CROI 2018 Abstract eBook

Abstract eBook

Poster Abstracts

998 HOW LOW CAN YOU GO? DIAGNOSIS OF ACUTE HIV INFECTION AT VERY LOW HIV RNA LEVELS Donn Colby 1 , Siriwat Akapirat 2 , Supanit Pattanachaiwit 1 , Rapee Trichaviroj 2 , Suteeraporn Pinyakorn 3 , Sasiwimol Ubolyam 1 , Eugène Kroon 1 , Carlo Sacdalan 1 , Nitiya Chomchey 1 , Nittaya Phanuphak 1 , Merlin L. Robb 4 , Praphan Phanunphak 1 , Jintanat Ananworanich 3 , Mark de Souza 1 1 Thai Red Cross AIDS Research Center, Bangkok, Thailand, 2 Armed Forces Research Institute of Medical Sciences in Bangkok, Bangkok, Thailand, 3 US Military HIV Research Program, Bethesda, MD, USA, 4 US Military HIV Research Program, Silver Spring, MD, USA Background: Diagnosis of acute HIV infection (AHI) prior to development of HIV antibodies (Ab) in blood can be challenging. In the earliest stages of AHI, diagnosis may be made solely by the detection of HIV RNA in plasma. Guidelines caution that HIV RNA <5,000 copies/ml may be a false positive result. However, those recommendations are based on data from>15 years ago, when testing methodologies were less specific than with current technology. Methods: The RV254/SEARCH010 cohort has recruited participants with AHI in Thailand since 2009. A total 230,036 screening tests were done and 462 AHI cases enrolled through April 2017. We longitudinally examined participants diagnosed with HIV infection based on initial low viral load (LVL-detectable plasma HIV RNA <5,000 copies/ml) with HIV Ab tests negative or inconclusive. HIV RNA (qualitative and quantitative) and HIV antibody [4th-generation (4thG) and 3rd-generation (3rdG) immunoassays (IA)] were performed at baseline, 12 and 24 weeks. Antiretroviral therapy (ART) was started on all at baseline. Data are presented as median (range) unless noted otherwise. Results: There were 54 (12%) participants diagnosed with AHI based on LVL alone. Initial HIV RNA was 753 (29-4865) copies/ml; 57%were <1,000 copies/ ml. Testing was repeated concurrently with starting ART after 3 (1-6) days with a median change of +1.3 (-0.4 to +3.4) log 10 copies/ml. No false positive HIV RNA tests were identified; all tests were confirmed by repeat HIV RNA, of which 65% were >5,000 copies/ml. Only 3 participants had a fall in HIV RNA at the second test: all had started 3-drug post-exposure prophylaxis (PEP) on the day of the initial test. HIV RNA became undetectable at 4 (IQR 2-8) weeks on ART. Repeat HIV serology with 3rdGIA showed HIV Ab seroconversion in 90% (44/49) at 12 weeks and 88% (43/49) at 24 weeks. Surprisingly, 4thGIA was less sensitive: 63% (31/49) and 51% (25/49) positive at 12 and 24 weeks, respectively. Only 3 participants, 2 of whom received PEP, had both repeat HIV RNA <5,000 copies/ ml and negative HIV Ab at 12 weeks. Conclusion: Diagnosis of AHI based on HIV RNA <5,000 copies/ml alone had a positive predictive value of 100%: no false positive tests were identified. ART can safely be started in these individuals concurrently with confirmatory HIV RNA and Ab testing and sequentially thereafter. The diagnosis of HIV may be more challenging with the use of PEP, and potentially PrEP, which may blunt or delay the HIV Ab response. 999 VALIDATION OF THE AMSTERDAM RISK SCORE FOR RECENT HIV INFECTION AMONG MSM Timothy Lin 1 , Maartje Dijkstra 2 , Godelieve J. de Bree 2 , Maarten F. Schim van der Loeff 2 , Martin Hoenigl 1 1 University of California San Diego, La Jolla, CA, USA, 2 Academic Medical Center, Amsterdam, Netherlands Background: Diagnosis of acute HIV infection (AHI) is challenging and resource-intensive. Dijkstra and colleagues recently described a risk- and symptom-based score that was moderately predictive for seroconversion in a 6 to 12-month period preceding a follow-up visit in men who have sex with men (MSM) in Amsterdam (“Amsterdam Score”). They propose that the Amsterdam Score may reduce HIV nucleotide amplification testing (NAT) and increase diagnostic yield. The hypothesis of this study was that the Amsterdam Score will also be at least moderately predictive for AHI in San Diego (receiver operating characteristic [ROC] area under the curve [AUC] >=0.70), in a cohort with 3-times more HIV-positive test events, but a similar age distribution (33 [interquartile range 27-44] vs 34 [29-41] in Amsterdam). Methods: MSM who tested positive for AHI (antibody-negative, HIV NAT- positive) from 2007 to 2017 or tested NAT-negative in 2017 with the Early Test community-based HIV screening program in San Diego were included for analysis. The Amsterdam Score was calculated for each participant using values described previously with minor adjustments (Table 1). Cases with missing variables were deleted listwise. The Amsterdam Score was assessed with ROC curves; an optimal cut-off score was determined with Youden’s index.

Results: 712 MSM (79 AHI, 633 HIV NAT-negative) were included in the analysis. The Amsterdam Score yielded a ROC curve with AUC of 0.89 (95%CI 0.86 to 0.93). The optimal cut-off score was >=1.8, yielding a sensitivity of 78.5%, specificity of 82.8%, positive predictive value of 36.3%, negative predictive value of 96.9%, positive likelihood ratio of 4.56, and negative likelihood ratio of 0.26. 24.0% of participants would have met this cut-off for NAT testing. Conclusion: The Amsterdam Score was highly predictive (AUC 0.89) of AHI in MSM in San Diego compared to moderately predictive (AUC 0.78) in the original validation cohort. The improved performance may be attributable to more stringent inclusion of only AHI or HIV NAT-negatives in the San Diego cohort. The higher optimal cut-off of >=1.8 (compared to >=1.5 in the Amsterdam cohort) may be explained by overall higher risk behavior in the Early test cohort. Combined risk- and symptom-based scores may demonstrate improved generalizability across different populations compared to existing risk-based scores, and may improve the yield of AHI-diagnosis strategies particularly in settings that do not, or cannot feasibly, routinely test for AHI.

Poster Abstracts

1000 DEVELOPMENT AND VALIDATION OF THE SAN DIEGO SYMPTOM SCORE FOR ACUTE HIV INFECTION Timothy Lin , Tara Tenenbaum, Sara Gianella, Susan J. Little, Martin Hoenigl University of California San Diego, La Jolla, CA, USA Background: Diagnosing and treating acute HIV infection (AHI) decreases transmission, preserves immune function, but is resource-intensive. Risk-based scores have been developed to predict for AHI, but symptom-based scores have the potential to be generalizable among different risk strata. The objective of this study was to derive and validate a “San Diego Symptom Score” (SDSS) which we hypothesized would be at least moderately predictive for AHI (receiver operating characteristic [ROC] area under the curve [AUC]≥0.7). Methods: Analysis included adults who self-presented to the Early Test community-based HIV screening program in San Diego and 1) tested positive for AHI (antibody-negative, HIV nucleotide amplification test [NAT]-positive) from 2007 to 2017 or 2) tested HIV NAT-negative in 2017. Participants were assessed for 11 symptoms in the 14 days prior to the testing event. The sample was retrospectively randomized 2:1 into a derivation and validation set. In the former, symptoms with p<0.2 for AHI in univariate logistic regression models were entered into a multivariate model. Symptoms with p<0.05 were then assigned a score value equivalent to its odds ratio. The score was assessed in the validation set using ROC curve AUC. A cut-off score was found using Youden’s index. Results: 1003 participants (738 men who have sex with men (MSM), 151 non-MSMmen, 111 ciswomen, 2 transwomen, 1 declined to disclose gender) were included, of which 114 had AHI (109 MSM, 3 non-MSMmen, 1 ciswoman and 1 transwoman). Compared to HIV-negative cases, AHI cases were of similar median age (32[interquartile range 25-42] vs 33[27-43], p = 0.11 by Mann- Whitney U) and reported more symptoms (4[2-6] vs 0[0-1], p<0.01). In HIV- negative cases, men and women reported similar a number of symptoms (0[0-1] vs 0[0-1], p=.850). This study sample was representative of the overall Early Test

CROI 2018 382