Research Article: Assessment of Recent HIV-1 Infection by a Line Immunoassay for HIV-1/2 Confirmation

Date Published: December 1, 2007

Publisher: Public Library of Science

Author(s): Jörg Schüpbach, Martin D Gebhardt, Zuzana Tomasik, Christoph Niederhauser, Sabine Yerly, Philippe Bürgisser, Lukas Matter, Meri Gorgievski, Rolf Dubs, Detlev Schultze, Ingrid Steffen, Corinne Andreutti, Gladys Martinetti, Bruno Güntert, Roger Staub, Synove Daneel, Pietro Vernazza, Walid Heneine

Abstract: BackgroundKnowledge of the number of recent HIV infections is important for epidemiologic surveillance. Over the past decade approaches have been developed to estimate this number by testing HIV-seropositive specimens with assays that discriminate the lower concentration and avidity of HIV antibodies in early infection. We have investigated whether this “recency” information can also be gained from an HIV confirmatory assay.Methods and FindingsThe ability of a line immunoassay (INNO-LIA HIV I/II Score, Innogenetics) to distinguish recent from older HIV-1 infection was evaluated in comparison with the Calypte HIV-1 BED Incidence enzyme immunoassay (BED-EIA). Both tests were conducted prospectively in all HIV infections newly diagnosed in Switzerland from July 2005 to June 2006. Clinical and laboratory information indicative of recent or older infection was obtained from physicians at the time of HIV diagnosis and used as the reference standard. BED-EIA and various recency algorithms utilizing the antibody reaction to INNO-LIA’s five HIV-1 antigen bands were evaluated by logistic regression analysis. A total of 765 HIV-1 infections, 748 (97.8%) with complete test results, were newly diagnosed during the study. A negative or indeterminate HIV antibody assay at diagnosis, symptoms of primary HIV infection, or a negative HIV test during the past 12 mo classified 195 infections (26.1%) as recent (≤ 12 mo). Symptoms of CDC stages B or C classified 161 infections as older (21.5%), and 392 patients with no symptoms remained unclassified. BED-EIA ruled 65% of the 195 recent infections as recent and 80% of the 161 older infections as older. Two INNO-LIA algorithms showed 50% and 40% sensitivity combined with 95% and 99% specificity, respectively. Estimation of recent infection in the entire study population, based on actual results of the three tests and adjusted for a test’s sensitivity and specificity, yielded 37% for BED-EIA compared to 35% and 33% for the two INNO-LIA algorithms. Window-based estimation with BED-EIA yielded 41% (95% confidence interval 36%–46%).ConclusionsRecency information can be extracted from INNO-LIA-based confirmatory testing at no additional costs. This method should improve epidemiologic surveillance in countries that routinely use INNO-LIA for HIV confirmation.

Partial Text: Assessment of the number of individuals with early HIV infection is needed for evaluation of the current HIV epidemic and preventive efforts targeted at the different transmission risk populations. Consequently, serologic testing algorithms for recent HIV seroconversion (STARHS) have been developed. These tests utilize the fact that both the concentration and affinity of HIV antibodies in early infection are lower than at later stages [1,2]. The fact that STARHS requires a special assay, which has a deliberately reduced sensitivity compared to HIV screening tests, restricts STARHS to epidemiologic studies. However, for systematic epidemiologic monitoring it would be advantageous if “recency” information (i.e. how recently the infection was acquired) could be simultaneously gained from the same tests being used to diagnose HIV infection.

A total of 765 newly diagnosed HIV infections were reported to the SFOPH during the 12 mo of the study (July 05–June 06). Seventeen cases were excluded due to missing INNO-LIA results, thus leaving a total of 748 patients for evaluation (97.8%). All infections were by HIV type 1. A total of 195 patients met the definition of a recent infection as described in Methods, and 161 patients met the definition of an older infection. Of these, 84 were reported to be in CDC stage B and 77 in stage C. Remaining unclassified were 392 patients, 252 without symptoms (stage A) and 140 with no information provided. The main characteristics of the study population are summarized in Table 6.

We demonstrate that a patient’s antibody reaction in the INNO-LIA HIV I/II Score assay contains information which, similar to the Calypte BED-EIA, allows a distinction to be made between recent and older HIV-1 infection, thus providing a tool for estimating HIV-1 incidence in a population. Because antibody reactions to the five HIV-1 antigens measured by the INNO-LIA evolve over time after the infection (Figure 1), we evaluated various algorithms for their ability to distinguish between recent and older infection in a cohort of 748 unselected HIV-1 infected patients newly diagnosed during a period of 12 mo in Switzerland. The best INNO-LIA algorithms ruled 40%–50% of the 195 infections defined as being recent (up to 12 mo duration of infection) as recent, while ruling 95%–99% of the 161 infections defined as being older as older. In comparison, a dedicated commercial recency test, the BED-EIA, ruled 65% of the recent samples as recent and 80% of the older samples as older (Table 1). Concordance of the two tests was 74%–86% based on stage (Table 2). Both tests exhibited about 80% sensitivity for the early phase of recent infection, while showing reduced sensitivity for the intermediate and, particularly, late phases of recent infection (Table 3). Based on the sensitivity and specificity rates thus obtained from patients of known recency status we estimated the proportion of recent infections in the entire cohort of 748 patients. Utilizing an adjusted model we obtained recency rates of 0.33 and 0.35 for two INNO-LIA algorithms and of 0.37 for the BED-EIA (Table 4), which was similar to the window-based estimate of 0.41 for the BED-EIA (Table 5). There was higher discrepancy for the INNO-LIA algorithms, due to still-unreliable window period estimates.



Leave a Reply

Your email address will not be published.