Cytokine Measurement Consistency

Multiplex cytokine immunoassays are valuable research tools for studying circulating cytokines, chemokines, and growth factors. Lot-to-lot consistency is an important requirement when performing cytokine analysis, with longitudinal studies of immune factors providing insight into immune response, inflammatory response, and disease. When the lot-to-lot performance of measuring cytokine concentrations is not consistent in immunoassays, longitudinal studies are not possible. This is because changes in assay performance such as sample drift can produce confounding or misleading results leading to incorrect conclusions. In this study comparing the performance of a MILLIPLEX^® multiplex cytokine assay to other assays, the focus was on two immunoassay performance indicators related to consistency: standard curve reproducibility and sample measurement.

Read more about

• Multiplexed Analysis of Cytokines with Different Immunoassay Brands
• Sample Preparation
• Immunoassays and Data Analysis
• Standard Curve Reproducibility
• Limit of Detection
• Sample Detectability
• Summary
• Related Cytokine Multiplex Assays
• Request Information

Multiplexed Analysis of Cytokines with Different Immunoassay Brands

The lot-to-lot performance of the MILLIPLEX^® Human Cytokine/Chemokine/Growth Factor Panel A 48-plex immunoassay was evaluated and compared to similar assays from leading competitors, referred to here as Brand X, Brand Y, and Brand Z, using matched human serum and plasma from healthy donors. Brand X and Brand Z are 45-plex immunoassays and Brand Y is a 48-plex immunoassay. Each assay uses magnetic microspheres from Luminex^® technology to simultaneously measure up to 48 cytokines.

Two different lots from each kit were tested with the same matched sample cohort and data were analyzed with Belysa^® Immunoassay Curve Fitting Software. All four of the kits share an overlapping set of 27 inflammatory cytokines and other analytes, including: Eotaxin/CCL11, FGF-2, GM-CSF, GROα, IFNα, IFNγ, IL-1α, IL-1β, IL-1Ra, IL-2, IL-4, IL-5, IL-6, IL-7, IL-8, IL-10, IL-12p70, IL-13, IL-15, IL-17A/CTLA8, IP-10/CXCL10, MCP-1, MIP-1α, MIP-1β, RANTES, TNFα, and VEGF-A.

Sample Preparation

Seven matched human plasma and serum samples were obtained from healthy donors from a commercial vendor (BioIVT, Westbury, NY). For serum samples, the blood was allowed to clot for 30 minutes before centrifugation for 10 minutes at 1,000 x g. The serum was removed and either assayed immediately or aliquoted and stored at -80 °C.

Plasma samples, with EDTA anticoagulant, were centrifuged at 1,000 x g within 30 minutes of blood collection. Plasma was removed and assayed immediately or aliquoted and stored at -80 °C. Frozen samples were thawed completely, vortexed, and centrifuged prior to use, to remove particulates.

Samples were run neat or diluted, according to the respective kit protocols.

Immunoassays and Data Analysis

Multiplex immunoassays were performed in 96-well plates according to each respective product manual. Two different kit lots were obtained from each vendor over the course of a 2-year period and were each analyzed according to the manufacturers’ protocols. All kits were run on the Luminex^® 200™ instrument and data was acquired via xPONENT^® v. 4.3 software.

Standard curve information was obtained from the MILLIPLEX^® Human Cytokine/Chemokine/Growth Factor Panel A protocol and the lot-specific Certificates of Analysis for the Brand X, Brand Y, and Brand Z kits. Data analysis was performed for all immunoassays using the Belysa^® Curve Fitting Software. Figures were prepared in GraphPad Prism and Microsoft Excel.

Standard Curve Reproducibility

Standard curves were generated from each kit lot using averaged Mean Fluorescence Intensity (MFI) from duplicate wells. Standard curve consistency is illustrated by overlaying the curves from both lots. Two representative analytes, shared by all four kits, are displayed in Figure 1. To see the standard curves for more of the shared analytes, download our application note.

Notably, the MILLIPLEX^® Human Cytokine/Chemokine/Growth Factor Panel A maintains a consistent standard curve range from one kit lot to another. In comparison, kits from Brand X, Brand Y, and Brand Z have value-assigned standards, such that each kit lot will have a different standard curve range for every analyte. For those reasons, the standard curves from the other three kits appear slightly staggered along the x-axis.

Relative potency was then determined using Belysa^® Immunoassay Curve Fitting Software. MILLIPLEX^® Human Cytokine/Chemokine/Growth Factor Panel A had a median relative potency of 0.80, compared to median relative potencies of 1.90, 0.669, and 3.09 for Brands X, Y, and Z, respectively. A chart of relative potency is shown for analytes common between the four kits in Figure 2.

Limit of Detection

Limit of Detection (LOD) provides a key indication of assay sensitivity, as the lowest concentration of an analyte that can be reliably detected from background noise. For each analyte the LOD was determined by Belysa^® Curve Fitting software using the following equation:

LOD = 3 𝑥 𝜎 (blank sample) / Slope_II , where Slope_II is the slope of the least squares regression line that best fits the first three standard groups with increasing concentrations starting from the blanks.

Because this calculation is dependent on the variation of the blank, LOD cannot be calculated when the MFI is identical between duplicate blank samples. Therefore, the LOD is listed as “N/A” for standard curves without variation in their blanks. Users may choose to run standard curves in triplicate to increase the chances of obtaining a tangible LOD value for each analyte.

For each of the four bead-based immunoassays tested, the LOD values for the 27 common analytes were compared between each lot and averaged to determine a mean LOD value. Data from 26 analytes from MILLIPLEX^® Human Cytokine/Chemokine/Growth Factor Panel A, 25 analytes each from Brand X and Brand Z, and 27 analytes from Brand Y were incorporated in the analysis, with analytes only omitted if the MFI in the duplicate background wells was identical in the runs from each lot, thereby producing a LOD of “N/A”.

The greatest sensitivity, on average, was observed with MILLIPLEX^® Human Cytokine/Chemokine/Growth Factor Panel A, which showed the lowest mean LOD values for the shared analytes: 2.7 pg/mL for the MILLIPLEX^® multiplex kit versus 5.6 pg/mL for Brand X, 7.1 pg/mL for Brand Y, and 35.0 pg/mL for Brand Z (Figure 3A). Several examples of analyte-specific LOD values are displayed in Table 1.

The %CV of the LOD (Figure 3B) gives an indication of the consistency at the lower end of the curve. As described above, this value could not be obtained for all analytes since runs with identical background MFIs had indeterminate LODs. Thus, only analytes where both runs produced a tangible LOD could be used to calculate the %CV. This included 18 analytes from MILLIPLEX^® Human Cytokine/Chemokine/Growth Factor Panel A, 17 analytes from Brand X, 15 analytes from Brand Y, and 13 analytes from Brand Z. The only kit for which the mean %CV value for the LOD for all eligible analytes was less than 50% was MILLIPLEX^® Human Cytokine/Chemokine/Growth Factor Panel A.

Sample Detectability

Healthy donor serum and EDTA plasma samples were measured on two occasions with the two lots of each kit. The ratio of sample values was assessed between lots to determine which kit gave the most consistent sample measurement (Figure 4A). Additionally, sample correlation between each lot was considered, with R² values generated in Excel (Figure 4B). For each metric, the closer the analytes cluster to the value of “1”, the better the sample ratio or correlation.

Direct comparison of MILLIPLEX^® Human Cytokine/Chemokine/Growth Factor Panel A with Brand X (Figure 5A), Brand Y (Figure 5B), and Brand Z (Figure 5C) is also shown with sample ratios plotted on the x-axis and R² values on the y-axis.

Three examples of sample correlation results between kit lots are shown for IL-13 (Figure 6A), MIP-1β (Figure 6B), and VEGF-A (Figure 6C).

For Research Use Only. Not For Use In Diagnostic Procedures.

Request Information

Have questions about MILLIPLEX^® multiplex assays? Find more information on our MILLIPLEX^® Immunology Panels or fill out the form by clicking on the button below.