Abstract Body

Background: Understanding the processes shaping epidemics is critical for their prediction and management. Phylogenetic inferences drawn from viral sequence data can provide a window into the temporal dynamics of regional epidemics, allowing estimation of rates of incidence, infection accumulation, and viral population demographic history. Within such regional epidemics, different risk factors in transmission dynamics and health care engagement predict corresponding different viral evolutionary dynamics measureable from the shape of phylogenetic trees. Methodology: We inferred 1,000 time-scaled phylogenetic trees for 27,296 HIV protease and RT sequences sampled from 7,747 patients in the BC Centre for Excellence in HIV/AIDS database in BC, Canada. All sequences were doubly anonymized to protect patient privacy. Estimates of epidemic age were calculated using the root age of each phylogenetic tree in our distribution. We visualized the accumulation of infections through time using lineage through time plots (LTT). We utilized maximum likelihood approaches for estimating rates of incidence and effective population size by fitting birth-death models across our distribution of phylogenetic trees for the epidemic as a whole and within risk factors. Results: Our analysis suggested 1897 (IQR: 1883-1923) as the estimated median time for the origin of the HIV epidemic overall, and 1941 (IQR: 1923-1958) for the subtype B clade. LTT plots reveal that within and between HIV subtypes, the BC epidemic is composed primarily of variation originating prior to 1990 (the advent of health care intervention). Birth-death models show the subtype B epidemic has undergone a recent decrease in the estimated rate of incidence. Skyline plots show corresponding declines in effective population size towards the present. Results are consistent with an epidemic founded by multiple sources and subsequent spread along local networks corresponding to different HIV risk factors, with later declines in spread. Comparative phylogenetic analyses among risk exposure categories reveal substantial differences in the timing of the HIV epidemic between intravenous drug users (IDU) and MSM. IDU experienced a dramatic increase in rate of lineage accumulation; incidence and effective population size in the mid-1990s whereas the MSM epidemic underwent rapid growth in the 1980’s followed by subsequent tapering off. Conclusions: Concordance of our population-level phylogenetic results with epidemiological data validates the use of phylogenetic methods in assessing the past and present dynamics of the HIV epidemic. Further, we show the potential of comparing inferences drawn from phylogenetic trees partitioned by different HIV risk exposure categories.