Bridging the Gap: A Scoping Review of Pre-Analytical Variability in Biofluid Metabolomics

Abstract

Metabolic profiling enables comprehensive characterisation of the small molecules that are part of the biochemical composition of biological fluids. The most widely profiled biofluids include serum and plasma. Additionally synovial fluid provides a direct reflection of the metabolomic environment of joints and holds promise for biomarker discovery in arthropathies. However, the reproducibility of metabolomics data is highly sensitive to pre-analytical variation, and at the present time, standardised protocols for synovial fluid remain underdeveloped. This review aims to identify and evaluate the existing literature on effects of biofluid pre-analytical handling treatments on metabolic profiles. This review was conducted and reported according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for Scoping Reviews (PRISMA-ScR) guidelines. A search was carried out to identify studies employing LC-MS, GC-MS, and NMR spectroscopy for the investigation of factors including sample collection variables, pre-centrifugation conditions, centrifugation parameters, post-centrifugations conditions, sample storage conditions, and freeze/thaw cycling. Best practice recommendations emerging from this review include the use of additive free serum and heparin plasma tubes, the centrifugation of samples within two hours of collection, immediate storage of samples at −80 °C, and avoidance of repeated freeze/thaw cycling. However, while pre-analytical influences have been extensively characterised for plasma and serum, evidence for synovial fluid remains limited. Overall, the findings highlight the existing recommendations for plasma and serum and demonstrate the need for standardised pre-analytical protocols and validation of quality control markers to advance synovial fluid metabolomics.

1. Introduction

Metabolomics is the comprehensive measurement of all metabolites and low molecular weight molecules (metabolites) that are found in a biological specimen to understand the biochemistry and physiological state [1,2]. As the downstream products following genomic and proteomic processes, the metabolic profile (also referred to as untargeted metabolic phenotyping, metabolomics, or metabotyping) is representative of both endogenous and exogenous factors such as a person’s diet, lifestyle, age, therapeutic interventions, and exposome, as well as the gut microbiome [3,4,5]. This makes metabolomics ideal for the characterisation of complex phenotypes, understanding mechanisms involved in disease pathophysiology, the discovery of biomarkers for disease monitoring, and understanding treatment outcomes [6,7,8,9]. Metabolomic platforms include liquid chromatography or gas chromatography coupled with mass spectrometry (LC-MS or GC-MS), as well as nuclear magnetic resonance (NMR) spectroscopy [10,11], all having different advantages. NMR is not as sensitive as GC and LC-MS platforms, and its applications are generally limited to the identification of highly abundant metabolites (≥1 µM) compared to GC/LC-MS approaches that can be used to profile metabolites at low concentrations (typically in the nano–picomolar range depending on the instrument used and in some targeted applications down to the femtomolar range) and offer a higher resolution (∼10³–10⁴), allowing for the profiling of low abundance metabolites including small metabolites (<650 Daltons) [12,13]. However, NMR is considered a non-destructive technique, generally requiring less sample preparation prior to analysis, therefore preserving metabolite structure and more readily enabling accurate metabolite quantification [14]. GC and LC-MS approaches can be applied for the analysis of metabolites present at low abundance. GC-MS has been shown to perform better when profiling volatile and thermally stable metabolites, whilst LC-MS outperforms GC-MS in the profiling of lipophilic and higher molecular weight metabolites [15,16,17].

Some of the most widely metabolomic profiled samples are bloods, including serum and plasma [18,19], some of which are large scale cohort studies [20,21]. Metabolomic profiling of blood samples is valuable for understanding the systemic responses and contributors of a disease status/phenotype [22]. For biomarker analysis, the detection of markers, which can be assessed via a simple blood test, is optimal for clinical utility [23,24,25]. However, for certain diseases, analysis of tissue-specific biofluids can be more informative of the disease status [26]. In recent years, the metabolomics of synovial fluid has grown in popularity, particularly for the diagnosis and study of disease mechanisms associated with joint pathologies such as osteoarthritis (OA), rheumatoid arthritis (RA), posttraumatic arthritis, and joint infections, as it is a locally derived biofluid that directly reflects the biochemical environment of the joint space [27,28,29,30,31,32,33]. Unlike blood-based biofluids, synovial fluid composition is influenced by resident and infiltrating cells, cartilage and bone turnover, inflammatory mediators, and mechanical loading [34,35,36]. Across biofluids, a significant challenge in metabolomics studies lies in the preservation of the authentic in vivo metabolite profile. This is because endogenous enzymatic activity continues following sample collection, and subsequent pre-analytical handling steps may further contribute towards changes that distort the metabolite levels [37]. As an extreme example, metabolic flux rates in central carbon metabolism are very fast: within glycolysis in trypanosomes, glucose is transformed into pyruvate in under a second [38]. Clearly biofluids cannot be analysed in these timescales; therefore, one must be pragmatic particularly if collecting samples in a clinical setting.

Pre-analytical factors such as the collection of the sample (e.g., time of collection, blood tubes used for collection, etc.), transport conditions, storage conditions, and sample handling can all impact the stability of the metabolome, therefore introducing variability and making the data collected less reproducible [39,40]. As a result, it is crucial to understand how delays in sample processing and the conditions these samples are exposed to impact the metabolome and our interpretation of the data. Additionally, synovial fluid also presents distinct pre-analytical challenges including low sample volumes (median total knee volume = 3.05 mL), high viscosity due to the presence of hyaluronic acid (3–4 mg/mL), variable dilution of the biofluid due to sample collection procedures, and limited opportunities for rapid processing in clinical settings [41,42,43]. These characteristics mean that pre-analytical factors established for blood-based metabolomics may not be directly transferrable to synovial fluid, underscoring the need for a focused evaluation of pre-analytical handling practices specific to synovial fluid metabolomics. The aim of this review is to identify and assess the current literature on the effects of pre-handling treatments on metabolic profiling of biofluids (whole blood, plasma, serum, and synovial fluid). By mapping the scope and depth of existing research, the review will highlight knowledge gaps and provide biofluid sample handling recommendations for untargeted metabolomics.

2. Materials and Methods

This review was conducted and reported according to the PRISMA-ScR statement and guidance [44]. The PRISMA-ScR checklist is presented in Supplementary Table S1. Searches were performed using a Boolean search string in online databases PubMed and Web of Science (Table S2). In addition to database searches, a snowballing approach was used to identify additional relevant studies. Reference lists of included papers were screened manually to capture studies missed in the initial search. These studies were screened against the same inclusion and exclusion criteria as the database identified records.

2.1. Inclusion and Exclusion Criteria

Only studies that tested the effects of pre-analytical processing using untargeted metabolomic profiling approaches (LC-MS, GC-MS, or NMR) of selected biofluids were included. Both human and animal studies were included. Only studies analysing synovial fluid, whole blood, serum, and plasma samples were included. Non-English language and studies where the full text was unavailable were excluded, and review/abstract only papers were excluded.

2.2. Data Extraction and Analysis

Full text reviewing of all papers meeting the inclusion criteria and data extraction was carried out by two independent researchers (Y.L and S.S). The data extracted included the type of biofluid analysed, source of the biofluid, pre-analytical factors tested, technique used to analyse samples, and the outcomes. Disagreements between the reviewers in the screening were resolved through discussion with a third reviewer (C.H.H). The data were summarised by describing the variables tested in the study and results concerning the impact on the metabolome in a table format.

3. Results

3.1. Study Characteristics

The PRISMA flowchart (Figure 1) outlines the study selection process. The initial search identified a total of 1042 studies. Following the removal of duplicate records (n = 174), the remaining records were screened by title and abstract. During title and abstract screening, records were excluded if they were not related to the metabolomic analysis of relevant biofluids (plasma, serum, or synovial fluid), or if they did not apply NMR, LC, or GC-MS analytical platforms. Full text reports were sought for retrieval for the remaining records (n = 135). These records were assessed for eligibility, and studies were excluded at this stage if they did not meet the inclusion criteria including the use of untargeted metabolomic profiling approaches and the experimental evaluation of pre-analytical variables (n = 89). A total of 46 studies were identified to meet the inclusion criteria, with an additional study identified via snowballing, resulting in a total of 47 studies included in the review. The majority of studies analysed blood-derived biofluids including plasma and serum, while a smaller subset focused specifically on synovial fluid. Analytical platforms employed across studies included NMR, GC-MS, and LC-MS, either individually or in combination. Studies included were published between the years 2000 and 2025.

3.2. Sample Collection Variables

Twelve papers (Table 1) explored the impact that blood collection tubes and anticoagulants have on the metabolomic profile.

In NMR studies, all common tube types used in blood collection (EDTA, citrate, heparin, and serum) consistently introduced intense background signals; however, the region of the signal varied depending on the anticoagulant [45]. Serum and heparin tubes (plasma) have been associated with demonstrating increased concentrations of lactate when analysed using NMR, whereas K2-EDTA and K3-EDTA (plasma) tubes cause increases in acetate and pyruvate levels, respectively [46]. Heparin (plasma) tubes generated broad overlapping peaks that are more difficult to distinguish from endogenous signals, compared to the characteristic peaks generated by EDTA tubes, which makes them easier to exclude [47]. Citrate and ACD (plasma) tubes have been found to typically present with intense signals overlapping with endogenous metabolites including citrate and glucose signals, therefore making it difficult to exclude these signals from analysis [48], which is important when there are differences within central carbon metabolite pathways and, in particular, glycolysis and TCA.

In MS based studies, the use of anticoagulants directly impacts the ionisation efficiency and causes matrix effects. This has been observed in relation to citrate and EDTA plasma tubes, as they form sodium and potassium formate clusters and can additionally cause ion suppression or the enhancement of metabolites co-eluting with citrate and EDTA peaks [49]. However, EDTA is a chelating agent; therefore, it can bind to interfering Ca²⁺ and Mg²⁺ ions, which is useful when analysing carbohydrates using electrospray ionisation (ESI) MS, as it removes metal adducts and enhances carbohydrate peaks [50,51]. The use of heparin plasma tubes, on the other hand, has been consistently identified to be associated with minimal contamination and matrix interference when samples are analysed using LC-MS; therefore, they are recommended over citrate and EDTA tubes [52,53,54]. With regards to GC-MS analysis, of the two papers identified to apply this analytical technique, the results suggest that, like with the NMR and LC-MS results, the peaks generated by EDTA are easily identifiable and can be excluded from analysis, but the peaks generated by the presence of sodium heparin overlap with endogenous metabolites [55].

Serum tubes when analysed using LC-MS have been found to exhibit increased concentrations of key endogenous metabolites including hypoxanthine, sphingosine-1-phosphate, linolenyl carnitine, and xanthine, where they were found to increase their concentrations. These metabolites have been proposed as biomarkers of coagulation and as indicators of delayed sample processing when serum is stored at room temperature, further highlighting the need for careful sample handling when interpreting biologically relevant changes of these specific metabolites, so as to distinguish them from artefactual concentration shifts [54,56]. GC-MS analysis of serum samples revealed that the presence of polymeric gels in serum collection tubes primarily impact metabolites involved in amino acid metabolism (alanine, proline and threonine), glycerolipid metabolism (monopalmitin and monostearin), and those involved in primary pathways including aconitic acid and lactic acid [57].

Together, these results suggest that the type of anticoagulant used in blood tubes was identified as a common source of variation in metabolomic data across both NMR and MS platforms. Acting as a source of exogenous compounds, anticoagulants can contaminate the sample causing matrix effects that can mask or alter endogenous metabolite signals, therefore potentially skewing the interpretation of the results. Notably, no studies were identified that directly investigated the impact of sample collection variables such as collection containers on synovial fluid metabolomic profiles. This contrasts with the extensive body of evidence available for blood-derived biofluids, therefore highlighting a clear gap in the literature underscoring the need to investigate how sample collection practices may influence synovial fluid metabolomics.

Table 1. Summary of studies investigating the impact of blood collection tube types and anticoagulants on metabolomic profiles. Results highlight differences introduced by tube additives and materials across serum and plasma samples using both NMR and MS platforms.

Study	Sample Type	Variable/s Tested	Sample Size (n)	Analytical Method Used	Conclusions
Yin et al. 2013 [56]	Plasma and Serum	Blood collection tube types/Anticoagulants, Haemolysis, Storage temperature and time	3 samples per anticoagulant type	Mass Spec (LC-MS)	Anticoagulants tested include lithium–heparinate, sodium fluoride, sodium citrate, potassium EDTA, and serum tubes. Lithium heparin and serum blood collection tubes interfere with metabolomic profiling. Hypoxanthine, sphingosine-1-phosphate, and linolenyl carnitine are identified as pre-analytical biomarkers of prolonged exposure of blood to room temperature. They can allow for the identification of systematic inaccuracies that arise during processing of whole blood that affect the quality of many samples and introduce random errors leading to outliers. Use of haemolysed samples should be avoided for metabolomic studies. Blood should be stored immediately in ice water after collection until further processing (for a fixed time; ideally not longer than 2 h).
Khadka et al. 2019 [58]	Plasma	Blood collection tube types/Anticoagulants	70 (split between anticoagulants not specified)	Mass Spec (LC-MS)	Anticoagulants tested include EDTA-K2 and sodium citrate. EDTA-K2 and citrate anticoagulants influence both lipidomic and metabolomic profiles. Lipids and amino acids are more abundant in EDTA tubes compared to citrate tubes. Samples are recommended to be collected in EDTA tubes.
Hebels et al. 2013 [53]	Plasma	Blood collection tube types/Anticoagulants	4 samples per anticoagulant type	Mass Spec (LC-MS)	Anticoagulants tested include citrate, EDTA, or heparin tubes. Anticoagulants strongly influence metabolite concentrations. Irrespective of the anticoagulant used, however, the main source of variation is from the donors and delayed processing time (up to 24 h, samples kept at room temperature). Overall heparin tubes are recommended over EDTA or citrate.
Zhou et al. 2017 [52]	Plasma and Serum (animal)	Blood collection tube types/Anticoagulants	12 samples per anticoagulant type	Mass Spec (LC-MS)	Anticoagulants tested include EDTA-K2, heparin lithium, heparin sodium, sodium citrate, and potassium oxalate for plasma samples and serum tube containing separation gel. Anticoagulants influence the MS detection of metabolites by introducing exogenous metabolites such as citric acid and edetic acid from serum collection tubes that can mask the detection of endogenous metabolite peaks. The use of separation gels can introduce intra-group variation, which can sway biomarker identification. The use of heparin plasma tubes is recommended, as they demonstrate wider metabolite coverage with the least influence from exogenous metabolite peaks.
Barri et al. 2013 [54]	Plasma and Serum	Blood collection tube types/Anticoagulants	21 samples per anticoagulant	Mass Spec (LC-MS)	Anticoagulants tested include EDTA-K2, lithium heparin, and serum tubes. Anticoagulants act as a source of exogenous contamination and impact the composition of both serum and plasma samples. Increased concentrations of hypoxanthine and xanthine can act as markers of coagulation in serum samples. Heparin plasma tubes are recommended for LC-MS analysis, as they demonstrated the least amount of contamination and matrix effect compared to citrate and EDTA plasma tubes. Serum samples are also recommended over plasma so long as all samples are collected and analysed following the same procedure.
Bando et al. 2010 [55]	Plasma (animal)	Blood collection tube types/Anticoagulants	8 samples per anticoagulant type	Mass Spec (GC-MS)	Anticoagulants tested include EDTA-K2 and sodium heparin. EDTA tubes are recommended over sodium heparin tubes for plasma collection, as the EDTA peak can easily be identified and excluded from analysis, whereas the sodium heparin-associated peaks are harder to identify, due to their overlap with endogenous metabolites.
López-Bascón et al. 2016 [57]	Plasma and Serum	Blood collection tube types/Anticoagulants	13 samples per anticoagulant type	Mass Spec (GC-MS)	Anticoagulants tested include plastic serum tubes with spray-coated silica (serum), spray-coated silica and a polymer gel (serum–gel), spray-coated silica tubes with heparin (plasma), and heparin tubes with polymer gel (plasma–gel). The use of polymeric gels impacts amino acid and glycerolipid metabolism.
Ghini et al. 2022 [45]	Plasma and Serum	Blood collection tube types/Anticoagulants	210 EDTA-plasma samples 24 citrate-plasma samples 30 serum samples	NMR	Both serum and plasma collection tubes and their anticoagulants introduce intense signals in NMR spectra.
Loo et al. 2020 [46]	Plasma and Serum	Blood collection tube types/Anticoagulants	5 samples per anticoagulant	NMR	Anticoagulants tested include SST II Advance gel and Plain clot (serum), Plasma lithium heparin (LH and LH PST II gel), and EDTA-K2 and K3. Type of blood collection tube used introduces intense signals associated with each tube including higher lactate (heparin and serum tubes), pyruvate (K3-EDTA tube), and acetate (K2-EDTA tube) levels.
Sotelo-Orozco et al. 2021 [48]	Plasma and Serum	Blood collection tube types/Anticoagulants	8 samples per anticoagulant	NMR	Anticoagulants tested include serum, ACD, sodium citrate, EDTA, sodium fluoride, and sodium heparin tubes. Heparin and EDTA tubes are comparable to serum tubes for NMR-based metabolomics. ACD and citrate plasma tubes introduce interfering peaks from citrate and glucose. Differences between subjects are preserved regardless of tube type used.
Xiong et al. 2024 [59]	Plasma and Serum	Blood collection tube types/Anticoagulants	12 samples per anticoagulant type	NMR	Anticoagulants tested include lithium heparin plasma, fluoride/oxalate plasma, or serum tubes. Metabolite profiles of serum and lithium heparin plasma are comparable and behave similarly in response to delayed erythrocyte separation.
Pinto et al. 2014 [47]	Plasma	Blood collection tube types/Anticoagulants	5 samples per anticoagulant	NMR	Anticoagulants tested include sodium heparin and EDTA tubes. Both heparin and EDTA tubes introduce interfering peaks that can overlap with sample peaks. Though EDTA tubes present more interfering peaks compared to heparin tubes, both anticoagulants are recommended for NMR analysis of the plasma metabolome, as the level of interfering peaks is lower than the peaks of interest.

Abbreviations: ACD: acid citrate dextrose; EDTA: ethylenediaminetetraacetic acid; GC-MS: gas chromatography–mass spectrometry; LC-MS: liquid chromatography–mass spectrometry; NMR: nuclear magnetic resonance spectroscopy.

Table 1. Summary of studies investigating the impact of blood collection tube types and anticoagulants on metabolomic profiles. Results highlight differences introduced by tube additives and materials across serum and plasma samples using both NMR and MS platforms.

Study	Sample Type	Variable/s Tested	Sample Size (n)	Analytical Method Used	Conclusions
Yin et al. 2013 [56]	Plasma and Serum	Blood collection tube types/Anticoagulants, Haemolysis, Storage temperature and time	3 samples per anticoagulant type	Mass Spec (LC-MS)	Anticoagulants tested include lithium–heparinate, sodium fluoride, sodium citrate, potassium EDTA, and serum tubes. Lithium heparin and serum blood collection tubes interfere with metabolomic profiling. Hypoxanthine, sphingosine-1-phosphate, and linolenyl carnitine are identified as pre-analytical biomarkers of prolonged exposure of blood to room temperature. They can allow for the identification of systematic inaccuracies that arise during processing of whole blood that affect the quality of many samples and introduce random errors leading to outliers. Use of haemolysed samples should be avoided for metabolomic studies. Blood should be stored immediately in ice water after collection until further processing (for a fixed time; ideally not longer than 2 h).
Khadka et al. 2019 [58]	Plasma	Blood collection tube types/Anticoagulants	70 (split between anticoagulants not specified)	Mass Spec (LC-MS)	Anticoagulants tested include EDTA-K2 and sodium citrate. EDTA-K2 and citrate anticoagulants influence both lipidomic and metabolomic profiles. Lipids and amino acids are more abundant in EDTA tubes compared to citrate tubes. Samples are recommended to be collected in EDTA tubes.
Hebels et al. 2013 [53]	Plasma	Blood collection tube types/Anticoagulants	4 samples per anticoagulant type	Mass Spec (LC-MS)	Anticoagulants tested include citrate, EDTA, or heparin tubes. Anticoagulants strongly influence metabolite concentrations. Irrespective of the anticoagulant used, however, the main source of variation is from the donors and delayed processing time (up to 24 h, samples kept at room temperature). Overall heparin tubes are recommended over EDTA or citrate.
Zhou et al. 2017 [52]	Plasma and Serum (animal)	Blood collection tube types/Anticoagulants	12 samples per anticoagulant type	Mass Spec (LC-MS)	Anticoagulants tested include EDTA-K2, heparin lithium, heparin sodium, sodium citrate, and potassium oxalate for plasma samples and serum tube containing separation gel. Anticoagulants influence the MS detection of metabolites by introducing exogenous metabolites such as citric acid and edetic acid from serum collection tubes that can mask the detection of endogenous metabolite peaks. The use of separation gels can introduce intra-group variation, which can sway biomarker identification. The use of heparin plasma tubes is recommended, as they demonstrate wider metabolite coverage with the least influence from exogenous metabolite peaks.
Barri et al. 2013 [54]	Plasma and Serum	Blood collection tube types/Anticoagulants	21 samples per anticoagulant	Mass Spec (LC-MS)	Anticoagulants tested include EDTA-K2, lithium heparin, and serum tubes. Anticoagulants act as a source of exogenous contamination and impact the composition of both serum and plasma samples. Increased concentrations of hypoxanthine and xanthine can act as markers of coagulation in serum samples. Heparin plasma tubes are recommended for LC-MS analysis, as they demonstrated the least amount of contamination and matrix effect compared to citrate and EDTA plasma tubes. Serum samples are also recommended over plasma so long as all samples are collected and analysed following the same procedure.
Bando et al. 2010 [55]	Plasma (animal)	Blood collection tube types/Anticoagulants	8 samples per anticoagulant type	Mass Spec (GC-MS)	Anticoagulants tested include EDTA-K2 and sodium heparin. EDTA tubes are recommended over sodium heparin tubes for plasma collection, as the EDTA peak can easily be identified and excluded from analysis, whereas the sodium heparin-associated peaks are harder to identify, due to their overlap with endogenous metabolites.
López-Bascón et al. 2016 [57]	Plasma and Serum	Blood collection tube types/Anticoagulants	13 samples per anticoagulant type	Mass Spec (GC-MS)	Anticoagulants tested include plastic serum tubes with spray-coated silica (serum), spray-coated silica and a polymer gel (serum–gel), spray-coated silica tubes with heparin (plasma), and heparin tubes with polymer gel (plasma–gel). The use of polymeric gels impacts amino acid and glycerolipid metabolism.
Ghini et al. 2022 [45]	Plasma and Serum	Blood collection tube types/Anticoagulants	210 EDTA-plasma samples 24 citrate-plasma samples 30 serum samples	NMR	Both serum and plasma collection tubes and their anticoagulants introduce intense signals in NMR spectra.
Loo et al. 2020 [46]	Plasma and Serum	Blood collection tube types/Anticoagulants	5 samples per anticoagulant	NMR	Anticoagulants tested include SST II Advance gel and Plain clot (serum), Plasma lithium heparin (LH and LH PST II gel), and EDTA-K2 and K3. Type of blood collection tube used introduces intense signals associated with each tube including higher lactate (heparin and serum tubes), pyruvate (K3-EDTA tube), and acetate (K2-EDTA tube) levels.
Sotelo-Orozco et al. 2021 [48]	Plasma and Serum	Blood collection tube types/Anticoagulants	8 samples per anticoagulant	NMR	Anticoagulants tested include serum, ACD, sodium citrate, EDTA, sodium fluoride, and sodium heparin tubes. Heparin and EDTA tubes are comparable to serum tubes for NMR-based metabolomics. ACD and citrate plasma tubes introduce interfering peaks from citrate and glucose. Differences between subjects are preserved regardless of tube type used.
Xiong et al. 2024 [59]	Plasma and Serum	Blood collection tube types/Anticoagulants	12 samples per anticoagulant type	NMR	Anticoagulants tested include lithium heparin plasma, fluoride/oxalate plasma, or serum tubes. Metabolite profiles of serum and lithium heparin plasma are comparable and behave similarly in response to delayed erythrocyte separation.
Pinto et al. 2014 [47]	Plasma	Blood collection tube types/Anticoagulants	5 samples per anticoagulant	NMR	Anticoagulants tested include sodium heparin and EDTA tubes. Both heparin and EDTA tubes introduce interfering peaks that can overlap with sample peaks. Though EDTA tubes present more interfering peaks compared to heparin tubes, both anticoagulants are recommended for NMR analysis of the plasma metabolome, as the level of interfering peaks is lower than the peaks of interest.

Abbreviations: ACD: acid citrate dextrose; EDTA: ethylenediaminetetraacetic acid; GC-MS: gas chromatography–mass spectrometry; LC-MS: liquid chromatography–mass spectrometry; NMR: nuclear magnetic resonance spectroscopy.

3.3. Pre-Centrifugation Conditions

Pre-centrifugation conditions including temperature, time delay, light exposure, and heat treatment were frequently explored as sources of pre-analytical variation in metabolomics workflows. Twenty-one papers (Table 2) were identified, which explored these factors, with 12 studies applying NMR [45,46,59,60,61,62,63,64,65,66,67,68], five applying LC-MS only [58,69,70,71,72], two applying GC-MS only [73,74], and two applying both LC-MS and GC-MS [40,75]. Delayed processing time and storage temperature prior to centrifugation were two commonly explored themes within this category.

NMR-based studies [45,46,59,60,61,62,63,64,65,66,67,68] agreed that the delayed processing time significantly impacted the metabolome in both plasma and serum samples; however, the degree of impact was time- and anticoagulant-dependent. One study found that plasma samples collected using tubes containing fluoride/oxalate additives maintained metabolite stability when stored at 4 °C for 24 h [59]. Studies have identified common metabolites that can act as potential biomarkers of delayed sample processing including glucose and lactate or the ratio between glucose and lactate in both plasma and serum samples [45,62,65,67,68]. Prolonged pre-processing time (>4 h) was found to result in increased lactate and decreased glucose concentrations particularly in serum samples. Ornithine was also identified in plasma samples as a potential marker of delayed processing time [61]. However, these results can also be influenced by the temperature the samples are stored at, with studies suggesting an interplay between storage temperature and duration. Generally, studies were in agreement that the storage of plasma and serum samples at either 4 °C or at room temperature (20–25 °C) for up to 48 h significantly influences individual metabolite abundances as opposed to alterations in the global multivariate structure of the metabolome; however, the degree of impact is more pronounced in samples stored at room temperature therefore the recommendation would be to limit the time between plasma or serum collection and centrifugation and store samples at 4 °C during this time [46,61,62,63,65,66]. Ornithine, pyruvate, and lactic acid were specifically identified as temperature sensitive markers with altered levels observed when samples were stored at 22 °C but were more stable at 4 °C [61,65]. One study found that the storage of samples at 4 °C can lead to decreased levels of short-chain fatty acids (SCFA); however, this effect could be reversed via the warming of the sample to 22 °C prior to centrifugation [61]. This study also evaluated the impact of light exposure during the pre-centrifugation delay period in plasma samples; however, no significant changes were observed in the metabolome of light-exposed samples compared to those shielded from the light, suggesting minimal interference from light exposure [61]. One study assessed the effect of heat activation (56 °C for 30 min) on plasma samples, which caused significant alterations in metabolite profiles including a significant decrease in α-1-acid-glycoprotein and increases in triglycerides, lactate, alanine, and glycerol; therefore, it should be avoided [46]. These variables should also be tested for serum and synovial fluid samples.

In MS-based metabolomic analysis (GC and LC-MS) [40,58,69,70,71,72,73,74,75], pre-centrifugation conditions were found to significantly impact both plasma and serum samples; however, these factors primarily affected plasma samples. Delays in sample processing were found to significantly impact the metabolome. In plasma, studies demonstrated that delays of up to 4 h had minimal effects on lipid classes including phosphatidylinositol, sphingomyelin (SM), and hexosylceramides, but prolonged delays of 20–48 h significantly impacted metabolites involved in glycolysis/gluconeogenesis, the TCA cycle, γ-glutamyl amino acid metabolism, and the glutathione pathway [58,69,70,71,72]. Key metabolites identified in studies as markers of processing delays in plasma samples included glucose, lactate, pyruvate, fumarate, 5-oxoproline, and the ratio between arginine and ornithine [70,72]. The storage temperature prior to centrifugation was also found to significantly influence plasma metabolome stability, with studies identifying temperature sensitive metabolites. At room temperature (20–25 °C), plasma samples presented with significant changes in lactic acid, ascorbic acid, sucrose, pyruvic acid, and hypoxanthine levels [73,75]. One study also found elevated lipid levels associated with EDTA plasma when stored at room temperature, suggesting the storage temperature can influence lipid metabolism in plasma [58]. Based on these findings, pre-centrifugation delays should be minimised to under 3 h with storage at 4 °C for the metabolomic analysis of plasma samples when applying MS-based analytical techniques. In serum samples, pre-centrifugation delays of up to 24 h were found to have minimal impact on the metabolome when samples were stored at 4 °C with samples demonstrating no detectable variability in metabolite concentrations compared to those frozen immediately following collection [74]. Increasing the clotting time of serum samples (up to 2 h) was found to significantly impact the metabolome particularly impacting amino acids and peptides possibly associated with the breakdown of red blood cells (RBCs) or proteolysis [40]. These findings suggest that the serum clotting time should be standardised and minimised with samples then being stored at 4 °C for up to 24 h or frozen immediately to maintain metabolite integrity.

Despite multiple studies evaluating the impact of pre-centrifugation conditions on plasma and serum metabolomic profiles, no studies were identified that directly assessed these effects in synovial fluid. This highlights a clear gap in the existing literature, particularly considering the distinct physiochemical properties of synovial fluid, which may limit the direct transferability of findings from blood-derived biofluids.

Table 2. Overview of studies assessing the effect of pre-centrifugation factors on the metabolome stability of plasma and serum samples using NMR- and MS-based analytical techniques.

Study	Sample Type	Variable/s Tested	Sample Size (n)	Analytical Method Used	Conclusions
Niemuth et al. 2015 [60]	Plasma and Whole Blood (animal)	Delayed processing time	5 whole blood samples 20 plasma samples	NMR	Clear differences between whole blood and plasma observed, but changes associated with plasma processing delays would only affect the interpretation of a few metabolites at the level of clinical relevance including glutathione and taurine.
Ghini et al. 2022 [45]	Plasma and Serum	Delayed processing time and storage temperature	234 plasma and 150 serum samples	NMR	Delayed blood processing significantly alters lactate and glucose levels in plasma, suggesting delayed blood processing negatively affects plasma quality.
Malmodin et al. 2024 [61]	Plasma	Pre-centrifugation storage temperature, delayed processing time, and exposure to light	951	NMR	SCFA, pyruvate, glucose, lactate, and ornithine demonstrate distinct incubation temperature-dependent characteristics. Decrease in SCFA at 4 °C was found to be reversible through the increase in temperature to 22 °C prior to centrifugation. Light exposure has minimal effect. Plasma sampling should be conducted immediately at room temperature or stored at 4 °C followed by warming to room temperature prior to centrifugation.
Bernini et al. 2011 [62]	Plasma and Serum	Delayed processing time and temperature	10 plasma and 10 serum samples	NMR	The metabolite degradation processes are time-dependent and temperature-dependent for both plasma and serum: incubation at 25 °C causes more profound changes in the NMR profile. Notably, a decrease in glucose concentrations was observed in both sample types and an increase in lactate. Time between sample collection and processing should not exceed 2 h with samples being stored at 4 °C.
Brunius et al. 2017 [63]	Plasma	Delayed processing time and temperature	111	NMR	Pre-centrifugation delays and storage temperature significantly impacted metabolite profiles with most metabolite drift patterns being highly reproducible between individuals. Alterations were more pronounced at 22 °C compared to 4 °C with associations to specific groups of metabolites; therefore, pre-centrifugation storage at 22 °C should be avoided. Pyruvate, lactate, and ornithine were identified as markers for discriminating between pre-centrifugation temperatures. Pre-centrifugation delays of up to 36 h at 4 °C have minimal effects on overall metabolite changes.
Debik et al. 2022 [64]	Plasma and Serum	Delayed processing time	20 plasma and 20 serum samples	NMR	Centrifugation delay significantly impacts metabolite profiles, especially those associated with anaerobic glycolysis. Plasma samples were more robust against centrifugation delays compared to serum samples.
Santos Ferreira et al. 2019 [65]	Plasma and Serum	Delayed processing time, pre-centrifugation storage temperature	37 plasma and 37 serum samples	NMR	Overall, delays of up to 48 h and storage at either 4 °C or 21 °C prior to centrifugation had minimal effects on metabolite concentrations. Individual metabolites identified to be sensitive to processing delays and storage temperature included pyruvate, with mean concentrations decreasing when stored for 24 h at 4 °C but increasing when stored for 24 h at 21 °C compared to control.
Jobard et al. 2016 [66]	Plasma and Serum	Delayed processing time, pre-centrifugation storage temperature	189 plasma and 192 serum samples	NMR	Storage of samples at 4 °C following collection slowed down the metabolome degradation process in both plasma and serum samples compared to room temperature storage for up to 6 h prior to centrifugation. Increased serum clot contact time at room temperature significantly altered the metabolome of serum samples.
Loo et al. 2020 [46]	Plasma and Serum	Pre-centrifugation storage temperature (serum) and sample heating (plasma)	15 serum and 15 plasma samples	NMR	Storage of serum samples at 20 °C for up to 48 h prior to processing impacted individual metabolite levels (glucose decreased and lactate increased). Heating of plasma samples at 56 °C for 30 min prior to processing caused major systematic changes in metabolite profiles increasing levels of triglycerides, lactate, alanine, and glycerol post heating and decreased relative concentrations of α-1-acid-glycoprotein; therefore, it should be avoided.
Xiong et al. 2024 [59]	Plasma and Serum	Delayed processing time	168 plasma and 84 serum samples	NMR	The storage of samples at 4 °C for 24 h with a fluoride/oxalate additive resulted in the least impact on metabolite profiles; however, the overall impact of processing delays was influenced by the blood tube/anticoagulant used.
Altmann et al. 2025 [67]	Plasma and Serum	Delayed processing time	60 plasma and 30 serum samples	NMR	The storage of samples at room temperature for <8 h prior to centrifugation affects plasma and serum samples differently, with metabolites demonstrating differences in stability depending on the blood collection tube type used. Lactic acid and the ratio between lactic acid and glucose were found to be the least stable with a stability time of 24–45 min and 21–32 min, respectively; therefore, they can be potentially used as markers of sample quality.
Fliniaux et al. 2011 [68]	Serum	Delayed processing time and storage temperature	42	NMR	Prolonged storage (>4 h) of blood at room temperature can be characterised by increased concentrations of lactate and decreased concentrations of glucose in serum samples.
Trezzi et al. 2016 [73]	Plasma	Pre-centrifugation storage temperature and delayed processing time	18	Mass Spec (GC-MS)	Pre-centrifugation delays had a minor impact on metabolite concentration. Temperature at which samples are stored between time of collection and centrifugation is a key factor that can influence metabolite concentrations—recommend storing samples on ice following collection. Lactic acid and ascorbic acid identified to be temperature-sensitive metabolites. Samples of high pre-centrifugation quality identified as those that were stored either on ice for ≥3 h or at room temp for <3 h.
Dunn et al. 2008 [74]	Serum	Pre-centrifugation storage time	40	Mass Spec (GC-MS)	Changes to the serum metabolome caused by storage at 4 °C for up to 24 h were minimal with no detectable variability in comparison to samples frozen immediately at −80 °C post collection.
Khadka et al. 2019 [58]	Plasma	Delayed processing time and temperature	70	Mass Spec (LC-MS)	Storage for up to 24 h at room temperature or at 4 °C did not significantly impact the metabolome in both plasma and serum samples; however, more lipids (e.g., lysophosphatidylcholines, phosphatidylethanolamines) were elevated in EDTA tubes when stored at room temperature compared to 4 °C. Sample processing recommended to be completed within 24 h post collection at 4 °C.
Wang et al. 2023 [69]	Plasma	Delayed processing time, pre-centrifugation storage temperature	829	Mass Spec (LC-MS)	Three lipid classes (phosphatidylinositol, hexosylceramide and SM) identified to be stable against delayed processing time (≤4 h) and storage temperature (4 °C, 21 °C or 30 °C). The metabolome in samples cooled immediately to 4 °C following collection and storage for up to 24 h demonstrated the most stability in lipid profiles.
Wang et al. 2018 [70]	Plasma	Delayed processing time	72	Mass Spec (LC-MS)	Glucose, lactate, pyruvate, and 5-oxoproline concentrations were significantly altered because of processing delays (up to 48 h, stored at 4 °C). Prolonged processing delays resulted in poor reproducibility and impacted concentrations of metabolites involved in glycolysis, TCA cycle, glutathione metabolism, γ-glutamyl amino acid, and several fatty acid subtypes.
Zheng et al. 2021 [71]	Plasma	Pre-centrifugation storage temperature and delayed processing time	471	Mass Spec (LC-MS)	Delays in plasma centrifugation (up to 3.5 h) significantly impacted the metabolome regardless of the storage temperature (4 °C, 25 °C, 37 °C); therefore, pre-centrifugation delay should be as short as possible.
Jain et al. 2017 [72]	Plasma	Delayed processing time and storage temperature	9	Mass Spec (LC-MS)	Centrifugation delays of up to 20 h (storage at room temperature) significantly impacted metabolites from select pathways including glycolysis/gluconeogenesis/and pyruvate metabolism, TCA cycle, and gamma-glutamyl amino acid; however, the extent of the impact is influenced by inter-individual variation. Metabolites identified as potential markers of processing delays include 5-oxoproline, lactate, pyruvate, and fumarate, as well as the ratio between arginine and ornithine
Nishiumi et al. 2018 [75]	Plasma	Delayed processing time and storage temperature	6	Mass Spec (GC and LC-MS)	Storage time and temperature significantly influenced individual metabolite concentrations in plasma samples with increased levels of sucrose and pyruvic acid identified as potential markers of storage at room temperature and decreased pyruvic acid as a marker for cold temperature storage. Hypoxanthine has been identified as a potential marker for storage time with concentrations increasing with increased time. Immediate cooling and centrifugation of blood samples is recommended for the metabolomic analysis of plasma using both GC and LC-MS platforms.
McClain et al. 2021 [40]	Serum	Pre-centrifugation clotting time	26	Mass Spec (GC-MS and LC-MS)	Increased clotting time (up to 2 h) significantly impacts metabolite levels.

Table 2. Overview of studies assessing the effect of pre-centrifugation factors on the metabolome stability of plasma and serum samples using NMR- and MS-based analytical techniques.

Study	Sample Type	Variable/s Tested	Sample Size (n)	Analytical Method Used	Conclusions
Niemuth et al. 2015 [60]	Plasma and Whole Blood (animal)	Delayed processing time	5 whole blood samples 20 plasma samples	NMR	Clear differences between whole blood and plasma observed, but changes associated with plasma processing delays would only affect the interpretation of a few metabolites at the level of clinical relevance including glutathione and taurine.
Ghini et al. 2022 [45]	Plasma and Serum	Delayed processing time and storage temperature	234 plasma and 150 serum samples	NMR	Delayed blood processing significantly alters lactate and glucose levels in plasma, suggesting delayed blood processing negatively affects plasma quality.
Malmodin et al. 2024 [61]	Plasma	Pre-centrifugation storage temperature, delayed processing time, and exposure to light	951	NMR	SCFA, pyruvate, glucose, lactate, and ornithine demonstrate distinct incubation temperature-dependent characteristics. Decrease in SCFA at 4 °C was found to be reversible through the increase in temperature to 22 °C prior to centrifugation. Light exposure has minimal effect. Plasma sampling should be conducted immediately at room temperature or stored at 4 °C followed by warming to room temperature prior to centrifugation.
Bernini et al. 2011 [62]	Plasma and Serum	Delayed processing time and temperature	10 plasma and 10 serum samples	NMR	The metabolite degradation processes are time-dependent and temperature-dependent for both plasma and serum: incubation at 25 °C causes more profound changes in the NMR profile. Notably, a decrease in glucose concentrations was observed in both sample types and an increase in lactate. Time between sample collection and processing should not exceed 2 h with samples being stored at 4 °C.
Brunius et al. 2017 [63]	Plasma	Delayed processing time and temperature	111	NMR	Pre-centrifugation delays and storage temperature significantly impacted metabolite profiles with most metabolite drift patterns being highly reproducible between individuals. Alterations were more pronounced at 22 °C compared to 4 °C with associations to specific groups of metabolites; therefore, pre-centrifugation storage at 22 °C should be avoided. Pyruvate, lactate, and ornithine were identified as markers for discriminating between pre-centrifugation temperatures. Pre-centrifugation delays of up to 36 h at 4 °C have minimal effects on overall metabolite changes.
Debik et al. 2022 [64]	Plasma and Serum	Delayed processing time	20 plasma and 20 serum samples	NMR	Centrifugation delay significantly impacts metabolite profiles, especially those associated with anaerobic glycolysis. Plasma samples were more robust against centrifugation delays compared to serum samples.
Santos Ferreira et al. 2019 [65]	Plasma and Serum	Delayed processing time, pre-centrifugation storage temperature	37 plasma and 37 serum samples	NMR	Overall, delays of up to 48 h and storage at either 4 °C or 21 °C prior to centrifugation had minimal effects on metabolite concentrations. Individual metabolites identified to be sensitive to processing delays and storage temperature included pyruvate, with mean concentrations decreasing when stored for 24 h at 4 °C but increasing when stored for 24 h at 21 °C compared to control.
Jobard et al. 2016 [66]	Plasma and Serum	Delayed processing time, pre-centrifugation storage temperature	189 plasma and 192 serum samples	NMR	Storage of samples at 4 °C following collection slowed down the metabolome degradation process in both plasma and serum samples compared to room temperature storage for up to 6 h prior to centrifugation. Increased serum clot contact time at room temperature significantly altered the metabolome of serum samples.
Loo et al. 2020 [46]	Plasma and Serum	Pre-centrifugation storage temperature (serum) and sample heating (plasma)	15 serum and 15 plasma samples	NMR	Storage of serum samples at 20 °C for up to 48 h prior to processing impacted individual metabolite levels (glucose decreased and lactate increased). Heating of plasma samples at 56 °C for 30 min prior to processing caused major systematic changes in metabolite profiles increasing levels of triglycerides, lactate, alanine, and glycerol post heating and decreased relative concentrations of α-1-acid-glycoprotein; therefore, it should be avoided.
Xiong et al. 2024 [59]	Plasma and Serum	Delayed processing time	168 plasma and 84 serum samples	NMR	The storage of samples at 4 °C for 24 h with a fluoride/oxalate additive resulted in the least impact on metabolite profiles; however, the overall impact of processing delays was influenced by the blood tube/anticoagulant used.
Altmann et al. 2025 [67]	Plasma and Serum	Delayed processing time	60 plasma and 30 serum samples	NMR	The storage of samples at room temperature for <8 h prior to centrifugation affects plasma and serum samples differently, with metabolites demonstrating differences in stability depending on the blood collection tube type used. Lactic acid and the ratio between lactic acid and glucose were found to be the least stable with a stability time of 24–45 min and 21–32 min, respectively; therefore, they can be potentially used as markers of sample quality.
Fliniaux et al. 2011 [68]	Serum	Delayed processing time and storage temperature	42	NMR	Prolonged storage (>4 h) of blood at room temperature can be characterised by increased concentrations of lactate and decreased concentrations of glucose in serum samples.
Trezzi et al. 2016 [73]	Plasma	Pre-centrifugation storage temperature and delayed processing time	18	Mass Spec (GC-MS)	Pre-centrifugation delays had a minor impact on metabolite concentration. Temperature at which samples are stored between time of collection and centrifugation is a key factor that can influence metabolite concentrations—recommend storing samples on ice following collection. Lactic acid and ascorbic acid identified to be temperature-sensitive metabolites. Samples of high pre-centrifugation quality identified as those that were stored either on ice for ≥3 h or at room temp for <3 h.
Dunn et al. 2008 [74]	Serum	Pre-centrifugation storage time	40	Mass Spec (GC-MS)	Changes to the serum metabolome caused by storage at 4 °C for up to 24 h were minimal with no detectable variability in comparison to samples frozen immediately at −80 °C post collection.
Khadka et al. 2019 [58]	Plasma	Delayed processing time and temperature	70	Mass Spec (LC-MS)	Storage for up to 24 h at room temperature or at 4 °C did not significantly impact the metabolome in both plasma and serum samples; however, more lipids (e.g., lysophosphatidylcholines, phosphatidylethanolamines) were elevated in EDTA tubes when stored at room temperature compared to 4 °C. Sample processing recommended to be completed within 24 h post collection at 4 °C.
Wang et al. 2023 [69]	Plasma	Delayed processing time, pre-centrifugation storage temperature	829	Mass Spec (LC-MS)	Three lipid classes (phosphatidylinositol, hexosylceramide and SM) identified to be stable against delayed processing time (≤4 h) and storage temperature (4 °C, 21 °C or 30 °C). The metabolome in samples cooled immediately to 4 °C following collection and storage for up to 24 h demonstrated the most stability in lipid profiles.
Wang et al. 2018 [70]	Plasma	Delayed processing time	72	Mass Spec (LC-MS)	Glucose, lactate, pyruvate, and 5-oxoproline concentrations were significantly altered because of processing delays (up to 48 h, stored at 4 °C). Prolonged processing delays resulted in poor reproducibility and impacted concentrations of metabolites involved in glycolysis, TCA cycle, glutathione metabolism, γ-glutamyl amino acid, and several fatty acid subtypes.
Zheng et al. 2021 [71]	Plasma	Pre-centrifugation storage temperature and delayed processing time	471	Mass Spec (LC-MS)	Delays in plasma centrifugation (up to 3.5 h) significantly impacted the metabolome regardless of the storage temperature (4 °C, 25 °C, 37 °C); therefore, pre-centrifugation delay should be as short as possible.
Jain et al. 2017 [72]	Plasma	Delayed processing time and storage temperature	9	Mass Spec (LC-MS)	Centrifugation delays of up to 20 h (storage at room temperature) significantly impacted metabolites from select pathways including glycolysis/gluconeogenesis/and pyruvate metabolism, TCA cycle, and gamma-glutamyl amino acid; however, the extent of the impact is influenced by inter-individual variation. Metabolites identified as potential markers of processing delays include 5-oxoproline, lactate, pyruvate, and fumarate, as well as the ratio between arginine and ornithine
Nishiumi et al. 2018 [75]	Plasma	Delayed processing time and storage temperature	6	Mass Spec (GC and LC-MS)	Storage time and temperature significantly influenced individual metabolite concentrations in plasma samples with increased levels of sucrose and pyruvic acid identified as potential markers of storage at room temperature and decreased pyruvic acid as a marker for cold temperature storage. Hypoxanthine has been identified as a potential marker for storage time with concentrations increasing with increased time. Immediate cooling and centrifugation of blood samples is recommended for the metabolomic analysis of plasma using both GC and LC-MS platforms.
McClain et al. 2021 [40]	Serum	Pre-centrifugation clotting time	26	Mass Spec (GC-MS and LC-MS)	Increased clotting time (up to 2 h) significantly impacts metabolite levels.

3.4. Centrifugation Parameters

Centrifugation is a key step in the preparation of biofluids for metabolomic analysis, used to separate cells from the fluid component before storage and analysis [76]. Three studies [66,77,78] (Table 3) were identified to explore this theme in plasma, serum, and synovial fluid samples using NMR- and MS-based approaches. The findings from studies applying NMR and MS (LC-MS) on plasma and serum samples were mixed. Jobard et al. [66] compared centrifugation temperature (20 °C vs. 4 °C), speed (2000× g vs. 3000× g), and time (10 vs. 20 min) on plasma and serum NMR profiles, where no significant changes were observed as a result of these parameters. Lesche et al. [77] compared speed (1500× g vs. 3000× g) and time (5 vs. 10 min) at 20 °C using both NMR and LC-MS on plasma samples. Both the centrifugation speed and time altered the plasma metabolomic profiles. The study demonstrated that the centrifugation conditions influence glutamine levels, as measured by NMR analysis. These differences were attributed to residual platelets remaining in the samples centrifuged at 1500× g for 10 min, resulting in higher glutamine concentrations compared with samples centrifuged at 3000× g for 5 min. In LC-MS analysis, sphingomyelin SM (40:1) and the sodium adduct of sphingomyelin SM (42:2) were identified as key contributors to the separation of samples processed under the two centrifugation conditions in the sPLS-DA model [77].

Anderson and colleagues [78] compared the impact of centrifuging synovial fluid samples prior to freezing using NMR analysis. The study found that the centrifuging of synovial fluid prior to freezing was beneficial and resulted in improved sample clarity and analytical reproducibility; therefore, it is a recommended step for the metabolomic analysis of synovial fluid [78].

Table 3. Summary of the studies evaluating the influence of centrifugation parameters on plasma, serum, and synovial fluid metabolomes.

Study	Sample Type	Variable/s Tested	Sample Size (n)	Analytical Method Used	Conclusions
Jobard et al. 2016 [66]	Plasma and Serum	Impact of centrifugation parameters	189 plasma and 192 serum samples	NMR	Centrifugation temperature (20 °C vs. 4 °C), rotational speed (2000 vs. 3000× g), and time (10 vs. 20 min) did not significantly impact plasma and serum metabolome profiles.
Anderson et al. 2020 [78]	Synovial Fluid (animal)	Centrifuging vs. not centrifuging samples	18	NMR	Centrifuging synovial fluid samples prior to freezing aided in the removal of contaminants in the supernatant associated with intracellular material and is, therefore, recommended.
Lesche et al. 2016 [77]	Plasma	Impact of centrifugation parameters	20	NMR and Mass Spec (LC-MS)	Centrifugation of samples at 1500× g for 10 min or at 3000× g for 5 min at 20 °C significantly impacted plasma metabolomic profiles when analysed using NMR and LC-MS.

Table 3. Summary of the studies evaluating the influence of centrifugation parameters on plasma, serum, and synovial fluid metabolomes.

Study	Sample Type	Variable/s Tested	Sample Size (n)	Analytical Method Used	Conclusions
Jobard et al. 2016 [66]	Plasma and Serum	Impact of centrifugation parameters	189 plasma and 192 serum samples	NMR	Centrifugation temperature (20 °C vs. 4 °C), rotational speed (2000 vs. 3000× g), and time (10 vs. 20 min) did not significantly impact plasma and serum metabolome profiles.
Anderson et al. 2020 [78]	Synovial Fluid (animal)	Centrifuging vs. not centrifuging samples	18	NMR	Centrifuging synovial fluid samples prior to freezing aided in the removal of contaminants in the supernatant associated with intracellular material and is, therefore, recommended.
Lesche et al. 2016 [77]	Plasma	Impact of centrifugation parameters	20	NMR and Mass Spec (LC-MS)	Centrifugation of samples at 1500× g for 10 min or at 3000× g for 5 min at 20 °C significantly impacted plasma metabolomic profiles when analysed using NMR and LC-MS.

3.5. Post-Centrifugation Handling and Freezing Delays

Studies using NMR spectroscopy (Table 4) have assessed the impact of the time of delay to freezing and the temperature during the delay on plasma and serum metabolomes. One study found short delays of <1 h did not cause significant changes to plasma and serum metabolomes, suggesting that a short post-processing delay may be tolerable [66]. However, moderate to long delays between 2 and 24 h have been found to result in progressive metabolite degradation, especially in serum samples [62]. Delays of 2 h significantly impacted serum lipid concentrations, whilst delays of up to 24 h were associated with significant decreases in triglycerides, citrate, proline, choline, histidine, and lactic acid concentrations in both plasma and serum samples [45,62,67]. Studies also noted that the impact of post-centrifugation processing delays can also be influenced by the type of anticoagulant tube used, with proline and choline levels demonstrating better stability in EDTA plasma tubes compared to fluoride/oxalate treated plasma [62]. Additionally, one study suggested that metabolites in lithium–heparin plasma tubes demonstrated higher tolerance to post-centrifugation delays compared to EDTA tubes [67]. Overall, plasma samples were found to generally be more stable than serum samples to post-centrifugation delays; therefore, they are recommended over serum samples in NMR workflows that involve delayed freezing. These findings suggest that the period between sample centrifugation and freezing represents a critical window during which continued enzymatic activity, oxidation, and metabolite degradation can still occur; therefore, delays in sample processing following centrifugation should be limited. Ideally, samples should be frozen immediately at −80 °C post-centrifugation where possible, and delays should be limited to <1 h particularly for serum samples.

Four studies (Table 4) were identified that had applied MS analysis to test the impact of post-centrifugation delay. One study tested the impact of a delayed freezing time of 3 days at room temperature, followed by 9 days at 4 °C, on serum samples using both GC and LC-MS analytical techniques [79]. The results of the study indicated significant metabolite class specific alterations, which are perhaps not surprising given the long delays during sample processing; however, overall, the metabolome demonstrated stability despite the prolonged delay [79]. Similar findings were observed in a study that stored serum samples at 4 °C for up to 24 h before freezing [40]. In another study, the impact of plasma sample drying following centrifugation was assessed, where the time of drying was identified to significantly influence the metabolome, with metabolites exhibiting rapid degradation or generation over the first 48 h at room temperature. Plasma metabolites were significantly impacted by extended drying periods including methionine sulfoxide, glutamic acid, and histidine [80]. The short-term storage of plasma samples (up to 16 h) at either 4 or 12 °C resulted in less variation compared to plasma samples stored at room temperature [81]. These findings suggest that the plasma metabolome is more stable when stored at low temperatures following centrifugation, whilst serum samples demonstrate broader temperature stability even under prolonged delays.

Despite multiple studies assessing post-centrifugation handling and freezing delays in plasma and serum, no studies have directly examined these effects in synovial fluid.

Table 4. Summary of the studies exploring the impact of post-centrifugation sample handling and delays in freezing on the metabolome.

Study	Sample Type	Variable/s Tested	Sample Size (n)	Analytical Method Used	Conclusions
Ghini et al. 2022 [45]	Plasma and Serum	Post-centrifugation processing delay	234 plasma and 150 serum samples	NMR	Delayed time between centrifugation and freezing resulted in oxidative and redox-related degradation. This degradation caused a decrease in citrate and proline levels, which can act as indicators of delayed freezing.
Bernini et al. 2011 [62]	Plasma and Serum	Post-centrifugation freezing delay	4 plasma and 5 serum samples	NMR	Freezing delays of up to 24 h resulted in a progressive decreased concentration of triglycerides, proline, choline, citrate, and histidine. Citrate showed chemical shift changes, while proline and choline were more stable in plasma EDTA than fluoride/oxalate-treated samples. Plasma was more stable when kept at room temperature than serum. To minimise metabolome degradation, samples should be immediately frozen at −80 °C after centrifugation.
Jobard et al. 2016 [66]	Plasma and Serum	Post-centrifugation freezing delay	189 plasma and 192 serum samples	NMR	Post-centrifugation delays in freezing (15 min vs. 1 h) did not significantly impact plasma and serum metabolomes.
Altmann et al. 2025 [67]	Plasma and Serum	Post-centrifugation freezing delay	27 plasma and 27 serum samples	NMR	Post-centrifugation delays in freezing (up to 8 h) significantly impacted lipid concentrations in serum samples observed at the 2 h delay time point. Lactic acid, the least stable metabolite in plasma samples, was found to be stable for up to 4.3 h in serum. Samples collected in lithium–heparin tubes demonstrated higher metabolite stability to post-centrifugation delay compared to EDTA tubes.
Volani et al. 2017 [80]	Plasma	Post-centrifugation processing delay (sample drying)	69	Mass Spec (LC-MS)	Time of drying of samples prior to storage influences the final metabolome, with some metabolites demonstrating rapid degradation or generation over the first 48 h at room temperature. Therefore, having a longer drying step should be avoided, as it would significantly change the concentrations of these metabolites e.g., methionine sulfoxide, glutamic acid, and histidine.
McClain et al. 2021 [40]	Serum	Post-centrifugation storage time	52	Mass Spec (GC-MS and LC-MS)	Storage in the fridge for up to 24 h caused moderate changes to the overall metabolome; however, the degree of impact changed depending on individual metabolites.
La Frano et al. 2018 [79]	Serum	Post-centrifugation freezing delay	20	Mass Spec (GC and LC-MS)	Delayed freezing (three days at room temperature followed by nine days at 4 °C) following centrifugation caused significant changes on metabolite concentrations, which varied depending on the metabolite class. The serum metabolome, however, demonstrates stability despite delays in freezing.
Kamlage et al. 2014 [81]	Plasma	Post-centrifugation short term storage (wet ice, room temperature or at 12 °C for up to 16 h)	23	Mass Spec (GC and LC-MS)	The short-term storage of plasma at 4 °C or 12 °C resulted in the less variation in the metabolome compared to storage at room temperature. However, all storage temperatures caused distinct changes to the metabolome compared to immediate processing.

Table 4. Summary of the studies exploring the impact of post-centrifugation sample handling and delays in freezing on the metabolome.

Study	Sample Type	Variable/s Tested	Sample Size (n)	Analytical Method Used	Conclusions
Ghini et al. 2022 [45]	Plasma and Serum	Post-centrifugation processing delay	234 plasma and 150 serum samples	NMR	Delayed time between centrifugation and freezing resulted in oxidative and redox-related degradation. This degradation caused a decrease in citrate and proline levels, which can act as indicators of delayed freezing.
Bernini et al. 2011 [62]	Plasma and Serum	Post-centrifugation freezing delay	4 plasma and 5 serum samples	NMR	Freezing delays of up to 24 h resulted in a progressive decreased concentration of triglycerides, proline, choline, citrate, and histidine. Citrate showed chemical shift changes, while proline and choline were more stable in plasma EDTA than fluoride/oxalate-treated samples. Plasma was more stable when kept at room temperature than serum. To minimise metabolome degradation, samples should be immediately frozen at −80 °C after centrifugation.
Jobard et al. 2016 [66]	Plasma and Serum	Post-centrifugation freezing delay	189 plasma and 192 serum samples	NMR	Post-centrifugation delays in freezing (15 min vs. 1 h) did not significantly impact plasma and serum metabolomes.
Altmann et al. 2025 [67]	Plasma and Serum	Post-centrifugation freezing delay	27 plasma and 27 serum samples	NMR	Post-centrifugation delays in freezing (up to 8 h) significantly impacted lipid concentrations in serum samples observed at the 2 h delay time point. Lactic acid, the least stable metabolite in plasma samples, was found to be stable for up to 4.3 h in serum. Samples collected in lithium–heparin tubes demonstrated higher metabolite stability to post-centrifugation delay compared to EDTA tubes.
Volani et al. 2017 [80]	Plasma	Post-centrifugation processing delay (sample drying)	69	Mass Spec (LC-MS)	Time of drying of samples prior to storage influences the final metabolome, with some metabolites demonstrating rapid degradation or generation over the first 48 h at room temperature. Therefore, having a longer drying step should be avoided, as it would significantly change the concentrations of these metabolites e.g., methionine sulfoxide, glutamic acid, and histidine.
McClain et al. 2021 [40]	Serum	Post-centrifugation storage time	52	Mass Spec (GC-MS and LC-MS)	Storage in the fridge for up to 24 h caused moderate changes to the overall metabolome; however, the degree of impact changed depending on individual metabolites.
La Frano et al. 2018 [79]	Serum	Post-centrifugation freezing delay	20	Mass Spec (GC and LC-MS)	Delayed freezing (three days at room temperature followed by nine days at 4 °C) following centrifugation caused significant changes on metabolite concentrations, which varied depending on the metabolite class. The serum metabolome, however, demonstrates stability despite delays in freezing.
Kamlage et al. 2014 [81]	Plasma	Post-centrifugation short term storage (wet ice, room temperature or at 12 °C for up to 16 h)	23	Mass Spec (GC and LC-MS)	The short-term storage of plasma at 4 °C or 12 °C resulted in the less variation in the metabolome compared to storage at room temperature. However, all storage temperatures caused distinct changes to the metabolome compared to immediate processing.

3.6. Storage Conditions

The long-term preservation of the metabolome is crucial for ensuring biofluids are viable for metabolomic analysis after they have been kept in either short- and long-term storage. Clinical samples, such as from trials, have to be collected over a period of time prior to all samples being available for metabolomics analysis; hence, storage is unavoidable. Further, samples collected through biobanks may have been stored for long time periods. Variations in storage conditions can potentially lead to chemical degradation or enzymatic activity, which can alter the metabolomic profile of the sample. Four studies (Table 5) were identified that assessed the impact of the storage duration, temperature, and sample handling on the metabolite stability using NMR spectroscopy. One study investigated the impact of long-term storage (2–15 years) at −20 °C on animal-derived plasma samples [82]. This study was able to identify a selective few metabolites that demonstrated significant correlations with storage durations including betaine CH₃ group, betaine CH₂ group, acetoacetate, and imidazole compounds, which significantly decreased over time, and glycerol and glucose, which significantly increased over time [82]. The short-term storage of plasma samples was investigated in one study, which tested the impact of storage at room temperature and at −20 °C for 7 days and 1 month. The results demonstrated that, following storage at room temperature for 2.5 h, plasma samples exhibited significant alterations in lipid profiles, and storage at −20 °C for 1 month led to increased concentrations of proline and glucose levels. No significant changes were observed in samples stored at −20 °C for 7 days [47]. Two studies investigated the impact of short-term storage at −80 °C on both plasma and serum metabolomes. The results of both studies agreed, with no significant changes being observed in the metabolomes [47,66]. These findings suggest that the short-term storage of plasma samples can be done at −20 °C, avoiding storage at room temperature, but early transfer to −80 °C is preferable, and long-term storage of both plasma and serum samples should be at −80 °C. The quenching and extraction of the plasma metabolome was also explored in one study prior to freezing [83]. This study tested the use of methanol, ethanol, or a mixture of methanol and chloroform on sample extraction followed by incubation on ice. The results demonstrated that the use of a methanol: chloroform mix improved the extraction of coenzymes including NAD⁺; however, extended incubation on ice increased the conversion of NAD⁺ to NADH, decreased ATP levels, and sped up coenzyme degradation. These findings suggest that the rapid quenching of the metabolome prior to sample freezing is ideal for capturing labile coenzymes; however, prolonged incubation on ice should be avoided if quantifying redox cofactors and ATP [83].

Six studies (Table 5) were identified that had applied MS for the analysis of plasma and serum samples. Of these identified studies, the impact of long-term storage was assessed only on the plasma metabolome, where it was found that the metabolome demonstrated stability for up to 7 years in −80 °C; however, prolonged storage beyond 7 years was associated with significant metabolomic alterations [53,80,84,85,86]. The short-term storage of plasma samples at room temperature has been shown to cause substantial metabolomic alterations; therefore, it should be avoided where possible [80]. Metabolite stability was found to be class-specific, with specific metabolite classes such as oxylipins identified to demonstrate stability across a range of storage conditions [80,85]. Only one study investigated the impact of storage on serum samples, which found that the short-term storage of serum samples at either 4 °C or 22 °C for up to 24 h caused notable changes in amino acid concentrations suggesting the short-term temperature sensitivity of the serum metabolome [87].

While the majority of studies focused on plasma and serum, notably, two studies were identified that examined the impact of storage conditions on synovial fluid metabolomics using NMR platforms. In one study, animal-derived synovial fluid samples were either snap frozen in liquid nitrogen followed by storage in −80 °C or frozen directly in either −20 °C, −80 °C, or on dry ice. Samples that were snap frozen first, prior to freezing at −80 °C, demonstrated the least variability in metabolite profile compared to other freezing techniques, suggesting rapid freezing can help preserve the synovial fluid metabolome [78]. Two studies evaluated the short- and long-term storage of synovial fluid samples [88,89]. The results suggested that short-term storage at room temperature or at 4 °C for up to 12 h primarily affects metabolites associated with energy synthesis, whilst long-term storage at either −75 °C or −80 °C resulted in significant changes in metabolite concentrations. When stored at −80 °C for 10–12 months, significant decreases in concentrations of 3-hydroxybutyrate, acetate, succinate, creatine, dimethyl sulfone, and N-N dimethylglycine were observed and an increase in acetoacetate [88]. Storage at −75 °C for 1 year resulted in decreased concentrations of glucose, N-acetyl glycoproteins, CH₂-chain, and CH₃-terminal and resonances of lipoproteins, valine, leucine, and isoleucine concentrations [89]. These findings suggest that the short-term storage of synovial fluid samples should be limited, with samples being stored at −80 °C if stored for periods exceeding 12 h.

Table 5. Summary of the studies exploring the impact of samples storage on plasma, serum, and synovial fluid metabolomes.

Study	Sample Type	Variable/s Tested	Sample Size (n)	Analytical Method Used	Conclusions
Jaggard et al. 2021 [88]	Synovial Fluid	Impact of short- and long-term storage	75	NMR	The majority of metabolites affected by short term storage (up to 12 h either at room temperature or at 4 °C) were associated with energy synthesis (2-ketoisovalerate, valine, dimethylamine, succinate, 2-hydroxybutyrate, and acetaminophen glucuronide). Long term storage (10–12 months at −80 °C) significantly impacted the concentrations of 3-hydroxybutyrate, acetate, succinate, creatine, dimethyl sulfone, and N-N dimethylglycine, which were observed to decrease, whilst an increase in acetoacetate was observed.
Damyanovich et al. 2000 [89]	Synovial Fluid	Impact of short- and long-term storage	12	NMR	Prolonged storage (>1 year at −75 °C) caused significant signal reductions for several metabolites including glucose, N-acetyl glycoproteins, CH2-chain, and CH3-terminal and resonances of lipoproteins, valine, leucine, and isoleucine. Short-term storage (<24 h) did not significantly impact the metabolome.
Trabi et al. 2013 [82]	Plasma (animal)	Impact of long-term storage	22	NMR	Metabolome was stable for up to 15 years in storage at −20 °C with only a few metabolites identified as sensitive to long-term storage including Betaine CH3 group, Betaine CH2 group, acetoacetate, imidazole compounds (negative correlation to storage time), and glycerol and glucose (positive correlation to storage time).
Jobard et al. 2016 [66]	Plasma and Serum	Impact of storage time	189 plasma and 192 serum samples	NMR	Storage of plasma or serum samples at −80 °C for up to 3 months did not significantly impact their metabolomes.
Pinto et al. 2014 [47]	Plasma	Impact of storage temperature and time	52	NMR	The short-term storage of plasma samples at room temperature caused significant changes to the lipidomic profile of the plasma following >2.5 h of storage. Storage of plasma samples at −20 °C for up to 7 days did not cause any significant changes to the metabolome; however, following storage for 1 month, significant increases in levels of proline and glucose are observed. Storage of plasma samples at −80 °C demonstrate more stability and is, therefore, recommended over storage at −20 °C for the long-term storage of plasma.
An et al. 2021 [87]	Serum	Impact of storage time and temperature	165	Mass Spec (LC-MS)	Storage of serum samples at either 4 °C or 22 °C for up to 24 h caused significant changes to amino acid concentrations.
Volani et al. 2017 [80]	Plasma	Impact of storage temperature and time	69	Mass Spec (LC-MS)	While storage at room temperature caused significant changes to the metabolome, storage at −80 °C demonstrated metabolome stability for up to 6 months. Metabolite stability is both chemical class- and metabolite-dependent.
Moran-Garrido et al. 2025 [85]	Plasma	Impact of short- and long-term storage	9	Mass Spec (LC-MS)	Plasma oxylipins remained stable when stored at 4 °C for up to 120 h and at −80 °C for 98 days, and the addition of butylated hydroxytoluene helped preserve oxylipins when stored at −20 °C for 98 days.
Hebels et al. 2013 [53]	Plasma	Storage temperature	6	Mass Spec (LC-MS)	Storage of samples at −80 °C vs. liquid nitrogen did not cause substantial changes to plasma metabolite profiles.
Wagner-Golbs et al. 2019 [84]	Plasma	Impact of long-term storage at −80 °C	2398	Mass Spec (GC and LC-MS)	The plasma metabolome demonstrated stability when stored at −80 °C for up to 7 years; however, further prolonged storage resulted in significant changes to the metabolome; therefore, the metabolomic profiling of plasma samples stored for longer than 7 years should be avoided where possible.
Petrick et al. 2024 [86]	Plasma	Impact of storage temperature and time	12	Mass Spec (LC-MS)	The plasma metabolome demonstrated more stability when stored at −80 °C compared to storage at −20 °C or 4 °C for up to 168 days; therefore, the long-term storage of plasma samples at −80 °C is recommended over storage at −20 °C or 4 °C.
Nagana Gowda et al. 2023 [83]	Plasma	Impact of sample quenching, incubation temperature and time	84	NMR	The use of a mixture of methanol and chloroform enhanced coenzyme extraction compared to the use of methanol or ethanol; therefore, rapid quenching using an organic solvent is recommended for the analysis of labile metabolites. The incubation of samples on ice increased the rate of NAD+ to NADH conversion and the rate at which coenzyme levels changed irrespective of the extraction solvent used. Incubation on ice also decreased levels of ATP in the samples.
Anderson et al. 2020 [78]	Synovial Fluid (animal)	Post-centrifugation freezing method	32	NMR	Synovial fluid samples immediately frozen using liquid nitrogen followed by storage at −80° demonstrated the least variability compared to freezing at −20 °C, −80 °C, or using dry ice.

Table 5. Summary of the studies exploring the impact of samples storage on plasma, serum, and synovial fluid metabolomes.

Study	Sample Type	Variable/s Tested	Sample Size (n)	Analytical Method Used	Conclusions
Jaggard et al. 2021 [88]	Synovial Fluid	Impact of short- and long-term storage	75	NMR	The majority of metabolites affected by short term storage (up to 12 h either at room temperature or at 4 °C) were associated with energy synthesis (2-ketoisovalerate, valine, dimethylamine, succinate, 2-hydroxybutyrate, and acetaminophen glucuronide). Long term storage (10–12 months at −80 °C) significantly impacted the concentrations of 3-hydroxybutyrate, acetate, succinate, creatine, dimethyl sulfone, and N-N dimethylglycine, which were observed to decrease, whilst an increase in acetoacetate was observed.
Damyanovich et al. 2000 [89]	Synovial Fluid	Impact of short- and long-term storage	12	NMR	Prolonged storage (>1 year at −75 °C) caused significant signal reductions for several metabolites including glucose, N-acetyl glycoproteins, CH2-chain, and CH3-terminal and resonances of lipoproteins, valine, leucine, and isoleucine. Short-term storage (<24 h) did not significantly impact the metabolome.
Trabi et al. 2013 [82]	Plasma (animal)	Impact of long-term storage	22	NMR	Metabolome was stable for up to 15 years in storage at −20 °C with only a few metabolites identified as sensitive to long-term storage including Betaine CH3 group, Betaine CH2 group, acetoacetate, imidazole compounds (negative correlation to storage time), and glycerol and glucose (positive correlation to storage time).
Jobard et al. 2016 [66]	Plasma and Serum	Impact of storage time	189 plasma and 192 serum samples	NMR	Storage of plasma or serum samples at −80 °C for up to 3 months did not significantly impact their metabolomes.
Pinto et al. 2014 [47]	Plasma	Impact of storage temperature and time	52	NMR	The short-term storage of plasma samples at room temperature caused significant changes to the lipidomic profile of the plasma following >2.5 h of storage. Storage of plasma samples at −20 °C for up to 7 days did not cause any significant changes to the metabolome; however, following storage for 1 month, significant increases in levels of proline and glucose are observed. Storage of plasma samples at −80 °C demonstrate more stability and is, therefore, recommended over storage at −20 °C for the long-term storage of plasma.
An et al. 2021 [87]	Serum	Impact of storage time and temperature	165	Mass Spec (LC-MS)	Storage of serum samples at either 4 °C or 22 °C for up to 24 h caused significant changes to amino acid concentrations.
Volani et al. 2017 [80]	Plasma	Impact of storage temperature and time	69	Mass Spec (LC-MS)	While storage at room temperature caused significant changes to the metabolome, storage at −80 °C demonstrated metabolome stability for up to 6 months. Metabolite stability is both chemical class- and metabolite-dependent.
Moran-Garrido et al. 2025 [85]	Plasma	Impact of short- and long-term storage	9	Mass Spec (LC-MS)	Plasma oxylipins remained stable when stored at 4 °C for up to 120 h and at −80 °C for 98 days, and the addition of butylated hydroxytoluene helped preserve oxylipins when stored at −20 °C for 98 days.
Hebels et al. 2013 [53]	Plasma	Storage temperature	6	Mass Spec (LC-MS)	Storage of samples at −80 °C vs. liquid nitrogen did not cause substantial changes to plasma metabolite profiles.
Wagner-Golbs et al. 2019 [84]	Plasma	Impact of long-term storage at −80 °C	2398	Mass Spec (GC and LC-MS)	The plasma metabolome demonstrated stability when stored at −80 °C for up to 7 years; however, further prolonged storage resulted in significant changes to the metabolome; therefore, the metabolomic profiling of plasma samples stored for longer than 7 years should be avoided where possible.
Petrick et al. 2024 [86]	Plasma	Impact of storage temperature and time	12	Mass Spec (LC-MS)	The plasma metabolome demonstrated more stability when stored at −80 °C compared to storage at −20 °C or 4 °C for up to 168 days; therefore, the long-term storage of plasma samples at −80 °C is recommended over storage at −20 °C or 4 °C.
Nagana Gowda et al. 2023 [83]	Plasma	Impact of sample quenching, incubation temperature and time	84	NMR	The use of a mixture of methanol and chloroform enhanced coenzyme extraction compared to the use of methanol or ethanol; therefore, rapid quenching using an organic solvent is recommended for the analysis of labile metabolites. The incubation of samples on ice increased the rate of NAD+ to NADH conversion and the rate at which coenzyme levels changed irrespective of the extraction solvent used. Incubation on ice also decreased levels of ATP in the samples.
Anderson et al. 2020 [78]	Synovial Fluid (animal)	Post-centrifugation freezing method	32	NMR	Synovial fluid samples immediately frozen using liquid nitrogen followed by storage at −80° demonstrated the least variability compared to freezing at −20 °C, −80 °C, or using dry ice.

3.7. Freeze/Thaw Cycling

Biofluids undergoing F/T cycles are a common occurrence in metabolomic studies due to routine sample storage, transport, and the use of the same samples for multiple analytical procedures. The F/T process represents a critical source of pre-analytical variability in metabolomic studies. Each cycle and thawing can induce mechanical disruption, enzymatic reactivation, and chemical degradation, thus potentially affecting the concentration and detectability of metabolites. This section summarises the findings from studies that investigated the number of F/T cycles, thawing conditions, and sample-specific vulnerabilities. Only three studies (Table 6) were identified that applied NMR analysis for plasma, serum, and synovial fluid. One study evaluated the effect of up to 10 F/T cycles on serum samples with results demonstrating metabolome stability up until 5 F/T cycles [68]. Samples that underwent >5 F/T cycles demonstrated significantly altered concentrations of choline, glycerol, methanol, and ethanol, therefore acting as potential markers of F/T cycling [68]. In plasma the highest metabolomic variation was found to occur after four or more F/T cycles [47]. Synovial fluid analysis revealed the metabolome remained stable for up to 10 F/T cycles; however, eight metabolites were found to be significantly impacted by repeated F/T cycling, with changes observed following the first F/T cycle [88,89]. These include 2-ketoisovalerate, glycosaminoglycan, acetaminophen glucuronide, formate, arginine, glutamate, phenylalanine, and acetaminophen glucuronide [88]. Overall, these results suggest limiting F/T cycling of all three biofluids to no more than four cycles in order to preserve metabolome integrity.

Nine studies (Table 6) were identified that had applied MS platforms for the analysis of plasma and serum samples. In a study assessing the impact of up to three F/T cycles on plasma samples, five metabolites were identified that were consistently affected across all cycles including bilirubin (E, E), dihydroorotate, maltose, glycerol 3-phosphate, and sphingosine [90]. Another study assessed the impact of up to five cycles specifically on the stability of plasma oxylipins, with results suggesting higher stability of esterified oxylipin compared to free oxylipins [85]. In another study, up to four F/T cycles did not significantly impact the plasma metabolome [56]. A study assessing the impact of up to 10 F/T cycles on plasma and serum samples observed changes in both biofluids; however, the highest change was observed in plasma samples, with F/T primarily affecting metabolic pathways associated with peptides, low molecular weight lipids, and glycerolipid metabolites in plasma [91]. Pathways found to be impacted in both plasma and serum samples included pathways of cofactors and vitamins such as biliverdin [91]. In contrast, however, plasma samples that underwent thawing during transportation were found to demonstrate significant increases in the concentrations of amino acids, fatty acids, glycerol metabolites, and purine and pyrimidine metabolites [92]. However, several metabolites that demonstrate stability were also identified including glutamine, glycine, isoleucine, leucine, taurine, valine, and TCA cycle intermediates [92]. Serum metabolomic analysis demonstrated both resilience and vulnerability to F/T cycling depending on the metabolites measured. Up to three F/T cycles were found to significantly impact the concentrations of 11 amino acids including histidine, leucine, isoleucine, methionine, phenylalanine, glutamate, tryptophan, valine, taurine, tyrosine, and ornithine [87]. Lipids, however, were found to remain stable across three F/T cycles if samples were stored at −80 °C and underwent lipoprotein fractionation prior to freezing [93]. Serum samples exposed to up to five F/T cycles demonstrated the highest metabolomic alterations within the first three cycles. Notably, this study also reported that donor gender influenced the extent of the metabolomic variation observed following F/T cycling, an aspect that has not been as widely explored. These findings suggest that gender may represent an underexplored factor affecting sample stability and requires further research [94]. Additionally, one study evaluated the impact of the thawing temperature on serum samples and found that shorter thawing times of <20 min at room temperature had less impact on metabolite levels compared to longer thaws in the refrigerator or on ice [40]. Overall, the results suggest that plasma samples are more sensitive to F/T cycling compared to serum samples, suggesting the influence of the sample matrix on resilience against F/T cycling. Based on these findings, no more than three F/T cycles are recommended for both plasma and serum samples with consistent thawing protocols. The effect of F/T cycling on MS metabolomic analysis of synovial fluids is an area for further investigation.

Table 6. Summary of the studies investigating the impact of freeze/thaw cycling on the stability of plasma, serum, and synovial fluid metabolomes.

Study	Sample Type	Variable/s Tested	Sample Size (n)	Analytical Method Used	Conclusions
Jaggard et al. 2021 [88]	Synovial Fluid	Impact of up to five F/T cycles	15	NMR	Up to five F/T cycles had minor effects on the synovial fluid metabolome; however, the concentrations of eight metabolites were significantly impacted including 2-ketoisovalerate, glycosaminoglycan, acetaminophen glucuronide, formate, arginine, glutamate, phenylalanine, and acetaminophen glucuronide.
Damyanovich et al. 2000 [89]	Synovial Fluid	Impact of up to ten F/T cycles	12	NMR	Up to 10 freeze/thaw cycles had no significant impact on overall endogenous metabolite concentrations.
Fliniaux et al. 2011 [68]	Serum	Impact of up to ten F/T cycles	20	NMR	F/T cycling of >5 cycles significantly impacted the stability of the serum metabolome with changes in choline, glycerol, methanol, and ethanol identified as potential markers of F/T cycling.
Pinto et al. 2014 [47]	Plasma	Impact of up to five F/T cycles	15	NMR	Plasma metabolome stability to F/T cycling was sample-dependent; however, generally, the highest variation in the plasma metabolome occurred following four or more F/T cycles; therefore, the analysis of samples that have undergone no more than three F/T cycles is recommended.
Yin et al. 2013 [56]	Plasma and Serum	Impact of up to four F/T cycles	10	Mass Spec (LC-MS)	No significant changes observed in the metabolome up to four F/T cycles. Samples that have undergone the same number of F/T cycles should be used and avoid the use of samples from different F/T cycles.
An et al. 2021 [87]	Serum	Impact of up to three F/T cycles	8	Mass Spec (LC-MS)	Up to three F/T cycles significantly increased the concentrations of 11 amino acids including histidine, leucine, isoleucine, methionine, phenylalanine, glutamate, tryptophan, valine, taurine, tyrosine, and ornithine.
Chen et al. 2020 [94]	Serum	Impact of up to five F/T cycles	99	Mass Spec (LC-MS)	F/T cycling impacted the serum metabolome with the most change occurring in cycles one to three. Following the third F/T, changes to the metabolome were less prominent. Changes to the metabolome can be influenced by donor gender.
Goodman et al. 2021 [90]	Plasma	Impact of up to three F/T cycles	20	Mass Spec (LC-MS)	Up to three F/T cycles did not significantly impact overall plasma metabolite concentrations; however, five metabolites (bilirubin (E, E), dihydroorotate, maltose, glycerol 3-phosphate, and sphingosine) were consistently impacted across all three F/T cycles.
Moran-Garrido et al. 2025 [85]	Plasma	Impact of up to five F/T cycles	40	Mass Spec (LC-MS)	Esterified oxylipins in plasma were more stable for up to five F/T cycles compared to free oxylipins.
Torell et al. 2017 [92]	Plasma (animal)	Impact of F/T during transport	45	Mass Spec (GC-MS)	Metabolomic profiles of plasma were significantly altered as a result of unplanned thawing during transportation, with samples typically presenting increased concentrations of amino acids, fatty acids, glycerol metabolites, and purine and pyrimidine metabolites. Metabolites identified to remain stable included glutamine, glycine, isoleucine, leucine, taurine, and valine, as well as TCA cycle intermediates.
Saito et al. 2014 [91]	Plasma and Serum	Impact of up to 10 F/T cycles	30	Mass Spec (GC and LC-MS)	The plasma metabolome was more sensitive to F/T cycling compared to the serum metabolome. Metabolic pathways impacted in plasma samples were associate with peptides, low molecular weight lipids, and glycerolipid metabolites. Pathways affected in both plasma and serum samples included pathways of cofactors and vitamins such as biliverdin.
Zivkovic et al. 2009 [93]	Serum	Impact of up to three F/T cycles	27	Mass Spec (GC and LC-MS)	Up to three F/T cycles did not significantly impact the stability of lipids. The immediate storage of samples in small aliquots at −80 °C helped reduce variability in the quantitative measurement of lipids caused by the freezing of the sample. Density-based lipoprotein fractionation on serum samples should be carried out prior to freezing, as it can significantly impact lipid concentrations when carried out following sample freezing.
McClain et al. 2021 [40]	Serum	Impact of up to four F/T cycles and thawing temperature	416	Mass Spec (GC-MS and LC-MS)	Short (<20 min) thawing of samples at room temperature had minimal effects compared to longer thawing in the refrigerator or on ice.

Table 6. Summary of the studies investigating the impact of freeze/thaw cycling on the stability of plasma, serum, and synovial fluid metabolomes.

Study	Sample Type	Variable/s Tested	Sample Size (n)	Analytical Method Used	Conclusions
Jaggard et al. 2021 [88]	Synovial Fluid	Impact of up to five F/T cycles	15	NMR	Up to five F/T cycles had minor effects on the synovial fluid metabolome; however, the concentrations of eight metabolites were significantly impacted including 2-ketoisovalerate, glycosaminoglycan, acetaminophen glucuronide, formate, arginine, glutamate, phenylalanine, and acetaminophen glucuronide.
Damyanovich et al. 2000 [89]	Synovial Fluid	Impact of up to ten F/T cycles	12	NMR	Up to 10 freeze/thaw cycles had no significant impact on overall endogenous metabolite concentrations.
Fliniaux et al. 2011 [68]	Serum	Impact of up to ten F/T cycles	20	NMR	F/T cycling of >5 cycles significantly impacted the stability of the serum metabolome with changes in choline, glycerol, methanol, and ethanol identified as potential markers of F/T cycling.
Pinto et al. 2014 [47]	Plasma	Impact of up to five F/T cycles	15	NMR	Plasma metabolome stability to F/T cycling was sample-dependent; however, generally, the highest variation in the plasma metabolome occurred following four or more F/T cycles; therefore, the analysis of samples that have undergone no more than three F/T cycles is recommended.
Yin et al. 2013 [56]	Plasma and Serum	Impact of up to four F/T cycles	10	Mass Spec (LC-MS)	No significant changes observed in the metabolome up to four F/T cycles. Samples that have undergone the same number of F/T cycles should be used and avoid the use of samples from different F/T cycles.
An et al. 2021 [87]	Serum	Impact of up to three F/T cycles	8	Mass Spec (LC-MS)	Up to three F/T cycles significantly increased the concentrations of 11 amino acids including histidine, leucine, isoleucine, methionine, phenylalanine, glutamate, tryptophan, valine, taurine, tyrosine, and ornithine.
Chen et al. 2020 [94]	Serum	Impact of up to five F/T cycles	99	Mass Spec (LC-MS)	F/T cycling impacted the serum metabolome with the most change occurring in cycles one to three. Following the third F/T, changes to the metabolome were less prominent. Changes to the metabolome can be influenced by donor gender.
Goodman et al. 2021 [90]	Plasma	Impact of up to three F/T cycles	20	Mass Spec (LC-MS)	Up to three F/T cycles did not significantly impact overall plasma metabolite concentrations; however, five metabolites (bilirubin (E, E), dihydroorotate, maltose, glycerol 3-phosphate, and sphingosine) were consistently impacted across all three F/T cycles.
Moran-Garrido et al. 2025 [85]	Plasma	Impact of up to five F/T cycles	40	Mass Spec (LC-MS)	Esterified oxylipins in plasma were more stable for up to five F/T cycles compared to free oxylipins.
Torell et al. 2017 [92]	Plasma (animal)	Impact of F/T during transport	45	Mass Spec (GC-MS)	Metabolomic profiles of plasma were significantly altered as a result of unplanned thawing during transportation, with samples typically presenting increased concentrations of amino acids, fatty acids, glycerol metabolites, and purine and pyrimidine metabolites. Metabolites identified to remain stable included glutamine, glycine, isoleucine, leucine, taurine, and valine, as well as TCA cycle intermediates.
Saito et al. 2014 [91]	Plasma and Serum	Impact of up to 10 F/T cycles	30	Mass Spec (GC and LC-MS)	The plasma metabolome was more sensitive to F/T cycling compared to the serum metabolome. Metabolic pathways impacted in plasma samples were associate with peptides, low molecular weight lipids, and glycerolipid metabolites. Pathways affected in both plasma and serum samples included pathways of cofactors and vitamins such as biliverdin.
Zivkovic et al. 2009 [93]	Serum	Impact of up to three F/T cycles	27	Mass Spec (GC and LC-MS)	Up to three F/T cycles did not significantly impact the stability of lipids. The immediate storage of samples in small aliquots at −80 °C helped reduce variability in the quantitative measurement of lipids caused by the freezing of the sample. Density-based lipoprotein fractionation on serum samples should be carried out prior to freezing, as it can significantly impact lipid concentrations when carried out following sample freezing.
McClain et al. 2021 [40]	Serum	Impact of up to four F/T cycles and thawing temperature	416	Mass Spec (GC-MS and LC-MS)	Short (<20 min) thawing of samples at room temperature had minimal effects compared to longer thawing in the refrigerator or on ice.

3.8. Post Storage Handling

Two studies (

Table 7. Summary of the studies exploring the impact of post storage handling on plasma and serum metabolomes.

Study	Sample Type	Variable/s Tested	Sample Number (n)	Analytical Method Used	Conclusions
Santos Ferreira et al. 2019 [65]	Plasma and Serum	Buffer addition and NMR analysis delay	74	NMR	Delays in buffer addition and NMR analysis post storage impacted histidine and phenylalanine concentrations.
Goodman et al. 2021 [90]	Plasma	Delay in analysis following thawing of sample (up to 6 h)	15	Mass Spec (LC-MS)	Storage of plasma samples for up to 6 h in an ice bath following thawing did not significantly impact overall metabolite concentrations, suggesting metabolome stability for up to 6 h.

) were identified that evaluated the effects of delays between sample thawing and final analytical processing on the stability of metabolite profiles in plasma and serum samples [65,90]. One study applied NMR and observed the impact of a 24 h delay in buffer addition and NMR acquisition following thawing; delays resulted in changes in histidine and phenylalanine concentrations in plasma and serum samples [65]. In serum samples, histidine concentrations were found to decrease, and phenylalanine was found to increase following buffer (mean β values relative to reference = −0.00052 mmol/L, p = 2.2 × 10⁻⁴ and 0.0043 mmol/L, p = 2.1 × 10⁻⁵, respectively) and NMR (−0.00094 mmol/L, p = 0.28 and 0.0091, p < 1 × 10⁻²⁶ respectively) acquisition delays. Plasma (EDTA) samples also followed a similar trend as the serum, with histidine concentrations decreasing and phenylalanine concentrations increasing after buffer (−0.013 mmol/L, p = 4.8 × 10⁻¹¹ and 0.0043 mmol/L, p = 2.1 × 10⁻⁵, respectively) and NMR acquisition delay (−0.0044, p = 1.4 × 10⁻³ and 0.0091 mmol/L, p < 1 × 10⁻²⁶, respectively) [65]. The other study applying MS analysis on plasma samples investigated the storage of samples on ice for up to 6 h prior to analysis following thawing. The results revealed no significant changes to the metabolome, suggesting the metabolome remains stable for up to 6 h following thawing if kept cool [90]. However, given the limited number of studies and differences in analytical platforms and experimental designs, further investigations are required before definitive conclusions can be drawn regarding post-thaw handling effects.

4. Discussion

This review examined the impact of pre-analytical variables on the metabolomic profile of biological samples, focusing on plasma, serum, and synovial fluid. Accurate and reproducible metabolomic profiling depends on the preservation of in vivo metabolite concentrations, yet these are easily altered by sample handling factors such as the collection tube type, processing delays, temperature, storage conditions/duration, F/T cycling, and post storage steps. While specific effects varied depending on the analytical platform (NMR, GC- or LC-MS) used and biofluid type, several general patterns emerged.

Studies exploring sample collection agreed that anticoagulants introduce intense signals and interfere with endogenous metabolites. Heparin plasma and plain serum tubes (without additives or gels) were identified as the most compatible for metabolomics workflows, demonstrating the least interference. Delays in sample centrifugation following collection, especially when blood is stored at room temperature, were found to contribute towards major shifts in glycolysis-related metabolites, whilst prompt cooling and processing within two hours mitigated these changes. Centrifugation is a necessary step to obtain plasma and serum; studies testing the impact of centrifugation parameters found they have a relatively minor impact on plasma and serum metabolomes. However, one study identified centrifugation to be a beneficial step in the metabolomic analysis of synovial fluid, where it may not be typically applied. Post-centrifugation delays, even as short as 1–2 h, were found to cause oxidative degradation, with serum samples appearing more vulnerable than plasma. Sample storage conditions were also found to play a critical role: overall, samples stored at −80 °C demonstrated the highest stability both short and long term, whilst storage at −20 °C or room temperature was associated with progressive metabolite degradation. However, further investigations are required to assess the impact of long-term storage at −80 °C on the serum metabolome Additionally, none of the studies included in this review investigated the impact of the storage of the metabolome at −196 °C. F/T cycling of plasma and serum, particularly more than three cycles, was found to induce changes in specific amino acids, lipids, and redox sensitive metabolites, while short thawing times at room temperature were found to be less damaging than prolonged thawing on ice or refrigerator.

Based on the findings of this review, several best practice recommendations emerge (illustrated in Figure 2):

(1)

Plain serum collection tubes and heparin plasma tubes free of gels and citrate additives should be used for the collection of blood.

(2)

Blood samples are recommended to be centrifuged within 2 h of collection and stored on ice during this period.

(3)

Following centrifugation, samples should be frozen immediately at −80 °C and aliquoted to avoid multiple F/T cycles. Where F/T cycling is unavoidable, samples having undergone more than three cycles should be avoided for metabolomic analysis.

(4)

Metabolomic workflows should be standardised across study sites, with adequate documentation of pre-analytical handling conditions to support reproducibility and aid in cross-study comparisons.

Notably, several metabolites were identified to be consistently affected across pre-analytical handling variables. These metabolites were repeatedly impacted across different sample types (plasma and serum), platforms (NMR, GC-MS, and LC-MS), and handling steps. Their consistent appearance highlights both a vulnerability in metabolomic workflows but also a potential opportunity for developing internal quality control markers. These included glycolysis intermediates (glucose, lactate, and pyruvate) [88,89,90,91,92,93], redox-active compounds (histidine, proline, and choline) [95,96,97,98], and amino acids (arginine, ornithine, taurine, and glutamate) [99,100], each differentially affected by different pre-analytical handling conditions. This core set of metabolites may serve as candidate markers of sample quality, whose incorporation into routine quality control pipelines can enable researchers to flag compromised samples, correct for batch effects, or improve reproducibility across studies.

Figure 2. Pre analytical workflow highlighting best practice recommendations for the handling of blood based biofluids and synovial fluid for metabolomic analysis as well as highlighting evidence gaps for synovial fluid.

Figure 2. Pre analytical workflow highlighting best practice recommendations for the handling of blood based biofluids and synovial fluid for metabolomic analysis as well as highlighting evidence gaps for synovial fluid.

Despite these important insights, several limitations emerge in the current literature. The majority of identified studies focused on plasma and serum and addressed only a narrow range of pre-analytical variables. In several cases, findings were based on small sample sets thus limiting the generalisability of these results. Additional heterogeneity in study designs, sample preparation protocols, and analytical platforms make it difficult to synthesise findings quantitatively or compare across studies. Importantly, synovial fluid, a biofluid of growing interest in joint disease research remains under-investigated, with only three papers identified in this review. The absence of studies exploring sample collection variables, pre-centrifugation conditions, and post-centrifugation conditions on the synovial fluid metabolome contrasts with the body of evidence available for plasma and serum, which indicate that these factors can influence the metabolome. The unique biochemical profile of synovial fluid may render it more or less susceptible to degradation than plasma or serum, but this remains largely unknown. Addressing this gap is essential to ensure that the metabolomic analysis of synovial fluid yields reliable, reproducible, and clinically meaningful results.

5. Conclusions

In conclusion, while substantial progress has been made characterising the effects of pre-analytical handling on plasma and serum metabolomes, further work is required to standardise protocols, validate quality control markers, and expand the evidence base to include less studied biofluids like synovial fluid. In addition, we would advocate the inclusion of quality assurance and quality control practices throughout the whole metabolomics pipeline, with the inclusion of pooled quality control samples where possible, following the discussions outlined by Broadhurst and colleagues [101].