Use of Recombinant Envelope Proteins for Serological Diagnosis of Dengue Virus Infection in an Immunochromatographic Assay

RESULTS

The sensitivities obtained with the PanBio rapid test and the AFRIMS ELISA are summarized in Table 1. Of the 46 primary cases, 42 (91%) were detected via a positive IgM test line in the rapid test compared to 43 (93%) in the IgM ELISA. Thirty-one (86%) of the 36 secondary dengue cases were detected using the IgG rapid test line compared to 25 (69%) in the IgG ELISA. Of the secondary sera, 53% (19 of 36) were IgM positive in the rapid test compared to 50% (18 of 36) in the AFRIMS IgM ELISA. The overall sensitivity for dengue virus infections in the rapid test was not significantly different from that achieved using the ELISA (90 versus 87%, respectively; P = 0.4022).

Paired serum analysis revealed that the rapid test detected 18 of 20 (90%) primary and 14 of 16 (88%) secondary infections using the acute sample (S1 [Table 2]), with the ELISA detecting 19 of 20 (95%) primary and 8 of 16 (50%) secondary infections. Analysis of the discharge (S2) sera revealed that the rapid test detected 20 of 20 primary infections and 15 of 16 (94%) secondary infections (Table 2), while the ELISA also detected all primary infections and 14 of 16 (88%) secondary infections. Of the two triplicate samples taken from patients with primary dengue virus infection, one showed seroconversion with the second and third samples, while the other patient had detectable levels of IgM in all three samples. All four samples taken from the one patient with secondary dengue virus infection produced positive IgG results in the rapid test. Consequently, the rapid test detected 38 of 40 (95%) primary infections and 29 of 32 (91%) secondary infections compared to the ELISA, which detected 39 of 40 (98%) primary infections and 18 of 32 (56%) secondary infections.

The overall specificity of the rapid test (Table3) in nonflavivirus infections was 85% (76 of 89), with false-positive results occurring with sera from patients with NEI (n = 5), malaria (n = 2), leptospirosis (n = 5), and scrub typhus (n = 1) infections. The majority (12 of 13) of the false-positive results occurred at the IgM test line. No cross-reactivity with JE occurred at the IgM test line, with only two samples giving a positive IgG test line (89% specificity). The specificity was 89% for the IgM test line and 97% for the IgG test line, resulting in an overall specificity of 86% for the rapid test.

The individual results of the AFRIMS ELISA were well correlated using either the IgM or IgG tests (Fig. 2). There was a significant association between the mean ELISA ratio and the rapid score (analysis of variance, P < 0.0001 for IgM and P < 0.0001 for IgG). However, in the IgM and IgG assays it was apparent that a number of ELISA-negative samples were positive in the rapid test. The majority of discrepant sera gave low positive results in the rapid test, indicating that the cutoffs used in this assay are slightly lower than those used in the ELISAs.

• Open in new tab
• Download powerpoint

Fig. 2.

Correlation between the ELISA and the rapid test for IgM (a) and IgG (b). The broken lines represent the cutoff units used in the ELISA. The horizontal bars represent the mean ELISA ratio for each rapid score.

DISCUSSION

Due to the higher mortality associated with secondary infections (15), it is important to use diagnostic assays that are able to differentiate between the two forms of dengue virus infection. In endemic areas, secondary infections are most common, as the majority of children have antibodies against dengue by the time they are 5 years of age (18, 30, 36). As primary and secondary dengue virus infections show markedly different immunological responses (13), the detection of antibodies is a valuable procedure to diagnose and differentiate dengue virus infections. Traditionally, the HAI assay has been used as the “gold standard” serological test (39), though ELISA has recently been adopted in most laboratories due to its convenience and performance (13).

To further simplify diagnosis, a number of commercial lateral flow tests for the serological diagnosis of dengue virus infections are now available (from PanBio Ltd.; Cortez Diagnostics, Calabasas, Calif.; Glysby, Arcore, Italy; and AmeriTek, Seattle, Wash.). This format offers the benefit of short incubation times (5 to 30 min) and the ability to be used in field settings or in laboratories without adequate equipment or electricity. The PanBio Dengue Duo Rapid Strip Test is the first commercial lateral flow test for the diagnosis of dengue that incorporates recombinant proteins. The four proteins used (from dengue viruses 1, 2, 3, and 4) each represent the N-terminal 80% of the respective envelope glycoprotein. The expression and secretion of these recombinant subunits in the S2 cells has resulted in molecules that have retained native-like characteristics.

In the PanBio Rapid Test, IgM and IgG are detected simultaneously using a single addition of diluted serum. Therefore, differentiation between primary and secondary infections can be made through a single application of diluted serum rather than a series of dilutions, as is needed in the HAI assay (6). The sensitivity of this rapid test has been set so that in patients with primary dengue, IgM is positive while IgG is negative. In contrast, patients with secondary infections will have a positive IgG result with or without a positive IgM result.

In-house comparison studies between the whole virus and recombinant antigens in an IgM and IgG capture ELISA show a high level of agreement (94% for both). The data obtained suggest that these proteins should be useful in diagnostic assays for the detection of IgM and IgG to dengue virus (data not shown). In this study, the rapid test performed similarly to the in-house reference ELISA. The rapid test detected 91% of primary and 86% of secondary dengue cases compared to 93 and 69%, respectively, using the reference ELISA. There was no significant difference between the sensitivities in primary dengue virus infections (P = 1.0000), though there was a significant difference in the sensitivities in secondary dengue virus infections (P = 0.0063).

No significant differences were reported between the abilities of the rapid test and the reference ELISA to detect primary infections using the acute sample only (90 versus 85%, respectively; P= 1.0000). Based on these results, the rapid test was able to diagnose 90% of all primary cases at hospital admission. Eighty-eight percent of secondary cases were also diagnosed using only the acute sample in the rapid test. Consequently, 89% of all dengue patients were diagnosed using the acute sample in the rapid test, indicating that a second sample was necessary in only 11% of cases. A laboratory diagnosis of acute dengue virus infection at the time of admission will refine the differential diagnosis. However, a negative result early in the course of illness requires repeat testing, as does a diagnosis of secondary dengue virus infection if it is important to know the antibody response pattern.

There was no significant difference between the sensitivities when either the acute- or convalescent-phase sample was used for the detection of IgM in primary dengue (P = 0.4872) or IgG in secondary dengue (P = 1.0000). The only significant difference observed between the first and second samples was the sensitivity of IgM in secondary dengue (P = 0.0010). A sensitivity of 25% was achieved using the first sample, and this rose to 88% using the second sample. This supports theories stating that the IgM in secondary dengue is generally produced at lower levels and can take longer to reach detectable levels than IgM in primary infections (18, 30, 36).

As the recombinant proteins include 80% of the envelope protein, it is likely that they have a level of cross-reactivity similar to that of the whole virus. In this study, we found there were false positives with malaria, leptospirosis, and scrub typhus sera at the IgM test line, with the IgG test line being very specific. This pattern is different from that generally observed with whole virus (24, 31,37, 38).

The only cross-reactivity seen at the IgG test line in the rapid test was with sera from patients infected with JE virus. Consequently, this rapid test would be useful in areas where dengue and JE are endemic, as a positive IgM result in this assay is not likely to be due to an active JE virus infection. Cross-reactivity between flaviviruses at the IgG level is common and has been reported in diagnostic assays (7, 8, 20, 24).

The recombinant proteins were chosen because they represent the majority of the active section of the envelope protein, the target for neutralizing epitopes (16), and hence are more likely to mimic whole virus in serological assays. In addition, the envelope protein has been reported to elicit the first immune response in primary infections (5). As this response is also long lasting, the inclusion of recombinant glycoproteins in a dengue assay is a logical choice.

Several groups have expressed recombinant dengue envelope subunits as soluble secreted proteins using a variety of recombinant expression systems (9, 10, 23, 25, 28, 34). These are claimed to elicit neutralizing and HAI antibodies (11) and to induce a stronger and longer-lasting immune response in animals (29). These efforts to express soluble dengue envelope subunits have met with various degrees of success. In each case, efforts to establish that the expressed subunits have maintained native-like structure were made; however, it is difficult to assess the relative quality of one subunit versus another based on the in vitro assays used. The ability of the envelope subunits to stimulate the production of neutralizing antibodies and to provide a protective response when used to immunize mice or monkeys has been the key to assessing the quality of recombinant dengue envelope proteins. TheEscherichia coli-expressed E proteins (25), while reactive to appropriate monoclonal antibodies, failed to provide a protective response in a mouse challenge assay. The baculovirus-expressed E proteins (9, 10, 28) have resulted in various levels of neutralizing antibodies and protection in both mice and monkeys, indicating that these insect cell-produced subunits have a more native-like conformation. In similar studies with the S2-expressed 80% E subunits, the induction of levels of neutralizing antibodies and protection against challenge has been achieved in both mice and monkeys (B.-A. G. Coller, D. E. Clements, G. S. Bignami, I. D. Peters, J. R. Putnak, M. McDonell, and T. Humphreys, unpublished data). These results indicate that the S2-expressed 80% E subunits have maintained a native-like conformation. This study also confirmed that the 80% E subunit proteins expressed in the S2 expression system have native-like characteristics based on the ability of these recombinant subunits to produce sensitivities similar to those of whole-virus antigens in the assays described.

We have presented our results on the performance of the PanBio Duo Rapid Test using these recombinant dengue proteins and demonstrated a high level of sensitivity and specificity in diagnosing acute primary and secondary dengue. This test offers a safe alternative to diagnostic assays using whole-virus antigens and will provide the clinician with a reliable and rapid assay to distinguish acute dengue virus infections and identify secondary dengue virus infections and thus patients at risk for severe dengue. Due to these advantages, this rapid test is likely to be the first of the next generation of diagnostic assays for dengue.