Modelling longitudinally measured outcome HIV biomarkers with immuno genetic parameters.
According to the Joint United Nations Programme against HIV/AIDS 2009 AIDS epidemic update, there were a total of 33.3 million (31.4 million–35.3 million) people living with HIV worldwide in 2009. The majority of the epidemic occurs in Sub-Saharan Africa. Of the 33.3 million people living with HIV worldwide in 2009, a vast majority of 22.5 million (20.9 million-24.2 million) were from Sub-Saharan Africa. There were 1.8 million (1.6 million-2.0 million) new infections and 1.3 million (1.1 million-1.5 million) AIDS-related deaths in Sub-Saharan Africa in 2009 (UNAIDS, 2009). Statistical models and analysis are required in order to further understand the dynamics of HIV/AIDS and in the design of intervention and control strategies. Despite the prevalence of this disease, its pathogenesis is still poorly understood. A thorough understanding of HIV and factors that influence progression of the disease is required in order to prevent the further spread of the virus. Modelling provides us with a means to understand and predict the progression of the disease better. Certain genetic factors play a key role in the way the disease progresses in a human body. For example HLA-B types and IL-10 genotypes are some of the genetic factors that have been independently associated with the control of HIV infection. Both HLA-B and IL-10 may influence the quality and magnitude of immune responses and IL-10 has also been shown to down regulate the expression of certain HLA molecules. Studies are therefore required to investigate how HLA-B types and IL-10 genotypes may interact to affect HIV infection outcomes. This dissertation uses the Sinikithemba study data from the HIV Pathogenesis Programme (HPP) at the Medical School, University of KwaZulu-Natal involving 450 HIV positive and treatment naive individuals to model how certain outcome biomarkers (CD4+ counts and viral loads) are associated with immuno genetic parameters (HLA-B types and IL-10 genotypes). The work also seeks to exploit novel longitudinal data methods in Statistics in order to efficiently model longitudinally measured HIV outcome data. Statistical techniques such as linear mixed models and generalized estimating equations were used to model this data. The findings from the current work agree quite closely with what is expected from the biological understanding of the disease.