Pembrolizumab induces HIV latency reversal in people living with HIV and cancer on antiretroviral therapy

Abstract

In people living with HIV (PLWH) on antiretroviral therapy (ART), virus persists in a latent form where there is minimal transcription or protein expression. Latently infected cells are a major barrier to curing HIV. Increasing HIV transcription and viral production in latently infected cells could facilitate immune recognition and reduce the pool of infected cells that persist on ART. Given that programmed cell death protein 1 (PD-1) expressing CD4⁺ T cells are preferentially infected with HIV in PLWH on ART, we aimed to determine whether administration of antibodies targeting PD-1 would reverse HIV latency in vivo. We therefore evaluated the impact of intravenous administration of pembrolizumab every 3 weeks on HIV latency in 32 PLWH and cancer on ART. After the first infusion of anti-PD-1, we observed a median 1.32-fold increase in unspliced HIV RNA and 1.61-fold increase in unspliced RNA:DNA ratio in sorted blood CD4⁺ T cells compared to baseline. We also observed a 1.65-fold increase in plasma HIV RNA. The frequency of CD4⁺ T cells with inducible virus evaluated using the tat/rev limiting dilution assay was higher after 6 cycles compared to baseline. Phylogenetic analyses of HIV env sequences in a participant who developed low concentrations of HIV viremia after 6 cycles of pembrolizumab did not demonstrate clonal expansion of HIV-infected cells. These data are consistent with anti-PD-1 being able to reverse HIV latency in vivo and support the rationale for combining anti-PD-1 with other interventions to reduce the HIV reservoir.

Trial registration: ClinicalTrials.gov NCT02595866.

Fig. 1.. Study schema of specimen collection in relation to pembrolizumab administration.

Cycles of pembrolizumab (red triangles) were administered every 3 weeks on day 1 of a cycle for up to 35 cycles for participants benefiting from therapy. Participants discontinued treatment in the setting of unacceptable adverse events, progressive cancer, or participant preference. Blood was collected at regular time points (shown as dashed vertical lines), including before pembrolizumab administration on day 1, and plasma and peripheral blood mononuclear cells (PBMCs) were processed and cryopreserved. More frequent collection of specimens occurred after the first dose (red shaded box). DNA and RNA were extracted from CD4⁺ T cells isolated from PBMCs, and HIV DNA and unspliced and multiply spliced HIV RNA were quantified by PCR (shown as +). Plasma HIV RNA was quantified at the same time points. PBMCs collected at baseline, before cycle 7, and before cycle 13 were used for the inducible HIV using the tat/rev induced limiting dilution assay (TILDA; red tick). For participants receiving less than 7 cycles of therapy, TILDA was evaluated at the end of therapy.

Fig. 2.. Baseline measures demonstrated persistent forms of HIV on ART.

Baseline (day 1) plasma HIV RNA (copies/ml), unspliced HIV RNA (usRNA; copies per 10⁶ cells), HIV DNA (vDNA; copies per 10⁶ cells), and multiply spliced HIV RNA (msRNA; copies per 10⁶ cells) are shown for participants stratified by CD4⁺ T cell count. Values greater than zero were transformed by log₁₀(y) and grouped in 10 bins per log₁₀ unit, with exact zeros set apart for visualization.

Fig. 3.. Anti–PD-1 affects measures of HIV in CD4-sorted T cells.

Estimated within-participant changes from cycle 1, day 1 administration of pembrolizumab of (A) unspliced HIV RNA, (B) HIV DNA, (C) the ratio of unspliced HIV RNA:HIV DNA, and (D) plasma HIV RNA are shown. The y axis scale is log-transformed. Dots represent point estimates from regression analyses, and error bars indicate 95% confidence intervals; dashed line indicates no change from baseline; and exclusion of dashed line from confidence interval indicates P < 0.05 by Wald test of regression coefficient.

Fig. 4.. After a single infusion of pembrolizumab, there is an increase in unspliced HIV RNA and the ratio of unspliced HIV RNA:HIV DNA.

Waterfall plots of changes in (A) plasma HIV RNA, (B) unspliced HIV RNA, (C) HIV DNA, and (D) ratio of unspliced HIV RNA:HIV DNA from baseline administration of pembrolizumab on day 1 are shown for individual patients. Participants missing baseline or visit measures were excluded from the corresponding waterfall plots.

Fig. 5.. Repeated administration of anti–PD-1 over 6 cycles increased the inducible HIV reservoir in CD4⁺ T cells.

Changes in the frequency of cells with inducible HIV as measured by TILDA are shown stratified by initial comparisons performed at the prespecified cycle 7 (C7) time point versus an earlier time point due to dis-continuation of therapy. Orange triangles represent the final study time point. Cessation of the study participation occurred because of toxicity or progression of malignancy. EOT indicates end of therapy.

Fig. 6.. Comparison of HIV env sequences and clonal composition over time revealed no evidence of clonal expansion.

(A) The proportional distribution of 13 clones over four time points (cycle 1, cycle 4, cycle 7, and EOT) is shown. Clones are represented by color. (B) The percent of total env sequence at each time point are shown. Clone 12, which represented 34.4% of all clones at baseline, declined proportionally over time. (C) Pairwise distance analyses of distance from most recent common ancestor (MRCA) reveal no significant change over time. Data are presented as means and SD.

Fig. 7.. Phylogenetic analysis of HIV env sequences isolated from CD4⁺ T cell subsets over 2 years in a participant receiving pembrolizumab reveals no evidence of clonal expansion.

Sequences were categorized by time point (color) and T cell subset (shape). The outgroup is the prototypical HIV type 1 subtype B (isolate HXB2) (empty triangle). The scale indicates the number of nucleotide substitutions per site. Asterisk (*) indicates branches supported by >74% bootstrap replicates. NV, naïve; CM, central memory; TM, transitional memory; EM, effector memory; TD, terminally differentiated.