A multiplex immunoassay gives different results than singleplex immunoassays which may bias epidemiologic associations

Abstract

Objectives: Multiplex immunoassays are increasingly used in epidemiologic studies to measure inflammatory factors, however there are few published evaluations of this technology. Our objective was to compare a common multiplex immunoassay to singleplex immunoassays for measuring inflammatory factors, and to examine how combining data from each affects an epidemiologic association.

Design and methods: Plasma IL-1 beta, IFN-gamma, IL-6, and TNF-alpha were measured in 100 samples using a multiplex kit from Mesoscale Discovery (MSD) and singleplex ELISAs from R&D Systems. Separate samples (n=80) were collected to compare multiplex and singleplex assays from MSD. We simulated the effect of combining MSD multiplex and R&D singleplex data on the association between sugar sweetened beverage (SSB) intake and IL-6 in the Health Professionals Follow-up Study (HPFS; n=1314).

Results: Compared to R&D ELISAs, the MSD multiplex proportionally and significantly overestimated IL-1 beta (slope=1.2), and IFN-gamma (slope=2.9) but underestimated IL-6 (slope=0.5). Correlations were ≥ 0.81 except for TNF-alpha (r=0.31). Compared to MSD singleplex, the MSD multiplex proportionally underestimated IFN-gamma (slope=0.7) and TNF-alpha (slope=0.5). Correlations were ≥ 0.96. The association between sugar sweetened beverage intake and IL-6 in the HPFS (+0.16 pg/mL per serving/day, p=0.02, all singleplex) was gradually attenuated as multiplex data made an increasing contribution to the data-set. (+0.09 pg/mL [-45%], p=0.02, all multiplex)

Conclusions: A multiplex immunoassay for inflammatory factors yielded significantly different results than singleplex immunoassays-including those from the same company. Correlations were not consistently high, except among assays from the same company. Such differences may distort epidemiologic relationships if data from both methods are merged.