Statistical modelling to determine the influence of vaccine dose and prior Mycobacterium tuberculosis exposure on antigen-specific T cell responses

This dissertation investigates the effects of two subunit vaccines H1:IC31 and H56:IC31 as well as prior M.tb sensitization on the immune responses of three cohorts of South African adolescents and adults. The primary outcomes are frequencies of antigen-specific CD4 T cells expressing different combinations of immunological markers over three time points. Two M.tb antigens are investigated: Ag85B and ESAT-6. The dissertation compares the results produced by the standard procedures that would typically be employed in the immunology research community to investigate these aims with the results produced by employing a mixed effect modelling approach. Not only is it of interest to investigate whether the results agree, but also to investigate the difference in inference that one can make and whether the mixed effect modelling approach is able to provide greater insight into the data. Methods typically employed by the immunology community that are used in this thesis are non-parametric pair-wise tests and the data analysis pipelines mixture models for single-cell assays (MIMOSA) and combinatorial polyfunctionality analysis of single cells (COMPASS). For the mixed effect modelling approach, generalized linear mixed effect models with various hierarchical structures as well as latent variable models are employed. Results suggest that 5 μg of the vaccine induces the strongest immune response. The mixed effect modelling approach showed good potential in terms of depth of analysis and ease of interpretation, however many model assumptions were violated making inference difficult. The standard approaches where much more cumbersome to implement and interpret and resulted in significant multiple testing concerns.

Grantor dc:publisher.institution

Department of Statistical Sciences

Year dc:date.issued

2024

Author dc:creator

• Williams, Kelly

Advisors dc:contributor.advisor

• Little, Francesca
• Nemes, Elisa
• Gela , Anele

• Biostatistics

Language dc:language.iso

en

Handle dc:identifier.uri

http://hdl.handle.net/11427/41193

OAI identifier oai:identifier

oai:open.uct.ac.za:11427/41193