Computational prediction of vaccine strains for human influenza A (H3N2) viruses

Abstract

Human influenza A viruses are rapidly evolving pathogens that cause substantial morbidity and mortality in seasonal epidemics around the globe. To ensure continued protection, the strains used for the production of the seasonal influenza vaccine have to be regularly updated, which involves data collection and analysis by numerous experts worldwide. Computer-guided analysis is becoming increasingly important in this problem due to the vast amounts of generated data. We here describe a computational method for selecting a suitable strain for production of the human influenza A virus vaccine. It interprets available antigenic and genomic sequence data based on measures of antigenic novelty and rate of propagation of the viral strains throughout the population. For viral isolates sampled between 2002 and 2007, we used this method to predict the antigenic evolution of the H3N2 viruses in retrospective testing scenarios. When seasons were scored as true or false predictions, our method returned six true positives, three false negatives, eight true negatives, and one false positive, or 78% accuracy overall. In comparison to the recommendations by the WHO, we identified the correct antigenic variant once at the same time and twice one season ahead. Even though it cannot be ruled out that practical reasons such as lack of a sufficiently well-growing candidate strain may in some cases have prevented recommendation of the best-matching strain by the WHO, our computational decision procedure allows quantitative interpretation of the growing amounts of data and may help to match the vaccine better to predominating strains in seasonal influenza epidemics. Importance: Human influenza A viruses continuously change antigenically to circumvent the immune protection evoked by vaccination or previously circulating viral strains. To maintain vaccine protection and thereby reduce the mortality and morbidity caused by infections, regular updates of the vaccine strains are required. We have developed a data-driven framework for vaccine strain prediction which facilitates the computational analysis of genetic and antigenic data and does not rely on explicit evolutionary models. Our computational decision procedure generated good matches of the vaccine strain to the circulating predominant strain for most seasons and could be used to support the expert-guided prediction made by the WHO; it thus may allow an increase in vaccine efficacy.

FIG 1

Data analysis in influenza seasons with replacement of the predominant antigenic strain. Columns represent results for the 2002-2003 Northern Hemisphere influenza season (2003N), the 2004 Southern Hemisphere influenza season (2004S), and the 2005 Southern Hemisphere influenza season (2005S). Panels A, D, and G give the maximum likelihood phylogenetic trees inferred for the 2003N, 2004S, and 2005S influenza seasons, respectively. Colors represent known antigenic strains identified by key amino acid substitutions reported in the literature and used before (32): SY97/MO99/PA99 (light blue), FU02 (orange), WE04 (violet), CA04 (green), WI05 (dark blue), and BR07 (yellow). Horizontal bars indicate clades that contain viral isolates sampled in the relevant influenza season. Panels B, E, and H depict AD plots computed for the 2003N, 2004S, and 2005S influenza seasons, respectively. Alleles with a frequency of >90% or with a frequency increase of ≥10% in the relevant influenza season are shown in color (colors were arbitrarily chosen). Panels C, F, and I show the antigenic trees inferred for the 2003N, 2004S, and 2005S influenza seasons, respectively. Colors are as in panels A, D, and G. Branch lengths represent the maximum of the two branch weights (up- and down-weights). Weights for terminal branches and branches leading to subtrees without an isolate used as antiserum are set to 0 antigenic units for the sake of clarity.

FIG 2

Performance evaluation of antigenic allele-based computational prediction of vaccine strains (AACP) for human influenza A (H3N2) based on the combination of AD plots and antigenic trees and comparison with recommendations by the World Health Organization. In contrast to Table 1, now antigenic strains are shown for the seasons in which they were predominant; thus, for both methods, predictions are shown for the year in which the vaccine was made available, not for the year before (when it was to be produced). The top row shows a succession of predominant and antigenically distinct strains. The second row shows the recommendations made by the AACP, while the third row shows the recommendations made by the WHO, both for the 2003-2004 NH influenza season and for the season 2007-2008 NH influenza season. This figure illustrates that the predominant strains were predicted correctly in six out of nine seasons by the AACP and the recommendation made by the WHO matched four out of nine seasons. Both recommendations included one false positive, as the recommended 140E HA allele (A/Oklahoma/8/2004) predicted by AACP and the WE04 strain recommended by the WHO did not belong to the positive sample. Note that a vaccine update was necessary only if in the previous season a different strain was predominant.