Cluster analysis of schistosome-specific antibody responses partitions the population into distinct epidemiological groups

Immuno-epidemiological studies in schistosomiasis continue to generate large amounts of immunology data, whose analysis requires sophisticated statistical approaches. Here cluster analysis, is used to explore the relationship between immune responses and observed epidemiological patterns of schistosome infection in two Zimbabwean communities. Analysis of cross-sectional antibody data (IgA, IgE, IgG1, IgG2, IgG3, IgG4 and IgM directed against Schistosoma haematobium soluble egg antigen (SEA)) showed that cluster analysis partitioned the data into distinct epidemiological groups based on all seven antibody isotypes (defined by age, infection intensity, treatment status and history of infection) confirming an already known partitioning based on IgA/IgG1 production. All treated participants (children) changed cluster membership following treatment from clusters where IgA was the predominant antibody to clusters where IgG1 predominated. There was a differential distribution of IgE and IgG4 between clusters consistent with the recently proposed balance between T-helper cells (Th) 1, Th2 and regulatory T cells. The analysis suggested that naturally acquired anti-schistosome responses associated with resistance to infection were different from drug-induced responses associated with resistance to re-infection. Furthermore, the analysis suggested that parasite-specific immune responses were dynamic. The analysis conducted on data from participants resident in the S. mansoni endemic area who were all children partitioned the data into two clusters, one with predominately pre-treatment data (cluster 1) and the other with post-treatment data (cluster 2). The antibody profiles of both clusters were most similar to the profile of people with a modified Th2 response. Following treatment 43% of the children in cluster 1 moved to cluster 2, which generally had higher levels of antibodies. A detailed study of factors determining which children moved between the clusters showed that it was mostly the older, infected children who moved to cluster 2. The results of the analysis are discussed in terms of current theories of the development of acquired immunity to schistosomiasis. The relative merits of cluster analysis as a statistical tool for analysing these data are also discussed.