CROI 2018 Abstract eBook

Abstract eBook

Poster Abstracts

570 PERFORMANCE EVALUATION OF THE APTIMA HIV-1 QUANT ASSAY ON THE PANTHER SYSTEM Silvina Masciotra 1 , Wei Luo 1 , Tara Smith 2 , Vickie Sullivan 1 , William Fowler 1 , Steven Ethridge 1 , Laura Wesolowski 1 , Kevin P. Delaney 1 , WilliamM. Switzer 1 , S. Michele Owen 1 1 CDC, Atlanta, GA, USA, 2 Oak Ridge Institute for Science and Education, Atlanta, GA, USA Background: The APTIMA HIV-1 Quant Assay (APT-Quant) is FDA-approved for quantification of HIV-1 plasma RNA from 30-10 7 copies (cp)/ml. Outside the U.S. quantitative results can also be interpreted qualitatively. We evaluated the performance of APT-Quant for HIV-1 RNA quantification and for off-label diagnostic use. Methods: One AccuSpan HIV-1 RNA linearity panel (LP) (n=30 or n=43 with two kit lots), U.S. samples from early HIV-1 infections collected from 46 seroconverters (n=420, subtype B), Cameroonian antibody-positive samples (n=113, non-B subtypes and Group O), and HIV-negative samples (n=478) were tested with APT-Quant on the Panther system. Volume permitting, samples were also tested with APTIMA HIV-1 RNA Qualitative (APT-Qual), Roche COBAS AmpliPrep/COBAS TaqMan HIV-1 Test v2.0 (Roche viral load (VL)), and Abbott RealTi m e HIV-1 (Abbott VL) assays. We analyzed agreement between APT- Quant and other FDA-approved VL assays by linear regression and concordance correlation coefficient, calculated specificity in APT-Qual-nonreactive/HIV-1 WB-negative (HIV-negative) specimens and evaluated test reproducibility (standard deviation) using two kit lots with LPs. We analyzed APT-Quant for diagnostic use by comparing to APT-Qual results (McNemar’s paired test). Results: Table shows agreement between APT-Quant and other VL assays. Using LPs, the standard deviation ranged from 0.021-0.113 (lot 1) and 0.015- 0.489 (lot 2) with high variability at ~3 log(cp/ml). APT-Quant specificity was 99.79% [CI: 98.82- 99.96%]. APT-Quant detected virus in 34 more samples than APT-Qual in early stages of infection (n=417, p<0.0001) and in all 105 APT-Qual- reactive Cameroonian samples. Of 228 samples from seroconverters, APT-Quant detected virus in 180 samples, Roche VL in 163, and APT-Qual in 157. Both APT-Qual and Quant missed two infections with Roche VLs <20 cp/ml and one with 23 cp/ml. APT-Quant was nonreactive in two samples with VLs of 29 and 32 cp/ml. Controls failed in one run for LP and three panel members >7 log(cp/ml) were invalid needing 1:100 dilution (kit lot 2). Conclusion: APT-Quant showed good agreement with Roche and Abbott VL assays in Group M and O specimens and with high sensitivity and specificity. APT-Quant performance exceeds APT-Qual for detecting infections with low VL in early stages of infection and established infections. The APT-Quant assay could also be used for diagnostics if properly validated in a laboratory or via an FDA diagnostic claim.

571 EVALUATION OF THE APTIMA HIV-1 QUANT DX ASSAY FOR MONITORING OF HIV-1 VIRAL LOAD Mónica García-Álvarez , Sagrario Zurita, Francisca Gutiérrez, Paula Aranguren 1 , Samanta Aparicio, Rafael Delgado Hospital Universitario 12 de Octubre, Madrid, Spain Background: Viral load monitoring of HIV-1 is critical for effective HIV- infection management. This has led to the development of different HIV-1 quantitative assays. The Aptima HIV-1 Quant Dx is a quantitative assay based on real-time transcription mediated amplification and detection, run on the fully automated and random access Panther System instrument (Hologic). The objective of this study was to evaluate correlation, linearity, reproducibility and performance characteristics of the Aptima HIV-1 Quant Dx Assay (Aptima) in comparison to the reference method, Roche cobas 6800 System (c6800). Methods: For clinical reactivity correlation, 120 retrospective and 300 prospective collected plasma samples were tested in parallel in Aptima in side-by-side testing to the reference method cobas® HIV-1 6800. Linearity and reproducibility were assessed by replicate testing of serial dilutions of well- characterized clinical samples (5 replicates of seven dilution levels spanning the assay dynamic range), and an external proficiency panel (AcroMetrix™ HIV-1 panel). Subtype reactivity was addressed in Aptima by testing an external panel (SeraCare HIV-1 Genotype). A workflow study was conducted to define the performance characteristics: protocols were designed to test different laboratory workflows when running HIV, HCV and HBV viral load assays on 3 different molecular diagnostics instruments (c6800, Veris and Panther). Results: Correlation between Aptima and c6800 assays in a set of 100 clinical samples within the linear range of both techniques was 0.965. The Bland- Altman mean difference estimate between the two methods was 0.24 Log cp/ mL. Both systems demonstrated linearity over the dilution range (0.997 for c6800 and 0.998 for Aptima). The intra-assay variability was 1.24% for c6800 and 1.76% for Aptima, and the mean coefficient of variation inter-assay was 1.52%. Specificity for Aptima on negative samples was 100%. Aptima allowed detection and quantification of all HIV-1 subtypes tested (A, AG, B, C, D, AE, F, G and H). Panther system allowed rapid and fully automated testing: average time to first result was 161 minutes and turn around time was the shortest at different scenarios. Conclusion: The Aptima HIV-1 Quant Dx is a sensitive and reproducibility assay for monitoring of HIV-1 viral load. Random access, large capacity and improved time to result make the Aptima HIV-1 Quant Dx Assay run on the Panther System a good candidate for measuring HIV-1 viral load in HIV-1 infected patients. 572 CEPHEID XPERT HIV-1 VIRAL LOAD ASSAY PERFORMANCE EVALUATION Laura Wesolowski 1 , William Fowler 1 , Wei Luo 1 , Silvina Masciotra 1 , Kevin P. Delaney 1 , Emeka Oraka 2 , Pollyanna Chavez 1 , Steven Ethridge 1 , WilliamM. Switzer 1 , S. Michele Owen 1 1 CDC, Atlanta, GA, USA, 2 ICF International, Atlanta, GA, USA Background: The performance of the Cepheid Xpert HIV‐1 Viral Load (Xpert VL) has not been extensively evaluated. This new, simplified, automated single- use quantitative VL assay uses 1-1.2mL of plasma on the Cepheid GeneXpert Systemwith a reported limit of detection of 40 copies/ml and 100% specificity 95% CI (96.7–100%). Methods: Using an HIV-1 RNA linearity panel (AccuSpan) previously tested with Roche COBAS AmpliPrep/COBAS TaqMan HIV-1 Version 2.0 (Roche VL), we examined agreement between both VL assays by plotting the log VL of each concentration, fitting a regression line and calculating the correlation. We compared the proportion of 221 commercial plasma seroconverter specimens detected using Xpert VL and Roche VL. Among seroconverters, we calculated the proportion of specimens with an Xpert log 10 value ≥2.3 (threshold for virologic failure) and ≥3.0 (threshold for resistance testing) when the log 10 Roche VL was at least 2.3 and 3.0, respectively. We also tested 497 archived, uninfected plasma specimens with Xpert VL to measure assay specificity. Results: Roche and Xpert VLs were highly correlated (R²=0.994) (Figure 1). There were 6 (2.7%) Xpert VL errors among seroconverters, indicating that the assay was aborted. There was an invalid (HIV-1 RNA presence or absence could not be determined). These specimens lacked volume for repeat testing. Of 153 seroconverter specimens with virus detected by Roche VL, 145 (94.8%) were detected by Xpert VL. Seroconverter specimens detected by Roche VL but missed by Xpert VL were detectable but not quantifiable (n=2) or had log 10 VLs between 1.37 and 1.62 copies/mL (n=6). Of 61 seroconverter specimens not detected by Roche VL, 59 (96.7%) were not detected and 2 (3.3%) were

Poster Abstracts

CROI 2018 210