There is currently limited data on the antibody response following H1N1-2009 infection as detected on serological assays. In our study, antibody titers increased rapidly in the first 2 weeks, and collection of acute samples less than 5 days from illness onset, and convalescent samples more than 2 weeks after illness onset, maximized the proportion of RT-PCR confirmed infections which seroconverted on hemagglutination inhibition assays. In addition, we showed that more than 80% of RT-PCR confirmed H1N1-2009 cases seroconvert using hemagglutination inhibition and virus microneutralization assays, and demonstrated the development of cross-reactive antibodies to other influenza A strains following H1N1-2009 infection. These key characteristics of the antibody response have implications on the interpretation of serological assays for pandemic H1N1-2009.
Our finding, on the rapid increase of detectable antibodies by hemagglutination inhibition in the first two weeks, is fairly similar to what Miller and colleagues observed.
In addition to what they found, we were able to demonstrate how the timing of the blood sampling with respect to symptom onset affects the characteristics of paired serological assays. To optimize assay performance, we found that convalescent sera should be collected at least 2 weeks after illness onset. Testing samples collected less than 5 days after onset of illness maximized the proportion observed to seroconvert, although the cut-off point for acute sera was less critical. We also demonstrated that, with appropriately timed samples, 82% and 89% of the RT-PCR confirmed cases seroconverted on the HI and VM assays respectively. These results are fairly close to those described by Cowling et al, who reported that slightly less 80% and more than 95% of 19 RT-PCR confirmed pandemic H1N1-2009 cases with paired serology had four-fold or greater increase in titers on HI and VM assays respectively.
However, unlike Cowling who suggested that subjects given oseltamivir treatment early in the course of their disease might have a diminished convalescent antibody response, we did not find this to be so for oseltamivir use, nor for any of the participant characteristics investigated. It must be noted, however, that our study was under-powered to investigate key features of importance, such as co-morbid conditions which might depress immune function since such subjects were under-represented in our study.
Finally, we also documented the development of cross-reactive antibodies to other strains of influenza A following RT-PCR confirmed H1N1-2009 infection. While we cannot rule out the possibility that recent H3N2 or seasonal H1N1 infections preceding the episode of H1N1-2009 infection might have accounted for some of our findings, we note that there was little circulation of seasonal H1N1viruses throughout the study period, and H3N2 activity had also largely waned by the time our H1N1-2009 cases were sampled.
Heterotypic cross-reactive antibodies following suspected infection with pandemic influenza strains have been previously reported. For instance, one study on the 1968 influenza pandemic in Singapore reported that some patients with clinically diagnosed influenza who seroconverted to A/Singapore/1/68 (H3N2), the pandemic strain circulating in Singapore then, also seroconverted to A/Singapore/1/57 (H2N2), the causative strain of the previous pandemic about a decade earlier.
It is unclear as to the extent these cross-reactive antibodies are protective to the respective influenza A strains in vivo, but such antibodies could be one pathway for the heterotypic protection observed in an epidemiological study which showed how adults previously infected with H1N1 influenza in the years prior to the 1957 pandemic were protected during the 1957 H2N2 pandemic.
Our observations on cross-reactive antibodies also have implications for interpreting data from serologic surveys which assess relative infection rates of H3N2 and H1N1-2009 while the two strains continue to co-circulate.
Our study was based on participants aggregated from three separate sources which differed in their population characteristics and timing of sample collection. Due to low follow-up rates for the 2 to 4 week and 6 to 8 week samples, hospitalized cases were overrepresented in earlier samples and underrepresented in later samples. We suggest that the main bias that would arise from aggregating such subjects for analysis relates to the differences in the timing of baseline and follow-up samples for the three different groups, and we hence attempted to reduce such biases by restricting analyses to participants with baseline and follow-up sera collected within an appropriate time-frame. However, in doing so, the study sample size available for detecting differences in serological response by participant characteristics was reduced.
In spite of the above limitations, our study provides important information profiling key aspects of the serological response following infection with the pandemic H1N1-2009 strain. Our findings also have implications for the serological diagnosis of H1N1-2009 based on paired serum samples, as well as the conduct of future serological surveys for influenza. Further studies are needed to understand the significance and mechanisms of the cross-reactive antibody responses to other influenza strains that occur in some individuals following H1N1-2009 infection.