precentrifugation temperature were narrowed down using the MUVR-RF feature, followed by the rdCF-RF feature, and a validation set, resulting in a list of 14 features, 7 of which were annotated. The classification power was higher for specimens stored at 4°C or 25°C than 37°C. Precentrifugation storage duration was accurately predicted based on 44 metabolites, all of which were influenced by temperature in the GLM; larger shifts were observed when precentrifugation storage was at 37°C than when storage was at 25°C or 4°C. The metabolite features identified as being affected by a precentrifugation delay were narrowed down, resulting in a list of 8 features, 5 of which were annotated. Proper classification of specimens based on storage temperature was not affected by fasting status, but specimen classification based on precentrifugation storage duration was stratified by fasting status.
Studies (1)
Purpose
This paper compared the metabolome of EDTA plasma from blood stored at 4, 25, and 37°C for 5-210 min before plasma separation and analysis by untargeted LC-MS. Blood was collected from 28 healthy volunteers (fasting and nonfasting) into EDTA Vacutainer tubes and gently mixed and stored at 4, 25, and 37°C for 5 (for 4 and 25°C only), 90, 150, and 210 min. To validate the results, EDTA plasma samples were collected from 251 healthy donors with various fasting statuses at two sites and stored at 4°C or 25°C for as little as 1 min and as long as 12 days. Plasma was separated at each time point by centrifugation at 2000 g for 10 min at 4°C or room temperature and stored at -80°C. The metabolome was analyzed on a UHPLC-qTOF-MS system with all specimens from a volunteer included in a single batch. Raw data was converted into mzML format using ProteoWizard Version 3.0 and statistical and bioinformatic analysis was conducted using R software (version 3.4.4).
Summary of Findings
Precentrifugation storage temperature was accurately predicted based on 137 metabolite features (predictive power = 87%, permutation < 2.2e-16); most of the misclassified specimens were stored for ≤90 min (62.5% specimens stored at 4°C for 5 min and 41.5% of specimens stored at 37°C for 90 min were misclassified as being stored at 25°C). Proper classification of specimens based on storage temperature was not affected by fasting status. Precentrifugation storage duration was accurately predicted based on 44 metabolite features (predictive power = 82%, permutation < 2.2e-16), but was stratified by fasting status. The predictive power was higher for specimens stored at 4°C or 25°C than 37°C (82 and 83%, respectively versus 69%). Fasting status was accurately predicted with a predictive power of 73%. Of the 137 metabolite features that were predictive of precentrifugation storage temperature, 7 were unique to temperature and were not affected by precentrifugation storage duration in the generalized linear models (GLM). In contrast, all 44 metabolite features predictive of precentrifugation storage duration in the GLM were influenced by precentrifugation temperature, with the largest shifts based on precentrifugation storage duration observed when storage was at 37°C and, in general, no significant effects of storage at 25°C versus 4°C for 5 min occurred. Levels of linoleyl carnitine, linolenyl carnitine, o-oleoylcarnitine, 2-hydroxycinnamic acid, and lactate were stable when blood was stored at 4°C prior to centrifugation, but increased when blood was stored for 5-210 min at room temperature or 37°C. Levels of taurine, serotonin, and two unknown metabolite features changed more between 5 and 90 min of precentrifugation storage at 4°C than 25°C. The authors report that a lower percentage of metabolites had an intensity drift >10% but more had an intensity drift >30% and 60% when precentrifugation storage was at 4°C than at 25°C or 37°C. A higher percentage of metabolite features had a coefficient of variance >30% in blood specimens stored for up to 210 min prior to centrifugation when stored at 4°C compared to 25°C or 37°C (29.1% versus 20.1% and 14.8%, respectively). When features with similar intensity drift patterns were clustered, there were 30 clusters for the 4°C storage temperature, 50 at the 25°C storage temperature, and 50 at the 37°C storage temperature. The metabolite features identified as being affected by precentrifugation storage temperature were narrowed down by the MUVR-RF feature, followed by the rdCF-RF feature, and a validation set, resulting in a list of 14 features, 7 of which were annotated: lysophosphatidylethanolamine (P-16:0), serotonin, pyruvate, omega-(Hypoxanthin-9-yl)-pentoxycarbonyl-leucyl-methionine, fatty acyl 23:3;O6, lysophosphatidylcholine (lysoPC) 16:0 and sphingosine 1-phosphate. Similarly, the metabolite features identified as being affected by the duration of a precentrifugation delay were narrowed down, resulting in a list of 8 features, 5 of which were annotated: 3-(butyl sulfinyl)-alanine, svinylcysteine, linoleyl carnitine, fatty acyl 22:4;O4 and hypoxanthine.
Biospecimens
FluidBloodFluidPlasmaPreservation Types
FrozenDiagnoses
NormalPlatforms
Small moleculeLC-TOF-MSLipidLC-TOF-MSPre-analytical Factors
Classification Pre-Analytical Factor Value(s) Biospecimen Acquisition Time of biospecimen collection FastingNon-fastingStorage Storage duration 150 min210 min5 min90 minStorage Storage temperature 25°C37°C4°C