Background
Better characterization of cellular processes and cytokine pathways in a variety of diseases ranging from allergy and autoimmunity to cancer has lead to new treatments that use monoclonal antibodies which specifically block components of the human immune system including cytokine pathways [
1‐
6]. These new therapeutic strategies, which modulate inflammatory processes of the immune system, can induce major changes in the downstream cytokine milieu. Indeed, the aftermath of the TGN1412 phase I clinical trial in March 2006 revealed that the life threatening adverse events were the consequence of a rapid onset severe cytokine storm [
7,
8]. This example underscores the importance of monitoring cytokines during experimental therapies which are based on or could influence cytokine pathways or cytokine producing cells.
Cytokines are small secreted extra-cellular signaling (glyco-) proteins which regulate cell-mediated immune responses. They are effector molecules that can instantly alter the quality of the immune response. The effect of a particular cytokine on a given cell depends on the cytokine, its extra cellular abundance, the presence (or absence) of the complementary receptor on the cell surface, and downstream signals activated by receptor binding [
9]. As cytokines reflect the local or systemic inflammatory milieu, they could serve as biomarkers for potential clinical effect of the therapeutic interventions.
As cytokines act in networks, measurements of single cytokines is of limited value, emphasizing the need for simple, reliable, cost effective, and reproducible technology for the measurement of multiple cytokines. Several methodologies have been developed and employed for quantification of secreted cytokines. Immunoassays such as ELISA are currently the most commonly used techniques to quantify cytokines due to the high specificity and sensitivity [
10]. Built on the same principle, more rapid, automated, and high throughput methods have been developed [
11]. More recently a bead-based multiplex immunoassays (MIA) with the FlowMetrix (currently know as xMAP
tm technology, Luminex, Austin TX USA) has been increasingly used for detection of multiple cytokines in a single sample [
12].
A number of parameters can affect adequate and reliable measurements of cytokine levels in biological specimens collected in a (multicenter) clinical trial including the timing of sampling, sample handling and storage, and even the choice of plasma or serum (various blood collection tubes). In some cases, such as inflammatory diseases, numerous endogenous plasma proteins such as heterophilic antibodies, soluble receptors, complement, immune complexes, lysosyme, collectins (lectins) and some acute phase proteins can also interfere with immunoassays such as MIA and ELISA [
13].
We and others have previously shown that technical prerequisites for an "in-house" multiplex immunoassay have done comparison studies with ELISA's. In this study we set out to describe parameters which are critical for obtaining accurate cytokine measures from clinical samples, when using a multiplex cytokine detection platform, such as Luminex.
Discussion
Cytokines are involved in the effector phase of all inflammatory diseases, and thus, could serve both as biomarkers for disease severity and as targets for therapy. Anti cytokine therapy was first introduced in the early eighties of the last century, with the first successful application of an anti cytokine therapy trial involving anti-IFN-alpha antibodies in patients with Rheumatoid Arthritis [
19]. Since then, several monoclonal antibodies against cytokines or cytokine receptors as well as other immune modulating drugs are used for a variety of diseases [
1‐
6].
Because of the interconnectivity of immunoregulatory circuits, blocking or activation of specific components of the immune system can result in major changes in the cytokine milieu, which may lead to unexpected side effects. As an example, the development of systemic lupus erythematosus-like symptoms during anti TNFα therapy in Rheumatoid Arthritis patients is due to accumulation of another cytokine, IFN-α, a direct consequence of the TNFα blockade [
20,
21]. Careful monitoring of cytokines is necessary to assess the impact of drugs that influence the immune response, particularly during early phase clinical trials. As multiple cytokines may be effected, techniques for measuring panels of cytokines are of most value.
The microsphere-based multiplex immunoassay measurement allows the simultaneous analysis of multiple secreted cytokines in a variety of bodily fluids [
12]. This technology can detect up to 100 different mediators using as little as a 50 μl sample volume [
11,
22,
23], making this technology very useful in clinical trials, especially when volumes are limited.
For a multiplex assay to perform adequately the conditions should be optimal for every individual component. However, many clinical trial designs underestimate the impact of parameters which can impact reliable cytokine measurements such as sample handling and storage, and quality control of the platform used for cytokine detection.
First, timing of sampling is important. Cytokines are under neuroendocrine control and thus have a diurnal rhythm [
24,
25]. Many pro-inflammatory cytokines, such as IL-1, IL-6, TNFα and IFNγ are linked to melatonin and peak early in the morning [
26]. Rheumatoid arthritis patients have increased IL-6 levels in the morning which correlate with clinical symptoms [
27,
28]. Besides diurnal rhythms, exercise also impacts cytokine levels in blood, as after exercise, among others, muscle cells release IL-6 and which will result in temporarily elevated cytokine levels [
29,
30].
Secondly, proper sample handling and storage is critical for reliable measurement of circulating cytokines. When sampling takes place plasma and serum should be separated as soon as possible. In this study we kept our samples at room temperature and froze plasma and serum sample at -80°C within 1 hour after blood draw. A delay of sample processing containing cellular components will lead to different cytokine expression profiles by either degradation, absorption, or cellular production of cytokines [
31]. Furthermore, a careful consideration should be made for the use of blood collection tubes during trial design. During clotting or by centrifugation, platelets can release various cytokines such as IL-6 and CXCL8 [
32]. Furthermore cytokine binding to, or release from their soluble receptors can result in an under- or overestimation during spiking assays. For instances specific binding and release of IL-1, IL-6, and IL-18 towards their soluble receptor is well documented [
33,
34]. This phenomenon can at least partly account for the large variation found in our spiking assays (figure
2). We show here that the use of various blood collection tubes significantly affects cytokine measurement, with sodium heparin tubes showing most consistent cytokine recovery.
Cytokines should be stored below -80°C for longer time. Although degradation will occur also at this temperature after longer periods of storage (figure
3), sample storage at temperatures of -20°C degradation occurs at faster pace [
35]. Furthermore multiple freeze - thawing cycles also affect cytokine recovery in clinical samples. Although most cytokines are stable in a high protein matrix such as plasma, during the first freeze - thawing cycle an additional freeze - thawing cycle should be avoided at any time (figure
4).
The third critical factor is the quality control of the platform used. Obviously, assay performance is strongly dependent on the quantity of antibodies used as well matrix effects. We and others showed that quality control of different components (e.g. recombinant proteins, sample diluents, microsphere coupling etc.) of the multiplex immunoassay can be performed and implemented easily in a in-house assay [
15,
22,
36]. In commercially available kits internal QC samples are mostly lacking. However some commercial kits suggest to use several dilutions of the recombinant standard curves as an spiked in QC sample [
37]. Though, the amount of recombinant standards and diluents for construction of calibration curves are sufficient for a single kit and can not be used for long-term monitoring of performance [
38]. With the use of an in-house assay, standard curves can either be constructed with or calibrated on NIBSC/WHO standards. However, when comparing different commercial assays using these NIBSC/WHO standards, large differences in observed cytokine concentrations are found [
39,
40]. Use of different antibodies (clones) is only part of the explanation for this phenomenon. Other, more likely culprits are matrix effects, e.g. because of heterophilic antibodies that are often present in human inflammatory diseases [
41]. We have shown previously that removing antibodies with protein-L followed by blocking with 10% rat/mouse serum is sufficient to reduce such matrix effects [
14] where others demonstrated that blocking with fetal bovine serum (FBS), was effective in only a minority of the samples [
42]. Since FBS does not contain immunoglobulins, this underscores the possibility that proteins, other than antibodies, can interfere with the assay. Other matrix effects, such as protein content and pH can also influence the assay performances and should be standardized when using this assay for clinical trial design [
11,
43]. For samples with high protein content such as plasma or serum most commercial assays require dilutions for ideal assay performance, while an in-house assay condition can be set in such a way that samples do not have to be diluted.
The necessity of sample dilution is a major drawback, since it proportionally reduces assay sensitivity. This has major impactions for detection of cytokines such as IL-4, which is already biological active in the lower pg/ml range. Another disadvantage for dilution is that cytokines can be bound to their soluble receptors or binding proteins, resulting in biological non-active cytokines. This binding can, at least partly, be reversed when dilutions are applied, thus resulting in over estimation of the true cytokine concentration [
34,
44].
However, previous studies have shown that without addressing matrix effects in undiluted samples false positive findings are found in up to 90% of healthy individuals which could be reversed by incubating with rodent serum [
22]. However blocking with rodent serum alone is not sufficient to remove false positive values from patients which are positive for auto antibodies such as rheumatoid factor [
11]. Therefore in this study, we applied a relative small dilution (10 v/v %) with rodent serum after protein-L incubation to avoid a-specific binding of matrix proteins.
All points described concerning the biological properties of cytokines, such as long time stability and freeze thaw cycles are not only relevant multiplex immunoassays, but also be applied to other technology used for cytokine measurements in a clinical (trial) setting. Multiplex immunoassays have many advantages for biomarker development in clinical settings. It can measure multiple mediators at once and has the flexibility to easily add any new target protein to the assay. This study underlines that multiplex immunoassays have many advantages for the use in clinical trials, provided that a strict quality control system is in place.
Authors' contributions
WdJ performed research and analyzed data together with KB and GTR. BJP and VSM supervised the study. WdJ, BJP and VSM wrote the article. All authors had full access to all data, read and approved the final manuscript.