CROI 2015 Program and Abstracts

Abstract Listing

Poster Abstracts

245 Clustering of Swiss HIV Patients Not Enrolled in the Swiss HIV Cohort Study (SHCS) Mohaned Shilaih 1 ; Alex Marzel 1 ; Jörg Schüpbach 1 ; Jürg Böni 1 ; SabineYerly 2 ;Thomas Klimkait 4 ;Vincent Aubert 3 ; Huldrych F. Günthard 1 ; Roger Kouyos 1 The Swiss HIV Cohort Study (SHCS) 1 University Hospital Zurich, Zurich, Switzerland; 2 Geneva University Hospital, Geneva, Switzerland; 3 University Hospital Lausanne, Lausanne, Switzerland; 4 University of Basel, Basel, Switzerland; 5 University of Zurich, Zurich, Switzerland Background: One of the central challenges in HIV surveillance is that the surveyed population might not be representative of the entire HIV-infected population, especially with respect to marginalized populations. The SHCS is exceptionally representative (75% of HIV patients on ART), however the possibility remains that entire sub-epidemics might be missed by the cohort. A unique opportunity to assess the presence of such “under the radar” populations is provided by a database of all genotypic resistance tests performed in Switzerland, which includes both cohort and non-cohort patients. Methods: Phylogenetic cluster analysis was used to assess the presence of a hidden sub-epidemic. 11338 SHCS and 3099 Swiss non-SHCS sequences were pooled with 27803 background sequences from the Los Alamos database (10 best BLAST hits for each Swiss sequence). A maximum likelihood phylogenetic tree was built using FastTree. Clusters that were dominated by Swiss sequences (>=80%) were interpreted as Swiss transmission clusters. Results: Non-B subtypes were strongly overrepresented in the non-SHCS compared to the SHCS (OR 3.0, 95%CI 2.8-3.3). Moreover, non-SHCS patients were more likely to be female (OR 1.4, 95%CI 1.3-1.6). Transmission groups were assigned to non-cohort sequences based on phylogenetic proximity. This revealed that heterosexuals were more present among non-SHCS patients (OR 2.0, 95%CI 1.8-2.2; compared to MSM). Associations remained significant after adjusting for sex, test date, and subtype. We found 301 transmission clusters purely of non-SHCS patients. However, these clusters were small (median 4.5, IQR 3.25-5.75, max 9) compared to those consisting only of SHCS patients (median 7.5, IQR 4.75-10.2, max 17). Non-SHCS patients were more likely to be part of a transmission cluster compared to SHCS patients (OR 1.9, 95% CI 1.8-2.1). However, when sample date was included in the logistic regression model said clustering preference of non-SHCS markedly decreased (1.1, 95%CI 0.99-1.2). Conclusions: In this work we evaluated the coverage of the SHCS, one of the most representative HIV cohorts. We found an overrepresentation of non-B subtypes among non-SHCS patients suggesting that migrants might be underrepresented in the SHCS. We also observed transmission chains among non-SHCS patients, yet their limited size and frequency suggest that no major HIV outbreak in Switzerland is missed by the SHCS. More generally, this work shows the potential of sequence data to assess the representativeness of cohort studies. 246 HIV Transmission Network Structure Reveals Characteristics of Bridging Individuals Sanjay Mehta ; Joel O.Wertheim; Konrad Scheffler; Susan Little; Richard R. Garfein; Sergei L. Kosakovsky Pond; David M. Smith University of California San Diego, La Jolla, CA, US Background: Molecular epidemiology is often the only means to reveal difficult-to-measure patterns of HIV transmission in local epidemics. Of particular interest are “bridge” individuals that link otherwise disconnected network components, evaluated here for the inferred San Diego Primary Infection Cohort (SDPIC) HIV transmission network. Methods: We inferred a molecular transmission network (MTN) from a curated collection of 1024 partial pol sequences, representing 713 SDPIC participants sampled between 1996 and 2013, and 311 sequences from area chronically infected individuals. Two individuals (nodes) were linked if their sequences were <1.5% distant (TN93 metric). Network degree, mean path length (MPL), and betweenness centrality were computed for each network node, and statistical analyses were used to examine which sociodemographic factors associated with network properties. We also investigated the association of a novel “uniqueness” score with the various measures of network centrality. This score combined the demographic attributes of age, race, ethnicity, HIV risk factor, and location of residence into a single score, which was then used to compare all of the individuals within each transmission cluster relative to one another. Results: 42.1% of individuals in our study were linked to at least one other individual in the MTN. Age of individuals was the only sociodemographic measure marginally associated with centrality in univariate analyses [p=0.053, age vs clustering, t-test]. In clusters comprising 4 or more individuals, central nodes (low MPL) were significantly more likely [p=0.05, Fishers Exact Test] to have higher uniqueness scores, and the highest scoring individuals had significantly higher betweenness centrality [25.6% vs 8.2%, p=0.041, t-test]. Conclusions: In models of epidemiologic spread, individuals who serve as bridges between otherwise disconnected groups have been implicated as important drivers of HIV epidemics. Our analyses demonstrate in the San Diego HIV epidemic, bridging individuals were sociodemographically unique. Such uniqueness represents a higher degree of disassortative mixing by these key individuals, suggesting that disassortative partnerships may disproportionately drive HIV epidemics. 247 Phylodynamic Analysis of HIV Sub-Epidemics in Mochudi, Botswana Vladimir Novitsky 1 ; Denise Kuehnert 2 ; Sikhulile Moyo 3 ; Erik vanWidenfeldt 3 ; Lillian Okui 3 ; Max Essex 1 1 Harvard School of Public Health, Boston, MA, US; 2 ETH Zürich, Zürich, Switzerland; 3 Botswana-Harvard AIDS Institute Partnership, Gaborone, Botswana Background: The rapid intrinsic evolution of HIV-1 makes it possible to infer epidemiologic patterns from sequence data. The Bayesian birth-death skyline (BDSKY) plot was introduced recently as a model of virus transmission. We used BDSKY to estimate the effective reproductive number, R , and the timing of virus transmission, to distinguish “acute” HIV sub-epidemics ( with recent viral transmissions) from “historic” sub-epidemics ( without recent viral transmissions) in a southern-African community. Methods: Study subjects participated in enhanced household-based HIV Testing and Counselling in Mochudi, a peri-urban village in Botswana. The sampling density was around 70%. HIV-1C V1C5 sequences were generated for 1,248 residents of Mochudi. HIV-1C sub-epidemics were identified by a combination of bootstrapped maximum likelihood and internode certainty. For HIV sub-epidemics with 5+members, the epidemiological parameters were inferred from virus sequence data. The time line for each sub-epidemic was estimated by fitting the BDSKY model and inferring the tree height and internal node ages in the Maximum Clade Credibility time-trees using BEAST2. For each sub-epidemic we estimated the time interval the majority of HIV transmissions occurred in, with corresponding 95% HPD intervals. Results: We employed the BDSKY model to estimate effective reproductive number R and timing of HIV-1C transmissions within 15 sub-epidemics with 5+members. Only three of the 15 sub-epidemics were estimated as “acute” with recent HIV transmissions. The median estimates of R were 0.7–1.6. The V1C5-based informativeness of R estimates differed across sub-epidemics. The median peak duration of viral transmissions was 5.4 years (95% HPD, 3.8 to 8.6 years). The median life span of identified HIV sub-epidemics, i.e., the time from the cluster’s origin to its most recent sample, was 14.2 years (95% HPD, 9.8 to 16.4 years). The majority of viral transmissions within 15 identified HIV sub-epidemics in Mochudi occurred between 1997 and 2005. The generated data suggests that the time period during which infected people are infectious has decreased significantly since the introduction of ART in Botswana as the national program. Conclusions: Viral sequence data from a densely sampled community in Botswana allowed us to estimate the effective reproductive number, R , and timing of virus transmissions in 15 local HIV sub-epidemics. “Acute” sub-epidemics with recent HIV transmissions are likely to fuel local HIV/AIDS epidemics.

Poster Abstracts

221

CROI 2015