|Home | About | Journals | Submit | Contact Us | Français|
Cytokines are humoral regulatory molecules that act together in immunologic pathways. Monitoring cytokines and their variations within physiologic ranges is critical for biomarker discovery. Therefore, we evaluated the performance characteristics of 72 analytes measured by multiplex cytokine immunoassay, with an emphasis on the differences of analytes measured in serum compared to plasma, and, in plasma, on the impact of anticoagulants on the cytokine measurement.
We used fluorescent bead-based (Luminex) immunoassay kits to simultaneously measure 72 analytes. We tested serum and plasma samples from 11 matched donors. Plasma samples were anti-coagulated with sodium heparin, sodium citrate dextrose and ethylene diamine tetra-acetic acid (EDTA), respectively.
Of the 72 cytokines, 12 were undetectable in all types of specimen samples. Nineteen analytes, including PDGF-bb, IL-4, IL-8, IL-9, FGF-b, PAI-1, CXCL-5, CCL-5, CD40L, EGF, VEGF, IL-2ra, IL-3, SDF-1a, PCT, MCP-3, GIP, IL-16 and fibrinogen, showed significant differences between measurements in serum and all types of plasma, regardless of anticoagulant. Among plasma samples, 10 analytes (eotaxin, SCGF-b, MCP-1, SCF, MIP-1b, VEGF, RANTES, PDGF-b, PAI-1 and ITAC) showed significantly higher concentrations in heparinized plasma compared to citrated and EDTA plasma. IP-10, and CTAK were the only 2 cytokines that presented different concentrations in citrate and EDTA plasma.
With their small volume, low cost per test, and multiplex capacity, Luminex-based cytokine assays have enormous potential utility for screening in epidemiologic studies. In our study, we showed that many cytokines’ concentrations differed between serum and plasma samples, and that different anticoagulants used in preparation of plasma samples also affected the measurement of some cytokines. There was no optimal sample preparation that was clearly superior for the measurement of all analytes measured. Ultimately, the utility of cytokine measurement, as biomarker or to monitor the immune system, will depend on attention to detail in the collection and processing of samples in addition to assay precision.
Cytokines and chemokines are small molecules with large roles in modulating immune reactions . The measurement of cytokines permits assessment of the immune response in a disease or in response to a specific therapy. A particular cytokine may exert effects on multiple different types of cells, and multiple cytokines can exert various effects on a single cell type. Historically, most measurements of cytokines and chemokines have been made using single-plex enzyme-linked immunosorbent assay (ELISA) methods. This approach to cytokine measurement, while quite valuable, limited the scope of analytes that could be examined in any given trial, and thus studies of the cytokine network interactions were quite limited, often precluding comprehensive assessment of cytokines and chemokines for such studies.
Multiplex immunoassay technologies to monitor immune responses by measuring biomarkers such as cytokines have become increasingly available; they have the capability to measure up to 100 simultaneously in a minimal amount of biological sample . Determining the accuracy and reliability of the detection method used for each analyte in multiplex assays is key for the use of these assays in clinical trials. While analytical variables, such as matrix effects induced by multiplexing analyte assays, have been widely studied [3, 4], less attention has been paid to pre-analytical variables that can also affect multiplex cytokine assays. These can include assorted elements of the “preparasome”, such as how the specimen is stored, diurnal variations, and type of specimen (such as serum or plasma). Among pre-analytical variables, little is known regarding how the measurement of most cytokines is affected by the anticoagulant used for collection of blood from healthy individuals. In the few studies that have begun to address this issue [5, 6], only a small number of analytes were examined. As a result there are little, if any, data for most cytokines regarding how measurements are affected by anticoagulants. In the present study, we analyzed seventy-two analytes in serum and plasma from healthy individuals. Plasma collection used several anticoagulants, including sodium heparin, acid citrate dextrose, and EDTA. Our primary objective was to evaluate anticoagulant impact on the measured levels of cytokines in the peripheral circulation and to determine the best sample preparation for most cytokine measurements.
Samples were obtained concurrently from 11 healthy donors enrolled in protocol 09-H-0239, approved by the NHLBI institutional review board, and written informed consent was obtained from all participants. To assess cytokine levels in different preparations, we used matched samples collected the same day from the same patient. Table 1 summarizes the specifics of each type of sample collection tubes. Serum samples were isolated 2 hours after collection of blood using serum separating tubes (SST) (Beckton Dickinson, San Jose, CA, USA). Plasma was collected in 3 different ways depending on the type of anticoagulant present in the tube; i) Sodium heparin (“green top tube”; 15 IU/mL +/− 2.5 IU/mL), ii) ethylene diamine tetra-acetic acid (EDTA) (“lavender top tube”; 1.5 mg/mL +/− 0.35 mg/mL), iii) sodium citrate dextrose (“yellow top tube”; 0.109 mg/mL). Specimens used for this study were subjected to no more than 2 freeze/thaw cycles.
We used 6 different kits from 2 manufacturers; 5 kits were from Bio-rad (Hercules, CA, USA) and 1 kit from R&D (Minneapolis, MN, USA). All assays were performed according to Bio-Rad and R&D kit procedures.
Median fluorescence intensities were collected on a Luminex-100 instrument (Luminex, Bio-rad), using Bio-Plex Manager software version 6. Standard curves for each cytokine were generated using the premixed lyophilized standards provided in the kits. Cytokine concentrations in samples were determined from the appropriate standard curve using a 5 Point-regression to transform mean fluorescence intensities into concentrations. Each sample was run in duplicate and the average of the duplicate was used as the measured concentration. Three cytokines were measured in the kits from different vendors, IL-8, eotaxin and IP-10, and their results were presented separately.
Data obtained with serum/plasma from one donor were considered as one experiment (n). Results are shown as the mean of the observed concentration for each person, the standard of the mean and the percentage of coefficient of variation of the mean. Data were compared by nonparametric analyses with Wilcoxon’s matched pairs test, in which the median was used to calculate significant difference, p values significance level was set as p=0.05 and the actual p values were indicated for each series of experiments. Plasmaheparin was used as a control, as it is the most common type of anti-coagulant used for plasma preparation.
Statistical calculations were performed using Bio-plex data pro software (version 1.0 (Bio-rad)) and when serum was used as control instead of plasmaheparin, p values are showed in the figures. Heat maps were generated after normalization of the data, using log2 transformation of the observed concentration. The normalized data were then expressed as percentages of the highest value and thus, for each cytokine, with color gradient from low to high concentrations (green to red).
Numerous differences were observed among the cytokine measurements obtained with serum and the plasma samples collected with different anticoagulants. Table 2 displays a summary of the results obtained for each kit, independent from one another. There was no single method of specimen preparation that was clearly superior for the measurement of all 72 analytes tested. Of these, the following twelve were below the limit of detection in all four types of sample; IL-1a, IL-1b, IL-2, IL-5, IL-10, IL-13, IL-15, GM-CSF, LIF, M-CSF, TNF-a and TNF-b. The failure to detect these analytes in samples obtained from healthy volunteers was consistent with data previously reported by others [2, 7]. Further analysis on the influence of anticoagulants on the measurement of these analytes was therefore not possible.
Comparisons between serum and plasma were conducted by: 1) comparing serum to all plasma types, regardless of anticoagulant, and 2) comparing serum to each type of anti-coagulated plasma.
Thirty-two analytes (CTACK, GLP-1, SAA, CRP, Leptin, C-peptide, SAP, Haptoglobin, Ferritin, Insulin, TRAIL, Resistin, A2M, IFN-a2, MIF, Glucagon, IL-6, IFN-g, IL-7, Ghrelin, G-CSF, MIG, Visfatin, GROa, IL-1ra, MIP-1a, MIP-1b, IL-12 (p70), IL-18, tPA, HGF, IL-17) displayed no statistically significant differences when measured in serum or any type of plasma.
By contrast, 19 analytes, including PDGF-bb, IL-4, IL-8, IL-9, FGF-b, PAI-1, CXCL-5, CCL-5, CD40L, EGF, VEGF, IL-2ra, IL-3, SDF-1a, PCT, MCP-3, GIP, IL-16 and fibrinogen, showed statistically significant differences between measurements in serum or all types of plasma, regardless of anticoagulant, however the values measured were not equivalent in all types of plasma. The majority yielded higher values in serum than plasma (Fig. 1A), with the exception of IL-2ra, IL-3 and SDF-1a, which were not detectable in serum (Fig. 1B, Table 2) and PCT, MCP-3, IL-16, GIP and fibrinogen, which yielded lower values in serum than plasma (Fig. 1C, Table 2).
The remaining 9 analytes displayed statistically significant differences when measured in serum compared to at least one type of plasma sample, but not to all types of plasma samples (SCGF-b, RANTES, IL-17, eotaxin, b-NGF, IL-12 (p40), MCP-1, SCF, IP-10) (Table 2, Fig. 2).
Significant differences in the measurement of numerous analytes were observed in plasma samples obtained using different anticoagulants. EDTA and citrate yielded similar, but not identical, measurements of most of the 72 analytes assayed, with disparities found only for IP-10 and CTAK (Fig. 3). Plasma obtained using heparin was found to be significantly different from citrate and EDTA for the measurement of 10 analytes; eotaxin (R&D), SCGF-b, MCP-1, SCF, MIP-1b, VEGF, RANTES, PDGF-b, PAI-1 and ITAC showed significantly higher concentrations in heparin plasma compared to citrate and EDTA plasma. CTAK, tPA, b-NGF showed significantly lower concentrations in heparin plasma compared to citrate and EDTA plasma.
A heat map summarizing merged data from each type of samples was generated after normalization of the concentration using the log2 of the concentration (Fig. 4). In green, we confirmed that some cytokines were not detectable in all type of specimens, IL-1a, IL-1b, IL-2, IL-5, IL-10, IL-13, IL-15, GM-CSF, LIF, M-CSF, TNF-a and TNF-b, identified with an arrow bellow the heat map. The heat map also allowed us to rapidly identify which type of sample will provide the higher measurement, as for PAI-1, PDGF-bb, RANTES (marked with a yellow circle).
An 8-plex kit from R&D was run using serum and 3 types of plasma from 20 donors (different donors from those used with the Bio-Rad kits). These data revealed numerous significant differences as noted in Table 2 and Figure 5. This kit also shared three analytes - IL-8, eotaxin and IP-10 - which were measured in Bio-rad kits. Interestingly, for these analytes, the changes observed in the R&D kits were different from those seen in the Bio-rad kits (Fig. 5). IL-8 detection was consistent in both kits, showing a higher yield in serum compared to all plasma. Eotaxin and IP-10 presented discordant results, in both concentration values and yield in sample type. IP-10 from R&D could not be recovered from serum, and eotaxin from R&D was better detected in heparin plasma samples. With both Bio-rad kits, we observed a statistically significant difference for IP-10 between citrate and serum, and eotaxin in EDTA plasma showing lower yield compare to the other 3 types of samples.
We assessed variation in the measurement of circulating analytes in blood samples prepared by spontaneous clotting to serum and with three different methods of anticoagulation to create plasma. Regardless of the type of samples used, the best way to limit problems is to immediately start separating the plasma or serum (post clotting) from the cellular elements. An important goal in optimization of multiplex assay is to reduce the matrix effect by using the appropriate sample diluent, thereby decreasing non-specific antibodies interactions . Apart from the matrix effect, we studied the importance of the anticoagulants used for plasma.
The coagulation cascade necessary to produce clotting, and hence serum, is a complex biochemical process and it is not surprising that measurements of analytes made in plasma are generally not reflective of serum levels and thus are not interchangeable. Several studies have shown that plasma samples and serum collected from subjects at the same time yield different measurements of some cytokines [2, 7, 9]. Serum preparation includes removal of fibrinogen, platelets and other circulating proteins from the plasma that could influence the presence or detection of an analyte on such a sample. Also, during the cascade of coagulation, activation of cellular elements can release inflammatory mediators, that may affect cytokine levels [10, 11].
The reason for the differences in cytokine concentrations observed among the different types of plasma is not known. EDTA and citrate both remove calcium to prevent coagulation while heparin acts instead by binding to antithrombin III. Indeed, we observed only two differences in cytokine measurements between EDTA and citrate, while more differences were noted when comparing heparin to both EDTA and citrate.
Previously published work showed difference in CRP and IL-6 when comparing plasma to serum , but we did not observe such higher levels of IL-6 and CRP in plasma. It is difficult to speculate on the reason why we have different results, and the nature of the antibodies used in this assay cannot be ruled out, as the clones of antibodies used for IL-6 and CRP detection in the Bio-rad kits are not provided by the manufacturer.
In agreement with the literature, our results indicate that analytes such as CD40L, which are secreted by platelets upon activation during the coagulation reaction, are found at a higher concentration in serum [13–17]. It is been previously reported that EDTA plasma is not appropriate for measurement of platelet-derived cytokines .
Our results emphasize the need to consider experimental design carefully when a study requires measurement of cytokines in the peripheral circulation. The quality of cytokine measurements begin when the sample is collected and the choice of anticoagulant should be carefully considered. Additional factors such as sample preparation and time and temperature of storage have been reported to be important as well (19). Under no circumstances should cytokine measurements in a single study use samples collected in different anticoagulants or a combination of plasma and serum. Ultimately, the utility of cytokine measurement will depend on rigorous attention to detail in the collection and processing of samples in addition to assay precision.
Conflict of interest
All authors have no conflict of interest.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.