CROI 2018 Abstract eBook

Abstract eBook

Poster Abstracts

Results: FTFs and subtype D were independently associated with lower genotypic susceptibility scores ( P <10 –12 and P =0.02), which measures resistance to individual ART regimens. A/D recombinants were significantly less frequent in FTFs (odds ratio=0.53, P = 7.8×10 –7 ). No associations were observed for the other subtypes, implying that ART is uniformly effective in this setting. Power analysis estimated that our database conferred 80% power to detect an association between FTF and a subtype with an odds ratio of about 1.2. The number of inferred recombination breakpoints was negatively associated with FTFs, both within A/D recombinants (binomial generalized linear model, P =4.7×10 –3 ) and all recombinants ( P =2.9×10 –5 ). Conclusion: This is the largest database of FTFs in Uganda to date, with a predicted statistical power to detect odds ratios as low as 1.2. Our findings support the hypothesis that variation in rates of disease progression among subtypes does not have a major impact on response to ART, and suggest that inter-subtype recombination may on average represent a burden on viral fitness.

16 (18.6%) harbored no major resistance mutations, 11 (12.8%) harbored NNRTI resistance only, 1 (1.1%) harbored NRTI resistance only and 58 (67.4%) harbored both NRTI and NNRTI resistance. PI resistance was not observed. The most prevalent mutations were M184V (59.3%), K103N (31.4%),Y181C/V (26.7%) and G190A/S (N=24.4%). Conclusion: Dried-blood spot-based genotyping offers an alternative to plasma HIV RNA genotyping where the latter is not feasible. While nearly 70% of pediatric treatment failures in this Southern Ethiopian cohort occur with major NRTI and NNRTI resistance mutations, 30% occur with no resistance, or resistance to one drug class only. Enhanced access to resistance testing would greatly facilitate clinical-decision making capacity in this priority setting.

541 HIGH LEVELS OF HIV DRUG RESISTANCE AMONG ADULTS FAILING SECOND-LINE ART IN NAMIBIA Michael R. Jordan 1 , Ndapewa Hamunime 2 , Souleymane Sawadogo 3 , Simon Agolory 3 , Andreas Shiningawamwe 4 , Taffa Negussie 3 , Christa L. Fischer Walker 3 , Elliot G. Raizes 3 , Nicholus Mutenda 2 , Christian Hunter 5 , Natalie Dean 6 , Kim Steegen 7 , Vibha Kana 7 , Chunfu Yang 3 , Steven Y. Hong 1 1 Tufts University, Boston, MA, USA, 2 Ministry of Health and Social Services, Windhoek, Namibia, 3 CDC, Atlanta, GA, USA, 4 Namibia Institute of Pathology, Windhoek, Namibia, 5 University of Namibia, Windhoek, Namibia, 6 University of Florida, Gainesville, FL, USA, 7 University of the Witwatersrand, Johannesburg, South Africa Background: Global HIV drug resistance (HIVDR) surveillance efforts have largely focused on descriptions of HIVDR prior to treatment initiation or at time of first-line antiretroviral therapy (ART) failure. Data regarding the prevalence of HIVDR in people failing second-line ritonavir-boosted protease inhibitor (PI)-based ART in sub-Saharan Africa are limited and no data are available from Namibia. As of the end of 2015, Namibia had 136,324 people on first-line non- nucleoside reverse transcriptase inhibitor (NNRTI) based ART and 3,884 people on PI-based second-line regimens in its public health sector. We investigated the prevalence of HIVDR at second-line failure in Namibia. Methods: Between August 2016 and February 2017, HIV-infected people ≥15 years who failed PI-based second-line ART (2 consecutive HIV RNA tests >1000 copies/ml) were identified at 15 ART clinics in Namibia. The population on second-line ART at the 15 participating ART sites represented >70% of the total population on second-line in the country. At time of confirmed virological failure, dried blood spot specimens were collected for genotypic testing for drug resistance. HIV-1 Pol sequences were obtained using RT-PCR and Sanger sequencing. Sequences with low-, intermediate- or high-level HIVDR by the Stanford HIVdb were classified as resistant. Results: A total of 238 individuals were enrolled; 56.7%were female. The median age and duration on PI-based ART at time of failure were 37 years (IQR = 21 to 46) and 3.4 years (IQR 1.82 to 5.1), respectively. 237 individuals received lopinavir/ritonavir-based regimens; one received an atazanavir/ritonavir- based regimen. 160 (67.2%) individuals had sequencing data available. HIVDR prevalence data are presented in Table 1. No difference in the prevalence of HIVDR to any drug class by gender was observed. Conclusion: In this first study of HIVDR amongst patients failing PI/r-based ART in Namibia, significant levels of resistance are observed, suggesting prolonged virological failure and moderate to high levels of adherence. Nonetheless, up to 87%may achieve virological suppression with intensive strategies to improve adherence on current PI/r regimens. Drug resistance testing is needed to reduce unnecessary switches to more costly regimens

Poster Abstracts

540 FIRST-LINE HIV TREATMENT FAILURE IN NON-B SUBTYPES AND RECOMBINANTS IN UGANDA

Art Poon 1 , Emmanuel Ndashimye 1 , Mariano Avino 1 , Richard Gibson 1 , Fred Kyeyune 2 , Immaculate Nankya 2 , Miguel E. Quinones-Nateu 3 , Eric Arts 1 1 Western University, London, ON, Canada, 2 Joint Clinical Research Centre, Kampala, Uganda, 3 Case Western Reserve University, Cleveland, OH, USA Background: The enormous genetic diversity of HIV-1 subtypes has been a persistent concern for antiretroviral treatment (ART). HIV-1 infections in Uganda are predominated by subtypes A and D. While there is evidence for associations between HIV-1 subtypes and pathogenesis, finding associations with ART failure has been limited by sample population sizes in resource-limited settings. To maximize statistical power, we have combined HIV genotype and clinical data associated with first-line treatment failures across multiple cohorts and clinical sites in Uganda (Table 1). Methods: First-line treatment failures (FTFs, n =1,724) were defined by HIV genotype records with sample collection dates after the ART start dates in the database. We used a rules-based record linkage algorithm to associate genotype and clinical data (including plasma viral load, CD4 cell count, and first-line ART regimen). These data were combined with ART-naive records in the combined Uganda database ( n =968) and location-time matched records in the Stanford HIVdb database ( n =1,462) for a total of 4,134 records. Genotypic resistance predictions were generated with the Stanford HIVdb algorithm (version 8.1.1). HIV subtype predictions and recombination detection was performed with HyPhy SCUEAL on a local computing cluster.

CROI 2018 198