CROI 2018 Abstract eBook

Abstract eBook

Poster Abstracts

Sensitivity analyses of varying mobility definitions (e.g. away >3 weeks/year) did not change findings qualitatively Conclusion: Mobile individuals were significantly more likely to be viremic, a primary risk factor for HIV transmission. Health systems may need to better accommodate more mobile populations, to achieve high treatment and viral suppression targets.

938 EPIDEMIOLOGICAL STUDY OF TRANSMISSION CLUSTERS IN A LOCAL HIV-1 COHORT Carmen M. González Domenech 1 , Isabel Viciana-Ramos 1 , Gabriel Sena- Corrales 2 , Laura Mora Navas 1 , Guillermo Ojeda 1 , Enrique Nuño 1 , Encarnación Clavijo 1 , Rosario Palacios 1 , Jesus Santos 1 1 Hospital Virgen de la Victoria, Málaga, Spain, 2 Hospital Regional Carlos Haya, Malaga, Spain Background: Integration of molecular, clinical and demographic data represents a powerful tool to understand the dynamics of local transmission HIV-1 chains (TCs). The aim of our study was the phylogenetic analysis of the TCs within a HIV-1 cohort and the description of the relevant patients´ data within a TC. Methods: We performed a phylogenetic analysis of 757 sequences from newly HIV-1 diagnosed patients in Málaga (Southern Spain) during the period 2004-2015. We used the partial pol gene sequence in a preliminary phylogeny by Neighbour Joining method (MEGA v6.06 program) and after eliminating all those branches with bootstrap values<80%, we constructed a new phylogeny by Maximum likelihood method (FastTree program). We consider a TC any cluster with bootstrap values≥90%. Patients within and out TCs were compared. Resistance mutations in PR and RT sequences were analyzed by Stanford algorithm. Results: 451 out of 757 patients (59.6%) were grouped into 53 TCs, 17 of them with five or more subjects. The largest number of patients associated within a TC was 90. Patients younger than 40 years (OR 1.75, 95%CI 1.2-2.4, p=0.002), MSM (OR 2.14, 95%CI 1.3-3.2, p<0.0001), non-Spanish (OR 1.48, 95%CI 1.0-2.1, p=0.038), with a non-B subtype HIV-1 (OR 3.12, 95%CI 2.0-4.8, p<0.0001), and presenting primary resistance mutations (OR 14.1, 95%CI 3.1-62.6, p=0.001), were more likely to be associated within a cluster. 94 out 118 patients (79.6%) with transmission resistance mutations were included in some TC. The most frequent mutations associated with clusters were T69D/N, L210W and K219E/Q, for NRTIs, K103N and G190A/S for NNRTIs, and the I54L/M and L90Mmutations for PIs. The prevalence for resistance to NNRTIs in TCs was 13.7%. There were two TCs of peculiar non-B subtypes: CRF19_cpx, with 21 individuals, 16 of them (76.2%) with mutation G190A; and CRF51_01B with 39 patients, 20 of them with the K103N mutation. Conclusion: About 60% of newly HIV-1 diagnosed patients were included in a TC. Younger patients, MSM, non-Spanish, with non-B subtype HIV-1 and primary resistance mutations possessed more probability of belonging to a cluster. NNRTIs mutations were the most frequent ones among patients in TCs. We observed two TCs represented by infrequent non-B subtypes in our area, like CRF19_cpx and CRF51_01B, both of them associated to the transmission of primary resistances. 939 AGE-DEPENDENT RACIAL/ETHNIC DISPARITIES IN LONGITUDINAL HIV CARE INDICATORS Fidel A. Desir 1 , Catherine R. Lesko 1 , Richard D. Moore 1 , Michael J. Silverberg 2 , Peter F. Rebeiro 3 , Michael A. Horberg 4 , Mari Kitahata 5 , Stephen Crystal 6 , Angel Mayor 7 , Marina Klein 8 , Amy C. Justice 9 , Michael John Gill 10 , Keri N. Althoff 1 1 Johns Hopkins University, Baltimore, MD, USA, 2 Kaiser Permanente Northern California, Oakland, CA, USA, 3 Vanderbilt University, Nashville, TN, USA, 4 Kaiser Permanente Mid-Atlantic States, Rockville, MD, USA, 5 University of Washington, Seattle, WA, USA, 6 Rutgers University, Newark, NJ, USA, 7 Universidad Central del Caribe, Bayamon, Puerto Rico, 8 McGill University Health Centre Research Institute, Montreal, QC, Canada, 9 VA Connecticut Healthcare System, West Haven, CT, USA, 10 University of Calgary, Calgary, AB, Canada Background: Maximizing the amount of time spent in care, on antiretroviral therapy (ART), and with viral suppression (VS) after linkage to HIV care is critical to improving the health of persons with HIV. Although racial/ethnic disparities in these HIV care indicators have been described, the effect of age

937 NEW METHOD FOR RAPID DETECTION OF HIV TIME-SPACE CLUSTERS FOR PUBLIC HEALTH ACTION Laurie Linley, Arthur Fitzmaurice, Meg Watson, Chenhua Zhang, Riuguang Song, Anne Marie France , Alexandra M. Oster CDC, Atlanta, GA, USA Background: CDC has not previously developed systematic methods to use HIV diagnosis data in real time to detect possible outbreaks of HIV. We sought to determine whether we could apply methods of time-space cluster detection to U.S. HIV surveillance data and identify possible clusters of increased diagnoses to focus high-impact prevention efforts. Methods: We developed a systematic method for determining increased numbers of HIV diagnoses above expected baselines (“alerts”) in a given geographic area, as these might represent possible transmission clusters or outbreaks. Using National HIV Surveillance System data reported through December 31, 2016 from 51 jurisdictions (50 U.S. states and the District of Columbia), we compared the number of cases reported in 2016 to the previous 3-year baseline period by jurisdiction and county, both 1) for all diagnoses and 2) for those with a transmission risk category of injection drug use (IDU). An alert for a given area was generated when two criteria were met: a statistically greater number of cases for the most recent year (by 2 standard deviations) than the 3-year mean of the baseline period, and an increase of more than 2 diagnoses over the baseline mean. To improve sensitivity given possible reporting delays, the analyses were performed with and without lags of up to 3 months. Results: In analyses of all diagnoses by jurisdiction, alerts occurred for 12 (24%) of the 51 jurisdictions, of which 4 alerted without lags. At the county level, alerts occurred for 265/3,142 (8%) counties (143 without lags). The median and mean were 4 and 6 county alerts per jurisdiction, respectively. A higher percentage of counties with alerts than counties without alerts were located in the South (Table). For cases with IDU as a risk, alerts occurred for 7/51 (14%) jurisdictions, and 39/3,142 (1%) counties. Compared with counties without IDU alerts, a higher percentage of counties with IDU alerts were in the Northeast. Alerts were found in counties with low (<3), medium (3–9), and high (10+) baseline burden of HIV diagnoses (Table). Conclusion: This method of time-space cluster detection identifies significant increases in annual HIV cases across all regions and for counties with varying levels of disease burden. Use of this tool in near real time to provide systematic automated detection of possible increases in diagnoses that merit further investigation can serve to prioritize and focus prevention efforts in local areas for maximal public health impact.

Poster Abstracts

CROI 2018 358