Human T-cell lymphotropic virus type I (HTLV-I) is a persistent human retrovirus that infects an estimated 10-25 million individuals world-wide. Infection is life-long and carries a risk of developing one of two major, clinically independent diseases: an aggressive cancer of the blood or lymph nodes called adult T-cell leukaemia/lymphoma (ATLL), and a progressive neurological and inflammatory condition known as HAM/TSP. Despite the presence of a large, chronically activated cytotoxic T-lymphocyte (CTL) response directed against HTLV-I-infected cells, host immunity ultimately fails to effectively eliminate the virus. With respect to pathology, it is widely believed that the principal factor contributing to disease manifestation is a high proviral load. Yet, there exists a broad overlap in the proviral loads of both asymptomatic carriers and individuals with HTLV-I-associated disease, and a conclusive determinant of disease outcome remains to be identified.
It is becoming increasing clear that a more detailed understanding of the dynamic interactions between chronic HTLV-I infection and virus-specific host immunity is needed to help resolve the above issues. One important feature of HTLV-I that has not been considered in within-host mathematical models of HTLV-I infection is the simultaneous expression of multiple, antigenically variable viral proteins in the pool of infected target cells. Two such proteins, Tax and HBZ, have recently been identified as playing crucial roles in the establishment and maintenance of the infection. Anti-viral host immune responses acting on such a composite virus population then implies a corresponding measure of diversity in the pool of HTLV-I-specific CTLs.
In light of these observations, we develop a mathematical model for the within-host infection dynamics of HTLV-I that captures the inherent heterogeneity in provirus-positive target cells and host immune responses, thereby allowing us to explore the connections between abundance and structure in both the virus and CTL populations. The consideration of antigenic variation at multiple viral epitopes prompts the usage of a multi-locus modelling framework representing the complex host-virus interactions that define HTLV-I infection in vivo. The subsequent mathematical model is a high-dimensional non-linear system of ordinary differential equations. In this talk, I discuss our approach to modelling HTLV-I infection as a multi-locus system. Results from our model offer insights to the emergence and evolution of strain structure and highlight the significance of strain structure to the diagnosis and treatment of HTLV-I-associated disease.

Non-linear dynamical system; ODEs; multi-locus modelling framework; within-host viral dynamics; strain structure