A single case report from Belgium has traced an unusual virological failure of long-acting cabotegravir/rilpivirine therapy back to a resistance mutation that had been silently archived in the patient’s immune cells for nearly three and a half years. The patient, a 41-year-old man with HIV-1 subtype B, met every guideline-based eligibility criterion for the injectable two-drug regimen: years of durable viral suppression, no history of virological failure, a normal body mass index, and injections administered on schedule by certified providers. Yet in January 2024, confirmed virological failure set in, and retrospective sequencing revealed why.
The report, published as a case study in the journal Discover Viruses, describes how next-generation sequencing of proviral DNA from peripheral blood mononuclear cells detected the non-nucleoside reverse transcriptase inhibitor resistance-associated mutation 188L at a frequency of 14.1 percent, roughly 41 months before the patient began long-acting cabotegravir/rilpivirine. That mutation had been absent from baseline plasma genotyping performed years earlier, when the patient was first diagnosed. The finding suggests that resistant variants were either transmitted at the time of infection and later archived, or selected during a brief period of NNRTI-based treatment, and then remained hidden in the viral reservoir until the regimen switch gave them a selective advantage.
The patient’s treatment history, reconstructed in detail by the team at the University Hospital of Liège, began in 2013 with a diagnosis at a plasma HIV-1 RNA of 130,000 copies/mL. Baseline Sanger genotyping found no resistance-associated mutations in the protease and reverse transcriptase regions. He started abacavir/lamivudine plus ritonavir-boosted darunavir in November 2013 and achieved full suppression. In July 2016, therapy was simplified to rilpivirine plus dolutegravir for 14 months, during which two consecutive low-level blips of 74 and 68 copies/mL occurred at a two-week interval. He then moved to abacavir/lamivudine/dolutegravir and, from August 2020, to dolutegravir/lamivudine dual therapy, maintaining viral loads below 20 copies/mL throughout.
In October 2022, he initiated an oral lead-in of cabotegravir/rilpivirine and transitioned to intramuscular injections every eight weeks. A single blip of 66 copies/mL in February 2023 resolved spontaneously. Then, in January 2024, virological failure was confirmed with three consecutive HIV-1 RNA measurements of 728, 30,100 and 89,400 copies/mL. Sanger genotyping of the first two samples detected reverse transcriptase mutations 179E and 188L, along with integrase mutations 140S and 148R. The injectable regimen was discontinued and the patient was switched to darunavir/cobicistat/emtricitabine/tenofovir alafenamide, which produced a rapid virological decline.
A crucial element of the investigation was drug-level monitoring. Retrospective liquid chromatography-tandem mass spectrometry, performed at Lausanne University Hospital on stored EDTA plasma collected at time points ranging from hours to several weeks after injections and near the failure event, showed that both cabotegravir and rilpivirine concentrations remained above the suggested efficacy thresholds for trough plasma levels. Those thresholds correspond to four times the protein-adjusted concentrations required for 90 percent inhibition of viral replication, namely 664 ng/mL for cabotegravir and 50 ng/mL for rilpivirine. Suboptimal drug exposure, a factor repeatedly implicated in failures of this regimen, was therefore ruled out in this case.
Instead, the explanation lay in the proviral reservoir. Resistance-associated mutations were present at three of the four reservoir time points. Notably, 188L was already present in PBMCs 41 months before the switch to long-acting therapy, while the matched rectal GALT sample from the same visit lacked 188L but carried the integrase mutation 140R, which was absent from PBMCs at that timepoint. Mutational profiles thus diverged across compartments, though the authors caution that this could reflect true compartmentalization or simply differences in technical sensitivity and sampling depth.
Most proviruses in the reservoir are defective and often clonally expanded, so resistance calls supported mainly by reads harboring frameshifts, stop codons or APOBEC-related hypermutation signatures within the sequenced regions can mislead interpretation. The standard pipeline reported more mutations; for example, reverse transcriptase mutations 138K, 184I and 230I flagged in one GALT and one PBMC sample were eliminated by proviral filtering. Critically, 188L persisted after filtering in PBMCs, strengthening confidence in its clinical relevance.
To probe the mutation’s origin, the researchers attempted to retrieve the original 2013 baseline plasma sample for deep sequencing, but it could not be located in the biobank. They instead sequenced the earliest retrievable plasma sample, collected three months after diagnosis in 2014, when the viral load was 328 copies/mL. No integrase mutations were detected, as expected. Reverse transcriptase coverage fell below the threshold for automated variant calling, although the protease and integrase amplicons of the same sample yielded sufficient coverage, indicating a limitation specific to the RT amplicon rather than a global sample failure. Manual inspection of roughly 30 reads at codon 188 identified the mutant-compatible codon encoding 188L alongside wild-type reads, and 179E was similarly identified by manual review. Because of the low coverage, both findings were recorded strictly as qualitative, exploratory signals, and no reliable frequency could be derived. The authors also considered index hopping as a source of the low-level signal, but the only co-pooled sample in that sequencing run carried no resistance mutations, and the specimens harboring 188L were processed in a separate run.
Two scenarios could explain the mutation’s presence. The patient may have acquired a virus already carrying NNRTI resistance mutations at infection; in the absence of selective pressure, a fitter, drug-sensitive variant could have outcompeted it, explaining the negative baseline Sanger test and the rapid response to NNRTI-free regimens, while the resistant strain persisted long enough to become archived. Alternatively, the resistant variant may have emerged during the brief window of low-level viremia at the end of the 14-month dolutegravir/rilpivirine regimen and been archived before suppression was restored. The authors note that de novo selection of a high-impact mutation such as Y188L within such a short, low-level viremic window would be atypical, which they consider more consistent with the transmitted or early-archived virus hypothesis, though neither scenario can be formally excluded and the early plasma signal is supportive rather than conclusive.
The case carries broader implications for the expanding use of long-acting antiretroviral therapy. The authors argue that proviral next-generation sequencing should be considered before switching to long-acting cabotegravir/rilpivirine in patients with a history of NNRTI exposure, even when baseline plasma genotyping shows no resistance. In this case, had PBMC genotyping been performed in 2020, the archived mutation would likely have been detected and might have prompted a more cautious switch or closer virological monitoring, and earlier identification of elevated risk could have enabled a more proactive approach to partner protection. The findings also raise organizational questions, including whether routine biobanking of PBMCs or buffy coat should be integrated into care to create a durable genomic record, and they acknowledge real limitations: no validated frequency thresholds exist for clinically relevant minority variants in proviral DNA, short-read sequencing cannot establish genome-wide proviral intactness, estimated defective-read fractions varied widely across samples, and repeated testing across timepoints, while ideal, is constrained by cost. Proviral genotyping, the authors conclude, is neither a magic solution nor an academic curiosity, but a clinically powerful tool when optimized, contextualized and interpreted with caution, and it should guide personalized care rather than serve as a basis for excluding patients from long-acting therapy.
Subject of Research: A case report using longitudinal proviral DNA sequencing to identify archived HIV-1 resistance as the mechanism of virological failure on long-acting cabotegravir/rilpivirine therapy.
Article Title: Longitudinal profiling of blood-tissue HIV-1 reservoirs reveals the mechanism of failure to long-acting cabotegravir/rilpivirine: a case report
Article References: El Moussaoui, K., Aguilar Ortmans, D., Van Laethem, K., Ruelle, J., Mortier, V., Bontems, S., Durkin, K., Briki, M., Thoueille, P., Mzougui, S., Hayette, M.-P., Meuris, C., & Darcis, G. (2026). Longitudinal profiling of blood-tissue HIV-1 reservoirs reveals the mechanism of failure to long-acting cabotegravir/rilpivirine: a case report. Discover Viruses, 3(1), Article 23. https://doi.org/10.1007/s44370-026-00056-x
Image Credits: AI Generated
DOI: 10.1007/s44370-026-00056-x
Keywords: HIV-1, long-acting cabotegravir/rilpivirine, virological failure, proviral DNA, drug resistance, NNRTI resistance, Y188L, next-generation sequencing, viral reservoir, case report, antiretroviral therapy, gut-associated lymphoid tissue
News Source: Ophelia Keating. (October 28, 2026). Archived HIV Resistance Mutation Surfaced Years Later to Undermine Long-Acting Therapy. Scienmag.
