Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
We read with interest the multiomic studies by Kong et al and Chen et al examining gut microbiome–metabolome interactions and potential diagnostic and classification biomarkers in colorectal cancer.1 2 Large-scale, multicentre, multisample, longitudinal studies are imperative to understand complex relationships between the metabolome and digestive diseases.
Analysis of matched blood and stool can advance biomarker development and aid mechanistic exploration, providing samples are collected and stored appropriately. Buffered kits are commercially available for stable transfer of stool samples to storage.3 However, stabilisation of metabolites from blood requires freezing that may be confounded by transport and storage variables including time to whole blood centrifugation, time to freezing and freezing temperature. The gold standard of immediate sample separation and freezing4 5 must be carefully balanced with pragmatic protocols, needed to facilitate standardised, cost-effective collection by busy clinical research facilities across multiple recruiting sites.
In preparation for the CD-metaRESPONSE precision medicine multicentre study (www.ibd-response.co.uk), we undertook this research to assess the impact of sample collection, shipping and storage on circulating metabolites. We collected whole blood in lithium heparin tubes from five non-fasting adult participants. Ten conditions were tested to model varying storage times of whole blood at 4°C from collection to central lab centrifugation, subsequent storage time of plasma at 4°C before freezing and the impact of long-term storage at −20°C or −80°C. We performed metabolomics using both liquid chromatography–mass spectrometry (LC-MS) and nuclear magnetic resonance (NMR) spectrometry to capture a broad range of metabolite classes, including lipids and bile acids by reversed-phase chromatography (RPC) LC-MS method, and small molecule metabolites by NMR.6–8 Raw data were preprocessed and quality assessed, as previously described,9 to generate global profiling datasets and targeted extraction of predefined metabolite panels.8 10 Analyses were perfomed in MetaboAnalyst V.5.0 and R. For metabolite profile analysis, ‘adonis’ was used to generate estimated effect size (R2) and false discovery rate (FDR) adjusted p values (1000 permutations). For single metabolite analysis, p values are based on Kruskal-Wallis with Fishers least significant difference (LSD) post-hoc test.
Analysis was first performed on all 10 conditions to determine if metabolite profiles were driven by subject, storage time of whole blood or plasma at 4°C or final freezing temperature (each experimental condition detailed in figure 1A). LC-MS metabolite profiles clustered by subject (all p=0.002), irrespective of whole blood or plasma storage time or freezing temperature (figure 1B–D). Further, the variation in conditions was not significantly associated with positive or negative mode LC-MS (including bile acids). Analysis of whole blood storage time found all metabolites detected by LC-MS to be stable with the exception of lysophosphatidylserine (18:0/0:0) and lysophosphatidic acid (22:6/0:0) (both p<0.001), which were more abundant at 72 hours (negative mode only). Interindividual differences were maintained up to 72 hours at 4°C. No single LC-MS metabolite was linked to plasma storage time or freezing temperature.
Small molecule NMR metabolite profiles similarly clustered by subject (p<0.001; figure 2A); however also by condition (storage time/temperature) (p=0.001). Further analysis of different whole blood storage times showed lactic acid was significantly higher after 24 and 72 hours at 4°C (p<0.001; figure 2B,C). No other small molecule metabolite was correlated with plasma storage time or freezing temperature. Following bioinformatic removal of lactic acid, the NMR dataset continued to cluster significantly by subject (p=0.002) and was no longer significantly associated with whole blood or plasma storage time or freezing temperature (figure 2D). Bioinformatic removal of lactic acid may be necessary to overcome confounding by sample storage time in large scale studies.
Our data provide reassurance regarding variations in blood sample handling and storage required to allow collection of plasma in the context of pragmatic multicentre large-scale clinical studies. These results form a basis for designing multicentre metabolomic biomarker studies.
Patient consent for publication
This study involves human participants. Human samples used in this research project were obtained from the Imperial College Healthcare Tissue Bank (ICHTB). ICHTB is supported by the National Institute for Health Research (NIHR) Biomedical Research Centre based at Imperial College Healthcare NHS Trust and Imperial College London. ICHTB is approved by Wales REC3 to release human material for research (17/WA/0161). Participants gave informed consent to participate in the study before taking part.
This work is supported by a research grant from The Leona M. and Harry B. Helmsley Charitable Trust (grant #2002-04255). IBD-RESPONSE is supported by the Medical Research Council (grant MR/T032162/1). We are grateful for support from the Newcastle Clinical Trials Unit, the National Phenome Centre, Imperial College London and the NIHR Biomedical Research Centres from Newcastle, Imperial and Cambridge. JLA is the recipient of an NIHR Academic Clinical Lectureship (CL-2019-21-502), funded by Imperial College London and The Joyce and Norman Freed Charitable Trust. NJW is an Academic Clinical Fellow supported by the NIHR. LJD is supported by a Sir Henry Dale Fellowship jointly funded by the Wellcome Trust and the Royal Society (Grant Number 208750/Z/17/Z) and the Kennedy Trust for Rheumatology Research. CJS is supported by a Sir Henry Dale Fellowship jointly funded by the Wellcome Trust and the Royal Society (Grant Number 221745/Z/20/Z) and the 2021 Lister Institute Prize Fellow Award. The views expressed are those of the authors and not necessarily those of our funders, the NIHR or the Department of Health and Social Care. Figure 1A was created with BioRender.com.
CJS and CAL are joint senior authors.
JLA and NJW are joint first authors.
Twitter @wyatt_nic, @carosands, @tariqahmadIBD, @DrNickKennedy, @charlie_lees, @mcintyre_re, @natter5, @IBD_MB, @lukejostins, @NickPowellLab, @gut_health, @CJStewart7, @DrChrisLamb
Contributors CAL, CStewart, JM and NP led study design with input from all authors. Participant recruitment was undertaken by JLA. Wet lab work was undertaken by JLA, SC, EC, BJ, CSands and PT. Analysis was undertaken by JLA, NJW, LJ, CStewart and CAL. Initial manuscript drafting was led by CAL, JLA, NJW and CStewart with subsequent critical review and revision by all authors.
Funding The Leona M. and Harry B. Helmsley Charitable Trust (2002-04255).
Competing interests None declared.
Provenance and peer review Not commissioned; internally peer reviewed.