CROI 2015 Program and Abstracts

Abstract Listing

Poster Abstracts

Conclusions: The performance of each of the three assays varied by HIV subtype and subtype D had the highest false recent rates. These results highlight the need to optimize and validate testing algorithms for cross-sectional HIV incidence estimation in populations with the relevant HIV subtype distributions 623 Avidity Assay for Cross-Sectional Incidence Based on a 4 th -Generation Combo Ag/Ab EIA Allison R. Kirkpatrick 1 ; Eshan U. Patel 1 ; Connie L. Celum 2 ; Richard D. Moore 3 ; Joel N. Blankson 3 ; Shruti H. Mehta 4 ; Gregory D. Kirk 4 ;Thomas C. Quinn 1 ; Susan H. Eshleman 2 ; Oliver B. Laeyendecker 1 1 National Institutes of Health, Baltimore, MD, US; 2 University of Washington, Seattle, WA, US; 3 Johns Hopkins University School of Medicine, Baltimore, MD, US; 4 Johns Hopkins Bloomberg School of Public Health, Baltimore, MD, US; 5 Johns Hopkins University School of Medicine, Baltimore, MD, US Background: Accurate methods for cross-sectional incidence estimation are needed for HIV surveillance and HIV prevention research. Third generation HIV screening tests, such as Genetic Systems (GS) HIV 1 /2 +O ELISA, have been modified for use as incidence assays, but are not available in many countries and may be replaced with more sensitive 4 th generation HIV assays. We developed an HIV avidity assay based on the 4 th generation GS HIV Combo Ag/Ab EIA (BioRad Combo assay) and evaluated its performance. Methods: The BioRad Combo assay was modified by diluting samples 1:10 and incubating them for 30 minutes at 4°C, with and without 0.025 M diethylamine (DEA). The avidity index (AI) was calculated as the optical density ratio of the DEA-treated to the untreated well. We analyzed 1,541 samples from the United States that were collected in the HIV Network for Prevention Trials (HIVNET) 001/001.1 study, the AIDS Linked to the Intravenous Experience Study (ALIVE) cohort, the Johns Hopkins Hospital Clinical Cohort (JHHCC), and the Johns Hopkins Elite Suppressor cohort (ES). Assay performance was assessed using the modified BioRad Combo assay alone, or with a viral load (VL) assay (classifying samples with a VL <400 copies/mL as non-recent). HIVNET 001 samples were used to estimate the mean duration of recent infection (MDRI) using a time window of 2 years for AI cut-offs of 20-90%. The false recent rate (FRR, fraction of samples misclassified as recent), was calculated for the ALIVE and JHHCC samples. Assay reproducibility was also evaluated. Results: Table 1 shows the performance characteristics of the modified BioRad Combo assay at different AI cutoffs alone and in combination with VL. Neither of these approaches provided a MDRI >100 days with a FRR <1%. All samples from elite suppressors had an AI >80%. AI values were significantly correlated between two technologists (r=0.93; p<0.01).

Conclusions: The performance characteristics of this assay suggest that it may be useful as one component of a multi-assay algorithm (MAA) for HIV incidence estimation. When combined with a VL exclusion the FRR was halved but MDRI estimates were reduced by approximately 10%. This may provide a useful tool for HIV incidence estimation in the future, if 3 rd generation HIV screening tests are replaced with 4 th generation tests for HIV diagnosis. Further research is needed to evaluate the modified BioRad Combo assay in

combination with other assays to accurately estimate HIV incidence based on 4 th generation HIV tests. 624 Estimation of HIV Incidence in a High-HIV-Prevalence Setting, South Nyanza, Kenya, 2012

Poster Abstracts

Andrea A. Kim 1 ; David Maman 2 ; Harrison Fredrick Omondi 3 ; Alex Morwabe 3 ; Irene Mukui 4 ;Valarie Opollo 3 ; Beatrice Kirubi 5 ; Jean-François Etard 2 ; MartinusW. Borgdorff 7 ; Clement Zeh 7 1 US Centers for Disease Control and Prevention, Dpo, AE, US; 2 Médecins Sans Frontières, Paris, France; 3 Kenya Medical Research Institute (KEMRI)/CDC Research and Public Health Collaboration, Kisumu, Kenya; 4 Kenya Ministry of Health, Nairobi, Kenya; 5 Médecins Sans Frontières, Nairobi, Kenya; 6 US Centers for Disease Control and Prevention (CDC), Kisumu, Kenya; 7 US Centers for Disease Control and Prevention (CDC), Kisumu, Kenya Background: Tests for recent infection (TRI) use immunological markers to distinguish recent from chronic HIV infection, providing the ability to estimate HIV incidence and evaluate HIV prevention interventions. Several candidate TRIs, including the Limiting Antigen Avidity Enzyme Immunoassay (LAg) and Bio-Rad Avidity Enzyme Immunoassay (Bio- Rad) are being validated for field performance. We evaluated the performance of the LAg and Bio-Rad in Ndiwha District in South Nyanza, Kenya, a setting with a HIV prevalence of 24% and availability of high-impact HIV prevention interventions. Methods: In 2012, we conducted a two-stage cluster sample household survey of persons aged 15-59 years residing in Ndhiwa District to estimate HIV incidence. Participants testing HIV+ using the national rapid HIV testing algorithm provided a blood sample to test for viral load and recent infection. A person was determined to have recent infection if they 1) tested recent on the TRI; 2) were not virally suppressed, i.e., had an HIV-1 RNA concentration ≥ 400 copies/mL; and, 3) were HIV treatment naive. Cross-sectional incidence was calculated using a mean duration of recent infection of 130 days [95% confidence interval (CI) 118-142] for LAg and 239 days (CI 214-265) for Bio-Rad, and a false-recent rate (FRR) of 0.5% (CI 0.01-0.95) for LAg and 2.4% (CI 1.4-3.4) for Bio-Rad. An incidence to prevalence (I:P) ratio was calculated to assess plausibility of the incidence estimate. Results: Among 6,095 participants, 1,465 were HIV+; 29 tested recent with LAg and 68 with Bio-Rad. HIV incidence was 1.5% (CI 0.6–2.3) with LAg and 1.3% (CI 0.5–2.1) with Bio-Rad. HIV incidence among women was two times higher than among men, respectively, with LAg (2.0%, CI 0.9–3.2 vs. 0.7%, CI 0–1.5) and Bio-Rad (1.6%, CI 0.6-2.0 vs. 0.8%, CI 0–1.7). HIV incidence peaked among persons aged 25-34 years with LAg (1.6%, CI 0.5–2.7) and persons aged 35-45 years (2.2%, CI 0.4–3.7) with Bio-Rad. Among persons aged 45+ years, HIV incidence was negative with Bio-Rad (-0.3%, CI -1.4-0.7) and 0.3% (CI 0-1.4) with LAg. The I:P ratio increased monotonically with age with LAg but not with Bio-Rad. Conclusions: The LAg and Bio-Rad produced similar HIV incidence estimates overall and by sex. Age-specific trends in incidence and prevalence were epidemiologically plausible with LAg but not with Bio-Rad. Negative incidence with Bio-Rad suggests that the sample size was either too low or FRR too high for generating reliable age-specific incidence estimates with Bio-Rad.

393

CROI 2015