CROI 2019 Abstract eBook

Abstract eBook

Poster Abstracts

Conclusion: The distribution of genetically intact proviruses within T cell subsets in blood do not necessarily reflect their distribution in lymph nodes. In lymphoid tissues, the frequency of intact proviruses is highest in less differentiated cells such as NV cells, while in blood the frequency is highest in more differentiated EM cells. Tissue-based NV T cells may act as progenitors of the total reservoir during ART, whereas in the periphery this reservoir is maintained within the EM T cell population, perhaps by clonal proliferation. 348 PERSISTENT HIV LOW-LEVEL VIREMIA CAN ARISE FROM AN ACTIVE PROVIRAL CLONE Xin Zhang 1 , Radwa Sharaf 1 , Behzad Etemad 1 , Ce Gao 2 , Kevin Einkauf 2 , Sigal Yawetz 1 , Guinevere Q. Lee 2 , Jackson W. Lau 1 , Zixin Hu 1 , Colline Wong 1 , Alexandra Rosenthal 1 , Daniel R. Kuritzkes 1 , Athe Tsibris 1 , Mathias Lichterfeld 2 , Jonathan Z. Li 1 1 Brigham and Women’s Hospital, Boston, MA, USA, 2 Ragon Institute of MGH, MIT and Harvard, Cambridge, MA, USA Background: Persistent low-level viremia (LLV) is not uncommon among patients with HIV despite receiving continuous antiretroviral therapy (ART), but the mechanism behind this finding remains unclear. We describe one individual with persistent low-level viremia (200-700 copies/ml) across 16 viral load measurements over >3 years despite ART intensification to a DTG, DRV/r, TAF/ FTC ART regimen. We hypothesized that the persistent LLV arose either from an expanded clone of transcriptionally-active reservoir cells or from ongoing viral replication. Methods: Commercial ARV drug levels and resistance genotyping were performed at multiple time points. We performed plasma single-genome sequencing for the Pro-RT region at 3 different timepoints, each 1 year apart. Confirmatory near-full length plasma sequences were obtained at the first time point. We also performed a novel next-generation single-genome proviral sequencing (NG-SGS) assay from PBMCs that combines near-full length proviral amplification and integration site analysis. Results: The LLV persisted despite detectable plasma ARV levels and the presence of at least 2 fully active ARVs by resistance genotyping. Across all 3 timepoints, 86% of all single-genome plasma sequences were comprised of one viral clone (range 67% - 100% at each time point). Intact near-full length proviruses exactly matching the majority plasma clone were identified, which constituted only 6% of all intact proviruses. Near-full length plasma HIV sequences confirmed the clonality of this population and the lack of known drug resistance mutations. Integration site analysis showed that this provirus is integrated into CD200R1, a gene encoding a transmembrane receptor expressed by CD4+ cells. Interestingly, the majority of intact proviruses consisted of one clonal proviral sequence, constituting 54% of all intact proviruses but only 9% of plasma variants. This intact provirus is integrated into the STAG2 gene, which has critical roles in regulating the chromosome structure and cell division. No evidence of viral evolution or emergence of new drug resistance mutations were detected in plasma over time. Conclusion: Persistent LLV can arise from the integration of HIV into a transcriptionally-active region of a clonally-expanded CD4+ population without evidence of ongoing viral replication. In this setting, further intensification of the ART regimen is unlikely to be effective and suppression of the LLV will require targeting of this transcriptionally-active reservoir. 349 PREDICTING HIV REBOUND IN VIVO BASED ON EX VIVO CD4+ T CELL LATENCY DISRUPTION Jason M. Hataye 1 , Joseph P. Casazza 1 , Alan S. Perelson 2 , Richard A. Koup 1 1 National Institute of Allergy and Infectious Diseases, Bethesda, MD, USA, 2 Los Alamos National Laboratory, Los Alamos, NM, USA Background: Following activation of ex vivo latently-infected CD4+ T cells from HIV-infected individuals on antiretroviral therapy (ART), we reported cell- to-cell variability for HIV release, and given such virus release, the probability of establishing exponential viral growth. From these results, we fit a population dynamic model that indicated a critical initial viral population size with a rare latently-infected cell lineage breaking through this threshold and establishing exponential viral growth, attributed to stochastic emergence of superspreading. Methods: With minimal modification, we extended our model to in vivo viral rebound following ART cessation. We assumed a latently-infected CD4+ T cell population half-life of 44 months and an initial total-body census of 1.5 million cells. To allow for potential synergy among near-simultaneous reactivations, we distributed these into 150,000 compartments of 10 cells each. We assumed

virus was randomly distributed in a volume of 15 liters. We performed Gillespie simulations, to obtain an estimate for the rate of latent cell reactivation on ART and for rebound off ART. Results: A rate of latent cell reactivation of ~ 1 x 10-6 /day resulted in simulated virus production that ranged between 0.5 and 4 HIV RNA copies /ml, consistent with in vivo ultrasensitive viral load quantitation on ART. Simulating a single HIV-infected individual for 100 days off ART, 234 viral reactivations occurred, with 26 that transitioned to exponential viral growth, 5 appearing in the first 20 days. This result was typical of 9 other simulated HIV-infected individuals. The mean time to greater than 100 HIV RNA copies /ml was 19 days (SD 3 days). The time interval between 100 HIV RNA copies /ml to 1 x 105 copies / ml was ~ 1 week. After a 1200-fold initial reservoir reduction, 5 of 10 individuals had rebound during 41 years. Conclusion: The frequency of rebound seeding reactivations predicted here within single simulated individuals is consistent with that estimated previously in vivo. Synergy between two reactivations was very rare, however, reactivations may not be independent, particularly for in vivo expanded clonal populations. The rate of simulated viral rebound, once virus was clinically detected, was faster than that documented in vivo, perhaps because parameters were estimated from ex vivo cultures that used maximally stimulated target cells. In addition, immune responses, not considered here, could decrease the rate of viral reactivation or rebound. 350LB VRC01 EPITOPE MOTIFS PREDICTED REBOUND KINETICS AFTER VRCO1/ TREATMENT INTERRUPTION Evan Cale 1 , Meera Bose 2 , Hongjun Bai 2 , Michael Messina 1 , Donn J. Colby 3 , Nittaya Phanuphak 3 , Nelson L. Michael 2 , Merlin L. Robb 2 , Nicole Doria- Rose 1 , Trevor A. Crowell 2 , John R. Mascola 1 , Jintanat Ananworanich 2 , Sodsai Tovanabutra 2 , Morgane Rolland 2 , for the RV397 Study Group 1 Vaccine Research Center, NIAID, Bethesda, MD, USA, 2 US Military HIV Research Program, Silver Spring, MD, USA, 3 SEARCH, Bangkok, Thailand Background: Ongoing antibody-mediated prevention clinical trials are testing infusions of the broadly neutralizing antibody VRC01 as a strategy for HIV prevention, yet the impact of VRC01 on founder viruses is unknown. We evaluated the impact of VRC01 on homogeneous HIV populations characteristic of acute infection through an Analytic Treatment Interruption (ATI) study (RV397) that enrolled 18 acutely-treated participants who had been on antiretroviral therapy (ART) for over 2 years, randomizing 13 to receive VRC01 and 5 placebo. All participants rebounded between 9 and 296 days after ATI. We investigated the impact of HIV genetic features and VRC01 neutralization sensitivity on rebound kinetics. Methods: HIV sequencing was performed via endpoint-dilution on plasma samples in Fiebig I-III acute infection (10 genomes) and post-rebound (~15 pol and env). Env were tested for VRC01 neutralization sensitivity using the TZM-bl neutralization assay. Results: After ATI and concurrent VRC01 infusion, viral rebound was modestly delayed in the VRC01 group (median: 29 vs 14 days, p=0.051). Post-rebound, pol and env sequences differed by 1-2 nucleotides from the founder sequence derived pre-ART and all sequences were intermingled in phylogenetic trees demonstrating no evidence of VRC01-mediated escape during the ATI. For each participant, VRC01 neutralization sensitivity did not differ between acute infection and post-rebound (p=0.875). However, viral strains differed in their sensitivity to VRC01 neutralization, with two infections with VRC01-resistant viruses. The most sensitive strains tended to rebound slower than less sensitive strains (Rho=-0.62, p=0.033). We developed an epitope similarity score that weighted sites based on their importance in the VRC01/Env interaction and compared our sequences to known VRC01-susceptible strains. Our predictor was associated with the time to rebound (Rho=-0.70, p=0.007) and with neutralization results (Rho= 0.59, p=0.045). Sequences from participants who rebounded early were enriched for D at site 279, compared to N in late rebounders, placebo and known VRC01-sensitive strains. Conclusion: Clearing the latent reservoir to induce drug-free viral control remains a challenge, even if our findings showed that the presence of VRC01 for a few weeks, in the absence of standing HIV variation, did not select for escape. Our ability to predict how the responsiveness to VRC01 infusion varies across viral sequences could be useful to interpret results of future trials.

Poster Abstracts

CROI 2019 127