Simultaneous Cocirculation of 2 Genotypes of Dengue Virus Serotype 3 Causing a Large Outbreak in Sri Lanka in 2023

Abstract

Background

We observed a discrepancy between dengue NS1 antigen test and molecular diagnostics, with the emergence of dengue virus (DENV) serotype 3 in Sri Lanka, and sought to understand the cause for the rise in cases and high failure rates of molecular diagnostics.

Methods

Whole-genome sequencing was carried out in 22 DENV-3 samples. Phylogenetic and molecular clock analyses were done for genotype assignment and to understand the rate of evolution. Mutation analysis was done to understand the reasons for polymerase chain reaction (PCR) nondetection.

Results

We identified 2 DENV-3 genotypes (I and III) cocirculating. DENV-3 genotype III strains shared a common ancestor with a sequence from India collected in 2022, while DENV-3 genotype I, was found to share a common ancestor with DENV-3 sequences from China. DENV-3 genotype III was detected by the modified Centers for Disease Control and Prevention DENV-3 primers, whereas genotype I evaded detection due to key mutations at forward and reverse primer binding sites. We identified point mutations C744T and A756G in the forward primer binding sites and G795A in the reverse primer binding sites, which were not identified in DENV-3 genotype III. Furthermore, our Sri Lankan DENV-3 strains demonstrated a high root to tip ratio compared to the previous DENV-3 sequences, indicating a high mutation rate during the time of sampling (2017 to 2023).

Conclusions

The cocirculation of multiple genotypes associated with an increase in cases highlights the importance of continuous surveillance of DENVs to identify mutations resulting in nondetection by diagnostics and differences in virulence.

Dengue is one of the most rapidly emerging mosquito-borne viral infections globally, with significant increases in morbidity and mortality [1]. An emergency health warning was issued by the World Health Organization (WHO) in 2024 due to the exponential rise in cases [2]. The WHO named dengue as one of the top 10 threats to global health in 2019 [3]. However, during the coronavirus disease 2019 (COVID-19) pandemic, many countries observed a decrease in dengue incidence, reporting low to moderate levels of transmission [4]. With the cessation of the global health emergency for COVID-19, the incidence of dengue has markedly increased, attributed to climate change, fragile health systems, political, economic, and humanitarian crises, unplanned urbanization, and population movement [5–7].

Sri Lanka has experienced regular outbreaks of dengue since 1989, with the number of cases gradually increasing over the years and spreading into different geographical locations within the country [8]. The largest outbreak due to dengue was reported in 2017 with 186 101 cases, associated with the cosmopolitan strain, the most extensively spread nonsylvatic genotype of the dengue virus (DENV) serotype 2 [4, 9, 10]. DENV-2 continued to be the predominant circulating serotype until October 2019. DENV-3 began to emerge towards the end of October, with an increase in the number of dengue cases, and a total of 105 049 cases reported [4]. Of these cases, 20 718 (one-fifth) were reported from Colombo, which is usually the district with the highest incidence. However, with the onset of the COVID-19 pandemic, the number of dengue cases decreased drastically in 2020 (4257 reported in Colombo) and in 2021 (11 401 reported in Colombo) [11]. During these 2 years, DENV-2 accounted for the majority of cases (46%), followed by DENV-3 (27.9%) and DENV-1 (24.1%) [4]. In Sri Lanka, the number of cases began to gradually increase from June 2022 onwards with a total number of 89 799 cases reported in 2023, with 18 650 from Colombo [12]. Usually, Sri Lanka has 2 seasons of intensified dengue activity coinciding with the monsoon seasons [8, 12]. One season typically spans November to early February and the second season runs from May to July. However, 2023 saw an unusual pattern, because outbreak-level transmission persisted from November 2022 until July 2023 without intermonsoon respite.

As we have been carrying out dengue surveillance activities in Colombo, Sri Lanka for many years, we tracked these dynamics and noted an unusual discrepancy between nonstructural protein 1 (NS1) antigen test results and real-time polymerase chain reaction (RT-PCR) results for serum samples from patients with a suspected acute dengue infection. Our surveillance activities show that the outbreak coincided with the emergence of DENV-3 serotype as the predominant circulating serotype in Colombo. The aim of this study was to understand the unusual DENV-3 outbreak pattern as well as the discrepancy between point-of-care diagnostics and molecular diagnostics. To this end, we sequenced virus-positive samples to identify the serotype of infecting DENVs along with other genetic information.

METHODS

Patients

From June 2022 to November 2023, we tested serum samples of 210 patients with a suspected acute dengue infection admitted to the National Institute of Infectious Diseases, Colombo, following informed written consent. All samples were collected from patients presenting with symptoms suggestive of an acute dengue infection, within the first 5 days since onset of illness. The NS1 rapid antigen test (SD Biosensor) was conducted for a randomly selected subcohort of patients at the time of recruitment. The samples were transported to our laboratory and serum was separated and stored at −80°C.

Ethics Approval

Ethical approval was obtained from the Ethics Review Committee of the Faculty of Medical Sciences, University of Sri Jayewardenepura.

Quantitative Real-Time PCR Assays for Detection of the DENV Serotypes

Viral RNA was extracted from serum samples using MagMAX viral nucleic acid isolation kit (catalog No. A48310; Applied Biosystems) and an automated KingFisher Flex purification system (Applied Biosystems). Quantitative RT-PCR was performed using oligonucleotide primers and a dual-labelled probe for detection of DENV serotype 1 to 4, published by the Centers for Disease Control and Prevention (CDC) with slight modifications as described previously [4, 13]. However, in 2023, 50% of the NS1 antigen-positive samples (Table 1) gave a negative result with the quantitative RT-PCR (qRT-PCR) protocol used by us since 2014. We also implemented the Realstar Dengue Type RT-PCR Kit 1.0 (Altona Diagnostics) in a subset of NS1-positive, but CDC PCR protocol-negative, samples (n = 50) according to manufacturer's instructions.

Library Preparation and Next-Generation Sequencing of Dengue 3

Extracted viral RNA (QIAamp Viral RNA Mini Kit; Qiagen) was reverse transcribed using LunaScript RT SuperMix. Tiling PCR was then performed using specific dengue primers for DENV-3 as previously published [14]. Briefly, 2 primer pools (10 µM) A and B were prepared, and subjected to rapid barcode ligation using Rapid Barcoding kit (SQK-RBK110.96; Oxford Nanopore Technologies). MinION flow cell (FLO-MIN 106; Oxford Nanopore Technologies) and run on a MinION Mk1C device (Oxford Nanopore Technologies). Base calling was performed using Guppy (version 6.5.7) with Fast model, 450 bps base calling model. Reads were aligned with DENV-3 reference genome (National Center for Biotechnology Information accession No. NC_001475.2 dengue virus 3) using the wf-alignment workflow. Next, Samtools (version 1.16.1) was used to construct consensus sequences (minimum quality score threshold per base 20, and minimum depth coverage per base for consensus, 10). Genotype assignment was performed with the Flavivirus Genotyping Tool version 0.1.

DENV-3 Time-Resolved Phylogenetic Tree

We subjected 6 whole-genome sequences with >70% coverage of the DENV-3 genome (NC_001475.2) and >90% coverage of the E and NS1 regions to phylogenetic analysis. Accession numbers included in the analysis are as follows: SRA PRJNA1108507 (genome with full coverage); and PP767447, PP767459, PP766873, PP770477, PP770484, and PP770483 (genomes with complete coverage of NS1 and regions). To carry out phylogenetic analysis, a total of 1316 DENV-3 complete genomes were downloaded from the Genbank database. Matching of the whole genomes with the corresponding metadata was perform using R\Biostrings (version 2.70.1) and R\BSgenome packages (version 1.70.1) run on R version 4.3.2. Multiple sequence alignment was generated using MAFFT version 7.508 employing the FFT-NS-i algorithm. Subsequently, this was used to infer a Randomized Axelerated Maximum Likelihood (RAxML) phylogenetic tree using RAxML (version 8.2.12) with GTRGAMMA substitution model and bootstrap of 1000 replicates [15]. The best-fit model GTR + F + R5 was chosen using ModelFinder [16]. Subsequently, a molecular clock phylogeny was generated using Clockor2 (version 1.6.6). Final visualizations of the phylogenetic tree were done using R\ggtree, R\ape, and R\ggstar packages (R version 4.1.2).

Mutation Analysis

We performed a mutation analysis of the primer and probe binding regions of the CDC primers, located between nucleotides 740 and 813 (Supplementary Figure 2). We used the multiple sequence alignment to detect the mutations within the DENV-3 strains identifies in Sri Lanka in 2013, in comparison to sequences from previous years in Sri Lanka and globally. We incorporated the mutation patterns in the primer binding regions and included a minimum of 5 levels of ancestral branches of our samples in the phylogenetic tree constructed in the previous analysis. Sequence visualizations of the phylogenetic tree were generated using R\ggtree, R\ggplot, and R\msa packages (R version 4.1.2).

RESULTS

Serotype Detection

We have conducted DENV surveillance activities at the National Institute of Infectious Diseases, Colombo since 2014, which is one of the largest infectious disease hospitals in Sri Lanka. Until the later part of 2022, the proportion of dengue NS1-positive samples that were negative by PCR remained below 16.1% (Table 1). The rates of samples positive for NS1 antigen but negative by PCR were 18.9% (10/53) in 2022 and rose to 50% (68/136) by November 2023 (Table 1). Using the Realstar Dengue Type RT-PCR Kit 1.0 (Altona Diagnostics) the DENV serotype could be identified in 31 of 50 of the samples previously negative according to the CDC primer-based RT-PCR (Supplementary Table 1). From these, 15 of 50 samples were identified as DENV-3, 14 samples as DENV-2, and 2 samples as DENV-1.

Phylogenetic Analysis of the DENV-3 Sequences in Sri Lanka

Sequencing was carried out in a total of 22 samples; 12 of these samples were CDC primer PCR positive and 10 were PCR negative using CDC primers. Of these, only 14 had adequate coverage and were assigned to their corresponding genotype. We found that 11 samples positive by the CDC DENV-3 primers were of DENV-3 genotype III, whereas the DENV-3 sequences that were not detected by the CDC DENV-3 primers corresponded to genotype I (n = 3). Six of these sample genomes had a coverage greater than 70% of the whole DENV-3 genome and >90% coverage of the E and NS1 regions. These genomes were included in the phylogenetic analysis (Figure 1).

Three main clusters (clade I, clade II, clade III) were identified in the phylogenetic tree of DENV-3 genotype I (Figure 2). Three subclusters were identified within clade I, which includes sequences from China, Sri Lanka, Bangladesh, and Papua New Guinea. Genotype I of DENV-3 previously reported from Sri Lanka in 2017 was found to be closely related to the current 2023 strains of genotype I from Sri Lanka (Figure 2). The Sri Lankan DENV-3 genotype I sequences from 2017 and 2023 share a common ancestor with DENV-3 sequences from China (clade I).

The other cocirculating genotype in Sri Lanka in 2023, DENV-3 genotype III strains, shares a common ancestor with a sequence from India collected in 2022 and shows that the Sri Lankan samples closely relate to global samples (clade I) (Figure 3). Three subclusters were identified within clade I, which include Indian sequences that are similar to the Sri Lankan sequences identified in 2023, a subcluster consisting of sequences from China and Singapore, and a subcluster with Indian sequences alone. Clade II and clade III in Figure 3 represent 2 different time zones in the evolution. Sri Lankan DENV-3 genotype III strains before year 2000 were assigned to clade III.

Mutation Analysis of Primer Binding Regions

As genotype I of DENV-3 was not detected by CDC DENV-3 primers, we proceeded to investigate the mutations that could be responsible for evading detection. We also investigated any mutations within the genotype III sequences of DENV-3. The regions of both genotypes I and III were aligned with the forward and reverse primer binding sites and also the binding region of the probe (Supplementary Figures 1 and 2). We found a single nucleotide substitution of C to T at position 744 and A to G at position 756, and a G to A substitution at position 795 in genotype I compared to the reference DENV-3 genotype I genome NC_001475.2. These mutations were located within the forward primer and reverse binding sites of the CDC DENV-3 primers (Figure 4A and 4B) in Sri Lanka virus strains after 2017 and was also seen in the Sri Lankan genotype I, identified in 2023 (GenBank accession numbers: PP766873; SRA: PRJNA1108507 AICBU07.03_2023-06-05_SriLanka). These point mutations were detected within the Sri Lankan genotype I (2023) strain at positions C744T and A756G of the forward primer binding sites and at position G795A of the reverse primer binding sites. These mutations were not found in genotype III. The 740 to 820 regions in genotype III were found to be more conserved compared to genotype I. There were no mutations detected within the probe binding region in genotype I or III (Figure 4C). DENV-3 strains with mutations within both the forward and reverse primer binding sites, as seen with the DENV-3 genotype I strain, have not been detected elsewhere, based on published DENV-3 sequences.

DISCUSSION

In this study we found that the large dengue outbreak that occurred in 2023 in Sri Lanka was due to emergence of DENV-3, with simultaneous cocirculation of 2 genotypes (genotype I and III), with genotype I evading detection by the CDC primers used by us [4]. From 2016 to the end of 2019, the predominant serotype was the cosmopolitan strain of DENV-2, which was responsible for the largest dengue outbreak, seen in 2017 [9]. DENV-3 emerged towards the later part of 2019, with an increase in number of cases until January 2020. However, with the lock downs and other social restrictions implemented due to the COVID-19 pandemic, a very low number of cases were seen in Sri Lanka from 2020 to mid-2022, except in 1 city in the Eastern province [4]. During 2020 to mid-2022, DENV-3 was not detected and DENV-2 was the only detected serotype in the limited number of dengue cases [4]. With the lifting of the social restrictions, dengue reemerged in mid-2022 and DENV-3 emerged as the predominant serotype by 2023. Infections due to DENV-3 were seen prior to 2009 and only for 4 months during the later part of 2019 [4, 17]. Therefore, reintroduction of DENV-3 to a population that had not been exposed to DENV-3 for almost 15 years could have led to the large outbreak seen in 2023.

During the DENV-3 outbreak, we observed a discrepancy between the NS1 antigen positivity rates and PCR positivity rates. The RT-PCR for dengue is more sensitive and specific compared to the NS1 antigen rapid test for diagnosis of acute dengue [18, 19]. Based on our laboratory results prior to 2019, only 8.1% of samples that gave a positive result for the NS1 antigen test gave a negative result for PCR. However, in 2023, 50% of the samples that tested positive by NS1 rapid antigen test gave a negative result by PCR using the CDC primers. We noted that samples positive by NS1 antigen, but PCR negative, were more likely to be DENV-3 genotype I whereas those that were PCR positive were DENV-3 genotype III. NS1 antigen could persist longer than the DENV in patients with acute dengue and therefore NS1 antigen detection methods such as enzyme-linked immunosorbent assay (ELISA) could be more sensitive than PCR in detecting acute infection [20, 21]. However, all samples tested in our study were collected during early illness (≤4 days of illness) and therefore PCR has shown to have a higher sensitivity than detection of NS1 in early illness. So far, such a discrepancy (50%) between PCR and NS1 results has not been reported elsewhere.

We identified point mutations, C744T and A756G, of the forward primer binding sites and in position G795A of the reverse primer binding sites, which were not identified in DENV-3 genotype III. In fact, DENV-3 strains with all 3 mutations were not identified elsewhere, based on published sequences. These critical mutations in the forward and reverse primer binding sites, simultaneously occurring in DENV-3 genotype I, are likely to have led to the PCR dropouts, which have not yet been reported elsewhere.

While 1 sequence of DENV-3 genotype I from Sri Lanka in 2017 has been recorded, DENV-3 was not detected in patients with acute dengue until the later part of 2019. Therefore, it is possible that DENV-3 genotype I was circulating in Sri Lanka, causing infection, but remained undetected by us and others due to PCR failure based on the primers used. Our results highlight the importance of systematic DENV surveillance including serotyping, genotyping, and sequencing. Given the intense transmission seen today in most countries, accompanied by high rates of virus evolution, sequencing is critical to understand disease dynamics [6, 22, 23].

The genotype III of DENV-3 was shown to be responsible for the initiation of dengue outbreaks in Sri Lanka in the 1980s, which also caused dengue outbreaks globally [24]. However, the current clade of the DENV-3 genotype III differs from strains circulating from 1980 to 1990 and is similar to those of the Indian subcontinent [25]. The genotype III of DENV-3 causing outbreaks in Brazil in 2023 was also shown to originate from the Indian subcontinent via the Caribbean [26]. Both genotypes I and III of DENV-3 have been responsible for recent outbreaks in Bangladesh, genotype III in India, and genotype I in Indonesia, although so far these have not been reported in many other Asian countries [25, 27, 28]. DENV-3 genotype II has so far been responsible for most outbreaks in Southeast Asia [29]. Although many other countries experienced large dengue outbreaks globally and in South Asia due to DENV-3 [27], the genotypes and the clades responsible for these outbreaks in 2022 and 2023 have not been reported.

In this study we show that the large dengue outbreak that occurred in 2023, post–COVID-19 in Sri Lanka was due to emergence of DENV-3, with cocirculation of 2 genotypes. We found that the 2 genotypes are likely to have originated from 2 locations, and most importantly 1 of the genotypes evaded detection by primers used by us for over 10 years. This highlights the importance of continued surveillance activities including genomic sequencing. Future studies should address the potential differences in phenotype of these 2 genotypes, for example differences in clinical disease severity and/or the vector competence. Furthermore, because many of the dengue vaccines incorporate only 1 genotype representing each of the DENV serotypes, it would be important to closely monitor the potential evolution of different genotypes to understand vaccine efficacy and possible immune evasion. As we have shown, characterizing sequence evolution is also critical for assessing the performance of molecular diagnostic tests.

Supplementary Data

Supplementary materials are available at The Journal of Infectious Diseases online (http://jid.oxfordjournals.org/). Supplementary materials consist of data provided by the author that are published to benefit the reader. The posted materials are not copyedited. The contents of all supplementary data are the sole responsibility of the authors. Questions or messages regarding errors should be addressed to the author.

Notes

Author contributions. D. A., G. N. M., and C. J. contributed conceptualization. G. N. M., C. J., and A. W. contributed project administration. D. A., H. K., and A. W. performed data curation. D. A., H. K., T. T. P. J., L. G., and F. B. performed laboratory assays. B. S., D. A., S. M. J., S. B., and D. R. performed data analysis. G. N. M. and C. J. acquired funding. D. A., B. S., and G. N. M. wrote the original draft. S. M. J., S. B., and G. N. M. reviewed and edited the manuscript.

Financial support. This work was supported by the National Institute of Allergy and Infectious Diseases (grant number 5U01AI151788-02).

References

Author notes