Viral linkage in HIV-1 seroconverters and their partners in an HIV-1 prevention clinical trial

Abstract

Background: Characterization of viruses in HIV-1 transmission pairs will help identify biological determinants of infectiousness and evaluate candidate interventions to reduce transmission. Although HIV-1 sequencing is frequently used to substantiate linkage between newly HIV-1 infected individuals and their sexual partners in epidemiologic and forensic studies, viral sequencing is seldom applied in HIV-1 prevention trials. The Partners in Prevention HSV/HIV Transmission Study (ClinicalTrials.gov #NCT00194519) was a prospective randomized placebo-controlled trial that enrolled serodiscordant heterosexual couples to determine the efficacy of genital herpes suppression in reducing HIV-1 transmission; as part of the study analysis, HIV-1 sequences were examined for genetic linkage between seroconverters and their enrolled partners.

Methodology/principal findings: We obtained partial consensus HIV-1 env and gag sequences from blood plasma for 151 transmission pairs and performed deep sequencing of env in some cases. We analyzed sequences with phylogenetic techniques and developed a Bayesian algorithm to evaluate the probability of linkage. For linkage, we required monophyletic clustering between enrolled partners' sequences and a Bayesian posterior probability of ≥ 50%. Adjudicators classified each seroconversion, finding 108 (71.5%) linked, 40 (26.5%) unlinked, and 3 (2.0%) indeterminate transmissions, with linkage determined by consensus env sequencing in 91 (84%). Male seroconverters had a higher frequency of unlinked transmissions than female seroconverters. The likelihood of transmission from the enrolled partner was related to time on study, with increasing numbers of unlinked transmissions occurring after longer observation periods. Finally, baseline viral load was found to be significantly higher among linked transmitters.

Conclusions/significance: In this first use of HIV-1 sequencing to establish endpoints in a large clinical trial, more than one-fourth of transmissions were unlinked to the enrolled partner, illustrating the relevance of these methods in the design of future HIV-1 prevention trials in serodiscordant couples. A hierarchy of sequencing techniques, analysis methods, and expert adjudication contributed to the linkage determination process.

Figure 1. Adjudication criteria used in assigning transmission linkages.

For each pair, adjudicators evaluated monophyly (yes/no), genetic distance, and Bayesian posterior probability (≥0.5 or <0.5) and classified the pair as ‘linked’, ‘unlinked’, or ‘indeterminate’. Further evaluation of ‘unlinked’ or ‘indeterminate’ pairs involved gathering additional data, including sequencing of consensus gag and/or clonal, single molecule or pyrosequencing of env, as well as obtaining sequences from non-transmitting HIV-1 infected participants from the same study site. New trees, distance distributions and Bayesian priors were generated and each pair was re-adjudicated to make final linkage assignments.

Figure 2. Examples of Phylogenetically Linked and Unlinked Transmission Events.

A section of a phylogenetic tree showing examples of linked monophyletic (PP73 and PP82) and unlinked polyphyletic (PP45) pairs are shown, along with the adjudication criteria for each.

Figure 3. Pairwise Genetic Distances for env.

Distributions of pairwise genetic distances for env reference datasets, within acutely infected individuals from the Multicenter AIDS Cohort Study at different intervals post infection, between epidemiologically-unlinked pairs of sequences from the HIVDB of subtypes A, C, and D (lines) and between enrolled partner-pairs from the Partners in Prevention HSV/HIV Transmission Study cohort that were adjudicated as linked (red bars) and unlinked (blue bars) through sequencing of env, gag, or both. To improve visibility of the data, the y-axis scale ranges from 0 to 0.25 for bars representing the Partners in Prevention HSV/HIV Transmission Study cohort.

Figure 4. Bayesian Posterior Probabilities for env and gag Datasets.

Plot showing relationship between Bayesian posterior probabilities and genetic distance between partner pairs from the Partners in Prevention HSV/HIV Transmission Study cohort in env and gag.

Figure 5. Examples of HIV-1 Transmission Classified as Linked by SM Sequencing.

Example of a pair (Pair 17) whose consensus env sequences were unlinked, with linkage subsequently determined by single molecule (SM) env sequences. The linkage criteria used during adjudication are displayed in the table. Three linked sequences from the HIV-1 infected partner, PP17A variant 1, along with the sequences from the seroconverting partner PP17B are bounded by the solid rectangle. Unlinked sequences from the HIV-1 infected partner, PP17A variant 2 are delineated by the dotted rectangle.