Visualisation and analysis of patterns in serological data using "antibody landscapes"

In this thesis I develop and implement “antibody landscapes”, a method to profile immunity against a pathogen as a function of antigenic differences between a range of strains. Theoretically applicable to any antigenically variable pathogen and measurement of immunity, the work here focusses on antibody-mediated immunity against the A/H3N2 influenza subtype. Applying the methodology to study annual serum samples from individuals monitored for influenza infection over a period of six years, patterns of influenza immunity were found to be remarkably distinct and maintained almost unchanged over time in the absence of influenza exposure. Upon infection, the initial response is strikingly antigenically broad, including responses against viruses far beyond the extent of cross-reactivity observed after a primary infection. Analysis of two vaccination cohorts, one receiving an antigenically advanced vaccine strain and one a more typical vaccine strain choice, revealed many of the same patterns of response as seen with infection. Antigenically advanced vaccination generated greater responses against later strains but surprisingly, due to equivalent boosting of prior immunity, this came at no cost to responses generated against contemporary or older strains. Exploring in more detail the development of immunity over time, analysis of a cohort of children demonstrated that - in contrast to adults with diverse exposure histories - antibody responses to a first infections were remarkably similar in pattern and magnitude. Interestingly, for second infections, although post-infection antibody titres against circulating strains were comparable to those after first infections, overall cross-reactivity of the response against future antigenic variants appeared to be diminished. The findings here underline the significant role prior-immunity plays in affecting the response to new exposures and the importance of understanding it. An important conclusion is that by failing to account for it, current approaches to influenza vaccine strain selection may be suboptimal and pre-emptive vaccine strain updates may improve overall vaccine efficacy where immunity to current strains already exists in the population. Building on the work presented here should help to optimise strain choice and vaccine efficacy even further.

Part funded by an MRC studentship