CROI 2020 Abstract eBook

Abstract eBook

Poster Abstracts

individuals who were subjected to STIs prior to a prolonged analytical treatment interruption (ATI). We aimed to identify virologic determinants of post-treatment control in the participants of A5068 who underwent multiple TIs. Methods: A5068 participants in the STI arms underwent two short (~4 weeks) TIs (STI 1 and 2) and a subsequent extended ATI. Both STI 1 and 2 were followed by 16 weeks of antiretroviral therapy. We compared plasma viral load (pVL) dynamics after each STI between PTCs and post-treatment non-controllers (NCs). Single-genome sequencing (SGS) of the pol region from plasma HIV RNA was performed for 6 PTCs and 7 NCs. Confirmatory long-range SGS of the pol-env region was performed for a subset of time points. Viral diversity was calculated by the average pairwise distance at one time point and viral divergence was calculated by the average pairwise distance between sequences of different time points. Results: pVLs were significantly lower during STI 2 compared to STI 1 for both PTCs (n=6) and NCs (n=27). For both the first and second STI, PTCs had significantly lower peak pVLs compared to NCs (median pVL [Q1, Q3] for PTCs vs. NCs at the first STI: 1,270 [536, 5,593] vs. 37,506 [1,643, 66,579] HIV-1 RNA copies/mL, p=0.001; and second STI: 199 [<50, 424] vs. 14,562 [7,870, 33,031] HIV-1 RNA copies/mL, p=0.001). An algorithm that used a combination of peak pVL<10,000 HIV-1 RNA copies/mL during STI 1 and peak pVL<1,000 HIV-1 RNA copies/mL during STI 2 accurately predicted that all 6 PTCs would achieve HIV control and that 26/27 NCs would not. In addition, we have generated >500 plasma HIV single-genome sequences for the PTCs and NCs during the STIs and ATI. Among all participants, higher plasma HIV diversity during STI 1 predicted higher viral diversity in ATI (Spearman r=0.67, p=0.02). Increasing viral divergence from STI 1 to ATI was associated with a higher peak pVL at ATI (Spearman r=0.69, p=0.02). Conclusion: In participants undergoing STIs, lower peak pVLs during the first two short TIs may predict post-treatment control. Emergence of divergent viral populations during the third TI may compromise the ability to achieve viral control. 320 VIRAL REBOUND KINETICS FOLLOWING SINGLE AND COMBINATION IMMUNOTHERAPY FOR HIV/SIV Melanie Prague 1 , Jeffrey Gerold 2 , Irene Balelli 1 , Chloé Pasin 1 , Jonathan Z. Li 3 , Dan Barouch 3 , James Whitney 3 , Alison L. Hill 2 1 L'Université de Bordeaux, Bordeaux, France, 2 Harvard University, Cambridge, MA, USA, 3 Harvard Medical School, Boston, MA, USA Background: HIV infection can be treated but not cured with antiretroviral therapy, motivating the development of new therapies that instead target host immune responses. Three such immunotherapies were recently tested in non-human primates – a TLR7-agonist, therapeutic vaccine (Ad26/MVA), and broadly-neutralizing antibody (PGT121) – and cured a subset of animals by preventing or controlling viral rebound after antiretrovirals were stopped. The goal of this study was to use viral dynamics modeling to infer the mechanisms of action of these therapies and predict outcomes in human trials. In addition, we examined whether they reduced the pool of latently-infected cells versus boosted antiviral immunity, and whether they acted independently or synergistically. Methods: Here we conduct a detailed analysis of the kinetics of viral rebound after immunotherapy. We introduce a newmathematical model of viral dynamics that incorporates both the stochastic and deterministic regimes of rebound and includes the action of adaptive immune responses. This model is fit to data from 99 macaques across three separate studies using a non-linear mixed-effects statistical approach. A rigorous model comparison procedure was designed to identify the effects of each intervention and quantify the impact on viral dynamics. To predict the impact of these immunotherapies in clinical trials, we calibrated the model to HIV rebound in human treatment interruption trials and simulated the effect of adding each immunotherapy. Results: We find that the vaccine reduced reactivation of latent virus by 4-fold (95% CI [2,8]), and boosted the avidity of antiviral immune responses by 17-fold when alone [5, 67] and 210-fold [30, 1400] when combined with the TLR7-agonist. In the context of later initiation of antiretroviral therapy only (9 vs 1 week after infection), the TLR7-agonist reduced latent reservoir reactivation by 8-fold [4, 16], but also slightly increased target cell availability (1.5-fold). The antibody boosted immune response avidity 8-fold [3,16] and displayed no detectable synergy with the TLR7-agonist. In humans, the TLR7-agonist alone, TLR7+vaccine, and TLR7+antibody are expected to lead to control of rebound in some patients (~5%, 55%, 90% respectively), but often after a high peak viral

load. Heterogeneity in rebound time and peak/setpoint viral loads between patients is predicted to be very high. Conclusion: Overall, our results provide a framework for understanding the relative contributions of different mechanisms of preventing viral rebound and highlight the multifaceted roles of TLR7-agonists for HIV/SIV cure. 321 FREQUENCY OF POSTTREATMENT CONTROL VARIES BY ART RESTART AND VIRAL LOAD CRITERIA Jesse Fajnzylber 1 , Radwa Sharaf 1 , Evgenia Aga 2 , Ronald Bosch 2 , Jeffrey M. Jacobson 3 , Elizabeth Connick 4 , Daniel Skiest 5 , Michael Sneller 6 , Ronald T. Mitsuyasu 7 , Keith Henry 8 , Tae-Wook Chun 6 , Ann Collier 9 , Frederick M. Hecht 10 , Jonathan Z.Li 11 , for the CHAMP Study Team 1 Brigham and Women's Hospital, Boston, MA, USA, 2 Harvard T.H. Chan School of Public Health, Boston, MA, USA, 3 Case Western Reserve University, Cleveland, OH, USA, 4 University of Arizona, Tucson, AZ, USA, 5 Baystate Health, Springfield, MA, USA, 6 NIAID, Bethesda, MD, USA, 7 University of California Los Angeles, Los Angeles, CA, USA, 8 University of Minnesota, Minneapolis, MN, USA, 9 University of Washington, Seattle, WA, USA, 10 University of California San Francisco, San Francisco, CA, USA, 11 Brigham and Women's Hospital, Boston, MA, USA Background: Clinical trials including an analytic treatment interruption (ATI) are vital to evaluating the efficacy of strategies for HIV remissions. Determining the optimal ART-restart criteria that minimizes exposure to high-level viremia and maximizes detection of post-treatment controllers (PTCs) remains challenging. We present an interactive online tool for predicting viral rebound timing in ATI trials and describe the impact of PTC definitions on PTC frequency estimates. Methods: The interactive viral rebound calculator (http://jonathanlilab.bwh. harvard.edu/rebound-calc/) was created with a pooled analysis of plasma viral loads (pVLs) of >700 participants from 10 ATI trials. The tool allows the user to set the ART restart criteria based on a single or multiweek pVL criteria and to customize results by the timing of ART initiation, ART regimen, and PTC frequency (default is the CHAMP study criteria: pVL<400 cps/mL at ≥2/3 time points for ≥24 wks post-ATI). Results: We compared the impact of several commonly used ART restart criteria (1,000 pVL x 1 wk, 1,000 pVL x 2 wks, 1,000 pVL x 4 wks, and 50,000 pVL x 4 wks) on the ability of a hypothetical ATI trial to detect PTCs. Our calculator estimates that these criteria fail to identify 30%, 10%, 0%, and 0% of PTCs, respectively, due to premature ART restart. The sensitivity and specificity of PTC detection also varied by ART restart criteria. Of the 4 criteria, the 1,000 pVL x 1 wk criteria had high specificity (99%), but low sensitivity (43%), while the 50,000 pVL x 4 wks criteria had low specificity (15%), but high sensitivity (100%). The 1,000 pVL x 4 wks criteria achieved a balance with 91% specificity and 93% sensitivity for identifying PTCs. Using high pVL thresholds (≥10,000 cps/mL) for ART restart substantially reduces the specificity of PTC identification in early-treated participants, likely related to their overall lower pVL peaks compared to chronic- treated participants. The expected frequency of PTCs varied dramatically by the PTC definitions (Figure). In almost all scenarios, PTC frequency was significantly higher in early-treated individuals. Conclusion: This calculator provides the first interactive tool for estimating viral rebound outcomes and supporting the design of ATI trials. A multi-week ART restart criteria of 1,000 pVL provides high sensitivity and specificity for PTC detection. However, the expected frequency of PTC identification in ATI trials can vary dramatically by the definition of post-treatment control.

Poster Abstracts

CROI 2020 110