Samples were submitted to the National Microbiology Laboratory of the Public Health Agency of Canada (Winnipeg, Manitoba, Canada) for arbovirus diagnostic testing. All samples were obtained from Canadian travelers who visited areas with known Zika virus transmission and consulted their physicians after symptoms consistent with Zika virus infection developed upon return.

We obtained deidentified samples from 75 patients. Thirty samples were from patients with serologically confirmed Zika virus infections; 10 from patients with confirmed Zika virus infections identified by 2-target real-time reverse transcription PCR (RT-PCR); 10 from patients with suspected Zika virus infections, which were subsequently identified as DENV infections; and 25 acute-phase samples from flavivirus-negative persons tested by Zika virus RT-PCR and MAC-ELISA. Primary Zika virus diagnostic testing for all samples was conducted by using an in-house CDC-based MAC-ELISA and subsequent confirmation of Zika virus infection by cross-PRNTs for Zika virus and DENV, or molecular assays as described (5).

We evaluated 5 Zika virus immunoassays in this study. We tested a conventional IgM ELISA (EI 2668-9601 M; Euroimmun AG, Luebeck, Germany) and 3 MAC-ELISAs: Zika Virus Detect (InBios International Inc., Seattle, WA, USA); Ab213327 (Abcam, Cambridge, UK); and NovaLisa ZVM0790 (Novatec Inc., Baltimore, MD, USA). On the basis of preliminary testing, we also tested the Euroimmun IgM ELISA in parallel with the Euroimmun conventional Zika virus IgG ELISA (EI 2668-9601 G). Both Euroimmun assays use recombinant Zika virus nonstructural protein 1 as the antigen; the InBios Zika Virus Detect uses a recombinant Zika virus envelope glycoprotein as the positive antigen, an unspecified cross-reactive control, and reference cell antigens; and the Novatec and Abcam ELISAs use an unspecified Zika virus antigen.

Most tests evaluated provided algorithms that resulted in positive, negative, or equivocal results. However, the InBios kits account for antigenics associated with secondary flavivirus infections and reports results as Zika virus positive, possible Zika virus positive, or presumptive other flavivirus positive or negative on the basis of calculations of optical density ratios obtained from a sample with the 3 different antigens. Two independent laboratory technicians blindly evaluated the 5 assays by using the panel outlined, according to the manufacturer’s instructions. Comparisons and performance calculations were conducted by the Quality Control Office of the National Microbiology Laboratory.

The assays generally showed reproducible results during independent evaluations, although specificity and sensitivity of each varied (Table 1). The Euroimmun IgM and IgG ELISAs and the Abcam IgM ELISA showed a specificity of 100% for negative specimens with similar results (>90%) for confirmed DENV-positive samples (Table 2). The NovaTec ELISA showed a specificity of 66% for negative specimens and 70% for DENV-positive specimens. Although the InBios Zika Detect ELISA showed similar specificity results for flavivirus-seronegative specimens, it showed decreased specificity for DENV-positive samples. This assay incorrectly identified 40% of these samples as Zika virus IgM positive and 40% as possible Zika virus positive.

Although specificity is a key factor, for a front-line diagnostic test, sensitivity is a major factor in determining its usefulness. With appropriate diagnostic testing in place, including use of Zika virus conformational cross-PRNTs, false-positive results caused by specificity issues can usually be overcome. However, poor sensitivity will lead to false-negative results that might not be followed up by testing of additional sample collections. When compared with the in-house diagnostics (MAC-ELISA with PRNT confirmation), the IgM assays of Euroimmun, Abcam, and Novatec demonstrated sensitivities of 37%, 57%, and 65%, respectively. When we combined results of the Euroimmun IgM and IgG ELISAs, sensitivity increased to 82%. The InBios Zika Virus Detect IgM assay correctly identified all confirmed Zika IgM-positive samples identified by the recommended diagnostic assays, resulting in a sensitivity of 100%. The InBios ELISA also detected IgM in 50% of samples that were positive for Zika virus by RT-PCR, whereas the other assays did not detect IgM in most of these samples (Table 3).

The low sensitivity of most immunoassays evaluated could be improved by testing a repeat sample collected a few weeks after the initial specimen, although this sampling is not always practical, particularly if resources are limited. When performed in combination, the Euroimmun Zika Virus IgM and IgG ELISAs provide improved sensitivity. However, interpretation of recent versus past infections could be problematic, particularly when IgM results are negative and IgG results are positive. On the basis of our findings, the InBios Zika Virus Detect MAC-ELISA provides diagnostic results comparable to the CDC-based in-house MAC-ELISA for specimens collected from patients with primary flavivirus exposures (i.e., no detectable background immunity to DENV). A needed follow-up to our study will be further evaluation of IgM detection by commercial ELISAs involving cases of secondary flavivirus exposures or previous immunization to related viruses, such as yellow fever virus.