Comparison of the Avidity Index Method and the Serologic Testing Algorithm for Recent Human Immunodeficiency Virus (HIV) Seroconversion, Two Methods Using a Single Serum Sample for Identification of Recent HIV Infections

Abstract

A study was designed to compare an avidity index method to the serologic testing algorithm for recent human immunodeficiency virus (HIV) seroconversion (STARHS) for the detection of recent HIV infection. One hundred sixty HIV-positive sera were tested. Both techniques performed similarly in identifying recent infections, although STARHS tended to misclassify more individuals that had long-standing infection as being recently infected.

Increasing interest in estimating the incidence of human immunodeficiency virus (HIV) infection by using cross-sectional specimens has led to the development of several laboratory methods. A sensitive/less-sensitive serologic strategy (the serologic testing algorithm for recent HIV seroconversion [STARHS]) was developed in 1998 (4) at the Centers for Disease Control and Prevention (CDC), Atlanta, Ga., to identify recently infected individuals and has been used in numerous studies (3, 9-11, 13-15). In Catalonia, Spain, STARHS has been introduced as part of enhanced surveillance of HIV and sexually transmitted infections since 2002 (1). Antibody avidity has long been used as a marker for recent infection (2, 5, 8), and in 2002, an automated technique was developed for HIV (17, 18) based on the rationale that in the early phase of infection antibodies show a low avidity for the antigen. In this study, we report the first comparison of these two methods for the detection of recent HIV infection.

A total of 160 serum specimens from HIV-positive subjects (identified according to standard HIV testing algorithms for screening and confirmatory testing) were included in the study. For 66 specimens, the seroconversion date or clinical stage of infection was not available. For 70 specimens, a seroconversion date could be estimated as the midpoint between the last HIV-negative test and the first HIV-positive test (interval between the two tests of <6 months). These specimens were sorted into two groups. Group 1 included 13 specimens from 12 recently infected individuals (sera drawn ≤6 months after the seroconversion date). Group 2 comprised 57 specimens from 26 individuals with long-standing infection (sera drawn >6 months after the diagnosis of HIV infection). Group 3 consisted of 24 individuals with a diagnosis of AIDS.

STARHS was implemented whereby all specimens were first tested using Biotest HIV-TETRA HIV 1 + 2 (Biotest; Diasorin Diagnostics S.R.L., Italy) as the sensitive enzyme-linked immunosorbent assay (ELISA). The Vironostika HIV-1 Microelisa System (bioMerieux, Durham, N.C.) was used as the less-sensitive ELISA after a 1:20,000 three-step dilution as previously described (12). Appropriate controls and calibrator provided by the CDC were used in each run. First, a screening assay using one well per specimen was performed to allow for the identification of probable recent seroconverters according to standardized optical density (SOD), calculated as (patient OD − negative control OD)/(calibrator OD − negative control OD). Specimens with an SOD of <2.000 were then retested in triplicate (confirmatory assay). Specimens with an SOD of <1.000 in the confirmatory assay were considered as belonging to subjects who had seroconverted within the previous 170 days (95% confidence interval, 145 to 200).

For the avidity method, two aliquots of each sample were subjected to a pretest 1:10 dilution. One aliquot was diluted in phosphate-buffered saline (PBS) and the other one in 1 M guanidine hydrochloride (G; Pierce Chemicals, Dallas, Tex.). After incubation at room temperature, both aliquots were tested by an automated microparticle immunoassay (AxSYM HIV 1/2 gO; Abbott Diagnostics Division, Delkenheim, Germany) as recommended by the manufacturer. Sample/cutoff (S/CO) ratios were calculated, and the avidity index of HIV antibodies was computed as (S/CO ratio of the G aliquot)/(S/CO ratio of the PBS aliquot). A cutoff of 0.85 was selected based on previous results (16).

The overall agreement between both techniques was good (κ = 0.67, P < 0.001), where both techniques were concordant in classifying 25 specimens as recent infections and 117 as long-standing infections. Fourteen specimens were considered as recent infections by STARHS while 4 specimens were considered recent infections by the avidity index. Although the avidity technique showed better intra- and interday coefficients of variation (<10% versus 10 to 30%) (17), a number of specimens were close to the cutoff (Fig. 1). Results according to clinical information are shown in Table 1. Both techniques misclassified one out of 13 specimens corresponding to recent infections as being long-standing (the misclassified specimens were different from each other). Six specimens from long-standing infections were misclassified as recent infections by STARHS; one of them had been collected 7 months after the first HIV-positive test and was from a patient on highly active antiretroviral treatment (HAART), while the other five were obtained from subjects who had been infected with HIV for over a year. The only specimen misclassified by the avidity method had been collected 1 year after the first HIV-positive test. Another two specimens from subjects with long-standing infection were from subjects on HAART and were correctly identified as old infections by both techniques. Among the 24 AIDS patients, one was misclassified by STARHS only and another two were misclassified by both assays. Neither of the two tests seemed to be affected by a low CD4 count.

FIG. 1.

Distribution of SOD results for STARHS and avidity indexes according to the three groups of specimens. Dashed line, cutoff.

TABLE 1.

Characteristics of study specimens according to clinical and laboratory information

Group (diagnosis)	No. of samples	No. of recent infections by:
		STARHS	Avidity index	Both techniques
1 (known recent HIV infection)	13	12	12	11
2 (known long-standing HIV infection)	57	6	1	1
3 (AIDS)	24	3	2	2
Unclassified specimensa	66	18	14	11

^aSeroconversion dates or clinical stages of infection were not available.

While STARHS has a proven value, several drawbacks limit its use. First, even small errors in the three-step specimen dilution magnify differences in SOD values, leading to a high variability among replicates and requiring additional testing (6). Secondly, a common calibrator and quality control material, along with a software package, were developed and must be provided by the CDC. Third, STARHS results vary according to HIV subtype, requiring the adoption of different cutoffs (19). Finally, problems with the availability of the commercialized test used for this technique have occurred in the past and will arise again in the near future.

The avidity index method shows several advantages over STARHS, as previously suggested (18). After a single 1:10 dilution, the technique can be fully automated using the AxSYM platform, which leads to better reproducibility. This instrument is available among clinical laboratories in Europe and in most countries worldwide. While STARHS uses a second-generation ELISA based on a virus lysate (HIV-1 subtype B), the avidity index assay is a third-generation enzyme immunoassay that is more sensitive and detects antibodies earlier (it is able to recognize all antibody classes). Moreover, this assay allows the detection of HIV-1, groups M and O, and HIV-2. In addition, the selection of different avidity index cutoffs offers different sensitivities and specificities, depending on the study purpose. Moreover, the avidity index method showed more accurate results than STARHS among patients with long-standing HIV infection or AIDS. In our study there were too few subjects on antiretroviral therapy to evaluate its effect on test performance. However, STARHS tends to misclassify the subjects on therapy as being recently infected, since long-term HAART may lead to a decrease in antibody titers (4, 7, 14). Recently, a longitudinal study that compared STARHS and the avidity index assay for patients who had started antiretroviral therapy (G. Murphy, A. Charlett, G. L. Dean, M. Fisher, D. McElborough, and J. V. Parry, Abstr. XV Int. AIDS Conf., abstr. C11390, 2004) concluded that this therapy had a marked effect on STARHS but had only minor effects on the avidity assay. The avidity index method is easily exportable and does not require any specific training other than that provided by the manufacturer. Finally, the avidity index method does not need official approval by the regulatory authority in any country because it is not a new assay or a modification of an already existing assay and only a preparation of the sample is needed.

It is appropriate to stress that currently the results of both of these tests cannot be interpreted at an individual level. The discrimination between recent and long-standing HIV infection is based on a cutoff that defines the time period from seroconversion to either high antibody titers (STARHS) or antibody maturation (avidity index). However, this time period may vary among individuals, causing misclassification in some cases. Only at the population level will the false-negative and false-positive results balance each other.