Cure research at CROI 2018: defining and reducing the reservoir and the risks from interrupting treatment

Simon Collins, HIV i-Base

Although most news from CROI covers clinical research, almost a third of the 1100 studies at the conference were on basic science, many of which are linked to cure research.

This year this included several areas of cure research that are important to report.

  • Studies defining, measuring and identifying the viral reservoir.
  • Research in people who started ART in very early acute infection.
  • The risks and safety of treatment interruptions in cure research.
  • Whether current ART needs to be more active.
  • Optimistic results regarding monkeys enabled to stop ART without viral load rebounding.


A good update on progress in cure research that also highlighted studies at CROI 2018 was given by Huldrych Günthard as part of the pre-CROI workshop for young investigators, and is now online as a webcast. [1]

As an introduction, much cure research is focused on the latently infected reservoir of resting CD4 T-cells that is established in early infection. These cells predominate in lymph tissue throughout the body but the reservoir also includes other body sites and possibly other immune cells. This pool is established within three weeks of infection (by Fiebig stage 1), persists during untreated infection, and only gradually declines cover decades on ART.

Many presentations focused on defining and measuring this this reservoir, an essential research goal to judge whether interventions are having any impact. Outgrowth tests are a specialist research tool and underestimate replication-competent virus while PCR tests (generally for integrated HIV DNA in PBMCs or tissue as a surrogate but other methods are also used) also measure defective virus. More than 90% of HIV-identified particles picked up by PCR tests are sufficiently defective as to no longer be infectious. (As reference, several years ago, Eriksson et al published a nice review paper, online as open access, that compares ways to measure the reservoir. [2])

By definition, these HIV-containing resting cells are not affected ART which targets active stages of the HIV lifecycle. Although the resting cells periodically and unpredictably wake, slowly reducing the pool, they are so long-lived that it would take many decades for the body to completely clear them on ART alone. In someone with a small reservoir (ie from starting ART within weeks of infection) this would take 60 years and considerably longer for someone who started ART in chronic infection.

So other studies reported on ways to activate these resting cells. Even if HIV is not eradicated, significantly reducing the pool might allow for a shorter period during which ART alone might provide a cure.

Although many existing drugs are, and have been, studied for the potential to activate resting cells, at least in vitro, Deborah McMahon from the University of Pittsburgh presented results from a placebo controlled ACTG study showing that none of the three single infusion doses of the approved HDAC inhibitor romidepsin were successful at inducing HIV in adults on ART, despite potent results ex vivo. [3]

Also, as any intervention needs to selectively activate only cells that contain HIV, another research goal is to find a way to identify latently infected cells in vivo. Two studies at CROI 2018, showed that the initial hope that CD32+ cells might be a biomarker for latently infected cells was not supported in practice. [4, 5]

Studies in acute infection

As a smaller reservoir might be easier to cure, many researcher groups are looking to study people who started ART during acute infection.

An overview of such interventions was presented by Eugène Kroon from the Thai RV-254/SEARCH 010 study, now also webcast. This ongoing cohort includes 527 participants who were diagnosed as part of an intensive HIV testing programme and who started ART during acute HIV infection, many while still HIV antigen negative. [6]

This study has already reported results from three small single-intervention cure studies – using early ART, the HDAC inhibitor vorinostat and the bNAb VRC01, respectively. Unfortunately, after treatment was interrupted, all 40 participants had rapid viral rebound (at median 22 days, range 9 to 296). The treatment interruption protocol in these studies was to closely monitor viral load (every 3-7 days) and to restart ART after rebound is confirmed >1000 copies/mL. A fourth ongoing study is using an Ad28/MVA therapeutic vaccine.

Although none of these interventions were effective – VRC01 having the most promising results in terms of delayed viral rebound – safety results for participants included no CD4 decline, no drug resistance or acute HIV symptoms, and that all participants resuppressed HIV after restarting with the same ART regimen. However, the interruption did cause HIV seroconversions in 70% of the participants who until the study had remained HIV antibody negative on 4th generation HIV tests. [7] These participants had the negative outcome of now testing HIV antibody positive in Thailand where this outcome has social consequences related to employment and insurance. Also, despite the early restart protocol, viral load had rebounded to more than 10,000 copies/mL for some participants before ART was resumed, and two had CD4 declines of more than 20%.

In another presentation from the same cohort, Timothy Schacker from the University of Minnesota presented results on how very early ART can reduce the reservoir (and can reduce lymph node fibrosis), but that this needs to be during Fiebig stages 1 and 2. There was little difference to starting in chronic infection for people starting in Fiebig stage 3 or later (approximately 4-5 weeks post infection). [8]

Treatment interruptions in cure studies

The inclusion of treatment interruptions in research is complicated by safety concerns given that continuous ART is recommended in all guidelines. These include:

  • Reseeding or expanding the latent reservoir.
  • Clinical outcomes from increased immune activation and inflammation.
  • Impact in compartment sites, especially the brain/CSF.
  • The risk of seroconversion symptoms after a long period on ART.
  • CD4 declines that might not fully recover.

Two posters at CROI 2018 also reported reassuring data that concerns about reseeding the reservoir after a short treatment interruption might only have limited short-term risks.

Katherine Clarridge and colleagues from NIAID, reported on the impact of short treatment interruptions (maximum 16 weeks with protocol to restart with rebound >1,000 copies/mL) in ten people who started ART in chronic infection. [9]

These participants in bNAb VRC01 study had a median CD4 nadir of 345 cells/mm3 (but this ranged from 16 to 900) and had been suppressed on ART for a median of 8 years (range: 3 to 17). The poster reported that the size of the HIV reservoir and immune parameters returned within 6-12 months of restarting ART. However, despite a protocol with a low threshold to restart ART, median levels reached 30,000 copies/mL (mean 50,000; range 340–273,000) when ART was restarted. Details are available in the open access paper published earlier this year in PLoS Pathogens. [10]

Also related, a recently published paper from a small vaccine study using a 16-week treatment interruption (where viral load rebounding to >50,000 copies/mL). Although the vaccine was unsuccessful, the longer time taken to restart ART revealed higher than expected rates of spontaneous viral control without ART in four participants. That these participants were all in the placebo arm, shows the uncertainties linked to small numbers of people in this research. This level of rebound is higher than many participants, researchers and advocates would support, but also shows the lack of consensus for the time that might be needed to observe an active result. [11]

A related poster from Zachary Strongin and colleagues reported an analysis of a 12-week treatment interruption on the viral reservoir in 12 participants in ACTG studies. Although the poster included limited details about the interruption (no viral rebound or CD4 results), the results on HIV DNA showed a return to pre-interruption levels six months after restarting ART. [12]

Although these results from short treatment interruptions are encouraging, they only include small numbers of participants. Serious seroconversion symptoms have previously been reported from stopping ART, and the concern that renewed inflammation in other sites, principally CSF, was referred to in several presentations. Neurological complications are always a focus at CROI and this year an oral presentation by Serena Spudich from Yale University reported that HIV is detectable in CSF cells in up to half of participants on long-term suppressive ART. [13]

The median age of the 69 participants (97% male) was 50 years, with current and nadir CD4 counts of approximately 700 and 300 cells/mm3 respectively. Median duration on ART was 8.6 years (range 5.4-16.4).

Persistence of HIV-containing cells in CSF was not associated with higher levels of HIV DNA in blood or with levels of inflammation in blood or CSF. However, plasma viral load was associated with both cell-free HIV RNA and myeloid cell activation in CSF.

A timely oral presentation by Joseph Mankowski from Johns Hopkins included results that could inform the risk of treatment interruptions in this site. This macaque study looked at the CSF reservoir, comparing four animals euthanised on ART with six animal who first interrupted treatment for four months. [14]

As well as showing that SIV RNA rebounded in CSF during the treatment interruption, they found high SIV DNA in spinal cord even on ART, and that rebound was high in the newly reported site during the treatment interruptions. Viral rebound in CSF during the treatment interruptions came from SIV that was already in these compartments rather than being reseeded. These researchers concluded that their result showed that HIV drugs should be developed to better target this site.

One limitation of this study is that it involved a model of dual SIV infection in which one of the viruses is known to be very virulent in the brain. Macaques infected with these viruses all develop central nervous system disease and progress to AIDS within three months. The model was designed to facilitate studies of AIDS-related dementia rather than answer questions related to curing HIV, so the relevance of the results to the human situation will need to be confirmed.

Does HIV continue to replicate on ART?

The underlying question of whether the reservoir is relatively fixed (archived from when ART is started, and maintained by clonal proliferation) or whether it is continuously repopulated and evolving (reseeded by ongoing replication from body sites not reached by ART) has important implications for cure strategies. Two studies at CROI 2018, both reported in detail in an earlier HTB report, failed to find evidence of ongoing replication on ART. [15]

Mary Kearney and colleagues used phylogenetic analysis to show identical HIV retained viral sequences from plasma and lymph tissue, even after many years on ART, including using the same integration sites, including in longitudinal lymph samples. [16]

Thomas Rasmussen and colleagues showed no impact from ART intensification in a randomised placebo-controlled study from adding dolutegravir in 40 participants who had been suppressed on standard ART for more than three years. [17]

Both studies concluded that from a virological perspective, current ART doesn’t need to be more potent or effective.

Results from a large observational study were important for showing the range of individual responses. An oral presentation from Nadine Bachmann and colleagues from the Swiss HIV Cohort reported that 25% of people on long-term suppressive continuous ART (defined as >5 years without two having consecutive viral load results >200 copies/mL) did not have reductions in their reservoir size. In these participants, total HIV DNA remained either stable over time or increased. Although early viral blips were associated with a lower rate of reservoir decay, the observational nature of the study means that cause and effect can’t be shown (or even inferred). [18]

This was a large study in 1078 HIV positive participants with three or more PBMC samples from the first 5 years on ART) and 429 people with an additional late sample > 8 years (median 10.5 years).

The mean reservoir size declined very slowly and steadily (by 0.05 log/year) over five years and more slowly still over subsequent years. However, there were very wide differences in reservoir size between individuals. In multivariate analysis, having an early viral load blip over the first 18 months of ART, was associated with a higher reservoir, with earlier ART, faster time to suppression, higher CD4 count and non-B sub-type associated with a lower reservoir. Blips were significantly associated with a slower decay of the reservoir.

Successful results with bNAb PGT-121 and a TLR-7 agonist

And then dozens of other studies were presented as posters, and highlighted in six poster discussion sessions on defining the reservoirs, insights from viral rebound, the reservoir in acute infection and the impact of ART in both human and animal studies.

In addition to ART, the range of therapeutic interventions includes ‘shock and kill’ approaches with various latency inducing drugs, vaccines, bNAbs with effector functions, immune modulation and direct targeting of infected cells, many of which were reported at CROI, or more likely, combination approaches.

The most encouraging of these was included in the first press conference at CROI 2018, before the results were shown at the main conference. The results in a macaque study showed significantly delayed time to viral load rebound off ART following dual treatment with the bNAb PGT-121 and a TLR-7 agonist. The full results were presented as a late breaker oral abstract by Dan Barouch from Beth Israel Deaconess Medical Centre, Boston, and are reported in detail in an earlier HTB report. In the group receiving dual therapy, 5/11 animals maintained undetectable viral load for six months without restarting ART. [19, 20]

However, even these most promising results were presented as very preliminary stages for an HIV cure.

Thanks to Richard Jefferys for editorial comments to this article.


Please see later in this issue of HTB for an excellent article by Richard Jefferys on assessing treatment interruptions in cure research studies. [21]


Unless stated otherwise, all references are to the Programme and Abstracts of the 25th Conference on Retroviruses and Opportunistic Infections (CROI 2018), 4–7 March 2018, Boston.

  1. Günthard HK. Advances in HIV cure. Pre-CROI workshop for trainees and young investigators, 3 March 2018, Boston. (webcast)
  2. Eriksson S et al. Comparative analysis of measures of viral reservoirs in HIV-1 eradication studies. PLoS Pathogens. (14 February 2013).
  3. McMahon D et al. Single romidepsin infusions do not increase HIV expression in persons on ART (A5315). 25th CROI, 4-7 March 2018, Boston. Oral abstract 72. (abstract) (webcast)
  4. Noto L et al. CD32+PD-1+ TFH cells are the major HIV reservoir in long-term ART-treated individuals. 25th CROI, 4-7 March 2018, Boston. Oral abstract 156. (abstract) (webcasts)
  5. Bertagnolli LN et al Majority of the latent reservoir resides in CD32A negative CD4 T cells. 25th CROI, 4-7 March 2018, Boston. Oral abstract 156.’ (abstract) (webcast)
  6. Kroon E et al. Stop it now: targeting early infection events. 25th CROI, 4-7 March 2018, Boston. Oral abstract 51. (abstract) (webcast)
  7. Trichavaroj R et al. 343 HIV Serology following treatment interruption in very early treated people. 25th CROI, 4-7 March 2018, Boston. Poster abstract 343. (abstract and poster)
  8. Schacker T et al. Persistent detection of HIV RNA+ cells with art started in Fiebig 1&2 vs Fiebig 3-5. 25th CROI 2018, Boston. Oral abstract 66. (abstract) (webcast)
  9. Clarridge K et al. Impact of treatment interruption on HIV reservoirs and immunologic parameters. 25th CROI, 4-7 2018. Poster abstract 334. (abstract and poster)
  10. Clarridge JE et al. Effect of analytical treatment interruption and reinitiation of antiretroviral therapy on HIV reservoirs and immunologic parameters in infected individuals. PLoS Pathogens, January 2018.
  11. Sneller et al. A randomized controlled safety/efficacy trial of therapeutic vaccination in HIV-infected individuals who initiated antiretroviral therapy early in infection. Sci Transl Med. 2017 Dec 6;9(419). pii: eaan8848. doi: 10.1126/scitranslmed.aan8848.
  12. Strongin Z et al. Short-term art interruption has little effect on levels of integrated proviral DNA. Poster abstract 335.
  13. Spudich S et al. HIV-1 persists in CSF cells in half of individuals on long-term ART. 25th CROI 2018, Boston. Oral abstract 119. (abstract) (webcast)
  14. McCarron et al. SIV rebound in the spinal cord after stopping ART: a novel CNS reservoir. 25th CROI 2018, Boston. Oral abstract 124. (abstract) (webcast)
  15. Collins S. No HIV evolution in plasma or lymph nodes on suppressive ART and no impact from further intensification. HTB i-Base report.
  16. McManus WR et al. No evidence for ongoing HIV replication in lymph nodes during suppressive ART. 25th CROI, 4-7 March 2018, Boston. Oral abstract 70. (abstract) (webcast)
  17. Rasmussen T et al. No residual virus replication in a randomised trial of dolutegravir intensification
. 25th CROI, 4-7 2018. Oral abstract 71. (abstract) (webcast)
  18. Bachmann N et al. Determinants of HIV-1 reservoir size and long-term dynamics under suppressive ART. 25th CROI, 4-7 2018. Late breaking oral abstract 69LB. (abstract) (webcast)
  19. Collins S. Inching towards an HIV cure: bNAb and TLR-7 agonist reduce viral rebound off-ART in macaques. HTB (9 March 2018).
  20. Borducchi E et al. PGT121 combined with GS-9620 delays viral rebound in SHIV-infected rhesus monkeys. 25th CROI, 4-7 2018. Oral late-breaker abstract 73LB. (abstract) (webcast)
  21. Jefferys R. Assessing antiretroviral therapy interruptions in HIV cure research. HTB 2018. (30 March 2018).

Links to other websites are current at date of posting but not maintained.