Plasma versus serum for extracellular vesicle (EV) isolation: A duel for reproducibility and accuracy for CNS‐originating EVs biomarker analysis

Blood‐derived extracellular vesicles (EVs) are a popular source of biomarkers for central nervous system (CNS) diseases, but inconsistencies in isolation and analysis hinder their clinical translation. This review summarizes recent studies that investigate the impact of different anticoagulated plasma and serum on the yield, purity, and molecular content of EVs. Specifically, the studies compare ethylenediaminetetraacetic acid (EDTA), citrate, heparin plasma, and serum and highlight the risk of contamination from platelet‐derived EVs. Here, I offer practical guidelines for standardizing EV isolation and analysis, recommending the use of plasma anticoagulated with acid‐citrate‐dextrose (ACD) or citrate followed by EDTA and heparin, subgroup analyses for samples from different biobank repositories, and avoiding serum and plasma‐to‐serum transformation. Other factors like illness, age, gender, meal timing, exercise, circadian timing, and arm pressure during blood draw can alter EV signatures. Yet, how these variables interact with different anticoagulated plasma or serum samples is unclear, necessitating further research. Furthermore, whether the changes are dependent on the isolation or quantification methodology remains an area of investigation. Importantly, the perspective emphasizes the need for consistency in experimental methodologies to improve the reproducibility and clinical applicability of CNS‐originating EV biomarker studies. The proposed guidelines, along with ongoing efforts to standardize blood sample handling and collection, may facilitate the development of more reliable and informative CNS‐originating EV biomarkers for diagnosis, prognosis, and treatment monitoring of CNS diseases.

Most studies investigating CNS-originating EV biomarkers have used serum (Dutta et al., 2021;Dutta, Hornung, Taha, Biggs, et al., 2023;Jiang et al., 2021;Taha & Ati, 2023) and/or plasma (Fiandaca et al., 2015;Pulliam et al., 2019;Shi et al., 2014;Taha & Ati, 2023) as media for immunoprecipitation of the EVs directly or as part of a two-step procedure where EVs are first isolated and CNS-originating EVs are then immunoprecipitated using specific antibodies. However, potential differences between these two media and their implications for downstream analysis have not been thoroughly investigated despite previous studies showing that preanalytical variations can affect the number of largersized and smaller-sized EVs isolated from blood (Baek et al., 2016;György et al., 2014;Jayachandran et al., 2012;Karimi et al., 2022;Lacroix et al., 2012;Liu et al., 2018;Palviainen et al., 2020;Zhang et al., 2022).
Blood consists of erythrocytes, white blood cells, and plasma, which is the liquid component that carries micro and macromolecules, including EVs. The coagulation cascade is a dynamic process that requires a balance between procoagulant activity and protease inhibitors. It involves the activation of multiple enzymes and cofactors through three pathways: intrinsic, extrinsic, and common.
Detailed information on the coagulation cascade can be found elsewhere (Davie et al., 1991). To prevent the coagulation cascade in plasma, biobank repositories typically use anticoagulant molecules such as ethylenediaminetetraacetic acid (EDTA), citrate, or heparin, sometimes in combination with other molecules. EDTA and citrate function by chelating calcium cofactors required for holoenzyme formation and the progression of the coagulation cascade, while heparin activates the antithrombin enzyme to inhibit blood clot formation. If blood clot formation is allowed to occur and the coagulation factors are depleted, plasma is then called serum.
Therefore, the main objective of this review is to assess the impact of preanalytical variables on the concentration of ptEVs and EVs derived from either plasma (pEVs) or serum (sEVs). These preanalytical variables include the choice of anticoagulation agent mixed with plasma, the time of preparation, centrifugation methodology, the nature of transport, the number of freeze-thaw cycles, storage conditions, temperature, and the type of collection tube. These preanalytical variations are often under-reported in biomarker studies of CNS-originating EVs (Gomes & Witwer, 2022), hindering the reproducibility and the generalizability of the results to real-world and clinical settings. Two recent meta-analyses evaluating α-synuclein levels and diagnostic accuracy of biomarkers in CNS-originating EVs highlighted high heterogeneity across CNS-originating EVs studies (Taha & Ati, 2023;Taha & Bogoniewski, 2023).
This review aims to address the impact of preanalytical variables on the reproducibility and accuracy of EV isolation and make recommendations for studies using CNS-originating EVs as biomarkers in plasma and serum. I will identify significant differences between these media and provide recommendations for the isolation of CNS-originating EVs to improve the reproducibility and accuracy of biomarker analysis. Additionally, I will highlight a few remaining questions that have not yet been addressed in the context of cellspecific EV isolation, especially CNS-originating EVs.

| ANTICOAG UL ATED PL A S MA VER SUS S ERU M
A comprehensive summary of each study, along with corresponding recommendations, is presented in Table 1. George et al. conducted the first study to evaluate preanalytical variations in the early 1980s (George et al., 1982). They used ultracentrifugation to isolate putative larger-sized pEVs and sEVs from platelet-free (depleted using centrifugation at 5000 g for 15 min) plasma mixed with the anticoagulant acid-citrate-dextrose (ACD) and serum, respectively, to quantify differences in putative largersized ptEVs (GP IIb/IIa + ) concentrations. Immunoelectrophoresis revealed that despite platelet depletion, ptEVs were still present in pEV and sEV preparations. Notably, higher concentrations of at least 10-fold ptEVs were found in serum compared to ACD plasma, which aligns with the observation that converting plasma to serum Significance Extracellular vesicles (EVs) transmit vital messages between cells such as ongoing trauma. The blood-brain barrier (BBB) separates the central nervous system (CNS) from the blood, complicating the study of brain biochemistry in CNS diseases. CNS-EVs can cross the BBB, making them promising biomarkers. However, clinical application is hindered by inconsistencies in isolation and analysis. This perspective suggests guidelines for standardizing EV isolation from plasma and serum, stressing consistent methodologies to boost reproducibility in CNS biomarker research.
By adhering to these guidelines, scientists can create more reliable biomarkers for diagnosing, monitoring, and treating CNS diseases, benefiting the public and researchers. activates platelets (Siljander et al., 1996), leading to the release of ptEVs. The study did not quantify differences in total EV count.
Following these observations, Jayachandran et al. investigated the effects of various preanalytical factors on the count of putative larger-sized (Annexin-V + ) ptEVs (CD42a + ) and pEVs using plateletrich and -free (depleted using two cycles of centrifugation at 2500 g for 15 min each) anticoagulated plasma samples (Jayachandran et al., 2012), respectively. Using flow cytometry, they found out that Ca 2+ chelating agents, such as EDTA and citrate, decreased ptEVs in comparison to protease inhibitors, like heparin. Moreover, plasma incubated with chelating agents at temperatures of 22 or 33°C for 15 min to 1 h reduced the concentration of both ptEVs and pEVs, whereas heparin plasma increased the fraction of both ptEVs and pEVs significantly after 1 h, with larger increases observed at 22°C.
Therefore, to ensure the accuracy of the total number of EVs isolated, plasma samples should be kept at lower temperatures and pEVs should be isolated as soon as possible. Additionally, differences in centrifugation force and time affected the count of ptEVs recovery in plasma inconclusively-suggesting that centrifugation time and speed should be kept constant across trials. Importantly, the count of ptEVs and pEVs recovered after EV isolation in chelating agents and protease inhibitors was found to decrease after freeze-thaw cycles, whereas up to three freeze-thaw cycles prior to EV isolation did not influence EV content.

Similarly, Gelibter et al. recently conducted a study in which they
isolated EVs from platelet-free human EDTA anticoagulated plasma and murine microglia culture using size exclusion chromatography (SEC) and found that freezing the EV suspension at −80°C resulted in a reduction of EV count and purity, as well as increased occurrence of fusion events observed as double-fluorescently tagged EV populations (Gelibter et al., 2022). These results suggest that EVs isolated from plasma should be used immediately for downstream analyses and that freeze-thaw cycles should be avoided after EV isolation has occurred. It should be noted, however, that the content of EVs in plasma appears to remain stable up to three freeze-thaw cycles, provided that EVs have not yet been isolated.
During the same time, Lacroix et al. evaluated the influence of anticoagulant molecules (EDTA, citrate, and heparin) on putative larger-sized pEVs (Annexin-V + ), erythrocyte-derived EVs (eryEV; CD235a + ), and ptEVs (CD41 + ) isolated from platelet-free (depleted using two cycles of centrifugation at 2500 g for 15 min each) plasma and quantified using flow cytometry (Lacroix et al., 2012). The blood was drawn and allowed to incubate for one hour at room temperature (RT) before being centrifuged and separated into three layers.
Findings suggested that EDTA significantly affected the number of pEVs counted after one hour, followed by heparin, while citrate did not. Furthermore, the authors claimed that EDTA affected both er-yEVs and ptEVs count, but the data were not reported. Therefore, the study recommended the use of citrate tubes as they are less likely to interfere with the quantification of pEVs up to 2 h when freshly collected before separating the plasma from other blood components. Additionally, the study mimicked the unsupported nature of blood transportation commonly seen in hospitals, where blood is transported in a plastic bag. The results showed that these conditions significantly altered the isolated pEV components, resulting in 80% larger counts and 30% decreased clotting times. Interestingly, when blood was transported in supported conditions, horizontally positioned tubes gave similar results to unsupported transportation, whereas vertically positioned tubes prevented the differences observed in pEVs count.
Moreover, the study examined the effects of centrifugation force and time, storage time, and freeze-thaw cycles on the composition of EVs. The results showed that the force of the initial centrifugation and the precentrifugation elapsed time before initial centrifugation affected the EV composition, while storage time and freeze-thaw cycles did not result in significant differences. Based on these findings, the study recommended several guidelines for pEV isolation and analysis: (1) the use of citrate tubes for blood collection, (2) the use of large needles for blood drawing and discarding the first 1 mL, (3) limiting the time delay before the first centrifugation to ≤2 h, and (4) maintaining the tubes vertically during transportation. However, it was unclear from the study whether the findings supporting recommendations (2) to (4) were based on observations of citrated plasma or other tubes.
Additionally, no direct comparisons were made among the anticoagulation molecules for recommendations (2) to (4). It can likely be assumed that these findings were obtained from citrated tubes.
In support of previous findings, a follow-up study (György et al., 2014) compared the effects of different anticoagulated plasma tubes on the quantification of putative larger-sized (Annexin-V + ) ptEVs (CD42 + ) and pEVs using platelet-free (depleted using two cycles of centrifugation at 2500 g for 15 min each) plasma. The study used flow cytometry to show that the protease inhibitor heparin increased the number of ptEVs and pEVs compared to the chelating agent citrate, with the lowest number and standard deviations found in ACD tubes.
The results were replicated using a Zymuphen assay, a special procoagulant assay, as a proxy for ptEVs and resistive pulse sensing for the size and diameter of EVs isolated. Furthermore, incubations of these  , 2019). However, it is worth noting that this method is considered to be much more expensive and significantly less pure than other EV isolation methods, such as ultracentrifugation and SEC (Liangsupree et al., 2021).
To test whether the conclusions may also be applied to smaller-  The study also compared two different centrifugation protocols for the removal of platelets, using a single centrifugation cycle at 1800 g for 6 min or two centrifugation cycles at 2000 g for 30 min, and found that the two-step platelet depletion protocol was more effective in reducing background signals, particularly in EDTA plasma tubes. Additionally, the study's findings (Baek et al., 2016) were consistent with those of Jayachandran et al. (2012), demonstrating that up to four freeze-thaw cycles did not affect the EV content in any plasma or serum tubes only before EV isolation, further reinforcing the importance of avoiding freezing isolated EVs. A comparison of EVs isolated using ExoQuick versus those found in the crude plasma/serum showed a reduction of tetraspanin exosomal markers only evident in heparin plasma tubes. No differences were found in EDTA, citrate, or serum samples, suggesting that ExoQuick in these samples was not useful for the specific enrichment of smaller-sized EVs. It is unclear why this is the case and requires further investigation. Although the results from this study suggested that heparin plasma tubes are useful when the goal is to target smaller-sized EVs, the choice is offset by an increased risk of platelet activation (Gao et al., 2011). In contrast to the previously mentioned studies analyzing putative larger-sized EVs (György et al., 2014;Lacroix et al., 2012), the authors concluded that EDTA plasma tubes were the most troublesome to work with when evaluating smaller-sized EVs, while heparin tubes seemed to be the most plausible.
Following, Palviainen et al. aimed to find differences among plasma anticoagulated with EDTA, ACD, citrate, or serum using platelet-free (depleted using two cycles of centrifugation at 2500 g for 15 min each) samples with contemporary methods (Palviainen et al., 2020). Nanoparticle tracking revealed higher concentrations of smaller-sized EVs (size = 0-200 nm) in serum compared to all anticoagulated plasma samples. SDS-PAGE analysis showed similar protein content patterns between ACD and citrated pEVs that differed from sEVs and EDTA pEVs. Additionally, as George et al. reported higher ptEVs and pEVs in serum than in ACD plasma (George et al., 1982), this prompted Palviainen et al. to use ultracentrifugation to isolate EVs and quantify ptEVs (CD61 + ), eryEVs (CD235a + ), and putative larger sized EVs (Annexin-V + ). In all the samples, the proportion of ptEVs to eryEVs was lower, but more ptEVs were found to be present in serum compared to ACD and EDTA plasma, as determined by both flow cytometry and procoagulant activity as a proxy for ptEVs.

Flu-SEC and Western blot experiments yielded similar findings, fur-
ther strengthening the robustness of the results. Therefore, when studying CNS-originating EVs for biomarkers, subgroup analyses should be conducted according to whether the blood was collected as serum or anticoagulated with EDTA, heparin, and/or ACD/citrate.
In another study, Liu et al. used proteomics and showed that there is an increased level of platelet-derived metabolites in serum but not in plasma (Liu et al., 2018). This difference was also shown to be dependent on the type of clotting tube used for serum, further weakening the usage of serum as a medium for EV isolation and biomarker discovery due to possible reproducibility concerns that are strenuous to control.
In another study focusing on the isolation of putative smallersized EVs through differential ultracentrifugation, Zhang et al.
compared differences among EDTA, citrate, and heparin anticoagulated plasma versus serum isolated from C57BL/6J mice . They reported that sEVs and pEVs were different in two ways: (1) the sEVs contained more particles within the same diameter when compared to pEVs; while no inter-anticoagulation differences were found, as assessed by nanoparticle tracking and (2) protein contents of sEVs and pEVs were substantially different with larger platelet-associated proteins found in sEVs, as assessed by proteomic analyses using liquid chromatography-tandem mass spectrometry. Isolated sEVs and pEVs from human blood using differential ultracentrifugation and further Western blotting also revealed higher platelet-associated proteins in sEVs versus pEVs. Zhang et al. particularly showed that findings between EVs isolated from mice cannot be generalized to serum/plasma isolated from humans and that sEVs from both mice and humans contain more ptEVs and ptEVs-associated proteins than their pEVs counterparts. The study concluded that the usage of plasma is the most ideal for accurate EV measurements, and comparisons between sEVs and pEVs align with both Liu et al. (2018) and Palviainen et al. (2020), discouraging the use of serum for EV isolation.
Furthermore, Karimi et al. compared platelet-free (depleted using two cycles of centrifugation at 2500 g for 15 min each) anticoagulated EDTA and ACD plasma and serum for differences in putative smaller-sized EVs that are positive for at least one of the three tetraspanin markers and those that originated from ptEVs using various methodologies such as cushion, SEC, density cushion + SEC, or immunoprecipitation directly with one of the tetraspanin markers (Karimi et al., 2022). The number of CD9 + and CD41a + EVs found in EDTA plasma was higher compared to ACD plasma and serum, suggesting they originated from ptEVs. Additionally, the number of EVs in the blood that were double-positive for CD63 and CD81 was very low. CD63 + EVs were more abundant in serum, while CD81 + vesicles were less abundant in both plasma and serum. However, CD81 + EVs likely did not originate from platelets, as they were not positive for CD41a + . EDTA plasma had more residual platelets than ACD-plasma or serum, and two centrifugation steps were necessary to reduce the number of platelets before isolating EVs. Interestingly, the study BPIs on a diverse set of metrics reflecting blood sample quality, ex-vivo generation of blood-cell-derived EVs, EV recovery, and EVassociated molecular signatures. Although the study is too long for detailed discussion in this review, the authors recommended several guidelines in support of previous findings.  (Shah et al., 2018), age (Eitan et al., 2017), gender (Noren Hooten et al., 2022), prandial state (Mork et al., 2018), exercise (Pierdoná et al., 2022), circadian timing (Yeung et al., 2022), and arm pressure during blood collection (Jenkins et al., 2013). However, it remains unclear how these factors affect different anticoagulated plasma or serum samples, as such, more research is needed.

| FURTHER CONS IDER ATIONS
In light of the discussed results above, a few questions remain. The possible contamination of ptEVs in these preparations becomes a serious concern and may confound the results seen and influence the lack of reproducibility that currently exists in the CNS-originating biomarker field. Furthermore, no study to date has tested the possibility of using CNS-immunoprecipitation antibodies (e.g., anti-L1CAM) to cross-capture ptEVs and/or any of the present differences for Addressing these challenges is paramount to harnessing the full potential of these biomarkers and advancing the field.
Additionally, plasma can be readily transformed to serum by using thrombin or silica beads followed by a high-speed centrifugation step to precipitate the formed fibrin pellet. However, this also activates platelets (Siljander et al., 1996). Activated platelets increase the release of a variety of biomolecules, including both largesized (i.e., ectosomes) and smaller-sized (i.e., exosomes) EVs (van der Meijden & Heemskerk, 2019). Regardless, in the blood, ptEVs also serve as the most abundant source of EVs (Tao et al., 2017). It is important to also note that EVs are thought to have a "corona," which consists of biomolecules carried on their outer lipid bilayer surface (Tóth et al., 2021), as well as engage in fusion events (Morandi et al., 2022).
Studies focusing on CNS-originating biomarkers often prefer treating plasma with thrombin before EV isolation. This preference arises because polymer-based precipitation techniques, such as ExoQuick, when applied to plasma, can lead to the precipitation of an insoluble fibrin pellet. This pellet, a by-product of fibrinogen present in plasma, can interfere with the purity and yield of isolated EVs. However, this process can lead to increased platelet activation, resulting in the release of biomolecules and ptEVs, which can cause two significant issues: (1) the released biomolecules may attach to the "corona" surface of CNS-originating EVs (Palviainen et al., 2020) and (2) fusion of ptEVs with CNS-originating EVs, which may lead to loss of the original message sought after and the results becoming nothing more than a random chance (Morandi et al., 2022).
Moreover, no literature exists to explain how the precipitated fibrin pellet may impact the purity of CNS-originating EVs. It is currently unknown whether the precipitated fibrin pellet may bind to specific proteoforms of the protein of interest, thereby affecting downstream analysis. It is also unclear if this is influenced by the specific anticoagulant molecules (i.e., EDTA, citrate, or heparin) used for plasma.

| CON CLUS ION
Standardized experimental methodologies should be used whenever possible for all CNS-originating EVs biomarker studies. Isolating cell-specific EVs is best done using ACD plasma tubes followed by citrated plasma and EDTA tubes, while heparin plasma tubes are discouraged. If samples must be obtained from different biobank repositories, subgroup analyses should be conducted separately for each source from where the sample was acquired to account for possible confounding factors. Also, the EV suspension should never be frozen under any circumstance after isolation, while freezing the samples before isolation is acceptable. Lastly, the use of serum for EV isolation should be avoided whenever possible, and plasma should not be treated with thrombin without careful characterization of the resulting fibrin pellet. Current efforts by ISEV (Théry et al., 2018) and others (Clayton et al., 2019) are aiming to standardize blood sample handling and collection to generate a harmonizing methodology for the field. Studies are also encouraged to report their detailed methodologies using EV-TRACK (EV-TRACK Consortium et al., 2017). It is hoped that further consideration of these critical issues may further help the standardization and reproducibility of CNS-originating EV biomarker studies.

AUTH O R CO NTR I B UTI O N S
Hash Brown Taha conceptualized, designed, and wrote the manuscript.

ACK N OWLED G M ENTS
I thank the International Society for Extracellular Vesicles, in particular, Edit Buzás and Rienk Nieuwland, for their inspiration.

FU N D I N G I N FO R M ATI O N
The authors declare that no funds, grants, or other support were received during the preparation of this manuscript.

CO N FLI C T O F I NTE R E S T S TATE M E NT
The authors have no relevant financial or nonfinancial interests to disclose.

PE E R R E V I E W
The peer review history for this article is available at https:// www.webof scien ce.com/api/gatew ay/wos/peer-revie w/10.1002/ jnr.25231.