CROI 2017 Abstract e-Book

Abstract eBook

Poster and Themed Discussion Abstracts

a 2% threshold was used to assess minority drug resistant variants defined as representing <20% of the viral population. TDR was assessed using the WHO 2009 list of mutations. Cox regression was used to estimate hazard ratios. Adherence was measured by visual analogue scale Results: 921 (854/1064 CIn (81%); 67/81 (83%) RIn) of 1,145 adults with sequences initiated ART. 838 (783/854 CIn (92%); 55/67 RIn (82%)) had follow up VL and contributed to the analysis. Median age was 35 years (y) (IQR 28, 47) in CIn and 27 y (23, 38) in RIn; 71% and 84%were female, respectively; 97%were on fixed dose combination of tenofovir/ emtricitabine/efavirenz. In RIn, the prevalence of any TDR was 14.6%(95%CI 8.3-24.2) in majority virus and 23.6%(95% CI 15.2-34.9) in minority virus. In CIn, PDR prevalence was 8.8%(7.2-10.7) and 18.7%(16.2-21.3) in majority and minority virus, respectively. The K103N mutation was the most common in majority and minority viruses. Cumulative suppression at 12 months (m) was 97%. Median time to VS was 2.96m (IQR 2.76-3.94). After adjusting for sex, age, baseline VL and adherence, there was no evidence that PDR was associated with VL suppression (adjusted (a)HR 0.96, 95%CI 0.75-1.23 and aHR=1.12, 95%CI 0.89-1.42, for majority virus and minority virus, respectively, vs no mutations). High baseline VL (>100,000 vs <10,000) was associated with a decreased rate of VS (aHR 0.75; 0.62-0.91) and good adherence (≥95% vs <95%) was associated with an increased rate of VS (aHR 1.36; 1.11-1.66) Conclusion: The prevalence of TDR is above the WHO threshold of 5% in this rural South African setting. With local guidelines recommending test and treat, strengthening early warning TDR indicators is required. However, our data suggest the clinical impact of this may be limited, if adherence to current 1st-line ART is good 492 ASSOCIATION BETWEEN HIV DRUG RESISTANCE AND BASELINE FAILURE IN THE ACTG PEARLS STUDY Rami Kantor 1 , Saran Vardhanabhuti 2 , Srikanth Tripathy 3 , Mariza Morgado 4 , Shanmugam Saravanan 5 , Carole Wallis 6 , Sarah E. Hudelson 7 , Susan H. Eshleman 7 , Thomas Campbell 8 1 Brown Univ, Providence, RI, USA, 2 Harvard Univ, Cambridge, MA, USA, 3 Natl Inst for Rsr in TB, Chennai, India, 4 Oswaldo Cruz Inst, Rio de Janeiro, Brazil, 5 YRG CARE, Chennai, Tamilnadu, 6 Lancet Labs and BARC-SA, Johannesburg, South Africa, 7 The Johns Hopkins Univ, Baltimore, MD, USA, 9 Univ of Colorado Denver, Aurora, CO, USA Background: Pre-therapy genotyping is recommended in developed countries but supporting data in global settings are lacking. We examined the association of pre-therapy baseline (BL) DR mutations (DRMs) with resistance at virologic failure (VF) in a multi-national, prospective, randomized trial, in 9 countries on 4 continents. Methods: 1571 drug-naïve HIV-infected adults were randomly assigned to 3 Arms: efavirenz+lamivudine+zidovudine (Arm A; World Health Organization (WHO) alternative option), atazanavir+didanosine+emtricitabine (Arm B), or efavirenz+emtricitabine+tenofovir (Arm C; WHO preferred option). VF was defined as confirmed viral load (VL) >1000 copies/mL at/after 14 weeks. Reverse transcriptase and protease genotyping was performed with the ViroSeq HIV Genotyping System. DR was assessed using the 2015 International Antiviral Society-USA list. Clinically significant DR was defined as intermediate/high resistance to ≥1 drug based on the Stanford Database. Associations of DR at VF were estimated by Chi-Square or Wilcoxon tests for baseline dichotomous or continuous variables respectively. Results: BL and at-failure genotypes were available for 238/270 participants with VF (74 Arm A, 95 Arm B, 69 Arm C); 44%women; median age 32 years; median BL-at failure CD4 162-240 cells/µL and VL 5.2-4.2 log10 copies/mL; median time to VF 32 weeks; subtypes: 56% C, 39% B, 5% other. At least one DRM was found at VF in 70% of participants across Arms; 47% NRTI, most common M184V (40%), 51% NNRTI, most common K103N (31%); 7% PI in Arm B, most common I50L (6%); 38% single, 29% dual, 3% triple-class. Of those with BL DR, 98% (42/43) had any DR at VF and 63% had DRMs at failure that were not detected at BL. Of those without BL DR, 63% (122/195) had any DR at VF. DR at VF was associated with low BL CD4, greater CD4 change from BL, high BL VL, low at-failure VL and BL DR (p≤.002 for all). Of those with any DR at VF 63% had clinically significant DR to study regimens and 15% had DR to future options; and was associated with males, low BL CD4, greater CD4 change from BL, low BL VL, low at failure VL and BL DR (p≤.003 for all). Conclusion: In this multi-national clinical trial with 60% non-subtype B HIV-1 infection, BL DR was associated not only with VF as previously reported, but also with clinically- significant DR at VF that impacted treatment options. 493 Background: Treatment guidelines recommend baseline genotype resistance testing in new HIV diagnoses to guide selection of 1st-line therapy. While integrase strand transfer inhibitors (INSTIs) are recommended for initial treatment, most genotype tests do not assess INSTI resistance (INSTI-R). Given the low rate of transmitted INSTI-R, it is unclear if baseline testing would provide value. We examine the conditions under which INSTI-R testing might be cost-effective. Methods: We use a decision tree to examine the incremental cost-effectiveness ratio (ICER) of INSTI-R testing vs. no testing. Results are based on differences in 96-wk quality- adjusted life expectancy (QALE), assuming equivalent outcomes thereafter. In the base case, the prevalence of transmitted INSTI-R is 0.1%, derived from published studies. Those who undergo INSTI-R testing and have resistance detected at baseline start a DRV/r-based regimen. The FLAMINGO trial provides estimates of 96-wk HIV RNA suppression rates with DTG (80.1%) and DRV/r (67.8%)-based regimens. For those with INSTI-R but without an INSTI-R test, we presume a 30% probability of success on DTG, since many INSTI-R isolates remain susceptible to DTG. In the base case, we presume a quality of life (QoL) reduction for those experiencing virologic failure (~3%). Costs include: INSTI-R $250/test; DTG-based ART ($38,150/yr) and DRV/r-based ART ($44,400/yr). In sensitivity analyses, we examine how changes in these parameters influence the conclusions. Results: Compared to a no test strategy, the INSTI-R test decreased QALE by 1.23x10^-6 QALY and increased per person costs by $250 at 96-wk; hence, not performing the test was the preferred strategy. Reducing QoL for DRV/r treatment further strengthened the preference for no test, even at implausibly high levels of transmitted INSTI-R. In sensitivity analyses, reducing DTG suppression with undiagnosed INSTI-R to <20% and increasing PI-regimen efficacy to >80% yielded improved clinical outcomes, but the ICER of INSTI-R testing was not cost effective, well exceeding $1 million/QALY. Base-case results were similar when an NNRTI was substituted for DRV/r. Conclusion: At an estimated INSTI-R prevalence of 0.1% and a current INSTI-R test cost of $250, baseline INSTI-R testing results in worse clinical outcomes and higher costs than no test. A no test strategy remains preferred if there is a possibility of INSTI-suppression despite detected INSTI-R, or if there is a QoL benefit to PI avoidance, regardless of the prevalence of INSTI-R. 494 COST-EFFECTIVENESS ANALYSIS OF PRE-ART HIV DRUG RESISTANCE TESTING IN KENYA Horacio A. Duarte 1 , Joseph Babigumira 2 , David Stauffer 3 , Robert Shafer 3 , Ingrid Beck 4 , Louis Garrison 2 , Michael H Chung 2 , Lisa Frenkel 1 , Eran Bendavid 3 1 Univ of Washington/Seattle Children’s Rsr Inst, Seattle, WA, USA, 2 Univ of Washington, Seattle, WA, USA, 3 Stanford Univ, Stanford, CA, USA, 4 Seattle Children’s Rsr Inst, Seattle, WA, USA Background: The prevalence of pre-treatment HIV drug resistance (PDR) is increasing in East Africa, which may decrease the effectiveness of antiretroviral therapy (ART) programs. Testing for PDR with an oligonucleotide ligation assay (OLA), a low-cost point mutation assay, is a potential strategy to address the challenges posed by PDR in resource- poor settings. Methods: We developed an HIV drug resistance model that simulated the emergence and transmission of resistance mutations, calibrated to the Kenyan epidemic. We implemented 6 care strategies for PDR testing: no testing (NT), OLA, or consensus sequencing (CS), each with initial viral load (VL) testing at either 6 or 12 months (NT+VL6, NT+VL12, OLA+VL6, OLA+VL12, CS+VL6, and CS+VL12). We assumed the cost of OLA and CS were $30 and $300, respectively, per test. This model was used to evaluate the health outcomes, lifetime costs, and cost-effectiveness of the strategies over a 15-year time horizon. Health outcomes included quality-adjusted life years (QALYs), new HIV infections and opportunistic infections, rates of ART suppression of HIV replication, and rates of drug resistance. Results: The OLA+VL12 strategy provided 23 additional QALYs at an additional cost of $5,297 per 1,000 population, compared to NT+VL12 (an incremental cost-effectiveness ratio (ICER) of $230/QALY gained). The current Kenyan practice (NT+VL6) resulted in 22 fewer QALYs and cost an additional $3,889 per 1,000 population, compared to OLA+VL12. Strategies that used CS for PDR testing were not cost-effective by national income standards. OLA+VL12 resulted in more patients maintaining viral suppression than NT+VL6 (73% vs. 60%, respectively). OLA+VL12 also had 7.6% fewer new HIV infections than NT+VL6. Initial PDR prevalence was 8.0% in 2013. By 2028, this prevalence increased to 34.8% with NT+VL6, but only to 25.5%with OLA+VL12. The probability of virologic failure at 12 months among patients with PDR is an important factor in the cost-effectiveness of SHOULD WE BE TESTING FOR BASELINE INTEGRASE RESISTANCE? Yiannis Koullias 1 , Paul E. Sax 1 , Rochelle P. Walensky 2 , Emily P. Hyle 2 1 Brigham and Women’s Hosp, Boston, MA, USA, 3 Massachusetts General Hosp, Boston, MA, USA

Poster and Themed Discussion Abstracts

CROI 2017 206