Immunotherapy for HIV Treatment and Cure in the Era of Personalized Medicine

1. Introduction

While antiretroviral therapy (ART) has transformed the clinical course of human immunodeficiency virus (HIV) infection, markedly improving survival [1] and preventing sexual transmission during sustained viral suppression [2], it does not cure the infection. Viral reservoirs persist despite treatment [3-5], and immune recovery can remain incomplete. Adverse effects and continuing adherence demands add to the burden of lifelong therapy [6].

Concern about long-term toxicity helped motivate SMART (Strategies for Management of Antiretroviral Therapy), which tested whether episodic, CD4-guided treatment could reduce drug exposure without compromising clinical outcomes. Interruption instead increased opportunistic disease or death and major cardiovascular, renal or hepatic disease, without reducing the adverse events associated with ART [6].

Immunotherapy offers a way to address these limitations: sustaining antiviral activity with less frequent treatment, improving immune recovery, or enabling lasting control after ART withdrawal. Trials pursuing the latter use analytical treatment interruption (ATI), with close virological monitoring and prespecified restart criteria [7]. The aim is sustained remission, in which control continues after ART stops despite potentially persistent replication-competent virus, or ultimately eradication of all infection capable of restarting systemic replication [8]. Long-acting antibodies and persistent cellular products may themselves sustain suppression after ART withdrawal, making their duration of activity important to the interpretation of control [7,9]. The durability of control depends on both the reservoir that remains and the host response capable of containing it.

2. Viral persistence and immune control

The difficulty of clearing HIV-1 is rooted in its biology as a member of the genus Lentivirus within Retroviridae. Integration into the host genome, CD4-cell tropism, mucosal injury and immune dysregulation link viral persistence to the progressive impairment of host defense [10,11]. ART suppresses productive infection while leaving a latent reservoir [3,4]; inducible, infectious virus remains recoverable even after decades of treatment, sustained partly by infected-cell proliferation [5].

Proviral integrity, inducibility, integration context, antigen expression and tissue distribution influence both viral rebound and the response to immunotherapy. Profiling studies reveal persistent cells with distinct inhibitory and survival characteristics [12,13]. Blood DNA captures only part of this biology. Clearance depends on whether persistent infection can resume transcription and whether immune effectors can reach, recognize and kill the infected cells [14].

Despite these obstacles, natural controllers demonstrate that prolonged containment is biologically possible. Elite controllers maintain very low or undetectable viraemia without ART, whereas viremic controllers sustain low but measurable virus [15,16]. Post-treatment controllers maintain control after stopping previous ART [17]. Controller-associated natural killer (NK) phenotypes and epigenetic states, together with cellular and humoral findings in post-treatment controllers, suggest multiple protective mechanisms [18-20]. Transient viral exposure can strengthen antibody responses, linking humoral immunity to the course of rebound.

Yet some elite controllers retain immune activation and vascular abnormalities despite low viraemia, with inconsistent hospitalization findings across cohorts [21-23]. Reproducing their viral control alone may leave inflammation and morbidity unresolved [24], strengthening the rationale for interventions that improve immune health as well as suppress replication.

3. Clinical Evidence

Antibody-based approaches have brought this therapeutic ambition closest to clinical application. Broadly neutralizing antibodies (bNAbs) act directly against susceptible virus when exposure remains sufficient. Antibody-based salvage is already clinically relevant: ibalizumab, a CD4-directed post-attachment inhibitor, demonstrated activity against multidrug-resistant HIV alongside optimized background treatment [25]. Leronlimab, also called PRO 140, has shown short-term antiviral activity through blockade of C–C chemokine receptor type 5 (CCR5) [26].

The potential to extend antibody-mediated control beyond conventional ART was tested in RIO, where 68 early-treated men received 3BNC117-LS plus 10-1074-LS or placebo before ATI. At week 20, estimated freedom from protocol-defined rebound was 75% versus 11%, with a hazard ratio of 0.09 (95% confidence interval 0.04–0.21) [27]. The endpoint—six consecutive HIV RNA measurements above 1,000 copies/mL or two above 100,000 copies/mL—allowed periods of detectable viraemia. Residual antibody activity overlapped the period of control, while subsequent kinetics bear on the possibility of longer-lasting remission. A companion analysis associated outcomes with viral sensitivity and autologous antibody activity [28].

Long-acting maintenance regimens offer a more immediate route to reducing treatment burden. Lenacapavir with teropavimab and zinlirvimab, and cabotegravir with VRC07-523LS, have been evaluated for sustained antiviral suppression [29,30]. In Botswana, Tatelo achieved antibody-supported suppression in selected early-treated children with favorable reservoir characteristics [31]. These pediatric findings also highlight the importance of age-dependent pharmacokinetics and early reservoir establishment.

Therapeutic vaccines seek to strengthen endogenous immunity and make control less dependent on passive antibody administration. Their strong immune responses have translated less consistently into virological benefit. AELIX-002 enrolled 44 men and one woman; all participants entering ATI rebounded despite broad responses to HIVACAT T-cell immunogen (HTI) vaccines. In AELIX-003, HTI vaccination plus vesatolimod, a Toll-like receptor 7 agonist, did not significantly improve the principal interruption outcomes, and all 47 participants entering ATI rebounded [32,33]. RIVER (Research in Viral Eradication of HIV Reservoirs) found no reservoir benefit from vaccination and latency reversal over ART alone in recently infected men [34]. Improving recognition remains biologically attractive, but antigen availability and effector function may limit the benefit of vaccination alone.

Tat vaccination has also been evaluated for immune recovery during continued ART. A South African randomized phase II trial, conducted predominantly in women with clade C infection, reported anti-Tat responses and CD4 increases [35]. Its 12-year follow-up described earlier immune and blood-DNA changes, although attrition, ART changes and generally nonsignificant overall between-group comparisons weaken the evidence for a durable vaccine effect [36].

The clinical value of immune recovery remains central to interpreting these findings. In ESPRIT (Evaluation of Subcutaneous Proleukin in a Randomized International Trial) and SILCAAT (Subcutaneous Recombinant Human Interleukin-2 in HIV-Infected Patients with Low CD4 Counts under Active Antiretroviral Therapy), interleukin-2 increased CD4 counts without reducing opportunistic disease or death [37].

4. Combination Immunotherapy

For cure-directed treatment, combining interventions could address the barriers left unresolved by individual approaches, improving antigen expression, recognition, killing or tissue access while restricting viral escape (Figure 1; Table 1). TITAN (Combining a TLR9 Agonist with Broadly Neutralizing Antibodies for Reservoir Reduction and Immunological Control of HIV Infection) tested this rationale in a four-arm randomized study of lefitolimod, a Toll-like receptor 9 agonist, and 3BNC117 plus 10-1074. Antibodies prolonged control, but lefitolimod added no clear benefit [38]. Additional immune stimulation may therefore be useful only when it addresses a limiting mechanism.

Peluso and colleagues tested a broader regimen combining conserved-element vaccination, interleukin-12-supported priming, lefitolimod and dual antibodies in ten participants [39]. Seven showed partial or complete post-intervention control, including six who first rebounded. Activated CD8 responses near rebound correlated with a lower subsequent set point, consistent with a contribution from recall immunity after initial viral restraint. Without a randomized comparator, the contribution of each component and the influence of favorable baseline biology remain uncertain.

Other regimens seek to restore the function of existing effector cells through checkpoint modulation, which has produced control signals alongside concerns about toxicity. In a randomized phase 1b study of budigalimab, an antibody against programmed cell death protein 1 (PD-1), six of eleven recipients in the four-dose 10-mg cohort met an exploratory composite involving delayed rebound, lower peak or subsequent control; two maintained control through study end [40]. A larger 2026 phase II conference report did not establish a significant advantage for its week-24 control endpoint and reported treatment-related serious adverse events [41,42]. Persistent immune-related toxicity is especially consequential in otherwise stable people receiving effective ART, whose risk–benefit balance differs from that of patients treated for other diseases, including cancer.

Effector activity can also be redirected toward infected cells using engineered cellular products or bispecific molecules. In the VRC01-derived chimeric antigen receptor T-cell (CAR-T) study, all six participants undergoing ATI rebounded despite reservoir-associated changes [43]. A first-in-human Gag-directed T-cell receptor bispecific study enrolled people with compatible human leukocyte antigen (HLA)-A*02:01 and peptide recognition; it demonstrated pharmacodynamic activity without reservoir depletion at the studied doses [44]. Both approaches still depend on antigen accessibility, tissue penetration and functional effectors.

Strategies to improve effector activity and reservoir access are being explored in preclinical studies. A scaffold combining interleukin-15 superagonism with neutralization of transforming growth factor-β enhanced HIV-specific CAR-T activity in human-cell experiments [45]. Venetoclax reduced intact reservoirs during continued ART in macaques infected with barcoded simian immunodeficiency virus (SIV) strain SIVmac239M [46]. Studies in infant simian–human immunodeficiency virus (SHIV)-SF162P3 infection combined antibodies, ART and CCR5 blockade, whereas neonatal SIVmac251 experiments depleted CCR5-expressing cells transiently [47,48]. Together, these experiments suggest that intervening during reservoir establishment can amplify the effect of immune and entry-directed treatment; whether similar gains are achievable in chronic human infection remains uncertain.

In humans, prolonged remission after allogeneic transplantation shows how profound alteration of the reservoir and host environment can prevent rebound. Patients undergoing hematopoietic stem cell transplantation, including the Berlin patient, the Geneva patient who received donor cells with wild-type CCR5, the recipient of CCR5Δ32 heterozygous donor cells, and the Oslo patient, provide insights into the varying contributions of reservoir depletion, immune system replacement, and resistance to viral entry in achieving sustained HIV remission or cure [49,50,51,52]. Successful-case analyses and the prospective IciStem cohort support this multifactorial interpretation [53,54]. However, transplant toxicity restricts the procedure to patients with an independent indication, leaving the challenge of reproducing useful mechanisms through less hazardous interventions.

5. Immune Phenotypes and Treatment Response

Immune profiling offers a way to investigate the host biology underlying variable treatment responses, revealing substantial heterogeneity within conventional clinical groups. Ramirez and colleagues identified five immunological clusters, including separate controller-enriched and nonresponder-enriched groups [55]. A 164-person reservoir study found five profiles driven mainly by monocyte HIV-DNA patterns; a separate 154-person CD4 transcriptomic study identified three groups with different inflammatory signatures [56,57]. In a small proteomic study, two inflammatory biotypes crossed treatment and controller categories [58]. Together, these studies relate reservoir heterogeneity to inflammatory and cellular variation (Table 2).

Single-cell studies connect these broader profiles to infected cells with particular survival and immune-evasion properties [12,13]. The emerging therapeutic question is whether such features identify a vulnerability—for example, insufficient antigen expression or resistance to cytotoxic killing—that a selected intervention can overcome.

Tissue and developmental context add further variation. Paired samples from ART-treated women show differences in transcription and persistence across blood, gut, genital tract and liver [59]. Young adults with perinatally acquired HIV have immune and proviral features shaped by developmental exposure and prolonged treatment [60]. These findings make tissue history, age and sex relevant to the interpretation of circulating markers and to the populations in which a proposed treatment match is tested.

Repeated sampling can show how an immune profile changes with viral exposure and treatment. Receptor measurements also require care when the therapeutic antibody itself alters staining [61].

6. Clinical Validation of Personalization

Translating these profiles into treatment choices requires linking the observed biology to a mechanism that can be targeted. Viral susceptibility, CCR5 tropism and HLA–peptide compatibility already provide a basis for matching products to patients. Extending it to immune phenotypes will require evidence that a profile identifies greater benefit from a particular intervention. A small reservoir, for example, may predict slower rebound across several regimens without favoring one treatment over another.

HIV pharmacogenetics provides a precedent. PREDICT-1 (Prospective Randomized Evaluation of DNA Screening in a Clinical Trial) randomized HLA-B*57:01 screening before abacavir and eliminated immunologically confirmed hypersensitivity in the screening arm, compared with 2.7% in controls [62]. CYP2B6-guided efavirenz dosing has also undergone prospective randomized pilot testing, although the small sample limited clinical inference [63].

Immunogenetics already influences eligibility for HLA-restricted interventions such as the Gag-directed bispecific [44]. Host repertoire diversity also influences vaccine responses. In adults without HIV, International AIDS Vaccine Initiative (IAVI) G001 used the eOD-GT8 60mer immunogen to prime B cells with the potential to develop into VRC01-class bNAbs [64]. Follow-up analysis showed that variation in the immunoglobulin heavy-chain variable gene IGHV1-2 explained apparent differences between vaccine-dose groups better than dose itself, largely through differences in the baseline frequency of compatible B-cell precursors. The only vaccine recipient without a detectable VRC01-class response lacked the required IGHV1-2*02 or *04 alleles [65]. These findings suggest that immunoglobulin genotyping and precursor measurements could inform immunogen selection and the interpretation of dose responses across populations. However, therapeutic application of such information remains less developed. Comparable randomized evidence that host immunogenetic or immunophenotypic profiling improves HIV immunotherapy selection is still lacking.

Randomizing treatments within prespecified biomarker groups would reveal whether their effects vary with the profile. Comparing guided treatment with conventional selection would then test the clinical value of using that information. Reproducible assays and adequately powered comparisons are essential, particularly when a favorable profile already predicts a good outcome.

7. Mathematical Modelling and Longitudinal Treatment Decisions

Longitudinal profiling can extend this approach by tracking declining antibody concentrations, expanding resistant lineages and evolving immune responses during treatment. ART timing also changes the relationship between reservoir measurements and rebound [66]. These dynamics make a single pretreatment profile an incomplete basis for personalization [67]. Mathematical models can integrate repeated measurements and could inform treatment choices as exposure and susceptibility change (Figure 2; Table 3).

Stochastic models connect successful reservoir reactivation with the distribution of rebound times. Hill and colleagues examined how a small residual reservoir produces variable waiting times and subsequently used continuing aviraemia to update estimates of residual infection [68,69]. Deterministic virus–immune models describe conditions under which replication-competent virus persists under immune control [70]. The models connect the timing and course of rebound to reservoir reactivation, viral expansion and immune containment (Figure 3).

Models incorporating cell-associated RNA and treatment history have estimated individual rebound distributions; evolutionary models have compared antibody combinations using baseline diversity, escape opportunities and resistance costs [71,72]. Clinical translation remains limited by unmeasured tissue reservoirs and by different parameter combinations that reproduce similar viral trajectories. Validation in independent cohorts is needed before these predictions can reliably guide care.

Cassidy and colleagues modelled PGT121-sensitive and resistant viral populations using clinical data [73]. Baseline resistance and the relative fitness of competing variants helped explain rebound. In two recipients with prolonged suppression, potent neutralization and latent-cell dynamics could reproduce the observations without assuming eradication. Simulated dose escalation had participant-dependent effects, suggesting that dose, epitope coverage and alternative mechanisms merit comparison. Susceptibility, sequence and exposure measurements helped constrain explanations that viral-load trajectories alone could not resolve.

Pharmacokinetics creates an immediate application for such work. Unequal decay can leave one antibody effectively acting alone against the individual's virus. Joint models of component-specific exposure and resistance could compare products or dosing intervals more meaningfully than total antibody concentration alone. Cellular interventions add expansion, functional persistence and tissue access to this problem. Immune endotypes might help explain variation in these processes, once their relationships have been established independently.

ATI studies allow models to be tested against the onset of rebound and subsequent containment within protocol-defined monitoring and restart rules [7,9]. Early kinetics may predict later control [74]. Whether these predictions can improve intervention timing or selection requires prospective evaluation within closely monitored trials.

8. Discussion

Across these studies, the clearest clinical gains have come from antibody-mediated antiviral activity and delayed rebound. RIO demonstrated substantial benefit, while TITAN found no further improvement from adding an innate agonist [27,38]. Future combinations will need to address the mechanisms limiting control in the individual, whether resistance, inadequate antigen recognition, impaired killing or poor tissue access.

Virological suppression can coexist with immune activation and vascular abnormalities in elite controllers, while the interleukin-2 trials demonstrate that improved CD4 counts can occur without fewer serious clinical events [21-23,37]. The clinical value of remission will depend on whether sustained control also restores immune health and reduces morbidity [24].

Reservoir and immune profiling can help explain this dissociation. Rebound potential depends on proviral integrity, inducibility, integration context, clonal expansion and tissue distribution, as well as the ability of immune effectors to reach and kill infected cells [12-14]. Published clusters link this biology to measurable immune variation [55-58]. They offer candidate treatment targets, although a reproducible profile that selects the most effective immunotherapy has yet to emerge.

The decisive evidence will come from randomized comparisons that test whether biological differences modify treatment benefit. Prespecified treatment–biomarker interactions can identify promising matches, and trials of guided selection can establish whether acting on them improves care. HIV pharmacogenetics shows that this progression is feasible [62].

Personalization is also a longitudinal problem. Antibody exposure, viral escape, reservoir composition and effector competence change during intervention and ATI. Modeling offers ways to integrate serial observations and compare dosing, combinations, sequencing or monitoring [69-73]. Serial data could therefore refine dosing and intervention choice as the biological conditions change. Testing these decisions prospectively would give modelling a direct clinical role alongside the interpretation of rebound [7,9].

The practical comparator remains effective, accessible ART. Persistent autoimmunity, conditioning toxicity or intensive monitoring may outweigh reduced treatment exposure for some people; others may place substantial value on less frequent suppressive treatment. Quality of life, preferences, cardiovascular and other morbidity, and equitable access belong alongside virological efficacy in assessing benefit. Women, children, older adults and people treated during chronic infection require appropriate validation, since developmental history, tissue compartment and viral subtype can alter both the mechanism and the clinical value of an intervention.

9. Conclusions

HIV immunotherapy is increasingly guided by the compatibility of the virus, reservoir and immune response with a particular intervention. Antibodies have established antiviral activity and prolonged control after ART withdrawal, while durable intervention-free remission remains investigational outside exceptional settings. Rational combinations and immune profiling may extend these gains when they address the mechanisms limiting control in the individual.

Personalization will become clinically useful when a reproducible biological profile identifies the treatment offering the greatest benefit. Randomized biomarker studies and trials of guided selection can establish that value. Longitudinal modelling may refine decisions as exposure, resistance and immunity evolve. The goal is durable virological control accompanied by immune health, acceptable toxicity, better quality of life and accessible care.