CROI 2016 Abstract eBook

Abstract Listing

Poster Abstracts

950 No Selection of X4 Viruses by Maraviroc in Cell Reservoirs in R5X4 HIV Infections Stéphanie Raymond ; Romain Carcenac; Florence Nicot; Nicolas Jeanne; Michelle Cazabat; Mary Requena; Lise Cuzin; Pierre Delobel; Jacques Izopet CHU Toulouse, Toulouse, France Background: The CCR5 antagonist maraviroc, used in combination with antiretroviral drugs in HIV-infected patients, is only active against CCR5-using viruses. However, some patients were given maraviroc although they were infected by R5X4 dual-mixed viruses. This was to assess the immunological benefit of maraviroc therapy. In the MARIMUNO study, patients with undetectable plasma HIV RNA received a 24-week maraviroc supplement to an efficient antiretroviral therapy. Since the positive selection of CXCR4-using viruses in cell reservoirs may influence any response to later treatment, we investigated how the frequency of CXCR4-using variants in R5X4 dual-mixed virus populations responded to maraviroc selection pressure using ultra-deep sequencing (UDS). Methods: We explored 22 patients from the MARIMUNO study infected with R5X4 dual-mixed viruses according to the recombinant virus assay Toulouse Tropism Test. The frequency of CXCR4-using variants was determined in peripheral blood mononuclear cells (PBMCs) before maraviroc intensification (week 0) and after 24 weeks of maraviroc (week 24); both samples were tested in the same run. UDS was performed on a 454 GS Junior system. The sequence reads of the V3 env regions were analysed with PyroVir software (Inserm-Transfert) developed to provide a fast and automated position-specific process for inferring HIV-1 tropism from V3 env 454 ultra-deep pyrosequencing data. Results: The mean total HIV-1 DNA before maraviroc intensification was 2.4 log copies/10 6 cells; it was 2.5 log copies/10 6 cells 24 weeks later (Wilcoxon rank test, P=0.3). UDS with the PyroVir genotypic algorithm detected CXCR4-using viruses in the 22 R5X4 infected patients at week 0 with a mean frequency of 59% (range: 3-100%). CXCR4-using viruses were detected in 21/22 patients at week 24 with a mean frequency of 52% (range: 10-92%). We found no correlation between the HIV DNA concentration in PBMCs and the number of CXCR4-using variants or their frequency. The frequency of CXCR4-using variants did not increase between weeks 0 and 24 except in patient 4 whose increase was 32%. The mean numbers of unique amino acid variants and X4 amino acid variants were similar at both stages. Conclusions: A 24-week course of a CCR5 antagonist does not select CXCR4-using viruses in the PBMCs of patients on suppressive therapy infected with R5X4 dual-mixed viruses. These results indicate little or no residual HIV replication that could be subjected to selection pressure. 951 Failure Rate in NRTI-Free Treatment Is Higher in 2 Compared to 3 Class Regimens Alexandra Scherrer 1 ; Jürg Böni 2 ; SabineYerly 3 ;Thomas Klimkait 4 ;Vincent Aubert 5 ; Matthias Cavassini 6 ; Manuel Battegay 7 ; Christoph Hauser 8 ; Enos Bernasconi 9 ; Huldrych F. Günthard 10 ; for the Swiss HIV Cohort Study 1 Univ Hosp Zürich, Zürich, Switzerland; 2 Inst of Med Virology, Swiss Natl Cntr for Retroviruses, Zurich, Switzerland; 3 Geneva Univ Hosp, Geneva, Switzerland; 4 Univ of Basel, Basel, Switzerland; 5 Univ Hosp Lausanne, Lausanne, Switzerland; 6 Univ Hosp Lausanne, Univ of Lausanne, Lausanne, Switzerland; 7 Univ Hosp Basel, Basel, Switzerland; 8 Bern Univ Hosp and Univ of Bern, Bern, Switzerland; 9 Regional Hosp Lugano, Lugano, Switzerland; 10 Univ Hosp Zurich, Zurich, Switzerland Background: Little knowledge exists about the optimal drug combination in nuclesoside/tide (NRTI)-free antiretroviral treatment (ART). We studied the effectiveness of NRTI- free ART containing two compared to three drug classes. Methods: We studied time to virological failure (VF) in 503 NRTI-free treatment episodes from 342 patients participating in the Swiss HIV Cohort Study (ART start after 01/01/2007). VF was defined as two consecutive viral loads (VL) >50 copies/mL or 1 VL>50 copies/mL followed by a treatment change or interruption after 180 days of continuous treatment or previous viral suppression. Cox proportional hazards models with robust standard errors to account for intra-patient correlation were performed. The following variables were considered as potential confounders: VL before start of NRTI-free ART, CD4 cell count, gender, risk group, age, previously reported non-adherence, previously experienced VF, year of first ART initiation, genotypic sensitivity score of NRTI-free ART (Stanford algorithm version 7.0) and presence of major drug resistance mutations (IAS-USA list 2014). In the multivariable model, all variables with a p-value <0.2 were included. Results: A comparable number of NRTI-free episodes included two and three drug classes, 252 (50.1%) and 251 (49.9%), respectively. Most patients were ART-experienced (99%) and in 65% of patients VL was <50 copies/mL when starting the NRTI-free episode. Overall, 7% of patients experienced a treatment failure. Patients treated with two drug classes had a higher risk for ART failure, the multivariable hazard ratio (mHR) was 3.0 (95% CI: 1.4-6.6, p=0.007). Another important risk factor for VF was the VL level before NRTI-free ART initiation. The mHR was 3.2 (95% CI: 1.1-9.1, p=0.032) and 6.7 (95% CI: 2.6-17.7, p<0.001) for VL 50-10,000 and >10,000 copies/mL, respectively (reference <50 copies/mL). In addition, we found that male participants were more likely to fail NRTI-free ART (mHR: 3.0, 95% CI: 1.1-8.0). No other factor was significantly associated with VF. Conclusions: NRTI-free ART with three drug classes was more effective compared to two drug classes. Thus, NRTI-free ART with two classes should be prescribed very cautiously, if at all, especially for patients with a detectable VL before starting NRTI-free treatment. 952 The Impact of Medication Adherence on Virologic Failure in A5202 Robert A. Parker 1 ; Dustin J. Rabideau 1 ; Paul E. Sax 2 ; CamlinTierney 3 ; Eric Daar 4 ; Ann Collier 5 ; Elena Losina 2 ; Kenneth A. Freedberg 1 1 Massachusetts General Hosp, Boston, MA, USA; 2 Brigham and Women’s Hosp, Harvard Med Sch, Boston, MA, USA; 3 Harvard Sch of PH, Boston, MA, USA; 4 Harbor Univ of California Los Angeles Med Cntr, Torrance, CA, USA; 5 Univ of Washington, Seattle, WA, USA Background: Poor adherence is widely recognized to decrease the efficacy of antiretroviral therapy (ART). This effect may vary depending on regimen. To quantify the impact of self-reported adherence on virologic failure (VF) in ART-naïve participants, we evaluated data from ACTG A5202. We also assessed the impact of adherence on the previously reported increased rate of VF in the abacavir-lamivudine (ABC-3TC) group compared to tenofovir–emtricitabine (TDF-FTC) in participants with screening HIV RNA ≥100,000 copies/ mL. Methods: We limited our analysis to the 1,798 (of 1,857 eligible; 97%) participants with adherence data. There were 11,303 self-reports, assessed at follow-up weeks 8, 24, and every 24 weeks thereafter, of when ART was most recently missed using a 6-point scale ranging from “never” to “within the last week”. We used adherence as a time-varying covariate in a Cox proportional hazards model of time to VF in the entire cohort to estimate the hazard ratio (HR) with reduced adherence. We did additional analyses incorporating the randomized NRTI assignment as a fixed covariate in the 761 participants (of 798; 95%) with high screening viral load (HI-VL) and adherence data. Results: Overall, 15% had study-defined VF. Adherence was a strong predictor of time to VF (Figure). With the “never skipped medications” and missed “more than 3 months ago” (rarely) reports pooled post hoc as the reference group, the HR increased monotonically from 1.6 (95%CI: 1.04-2.5) for the “1-3 months ago” group to 3.4 (95% CI: 2.4-4.8) for the “within the last week” group. To illustrate this impact, at 96-weeks 20% (95% CI: 12-29%) of participants who consistently missed pills “1 to 2 weeks ago” would be expected to have failed, compared to 9% (95%: 7-11%) of participants who rarely or never missed pills. In the analysis of participants with HI-VL, those randomized to ABC-3TC had a significantly increased HR of VF compared to TDF-FTC (1.70; 95% CI: 1.19-2.41, P=0.003). Adjusting for self-reported adherence in the period prior to VF partially accounts for this effect (HR: 1.56; 95% CI: 1.09-2.23, P=0.014). Conclusions: Recent self-reported reduced adherence more than tripled the risk of VF in A5202. Although lower adherence partially accounted for the increased risk of VF in participants starting initial treatment with a HI-VL receiving ABC-3TC compared to TDF-FTC, significantly elevated risk of VF remained. Continued efforts to improve adherence will increase rates of virologic suppression with these regimens.

Poster Abstracts

403

CROI 2016