Bead-based multiplex immuno-assays for cytokines, chemokines, growth factors and other analytes: median fluorescence intensities versus their derived absolute concentration values for statistical analysis.

Within the scientific literature, analyses of data from bead based multiplex immunoassays are based on either median fluorescence intensities (MFI) or derived absolute concentration values (ACV) but no consideration of which set of data is the most appropriate for analysis has been published. Here we look at the variance of MFI versus their ACV from the expression of 14 analytes in plasma, using 6 commercially available kits, across 177 patients, recorded at two time points and the associated analyte standards. In total 60 micro titre plates were used resulting in 4965 MFI. In doing so we develop a new background subtraction procedure that reduced by 50% the number of out-of-range values observed in our data set. Using a linear mixed-effect model, which normalizes for assay-to-assay variation, MFI produced similar significant differences than that observed using absolute concentration values. We show that subtracting analyte blanks produces 15% negative MFI resulting in uncertainty of the data being analysed. We argue for analysis of protein expression values MFI are generally a better choice than absolute concentration values. It is argued that analyte standards are not required on each plate, or not at all, in multi-plate experiments, but knowledge of the concentration curve and the range of MFI values that fall within the limits of this curve for each analyte is required. The significance of using MFI over concentration values for the life scientist means higher statistical power and lower costs.