Background:  Approximately 32,000 men who have sex with men (MSM) in the UK are infected with HIV and recent data suggest that infection prevalence doubled between 1999 and 2007. This trend highlights the need for more effective control strategies and a better understanding of the dynamics of transmission within this risk group. Research focusing on the global pandemic of HIV indicates strong spatial structuring of different HIV subtypes, but less is known about viral gene flow on smaller temporal and spatial scales.

Methods:  Epidemiologically linked clusters were identified from 14,560 subtype B HIV partial pol sequences by finding sequences with a close genetic match and using these sequences to build a bootstrapped Neighbor-Joining phylogeny. A newly developed Bayesian phylodynamic method was then applied to 97 clusters that had 10 or more individuals (giving a total of 1673 individuals) to estimate the rate of geographical spread between 17 locations from the UK over the previous 10 years.

Results:  For more than 70% of clusters, at least half of all individuals came from one location, supporting the assumption that our method accurately detects epidemiologically linked individuals. Of clusters that were more heterogeneous, time-scaled phylogenies showed that within the cluster, related viruses are generally found in the same location. This reveals the areas where there is significant ongoing transmission and the date since this has been occurring. However, there were also some areas within clusters where infection was clearly seeded from elsewhere. The highest rates of migration of viral strains were observed from London to Brighton and London to Manchester. Analyses of all other rates showed that most other infections in the UK originated from these 3 key areas.

Conclusions:  In contrast to the early UK epidemic, where infection was heavily London-based, our results suggest there is ongoing infection in other areas of the country. However, analyses of geographical rate show that there are still only a few areas where outward spread to other areas is significant. These results have implications to help us in the future optimise where preventative strategies will be most effective.