The Value of Neuraminidase Inhibition Antibody Titers in Influenza Seroepidemiology Free

(See the Major Article by Huang et al on pages 347–57.)

In an article in this issue of the Journal of Infectious Diseases, Huang et al report the results of an influenza serologic study conducted in New Zealand in 2015 [1]. The study period covered a season dominated by influenza A(H3N2) and B, which had a moderate impact based on influenza-like illness and severe acute respiratory illness rates [2]. Huang et al found that some individuals had evidence of influenza A(H3N2) virus infections based on rises in neuraminidase inhibition (NAI) antibody titers without corresponding rises in hemagglutination inhibition (HAI) titers. When data on ≥4-fold rises in NAI titers were used in addition to ≥4-fold rises in HAI titers, estimates of the cumulative incidence of influenza A(H3N2) infections in 2015 were around 45% higher than when based on ≥4-fold rises in HAI titers alone.

Serologic studies have been used for decades to estimate the proportion of the population that have been infected in influenza epidemics. One of the earliest such studies was reported by Widelock et al in 1959, who used HAI assays on sera collected from adults in 1957–1958 by the syphilis laboratory in New York [3]. They reported an increase in seroprevalence of A(H2N2) antibodies from 4% to 35% during the first wave and to 70% by the end of the second wave of that pandemic in New York, with evidence of high rates of infection across all age groups. More recently, serologic studies were conducted in many countries during and after the 2009 pandemic, and one metaanalysis reported that an overall estimate of the cumulative incidence of influenza A(H1N1)pdm09 infection from those studies was 24% [4].

Serologic studies of influenza have typically analyzed HAI titers against a single reference strain to provide information on cumulative incidence in an epidemic [4]. In some cases, the virus microneutralization assay has been used in addition to, or in place of, the HAI assay [5, 6]. Huang et al went further and used a second assay, in this case NAI, and were able to identify an additional 11% of the cohort as infected based on ≥4-fold rises in NAI titers but not HAI titers [1]. The observation that a fraction of infected persons do not have ≥4-fold rises in HAI titer has been made before. For example, Miller and colleagues noted that, in 2009, 10% of confirmed H1N1pdm09 cases did not have a ≥4-fold rise in HAI titer at convalescence [7], and Freeman et al noted that this proportion was as high as 50% for polymerase chain reaction (PCR)-confirmed A(H3N2) infections in 2009–2013 [8]. Cauchemez et al noted that a greater fraction of paired sera showed 2-fold rises in HAI titers than 2-fold drops, and inferred that some of the 2-fold rises might have occurred in infected persons [9].

A number of problems with the HAI assay, particularly for A(H3N2) viruses, limit interpretation of current HAI data. The reference antigens used in serological studies are often egg-grown vaccine viruses, known to elicit higher titers than their cell-grown counterparts. Cell-grown A(H3N2) viruses are harder to grow due to interference from the neuraminidase [10]. In contrast, egg-grown A(H3N2) viruses are less vulnerable to neuraminidase interference but tend to have acquired mutations that alter antigenicity [11]. In the study by Huang et al[1], the primary reference virus used was the egg-grown A/Switzerland/9715293/2013-like virus, which appears to have predominated in New Zealand in 2015 [2]. However, genetic data tracked on the nextflu website (https://nextstrain.org/flu/) [12] suggest that circulating viruses at that time may have been A/Hong Kong/4801/2014-like, with most clustering within the 3C.2a genetic group, rather than the 3C.3a genetic group to which A/Switzerland/9715293/2013 belongs. Although there is currently no clear consensus on the concordance of antigenic and genetic drift, it is not so surprising that some persons infected with A(H3N2) viruses might have lower rises in HAI titers to the A/Switzerland/9715293/2013-like strain. Comparison data from cell-grown reference antigens representative of the genetic groups that circulated in 2015 may have provided a better indicator of exposure to circulating A(H3N2) viruses and their hemagglutinin antigens.

As with the hemagglutinin (HA) gene, the neuraminidase (NA) gene shows considerable genetic diversity. However, assignment into genetic groups is based on the HA, and the phylogeny of the NA gene may follow a different pattern, with clustering in different genetic groups. It is, therefore, potentially remiss to only examine infections against one reference strain, the sensitivity of which may be limited if circulating strains have drifted sufficiently from the reference strain. A better approach may be to assess responses to a panel of HA (cell-grown) and NA reference antigens representative of circulating genetic clusters. Furthermore, it is known that infections by a particular strain will not only generate detectable responses to that strain, but will boost the titers of antibodies to older strains [13]. As antibody generation diminishes with increasing age or number of prior vaccinations [14], exposures that boost previously acquired antibodies but do not generate new ones may still be indicative of infection for the purposes of seroepidemiology. This would allow identification of a higher proportion of all infections or, conversely, may reveal that so-called serodiscordant participants (ie, those who seroconverted for NA but not HA) are in fact seroconcordant for different pairs of HA and NA.

The observation that NAI titers can be used to identify infections leads us to suggest another potential application. It is well known that HAI titers are of limited value in identifying infections in vaccinated persons [15]. Current split-virion and subunit inactivated influenza vaccines stimulate strong rises in HAI titers. However, they have varying, but generally low, neuraminidase content and may elicit detectable rises in NAI titers in a small proportion of vaccinees [16, 17]. Therefore, could NAI titers be of value in identifying influenza virus infections in vaccinated persons? Inability to estimate the cumulative incidence of infections in the vaccinated segment of the population was a limitation of the Huang et al study [1]. An important next step would be to examine changes in NAI titers in vaccine failures. Even if this does not have direct applicability to standard inactivated influenza vaccines, trials of vaccines that have no NA content, such as FluBlok (Sanofi Pasteur), might be able to use the NAI as a serological indicator of infection.

Following from the previous point, a more general limitation of the present study is the lack of information on changes in NAI titers after documented influenza, and specifically documented influenza with A/Switzerland/9715293/2013-like viruses. When embarking on a longitudinal serologic study, it can be valuable to have a companion study in which acute and convalescent sera are collected from virologically confirmed influenza cases, to provide information on the distribution of antibody titers rises in confirmed infections [7].

In conclusion, the study by Huang et al is important because it shows, for the first time, that NAI titers can complement HAI titers in serological studies of the cumulative incidence of influenza virus infections in populations. Information on changes in HAI and NAI titers after confirmed infections would aid interpretation of these results, and indicate whether NAI titers would be informative in serologic studies of influenza A(H1N1) and B. Two future directions are clear. First, determining whether other measures of antibody titers could further improve the resolution of serologic studies, for example, adding data on HAI titers against a range of contemporary strains. Second, determining whether NAI titers could provide reliable information on infections in vaccinated persons, given that HAI titers are unsuitable for this purpose.

Notes

Financial support. The WHO Collaborating Centre for Reference and Research on Influenza is supported by the Australian Government Department of Health.

Potential conflicts of interest. B. J. C. has received research funding from Sanofi Pasteur for a study of influenza vaccination effectiveness, and honoraria from Roche and Sanofi Pasteur. S. G. S. reports no potential conflicts. Both authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.

References