Safety and antiviral effect of a triple combination of HIV-1 broadly neutralizing antibodies: a phase 1/2a trial

Abstract

Human immunodeficiency virus type 1 (HIV-1)-specific broadly neutralizing monoclonal antibodies (bNAbs) have to date shown transient viral suppression when administered as monotherapy or as a cocktail of two antibodies^1-4. A combination of three bNAbs provides improved neutralization coverage of global viruses, which may more potently suppress viral escape and rebound^5-7. Here we performed an open-label, two-part study evaluating a single intravenous dose of HIV-1 bNAbs, PGT121, PGDM1400 and VRC07-523LS, in six adults without HIV in part 1 and a multicenter trial of up to six monthly infusions of these three bNAbs in 12 people living with HIV with an antiretroviral therapy (ART) interruption in part 2. The primary endpoints were safety, tolerability and pharmacokinetics, and the secondary endpoints in part 2 were antiviral activity following ART discontinuation, changes in CD4⁺ T cell counts and development of HIV-1 sequence mutations associated with bNAb resistance. The trial met its prespecified endpoints. The bNAb treatment was generally safe and well tolerated. In part 2, 83% of participants (10 of 12) maintained virologic suppression for the duration of antibody therapy for at least 28 weeks, and 42% of participants (5 of 12) showed virologic suppression for at least 38-44 weeks, despite the decline of serum bNAb concentrations to low or undetectable levels. In exploratory analyses, early viral rebound in two individuals correlated with baseline resistance to PGT121 and PGDM1400, whereas long-term virologic control in five individuals correlated with reduced immune activation, T cell exhaustion and proinflammatory signaling following bNAb therapy. Our data show the potential of a triple bNAb cocktail to suppress HIV-1 in the absence of ART. ClinicalTrials.gov registration: NCT03721510 .

Fig. 1. CONSORT diagram.

Participant recruitment, randomization and follow-up are depicted. Participants were enrolled at Beth Israel Deaconess Medical Center (BIDMC), Orlando Immunology Center (OIC) and Houston AIDS Research Team (HART), McGovern Medical School at The University of Texas Health Science Center). ^aNote that one volunteer from group 2 who received a partial infusion (one IP of the assigned three-IP combination was not given) at day 0 and no subsequent infusions had limited safety follow-up in the study. This volunteer will be left out of the safety data displays.

Fig. 2. Study design, pharmacokinetics and therapeutic efficacy of PGT121, PGDM1400 and VRC07-523LS.

a, Study design. Groups 1A and 1B included participants without HIV-1 that received a single IV dose of either PGT121 and VRC07-523LS (green triangle) or PGT121, PGDM1400 and VRC07-523LS (blue triangle). Group 2 PLWH that were virologically suppressed on ART received IV infusions of PGT121, PGDM1400 and VRC07-523LS monthly for 3 months (blue triangles), with the option to continue monthly infusions for another 3 months (light blue triangles) in the presence of ongoing virological suppression. ART was stopped on day 2 after the initial bNAb infusion on day 0 and was reinitiated if participants met predefined restart criteria or at the end of the study on day 308. D, day(s). b, Levels of VRC07-523LS, PGDM1400 and PGT121 in serum of participants in group 2 that received all six infusions of the three bNAbs (n = 8 participants), as determined by anti-idiotype-specific quantitative ELISAs (PGT121, PGDM1400) and electrochemiluminescence immunoassay (VRC07-523LS). Data are mean ± s.d. Mean t_1/2 of each bNAb is indicated in days. c, HIV-1 RNA levels. Plasma HIV-1 RNA levels (log₁₀(RNA copies per ml)) over the course of the study are shown for each of the 12 group 2 participants (color coded). The vertical dotted lines indicate the timepoints when the three bNAbs PGDM1400, PGT121 and VRC07-523LS were administered. ART was stopped 2 days after the first triple-bNAb infusion (as indicated by the gray box). One participant, 98063, was lost to follow-up. The horizontal dotted line indicates the lower limit of quantification for HIV-1 RNA levels. All participants received six infusions except for 65021 who received two and 65015, 98056 and 98059 who received three infusions.

Fig. 3. Virological and pharmacokinetic follow-up of individual participants.

Plasma HIV-1 RNA levels (red line; right y-axis) and bNAb serum concentrations (VRC07-523LS (green), PGDM1400 (blue) and PGT121 (red); left y-axis) of group 2 participants. Gray-shaded areas indicate time on ART. Participant 65021 resumed ART but was lost to follow-up (LTFU), and participant 98063 moved out of state and had to exit the study prematurely. The red dotted horizontal line indicates the limit of detection.

Fig. 4. Viral reservoir and T cell immunity.

a, Baseline (before the first bNAb infusion) frequencies of CD4⁺ T cells carrying intact proviral HIV DNA by the IPDA assay in study participants that had either early, late or no rebound of plasma viremia during study follow-up. No statistically significant differences were observed. b, Dynamics of IPDA reservoir virus in the CD4⁺ T cells of participants with late plasma viral rebound (left panel) at baseline and at month 6 (before plasma viremia) and in participants with no viral rebound (right panel) at baseline and at the end of study follow-up. c, Frequency of T cell responses to HIV PTE peptide pools (Gag1, Gag2, Env1, Env2 and Env3) at baseline as measured by the IFNγ ELISpot assay. Shown are the means of responses from each participant group to each peptide pool. d, Frequencies of HIV-specific IFNγ⁺ CD8⁺ T cell subsets across early, late and no rebounders at baseline as determined by flow cytometry. e, Dynamics of HIV-specific T cell responses at baseline and 6 months in participants with late viral rebound and no viral rebound as quantified by IFNγ ELISpot assay. f,g, Dynamics of HIV-specific IFNγ⁺ CD8⁺ T cells (top panel) and CD4⁺ T cells (bottom panel) in participants with late viral rebound (f) and no viral rebound (g) as determined by intracellular cytokine staining. SFC, spot-forming cells; f/u, follow-up.

Fig. 5. Proteomic profiles.

a, Proteomic pathways at week 24 compared with the baseline in the 10 participants who showed virologic control for the duration of bNAb therapy and who showed long-term virologic control for the duration of follow-up (controllers; n = 5) or who showed viral rebound (rebounders; n = 5). The size of the circle indicates the level of pathway enrichment; red refers to increased and blue refers to decreased levels for each pathway. b, Proteomic pathways modulated in controllers compared with rebounders at baseline before the initiation of bNAb infusions and at week 24 following completion of bNAb infusions but before viral rebound. Protein levels were averaged across individuals for each group. The color gradient indicates the log₂ transformation of the fold change, ranging from blue (decreased) to red (increased) protein levels.

Fig. 6. Rebound virus sequences.

Phylogenetic tree of participant baseline and viral rebound Envs. Baseline Envs are colored blue and rebound Envs colored red. Orange indicates Env sequences derived from PBMCs during the absence of plasma viral rebound. Participant identification numbers are shown at the root of each participant Env cluster. Envs that were tested for neutralization are shown by filled circles or triangles, and their neutralization IC₈₀ titers for each of the three bNAbs are shown on the right using colored boxes, with color coding per legend below.

Extended Data Fig. 1. CD4 + T-cell counts overtime and Group 1 serum bNAb pharmacokinetics.

CD4 + T-cell Counts over Time in participants that had received 3 (a) or 6 bNAb infusions (b). Pharmacokinetics of PGT121, VRC07-523LS and PGDM1400 in Group 1 A and 1B. Levels of VRC07-523LS (green), PGDM1400 (blue) and PGT121 (red) in serum in participants in group 1 A after a single infusion of PGT121 and VRC07-523LS at 30 mg/kg each (c) or in group 1B after infusion of PGT121, VRC07-523LS and PGDM1400 at 20 mg/kg each (d), as determined by anti-idiotype specific Binding Antibody Multiplex Assay. Data are mean ± s.d. Mean half-life (t1/2) of each bNAb is indicated in days.

Extended Data Fig. 2. Participant 65021: viral characteristics.

(a) Virological and pharmacokinetic dynamics (see Fig. 3 for details). (b) Phylogenetic tree with neutralization data and (c) Env evolution between baseline & rebound in the bNAb epitopes. Data demonstrates that pseudoviruses were completely sensitive to VRC07-523LS and resistant to PGDM1400 at baseline and rebound. PGDM1400 resistance might be explained by the presence of the resistance signatures N-130 and I-165 in all sequences. For PGT121, two baseline sequences were sensitive while the rebound sequences all likely arose from a resistant baseline variant carrying the resistance conferring mutation Y-330, consistent with PGT121 selection.

Extended Data Fig. 3. Participant 98059: viral characteristics.

(a) Virological and pharmacokinetic dynamics (see Fig. 3 for details). (b) Phylogenetic tree with neutralization data and (c) Env evolution between baseline & rebound in the bNAb epitopes. Participant 98059 was the only participant that demonstrated fully resistant virus at rebound, although one of two baseline viruses was sensitive to all three antibodies, while the other one showed resistance to PGDM1400. Rebound viruses carried the G459D signature, associated with VRC07-523LS resistance, and the G324R mutation in the key ³²⁴GDIR³²⁷ motif in the bNAb epitope known to confer PGT121 resistance. There were no clear PGDM1400 resistance signatures that differentiated between the sensitive baseline virus (V1-D6) and resistant baseline virus (V1-C4), however in V1-D6 there was an atypical additional pair of Cysteines in the hypervariable section of the V1 loop, which could have formed a disulfide bridge thus stabilizing the V1 loop structure, a phenomenon which has been associated with increasing sensitivity to V2-Apex antibodies. None of the rebound viruses carried the additional pair of Cysteines found in the sensitive baseline virus V1-D6.

Extended Data Fig. 4. Participant 98050: viral characteristics.

(a) Virological and pharmacokinetic dynamics (see Fig. 3 for details). (b) Phylogenetic tree with neutralization data and (c) Env evolution between baseline & rebound in the bNAb epitopes. All rebound Env sequences were resistant to PGT121, attributable to the absence of N332 glycan in each of the viruses, and all rebound sequences showed PGDM1400 resistance, attributable to resistance signatures V-169 and N130. Baseline sequence data demonstrated that Envs had identical sequences to the rebounding Envs at PGDM1400 and VRC07-523LS contact sites, suggesting that PGDM1400 resistance existed at baseline and that VRC07-523LS did not exert any significant selection pressure.

Extended Data Fig. 5. Participant 98058 and 98060: viral characteristics.

Virological and pharmacokinetic dynamics (see Fig. 3 for details), phylogenetic tree and Env evolution. (a) 98058 and (b) 98060 had been infected with HIV-1 at the same time and from the same individual, explaining the substantial HIV-1 sequence similarities that were observed (Fig. 6). All rebound Env sequences were resistant to PGT121, attributable to the absence of N332 glycan in each of the viruses, and all except 1 rebound sequence in participant 98060 showed PGDM1400 resistance, attributable to resistance signatures N130.

Extended Data Fig. 6. Participant 65010: viral characteristics.

(a) Virological and pharmacokinetic dynamics (see Fig. 3 for details). (b) Phylogenetic tree with neutralization data and (c) Env evolution between baseline & rebound in the bNAb epitopes. Rebounding virus was in the majority fully sensitive to all three bNAbs and demonstrated only single variants (that were resistant to PGDM1400 (2 of 7 tested pseudoviruses) or PGT121 (1 of 7 tested pseudoviruses). Virus was highly diverse at baseline, but only two variant sublineages were recovered in the rebound quasispecies.

Extended Data Fig. 7. Participant 65015: viral characteristics.

(a) Virological and pharmacokinetic dynamics (see Fig. 3 for details). (b) Phylogenetic tree with neutralization data and (c) Env evolution between baseline & rebound in the bNAb epitopes. Rebounding virus was fully sensitive to all three bNAbs.

Extended Data Fig. 8. Participant 98056: viral characteristics.

(a) Virological and pharmacokinetic dynamics (see Fig. 3 for details). (b) Phylogenetic tree and (c) Env evolution between baseline & PBMC proviral DNA at week 24 (approximately 16 weeks after the last bNAb administration) in the bNAb epitopes. The latter showed identical sequences to the baseline Envs at each bNAb contact site suggesting the absence of any selection pressure in this individual’s viral reservoir. Baseline resistance signatures in the core C-strand contact region of PGDM1400 (position 161, 165 and 169) was observed.

Extended Data Fig. 9. Participant 98005 and 98063: viral characteristics.

(a) Participant 98055 and (b) participant 98063. Virological and pharmacokinetic dynamics (see Fig. 3 for details), phylogenetic tree and Env evolution.

Extended Data Fig. 10. Neutralization profiles and complementarity of coverage for triple and dual bNAb combinations.

(a) IC80 titers for single and combination bNAbs against a heterologous panel of 374 pseudoviruses^,. Each row represents a pseudovirus, and the ordering of pseudoviruses for the left and right heatmaps is different. IC80 titers are colored per legend. For 2 and 3 bNAb combinations, IC80 titers were predicted using Bliss-Hill model on individual bNAb data using CombiNAber (https://www.hiv.lanl.gov/content/sequence/COMBINABER/combinaber.html). The left-most column indicates subtype of the virus colored per legend. (b) Percent coverage of sensitive viruses to only one bNAb in the combination (light green), and to two or more bNAbs is shown (dark green). Sensitivity threshold of IC80 < 1ug/ml is used. (c) Breadth-potency curves for triple and dual bNAb combinations showing overall coverage (light shades) or dual coverage (dark shades).