Comparison of Second- and Third-Generation Enzyme Immunoassays for Detecting Antibodies to Hepatitis C Virus

MATERIALS AND METHODS

We tested 1,134 serum samples collected during the 2nd year of a longitudinal community-based study in Assiut Governorate, Egypt (11), with both G2 and G3 EIAs from Abbott Laboratories (Wiesbaden, Delknheim, Germany) following the manufacturer's instructions and compared the tests' results. To further evaluate discordant G2/G3 assay results, we defined samples as truly positive or negative based on an algorithm representing the following elements (Fig. 1): (i) an in-house RT-PCR generated from the 5′-untranslated region of the HCV genome (1) performed with the same sample and additional samples from the same individual in the prior or subsequent years, (ii) G3 results from prior or subsequent years, and (iii) Micro-Particle Immunoassay (M-EIA; Abbott Laboratories) results for the discordant samples. The sensitivity and specificity of the G2 and G3 assays were initially calculated by assuming true serostatus according to the algorithm. To further support the assessment of sensitivity and specificity of the G2 and G3 assays, discordant samples were subsequently tested with the G3 RIBA (Chiron, Emeryville, Calif.). The sensitivity and specificity of the G2 and G3 assays were once again calculated on the basis of the RIBA, and the two methods of assessing sensitivity and specificity were compared. For all calculations of sensitivity and specificity, the samples with concordant G2/G3 assay results were included with the assumption that those results represented true positives and true negatives.

• Open in new tab
• Download powerpoint

FIG. 1.

Algorithm to assign serostatus (shaded boxes) of discordant samples on the basis of PCR, M-EIA, and testing of additional samples from the same individuals. S1, serum sample from the year before discordant samples; S3, serum sample from the year following discordant samples.

This study was reviewed and approved by the Institutional Review Board of the University of Maryland and the Egyptian Ministry of Health and Population.

RESULTS

Eighty-seven (7.7%) of the serum samples were positive by both the G2 and G3 EIAs, while 1,012 (89.2%) were negative by both tests, with a resultant concordance of 96.9% (Table 1). Among the 35 discordant G2/G3 EIA results, there were 17 specimens G2 assay negative and G3 assay positive, while 18 specimens were G2 assay positive and G3 assay negative. Results of M-EIA testing of the 35 discordant samples were more frequently in agreement with the G3 EIA result (concordance of 88.6% [31 of 35]) than with the G2 EIA result (11.4% [4 of 35]). Among the 18 samples with G2 assay-positive and G3 assay-negative results, 2 were M-EIA positive, while 15 of the 17 sera that were G3 assay positive and G2 assay negative were M-EIA positive (Table 2). The results of HCV RNA testing of the discordant samples also showed better agreement with the G3 EIA results. All 18 samples that tested positive by the G2 EIA, but negative by the G3 EIA, were negative by RT-PCR. However, 7 (41.2%) of the 17 individuals with G3 assay-positive and G2 assay-negative results had detectable HCV RNA in the same sample or in a sample from the same individual taken a year earlier.

Among the 35 discordant samples, RIBA was positive in 11 samples, negative in 13, and indeterminate in 11 (Table 3). G3 EIA results agreed with 87.5% (21 of 24) of interpretable RIBA results. Ten samples were positive by both RIBA and G3 EIA, and 11 were negative by both. Two samples were negative by RIBA, but positive by G3 EIA, while one sample was positive by RIBA and negative by G3 EIA. However, G2 EIA and RIBA results were concordant in only 3 of 24 (12.5%) samples. Excluding the 11 samples with indeterminate RIBA results, there was only one discordant result between RIBA and M-EIA (Table 4). This sample was positive by M-EIA and G3 EIA, but negative by RIBA. All samples with indeterminate RIBA results were PCR negative.

Based on the algorithm (Fig. 1) used to define “true” positive and negative serostatus, the sensitivity of the G3 EIA was 98.1% (95% confidence interval [CI], 92.6 to 99.7), compared to 85.6% (95% CI, 77.0 to 91.5) for the G2 EIA, while the specificity of the G3 EIA was 99.8% (95% CI, 99.2 to 99.9), compared to 98.4% (95% CI, 97.4 to 99.0) for the G2 EIA. The assessment on the basis of the RIBA, excluding samples with indeterminate RIBA results, was similar: the sensitivity of the G3 EIA was 99.0% (95% CI, 93.7 to 99.9), compared to 89.8% (95% CI, 81.6 to 94.7) for the G2 EIA, while the specificity of the G3 EIA was 99.8% (95% CI, 99.2 to 99.9), compared to 98.9% (95% CI, 98.0 to 99.4) for the G2 EIA.

DISCUSSION

This investigation with sera from over 1,100 subjects collected during a cross-sectional serosurvey in Egypt shows that the sensitivity and specificity of the G3 EIA are outstanding, a finding consistent with a recent review of predominantly small studies (6).

The G3 EIA clearly demonstrated sensitivity and specificity superior to those of the G2 EIA. For the subset of problematic, discordant samples, the M-EIA agreed with the G3 assay eight times as frequently as it did with the G2 assay.

The excellent agreement between the results of the G3 EIA and M-EIA is not surprising, since both include the NS5 epitope, which is not included in the G2 assay. Because of its lower sensitivity, surveillance data based on G2 EIA testing may result in a spuriously low prevalence of anti-HCV antibodies, especially in populations with relatively high prevalence of infection, in which false-positives are less of a concern.

The concordance between RIBA and M-EIA was almost complete for the samples with definitive RIBA results. RIBA results did not prove more helpful than M-EIA, because the calculations of G2 and G3 EIA reliability were not substantially changed when comparisons were made with RIBA results. Furthermore, one-third of the RIBA tests had indeterminate results, greatly limiting its usefulness for confirmation of EIA results, consistent with previous assessments (13).

In developing countries, where resources are scarce, use of supplemental tests (e.g., RIBA) is generally not fiscally possible. In this context, G3 EIA results appear sufficiently reliable for detecting exposure to HCV without supplemental assays. The decision to use these supplemental tests for clinical situations in developed countries, where resources are more readily available, must be based on the individual situation. It has been suggested that supplemental tests be used only in populations with expected low prevalence, such as blood banks (3, 10). RIBA appears to have no advantage over the less expensive M-EIA in supporting positive screening test results, and the crucial clinical information needed to determine who has chronic infection is provided by RT-PCR.