CROI 2019 Abstract eBook

Abstract eBook

Poster Abstracts

antigen/antibody (Ag/Ab) test result were tested with a nucleic acid test (NAT) in 10-member pools. Specimens with reactive Ag/Ab results and negative or indeterminate supplemental antibody test results, were tested individually using a NAT. For both situations, specimens with a negative NAT result were classified as HIV-uninfected. Specificity of the POC tests with exact 95% confidence intervals (CI) were calculated based on the HIV-uninfected status of the specimen stratified by participant’s report of current PrEP use. Results: Among 1,434 HIV-uninfected specimens, 16.7%were from persons on PrEP at the time of the clinic visit, 80% from persons not currently on PrEP, and 3.5%missing data on PrEP status. There were 8 specimens with false-positive results, 2 from persons on PrEP. No specimen tested false positive on more than one test. False-positivity rates were 0.4% for Determine and 0.1% for INSTI. DPP and OraQuick performed on whole blood produced no false-positive test results. Specificity was high and comparable for all tests and was not affected by PrEP use. Conclusion: Point estimates for specificity are higher than what we have previously published. The high specificity of these HIV POC tests, including when used with participants taking PrEP, should reassure organizations implementing rapid HIV testing using whole blood specimens. However, the possibility of false-positive results should still prompt organizations to establish mechanisms for either additional HIV testing onsite (using a different rapid HIV test) or follow-up laboratory testing to confirm any positive result.

the median anti-Env titer was 3200 (IQR; 2400-6400) in participants with VISR according to the Tanzanian algorithm and 800 (IQR; 400-1600) in those without VISR, p<0.0001. Conclusion: HIV diagnostic algorithms currently used in sub-Saharan Africa will misclassify a proportion of HIV vaccine recipients, but fewer than the Enzygnost Integral ELISA. The Mozambican HIV rapid test algorithmwas significantly more accurate than the Tanzanian algorithm. Development of HIV rapid assays that can adequately differentiate VISR from true HIV infection should be prioritized. 556 DIFFERENTIATION CAPABILITY OF THE GEENIUS ASSAY FOR HIV-2 AND HIV-1/2 DUAL INFECTIONS Ming Chang 1 , Eugene Deng 1 , Jose Ortega 1 , Dana Raugi 1 , Babacar Faya 2 , Fatou Kine Simal 2 , Samba Cisse 2 , Khadim Faye 3 , Robert A. Smith 1 , Moussa Seydi 2 , Geoffrey S. Gottlieb 1 , Robert Coombs 1 1 University of Washington, Seattle, WA, USA, 2 CHU de Fann, Dakar, Senegal, 3 Ministere de la Sante et de l’action sociale du Senegal, Dakar, Senegal Background: Detection and discrimination of HIV-1 and HIV-2 antibodies is a key component of the US CDC HIV diagnosis algorithm. However, differentiation between HIV-2 single- and HIV-1/HIV-2 dual infection by serology alone is challenging. The Bio-Rad Geenius HIV 1/2 supplemental assay (Geenius) is a commonly used, US FDA-approved, assay for HIV-1 and HIV-2 immuno- differentiation. In this study, we evaluated the Geenius assay’s output characteristics in the United States (US) HIV positive patient plasma samples that had a clinical diagnosis of HIV-2. Methods: HIV-2 patients’ plasma samples, originating from US clinics and laboratories that were referred for HIV-2 quantitative RNA viral load testing to the University of Washington Retrovirology Laboratory between 2011 to 2018, were retrospectively tested by the Geenius assay. Results were read and interpreted by the Geenius Reader with the proprietary US software (Bio-Rad). Results: Senegalese plasma samples from known HIV-2-infected (n=20) and HIV-1/HIV-2 dually-infected (n= 8) subjects were used to verify the Geenius assay (Table 1). The Geenius assay algorithm output from 65 US patients’ plasma samples with clinically diagnosed HIV-2 was as follows: 27 (41.5%) were HIV-2 positive; 31 (48%) were HIV-2 positive with HIV-1 cross-reactivity; 6 (9%) were HIV positive-untypable; and 1 (1.5%) was HIV-2 indeterminate (Table 1). Notably, 7 samples designated by Geenius as HIV-2 positive with HIV-1 cross-reactivity were reactive to all HIV-2 gp36, gp140 and HIV-1 p31, gp160, p24 and gp41 antigen (Ag) bands. The Geenius interpretation for 4 HIV positive- untypable and one HIV-2 Indeterminate samples were confirmed by additional plasma samples from subsequent dates. Conclusion: Although the Geenius assay confirmed 20 HIV-2 single- and 8 HIV-1/-2 dual- infection diagnosed from Senegalese plasma samples; nearly half of HIV-2 single-infection plasma samples were also reactive to HIV-1 Ag bands. Variable results were also obtained by Geenius for HIV-2 samples collected from the US, with 7/65 (10.8%; 95% CI 4.4-20.9%) giving untypable or indeterminate results and nearly half showing some cross-reactivity to HIV-1. Additional tests are needed for confirming HIV-2 single infection and differentiating HIV-1/HIV-2 dual infections. Validated nucleic acid amplification testing for HIV-2 and HIV-1/ HIV-2 dual infection may improve the CDC algorithm in this patient population.

Poster Abstracts

555 PERFORMANCE OF HIV DIAGNOSTIC ALGORITHMS IN THE PRESENCE OF VACCINE-INDUCED IMMUNITY Frank Msafiri 1 , Alice Manjate 2 , Sarah Lindroth 3 , Said Aboud 1 , Eligius Lyamuya 1 , Sören Andersson 1 , Charlotta Nilsson 4 1 Muhimbili University of Health and Allied Sciences, Dar es Salaam, Tanzania, United Republic of, 2 Universidade Eduardo Mondlane, Maputo, Mozambique , 3 Örebro University, Örebro, Mozambique, 4 Karolinska Institute, Stockholm, Sweden Background: Participants in HIV vaccine trials are at risk of being misclassified as HIV-infected since routine tests may fail to distinguish vaccine induced antibodies from those elicited by infection. We assessed the performance of HIV testing algorithms to distinguish vaccine-induced seroreactivity (VISR) from true infection. Methods: Stored serum/plasma samples from healthy Swedish and Tanzanian volunteers who participated in any of three previously conducted phase I/II vaccine trials evaluating an HIV-DNA prime HIV-modified vaccinia virus Ankara (MVA) boost strategy were analyzed. HIV infection in participants was ruled out by HIV RNA PCR. Samples were tested for VISR using the HIV testing algorithms of Tanzania and Mozambique, which use two sequential rapid diagnostic tests. SD Bioline HIV1/2 (Standard Diagnostic Inc, Republic of Korea) for screening and Uni-Gold HIV-1/2 (Trinity Biotech, Ireland) for confirmation of HIV infection in Tanzania. Determine HIV-1/2 (Alere Medical Co. Ltd, Japan) for screening and Uni-Gold HIV-1/2 for confirmation of HIV infection in Mozambique. In both countries, patients are considered HIV-infected if both assays are reactive, and discrepant results are resolved by repeated testing. The vaccinees’ samples were also tested for VISR using Enzygnost HIV Integral 4 ELISA (Siemens, Germany). Antibodies to subtype C gp140 were determined using an in-house ELISA. Results: VISR as determined by the Enzygnost HIV Integral ELISA was 92% (61/66). The proportion of vaccine recipients that would have been falsely labeled as HIV positive by the HIV diagnostic algorithm used in Mozambique was half of that by the Tanzanian algorithm, 10/66 (15%) and 21/66 (32%), respectively, p= 0.039. The median anti-Env titer was 3200 (IQR; 3200-12800) in vaccinees with VISR according to the Mozambican algorithm compared to median 800 (IQR; 400-1600) in participants without VISR, p<0.0001. Similarly,

557 HIV-1 RNA DETECTION BY ABBOTT M2000 CORRELATES WITH INTEGRASE SINGLE-COPY ASSAY Melissa A. Tosiano 1 , Hanna Mar 2 , Joshua C. Cyktor 1 , Dianna L. Koontz 1 , Joseph J. Eron 3 , Rajesh T. Gandhi 4 , Deborah McMahon 1 , Ronald Bosch 2 , John W. Mellors 1 , for the ACTG A5321 Team

CROI 2019 208