The family Flaviviridae, genus Flavivirus, holds many of the world’s most prevalent arboviral diseases that are also considered the most important travel related arboviral infections.[1] As the geographic distribution and symptoms caused by these viruses overlap, detection requires differential diagnostic algorithms that include multiple flaviviruses.[2] Increase in travel expands the need for rapid and reliable multiplex diagnostic tests in non-endemic countries to supplement currently used methods.[3, 4]

Flaviviruses are single stranded enveloped viruses with an RNA genome of about 11 kb length. The genome is composed of three structural (Envelope, Capsid and Precursor-membrane) and seven non-structural proteins (NS1, NS2A, NS2B, NS3, NS4A, NS4B and NS5).[5] Diagnosis is primarily based on serology through detection of IgM and IgG antibodies, as viremia typically has been reduced to undetectable levels when symptoms set in and patients seek medical attention.[5–7]

The genus is divided into serocomplexes that are distinguished based on neutralizing antibody reactivity (Fig. 1). The amino acid homology of the envelope (E) protein (which is the immunodominant antigen for neutralizing antibody assays) ranges from 40–50% between serocomplexes and 70–80% for virus species within a serocomplex.[5, 8] Antibodies to flaviviruses are known to cross-react extensively within, and to a certain extent between, serocomplexes when using traditional antibody assays.[9–12] Cross-reactivity occurs also if patients have been vaccinated against flaviviruses such as yellow fever virus (YFV), tick-borne encephalitis virus (TBEV) and/or Japanese encephalitis virus (JEV) or after secondary infection with a different flavivirus.[9, 10, 13]

To overcome flavivirus cross-reactivity in diagnostics the use of recombinant antigens in ELISA is to be preferred over whole virus as it increases specificity.[14–16] Envelope, pre M and NS1 recombinant proteins are the most commonly used.[14–16] Of these, the NS1 has shown to be highly immunogenic and important in the development of non-neutralizing protective antibodies.[17, 18] NS1 is thought to contain more species specific epitopes than the envelope protein, although some cross-reactivity is seen between NS1 proteins.[19–21] NS1 in its natural conformation is thought to elicit a more specific immune response.[22, 23] The absence of NS1 proteins in inactivated JEV vaccines offers further potential for serological diagnosis through allowing differentiation between vaccinated and infected patients.[24] Thus, NS1 protein shows potential to use in serological differentiation between flavivirus infections.[25, 26]

To enable fast, syndrome based laboratory testing that focuses on multiple rather than individual viruses, we developed a protein microarray, using recombinant NS1 proteins, as a serological test for medically important viruses within the Flavivirus genus.

We developed a first generation protein microarray for rapid, multiplex and virus-specific IgM and IgG tests for diagnosis of flavivirus infections that could differentiate between virus species, even within flavivirus serocomplexes, as well as between vaccinated and infected individuals, with the exception of YFV vaccinated individuals. Initial validation shows that this microarray can be used for flavivirus surveillance in travelers and potentially in regions with co-circulation of multiple flaviviruses. To establish this we first investigated the sensitivity of NS1 antigens to their homologous sera on a protein microarray. The results show good IgG and IgM sensitivity for both JEV and DENV serocomplex viruses. The sensitivity is comparable to current IFA and ELISA commercial kits. However, our sample selection was tested with a multitude of standard serological assays. Comparing our results to commercial ELISA or IFA kits should therefore be done with caution. Our results confirm that NS1 provides a good sensitive and specific antigen tool for serological diagnosis.

For DENV, the sensitivity of the IgM assay was lower than for IgG. There are several possible explanations for this finding. First, lower DENV IgM sensitivity may have resulted because five to ten days post onset of symptoms may be too early for detecting seroconversion in serum samples.[32] Evaluations of commercial flavivirus IgM diagnostics kits showed varying sensitivities ranging from 58% to 98%, partially due to sample timing.[32–34] Five days post onset of symptoms, around 50–80% of patients on average have detectable IgM antibodies. This increases to 99% 10 days post onset of symptoms.[35]

Second, a study of antibody reactivity to individual DENV proteins has found that the measured mean OD is lower for NS1 compared to E-protein with many samples clustering closer to the cut-off compared to E-protein antibodies.[18] This makes NS1 antibody detection more susceptible to timing of sample taking and detection limit of test used.

Third, patient-to-patient variation in antibody responses to individual viral antigens may cause discrepancies of test results. NS1 based protein assays could potentially pick up infections missed by prM-E based front-line serological tests.[20] Antibodies to envelope protein were detected in 91% of the DENV cases while NS1 antibodies were detected in 99%, indicating that NS1 has a higher sensitivity.[18] Two of the nine DENV IgM positive samples from travelers that tested negative on DENV IgG ELISA, which is based on the prM-E antigen, tested IgG positive on our microarray, further supporting this assumption.

Finally, a number of the DENV cases may have been secondary, tertiary or quaternary infections. Several of the samples originated from DENV hyperendemic countries and showed high titers to multiple DENV and other flaviviruses, further supporting this assumption. Literature shows that IgM antibody titers produced during a secondary DENV infection are absent or lower and produced during a shorter period. This reduces the sensitivity of IgM antibodies in non-primary infections significantly.[35] Despite the above limitations, the NS1-based protein microarray performed well. Ideally, further validation is needed with sequentially sampled patient sera, allowing evaluation optimization of timing of sampling during infection, in relation to cut-off chosen.

The biggest challenge of current flavivirus serodiagnostics is virus specific differentiation.[3, 4] We analyzed the signals produced towards all antigens per sample to test our protein microarray’s capability to achieve this. With three exceptions (probable secondary infections), clear differentiation of antibody responses to the homologous antigen (defined as the virus the patient was confirmed infected with) was found for most patients. This can be clearly seen in the heatmap (Fig. 6). This is a big advantage over currently available serological tests, for which particularly flavivirus vaccination causes extensive cross-reactivity, mainly for IgG.[9–12] Previous epitope analysis of DENV NS1 and envelope protein indicated that NS1 has more virus specific epitopes and thus could be used for more specific serological tests.[16, 19–22] Our results support this showing good specificity for the NS1-based protein microarray.

The cause of the outliers shown in Fig. 5 is unclear as they may be the result of cross-reactive antibodies or previous exposure. Cross-reactive antibodies to NS1 proteins have been detected in previous studies, but why they are seen in some patients and not others remains unclear.[20] The reactivity to multiple antigens more likely reflects differences in exposure history.[10, 36] This assumption is supported by the fact that most patients with reactivity to multiple antigens originated from countries where multiple flaviviruses are endemic (Vietnam and Venezuela). The IgG titers of these patients are log multiplications higher than singular reactive titers and their corresponding IgM titers, which is highly indicative for secondary and frequent flavivirus infections.[16] Distinguishing primary from secondary or multiple flavivirus infections is serologically difficult.[16] Further investigation into patients with known multiple flavivirus infections will be needed to further define the uses of our multiplex array testing in such patient populations.

While we show excellent discrimination in antibody responses to viruses within the JEV complex, this is less straightforward for DENV. The protein microarray IgG antibody reactivity to individual DENV NS1 antigens shows capability to distinguish serotypes (as can be seen in Fig. 6), but this is not seen to the same extent for IgM antibodies. This was surprising as flavivirus IgM envelope antibodies are thought to be more specific than IgG antibodies.[16] However, to what extent this can also be said for NS1 is unclear. Here, future work will focus on more extensive evaluation, for which well characterized patient cohorts are needed.

Finally, we looked at the ability to rule out false positive reactions due to vaccination. This was not possible for YFV vaccination when testing against YFV NS1. YFV vaccine is a live attenuated vaccine and causes mild infection resulting in NS1 antigen and antibody production.[37] We do, however, show that when using NS1 antigens the cross-reactivity between the YFV serocomplex and other serocomplexes is absent.[9, 10, 13] JEV vaccine currently on the European and American market is an inactivated whole virus vaccine based on JEV virions.[38] This production technique makes NS1 a good target for vaccination versus infection differentiation and is the basis for current surveillance programs in Japan and surrounding countries.[39] However, live-attenuated (non-recombinant) JEV SA-14–14–2 vaccine is currently in use in a number of countries in Asia and has recently been WHO prequalified. Its use nullifies the ability to differentiate vaccination from infection, reducing public health surveillance options. New vaccine methods using chimeric virus vaccines are in development and might provide new opportunities for vaccination and infection differentiation through NS1. In veterinary vaccine development, good practice ensures the ability for such discrimination through the development of marker vaccines according to the DIVA (Differentiating Infected from Vaccinated Animals) principle.[40] Public health challenges associated with inability to reliably provide patient diagnostics in some instances due to vaccination, highlight the need for introducing this principle to the human vaccine market.