Dengue virus serotype did not contribute to clinical severity or mortality in Taiwan’s largest dengue outbreak in 2015

Dengue virus (DENV) belongs to the genus Flavivirus and family Flaviviridae, and there are four different serotypes of DENV (DENV-1 to DENV-4) according to serological tests [1]. Infection by DENV in humans can result in a wide range of clinical consequences, including asymptomatic infections, dengue with warning signs, dengue without warning signs, severe dengue according to the 2009 World Health Organization (WHO) dengue guidelines [2] or undifferentiated fevers, dengue fever (DF), life-threatening dengue hemorrhagic fever (DHF) and dengue shock syndrome (DSS) according to the 1997 WHO dengue guidelines [3]. The impact of dengue on humans has increased over the past five decades due to frequent international travel, travel between urban and rural areas, population increases, and global warming, resulting in the widespread distribution of Aedes mosquitoes [4].

Dengue epidemics were recorded in the southern regions of Taiwan as early as the early twentieth century. Six major dengue outbreaks have occurred in Taiwan over the past four decades. The first dengue outbreak (dominant serotype DENV-2) occurred in 1981. Subsequently, dengue outbreaks occurred in 1988 (dominant serotype DENV-1), 2002 (dominant serotype DENV-2), 2007 (dominant serotype DENV-1), and 2014 (dominant serotype DENV-1), and the largest outbreak after World War II occurred in 2015 (dominant serotype DENV-2) [5]. The 2014 epidemic of the DENV-1 genotype I continued and resulted in sporadic cases in southern Taiwan between January and April 2015 [6‐9]. The dengue outbreak in 2015 started in July in Tainan city. This DENV-2 cosmopolitan genotype outbreak spread from Tainan to its neighboring city, Kaohsiung, at the end of August [7]. This year's dengue outbreak was initiated by the importation of dengue cases. The results indicated that various factors may have been associated with this DF outbreak in Kaohsiung, such as higher monthly average temperatures, precipitation, consecutive rainfall, and elevated Breteau index. These factors may have increased mosquito breeding activity, thus aiding in the transmission of DENV [7]. In addition, a recent study by Pan et al. identified a possible relationship between the period of disease concealment and the number of imported dengue cases, which resulted in epidemics of indigenous dengue fever within local communities [10].

Previous studies suggested that DENV-1 and DENV-2 were the major DENV subtypes in the 2014 and 2015 outbreaks, respectively [5]. The mortality rate during the DF outbreak in 2015 (0.52%) was higher than that in 2014 (0.17%) [7]. It has been suggested that different serotypes of DENV are associated with different severities of dengue [11‐13]. A recent meta-analysis study suggested that certain serotypes increased the risk of severe dengue with geographical effects [14]. The serotype effect on the disease severity as well as on the clinical outcomes in the 2015 dengue outbreak had not been investigated using the 2009 dengue guideline proposed by the WHO [2]. In addition, cocirculation of multiple serotypes of DENV during a single season increased the incidence of coinfections with more than one dengue serotype and was correlated with severe dengue [15]. Coinfection with multiple serotypes of DENV may result in the emergence of recombinant virus strains, which possibly increase disease severity [16] and the difficulty of vaccine development. Moreover, monitoring dengue serotype prevalence assists healthcare providers in predicting the potential severity of dengue cases and delivering suitable clinical care. In this study, we questioned whether serotypes other than DENV-2 were cocirculating and hypothesized that dengue serotype might play a role in clinical severity and outcomes. Therefore, we identified the dengue serotypes using serum samples collected between September and December 2015 and analyzed the association of DENV serotypes with dengue severity and mortality in Kaohsiung city during that period.

Two of the largest dengue outbreaks ever recorded in Taiwan occurred consecutively in 2014 (DENV-1) and 2015 (DENV-2). The number of confirmed DF cases in Kaohsiung city was 15,043 and 19,784 in 2014 and 2015, respectively, and these DF cases comprised 58.5% of all DF cases nationwide in these two years. Our results suggested that DENV-1 still circulated at 4.0–15.4% between September and December 2015 in Kaohsiung city after the number of DF cases surged in September 2015, which is in contrast to a previous study that reported that DENV-1 was not detected after September 2015 [9]. That study enrolled hospitalized dengue patients and performed RT-PCR for serotyping. Their methods were similar to those used in this study. The reason why they did not detect DENV-1 might be that the sample size was relatively low in their study (n = 300), so they might have missed DENV-1 in surveillance. The reason for this prolonged circulation of DENV-1 might have been the constant importation of DENV-1 from neighboring Southeast Asian countries [8, 10]. A recent study by Pan et al. revealed that 63.9% of imported cases had a concealment period of > 3 days in 2015. This led to the possibility that DENV-1 may have circulated in the community [10]. Although cocirculation of multiple serotypes of DENV increased the risk of being coinfected with more than one dengue serotype and was correlated with severe dengue, we did not find concurrent infection in the same individual by multiple dengue virus serotypes. In this study, we did not find any patients infected with DENV-3 or DENV-4. Previous results suggested that DENV-2 correlated with severe dengue and was associated with a higher proportion of patients with severe dengue than the other serotypes [11]. Lin et al. suggested a cosmopolitan genotype DENV-2 had high transmissibility to Aedes mosquitoes and was highly virulent in type I and type II interferon-deficient mice, with robust replication in spleen, lung, and intestine [23]. In contrast, Yung et al. demonstrated that DENV-1 infection may be more severe than DENV-2 infection [12]. However, in our study, there was no significant difference in any demographic characteristic between those who were infected with DENV-1 and DENV-2. We also found that those who were infected with DENV-2 did not have a higher risk of developing severe dengue, resulting in death, than those who were infected with DENV-1 and vice versa. The results in this study agree with our previous results that there is no significant difference between DENV-2 and other serotypes (e.g., DENV-3) in the development of severe dengue [24]. The dengue virus serotype (DENV-1 and DENV-2) did not contribute to clinical severity and mortality in Taiwan’s largest dengue outbreak in 2015. Our results were not consistent with previous results, which showed that DENV-2 had a higher risk of DHF and severe dengue [11, 13]. Our observations suggested that DENV-2 was not responsible for the relatively high mortality rate in 2015 compared with 2014. We found that people with DM and CKD had a significantly higher risk of developing severe dengue, and those with severe dengue or higher CCI scores had a significantly higher risk of death. Rowe et al. reported that elderly individuals (≥ 60) suffered more severe disease with more DHF (29.2% vs. 21.4%, P = 0.002) and severe dengue (20.3% vs. 14.6%, P = 0.006) than adults (< 60) [25]. Our results revealed that the elderly suffered more severe dengue, which was similar to their results (74.3% vs. 25.7%, P < 0.001). Our results also suggested that elderly patients accounted for the majority of dengue deaths (86.0% vs. 14.0%, p < 0.001), while their results suggested that dengue death between the elderly and the adults had no significant difference (P = 1). However, the reason for the difference between the two findings is not clear. The results of a recent meta-analysis by Chagas et al. suggested that DSS and DM are associated with mortality in patients with dengue [26]. Our results suggested that DM and severe dengue were associated with dengue death. The term "severe dengue" is used to describe the most serious forms of dengue disease, which can include both DHF and DSS. Thus, our results agreed with their findings. In addition, our results conformed to the conclusion in a study conducted in Thailand by Huang et al. that “our results suggest a continuing increase in age of DHF cases, shifting the burden toward individuals with more comorbidity” [27].

In the subanalysis of DENV-2-infected patients, the results revealed that patients who developed severe dengue had a higher average age than those who developed DF. Patients with DM or moderate to severe CKD were at higher risk of developing severe dengue after infection with DENV-2. Furthermore, the results suggested that females and those who develop severe dengue are at higher risk of death after infection with DENV-2. An earlier study reported that girls suffered from more severe illness than boys [28]. In addition, the results from the study by Anders et al. suggested that girls had a significantly higher risk of DSS and death [29]. Sex hormones delineate the distinctions between males and females, and the varying endocrine conditions could potentially impact both human physiology and pathophysiology. Sex-specific disparities can be discerned in numerous immune-related conditions, although many of these aspects have not yet undergone comprehensive evaluation [30]. The mechanism(s) of why females are at higher risk for severe dengue and dengue-related death is still unknown. In addition, the presence of anti-dengue IgG correlated with severe dengue and mortality. The results in this study are different from the peak dengue hospitalization in children or young adults, and severe dengue cases predominate and occur more often in children than in the elderly [29, 31, 32]. The results of preexisting anti-dengue IgG in the current study were consistent with our previous results that anti-DENV IgG seroprevalence rates increased with age, from 2% (20–29 y/o), 4.8% (30–39 y/o), 32.1% (60–69 y/o), and 62.9% (70–79 y/o) in healthy participants living in a highly epidemic district, Sanmin District, and seroprevalence rates were 4.8% (20–49 y/o), 9.3% (60–69 y/o) and 36.4% (70–79 y/o) in a low epidemic district, Nanzih District [33]. Recent studies suggested that severe dengue was characterized by elderly patient cases and multiple comorbidities in the 2015 dengue outbreak in Tainan city and Kaohsiung city [9, 34, 35]. Wang et al. reported that age ≤ 60 years, chronic diseases (hypertension and diabetes), preexisting anti-dengue IgG antibodies, thrombocytopenia, AST elevation, and APTT prolongation were correlated with DHF during the 2015 DF outbreak in Kaohsiung city [9]. The results by Chang et al. revealed that relatively high percentages of diabetes mellitus, hypertension, cerebral vascular accident, proton pump inhibitor use, severe hepatitis, and acute renal failure and relatively high maximum AST, maximum ALT, and initial white blood cells were observed in DHF patients in 2015 compared with 2014 [35]. Hsieh et al. reported that organ failure (cardiac failure and renal failure) and prolongation of initial APTT were predictive factors for in-hospital fatality during the 2015 DF outbreak in Tainan city. In addition, their data suggested that the proportion of severe dengue patients with hypertension, diabetes, and dyslipidemia was not significantly different between survivors and nonsurvivors [34]. However, their analysis did not mention which serotype of DENV the patients were infected with. Taken together, these results and our results indicate that dengue patients with chronic diseases, especially DM and moderate to severe CKD, have a higher risk of developing severe dengue. Our results also suggested that surveillance of the prevalent dengue virus serotype(s) can alert healthcare facilities to prepare for a potential increase in severe dengue cases and allocate resources accordingly. Testing for preexisting anti-dengue IgG (Additional file 1: Table S1) in the blood of dengue fever patients may provide an early warning of the development of severe dengue when a serotype is prevalent after other dengue virus serotypes have been circulating for many years. Public health agencies and researchers can prioritize investigations of prevalent serotypes to gain a deeper understanding of their genetic characteristics and virulence factors. Tailoring public awareness campaigns to highlight the major dengue serotypes can educate communities about the specific risks associated with prevalent serotypes, thereby promoting relevant preventive measures.

The severe dengue incidence was relatively high (6.6%) in our study when compared to nationwide data (1.49%) [7]. This could be because DF patients have relatively severe manifestations compared to overall community cases seeking medical treatment in medical centers. In contrast, the DF death rate was relatively low (2.4%) in our study when compared to nationwide data (5.15%) [7]. Our results revealed that aged patients and those who had DM or CKD were prone to severe dengue. Notably, the prevalence of DM and CKD increases with age in Taiwan [36, 37]. It is likely that the comorbidity instead of severe dengue alone makes the patients die. Other factors associated with dengue disease severity and clinical outcome include innate immunity [38], genetic background and the presence of secondary infections [39]. Recent study suggested mechanical transmission of DENV without an incubation period is expected to influence not only the timing, but also the severity, of DENV outbreaks [40]. These factors might play a role in the relatively high mortality rate in the 2015 DF outbreak in Taiwan. However, we did not include immunity and genetic background in our study. Primary DENV infection is characterized by the presence of anti-DENV IgM antibodies at 3 to 5 days PSO, and anti-DENV IgG antibodies usually become detectable at 7 to 10 days PSO. A low level of anti-DENV IgG antibodies persisted for decades after recovery from DENV infection and was an indication of a past DENV infection. During secondary DENV infection, anti-DENV IgG antibodies become detectable at high levels even in the initial phase of infection [2, 41‐44]. It is suggested that heterotypic secondary infection might result in a more severe form of dengue, such as DHF. It is likely that the anti-DENV IgG detected as shown in Additional file 1: Table S1 was heterotypic to DENV-2 since the major DENV strains attributable to DF epidemics were DENV-1 and DENV-3 in Kaohsiung city between 2005 and 2014. Notably, the patients enrolled in this study were all hospitalized, so they were unlikely to be infected with the same serotype as the previous DENV infection. Thus, heterotypic secondary infection might play a role in the higher mortality rate in the 2015 dengue outbreak than in the 2014 outbreak. However, more studies need to be conducted to verify this inference.

Dengue is not endemic in Taiwan. Dengue outbreaks in Taiwan usually start with imported cases from Southeast Asian countries, spreading during rainy and warm weather starting in July and peaking between September and November [5]. Various serotypes and clades of DENV have been imported into Taiwan in the past twenty years from Southeast Asian countries, such as the Philippines, Thailand, Vietnam, Malaysia, Indonesia, Singapore, and Cambodia, and resulted in dengue epidemics of varying degrees [8, 45, 46]. It has been suggested that the business links and tourism between Taiwan and other Southeast Asian countries are responsible for the importation of multiple DENVs and dengue outbreaks in Taiwan [8]. To understand the possible origin of DENV-2 responsible for the DF outbreak in Taiwan in 2015, we retrieved genomic sequences of the DENV-2 E gene from GenBank to reconstruct the phylogenetic tree [47]. The clinical samples of these sequences were collected from dengue patients imported from abroad into Taiwan (virus name with the letters Tw) and from autochthonous DF patients in Taiwan (virus name with the letters TN or KH) in 2015 (Fig. 2a). Our results suggested that the DENV-2 circulated in Kaohsiung city (KH) was clustered with the DENV-2 circulated in Tainan city (TN), which was the genotype II – cosmopolitan clade – and possibly originated from Indonesia, while a previous report suggested that this DENV-2 is phylogenetically clustered with isolates from China and Thailand [7, 9]. Our results were consistent with the results from a previous study showing that the DENV-2 epidemic in Kaohsiung city was phylogenetically clustered with the DENV-2 epidemic in Tainan city [7]. In addition, most imported cases were from Indonesia, while the number of cases from Thailand and China was relatively low in 2015 (Additional file 1: Fig. S3). Furthermore, the possibility that the DENV-2 sequences included in the phylogenetic analysis in the previous studies [7, 9] were imported into Thailand and China from Indonesia had not been ruled out. Taken together, these results indicate that the DENV-2 in the 2015 DF outbreak in Taiwan possibly originated from Indonesia. On the other hand, DENV-1 sequences detected during the 2015 DF outbreak in Taiwan were imported from many Southeast Asian countries (Fig. 2b). Most of these sequences were phylogenetically close to the DENV-1 sequences detected in the dengue outbreak in 2014 from the same countries. In addition, our results confirmed that the 2014 epidemic DENV-1 in Taiwan originated from Indonesia [8].

There are strengths in our study. We have two novel findings from our study. First, we identified the serotypes of every enrolled dengue subject, which was not done in previous studies. Therefore, we found that DENV-1 circulation persisted in the community in addition to DENV-2 in the 2015 large outbreak. Second, we further found that severity and mortality were not associated with the dengue serotypes DENV-1 and DENV-2. Furthermore, we found that DENV-1 in 2015 evolved from that in 2014 from phylogenetic analysis.

There are some limitations to our study. First, we collected only single serum samples but not paired-serum samples in this study, so we could not perform anti-DENV antibody assays to include more potential DF cases using convalescence serum and follow-up on the symptoms, progression, and prognosis of dengue disease. Second, the sample size of DENV-1 is relatively low compared with that of DENV-2. Thus, we could not analyze the differences in clinical severity or clinical outcomes among hospitalized severe dengue patients infected with DENV-1. Third, the skewing of the data was due to collection only from hospitalized patients, which did not include many who would have been treated as outpatients and cases from the community. Those with milder disease or asymptomatic infections may have been missed. Due to these biases, our results reflected only those hospitalized patients and might not be representative of the overall outbreak in the community. Fourth, we reconstructed the phylogenetic trees using limited sequences and using the E gene but not the full-length DENV genome. Thus, the results in this study may not provide a complete picture of the evolutionary history of the dengue virus. Fifth, the dengue outbreak in Kaohsiung began in August 2015 and ended in March 2016 according to data from the Taiwan CDC. Although ~ 96% of the DF cases occurred during our study period, the timeframe did not capture the entirety of the outbreak. Thus, we might have missed earlier or later variations in serotype prevalence. Finally, the results of this cross-sectional study did not include factors such as urbanization, migration, changes in mosquito control strategies, climate change, and socioeconomic and ecological factors.

Our results confirmed DENV-2 as the major serotype in the dengue outbreak in Kaohsiung city in 2015, while DENV-1 still circulated in the community. Dengue serotypes (DENV-1 and DENV-2) did not contribute to either severity or mortality. Aged patients and those who had DM and moderate to severe CKD were at a higher risk for severe dengue. Higher CCI scores and dengue severity cast a higher risk of mortality. For DENV-2-infected patients, we found that age, the presence of preexisting anti-dengue IgG, DM, and moderate to severe CKD were associated with a significant risk for severity, and female sex and severity contributed to death. Therefore, we should focus on the meticulous management of dengue with DM and moderate to severe CKD to lower the severity and prevent mortality. In addition, testing for preexisting anti-dengue IgG in dengue patients may provide early warning of the development of severe dengue when one serotype is circulating years after other dengue virus serotypes circulated.