CROI 2017 Abstract e-Book

Abstract eBook

Poster and Themed Discussion Abstracts

avidity to multiple HIV antigens for determining recent HIV infection. Due to the sensitivity and dynamic range characteristics of the assay, we demonstrated that some antibody biomarkers are remarkably predictive of time since seroconversion which may have additional applications, such as inferring recent transmission events. Identification of early HIV infection (< 3 months since seroconversion) may be useful for determining the efficacy of emergency responses and other interventions in outbreaks, and to infer a timeline for the outbreak. Methods: Plasma samples from persons (n=142) who inject drugs involved in a recent HIV-1 outbreak in rural Indiana were tested with a customized HIV-1 multiplex assay, based on the Bio-Rad Bio-Plex platform, which measures the antibody response to gp120, gp160, and gp41 antigens. Assay cutoffs for each analyte were established to determine whether seroconversion occurred within 30, 60, or 90 days of the sample collection date. The cutoffs were estimated based on the assay values fromwell-characterized, longitudinal seroconversion panels (n=608) with known last negative/first positive antibody test dates. In addition to each individual analyte, an algorithm incorporating three different analytes and their respective cutoffs was applied to the Indiana data to determine early infection status. Results: Sensitivity/specificity of the multiplex assay for predicting seroconversion within 30, 60, 90 days, based on the training data set, was 90.5%/95.4%, 94.1%/90%, and 89.4%/82.9%, respectively. Of the 142 new diagnoses in Indiana between December 2014 and January 2016, the majority of early infections (≤3 months since seroconversion) were estimated to have occurred between February and May. During this period, 13 persons were inferred to have seroconverted within 1 month of the diagnosis, 27 within 2 months, and 42 within 3 months. Conclusion: The HIV-1 multiplex antibody assay can identify and monitor trends in recent infection during a localized epidemic, help assess the impact of public health interventions, and may also be useful for inferring a timeline for the outbreak.

Poster and Themed Discussion Abstracts

878 INFERENCE OF HIV-1 INFECTION DATES IN AN OUTBREAK USING ANTIBODY-BASED RECENCY ASSAYS Ellsworth Campbell 1 , Kelly Curtis 1 , Krystin A. Price 1 , Debra Hanson 1 , Sara J. Blosser 2 , Joan Duwve 3 , Pam Pontones 2 , Philip J. Peters 1 , WilliamM. Switzer 1 1 CDC, Atlanta, GA, USA, 2 Indiana State Hlth Dept, Indianapolis, IN, USA, 3 Indiana Univ, Indianapolis, IN, USA

Background: Serologic assays for determining recent HIV-1 infection are used to estimate HIV incidence by differentiating recent from long-term infection. While an effective public health tool at a population-level, they may also benefit outbreak investigations that are subject to common biases of respondent-driven sampling. Given dates of diagnosis and serologic results we used plasma samples collected from persons who injected drugs during a 2015 HIV outbreak in rural Indiana and from a seroconverter panel of data used to characterize the incidence assays to infer the incidence curve during the outbreak. Methods: Plasma samples (n=608) from recent seroconverters with known last negative/first positive test dates were tested with a customized HIV-1 multiplex assay which measures antibody avidity to three envelope (gp120, gp160, and gp41) antigens. A 4-parameter logistic (4PL) function was fit to a principal component score computed from multiplex assay analytes. Statistics from these training data and the principal component eigenvalues were applied to plasma samples from 142 persons involved in a recent HIV-1 outbreak, collected between November 2014 and March 2016, to compute scores used to predict duration of recent infection (DRI) from the 4PL model parameters. By subtracting the inferred DRI from the date of diagnosis we inferred possible dates of infection (DOI). Results: The earliest HIV-positive diagnoses during the HIV outbreak in rural Indiana occurred between November 2014 and January 2015. Beginning in January 2015, the shape of the curve of cumulative diagnoses over time is more representative of outbreak response efforts than actual incidence during the outbreak. Inferred DOIs suggest that over 70% of HIV infections occurred prior to the first diagnosis and 90% occurred prior to the identification of the transmission cluster. In March 2015, extensive HIV prevention measures were implemented. Although >50% of HIV infections were diagnosed after March 2015, this inferred DOI model indicates that all sampled HIV infections occurred prior to the implementation of these HIV prevention interventions. Conclusion: We developed a novel but simple algorithm for inferring HIV infection dates using results of a multiplex incidence assay. We show that serologic incidence assays can aid outbreak investigations by inferring estimates of infection duration and by overcoming sampling biases inherent to respondent driven investigations that can inaccurately represent true epidemic curves.

CROI 2017 380