CROI 2015 Program and Abstracts

Abstract Listing

Poster Abstracts

627 Use of the Sample-to-Cutoff Ratio (S/CO) to Identify Recency of HIV-1 Infection Eric M. Ramos ; José Ortega; Glenda Daza;Yuree Namkung; Socorro Harb; Joan Dragavon; RobertW. Coombs University of Washington, Seattle, WA, US

Background: There are two FDA-approved 4th generation assays available that have the capability to detect both HIV-1/2 specific antibodies and HIV-1 p24 antigen, allowing for the diagnosis of acute HIV-1 infection (AHI). Both assay are considered reactive at sample/cutoff ratio (S/CO) ≥ 1 and non-reactive at S/CO<1. Since the S/CO signal increases with the quantity of antigen and antibodies presents in the sample, it should be possible to use the S/CO range to differentiate between negative, AHI, recent and established HIV infection status. Methods: All the samples were run with the Abbott ARCHITECT HIV Ag/Ab Combo CMIA (ARCHITECT) and the Bio-Rad GS HIV Combo Ag/Ab EIA (GSCOMBO). The following testing algorithmwas used: S/CO<1 with negative nucleic acid amplification test (NAT): negative; S/CO ≥ 1 with Bio-Rad Multispot HIV-1/2 rapid test (MS) non-reactive and positive NAT: AHI; ARCHITECT or GSCOMBO reactive, MS-reactive with a confirming Western blot (WB) without or with the band p31+ present: recent or established infection, respectively. Results: A total of 150 clinical specimens were evaluated. Ninety-nine samples with a S/CO<1 were confirmed as negative with an ARCHITECT and GSCOMBO S/CO median and interquartile range [IQR] of 0.11 [0.09-0.13] and 0.27 [0.25-0.28], respectively. Fifty-one samples had a S/CO ≥ 1, of which 25 confirmed as AHI (Fiebig II) with ARCHITECT and GSCOMBO S/CO median [IQR] of 12.5 [4.7-74] and 11.7 [5.8-14], respectively, and with a viral load median [IQR] of 1.07x10^6 RNA copies/mL [6.90x10^5 - 10x10^6]. Of the 26 specimens that were MS reactive and WB positive, 21 specimens were confirmed as established infection (Fiebig VI) and only 5 lacked the WB-p31+ band, which indicated recent infection (Fiebig V). The ARCHITECT S/CO medians [IQR] for recent and established infection were 418 [384-449] and 914 [785-1061] respectively; both S/CO ratios were the same (14 [14]) for the GSCOMBO. The GSCOMBO S/CO also reached 14 in 40% of AHI (10 samples). There were statistically significant differences in the ARCHITECT S/CO median [IQR] between AHI, recent and long-term infection (Kruskal–Wallis, p<0.0001) but not for GSCOMBO (Figure 1).

Poster Abstracts

Conclusions: In this small study both the GSCOMBO and ARCHITECT identified AHI equally well but the ARCHITECT S/CO dynamic range was able to further differentiate between AHI, recent and established infection. The use of the ARCHITECT S/CO to identify recency of HIV-1 infection requires confirmation in a larger study. 628 An Abbott Architect Combo Signal to Cut-Off RatioWith Adequate PPV to ConfirmHIV Tomas O. Jensen 1 ; Peter Robertson 2 ; Jeffrey J. Post 1 1 Prince of Wales Hospital, Randwick, Australia; 2 South Eastern Area Laboratory Services, Prince of Wales Hospital, Sydney, Australia Background: Recent revisions of HIV testing algorithms have aimed to minimise the time to confirmation of the diagnosis without losing sensitivity and specificity which has clear advantages to both patients and clinicians. We hypothesised that the positive predictive value (PPV) of the Abbott Architect HIV Ag/Ab Combo Assay increases with the signal to cut-off (S/CO) ratio and that a S/CO ratio could be identified with a sufficiently high PPV to give a positive result to patients without waiting for the results of supplementary testing. Methods: All testing episodes were extracted from the laboratory database between March 2006 and March 2014. Samples came from a wide range of adult clinical services including medical and surgical inpatient departments, other laboratories, sexual health clinics, infectious diseases clinics, a women’s hospital, IVF clinics, general practitioners and prison health services. Positive tests were classified as true positive or non-true positive depending on synchronous and subsequent supplementary testing with a Western Blot (WB), a HIV p24 Ag assay (confirmed by neutralization), a HIV Ab assay (tested with an alternate EIA) and/or HIV viral load. The data were randomly allocated to two equally sized samples – a train sample and a test sample. Results: Of 138,911 testing episodes, 3,705 had a positive result. The true positive group included samples with : 1) a synchronous positive WB (N=1,989), 2) an initial negative or indeterminate WB followed by a positive WB within 6 months (N=79), 3) a negative or indeterminate WB and at least two positive results of the p24 Ag assay, the HIV Ab assay or the HIV viral load (N=435), 4) a negative or indeterminate WB and a positive p24 Ag assay without a HIV viral load result available (N=38).The highest non-true S/CO ratio in the train sample was 151.17 and when this was applied to the test sample it had a PPV of 100% and a sensitivity of 67.4%. S/CO ratios of 100 and 50 had a PPV and sensitivity of 99.7% and 73.8% and 99.0% and 80.6% respectively. The frequency distribution of the S/CO ratios of true and non-true positives in the total sample is shown in figure 1.

395

CROI 2015