Comparison of influenza-specific neutralizing antibody titers determined using different assay readouts and hemagglutination inhibition titers: good correlation but poor agreement

Determination of influenza-specific antibody titers is commonly done using the hemagglutination inhibition assay (HAI) and the viral microneutralization assay (MN). Both assays are characterized by high intra- and inter-laboratory variability. The HAI assay offers little opportunity for standardization. For the MN assay, variability might be due to the use of different assay protocols employing different readouts. We therefore aimed at investigating which of the MN assay readout methods currently in use would be the most suitable choice for a standardized MN assay that could serve as a substitute for the HAI assay. For this purpose, human serum samples were tested for the presence of influenza specific neutralizing antibodies against A/California/7/09 H1N1 (49 sera) or A/Hong Kong/4801/2014 (50 sera) using four different infection readout methods for the MN assay (cytopathic effect, hemagglutination, ELISA, RT qPCR) and using the HAI assay. The results were compared by correlation analysis and by determining the level of agreement before and after normalization to a standard serum. Titers as measured by the 4 MN assay readouts showed good correlation, with high Person's r for most comparisons. However, agreement between nominal titers varied with readouts compared and virus strain used. In addition, Pearson's correlation of MN titers with HAI titers was high but agreement of nominal titers was moderate and the average difference between the readings of two assays (bias) was virus strain-dependent. Normalization to a standard serum did not result in better agreement of assay results. Our study demonstrates that different MN readouts result in nominally different antibody titers. Accordingly, the use of a common and standardized MN assay protocol will be crucial to minimize inter-laboratory variability. Based on reproducibility, cost effectiveness and unbiased assessment of results we elected the MN assay with ELISA readout as most suitable for a possible replacement of the HAI assay