CROI 2019 Abstract eBook

Abstract eBook

Poster Abstracts

871 PHYLOGENETIC ANALYSIS OF HIV FROM PWID IN EASTERN EUROPE AND ASIA: HPTN074 Mariya V. Sivay 1 , Philip J. Palumbo 1 , Yinfeng Zhang 1 , Kate Grabowski 1 , Estelle Piwowar-Manning 1 , Xu Guo 2 , Erica Hamilton 3 , Vivian Go 4 , Kostyantyn Dumchev 5 , Zubairi Djoerban 6 , Brett Hanscom 7 , Irving Hoffman 4 , William C. Miller 8 , Susan H. Eshleman 1 , for the HPTN 074 Study Team 1 Johns Hopkins University School of Medicine, Baltimore, MD, USA, 2 Statistical Center for HIV/AIDS Research and Prevention, Seattle, WA, USA, 3 FHI 360, Durham, NC, USA, 4 University of North Carolina at Chapel Hill, Chapel Hill, NC, USA, 5 Ukrainian Institute on Public Health Policy, Kyiv, Ukraine, 6 University of Indonesia, Jakarta, Indonesia, 7 Fred Hutchinson Cancer Research Center, Seattle, WA, USA, 8 The Ohio State University, Columbus, OH, USA Background: HPTN 074 is evaluating the efficacy and feasibility of using an integrated intervention (support for HIV treatment and substance use treatment) to reduce HIV transmission among people who inject drugs (PWID) in Indonesia, Vietnam, and Ukraine. We used phylogenetic methods to evaluate the relationship between HIV strains in the HPTN 074 study population. Methods: HPTN 074 enrolled HIV-infected PWID (index participants) with up to five concurrent HIV-uninfected injection partners per index. Index-partner pairs were randomized 1:3 to the integrated intervention vs. standard of care study arms. HIV genotyping (pol region) was performed using the ViroSeq- 1 HIV Genotyping System for 502 index participants and 7 partners who seroconverted during the study. Phylogenetic analysis was performed using RAxML v8.10.2. Individuals were considered to form a transmission cluster if the genetic distance of their sequences was ≤0.015 with ≥90% branch support. HIV strains from participants in pol transmission clusters were also analyzed using next-generation sequencing (NGS, env region). Participants were considered to have linked infections if their env sequences formed a distinct monophyletic cluster with ≥90% bootstrap support. Results: Pol sequences were obtained for 452/509 HIV-infected participants (445 indexes, 7 seroconverters). Median pairwise genetic distances for sequences from each study site were 2.9%-3.3%; there was no evidence of large discrete subclades at any of the three sites. Linkage results from pol and NGS-env analyses were concordant for all 7 seroconverter cases. The index and partner were linked in 2/7 cases; in addition, two unrelated partners were linked. In addition, 15 index-index clusters were identified that included 36 indexes with 2-7 indexes per cluster (13 pairs, one triplet, one cluster of 7 indexes). Individuals in each cluster were from the same study site; the cluster of 7 indexes was from Vietnam. NGS-env data was available for 14/15 index-index clusters; this confirmed linkage for 11 (78.6%) of the 14 index-index clusters. Conclusion: Analyses of HIV pol and NGS-env sequences revealed linked infections involving index-partner pairs, a partner-partner pair, and index-index groups that included 2-7 individuals. These findings suggest that there are complex patterns of HIV transmission among PWID in these communities, which should be considered when designing interventions for HIV prevention. 872 COMBINING OBSERVATIONAL AND PUBLIC HEALTH PHYLOGENETIC DATA TO IMPROVE HIV PREVENTION Ethan Morgan 1 , Stephanie Schuette 2 , Nanette Benbow 1 , Richard D’Aquila 1 , Brian Mustanski 1 1 Northwestern University, Chicago, IL, USA, 2 Chicago Department of Public Health, Chicago, IL, USA Background: Phylogenetic analyses can be used to highlight differences between the genetic sequences of persons infected with HIV. The findings from these analyses, however, are often limited in their interpretation and generalizability due to inherent limitations of collecting data in either clinical research or public health settings. The goal of this analysis was to combine these complementary data sources and explore how they may be used to inform HIV transmission network dynamics in Chicago. Methods: Data came from the RADAR cohort study in Chicago as well as the Chicago Department of Public Health surveillance data. Participant plasma samples were collected and pairwise genetic distances of HIV pol sequences obtained from the protease and reverse transcriptase region. Viral genetic sequences ≤1.5% genetically distant defined putative molecular clusters comprising ≥2 persons. Information regarding demographic and HIV risk behavior data were included from each data source while detailed information regarding participant’s sexual networks were obtained from the RADAR study. Results: Among 2977 HIV viral sequences available, 72 clusters contained ≥2 sequences while 23 contained ≥5 sequences (Figure 1). Overall, 424

individuals were identified as members of molecular clusters with 717 total inferred ties between individuals. The race/ethnicity composition of partners differed significantly from that of RADAR participants in both the molecular clusters and sexual networks (p<0.001). In molecular networks, black and white participants’ partners consisted primarily of other black (83%) and white partners (75%), respectively, while for Hispanics they were more heterogenous and consisted primarily of Hispanic (39%) and black (38%) partners. In network analyses, significant assortativity was observed with regards to both age (r = -0.20; p-value = 0.026) and year of HIV diagnosis (r = -0.88, p-value <0.001). Additionally, participants who always used condoms in the past six months, compared to those who had condomless sex at least once, had a significantly greater proportion of black partners (73% vs 67%) and a lower proportion of white partners (2% vs 12%). Conclusion: These results highlight the utility of combining observational cohort and public health data in a multi-tiered approach towards HIV prevention. Further studies would strongly benefit from the combination of these data which have the potential to yield novel targets aimed at reducing HIV incidence.

Poster Abstracts

873 MOBILITY PATTERNS CREATE DYNAMIC WIDELY DISPERSED RISK NETWORKS IN NAMIBIA Eugenio Valdano 1 , Justin Okano 1 , Vittoria Colizza 2 , Sally Blower 1 1 University of California Los Angeles, Los Angeles, CA, USA, 2 INSERM, Paris, France Background: Namibia has a severe HIV epidemic: prevalence is ~14%. In Namibia, as in other countries in sub-Saharan Africa, the population is highly mobile and the epidemic is generalized. Consequently, individuals may be at risk of acquiring HIV both in their home community (from other residents or from visitors from other communities) or in other communities (from residents of these communities). We use data frommobile phones and the 2013 Namibia Demographic and Health Survey (NDHS) to estimate the risks of HIV acquisition for individuals in their home community and when they travel. Methods: The mobile phone data were collected between October 2010 and September 2011; the data set is based on 9 billion communications from 1.19 million unique SIM cards. These data were used to calculate a mobility network based on the average proportion of time residents spent in each of the 96 constituencies in Namibia that include cell towers. We coupled this network with demographic data from the 2011 census and HIV-testing data from 7,600 participants of the NDHS. We used the coupled epidemic-mobility network to calculate, for each constituency, the risk of acquiring HIV that was due to sexual contact with other residents (i.e., localized risk), with visitors, or when traveling. Results: The calculated mobility network is shown in Figure 1. We find significant geographic variation in prevalence: 6%-40% for women, 0%-24%

CROI 2019 339