Analysis of genetic linkage of HIV from couples enrolled in the HIV Prevention Trials Network 052 trial

Abstract

Background: The HIV Prevention Trials Network (HPTN) 052 trial demonstrated that early initiation of antiretroviral therapy (ART) reduces human immunodeficiency virus (HIV) transmission from HIV-infected adults (index participants) to their HIV-uninfected sexual partners. We analyzed HIV from 38 index-partner pairs and 80 unrelated index participants (controls) to assess the linkage of seroconversion events.

Methods: Linkage was assessed using phylogenetic analysis of HIV pol sequences and Bayesian analysis of genetic distances between pol sequences from index-partner pairs and controls. Selected samples were also analyzed using next-generation sequencing (env region).

Results: In 29 of the 38 (76.3%) cases analyzed, the index was the likely source of the partner's HIV infection (linked). In 7 cases (18.4%), the partner was most likely infected from a source other than the index participant (unlinked). In 2 cases (5.3%), linkage status could not be definitively established.

Conclusions: Nearly one-fifth of the seroconversion events in HPTN 052 were unlinked. The association of early ART and reduced HIV transmission was stronger when the analysis included only linked events. This underscores the importance of assessing the genetic linkage of HIV seroconversion events in HIV prevention studies involving serodiscordant couples.

Figure 1.

Outline of approach used for linkage analysis (see text).

Figure 2.

Example of phylogenetic analysis of pol sequences from index-partner pairs. A phylogenetic tree was generated for 5 seroconversion events from Africa. The tree includes sequences from index participants and their partners (colored text: I, index; P, partner). Sequences from 2 different study visits were available for 9 of the 10 participants analyzed (A, sequence from the index or partner collected at the earlier study visit; B, sequence from the index or partner collected at the later study visit). The tree also includes subtype-matched reference sequences (C, ref), 10 local control sequences (randomly selected index participants from the same study site [blue dots]) and an outgroup sequence (CPZ.GA.CPZGAB, SIV). Bootstrap values are shown for the grouping of index and partner sequences for each event based on 500 bootstrap replications. In all cases, paired sequences from a given participant grouped together. In 4 cases, all of the sequences from a given event (all index and partner sequences from an index-partner pair) grouped on a single branch with a bootstrap value of 100% (052-1692, 052-2899, 052-0621, and 052-2337), indicating that the events were linked. In 1 case, the 2 sequences from the index did not group with the 2 sequences from the partner (052–3172), indicating that the event was unlinked.

Figure 3.

Densities of similarities from linked and unlinked sequence pairs. A, Densities of similarities from linked and unlinked sequence pairs from 32 subtype C couples, along with subtype C training data. B, Densities of similarities from linked and unlinked sequence pairs from 6 couples with non–subtype C human immunodeficiency virus (HIV) infection, along with non-subtype C unlinked training data and all linked training data (there were insufficient data to estimate a non-subtype C linked density).

Figure 4.

Next-generation sequencing results for the env region (gp41). The figure shows phylogenetic trees generated using data from next-generation sequencing (NGS, env region [gp41]). Each panel shows results obtained for 1 seroconversion event where linkage of the index and partner was established based on NGS data. Merged neighbor-joining trees were constructed using subtype reference sequences (black text, subtypes A, B, C, and D) and gp41 consensus NGS sequences from the relevant index participant (2 samples) and partner (2 samples). Note that only consensus sequences with ≥10 primary NGS reads were used. Color-coding indicates sequences from individual samples (red, index sample 1; pink, index sample 2; dark blue, partner sample 1; teal, partner sample 2). Bootstrap values are shown for selected branches (percentage of trees where the sequences group together, among 500 trees generated). Genetic distance is shown (scale line). A, Data for the seroconversion event from couple 052-2989. For this event, the second index sample was collected 1211 days after the first index sample; the partner samples were collected 1169 and 1211 days after the first index sample. The number of primary sequence reads per sample for these 4 samples ranged from 6555 to 13696, and the number of consensus sequences per sample for these 4 samples ranged from 35 to 108. This event was provisionally characterized as unlinked based on data obtained by phylogenetic and Bayesian analysis of pol region sequences. The linkage status was changed to linked based on the results from NGS. B, Data for the seroconversion event from couple 052-1168. For this event, the second index sample was collected 466 days after the first index sample; the partner samples were collected 539 and 556 days after the first index sample. The number of primary sequence reads per sample for these 4 samples ranged from 6891 to 11700, and the number of consensus sequences per sample for these 4 samples ranged from 38 to 80. The linkage status of this event could not be determined based on data obtained by phylogenetic and Bayesian analysis of pol region sequences. The linkage status was changed to linked based on the results from NGS. This analysis was repeated using local control sequences in addition to subtype reference sequences. Both of these events (052-2989 and 052-1168) were from Malawi. The local control sequences used in the analysis of each event included the most prominent consensus sequence from each of 22 other individuals enrolled in the HIV Prevention Trials Network (HPTN) 052. Twenty of the 22 individuals were from Africa and had subtype C human immunodeficiency virus (HIV), and 12 of those individuals were from Malawi. The results obtained using subtype reference sequences only (A and B) or subtype reference sequences plus local control sequences (not shown) did not differ in any meaningful way.