CROI 2017 Abstract e-Book

Abstract eBook

Poster and Themed Discussion Abstracts

workers (FSW) compared to 16.3% in LP clusters, p=0.0064. More respondents in HP clusters had greater HIV acquisition risk perception or were already known to be HIV-infected than in LP clusters (p<0001). Conclusion: HIV infection in Kenya exhibits localized geographic clustering that is dependent on socio-demographic and behavioral factors revealing disproportionate exposure to higher HIV-risk. Identification of these clusters reveals the right places for targeting priority-tailored HIV interventions. 848 A FRAMEWORK FOR PREDICTING PHYLOGENETIC CLUSTERS OF HIV AT HIGH RISK FOR GROWTH David McVea 1 , Richard Liang 2 , Jeffrey Joy 2 , P. Richard Harrigan 2 , Art Poon 3 1 Univ of British Columbia, Vancouver, Canada, 2 BC Cntr for Excellence in HIV/AIDS, Vancouver, British Columbia, Canada, 3 Western Univ, London, Ontario, Canada Background: Phylogenetic clustering of HIV from infected individuals facilitates the rapid detection of outbreaks and may augment traditional surveillance. However, we lack a quantitative understanding of cluster behaviour. Here, we examine the growth and characteristics of 224 phylogenetic clusters of >5 individuals from British Columbia (BC). In particular, we sought characteristics that could identify clusters likely to grow rapidly in the short- and long-term. Methods: The BC phylogenetic research program uses anonymized genotypes from the BC drug treatment program annotated with demographic, risk-factor, and treatment data. Phylogenetic clusters are assembled from groups of >5 individuals based on short distances between sequences in a phylogenetic tree. We divided clusters into large (≥ 20 members) and small (< 20 members). We also divided clusters into currently growing rapidly (≥ 5 newmembers in the past year) and slowly (≤ 4 newmembers in the past year). We then compared cluster demographic and risk-factor data to identify which ones varied most between large- and small-, and slowly- and rapidly-, growing clusters. Results: Both cluster size at one year and the maximum-ever growth rate increased with larger cluster size. A threshold of 5 newmembers within the first year, and 0.6 new members per month, were most efficient at identifying eventual large clusters. Using these thresholds, clusters could be classified into large or small with sensitivity of 1 and specificity of 0.76 based on maximum ever growth rate, and with sensitivity of 0.43 and specificity of 0.95 based on size at one year of age. Regarding current growth, the ratio of MSM in each cluster increased with current growth rate, and a threshold ratio of 0.5 MSM/Non-MSMwas the most efficient at differentiating rapidly- from slowly-growing clusters. Using this threshold, clusters could be identified as rapidly- or slowly-growing with sensitivity of 1 and specificity of 0.54. Conclusion: As the bulk of the HIV epidemic slows in developed countries, further control will rely on targeted interventions, such as pre-exposure prophylaxis in limited sub-populations at elevated risk of HIV infection. Simple characteristics of phylogenetic clusters may predict short- and long-term growth, identifying sub-groups most at risk of infection. While such rules will need to be validated and refined in each unique context, they offer an opportunity to target interventions to maximize their impact and cost- effectiveness. 849 NETWORK VIRAL LOAD: A CRITICAL METRIC FOR HIV ELIMINATION Britt Skaathun 1 , Aditya Khanna 1 , Ethan Morgan 1 , Samuel R. Friedman 2 , John A. Schneider 1 1 Univ of Chicago, Chicago, IL, USA, 2 Natl Development & Rsr Insts, New York, NY, USA Background: Previous associations have been observed between an aggregate viral load measure, the community viral load (CVL) and new HIV diagnoses. The CVL, however, is prone to ecological fallacy due to the presumption that transmission occurs between individuals in the same community. We develop a new and more precise metric, the Network Viral Load (NVL) to measure the composite viral load within a risk network of an HIV negative individual. Methods: We examined the relationship between NVL and HIV infection among a population-based sample of Young Black Men who have Sex with Men (YBMSM) in Chicago. Networks were generated using Respondent Driven Sampling. NVL was defined as the average viral load of HIV-seropositive individuals from a sample of one’s risk network. Multivariate logistic regression analyses were performed to assess the association between NVL and HIV serostatus. Permutation tests were conducted to account for dependency in the sampling scheme. Results: Of 457 respondents, 100%were Black. HIV seroprevalence was 39%. After controlling for total connections, age, substance use during sex, syphilis diagnosis (previous 12 months), and frequency of anal sex (previous 6 months), we found a positive association between NVL and HIV infection. Compared to a network with all HIV-seronegative members, the odds of HIV infection with a NVL of <200 to <10k copies/mL were 2.17 times higher, the odds of a NVL of >10k to <60k copies/mL were 2.38 times higher, and a NVL of >60k copies/mL were 2.80 times higher (all 95% CIs between 1.08-7.25) in the multivariate regression analysis. Conclusion: We found a positive association between NVL and HIV seroprevalence. NVL may have substantial public health implications for HIV-seronegative persons most at risk for HIV infection given that this novel metric avoids overreliance on individual level behavior or broad community indices.

Poster and Themed Discussion Abstracts

850 IDENTIFYING A GEOSPATIAL RELATIONSHIP BETWEEN COMMUNITY-LEVEL RISK AND HIV PREVALENCE Laurence Palk , Sally Blower Univ of California Los Angeles, Los Angeles, CA, USA

Background: Many countries in sub-Saharan Africa show considerable geographic variation in the severity of their HIV epidemics. We hypothesize that geographic variability in HIV prevalence may be due to differences among communities in risk behavior. To test this hypothesis we determine if there is a significant geospatial association between the size of the high-risk group (in a community) and the prevalence of HIV. We use Malawi as a case study.

CROI 2017 367