Inferring HIV transmission networks from time-resolved viral phylogenies for epidemiological modelling

The HIV epidemic in the UK is growing, both among heterosexuals, which is a relatively new epidemic, and again among gay men, where new infections had previously stabilised. This proposal is intended to develop methods for finding out in what settings most new infections occur. We will use anonymised data on viral sequences to discover whether HIV transmission occurs steadily over time, or whether it occurs in bursts, through networks. Early studies suggest networks are important, so we need to be able to identify the key settings. We will do this by using a computer program which describes ( models ) the epidemic as far as we understand it, including settings such as bars, gyms and the internet, where people can meet sexual partners. The model will allow information from the viral sequence data to help us to predict which will be the most effective links to target to reduce new infections.

Much understanding of the epidemiology of sexually transmitted infections has come from analysis of sexual network models. These can either describe contacts between individuals or affiliations, where individuals are linked to venues rather than to each other. For genetically diverse pathogens, molecular typing can also inform epidemiological models. For HIV in particular, the high rate of evolution permits time-resolved viral phylogenies to be reconstructed in which transmission clusters can be identified and their size estimated. We propose to develop a statistically rigorous methodology for using time-resolved viral phylogenies to obtain an independent estimate of the transmission network within an infected community. In parallel we will develop models of the sexual mixing of the population through use of affiliation networks. Integrating the two will allow identification of the key elements of the affiliation network that potentiate transmission, as revealed by the structure of the viral transmission network. Development, and subsequent application of this methodology will allow critical features of the sexual contact structure for viral transmission to be identified and for the first time will provide a model with which intervention measures can be tested in the setting of a specific population.