Safety and antiviral activity of triple combination broadly neutralizing monoclonal antibody therapy against HIV-1: a phase 1 clinical trial

Abstract

HIV-1 therapy with single or dual broadly neutralizing antibodies (bNAbs) has shown viral escape, indicating that at least a triple bNAb therapy may be needed for robust suppression of viremia. We performed a two-part study consisting of a single-center, randomized, double-blind, dose-escalation, placebo-controlled first-in-human trial of the HIV-1 V2-glycan-specific antibody PGDM1400 alone or in combination with the V3-glycan-specific antibody PGT121 in 24 adults without HIV in part 1, as well as a multi-center, open-label trial of the combination of PGDM1400, PGT121 and the CD4-binding-site antibody VRC07-523LS in five viremic adults living with HIV not on antiretroviral therapy (ART) in part 2 ( NCT03205917 ). The primary endpoints were safety, tolerability and pharmacokinetics for both parts and antiviral activity among viremic adults living with HIV and not on ART for part 2 of the study. The secondary endpoints were changes in CD4+ T cell counts and development of HIV-1 sequence variations associated with PGDM1400, PGT121 and VRC07-523LS resistance in part 2. Intravenously administered PGDM1400 was safe and well-tolerated at doses up to 30 mg kg-1 and when given in combination with PGT121 and VRC07-523LS. A single intravenous infusion of 20 mg kg-1 of each of the three antibodies reduced plasma HIV RNA levels in viremic individuals by a maximum mean of 2.04 log10 copies per ml; however, viral rebound occurred in all participants within a median of 20 days after nadir. Rebound viruses demonstrated partial to complete resistance to PGDM1400 and PGT121 in vitro, whereas susceptibility to VRC07-523LS was preserved. Viral rebound occurred despite mean VRC07-523LS serum concentrations of 93 µg ml-1. The trial met the pre-specified endpoints. Our data suggest that future bNAb combinations likely need to achieve broad antiviral activity, while also maintaining high serum concentrations, to mediate viral control.

Conflict of interest statement

H.C.G. is an employee of Icosavax Inc. All other authors declare no competing interests.

© 2022. The Author(s).

Figures

Fig. 1. PGDM1400, PGT121 and VRC07-523LS pharmacokinetics.

Serum levels of PGDM1400, PGT121 and VRC07-523LS as determined by BAMA. Mean values for each dose group with s.e.m. for PGDM1400 dosed alone (group 1, adults without HIV) (a), PGDM1400 and PGT121 dosed sequentially (group 2, adults without HIV) (b), PGDM1400, PGT121 and VRC07-523LS dosed sequentially (group 3A, adults with HIV) (c) and PGDM1400 and PGT121 dosed sequentially (group 3B, adults with HIV) (d). Dotted lines at the bottom indicate lower limit of detection of the assays, color-coded according to antibody. Each sample was measured in duplicate. Serum t1/2 of PGDM1400 is 20.8 days in adults without HIV when dosed alone, 17.4 days in adults without HIV when dosed in combination with PGT121 (P = 0.05) and 11 days in adults with HIV when dosed in combination. (Supplementary Tables 7 and 8). e, Concordance among PGDM1400, PGT121 and VRC07-523LS concentrations measured by binding and neutralizing antibody assays, shown exemplary for group 3A participants. Neutralizing assays used 6540.v4.c1, CH505TF.N334S.N160A.N280D.1 and CAP220.2.00_A8_5B to measure PGDM1400, PGT121 and VRC07-523LS concentrations, respectively. The concordance for PGDM1400, PGT121 and VRC07-523LS were each ρc = 0.98 in group 3A (substantial agreement)(Pearson’s correlation). The dotted line is the identity line. The solid line is the trend line. Data are colored by bNAb.

Fig. 2. HIV-1 RNA levels.

Plasma HIV-1 RNA levels (RNA copies per ml) are shown after PGDM1400, PGT121 and VRC07-523LS infusion at 20 mg kg−1 each (group 3A) (a) and after PGDM1400 and PGT121 infusion at 30 mg kg−1 each (group 3B) (b) in viremic participants with HIV not on ART. The dotted line indicates the LLoQ for HIV-1 RNA levels (40 copies per ml). Dots indicates when a sample was collected for sequencing. If and when ART was started is indicated in the figures. The symbol ‘&’ indicates the time point when participant 693–2290 was lost to follow-up.

Fig. 3. Neutralization sensitivity to bNAbs and escape variants.

Left: For each participant, the pseudovirus IC50 and IC80 values (µg ml−1) for each bNAb are shown. Note: participant 693–7312 was treated with PGT121 and PGDM1400 dual therapy. Center: Highlighter plots showing amino acid Env mutations in participant viruses. The first baseline virus for each participant is treated as the reference sequence, and all amino acid mutations away from this reference Env are shown. Right: Env sequences for critical epitope sites for each of the bNAbs are shown. The first baseline Env for each participant is taken as the reference sequence, with dots for subsequent Envs indicating identity to the reference Env. Resistance mutations to each bNAb are highlighted in red. Gain or loss of PNGS as compared to the reference Env are highlighted with cyan or purple boxes, respectively. Note: loss of glycans 160 and 332 are associated with resistance to PGDM1400 and PGT121, respectively, whereas gain of glycan 234 and in hypervariable V5 loop are associated with resistance to VRC07-523LS. PTID, participant ID.

Fig. 4. Phylogenetic tree of participant baseline and viral rebound Envs.

Baseline Envs are colored by blue tips and rebound Envs by red. For Envs tested for bNAb neutralization (Fig. 3), boxes next to the tips indicate IC80 values for PGT121, PGDM1400 and VRC07-523LS, going from left to right, color-coded using the scheme in the legend. Participant IDs are shown at the root of each participant Env cluster.

Fig. 5. VRC07-523LS neutralizing activity for baseline and rebound viruses.

a, VRC07-523LS IC80 values (µg ml−1) are shown for each participant with baseline viruses shown in blue and rebound viruses in red (baseline and rebound viruses for 693–1969: nine and ten viruses; 693–2215: six and eight viruses; 693–7989: eight and ten viruses, respectively). Thick horizontal black lines indicate medians, and thin black lines indicate 25th and 75th percentiles. P values indicate significance from one-sided Wilcoxon rank-sum test. b, VRC07-523LS plasma ID80 titers are shown for each participant. Same color scheme, number of viruses examined and statistics as in a. c, Comparison of CD4bs bNAb rebound ID80 titers across different CD4bs bNAbs. Each point represents the per-participant median CD4bs bNAb ID80 titers for rebound viruses from each study. Medians are shown by thick black lines and 25th and 75th percentiles by thin black lines. No significant differences were observed between any two studies. Studies: Bar-On et al., Caskey et al. and Lynch et al..

Extended Data Fig. 1. Participant enrollment and study design.

A total of 62 volunteers without and with HIV were screened for study participation. In individuals without HIV, PGDM1400 was sequentially administered alone or followed by PGT121 in a single IV infusion at increasing doses of 3 mg/kg (Groups 1A and 2A), 10 mg/kg (Groups 1B and 2B), or 30 mg/kg (Groups 1C and 2C) for each bNAb. Viremic participants with HIV (Group 3) received a single IV dose of PGDM1400, PGT121 and VRC07-523LS at 20 mg/kg each (Group 3A) or a single dose of PGDM1400 and PGT121 at 30 mg/kg each (Group 3B). Of a total of 62 subjects screened, 24 participants without HIV and 5 participants with HIV were enrolled. Of the 29 participants enrolled, 26 completed the study on the planned schedule, and three terminated the study early (one Group 1B participant and one Group 3A participant were lost to follow-up and one Group 1C participant moved out of the area). However, no participants were excluded from the safety analyses as all participants had accrued follow-up time after IP administration.

Extended Data Fig. 2. Neutralization sensitivity to bNAbs.

For each participant, the pseudovirus IC50 and IC80 values (µg/ml) for the V3 glycan antibody 10-1074 and the CD4 binding site antibody 3BNC117 are shown.

Extended Data Fig. 3. Mutations at VRC07-523LS signature sites.

All HXB2 sites significantly associated with sensitivity/resistance to VRC07-523LS from Bricault et al. are shown. At these sites, amino acids are shown as logos with height of the letter indicating its frequency. O represents Asn in an N-linked glycan sequon (N-X-S/T where X is not Pro). For each participant, top logos show the baseline Envs, and bottom show rebound Envs. To highlight differences, amino acids in the rebound Envs that are invariant from the baseline Envs are whitened out. If the baseline Envs had two or more variants, from which one was found in the rebound Envs, those variants are still shown (for example site 31 in participant 693-2215). Blue amino acids are associated with sensitivity to VRC07-523LS, red with resistance, and black showed no significant associations.

Extended Data Fig. 4. Mutations at PGDM1400 signature sites.

All isolated Envs from each participant are used for these analyses. All HXB2 sites significantly associated with sensitivity/resistance to PGDM1400 from Bricault et al. are shown. At these sites, frequency of amino acids are shown as logos with height of the letter indicating its frequency. For each PTID, top logos show the baseline Envs, and bottom show rebound Envs. To highlight differences, amino acids in the rebound Envs that are invariant from the baseline Envs are whitened out. If the baseline Envs had two or more variants, from which one was found in the rebound Envs, those variants are still shown (for example site 211 in 693-7989). Blue amino acids are associated with sensitivity to PGDM1400, red with resistance, and black showed no significant associations. O represents Asn in an N-linked glycan sequon (N-X-S/T where X is not Pro). Grey box indicates deletion.

Extended Data Fig. 5. Mutations at PGT121 signature sites.

Same as Extended Data Figs. 3 and 4, but using PGT121 signature sites from Bricault et al.

Extended Data Fig. 6. Recombination detection in three out of five samples.

A total of 16 recombinants were detected in 3 out of 5 samples (A). Each row shows the total number of sequences at each visit per participant (NSeq), the total number of recombinants detected (NRecs), and the sequence names of the recombinants (last column on the right). No recombinants were detected in participants 693-7312 and 693-7989, although our power to detect recombination suffered from the low number of sequences (range 4-17 per participant, per visit; median=10). All recombinants were found to be statistically significant after multiple testing correction using a Wald-Wolfowitz Runs Test statistic implemented by the LANL tool RAPR. (B) Impact of recombination on participant 693-2215. The full Env genome of all recombinant triplets from participant 693-2215 is shown on the left, solid green and purple lines for the parental strains at the top, and gray for the recombinant below, with tic marks indicating the sites where the daughter strain differs from at least one parent, color coded as follows: green if the recombinant matches the green parent at that site, purple if the recombinant matches the purple parent, and black if it doesn’t match either parental strain. Vertical gray bands show the positions in the genome of the 234 glycan and the V5 glycan where we find evidence of recombination favoring the resistant form over the sensitive one. This is highlighted on the right, with the reference amino acids surrounding the two glycan regions shown at the top and below the corresponding sequence for each recombinant triplet. Dashes indicate amino acids were the sequences match the reference. Mutations at the glycans are shown in red if they confer resistance and blue if they confer sensitivity. The top triplet shows the only recombinant found at the first time point where no resistant mutations were detected. In the remaining triplets, recombination preferentially selects the resistant mutation 4 out of 4 times for the glycan at position 234, and 3 out of 4 times at the site of the V5 glycan.

Extended Data Fig. 7. Comparison of CD4bs bNAb and bNAb combination neutralizing activity from this study to other studies.

(a) CD4bs bNAb serum ID80 titers at baseline and rebound from each participant across the 4 studies–. Similar to Fig. 4b, see methods for ID80 calculation details. A one-sided Wilcoxon rank sum test was used to calculate statistical significance of differences between baseline and rebound titers for each patient, and p-values are shown on the top of the panel. All participants except 9341, 2C1 and 2E1, showed significantly lower serum ID80 titers at rebound as compared to baseline. (b) CD4bs bNAb IC80 values (µg/ml) as tested in vitro against baseline and rebound viruses from each participant across the studies. Similar to (A). Most participants showed significant difference (p < 0.05 using one-sided Wilcoxon rank sum test), with the exception of participants 693-7989, 91C22, 91C34, 91C35, 9342, 2C1 and 2E1 for whom no significant differences were found (p > 0.05). (c) Summary of in vitro IC80 values (µg/ml) of rebound viruses per study. Each point shows the per-participant median CD4bs bNAb IC80 value (µg/ml) for rebound viruses from each study. The only significant difference using two-sided Wilcoxon rank sum test was found to be between 3BNC117 + 10-1074 and VRC01 studies (p = 0.0134), and a trend when comparing this study to VRC01 (p = 0.0719); all other comparisons resulted in p > 0.17 (p-values not shown). (d) Same approach as (A) but using bNAb combination serum ID80 titers (see methods for combination ID80 titer calculation details). All participants showed significantly lower ID80 titers (that is due to higher resistance) for rebound viruses as compared to baseline viruses (p < 0.05 using one-sided Wilcoxon rank sum test). (e) Same approach (C) but summarizing the combination serum ID80 titers at rebound across studies. For 3BNC117 and VRC01 groups, single bNAb rebound ID80 titers are shown. The only significant difference was observed between the 3BNC117 + 10-1074 and VRC01 studies, with the former having significantly higher rebound ID80 titers than the latter (p = 0.0353 using two-sided Wilcoxon rank sum test). All other comparisons were not significant (p > 0.23; p-values not shown).

Extended Data Fig. 8. Neutralizing activity of bNAb combinations against global heterologous viruses.

(a) Geometric mean IC80 values (µg/ml) of baseline viruses for this study (top) and from Bar-On et al (bottom). The participants are ordered according to time to rebound. These data suggest that longer times to rebound were found for those participants whose baseline IC80 < 0.3 µg/ml for at least 2 bNAbs in the therapeutic combination. (b) IC80 values for individual and combination bNAbs. Data are shown as heatmaps with viruses on the rows. Viral subtypes and individual bNAb and bNAb combination IC80 values are shown as columns. Color coding of subtypes and IC80 values is shown on the bottom right. Viruses are ordered according to the number of bNAbs with IC80 < 0.3 µg/ml, with most number of bNAbs active on top and least on the bottom, and according to neutralization sensitivity to each bNAb. The ordering of viruses between the left and the right panels is different. The single bNAb neutralization data for these non-tier-1 374 viruses was obtained from CATNAP, as previously described6, combination IC80 values were predicted using Bliss-Hill model assuming each bNAb at equal concentrations. Combination IC80 values are the sum of concentrations of each bNAb in the combination for predicted 80% neutralization of a given virus. (c) Overall and dual coverage breadth potency curves for VRC07-523LS + PGT121 + PGDM1400 (salmon and red, respectively) and for 3BNC117 + 10-1074 (light and dark blue, respectively). For dual coverage a virus is considered only if it is neutralized by 2 or 3 bNAbs at individual bNAb IC80 < 0.3 µg/ml, based on the suggested trend of this metric associated with longer time of viral control from panel (A). (d) Subtype-specific coverage of single, dual and triple bNAb activity for VRC07-523LS + PGT121 + PGDM1400 (shades of red), and for 3BNC117 + 10-1074 (shades of blue). Single bNAb activity threshold is IC80 < 0.3 µg/ml.