CROI 2018 Abstract eBook

Abstract eBook

Oral Abstracts

of the HIV epidemic in Swiss MSM. To achieve a further decrease in infection rates, phylogenetic clusters could be used to identify social networks, in which one should intensify HIV-1 testing.

are some critical unexplored issues that need to be addressed to quantify the real impact of geographically targeted interventions on the epidemic. The spatial connectivity of the transmission network of an entire community has never been studied before, and the contribution of geographical clusters of HIV infections, or ‘hot-spots’ on the spread of the infection in the entire population is virtually unknown. To address these issues, a spatially explicit transmission network and the transmission intensity from an HIV hot-spot were analyzed. Methods: We examined a sample of 18,294 individuals located in a hyper- endemic rural community in South Africa, fromwhich 5,624 (4,279 females and 1,345 males) tested positive for HIV. We identified a geographical cluster with high numbers of HIV infections (HIV ‘hot-spot’) using spatial statistical analysis. Additionally, we genetically sequenced and geo-located 1,222 HIV-positive individuals, identified phylogenetic transmission clusters, and estimated the number of transmission links (individuals grouped in these transmission clusters) that arose from the HIV hot-spot. Results: From the 351 transmission links identified, 254 links (72.4%) included at least one individual located within the HIV hot-spot (Figure 1). The average distance between individuals genetically linked was 6.4 km. Results frommicrosimulation models indicated that the observed HIV transmission link configuration does not follow a random pattern, and the probability of transmission link formation is negatively affected by the distance between individuals and the HIV hot-spot. Conclusion: To our knowledge, this is the first time a geographical transmission network of an entire community was studied. We observed intense transmission dynamics between the HIV hot-spot and the rest of the community located outside this high HIV burden area. These results suggest that geographical hot-spots could have a similar role as behavioral core groups in transmission networks of concentrated epidemics. Targeting these geographical core groups, would not only impact HIV incidence within the hot-spot, but could also disrupt the transmission network of the entire community.

Oral Abstracts

42 PHYLOGENETIC PATTERNS OF HIV TRANSMISSION AMONG TRANSGENDER WOMEN IN LOS ANGELES Manon Ragonnet-Cronin 1 , Yunyin W. Hu 2 , Sheldon Morris 1 , Zhijuan Sheng 2 , Kathleen Poortinga 2 , Joel O. Wertheim 1 1 University of California San Diego, San Diego, CA, USA, 2 Los Angeles County Department of Health Services, Los Angeles, CA, USA Background: Transgender women (TGW) are among the groups at highest risk for HIV infection, with a prevalence of 22% in the US. Despite this high risk, TGW have documented a high rate of undiagnosed HIV infection. We propose that this disparity can be addressed by characterizing TGW in a molecular transmission network to inform a targeted public health response. Methods: Since 2005, Los Angeles County (LAC) has collected HIV pol sequences from drug resistance testing along with demographics and transmission risk factor data (including transgender status). We reconstructed a molecular transmission network using HIV-TRACE (pairwise genetic distance threshold of 0.015 substitutions/site) from the earliest pol sequences of 22,398 unique individuals in LAC, including 412 (2%) self-identified TGW. We examined the epidemiological predictors of clustering (a proxy for transmission risk) in the network using multivariate logistic regression (diagnosis age, race/ ethnicity, transmission risk factor, gender, and country of birth were included as covariates) and calculated assortativity-the tendency for nodes to link to other nodes with the same attributes-for each transmission risk group. Results: We found 1,722 molecular transmissions clusters, and 36% of individuals (8,133/22,398) clustered in the network. TGWwho indicated a sexual risk factor were the most likely to cluster in the network: 147/345 (42.6%) linked to at least one other person (AOR 2.20, p<0.001, reference group: cis-men reporting IDU). MSM also had high clustering odds (AOR 2.05, p<0.001); TGW who reported IDU did not have significantly elevated odds of clustering. Both MSM and TGF were highly assortative in the network (0.17 and 0.08, respectively; p<0.001), indicating that MSM and TGW tended to cluster with members of their own risk groups. TGWwere distributed across 126 clusters, and the presence of one TGW in a cluster increased the odds of there being another TGW in the same cluster 9-fold. Conclusion: TGW in LAC were more likely to cluster than other risk groups, suggesting high transmission rates-despite low representation of TGW in the database. TGW tended to be part of the same clusters, indicating shared risk activities (i.e. linked directly or through shared partners). This assortativity demonstrates the potential to use molecular epidemiology to both identify transmission clusters likely to include undiagnosed or undisclosed HIV-infected TGW and improve the targeting of public health prevention and treatment activities to TGW. 43 ASSESSING THE ROLE OF GEOGRAPHICAL HIV HOT-SPOTS IN THE SPREAD OF THE EPIDEMIC Diego F. Cuadros 1 , Tiago Graf 2 , Tulio de Oliveira 2 , Till Bärnighausen 3 , Frank Tanser 2 1 University of Cincinnati, Cincinnati, OH, USA, 2 University of KwaZulu-Natal, Durban, South Africa, 3 Harvard University, Cambridge, MA, USA Background: In the last few years, a radical shift in thinking about geographical targeted interventions has prompted several international stakeholders (UNAIDS, PEPFAR, GFATM) to include geographical prioritization as a key component of their overall HIV intervention strategy. However, there

44 USING SUCCESSIVE SURVEYS TO CALCULATE INCIDENCE AMONG AT-RISK POPULATIONS IN THE US Janet C. Burnett , Qian An, Dita Broz, Cyprian Wejnert CDC, Atlanta, GA, USA Background: Although high prevalence of HIV has been documented, very few incidence estimates exist among persons at high risk for HIV infection in the United States. Given the limitations in estimating HIV incidence, various methods should be triangulated to provide a well-supported estimate in a population of interest. National HIV Behavioral Surveillance (NHBS) conducts cross-sectional surveys every 3 years among men who have sex with men (MSM), persons who inject drugs (PWID) and sexually active heterosexuals of low socioeconomic status at increased risk for HIV infection (HET) for 20 cities with high prevalence of HIV. We propose using successive cross-sectional surveys to estimate a pooled HIV incidence for each population through a nested cohort comprised of participants with at least one repeat observation across three surveillance cycles (MSM: 2008-2014; PWID: 2009-2015; HET: 2010-2016). Methods: For each cycle, venue-based sampling (MSM) and respondent- driven sampling (PWID and HET) were used to recruit participants ≥18 years for interview and HIV testing. Repeat participants were identified using a combination of date of birth, race/ethnicity, city and gender, if applicable.

18

CROI 2018