One or two enzyme-linked immunosorbent assay tests on the first serum sample for initial diagnosis of HIV-1 infection?

In France, the first sample for the initial diagnosis of HIV-1 infection must be tested with two antibody assays: one being an enzyme-linked immunosorbent assay. If one is positive, confirmation tests are performed. We evaluated the performance of initial diagnostic strategies based on the use of one versus two enzyme-linked immunosorbent assay tests, either an antigen-antibody test or a simple antibody assay. We found that a single antigen-antibody test was more efficient than a combination of the two tests.

In 2006, the HIV screening rate in France was 80 per 1000 inhabitants, representing a total of 4.9 million tests and had increased by 4% per year since 2001 (http://www.ecdc.europa.eu/Health_topics/AIDS/Factsheet.htm). In France, outside the context of mother-to-child-transmission, two different HIV-1/HIV-2 antibody tests or combined p24 antigen-antibody tests must be used on the first sera sample for the initial diagnosis of HIV infection, and one must be an enzyme-linked immunosorbent assay (ELISA) ^[1]. A more specific test, such as a western blot or immunoblot, is used to confirm the infection. Depending on the results obtained, additional analyses such as p24 antigen and/or a plasma HIV-1-RNA assay may be performed. Finally, two additional antibody tests are performed on a second sample to confirm the diagnosis. France is one of the last countries to require the use of two different tests on the first diagnostic sample. Most of the other countries, including the USA, use a single ELISA first, followed by reflex testing of ELISA-reactive specimens with a more specific test such as western blot ^[2]. In 2006, the French National AIDS Council recommended that the use of two assays on the first sample for the diagnosis of HIV infection should be revisited as initially suggested by an Agence Nationale d'Accréditation et d'Evaluation en Santé expert group in 2000 ^[3].

The aim of this study was to compare the performance of one versus two tests for initial serodiagnosis of HIV infection; one test being a combined p24 antigen-antibody method and the other an antibody detection method.

All diagnostic samples tested for HIV in three Paris virology laboratories in 2006 were considered for this study. Each laboratory used a combined antigen-antibody assay (Vidas HIV Duo Ultra Biomerieux, Marcy l'Etoile, France; Genscreen plus HIV antigen-antibody, Bio-Rad, Marnes la Coquette, France; or Ag/Ac HIV Combo, Abbott, Rungis, France) and an antibody test (Genscreen HIV-1/2, Bio-Rad or Biotest anti-HIV Tetra Elisa Diasorin, Antony, France). On the basis of the final diagnosis of HIV-1 infection for each sample, we evaluated the sensitivity and specificity of three different strategies for the initial diagnostic testing, namely, an antigen-antibody test and an antibody test, the actual strategy; an antigen-antibody test alone; and an antibody test alone. Sensitivities and specificities of theses three strategies were compared using McNemar tests with Stata 10 (StataCorp, College Station, Texas, USA).

A total of 37 568 samples were tested for HIV infection during the study period, of which 1107 were HIV-1 positive, including 35 samples from recent seroconverters and 41 were HIV-2 positive, including three dual positive. For HIV-1 diagnosis (Table 1), the strategies based on an antigen-antibody test and an antibody test, and on an antigen-antibody test alone, were 100% sensitive in all three laboratories. The sensitivity of an antibody test alone ranged from 98.92% to 99.53%. An antibody assay alone was less sensitive than the other two strategies (99.4 versus 100%, P = 0.016). In each laboratory, the specificity of a single assay (antigen-antibody or antibody alone) was better than that of two assays combined (antigen-antibody and antibody). Antigen-antibody testing was more specific than combined antigen-antibody and antibody testing (P < 0.0001). Similar results were obtained in each laboratory, irrespective of the manufacturer of the combined antigen-antibody tests. Among the 35 seroconversion samples, 28 were positive with both the antigen-antibody test and the antibody test and seven were positive only with the antigen-antibody test [20%, 95% confidence interval (CI) = 8-39]. Twenty-six of these 35 samples were also tested with an HIV1/2 rapid test (Determine, Abbott); 18 were positive and eight negative (31%, 95% CI = 14-52). For HIV-2 infection, all of the 41 samples were positive by both tests (antigen-antibody and antibody).

Table 1

For the diagnosis of HIV-1 infection, we found that the sensitivity of an antigen-antibody HIV screening test was similar to that of an antigen-antibody test combined with an antibody test, and that its specificity was much higher, at least for the tests evaluated in this study. The single-test strategy also had very high sensitivity during the seroconversion phase. The antigen-antibody test has been reported to detect HIV-1 infection 5 days before an antibody test ^[4]. The new combined antigen-antibody assays available in 2008 have a detection limit between 10 and 20 pg of p24 Ag/ml, which is similar to that of dedicated p24 antigen assays ^[5]. This high sensitivity explains why the use of two tests was not more efficient than the use of a single combined test.

Using a single combined HIV-1/HIV-2 antigen-antibody test would substantially reduce the cost of HIV seroscreening. In particular, additional tests that are required when the antigen-antibody test and the antibody test give discordant results would no longer be necessary. Another advantage would be to avoid the anxiety that patients suffer when the results of the two initial ELISA tests are discordant.

In conclusion, this study supports the use of a single combined antigen-antibody assay for initial diagnosis of HIV-1 infection.

Acknowledgements

Conception and design: F. Brun-Vézinet, D. Costagliola, and F. Damond. Provision of study materials or patients: F. Brun-Vézinet, F. Damond, P. Palmer, and C. Rouzioux. Statistical expertise and analysis: D. Costagliola. Drafting of the article: F. Brun-Vézinet, D. Costagliola, and F. Damond. Interpretation of the data, critical revision of the article for important intellectual content, and final approval of the article: F Brun-Vézinet, D. Costagliola, F. Damond, P. Palmer, and C. Rouzioux.