Evaluation of United States-Licensed Human Immunodeficiency Virus Immunoassays for Detection of Group M Viral Variants

Abstract

Six Food and Drug Administration (FDA)-licensed human immunodeficiency virus type 1 (HIV-1) and HIV-1/2 immunoassays, including five enzyme immunoassays and one rapid test, were challenged with up to 250 serum samples collected from various global sites. The serum samples were from individuals known to be infected with variants of HIV-1 including group M subtypes A, B, B′, C, D, E, F, and G and group O. All immunoassays detected the vast majority of samples tested. Three samples produced low signal over cutoff values in one or more tests: a clade B sample, an untypeable sample with a low antibody titer, and a group O sample. It is concluded that HIV-1 immunoassays used in the United States are capable of detecting most HIV-1 group M variants.

Human immunodeficiency virus type 1 (HIV-1) is known to exhibit extreme genetic variability, due largely to high error rates (10⁻⁴) associated with the retroviral reverse transcriptase (5, 17). Indeed, it has been estimated that, with an average of 10⁸ viruses produced daily, virtually every position within the 9.4-kb genome is mutated daily (25). In addition to the HIV-1 group M (major) subtype B virus responsible for the pandemic in the United States, Europe, and Australia, two distinct but related HIVs are known: HIV-1 group O (outlier) and HIV-2 (5, 14). Within the HIV-1 group M family of viruses at least nine phylogenetically distinct subtypes, designated A to I, have been identified (14, 17). HIV-1 strains belonging to multiple group subtypes have been identified in South America, Southeast Asia, Central and Sub-Saharan Africa, and India (5, 14). Recent surveillance efforts have documented a gradual increase in the spread of particular viral variants between continents (27) (www.who.int/emc-hiv/global_report/index.html). Specifically, HIV-1 group M and group O variants have increasingly been identified in Europe (2, 9, 14, 26), and sentinel surveillance sites in the United States have detected various HIV-1 group M subtypes and two group O infections (23, 26). Thus, the global prevalence of HIV-1 group M subtypes appears to be increasing.

The inability of early versions of HIV-1 diagnostic tests to reliably detect HIV-2 led to the development of new test formats, incorporating antigens unique to these viral variants (1, 3, 4, 6). Similarly, the failure of some U.S.-licensed tests to reliably detect HIV-1 group O-infected specimens prompted the FDA to request manufacturers to include group O-specific antigens in future versions of their immunoassays (6, 23). Although a few studies have examined the ability of immunoassays to detect HIV-1 group M non-B subtypes (1, 3, 4, 7, 8, 10, 12, 16, 19), these studies have examined a limited variety of subtypes, and it is unknown how test sensitivity and specificity of currently licensed tests in the U.S. might be compromised by these viral variants. In this study we have examined six U.S.-licensed immunoassays, used to screen the blood supply and for routine diagnosis, for sensitivity in detecting antibodies directed towards HIV-1 group M subtypes. Two hundred forty-nine well-characterized samples representative of subtypes A, B, B′, C, D, E, F, G, and J, as well as six HIV-1 group O samples, were tested with the five HIV-1 or HIV-1/2 enzyme immunoassays (EIAs) and with one HIV-1 rapid test.

MATERIALS AND METHODS

Sera.

Two hundred forty-nine serum samples, originating from over 18 countries, were collected for testing (Table 1). Forty-five serum samples were purchased from BBI (the modified worldwide [WW] HIV-1 performance panel) or acquired as part of the global surveillance activities and HIV-1 domestic surveillance program of the Centers for Disease Control and Prevention (CDC). Samples tested in the present study are part of various ongoing studies throughout the world and were selected based on their HIV-1-positive results in various EIAs. Plasma specimens from 249 HIV-1 group M-infected individuals were selected from Argentina (18 subtype F specimens), Brazil (16 subtype B, 2 subtype C, and 8 subtype F specimens), Cameroon (4 group O specimens), China (6 subtype B specimens), Egypt (1 subtype B specimen), Ghana (5 subtype A, 2 subtype G, and 2 untypeable specimens), India (2 subtype C specimens), Ivory Coast (20 subtype A specimens), Lebanon (12 subtype A, 10 subtype B, and 1 each subtype C and G specimens), South Africa (1 subtype B and 4 subtype C specimens), Thailand (20 subtype B′ and 23 subtype E specimens), Uganda (23 subtype A, 24 subtype D, and 3 subtype C specimens), and Zimbabwe (4 subtype C specimens). In addition, 37 HIV-1-infected persons (16 subtype A, 6 subtype B, 7 subtype C, 2 subtype D, 1 subtype F2 [subcluster of subtype F], 1 subtype J, and 2 untypeable, as well as 2 group O) identified as part of an African surveillance program were included (26). All specimens analyzed in this study were previously typed at the CDC using DNA sequence analysis of the HIV-1 gp120 V3 loop or gp41 sequence analysis (28, 29). Samples were received frozen and had been thawed several times prior to testing.

TABLE 1.

Detection of HIV-1 antibodies by commercial tests

Subtype (n)	% Positivity
	AB HIV1/2	AB HIV1	OTC HIV1	GSC rLAV	GSC HIV1/2 peptide	Murex SUDS
Group M
A (76)	100	100	100	100	100	100
B/B′ (60)	100	100	100	98	98	100
C (23)	100	100	100	100	100	100
D (26)	100	100	100	100	96	100
E (23)	100	100	100	100	100	100
F/F2 (27)	100	100	100	100	100	100
G (3)	100	100	100	100	100	100
J (1)	100	100	100	100	100	100
Untypeable (4)	100	100	100	50	100	100
Group O (6)	100	100	83	83	67	100

PCR amplification and sequence analysis.

A highly sensitive assay based on a conserved sequence within the gp41 region was used for amplification of viral RNA from plasma for HIV-1-positive specimens representing different subtypes of HIV-1 group M (21, 28). Following amplification, DNA from the nested PCR was cycle sequenced (60 ng of DNA per sequencing reaction) with the ABI PRISM Dye Terminator Cycle Sequencing Ready Reaction kit according to the manufacturer's protocol (Perkin-Elmer, Foster City, Calif.) using the nested primers gp46F2 and gp47R2 (28). Sequencing reactions were run in an automated DNA sequencer (model 373; Applied Biosystems, Foster City, Calif.). Sequences were translated and aligned using DNASIS version 2.1 (Hitachi Software, San Bruno, Calif.). Consensus sequences for each subtype were obtained from the 1997 HIV-1 Molecular Immunology Database (Los Alamos National Laboratory, Los Alamos, N.Mex.).

Immunoassays.

AB HIV-1 EIA (Abbott Laboratories, Abbott Park, Ill.) (AB HIV1) is an indirect EIA incorporating two purified HIV-1 proteins and whole viral lysates, coated onto polystyrene beads. Anti-HIV antibodies bound to the HIV-1 antigen-bead complexes are detected by color development with o-phenylenediamine following binding of goat anti-human immunoglobulin G conjugated with horseradish peroxidase.

AB HIV-1/HIV-2 (rDNA) EIA (Abbott Laboratories) (AB HIV1/2) is a sandwich EIA that uses polystyrene beads coated with recombinant HIV-1 env and gag and HIV-2 gag proteins. Anti-HIV antibodies within a bead-antigen-antibody complex are detected by incubation with recombinant HIV-1 env- and gag- and HIV-2 gag-encoded proteins labeled with horseradish peroxidase and color development with o-phenylenediamine.

LAV EIA (Genetic Systems, Redmond, Wash.) (GSC rLAV) is an indirect EIA incorporating HIV-1 whole lysate adsorbed onto wells of a microwell plate. Anti-HIV antibodies bound to the HIV-1 antigen-bead complexes are detected by color development with tetramethylbenzidine following binding of peroxidase-labeled goat anti-human immunoglobulin.

HIV-1/HIV-2 peptide EIA (Genetic Systems) (GSC HIV1/2 peptide) is an indirect EIA incorporating a mixture of four synthetic peptide antigens, derived from highly conserved immunodominant regions of the env and pol gene products for HIV-1 and HIV-2, by adsorption to microwell plates. Antibodies to HIV-1 or HIV-2 are detected by binding of peroxidase-labeled goat anti-human immunoglobulin and color development using tetramethylbenzidine.

Vironostik HIV-1 Microelisa System (Organon Teknika Corp., Durham, N.C.) (OTC HIV1) is an enzyme-linked immunosorbent assay that incorporates HIV-1 whole lysate by adsorption to microtiter plate wells. HIV-1 antibodies in serum or plasma are detected by binding of goat anti-human immunoglobulin conjugated with horseradish peroxidase and color development with ABTS substrate (2,2′-azino-di-[3-ethylbenzthiazoline-6-sulfonate]).

SUDS HIV-1 test (Murex Diagnostics, Inc., Norcross, Ga.) is a rapid (10-min) microfiltration EIA that incorporates HIV-1 gag-encoded antigens, affinity purified from HIV-1 lysate, and a synthetic peptide representing a conserved and immunodominant sequence from the HIV-1 transmembrane protein on latex particles. Bound anti-HIV-1 antibodies are detected with alkaline phosphatase-labeled anti-human immunoglobulin conjugate and color development using 5-bromo-4-chloro-3-indolylphosphate.

Test performance.

All immunoassay testing was performed at the FDA Center for Biologics Evaluation and Research according to the manufacturers' instructions provided in test kit package inserts. For three of the six EIAs all 249 samples were tested once, while 218, 229, or 235 of the samples were tested in the three remaining immunoassays due to specimen volume limitations. For a few samples where low (<1.0) signal-to-cutoff ratios (S/CO) were obtained with an EIA, the test was repeated. In all cases where repeat tests were performed, the two results were in excellent agreement. All EIA results exceeding an S/CO of 1 were deemed positive. S/CO of 0.5 to 1 were deemed equivocal, while S/CO of less than 0.5 were deemed negative. For the Murex SUDS rapid test, the scale of 0 to 4 provided by the manufacturer was used, with all results of 1 or greater scored as positive.

Specimens that were missed by one or more FDA-licensed kits were analyzed by sequence analysis of the gp41 region (10, 15), in an attempt to identify amino acid substitutions in immunodominant domains that might explain reduced antibody reactivity.

RESULTS AND DISCUSSION

In general, all of the immunoassays exhibited excellent sensitivity with the samples tested. All HIV-1 group M specimens were reactive in AB HIV1/2 (243 of 243 specimens), AB HIV1 (236 of 236 specimens), and OTC HIV1 (243 of 243 specimens), giving test sensitivities of 100% (Table 1). In contrast, of the 243 HIV-1 group M specimens, 240 (99%) reacted with GSC rLAV and 241 (99%) reacted with the GSC HIV1/2 peptide EIA. The Murex SUDS rapid test detected all 208 HIV-1 group M samples tested. Importantly, analysis of 31 HIV-1 group M non-B subtypes identified by surveillance programs in the United States (23) revealed that all were detected by each commercial test. Likewise, while all six HIV-1 group O specimens were detectable by the AB HIV1/2 and AB HIV1 EIAs, one specimen from the United States (97US265) was missed by the remaining three tests and an additional specimen from Cameroon (97CM359) was missed by the GSC HIV1/2 peptide assay (Table 2).

TABLE 2.

S/CO for selected specimens^a

Specimen (type and country)	S/CO
	AB HIV1/2	AB HIV1	OTC HIV1	GSC rLAV	GSC HIV1/2 peptide	Murex SUDS
BBI36 (B; China)	7.7	11.9/11.5	1.3	1.3	0.8	1
BBI37 (B; China)	>17	>13	5.4	0.9	3.1	1
BBI20 (D; Uganda)	>17	>13	6.8	5.5	0.33	3
BBI9 (NA; Ghana)	>17	9.5	1.9	0.7	1.3	2
BBI10 (NA; Ghana)	>17	10.8	1.8	0.3	1.4	1
97CM359 (group O; Cameroon)	1.9	1.4	1	1.2	0.9	NT
97US265 (group O; U.S.)	4	8.7/5.7	0.6	0.2	0.2	1

^aNA, no amplification; NT, not tested. 

In most cases, the samples with negative EIA results were HIV-1 group M samples purchased as part of the BBI WW panel (some with low antibody titers) or a single HIV-1 group O sample previously reported as undetectable by several EIAs (22). As previously reported, the HIV-1 group O sample (97US265) was not detected by the OTC HIV1 and GSC rLAV tests or the new GSC HIV1/2 peptide assay. Specimens that were missed by one or more of the assays are listed in Table 2. One each of two clade B samples of Chinese origin included in the BBI WW panel, both exhibiting low antibody titers (BBI samples 36 and 37), were not detected as positive by the GSC rLAV or GSC HIV1/2 peptide assay (S/CO, 0.9 and 0.8, respectively). The equivocal borderline results obtained would, however, most likely trigger a repeat test. The GSC HIV1/2 peptide assay also failed to detect a single subtype C sample from Uganda included in the BBI WW panel (S/CO, 0.33). Two untypeable low-antibody-titer samples from Ghana included in the BBI WW panel gave low or equivocal S/CO (0.3 and 0.7) in the GSC rLAV test.

Such high sensitivity of detection of group M and O sera is in accordance with other studies, where most kits have very high sensitivity and specificity of HIV antibody detection (1, 3, 8, 10, 12, 16). More importantly, we also tested the sensitivity of various subtype specimens against the only rapid test (Murex SUDS) licensed by the FDA. While some tests have previously been shown to have lower sensitivities with some subtypes (16, 20), the Murex SUDS assay was highly sensitive in detecting all subtypes. These results are in general agreement with previous studies where various rapid test kits were found to have comparable sensitivities and specificities with standard EIA-Western blot algorithms (19, 20). Despite such high sensitivity of HIV-1 antibody detection, a very minor subset of HIV-1-infected persons remain seronegative despite active HIV-1 infection (11).

Specimens that were missed by one or more FDA-licensed kits were analyzed by sequence analysis of the gp41 region (10). Results for the amino-terminal immunodominant regions of gp41 are shown in Fig. 1. Analysis of cluster I (amino acids 581 to 615) and cluster II (amino acids 646 to 682) revealed minor amino acid substitutions for the three group M (BBI36, -37, and -20) HIV-1 specimens (10). Likewise, sequence analysis of a group O specimen from the United States revealed several amino acid substitutions (30); however, the direct impact of these changes on antibody detection is unknown. For comparison, gp41 sequences from another group O specimen that was detectable by the commercial tests are shown.

FIG. 1.

Amino acid sequences of specimens missed by EIAs.

Taken together these data provide evidence that most of the FDA-licensed kits, including the rapid test kit, are suitable for diagnosis of HIV-1 infection in seropositive individuals infected with divergent subtypes. However, the sensitivities of these assays for detection of HIV-1 non-clade B subtypes during seroconversion remain to be determined. While a few specimens were missed by one or the other assay, no specific amino acid mutation could be delineated that would account for the lack of antibody detection. More recently, a new variant of HIV-1, termed group N, has been identified (13, 24). Although group N sera show cross-reactivity with the group M test antigen (18, 24), continued effort and testing is needed to ensure detection of emerging variants of HIV.