The development of accurate models that permit prediction of biological responses upon perturbation has the potential to increase our mechanistic understanding of pathophysiology and contribute to the development of improved therapeutics. The human immune system provides an excellent context for developing such systems biology approaches: many immune cells and molecular components are readily accessible from blood, permitting collection of samples from individuals across multiple time-points, followed by in depth data generation and analyses. Furthermore, there is an increasing understanding that the immune system and inflammation contribute to the pathogenesis of multiple disorders. These include not only those classically considered to involve the immune system such as autoimmune and infectious diseases, but also cancer, cardiac disease, diabetes, obesity, neurodegeneration, and other chronic illnesses affecting a large segment of the population (Germain and Schwartzberg, Nat Immunol. 2011). Thus, a more comprehensive and quantitative understanding of how immune responses are orchestrated, together with identification of predictive molecular and cellular parameters of effective vs. damaging responses, could have major implications for the prevention and treatment of diverse diseases. To this end, I have coordinated one of the initial studies from the NIH Center for Human Immunology designed to help build a data base of normal human variation (the human immunome) and understand how variation in immune states contributes to immune reponses and disease (Tsang, Schwartzberg et al, submitted;DIckler, H. et al, Ann New York Acad Sci, 2013). As a first step towards modeling human immunity, we have analyzed immune parameters in depth both at baseline and in response to perturbation with influenza vaccination. Peripheral blood cell transcriptomes, serum cytokines, influenza titers, frequencies of 126 cell subpopulations, and B cell responses were assessed before and after vaccination in 63 individuals and used to develop a computational framework to dissect inter- and intra-individual variation and build predictive models of post-vaccination antibody responses. Similar to other vaccine studies we have been able to show post-vaccination gene expression signatures that correlated with vaccine responses, but furthermore have linked these to the expansion of B cell plasmablast populations. Importantly, using an approach that accounts for the influence of pre-existing serology, age, ethnicity and gender, we demonstrated that much of the post-vaccination responses to Influenza vaccination and predictive signatures are heavily influenced by pre-vaccination titers. Strikingly, independent of age and pre-existing antibody titers, we found that accurate models could be constructed using pre-perturbation parameters alone, which were validated using data from independent baseline time-points. Most of the parameters contributing to prediction delineated temporally-stable baseline differences in immune cell populations across individuals, raising the prospect of immune health monitoring before intervention (Tsang, Schwartzberg et al, under revision). The framework we detail provides a potential resource for studying human immunity in health and disease.
|Tsang, John S; Schwartzberg, Pamela L; Kotliarov, Yuri et al. (2014) Global analyses of human immune variation reveal baseline predictors of postvaccination responses. Cell 157:499-513|
|Dickler, Howard B; McCoy, J Philip; Nussenblatt, Robert et al. (2013) The National Institutes of Health Center for Human Immunology, Autoimmunity, and Inflammation: history and progress. Ann N Y Acad Sci 1285:133-47|