CROI 2019 Abstract eBook

Abstract eBook

Poster Abstracts

majority (63%) of identified HIV-1C clusters were dyads. In clusters spreading across BCPP communities (n=693), the median genotyping density was 13% (IQR 9–18%; range 3–47%). The median number of identified viral lineages was 38 (IQR 25–56) per community. Viral lineages spreading across multiple communities (median 32; IQR 22–47 per community) were identified more frequently than unique viral lineages seen in a single community (median 5; IQR 2–8 per community). Overall, 74% of HIV-1C lineages were spread across multiple BCPP communities, while only 26%were found in a single BCPP community. Men in clusters (median age 41 years old; IQR 34–47) were about 4 years older than women (median age 37 years old; IQR 31–44; p<0.001). Conclusion: The study revealed a large number of phylogenetically distinct HIV-1C lineages circulating in Botswana by mid-2018. The vast majority of identified viral lineages were spread across multiple communities highlighting the complexity of HIV-1C transmission network. Antoine Chaillon 1 , Margarita Matías-Florentino 2 , Santiago Avila-Rios 2 , Sanjay R. Mehta 1 , Hector E. Paz-Juarez 2 , Manuel A. Becerril-Rodríguez 2 , Silvia del Arenal 2 , Alicia Piñeirúa 3 , Verónica Ruiz 3 , Patricia Iracheta-Hernández 3 , Israel Macías-González 3 , Jehovani Tena-Sánchez 3 , Florentino Badial Hernandez 3 , Gustavo Reyes-Terán 2 1 University of California San Diego, La Jolla, CA, USA, 2 National Institute of Respiratory Diseases, Mexico City, Mexico, 3 Clinica Especializada Condesa, Mexico City, Mexico Background: Non-nucleoside reverse transcriptase inhibitor (NNRTI) pre- treatment HIV drug resistance (HIVDR) has consistently increased in Mexico during the last decade, approaching 10%. Mexico City concentrates a fifth of all persons living with HIV in Mexico and is a main hub for viral dissemination within Mexico. Combining HIV molecular data and epidemiologic information might help to understand HIVDR transmission dynamics. Methods: HIV pol sequences were obtained by next generation sequencing from 2,447 individuals initiating first-line antiretroviral therapy from 09/2016 to 06/2018 at Condesa Clinic, the largest HIV care provider in Mexico. Pre- treatment HIVDR was estimated using the Stanford Algorithmwith a minimum threshold of 20% and 2% to define low-frequency variants. Genetic networks were inferred with HIV-TRACE, establishing putative transmission links with genetic distances <1.5%. Newman’s assortativity coefficients for age and residence were estimated using igraph. Geospatial dispersal was determined by calculating the average distance between centroids of the municipalities of residence of linked individuals. Results: At 20% threshold, pre-treatment HIVDR reached 14.8% overall and 9.6% to NNRTI. K103N/S was the most frequent surveillance drug resistance mutation (DRM) with 7.1% frequency (7.6% at 2% threshold). Putative links with at least one other sequence were found for 963/2,447 (39%) sequences, forming 99 clusters ranging in size from 2 to 20 individuals (Fig. 1A). Clustering individuals were younger (adjusted odds ratio, aOR=0.96, p<0.0001), included a higher proportion of males (aOR=2.3, p=0.001) and a lower proportion of persons residing outside of the central metropolitan area (aOR=0.11, p=0.003). Among clustering individuals, 175/963 (18%) shared drug resistance mutations at 2% threshold in 66 clusters, with 63/175 (36%) sharing K103N/S in 24 clusters (Fig. 1B). Eight municipalities (out of 75) harboured 65% of persons sharing DRMs (Fig. 1C & 1D). The inferred transmission network was assortative by age and municipality (p<0.001). The residence of genetically linked individuals was closer than expected in a random distribution (median distance: 13 km vs. 65 km, p<0.01). Conclusion: DRMs (including low-frequency variants) are frequently transmitted in Mexico City metropolitan area, predominantly among recently diagnosed young men in a densely-sampled, geographically assortative network, warranting serious consideration of non NNRTI-based first-line regimens locally.

876 HIV MOLECULAR SURVEILLANCE AND PRETREATMENT DRUG RESISTANCE IN MEXICO CITY

Poster Abstracts

877 DETERMINANTS OF HIV DIFFUSION ACROSS MEXICO IDENTIFIED THROUGH SPATIAL EPIDEMIOLOGY Santiago Avila-Rios 1 , Simon Dellicour 2 , Claudia García-Morales 1 , Daniela Tapia- Trejo 1 , Margarita Matías-Florentino 1 , Hector E. Paz-Juarez 1 , Sanjay R. Mehta 3 , Davey M. Smith 3 , Gustavo Reyes-Terán 1 , Bram Vrancken 4 , Antoine Chaillon 3 , for the Mexican HIV ResNet Group 1 National Institute of Respiratory Diseases, Mexico City, Mexico, 2 Rega Institute for Medical Research, Leuven, Germany, 3 University of California San Diego, La Jolla, CA, USA, 4 University of Leuven, Leuven, Germany Background: The HIV epidemic in Mexico is highly complex. It is the result of multiple local but intermingled epidemics shaped by various factors including human migration. Here, we characterized the diffusion dynamics of HIV across Mexico. Methods: Using a comprehensive data set of HIV-1 subtype B pol sequences sampled from across Mexico, we applied a multistep phylogenetic approach: (1) we first performed maximum likelihood phylogenetic inference to identify well-supported monophyletic clades within our dataset; (2) all clades of size ≥ 3 identified in step (1) were used to perform a discrete phylogeographic inference to evaluate the dispersal history across the Mexico states; (3) using a generalized linear model (GLM) to test the association of epidemiologic factors (i.e. population size, human emigration and immigration flows) and connectivity (i.e. geographic distances and the intensity of air traffic passenger flow among locations) with lineage dispersal frequencies among the states (Fig 1A). Results: A total of 7,410 unique HIV subtype B partial pol sequences (HXB2 position 2253-3554) from participants originating fromMexico were collected between 2005-2018. After combining these with 2,629 publicly available HIV-1 B pol sequences with known sampling location, we identified 65 clades of size ≥ 3 that represent 798 sequences from 19 states. The discrete phylogeographic analysis based on these clades revealed high levels of virus exchange between Mexico state, Jalisco (including Guadalajara, the second largest city in Mexico) and the eastern touristic state of Quintana Roo (including Cancun), with the border state of Baja California (Fig. 1B). Furthermore, the GLM analysis also suggests that viral migration was strongly associated with the population density, and number of emigrants from the origin state (Bayes Factor, BF>105), the number of immigrants in the recipient state (BF>105), and the intensity of air passenger flows (BF=3.2). The distance between states was also negatively associated with viral movement (negative BF>105), Fig. 1C. Conclusion: This comprehensive analysis of HIV dynamics within Mexico emphasizes the key role of human migration in the diffusion of the HIV epidemic. These results may help to more efficiently allocate prevention

CROI 2019 341