Qualification of the hemagglutination inhibition assay in support of pandemic influenza vaccine licensure

Abstract

Continued outbreaks of highly pathogenic avian influenza over the past decade have spurred global efforts to develop antivirals and vaccines. As part of vaccine development, standard methods are needed for determining serum antibody titers in response to vaccination. Hemagglutination inhibition (HAI) assays are appropriate for assessing the immunogenicity of pandemic influenza vaccines in support of license approval. We demonstrate that a rigorous qualification of the HAI assay for H5N1 influenza virus, evaluating for precision, intermediate precision, linearity, range, specificity, and robustness, satisfies the intent of regulatory guidance for assay validation despite the lack of availability of specific reference standard antigens and antisera.

FIG. 1.

Test plate setup prior to the addition of virus. VC, virus controls; CC, cell controls.

FIG. 2.

Agglutination patterns observed with HAI plate tilted at a 60° angle.

FIG. 3.

Scatter plot and regression line for all data combined for predicted versus actual titers of positive control serum using a log₁₀ scale.

FIG. 4.

Scatter plots and regression lines by operator and day for predicted versus actual titers of positive control serum using a log₁₀ scale.

FIG. 5.

Performance of the control sera over time, as tracked by our historical database, indicates that carefully stored controls provide consistent results over a 2-year period that encompassed hundreds of assays and multiple analysts. Arrow A, implementation of a serum mixing step; arrow B, modification of the dilution factor used for the positive control serum.

FIG. 6.

Frequency distribution of transformed GMT collected for the precision assay meeting or failing qualification assay acceptance criteria across all 3 days.

FIG. 7.

Seven operators analyzed 72 nonclinical samples at 45 min, 90 min, or 135 min. (A) GMT from the data of all operators at each time point was converted to a log₁₀ scale and linear regression against predicted titers was performed. (B) GMT from the data of all seven operators at each time point is graphically displayed for titers of 80 or lower to underscore the increased frequency of detecting these lower titers with an increase in incubation time.