HIV-1 transmission patterns in antiretroviral therapy-naïve, HIV-infected North Americans based on phylogenetic analysis by population level and ultra-deep DNA sequencing

Abstract

Factors that contribute to the transmission of human immunodeficiency virus type 1 (HIV-1), especially drug-resistant HIV-1 variants remain a significant public health concern. In-depth phylogenetic analyses of viral sequences obtained in the screening phase from antiretroviral-naïve HIV-infected patients seeking enrollment in EPZ108859, a large open-label study in the USA, Canada and Puerto Rico (ClinicalTrials.gov NCT00440947) were examined for insights into the roles of drug resistance and epidemiological factors that could impact disease dissemination. Viral transmission clusters (VTCs) were initially predicted from a phylogenetic analysis of population level HIV-1 pol sequences obtained from 690 antiretroviral-naïve subjects in 2007. Subsequently, the predicted VTCs were tested for robustness by ultra deep sequencing (UDS) using pyrosequencing technology and further phylogenetic analyses. The demographic characteristics of clustered and non-clustered subjects were then compared. From 690 subjects, 69 were assigned to 1 of 30 VTCs, each containing 2 to 5 subjects. Race composition of VTCs were significantly more likely to be white (72% vs. 60%; p = 0.04). VTCs had fewer reverse transcriptase and major PI resistance mutations (9% vs. 24%; p = 0.002) than non-clustered sequences. Both men-who-have-sex-with-men (MSM) (68% vs. 48%; p = 0.001) and Canadians (29% vs. 14%; p = 0.03) were significantly more frequent in VTCs than non-clustered sequences. Of the 515 subjects who initiated antiretroviral therapy, 33 experienced confirmed virologic failure through 144 weeks while only 3/33 were from VTCs. Fewer VTCs subjects (as compared to those with non-clustering virus) had HIV-1 with resistance-associated mutations or experienced virologic failure during the course of the study. Our analysis shows specific geographical and drug resistance trends that correlate well with transmission clusters defined by HIV sequences of similarity. Furthermore, our study demonstrates the utility of molecular and epidemiological analysis of VTCs for identifying population-specific risks associated with HIV-1 transmission and developing effective local healthcare strategies.

Conflict of interest statement

Competing Interests: L.L.R. and M.S.S. are employed by ViiV Healthcare, whose company partly funded this study. J.U. was employed by BristolMyers-Squibb during the initial data collection. J.H., S.H. and J.B. are employed by GlaxoSmithKline. J.U. is an employee of Amicus Therapeutics. There are no patents, products in development or marketed products to declare as this analysis was performed using samples that were collected prior to any study drug administration. This does not alter the authors’ adherence to all the PLOS ONE policies on sharing data and materials.

Figures

Figure 1. Prediction of VTCs by two studies using HIV-1 sequences obtained from pre-therapy subjects.

Shown are comparisons of VTC numbers as predicted by NJ tree-reconstruction of either population-based, ultra-deep sequencing (UDS) or both datasets. See Methods and Materials for VTC prediction methodology.

Figure 2. Neighbor-joining (NJ) tree based on population-genotyped HIV sequences from 690 pre-therapy subjects.

High-confidence VTCs are labeled in red (1–30). These VTCs all represent 2 or more clustered subjects with a minimum confidence level of 95% in the (1000 bootstrap NJ replicates) for both population and UDS datasets. VTCs predicted only by the population or UDS data are shown in green and blue, respectively. VTCs where sequences are subsets of population or UDP VTCs have multiple colored labels. A more detailed phylogenetic tree with the OTUs labeled appears in Figure S1.