INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

The type-specific serodiagnosis of human immunodeficiency virus type 1 (HIV-1) and HIV-2 is a critical initial step in understanding the transmission, surveillance, and pathogenesis of HIV in geographic areas where both viruses are endemic. The classic strategy for the serodiagnosis of HIV infection includes screening sera by enzyme immunoassays and confirmation by Western blotting (WB). Because of the high cost and complexity of this confirmatory step, several simplified enzyme immunoassay-based alternative strategies have been proposed (11, 12, 15, 21). Although these strategies are both sensitive and specific, none of them takes into account the type-specific serodiagnosis of HIV-1 and HIV-2. Several methods exist that allow this type-specific serodiagnosis to be made. These include the WB assay, synthetic-peptide-based line immunoassays (LIAs) that recognize antibodies to the transmembrane glycoproteins of the two viruses (gp41 for HIV-1 and gp36 for HIV-2) (4), and the dot blot analysis, which uses recombinantly expressed env peptide (7). However, each of these strategies has limitations. WB testing lacks specificity because of the tendency of the HIV-2 glycoprotein 36 (gp36) to form trimers of about the same size as gp120 of the HIV-1 envelope and thereby cross-reacting in the HIV-1 WB (5, 13). For example, only about 45% of specimens classified as dually HIV-1 and HIV-2 (HIV-D) reactive by WB remain so by LIA (4). The high cost of LIAs prohibits their extensive use in resource-limited areas, and the dot blot analysis is not commercially available, thereby restricting its wide-scale application. Preliminary studies conducted in Europe have suggested that a strategy using a combination of monospecific enzyme-linked immunosorbent assays (ELISAs) (CME) can be a reliable and cost-effective means for the type-specific serodiagnosis of HIV (1, 16, 17, 19). However, this strategy has not been comprehensively evaluated in field settings in West Africa, where both HIV-1 and HIV-2 are endemic and where dual HIV-1-HIV-2 infection has been documented (6, 14). In this study, we report the evaluation of the application of this strategy to a large panel of sera from Abidjan, Ivory Coast.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Assays. For the detection of HIV-1 antibodies, we used Wellcozyme HIV Recombinant (Murex Biotech Limited, Dartford, United Kingdom), a commercially available, HIV-1-monospecific, competitive ELISA that costs $1.50 per test and that has been extensively used in several countries. The assay was performed according to the manufacturer's instructions.

Supplemental testing. HIV-1 WB (Genelabs Diagnostic, Singapore, Singapore) and HIV-2 WB (Sanofi Diagnostics Pasteur, Marnes-la-Coquette, France) were used as confirmatory tests and were interpreted according to the Centers for Disease Control and Prevention criteria (3). In addition, the only three commercially available LIAs, which are capable of differentiating HIV-1 and HIV-2 antibodies, were used. Peptilav 1-2 (Sanofi Diagnostics Pasteur), which recognizes antibodies to the transmembrane glycoproteins of the two viruses (gp41 for HIV-1 and gp36 for HIV-2), and the INNO-LIA HIV confirmation test (Innogenetics, Ghent, Belgium), which uses the HIV-1 synthetic peptide antigens p17, p24, p31, gp41, and gp120 and the HIV-2 antigens gp36 and gp105 (20) were used. Lastly, a strip immunoassay (RIBA HIV-1/HIV-2 SIA), a research LIA kit from Chiron (Emeryville, Calif.), was also used (10). This assay incorporates recombinant HIV-1 antigens (p24, p31, gp120, and gp41) and HIV-2 antigens (p26 and envelope synthetic peptide). The cost of these assays is $19 per test for Peptilav and $23 per test for INNO-LIA.

Sera for retrospective evaluation of CME. A panel of 1,608 sera was selected from sera that had been screened for HIV-1 and HIV-2 antibodies either by Genelavia (Sanofi Diagnostics Pasteur) or by HIV-1 and HIV-2 ELISA from Genetic Systems (Seattle, Wash.). Of the 1,608 sera, 448 (27.9%) were those that had been previously found to be nonreactive by ELISA, Peptilav, and WB. The remaining 1,160 (72.1%) sera were reactive by both ELISA and Peptilav. Sera that reacted to the Peptilav HIV-1 band only were classified as HIV-1 antibody positive, those that reacted to the HIV-2 band only were scored as HIV-2 antibody positive, and those sera with at least weak reactivity to both the Peptilav HIV-1 and HIV-2 bands were classified as HIV-D seroreactive. On the basis of this Peptilav interpretation, 428 (26.6%) sera were HIV-1 antibody positive, 363 (22.6%) were HIV-2 antibody positive, and 369 (22.9%) were HIV-D antibody positive. These sera had been collected from different populations in Abidjan, including tuberculosis patients, blood donors, hospitalized patients, persons who voluntarily sought HIV testing, pregnant women, and female sex workers.

View larger version (20K): [in this window] [in a new window]   FIG. 1.   Proposed algorithm for the type-specific serodiagnosis of HIV-1 and HIV-2 infections by CME. All sera with known serologic status based on Peptilav were tested initially by an ELISA sensitive to both HIV-1 and HIV-2; reactive sera were then tested by monospecific ELISAs (Wellcozyme HIV Recombinant ELISA and ICE*-HIV-2 ELISA).

Sera for prospective evaluation of CME. To assess the reliability and cost-effectiveness of using CME in a routine setting, 821 sera collected consecutively between July and August 1996 from among different populations (pregnant women, female sex workers, tuberculosis patients, and persons who voluntarily sought HIV testing) were screened simultaneously in two HIV-1-HIV-2 combined ELISAs (Enzygnost Anti-HIV1/2 Plus [Behring Diagnostic, Marburg, Germany] and ICE* 1.0.2 [Murex Diagnostic, Dartford, United Kingdom]). Sera that were reactive in the two ELISAs were systematically tested by Peptilav and by CME by using the two monospecific ELISAs.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Retrospective evaluation of CME. Overall, 1,535 (95.5%) of 1,608 sera gave concordant results in the CME and Peptilav tests; 73 sera (4.5%) yielded discordant results, one of which was a confirmed HIV antibody-negative specimen that was reactive in CME (Table 1). Of the 428 sera that were HIV-1 antibody positive by Peptilav, all (100%) were scored as HIV-1 antibody positive by CME, reacting with high OD ratios in the Wellcozyme assay (mean OD ratio ± standard deviation [SD], 7.61 ± 4.37) and OD values well below the cutoff value in the ICE*-HIV-2 assay (mean OD ratio ± SD, 0.21 ± 0.03). Thus, the sensitivity of the Wellcozyme assay was 100% and the specificity of the ICE*-HIV-2 assay was 100% for specimens that were HIV-1 antibody positive by Peptilav. All 361 Peptilav-confirmed HIV-2 antibody-positive sera were strongly positive by the ICE*-HIV-2 assay (mean OD ratio ± SD, 11.19 ± 2.39). However, 45 (12.5%) of these sera were also weakly reactive (mean OD ratio ± SD, 1.36 ± 0.39; range, 1.00 to 2.56) in the Wellcozyme test and as such were classified as HIV-D antibody positive by CME. Of the 371 sera classified by Peptilav as HIV-D antibody positive, 344 (92.7%) reacted with high OD ratios in both the Wellcozyme and ICE*-HIV-2 assays (mean OD ratio ± SD, 7.32 ± 4.75 and 8.88 ± 4.07, respectively) and thus were classified as HIV-D antibody positive; 21 (5.7%) reacted in the Wellcozyme assay only and were categorized as HIV-1 antibody positive by CME; and 6 (1.6%) reacted only in the ICE*-HIV-2 test and were classified as HIV-2 only by the CME (Table 1).

View larger version (22K): [in this window] [in a new window]   FIG. 2.   Distribution of OD ratios of sera yielding discordant results by Peptilav and CME in the retrospective (A) and prospective (B) evaluations. Horizontally ruled lines represent means; dashed lines represent cutoff values (COV). WEL, Wellcozyme HIV Recombinant ELISA; ICE-2, ICE*-HIV-2 ELISA.

Prospective evaluation of CME. Of the 821 consecutive sera that were tested prospectively in our laboratory between July and August 1996, 267 (32.5%) specimens were reactive in the HIV-1-HIV-2 mixed-antigen ELISAs (Enzygnost and ICE* 1.0.2 assays) and were tested by both Peptilav and CME. On the basis of the results of the retrospective study, only sera giving OD ratios of more than 3.0 in the Wellcozyme test were considered reactive in the test. Of the 267 reactive sera, 215 (80.5%) gave concordant results by Peptilav and CME (Table 2). All 106 sera scored by Peptilav as HIV-1 antibody positive were strongly reactive in the Wellcozyme assay (mean OD ratio ± SD, 15.86 ± 4.97) and all were negative in the ICE*-HIV-2 assay (mean OD ratio ± SD, 0.21 ± 0.01) and were thus classified as HIV-1 seroreactive in CME (Table 2). All the 41 Peptilav-confirmed HIV-2 sera were positive in the ICE*-HIV-2 assay (mean OD ratio ± SD, 10.87 ± 2.19). However, one serum specimen gave an OD ratio of 4.92 in the Wellcozyme test (Fig. 2B, panel 1) and was therefore scored as HIV-D by the CME strategy.

Cost of type-specific serodiagnosis by the CME strategy. In the prospective evaluation, the cost of screening the 821 sera simultaneously by two HIV-1-HIV-2 combined ELISAs (Enzygnost Anti-HIV-1/2 and ICE* 1.0.2) was $1,970; that of the type-specific serodiagnosis of the 267 HIV antibody-positive sera was $908 for the CME strategy and $5,073 by the Peptilav-based strategy. Considering the cost of the two HIV-1-HIV-2 combined ELISAs and the type-specific strategies, a substantial cost savings in reagents in favor of the CME strategy (59%) was realized.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

Our results indicate that a testing strategy comprising two monospecific ELISAs can serotype HIV-1 and HIV-2 infections in HIV-seropositive persons more accurately and economically than a strategy based on a LIA assay. In both the retrospective and the prospective evaluation, the overall concordance between CME and Peptilav testing was high, and when these two strategies gave discordant results, the results of the supplemental testing agreed with the results of the CME strategy. The results of this field evaluation, based on a large panel of sera and in which we used commercially available monospecific HIV ELISAs, corroborate and extend the results from studies conducted in Europe (2, 16, 17). In a French study (17), 375 sera were tested by a competitive HIV-1 ELISA and an indirect HIV-2 ELISA, and all 49 dually seroreactive sera were correctly identified. Berry et al. (2) have reported that a highly sensitive and specific recombinant-based competitive in-house HIV-2 ELISA using a Gambian HIV-2 isolate was able to speciate most of the HIV-1 and HIV-2 antibody-positive sera when used in conjunction with a competitive commercial HIV-1 ELISA.

Using CME offers two main advantages. First, the cost of this approach is much lower than the prohibitive cost of LIAs. For instance, in the prospective evaluation, which represents 1 month's activity in our laboratory, compared with the cost of Peptilav testing, the CME strategy resulted in substantial cost savings (59%) for reagents to serotype specimens that are reactive on mixed HIV-1 and HIV-2 screening ELISAs. Second, since the determination of reactivity is based on a calculated cutoff value, the subjectivity associated with interpreting weakly reactive LIA bands is eliminated.

The lack of a "gold standard" for the confirmation of dual HIV-D reactivity is highlighted in our study, in which the three LIAs and the CME strategy gave conflicting results for the three groups of discordant sera (Fig. 2). These discrepancies observed with the different assays clearly highlight the difficulties of serologically discriminating between infections with two closely related retroviruses. Although the exact meaning of HIV-D serologic reactivity is not fully known, it may imply infection with one HIV type with cross-reactivity to the other, dual infection with both HIV types, infection by one HIV type and exposure to a second type of HIV, or infection with an intermediate virus (9). Moreover, even PCR cannot be used as a standard for diagnosis of HIV-D infection, because some persons with dual infections but low proviral loads have antibody responses but have proviral virus undetectable by PCR (18). In view of these limitations, how can the three groups of discordant CME and Peptilav results be interpreted? First, cross-reactivity of HIV-2 sera in HIV-1-monospecific competitive ELISAs (i.e., CME HIV-D, but Peptilav HIV-2) has been previously reported. Simon et al. (17) found that 151 confirmed HIV-2 sera cross-reacted in an HIV-1 competitive ELISA, leading the investigators to suggest that only sera with OD ratios of more than 5.0 in the competitive HIV-1 ELISA should be considered positive. However, our results (Fig. 2, panel 1) suggest that only sera reacting in the competitive HIV-1 ELISA with OD ratios of more than 3.0, instead of those with OD ratios of more than 1.0, should be considered positive by the CME strategy. A cutoff of 3.0 would increase the specificity and the positive predictive value of CME for the serodiagnosis of HIV-D infection. This arbitrary cutoff value is justified by our observation that all 47 sera that reacted weakly (OD ratio <3.0) in the HIV-1 test were reactive only to HIV-2 antigens by INNO-LIA testing. Moreover, these weakly cross-reactive sera were not indicative of HIV-1 group O infection, since none of them was positive in the research HIV-1 group O ELISA.

The second group of sera that gave discordant results consisted of those with an HIV-D profile in Peptilav but with weak HIV-1 band reactivity that was HIV-2 positive by CME. Based on the high analytic sensitivity to low-titer HIV-1 antibodies demonstrated by the end point dilution experiments and the performance of the Wellcozyme assay with a panel of sera representing the HIV-2-to-HIV-D seroconversion, it is unlikely that this profile represents the inability of the HIV-1 competitive ELISA to detect sera with low-titer antibodies or that it represents HIV-2-infected persons in an early phase of seroconversion to HIV-D infection. Rather, it more probably suggests false HIV-D reactivity in Peptilav.

The third group of discordant specimens consisted of those that were scored as HIV-D by Peptilav but scored as only HIV-1 by CME. Furthermore, the difference in the proportions of HIV-D samples as scored by Peptilav in the retrospective and the prospective evaluations was striking. Our results suggest that this discrepancy is due to antibody cross-reactivity in the Peptilav assay rather than a lack of sensitivity in the HIV-2-monospecific ELISA, since the ICE*-HIV-2 assay showed a 100% sensitivity with the 361 HIV-2 Peptilav-confirmed sera and demonstrated high analytic sensitivity in the dilution experiments. Thus, the high rate of dual reactivity is due partly to the lack of reproducibility of the Peptilav assay for HIV-D samples, since 39% (16 of 41) of the HIV-D samples were scored as HIV-1 only when retested with a different lot of Peptilav. This lack of reproducibility of the Peptilav assay, together with the finding that most of these samples gave predominantly HIV-1 results when tested by INNO-LIA and RIBA, further strengthens the hypothesis that the lack of concordance between Peptilav and CME is due to false HIV-D reactivity in Peptilav.

In summary, we have shown that using a combination of highly sensitive and specific commercially available monospecific ELISAs is a reliable and cost-effective strategy and thus a recommended strategy for type-specific serodiagnosis of HIV-1 and HIV-2 infections in HIV-seropositive persons in areas where both viruses are prevalent.