Performance of commercial dengue NS1 ELISA and molecular analysis of NS1 gene of dengue viruses obtained during surveillance in Indonesia

Dengue is the most important arthropod-borne viral diseases in humans with a large global burden. There are an estimated 50 million infections per year occurring across approximately 100 countries in tropical and sub-tropical regions in the world and potential for further spread. The disease affects approximately 2.5 billion people living in Southeast Asia, the Pacific, and the Americas [1, 2]. Dengue disease causes variable clinical manifestations, ranging from an undifferentiated fever and dengue fever to the more severe forms of the disease, Dengue Hemorrhagic Fever (DHF) and Dengue Shock Syndrome (DSS) [3].

Dengue disease is caused by dengue virus (DENV), a member of Flaviviridae family, with a substantial genetic diversity shown by the presence of four serotypes (DENV-1, -2, -3, and −4) and multiple genotypes (or subtypes) within each serotype [4, 5]. DENV is transmitted through human-mosquito cycle by Aedes aegypti and A. albopictus mosquito vectors. The genome consists of single-stranded positive-sense RNA which encodes three structural (C, prM/M, E) and seven non-structural proteins (NS1, NS2A, NS2B, NS3, NS4A, NS4B, NS5) [1].

With the absence of licensed vaccines or specific antiviral therapies for dengue, patient management relies on good supportive care. Prompt and early diagnosis of dengue viral infection remains crucial. Laboratory confirmation is important due to difficulties in making accurate diagnosis due to the broad spectrum of clinical presentations. Among the available dengue diagnostic tools, the detection of virus encoded NS1 antigen has become the basis for commercial diagnostic kits and laboratories are increasingly using NS1 detection as the preferred diagnostic test [2]. NS1 is a glycoprotein essential for viral replication and viability. Assays have been developed to diagnose DENV infections by detection of NS1 protein in blood during acute phase [6]. High level early viremia and NS1 antigenemia has also been associated with more severe clinical presentations [7].

The diagnostic accuracy of commercial diagnostic assays based on DENV NS1 antigen detection in plasma/serum samples have been described [6, 8, 9]. A multi-country evaluation study reported that the best performing NS1 assay had only a moderate sensitivity (median 64%, range 34-76%), with 100% specificity. The poor sensitivity of the evaluated assay has been related to study sites in different geographical regions suggesting the need for further assessment [10].

Indonesia is the largest archipelago country in the world with over 17,000 islands, inhabited by around 240 million people. The commercial NS1 antigen detection assays have been increasingly used and are becoming the tool of choice among clinicians to confirm DENV infection in Indonesia. However, only limited data on the performance of the assays for Indonesian samples is available. We report here the performance of Panbio Dengue Early NS1 ELISA and IgM ELISA diagnostic assays in the detection of DENV infection from serum samples collected during our dengue surveillance study conducted in eight cities across Indonesian archipelago in 2010–2012. We also analyzed the NS1 gene sequences and amino acid polymorphisms in 48 DENV isolates from Indonesia to determine their contribution to the sensitivity of the NS1 ELISA assay.

We report here the use of Panbio Early Dengue NS1 ELISA assay in detecting dengue infection during surveillance in Indonesia in 2010–2012. This report reflected the performance of the NS1 ELISA assay in surveillance/field setting. We also report here the Panbio Dengue Duo IgM ELISA performance as supporting information. Our evaluation was based on the molecular detection results as gold standard. We analyzed samples which were positive for dengue by RT-PCR and/or virus isolation followed by NS1 gene sequencing. On these samples, NS1 antigen detection results were recorded and the positivity of the samples on NS1 antigen was compared. We observed a number of NS1-negative samples (n = 82) which were positive by RT-PCR. In order to confirm the accuracy of the detection results on those NS1-negative samples, NS1 ELISA was repeated. Our results detected overall sensitivity of 56.4% in the NS1 assay used. This Early Dengue ELISA NS1 assay has been previously assessed for its performance [22]. Compared to previous reports, this sensitivity obtained in our study was rather low. During our surveillance study, we also used the Panbio Dengue Duo IgM & IgG ELISA. The IgM ELISA detected 140 IgM-positive samples out of 188 samples (sensitivity = 73.7%). Although the sensitivity was higher for IgM assay compared to NS1 assay, this did not necessarily indicate that IgM assay had better sensitivity since we did not compare the duration of illness. When both NS1 antigen and IgM-positive results were combined together, the detection rate/sensitivity increased to 89.4%. The increased sensitivity of dengue detection using combination of NS1 and IgM has been previously reported [23–25] and our result was in accordance with those reports.

The specificity of the NS1 assay in this study was 100% when tested in samples confirmed for non-dengue infection and from healthy individuals. Unfortunately, because the lack of samples confirmed for other flavivirus infection, we were not able to assess for its specificity against other flaviviruses. Previous study has described the specificity of this NS1 assay against non-dengue infection including other flaviviruses [10].

The NS1 sensitivity was varied if samples were grouped according to the geographical regions/cities of origin and serotypes (Table 2). Guzman and co-workers also observed the different sensitivity of NS1 assays according to the geographical region of the patients [10]. Indonesia is an archipelago country in which spatial barriers exists between islands. Our samples were collected from eight cities reside in six different major islands, thus the different sensitivity observed in various cities may reflect the circulation of various serotypes and genotypes of DENV in different geographical regions of Indonesia. The possible differences in the background immunity of people in each city may also contribute to the variable sensitivity of the NS1 and IgM assays. Although we ensured the uniformity in the sample collection method, it is possible that different sample handling and storage in each region may also become other contributing factors in regards to geographical regions. Also, we are aware that the sample number and DENV serotype distributions according to cities were not equal, and some cities were only represented by small sample number. This observation warrants further confirmation on the genetic diversity of DENV in Indonesia.

In term of sensitivity in relation with infection status, the NS1 antigen assay evaluated here showed higher sensitivity in primary infection (67.6%) compared to secondary infection (48.2%) (Table 2), supported by ANOVA test (p-value = 0.015, Table 3). This result is in accordance with previous reports that observed the decreased sensitivity of NS1 tests in secondary infection [26, 27]. It has been proposed that the reduced sensitivity of NS1 antigen detection in secondary dengue infection occurred because NS1 protein was sequestered in immune complexes when a substantial level of DENV-reactive IgG was present [28]. Therefore, our NS1 result confirmed the previous findings about the significant association of the NS1 detection and infection status. Unlike the NS1 antigen detection, our supporting data on the IgM detection observed higher sensitivity in secondary infection, which is in accordance with previous report [27]. This was also supported by the significant result of the ANOVA test (p-value < 0.001). Therefore, we believed that the sensitivities of both NS1 and IgM assays were affected by the infection status of patients.

The severities of the dengue patients recruited in this study were graded according to WHO-SEARO Guidelines 2011 classification. Most patients were DHF, followed by DF and DSS (Table 2). A recent report described that the circulating NS1 protein levels were higher in patients with DHF than patients with DF [29]. In our study, when the sensitivity of the NS1 assay was correlated with the severity, we observed quite similar rates of detection on all grades of dengue severity (Table 2). This data demonstrated that the performance of this NS1 assay was not influenced by disease severity. Our data also confirmed the previous report on NS1 assays evaluation, in which no correlation between sensitivity and disease severity was observed [10].

In general, our results did not indicate that the DENV serotypes contributed to the sensitivity of NS1 based on the ANOVA test. However, we observed very low sensitivity of NS1 in samples infected by DENV-4 (Table 2), and that DENV-4 serotype was indeed significant in influencing the NS1 sensitivity (Additional file 1: Table S1). When DENV-4 data was excluded, the overall NS1 sensitivity increased to 65.7% (data not shown). Previous studies suggested that sensitivity of NS1 test differed according to the serotype, and lower NS1 sensitivity was observed for DENV-4 [10, 30, 31]. Our result was in accordance with the studies described above, but in contrast with previous observation of the NS1 rapid test sensitivity in Cambodia, which was highly sensitive for DENV-4 [27]. It is possible that the different genotypes of DENV contribute to the detection limit of the NS1 assay. Indeed, based on molecular phylogenetic data, the DENV-4 viruses in Cambodia and surrounding countries such as Thailand and Vietnam were grouped into Genotype I [32], while Indonesian DENV-4 viruses were grouped into Genotype II, based on Lanciotti [33] classification (Sasmono et al., in preparation). Overall, we do not know exactly why this particular serotype has significantly low sensitivity toward NS1 assay. Unfortunately, we only have limited genomic information of the DENV-4 from our samples, hence we cannot provide better explanation. Other factors that may impact the sensitivity include different virus characteristics, higher rates of immune complex bound antigens [28], or specimen handling and/or transport differences.

The low sensitivity of NS1 antigen detection in our samples prompted us to perform NS1 gene sequence analysis. This was carried out to further investigate the correlation between DENV genetic diversity and the NS1 ELISA sensitivity, as well as to study the genetic variation and selection pressure of the NS1 gene. We successfully sequenced 48 NS1 genes and aligned them according to their serotypes. The most polymorphic NS1 gene was observed for DENV-1, followed by DENV-3, DENV-2 and DENV-4 (Table 4). We are aware that the small sample number of DENV-2 and DENV-4 may contribute to the low polymorphism observed for these serotypes.

The different avidity of the anti-NS1 antibody used in the assay against epitopes of different serotypes/genotypes may also potentially be related to the assay’s sensitivity. The Panbio NS1 ELISA assay is designed to detect the NS1 antigen from all serotypes, in which the monoclonal antibody (MAb) that was used detected common epitope shared by all serotypes. Previous studies mapped the anti-NS1 MAb that cross-reacted with all serotypes into common NS1 gene epitope (¹¹¹LRYSWKTWGKA¹²¹) [34, 35]. When we aligned the NS1 AA sequences of the 48 Indonesian DENV isolates with the above epitope, we observed the presence of 1 to 4 AA mismatches in all isolates, with most substitutions occurred in AA 111–112 and 117 (Additional file 2: Figure S1). However, since these mismatches were shared by samples with both positive and negative NS1 antigen results, these mismatches were unlikely to influence the sensitivity of NS1 detection assay.

Further alignment of the Indonesian samples with other proposed common epitope regions of NS1 reported by Masrinoul et al. [36] showed that although the regions had limited polymorphisms compared to the common epitope sequences, the polymorphisms were retained on all of the samples (Additional file 2: Figure S1). The polymorphisms of these epitope regions were most likely not involved in the sensitivity of NS1 ELISA assay as the same substitutions were occurred on majority of the samples with either NS1 antigen negativity or positivity.

We observed 8 polymorphic positions that only occurred in NS1 antigen negative samples. From those positions, only one mutation (AA 168R) from a DENV-3 sample resided within the proposed epitope region of AA 141 – 168. However, we were not able to conclude whether mutations that occurred in those polymorphic positions were indeed affecting the results of NS1 sensitivity as those mutations were singletons.

We analyzed the genetic relationships of the Indonesia DENV based on the NS1 gene sequences. Other than the commonly used E protein gene, the NS1 protein gene has been used in the phylogenetic analysis to determine the genotype of DENV [37]. The non-structural (NS) proteins themselves may play an important role in DENV evolution, particularly NS1, NS2A and NS4B proteins [38]. Our phylogenetic analysis observed the clustering of the DENV-1 isolates into two groups with one isolate as an out-group (Figure 2A). This grouping was in accordance with grouping based on E protein gene sequences classified by Goncalvez et al. [39] (Sasmono et al., in preparation). Therefore, it is clear that during this study, multiple genotypes of DENV-1 were circulating in Indonesia. The predominant genotypes were the Genotype I and IV (Sasmono et al., in preparation). We also recently discovered an isolate belonging to the old genotype II of DENV-1 in Semarang [40], one of the cities involved in this study. The patient infected by this particular strain (SMG-SE003/2012) was negative for NS1 ELISA but positive for dengue IgM, IgG, and RT-PCR detection [40]. The phylogenetic analysis grouped all of DENV-2 isolates together in one group. Similarly, DENV-3 isolates were also clustered as a single group. The small number of DENV-4 isolates sequenced in this study was not sufficient for the phylogenetic analysis. However, based on the E protein sequence analysis, so far we only discovered a single genotype of DENV-4 circulating in Indonesia (i.e. genotype II based on Lanciotti’s classification [33]) (Sasmono et al., in preparation).

Another aspect that may affect the varied sensitivity on NS1 detection is the different magnitude of NS1 secretion which appeared to be strain dependent, as recently reported [41]. Indeed, our preliminary study detected different levels of NS1 expression in various strains of DENV in vitro (data not shown). Therefore, this phenomenon may also contribute to the varied sensitivities of the NS1 detections.

We did not detect any evidence of substantial positive selection in the evolution of NS1 in all serotypes, as the majority of sites were under strong negative or constraining selection. A previous study also observed that there was no evidence of positive selection in the structural proteins of DENV-2 including the NS1 protein [29]. This indicates that NS1 needs to retain its structure and amino acid sequences to be able to perform its function effectively.

In summary, we observed the low sensitivity of Panbio Early Dengue NS1 ELISA in detecting dengue infection in samples collected during our surveillance in Indonesia. The combination of both NS1 and IgM ELISAs increased the detection rate. We concluded that infection status contributed to the sensitivities of both NS1 and IgM assays. Geographical regions, which might reflect on the differences of sample handling and proportion on dengue virus serotype/genotype, were also contributed to the NS1 assay results. Although in general the serotypes were not influential in the sensitivity of NS1 assay, our data suggested that DENV-4 serotype was associated with the low sensitivity of the assay. Our genetic analysis of the NS1 genes of DENV isolates from Indonesia revealed that the observed polymorphisms of NS1 genes in Indonesia were less likely to contribute to the sensitivity of NS1 ELISA. The performance of dengue diagnostic assays should be continually assessed to ensure their accuracies in detecting dengue infection, especially in different geographical regions and infection status background which is eminent in dengue endemic countries.