Phylogenetic surveillance of viral genetic diversity and the evolving molecular epidemiology of human immunodeficiency virus type 1

Abstract

With ongoing generation of viral genetic diversity and increasing levels of migration, the global human immunodeficiency virus type 1 (HIV-1) epidemic is becoming increasingly heterogeneous. In this study, we investigate the epidemiological characteristics of 5,675 HIV-1 pol gene sequences sampled from distinct infections in the United Kingdom. These sequences were phylogenetically analyzed in conjunction with 976 complete-genome and 3,201 pol gene reference sequences sampled globally and representing the broad range of HIV-1 genetic diversity, allowing us to estimate the probable geographic origins of the various strains present in the United Kingdom. A statistical analysis of phylogenetic clustering in this data set identified several independent transmission chains within the United Kingdom involving recently introduced strains and indicated that strains more commonly associated with infections acquired heterosexually in East Africa are spreading among men who have sex with men. Coalescent approaches were also used and indicated that the transmission chains that we identify originated in the late 1980s to early 1990s. Similar changes in the epidemiological structuring of HIV epidemics are likely to be taking in place in other industrialized nations with large immigrant populations. The framework implemented here takes advantage of the vast amount of routinely generated HIV-1 sequence data and can provide epidemiological insights not readily obtainable through standard surveillance methods.

FIG. 1.

Strain-level classification of HIV-1 genetic diversity. Thirty-six strain-level groupings were defined by phylogenetic analysis of sequences from the complete HIV-1 genome and were recoverable using subgenomic fragments. In this example, phylograms illustrate that due to the presence of recombinant strains, the internal topologies of phylogenies (shown as dotted lines) differ depending on which subgenomic region is analyzed. Thus, the CRF03 strain illustrated, highlighted in trees by shaded circles, can be seen to group with subtype A in trees constructed using gag (A) and with subtype B in trees constructed using pol (B). However, the classification that we describe here focuses on closely related sequences sharing specific geographic associations and does not seek to represent deeper evolutionary relationships. As such, it is robust to the effects of recombination; CRF03 and all other CRFs included in the initial data set were recovered as robustly supported monophyletic lineages in trees constructed using subgenomic regions.

FIG. 2.

Clusters of epidemiologically linked HIV-1 infections in the United Kingdom involving recently introduced strains. Epidemiologically linked infections were identified by statistically significant (P < 0.01) clustering of sequences sampled in the United Kingdom (UK) relative to globally sampled sequences in phylogenies representative of characterized diversity among the HIV-1 strains involved (East African subtype A [97 sequences] and West African/Iberian subtype G [56 sequences]). These clusters are shown in the phylogeny above with a representative set of globally sampled sequences. Bootstrap support for clusters was >70%. Filled circles indicate sequences from the United Kingdom. Globally sampled reference sequences are labeled according to subtype and country of sampling. The estimated date of the most recent common ancestor (MRCA) is shown for each cluster.