The Effects of Human Immunodeficiency Virus Type 1 (HIV-1) Antigen-Expanded Specific T-Cell Therapy and Vorinostat on Persistent HIV-1 Infection in People With HIV on Antiretroviral Therapy

Abstract

Background

The histone deacetylase inhibitor vorinostat (VOR) can reverse human immunodeficiency virus type 1 (HIV-1) latency in vivo and allow T cells to clear infected cells in vitro. HIV-specific T cells (HXTCs) can be expanded ex vivo and have been safely administered to people with HIV (PWH) on antiretroviral therapy.

Methods

Six PWH received infusions of 2 × 10⁷ HXTCs/m² with VOR 400 mg, and 3 PWH received infusions of 10 × 10⁷ HXTCs/m² with VOR. The frequency of persistent HIV by multiple assays including quantitative viral outgrowth assay (QVOA) of resting CD4⁺ T cells was measured before and after study therapy.

Results

VOR and HXTCs were safe, and biomarkers of serial VOR effect were detected, but enhanced antiviral activity in circulating cells was not evident. After 2 × 10⁷ HXTCs/m² with VOR, 1 of 6 PWH exhibited a decrease in QVOA, and all 3 PWH exhibited such declines after 10 × 10⁷ HXTCs/m² and VOR. However, most declines did not exceed the 6-fold threshold needed to definitively attribute decline to the study intervention.

Conclusions

These modest effects provide support for the strategy of HIV latency reversal and reservoir clearance, but more effective interventions are needed to yield the profound depletion of persistent HIV likely to yield clinical benefit.

Clinical Trials Registration. NCT03212989.

The HIV latency reversal agent vorinostat was well tolerated when given with expanded, autologous, HIV-specific cytotoxic T cells, but only modest effects were seen on the pool of latently infected cells that persist in people with HIV on therapy.

Human immunodeficiency virus (HIV) persists within diverse populations of cells despite antiretroviral therapy (ART) in varying states of transcriptional activity or quiescence [1, 2]. However, the persistence of quiescent, latent infection in memory CD4 T cells has been documented over years of ART [3, 4]. Attempts to combine agents that induce expression of quiescent proviruses (latency reversal agents [LRAs]) with immunotherapies that might deplete persistent infection have thus far failed to deplete persistent infection to the extent thought to be required for viral eradication [5].

However, HIV-specific CD8 T-cell responses insufficient to eliminate persistent infection [6]. Viral persistence may be fueled both by cellular proliferation [7] and by ineffective immune-mediated clearance [8, 9].

Adoptive T-cell therapy using autologous T cells has been successfully employed against other viruses in hematopoietic stem cell transplant recipients [10, 11]. In the HIV setting, polyfunctional, potent HIV-specific expanded T cells (HXTCs) can be produced that target cells expressing multiple HIV epitopes [12, 13]. HXTCs can be safely administered to PWH on ART [14]. In this pilot study, we tested the ability of the coadministration of HXTCs and the LRA vorinostat (VOR) to deplete persistent HIV infection in a broadly relevant cohort of PWH on ART.

METHODS

HXTC Generation and Clinical Protocol

The study was approved by the relevant institutional review boards (IRBs) (ClinicalTrials.gov identifier NCT03212989), and all participants provided written informed consent. HXTCs were generated as previously described [12–14]. PWH ≥18 and <65 years of age, with durable viral suppression for ≥24 months and CD4 count ≥350 cells/μL, were studied. As the primary endpoint of the study was a decline of infectious units per million (IUPM) by quantitative viral outgrowth assay (QVOA) of >0.5 log, an IUPM ≥0.30 was required to allow a definitive measure of a decline of 0.5 log. Furthermore, given the potential risks of these interventions in stably treated PWH, a measurable induction of resting cell-associated HIV RNA levels (rca-RNA) was required following exposure of participant cells to VOR ex vivo [15].

Quantitative Viral Outgrowth Assay

Human resting lymphocytes were acquired using continuous-flow leukapheresis, and quantification of replication-competent virus was performed [16]. Outgrowth is reported at day 15 for all participants, except for VH-16 and VH-24 where it is reported at day 19. IUPM was estimated by maximum likelihood method [17].

Intact Proviral DNA Assay

Resting CD4⁺ T cells were isolated as described above, and intact proviral DNA assay (IPDA) was performed [18]. Median DNA shearing index was 0.33 [quartile 1, quartile 3 [Q1, Q3]: 0.32, 0.34). A median of 1.2 × 10⁶ (Q1, Q3: 1.1 × 10⁶, 1.2 × 10⁶) resting CD4 T-cell equivalents was evaluated per sample. Gating of droplets were performed using no template, HIV-seronegative CD4 T-cell DNA, and gblock controls (Integrated DNA Technologies, Morrisville, North Carolina). Intact proviral frequencies <5 × 10⁶ cells were censored as previously described [19].

Resting CD4 rca-RNA and Low-Level Plasma Viremia Single Copy Assay

Changes induced in rca-RNA in vivo in response to VOR and HXTCs were measured as described previously [15] with minor modifications (see Supplementary Methods). Low-level plasma viremia was measured from ultracentrifuged plasma samples [20].

Histone Deacetylase Inhibitor Host Biomarker Gene Expression Analysis

RNA was extracted (NucleoMag RNA extraction kit; Takara Bio USA, San Jose, California) from approximately 2 million peripheral blood mononuclear cells (PBMCs) per timepoint. Complementary DNA (cDNA) reactions for each sample of up to 1 µg were prepared with the Maxima cDNA Synthesis Kit with dsDNAse (ThermoFisher). cDNA was diluted 1:3 and subsequently used for quantitative polymerase chain reaction (PCR) in technical duplicates for H1F0 and IRGM (upregulated histone deacetylase inhibitor [HDACi] genes] and PHF15 and PRDM10 [downregulated HDACi genes] using QuantiTect Multiplex PCR NoROX Mastermix [Qiagen, Germantown, Maryland) and the primer/probes described previously [21]. Fold change in gene expression was determined using the 2^−ΔΔCt method with normalization to the reference gene RPL27 and to a baseline sample (before initiation of the 2 cycles of 10 q72h doses of VOR).

RESULTS

Characteristics of Participants

All 9 participants who received VOR and HXTCs (6 at 2 × 10⁷ HXTCs/m², 3 at 10 × 10⁷ HXTCs/m²) were males with a median age of 52 years (range, 25–63 years) and stably suppressed on ART at the time of study enrollment. The median duration of ART was 8 years; other characteristics are listed in Table 1 (also see Supplementary Methods). The median IUPM as measured by QVOA was 0.761 (range, 0.309–5.576) cells per million (Table 1).

Table 1.

Clinical Characteristics and Baseline Reservoir Size^a

Study ID	Age, y	Sex	Race/Ethnicity	Stage ART Initiatedb	CD4 Nadir	CD4 at Study Entry	Years on ART	IUPMc
VH-10	57	M	White	Chronic	195	701	30	0.761
VH-11	54	M	White	Chronic	174	785	4	1.112
VH-12	25	M	White	Chronic	323	740	2	5.576
VH-13	30	M	Black/African American	Chronic	467	642	6	0.608
VH-16	55	M	White	Chronic	NA	923	14	0.412
VH-17	51	M	Black/African American	Chronic	335	661	8	0.403
VH-22	63	M	Black/African American	Chronic	352	927	28	8.010
VH-23	33	M	Black/African American	Acute	795	1012	3	0.442
VH-24	52	M	Hispanic	Chronic	13	461	16	0.381

Abbreviations: ART, antiretroviral therapy; IUPM, infectious units per million; M, male; NA, not applicable.

^aAll values were taken at time of screening or baseline leukapheresis.

^bAcute human immunodeficiency virus infection was defined as in Gay et al [32].

^cAs measured by the quantitative outgrowth assay following maximal mitogenic stimulation with phytohemagglutinin as previously described [15].

HXTC Product Expansion, Phenotype, and Specificity

HXTCs from 7 of 10 participants were successfully expanded to numbers sufficient for use in the clinical trial after 2 rounds of stimulation. HXTC products for VH-12 and VH-20 required a third stimulation to reach the number of cells needed for dosing (Figure 1A). The median total fold expansion starting with bulk lymphocytes was 55.5-fold (range, 10.4- to 435.7-fold). HXTCs exhibited specificity for the Gag, Nef, and Pol peptide pools as measured using interferon gamma (IFN-γ) enzyme-linked immunosorbent spot assays, with negligible nonspecific activity (Figure 1B, Supplementary Figures 1 and 2, Supplementary Methods). The magnitude and breadth of HIV-specific T-cell responses within the HXTC product varied among participants. HXTCs from 5 participants showed T-cell responses to all 3 antigens, whereas VH-12 showed T-cell responses to 2 of the 3 peptide pools. The median magnitude of T-cell responses was 222.5. 57.3, and 434.4 IFN-γ spot-forming cells/10⁵ cells to Gag, Pol, and Nef, respectively.

Figure 1.

Characterization of human immunodeficiency virus (HIV)–specific T-cell (HXTC) products. A, HXTCs expanded to clinically relevant levels after 2 stimulations except for VH-22, which underwent 3 stimulations (second stimulation shown as .41 and third as .42). *VH-20 failed to expand and did not proceed in the study. B, HXTCs display HIV specificity. All products produced interferon-γ in response to HIV Gag, Nef, and/or Pol peptide mix stimulation, or actin (Act) as a negative control, and measured by enzyme-linked immunosorbent spot assay. VH-13 was tested as a pool of Gag/Pol/Nef and not individual peptides. C, HXTC product phenotyping was analyzed by fluorescence-activated cell sorting, to show subpopulations expressed as percentage of lymphocytes, T cells (CD45⁺CD3⁺), CD4 T cells (CD3⁺CD4⁺), CD8 T cells (CD3⁺CD8⁺), natural killer cells (CD3^–CD56⁺CD16⁺), monocytes (CD45⁺CD14⁺), B cells (CD19⁺), and dendritic cells (CD3^–CD83⁺CD16⁺). D, Exhaustion markers of HXTC by flow cytometry. Cell products displayed overall low expression of markers associated with exhaustion. Abbreviations; ELISpot, enzyme-linked immunosorbent spot assay; HXTC, human immunodeficiency virus–specific T cell; IFN-γ, interferon gamma; NK, natural killer; SFC, spot-forming cells.

HXTCs were predominantly CD8 T cells (median, 45.8% [range, 9.2%–91.2%]), although natural killer cells expanded within the HXTCs from some participants, with the largest proportion seen in HXTC-13 (89.4%) and HXTC-20 (71.6%) (Figure 1C) as has been reported with other antigen-specific T-cell products expanded in the presence of interleukin 15 [23, 24]. As expected, there were no residual dendritic cells or co-stimulatory K562 cells detected at the end of the expansion period. Despite the 2 rounds of stimulation, HXTC cells displayed detectable but low levels of immune exhaustion markers, including CD160, CTLA-4, LAG-3, PD-1, PDL1, TIGIT, and TIM-3 (Figure 1D). HXTCs also demonstrated cytotoxicity against autologous target cells pulsed with Gag, Pol, and Nef peptide pools in Chromium 51 release assays. Cytotoxicity was HIV-specific (Supplementary Figure 3).

HXTC Expression of Homing Markers to Gut and Lymph Node Tissues

Homing markers, including integrin β7 and the chemokine receptor CXCR5, were examined to determine whether HXTCs exhibited the potential to home to putative tissue sites of persistent HIV infection. CD8⁺ T cells were evaluated, chosen based on cell availability. The combination of CCR9⁺INTβ7⁺, associated with homing to the gut-associated lymphoid tissue, was expressed at a high range of frequencies in HXTCs (median, 60.8% of memory CD8 T cells [range, 29.4%–93.8%]) (Figure 1D). CXCR5 was only expressed on a minority of memory T cells (median, 0.3% [range, 0.1%–1%]).

Clinical Trial Protocol

Six participants received VOR 400 mg every 3 days for 10 doses, with two 2 × 10⁷ HXTCs/m² infusions at day 0 and week 2. As in prior studies, VOR was well tolerated, and after a rest period of 4 or more weeks, participants received a second cycle of VOR 400 mg every 3 days for 10 doses, with 3 infusions of 2 × 10⁷ HXTCs/m² at day 0, week 2, and week 4, over the month of VOR dosing. Three participants underwent initial evaluations but terminated study participation prior to production of HXTCs as latent reservoir frequency or inducibility was insufficient, and 2 participants withdrew consent following initial screening, during the coronavirus disease 2019 pandemic.

After an interim analysis of data in the first 6 participants, the protocol was modified with IRB approval to simplify study visits, remove the requirement for a test dose of VOR (as this was universally seen in participants where induction could be measured ex vivo), and increase the HXTC dose to 10 × 10⁷ HXTCs/m². Six additional participants were then enrolled and screened, but 2 participants had a baseline IUPM <0.3, and for 1 participant insufficient HXTCs were generated for study infusions. The 3 eligible participants received VOR and HXTCs as before, but up to 10 × 10⁷ HXTCs/m² during the 5 infusions in 2 cycles. In total, given the available HXTCs produced, VH-22, VH-23, and VH-24 received 13.08 × 10⁸, 13.08 × 10⁸, and 11.04 × 10⁸ HXTCs, respectively.

VOR Dosing, HXTC Infusion, and VOR Responses

Overall, VOR and cell infusions were safe and well tolerated. Participants had several grade 1 adverse events possibly related or related to VOR and/or HXTCs, but all were self-resolving and none were dose-limiting (53 events; median, 6 events per participant [range, 3–10 events per participant]). In 1 participant, a single episode of grade 2 increase of alanine aminotransferase and grade 4 increase of aspartate aminotransferase was noted, resolved with VOR interruption, and did not recur with VOR and HXTC rechallenge.

VOR dosing consistently modulated HDACi-responsive host genes, as assessed by expression analysis of 4 genes (H1F0, IRGM, PRDM10, and PHF15) [21]. We compared gene expression changes relative to a baseline pre-VOR//HXTC timepoint-level data for several VOR dose timepoints (3–6 hours postdose) and washout timepoints (>24 hours postdose) per participant, with remarkably consistent modulation of HDACi-responsive host genes (Figure 2A–D), whether examined by individual participant (Supplementary Figure 4) or individual VOR dose (Supplementary Figure 5).

Figure 2.

Histone deacetylase (HDAC)–responsive gene expression following vorinostat (VOR) administration. HDAC-responsive gene expression following VOR dose versus washout data for all timepoints and participants examined. Each plot represents combined cohort level data for a different HDAC inhibitor–inducible gene. The y-axes represent fold change in gene expression relative to baseline (before initiation of 2 × 10 q72h VOR cycles) timepoint. P values from Mann–Whitney U test.

Frequency of HIV-Specific T Cells as Measured by IFN-γ Release Following HXTC Infusion

In PWH on ART, HIV-specific T-cell responses are highly stable over time [25]. Based on this, our study had >90% power to detect a 2-fold change in the HIV-specific T-cell response [25]. Although HIV-reactive T-cell responses were detectable in all participants at baseline, a <2-fold change in T-cell response to Gag, Pol, or Nef proteins was observed following HXTC infusions in the majority of participants (Figures 3 and 4).

Figure 3.

Human immunodeficiency virus (HIV)–specific T-cell (HXTC) responses over time. T-cell responses to the HIV proteins Gag, Pol, Nef, and Env and HIV accessory proteins (ACC: Rev, Vif, Vpr, Tat) and a pool of influenza, Epstein-Barr virus, and human cytomegalovirus optimal CD8 T-cell epitopes (FEC55) were measured in peripheral blood mononuclear cells of 6 participants by ex vivo interferon-γ enzyme-linked immunosorbent spot assay. Mean and standard error of the mean are shown. The block areas indicate the first and second HXTC + vorinostat treatments, respectively. Gag, Pol, and Nef pools were obtained from JPT. Abbreviations: HXTC, human immunodeficiency virus–specific T cell; PBMCs, peripheral blood mononuclear cells; SFU, spot-forming cells; VOR, vorinostat.

Figure 4.

Human immunodeficiency virus (HIV)–specific T-cell (HXTC) responses over time following high-dose HXTC administration. T-cell responses to the HIV proteins Gag, Pol, Nef, and Env and HIV accessory proteins (ACC, Rev, Vif, Vpr, Tat) and a pool of influenza, Epstein-Barr virus, and human cytomegalovirus optimal CD8 T-cell epitopes (FEC55) were measured in the peripheral blood mononuclear cells of 3 participants by ex vivo interferon-γ. Mean and standard error of the mean are shown. The block areas indicate the first and second HXTC + vorinostat treatments, respectively. JPT refers to peptide source. Abbreviations: HXTC, human immunodeficiency virus–specific T cell; PBMCs, peripheral blood mononuclear cells; SFU, spot-forming cells; VOR, vorinostat.

Across participants receiving HXTCs and VOR (n = 9), treatments did not significantly impact the frequency of HIV Gag, Nef, or Pol-specific T-cell responses (all comparisons P > .5, Wilcoxon signed-rank test comparing baseline to 2–4 weeks after final treatment). In VH-12, HXTC-specific T-cell responses increased >2-fold following treatment but only to low, absolute frequencies, <100 spot-forming cells/10⁶ PBMC. In VH-17, a 2-fold increase in Nef-specific T-cell response was observed following the second HXTC treatment.

We assessed the antiviral activity of CD8 T cells isolated from PBMCs of participants following HXTC infusion [14]. Baseline antiviral activity varied, but no significant change in the antiviral activity of circulating cells was observed at timepoints up to 4 weeks following HXTC infusion (data not shown).

Increase and Persistence of HXTC T-Cell Receptors Over Time

Using T-cell receptor (TCR) vβ sequencing, TCR clonotypes present in the HXTC products, but not present in preinfusion samples, were tracked in the peripheral blood of participants postinfusion (see Supplementary Methods), showing unique clonotypes detected in the peripheral blood of participants after HXTC infusion (Supplementary Figure 6). Expansion of unique clones was demonstrated in all evaluable participants postinfusion. As a comparator (Supplementary Figure 7), we examined TCRs in participants who received VOR alone during screening but were not eligible to receive combination study therapy; a similar increase in TCR clonotypes was not observed.

Measures of HIV RNA, HIV DNA, and Persistent HIV Infection

Following 2 cycles of VOR and HXTC, resting CD4 T-cell replication-competent virus frequencies were determined using QVOA. The cells that are quantitated in this assay may be sensitive to latency reversal, and then are likely to be targets for immune-mediated clearance. In 5 of 6 participants in the 2 × 10⁷ HXTC cohort, no significant decline in viral outgrowth frequencies was seen. In VH-12, a 5-fold decline in QVOA was measured (Figure 5), a decline of a magnitude only seen in 2 of 123 pairs of consecutive QVOA measurements in a stably ART-treated cohort [3]. Intriguingly, all 3 participants in the 10 × 10⁷ HXTC cohort had declines of IUPM of 4.0-fold, 3.5-fold, and 2.6-fold (IUPMs: VH-22, 8.040 to 2.014; VH-23, 0.442 to 0.127; VH-24, 0.381 to 0.148).

Figure 5.

Impact of vorinostat and HXTC on human immunodeficiency virus (HIV) reservoir frequency. A, Change in HIV reservoir frequency from baseline to endpoint for each participant for both intact DNA (determined by intact proviral DNA assay, upper lines) and outgrowth virus (determined by quantitative viral outgrowth assay, lower lines). B, Change in infectious units per million resting CD4 (rCD4) cells for each individual (left) and change in intact DNA per million rCD4 cells from baseline to endpoint for each individual (right). Abbreviations: HIV, human immunodeficiency virus; IUMP, infectious units per million; rCD4, resting CD4 cells.

In this study, 2 treatment cycles were undertaken. To provide an estimate of reservoir reduction should the effects of study treatment be extended with further cycles of study therapy, we fitted a mathematical model to the QVOA changes observed in each participant (see Supplementary Methods). If a total of 4 cycles had been administered, the predicted decreases in VH-12, VH-22, and VH-23 would be 21-, 16-, and 17-fold, respectively, whereas if 6 total cycles were administered, we expect 98-, 63- and 69-fold decreases, respectively, although there exist large uncertainties in these predictions (Supplementary Figure 9).

HIV DNA (intact, 3′-defective, 5′-defective, and total) frequencies were also assessed using IPDA. While IUPM by QVOA is a minimal estimate of replication-competent provirus, some proviruses detected by IPDA may not be fully intact [18, 26] and may not express epitopes or antigens that would allow immune targeting. Changes in intact DNA frequencies were largely concordant with that observed for QVOA. Analysis of the entire cohort using Wilcoxon matched-pairs signed-rank test did not reveal statistically significant differences for intact DNA (median difference, −4; P = .21) or outgrowth virus (median difference, 0.23; P = .13). No clear differences in 3′-defective, 5′-defective, or total DNA were observed (Supplementary Figure 10).

Finally, changes in resting CD4⁺ T-cell–associated HIV unspliced gag RNA (rca-RNA) was assessed as another potential measure of persistent replication-competent provirus (Supplementary Figures 11 and 12). As a screening criterion in this study, a measurable upregulation of rca-RNA was validated in participants’ cells ex vivo.

Following 2 cycles of VOR and HXTC administration, 2 of the 5 participants without a substantial change in IUPM by QVOA also had no change in rca-RNA (Supplementary Figure 11). One participant had a small but statistically significant decline in rca-RNA. Two participants had a modest but statistically significant increase in unspliced gag HIV rca-RNA, a phenomenon that has been reported to persist following LRA exposure [27]. VH-12, in whom a moderate decline in IUPM by QVOA was measured, had a substantial and statistically significant decline in rca-RNA. However, none of the 3 participants in the high-dose HXTC cohort, all of whom had modest declines in IUPM, had a change in rca-RNA (Supplementary Figure 12).

Persistent low-level viremia can be detected despite clinically successful ART-mediated suppression, may emanate from persistently infected cells, and might be reduced if these cells were cleared [28]. In this study cohort, although viremia measured by a low-copy assay trended downward in some cases, there was no clear overall effect of study therapy on low-level viremia (Table 2). Further, the recent description of persistent “nonsuppressible” low-level plasma HIV RNA carried by defective HIV particles may confound this assessment [29].

Table 2.

Low-Level Viremia (Plasma Human Immunodeficiency Virus Type 1 RNA/µL)

Timepoint	VH-10	VH-11	VH-12	VH-13	VH-16	VH-17	VH-22	VH-23	VH-24
Prebaseline	1.0	2.4	4.0	1.8	0.51	4.7	0.55	<0.26	1.0
Baseline	0.64	2.1	5.6	15	<0.23	4.8	<0.20	0.23	0.80
End of first cycle	0.79	1.7	3.1	2.9	<0.19	4.1	0.30	<0.27	4.2
End of second cycle	<0.17	1.5	3.1	0.59	<0.23	NA	5.4	0.64	2.7

Abbreviations: NA, not applicable.

DISCUSSION

The testing of experimental approaches to deplete the persistent reservoir of HIV infection, toward the goal of viral eradication, requires careful study of the dosing and timing of LRA interventions and of immunotherapies for viral clearance. Such studies must be coupled with an array of assays that carefully measure changes in markers of persistent HIV infection [30]. In this study, HXTC infusions with concomitant VOR dosing were safe and well tolerated in HIV-infected, ART-suppressed PWH, documenting only transient grade 1 or 2 adverse events. Few mild treatment-related adverse events were observed. The safety and tolerability of this combination study are consistent with results of previous trials using these reagents alone [31].

As in other studies [21, 32, 33], serial VOR dosing reliably induced biomarkers of HDACi response, demonstrating target engagement. As in prior studies, intermittent induction of plasma viremia was not been observed. In fact, induction of transient viremia by LRAs has generally not been seen in human studies. More potent, novel LRAs, alone and in combination, can induce viremia in animal models [22], but these approaches have not yet been translated into human trials. However, clinically effective latency reversal—that which allows an infected cell to be recognized and cleared by an innate, adaptive, or engineered immune response—might occur at levels that do not result in detectable plasma viremia. It is with this possibility in mind that this and other studies have assessed the combined in vivo effect of an LRA and an immunotherapy.

A prior study tested VOR using the same dosing strategy in combination with a dendritic cell vaccine [32], but vaccine response was poor and no effect was measured on persistent infection. Another study tested VOR using the same dosing strategy, but in combination with an modified vaccine Ankara–vectored T-cell vaccine in recently infected, recently treated PWH. Vaccination was associated with increased circulating T-cell responses to HIV, but again no effect on persistent infection was observed. The substantial ART-induced decline in measures of persistent infection in early-treated individuals may have interfered with the ability to measure modest declines in persistence induced by study interventions [34]. A third study [35] tested the effect of romidepsin, another HDACi, with the broadly neutralizing antibody (bnAb) 3BNC117. Although the serial efficacy of repeated doses of romidepsin has yet to be clearly defined, measures of persistent infection did not convincingly decline in the combination arms of this study. The eCLEAR study [36] tested the effects of these same reagents in the setting of recent infection and early ART. Decreases in persistent infection were reported in those receiving bnAbs with early ART, but romidepsin did not contribute to this effect. This suggested that this effect was due to clearance of cells that were establishing latent infection, rather than durable latent infection seen after long-term ART. The VOR07 study [33] tested VOR with a bnAb after long-term ART and saw only a modest trend toward depletion in a minority of participants.

This study tested VOR with autologous T cells expanded ex vivo in the presence of viral peptides and cytokines, deriving pools of HXTCs. Modest declines of QVOA were seen in 3 of 6 participants receiving 2 × 10⁷/HXTC/m², but in only 1 of these 3 was this decline likely to be outside the range of assay variability. Changes in IPDA, a more tractable but less specific measure of persistent, latent HIV infection, were generally concordant with those seen in QVOA.

Notably, 3 of 3 participants who received 10 × 10⁷/HXTC/m² infusions exhibited declines of QVOA, although none quite exceeded the previously defined 6-fold threshold to definitively attribute the reservoir decline to the study intervention. By way of comparison, as such declines were seen in 21 of 123 pairs of consecutive QVOA measurements [3], the depletion of persistent infection measured in the 10 × 10⁷ HXTC cohort cannot be definitively attributed to the VOR/HXTC study intervention. However, longitudinal trends are consistent with a treatment effect in VH-12, VH-22, and perhaps VH-23 (Supplementary Figure 8).

Changes in IPDA were again generally concordant with those seen in QVOA. As these participants also participated in studies before or after this one, longitudinal QVOA and IPDA data are available in some cases (Supplementary Figure 3). Longitudinal measurements of QVOA suggest that the modest declines in VH-12 in the low-dose arm, and in VH-22 and VH-23 in the high-dose arm, may be related to study therapy. Predictions from a mathematical model suggest that additional cycles of treatment may lead to substantial declines in these 3 individuals.

Changes in rca-HIV RNA in both HXTC dosing arms were variable, as might be expected in the settings of combined interventions with opposing effects; LRAs might increase the level of viral RNA expression, and HXTCs might reduce the survival of cells expressing antigen products of some intact viral RNAs. Recent work [37] suggests that most viral transcripts are defective, and so it may be difficult to identify a cell as expressing antigen that would allow clearance by simple quantitation of viral transcripts. Conversely, changes in low-level viremia were not seen, suggesting that cells producing virions might persist despite the administration of HXTCs. However, other recent studies have also identified nonsuppressible viremia that originates from small clones of proliferating CD4⁺ cells [29].

We did not observe any consistent increase in the magnitude of the HIV-specific immune response, as measured by IFN-γ release, in circulating PBMCs following HXTC administration. This may be due to the absence of a sufficient in vivo stimulus for expansion of the cells, including the lack of concomitant robust HIV antigen exposure, no preinfusion lymphodepletion or myeloablative therapy, and no additional cytokine support that would promote in vivo expansion of the cells. Alternatively, HXTCs may be preferentially distributed within tissue, where most infected cells undergoing latency reversal would be expected to reside.

Overall, serial administration of HXTCs and VOR was well tolerated but did not substantially deplete markers of persistent infection. Any depletion of <6-fold in the current QVOA assay could be caused by biological or assay variation [3]; no declines of >6-fold in QVOA were observed in this study. Although there were no differences in latency reversal markers or anti-HIV immune responses in the 2 HXTC dosing groups, there were some modest trends toward viral clearance seen more consistently in participants who received a higher dose of HXTCs. However, even if these trends were validated in a larger, controlled study, we would still conclude that achieving an HIV cure requires more effective latency reversal coupled with efficacious immune interventions.

Supplementary Data

Supplementary materials are available at The Journal of Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author.

Notes

Acknowledgments. We thank Y. Xu, M. L. Clohosey, B. Allard, and K. James for technical assistance; F. Maldarelli and R. Gorelick at National Cancer Institute (NCI) Frederick for the HIV RNA single copy assay; R. Bosch, M. Hudgens, and K. Mollan for statistical advice; the University of North Carolina (UNC) Blood Bank for apheresis management; and A. Adamo, C. Baker, H. Thaxton, T. Whitaker, and the UNC Clinical Trials unit for trial execution. We also thank the study participants for their donations to this research.

Author contributions. C. L. G., J. D. K., S. M. P., J. J. E., and D. M. M. executed the clinical trial. C. M. M., C. L., P. C., J. T., A. S., A. D., C. D. M., U. S., P. J. H., and C. M. B. produced and analyzed the HXTCs. N. A., S. D. F., S. L. M. R., C. M. M., C. L., N. G., and D. M. M. designed, performed, and analyzed the results of cellular, virological, and immunological assays. All authors contributed to the writing and editing of the manuscript.

Financial support. This work was supported by National Institutes of Health (grant numbers R01 HL132791 to D. M. M. and C. M. B., UM1 AI164567 to D. M. M., F30 AI145588 to S. D. F., and R01 AI152703 to R. K.); the NCI (P30 CA016086 Cancer Center Core Support Grant to the UNC Lineberger Comprehensive Cancer Center for the UNC Flow Cytometry Core Facility); the UNC Center for AIDS Research (grant number P30 AI50410); and the National Heart, Lung, and Blood Institute/NIH (Production Assistance for Cell Therapy, grant number NHLB1-N01 HB37163 to Baylor College of Medicine).