Abstract Body

Background: The HVTN505 HIV vaccine study enrolled 2,504 US participants in a phase IIb test-of-concept efficacy trial of a DNA/rAd5 vaccine expressing Gag/Pol/Nef from subtype B and Env proteins from subtypes A/B/C. Although the vaccine showed no efficacy with respect to HIV acquisition or post-infection viral load (Hammer, NEJM, 2013), we analyzed breakthrough HIV isolates to evaluate potential sieve effects.

Methods: Sequencing was performed following PCR amplification of near full-length single HIV genomes derived from plasma specimens at HIV diagnosis. Phylogenetic and statistical analysis methods assessed individual genes, k-mers, and amino acid (AA) sites to compare breakthrough sequences from vaccine and placebo recipients, with multiplicity adjustment procedures (20% false discovery rate).

Results: We characterized 480 HIV genomes from 27 vaccine and 20 placebo recipients. All infections were with subtype B. Infections were established by multiple founder variants in 5 of 26 vaccine recipients (19%) and in 8 of 20 placebo recipients (40%) (Fisher’s exact test, p= 0.19). Two vaccine recipients were infected by two phylogenetically-unrelated viruses (env diversity= 8.9% vs 0.3% for other subjects). Intra-host diversity was significantly lower in vaccine than placebo recipients for Gag, Pol, Env-gp120 and Vif proteins (p≤ 0.04, q≤ 0.09).
The distance between founder and vaccine sequences was measured for each participant and compared across treatment groups. While differences in other proteins were not significant, Env-gp120 sequences from vaccine recipients were significantly more distant from the subtype B vaccine insert than sequences from placebo recipients (median = 0.30 vs 0.29; Mann-Whitney test, p= 0.01, q= 0.12).
Vaccine/placebo AA differences in Env-gp120 were identified at six 9-mers starting at HXB2 positions 28, 86, 193, 365, 427, and 467, and at site 133 (p< 0.005, q< 0.2). Mapping of the 9-mer starting at AA193 on a trimer structure showed these residues in contact with the tip of the V3 loop of another monomer, thus possibly affecting the binding of V3-specific antibodies (Fig.1) – an interesting finding as the HVTN505 vaccine regimen elicited strong anti-V3 responses.

Conclusions: Our results suggest an HVTN505 vaccine-driven sieve effect against Env-gp120. Analysis of breakthrough HIV isolates in vaccine efficacy trials is important to assess whether a vaccine elicited a selective pressure in vivo and to identify which responses are associated with efficacy and which are not.

viewer?TAG_ACTION=SECURE_FILE_DOWNLOAD&V
Figure 1. Env-gp120 residues corresponding to the 9-mer starting at HXB2 position 193 that distinguished the vaccine and placebo groups are figured with circles on the BG505.SOSIP structure (PDB ID: 4NCO). The start and end residues of the 9-mer (AA193 and 201) are numbered on the monomer on the left, the residues are colored in green and turquoise with a mirror image on another monomer. At the tip of the V3 loop, GPG is in magenta, with the 5 atoms in the proline’s ring shown as spheres (the N atom being a blue sphere, carbon atoms in magenta).