Characteristics of the initial dengue outbreaks in a region without dengue prior to mid-2009 in a dengue-endemic country

Introduction: The present study evaluated the characteristics of the initial dengue outbreaks in the Jaffna peninsula, a region without dengue prior to mid-2009 in dengue-endemic Sri Lanka, a tropical island nation. Methodology: This is a cross-sectional study conducted using a total of 765 dengue patients’ clinical data and samples collected from the Teaching Hospital, Jaffna during the initial dengue outbreaks. Clinical, non-specific, and specific virological laboratory characteristics including the platelet count, NS1 antigen, and anti-DENV IgM/IgG were evaluated as correlates of dengue virus (DENV) infection in the two initial outbreaks of 2009/2010 and 2011/2012 in Northern Sri Lanka. Results: Firstly, affected age and clinical characteristics were significantly different between the outbreaks ( p < 0.005). Secondly, NS1 antigen detection in patients with fever days < 5 was statistically significant ( p < 0.005). Thirdly, platelet count, detection of NS1 antigen, and anti-DENV IgM/IgG profiles were adequate to diagnose 90% of the patients; hepatomegaly and platelet count of < 25,000/mm 3 were identified as predictors of severe disease. Fourthly, secondary DENV infections were detected in the early stages of the illness in many patients. Finally, infecting DENV serotypes were different between the two outbreaks. Conclusions: Clinical and non-specific laboratory characteristics and the infecting DENV serotypes between the two initial outbreaks in Northern Sri Lanka were significantly different. NS1 antigen, anti-DENV IgM/IgG, and platelet counts were identified 90% of the dengue patients. Hepatomegaly and platelet count of < 25,000/mm 3 were able to predict the disease severity in this study.


Introduction
The clinical features of symptomatic dengue virus (DENV) infections vary and most patients experience a self-limiting febrile illness.In less than 5-10% of DENV-infected individuals, the infection causes severe disease, dengue hemorrhagic fever (DHF), and dengue shock syndrome (DSS).The DHF is mostly characterized by plasma leakage with or without hemorrhage [1].Identification of patients who are likely to progress to DF/DSS has been a challenge in many dengue endemic regions.Thus, it is necessary to evaluate the virological and clinical characteristics with non-specific laboratory data such as platelet count, white blood cell count (WBC), packed cell volume (PCV), urine output, and blood pressure during the course of illness for the effective management of patients.Progressive leucopenia followed by a drop in platelet count often leads to plasma leakage in DHF/DSS patients.Although virus culture and identification of DENV is the gold standard for laboratory confirmation of dengue, it is not a feasible diagnostic option due to low sensitivity to the disease progression and the long turnaround time [2].Conversely, DENV nucleic acid detection by RT-PCR is highly sensitive in the acute phase of the illness but RT-PCR requires specialized laboratory infrastructure and expertise, which are not widely available in many resource-limited dengue endemic countries [3].Hence, relatively less expensive laboratory tests with reasonable diagnostic precision and short turnaround time such as NS1 antigen or/and anti-DENV IgM detection would be suitable to confirm DENV infection in patients with clinical suspicion.
During a clinical suspicion, anti-DENV IgM detection is the most used diagnostic test and is carried out after fever days 5 or more.However, it is not suitable for the early diagnosis of DENV infection, especially when patients present during the early viremic phase, prior to fever day 5 [4].Moreover, there is a possibility of cross reactivity between anti-DENV IgG and other flaviviral antibodies such as antibodies against Japanese encephalitis (JEV), yellow fever and Zika viruses.The latter two infections have not been reported in Sri Lanka, however, the possibility for cross-reactivity with the flaviviral antibodies could not be excluded.
Presence of NS1 antigen, a conserved and secreted glycoprotein, has been used as a diagnostic marker for the early diagnosis of DENV infection in patients with clinical suspicion [3,5,6].NS1 is a long-lived antigen, which is detectable during the course of dengue disease [7,8], especially in the early viremic phase, when the fever days are ≤ 5 [9].Differences in NS1 antigenemia have been reported in patients from different geographical areas and in some patients, it is detectable only after fever day 5 [9].
Sri Lanka has been affected by dengue for several decades.Although serologically confirmed dengue has been described since 1960, it was only after 1989, dengue became a public health problem on the island.There was an island-wide outbreak reported between 1965 and 1968 with 51 DHF cases and 15 deaths [10,11].In 1989, during the first larger dengue outbreak there were 203 cases documented with 20 deaths [11,12].Subsequently, the number of dengue cases has been on the rise with the occurrence of annual outbreaks.In 2004 and 2009/2010, Sri Lanka experienced explosive dengue outbreaks with 15,457 and 35,008 reported cases and 88 and 346 deaths, respectively [11,13].Now dengue is a well-established vector-borne viral disease-causing significant morbidity and mortality in many parts of the country.Jaffna peninsula is geographically isolated from the mainland of Sri Lanka.The isolation was confounded by thirty years of political unrest and the travel between the Jaffna peninsula and the mainland of Sri Lanka was restricted.Unrestricted movement of people started between the two regions in mid-2009.Prior to 2009, there were < 50 presumed dengue cases have been reported from Jaffna.However, due to the lack of laboratory services during the conflict, data on molecular or serological tests of these cases are not available [14].Due to the non-availability of etiological diagnosis in these regions, physicians and the pediatricians rely on non-specific laboratory data such as platelet counts to support their clinical diagnosis.The current study aimed to assess the clinical, non-specific, and specific virological laboratory data including platelet counts, NS1 antigen, DENV nucleic acid, and anti-DENV IgM/IgG as correlates of DENV infections during the two initial outbreaks (2009/2010 and 2011/2012) in northern Sri Lanka.On the other hand, in this study, we describe the clinico-demographic characteristics and infecting DENV serotypes of the two initial dengue outbreaks preceding the current dengue epidemics.

Ethical approval
The study was approved by the Ethical Review Committee of the Faculty of Medicine, University of Peradeniya, Sri Lanka (EC No: 2010/EC/13).An informed written consent was obtained from the study participants prior to collect demographic data and blood samples.

Classification of dengue
This cross-sectional study was carried out utilizing clinical laboratory data of patients admitted to medical and pediatric wards of the Teaching Hospital, Jaffna, during the two initial dengue outbreaks in 2009/2010 (outbreak 1) and 2011/2012 (outbreak 2).A total of 765 patients, with clinically suspected dengue satisfying the case definition of Ministry of Health, Sri Lanka and WHO [15,16] were recruited.Demographic and clinical data were obtained from each patient using a pre-tested questionnaire.Non-specific laboratory data and platelet counts were obtained from the patient's clinical notes.Moreover, based on the dengue guidelines [1], the study sample was divided into two disease severity categories, dengue with or without warning signs and severe dengue cases [15].

Diagnosis of dengue
Of the 765 samples, 300 were from the 2009/2010 outbreak and 465 from the 2011/2012 outbreak.WBC and platelet counts had been done during the hospital stay of the patients using an automated hematological analyzer (Mindray, China), as a routine laboratory procedure for all dengue suspected patients.Serum samples were retrospectively tested for NS1 antigen by ELISA (Pan Bio Diagnostics, Australia), DENV nucleic acid using RT-PCR (Promega, USA), anti-DENV IgM and IgG using ELISA (Pan Bio Diagnostics, Australia).All the tests were performed according to the manufacturer's instructions.DENV RNA was extracted from patients' sera using a viral RNA extraction mini kit (Qiagen, Hilden, Germany, Cat No 5206).Primers (D1 and D2) described by Lanciotti et al (1992) were used in RT-PCR to amplify the junction of C/PrM genes of the DENV [17].

Statistical analysis
The data were analyzed using SPSS, Version 21.Mono-variate analysis was performed for all descriptive variables.Student's t-test was carried out to evaluate the association between DENV NS1 antigen positivity and the mean platelet counts.Differences in the mean platelet counts between DF and DHF patients were also determined using the Student's t-test.

Results
The present study assessed the clinical, non-specific and specific virological laboratory data including platelet counts, NS1 antigen, and anti-DENV IgM/IgG as correlates of symptomatic DENV infection of the two initial dengue outbreaks (2009/10 and 2011/2012) in a region without a past history of epidemic dengue prior to mid-2009.Of the 765 patients, 52.7% and 55.3% of patients were males in outbreaks 1 and 2, respectively.The mean age and the mean duration of fever in the study sample on admission were 26.23 ±   1 and 2).The number of adult and pediatric patients with different ranges of platelet counts and the number of patients with severe dengue in outbreaks 1 and 2 are given in Tables 1 and 2.

Discussion
Dengue was not a public health problem in the Jaffna District until 2009 due to the isolation of the population from the rest of the island since 1980 due to travel restrictions during the internal conflict.The end of the conflict in mid-2009 allowed free movement of people from the other parts of the country to Jaffna and vice versa and this contributed to a marked increase in the number of reported dengue cases from the Jaffna peninsula with larger outbreaks in 2009-2010 and 2011/2012 (Figure 1) compared to the previous years [11,14].To a lesser extent, factors such as clinicians' capacity to identify suspected dengue cases, the lack of the surveillance system, and diagnostic capacities/capabilities during the conflict era may have also contributed to the low case numbers reported prior to 2009.
In the 2009/2010 and 2011/2012 dengue outbreaks (n = 765), males were predominantly affected than females and this finding is similar to most other studies conducted elsewhere and in Sri Lanka [12,[18][19][20][21].The reason for high number of males being affected might be attributed to the fact that a significant proportion of males engage in agriculture and livestock production in Jaffna.They often work with an exposed torso increasing the risk of acquiring DENV infection through mosquito bites [11].However, a few studies have also reported a higher ratio of females with dengue [22,23].In both outbreaks, the mean age of the affected patients was almost similar (26.2 years) and such findings have been reported by previous studies as well [12,24].In both outbreaks, all clinically dengue suspected patients presented with fever and other associated symptoms.The percentage of patients with headache, retro-orbital pain, arthralgia, and flushed extremities, pallor, rash/hemorrhages, and effusion were significantly different between the two outbreaks (Table 1).The virological laboratory characteristics were also significantly different between the two outbreaks (Table 3).These differences might be due to the clinical evolution of the dengue disease and infecting DENV serotypes in a previously nonepidemic region.Moreover, the main reason for detecting virological parameters such as NS1 antigen and DENV nucleic acid in a significantly large number of patients may be due to the low mean duration of fever on admission, especially during the 2011/2012 outbreak.
In the present study, the majority of the patients were positive for NS1 antigen when the fever day was < 5, suggesting that NS1 antigen detection is better done at an early stage in dengue-endemic regions and similar finding has been reported by others as well [3,7].On the other hand, NS1 antigen has been shown to be positive for up to fever days 6 to 7 [25] and 9 [8].The evolution of dengue disease in different populations is believed to be the reason for the differences in NS1 antigenemia during the course of fever [25,26].
Of the 765 patients, 21.6% (n = 165) were positive for anti-DENV IgM only indicating a primary DENV in these patients.However, 19% (n = 146) of the patients were positive for NS1 antigen and anti-DENV IgM;7.7% (n = 59) were positive for NS1 alone.Thus, in total, ~30% of the study population (293%) had primary DENV infection due to the presence of NS1 antigen or anti DENV IgM or both [27].
Of the 765, over 30% of the cases were positive for anti-DENV IgG alone indicating past exposure or secondary DENV infections.The anti-DENV IgM disappears relatively faster than that in a primary infection [28].Less than 30% of the study sample was positive for both anti-DENV IgM and IgG indicating another subset of secondary DENV infections [2], however, some of these patients might have had a primary DENV infection in which both anti-DENV IgM and IgG were detectable at the time of sample collection [29].Overall, 60% were positive for anti-  This study describes the first two outbreaks 2009/2010 and 2011/2012 in the Jaffna Peninsula, the Northern Sri Lanka.However, the mainland of Sri Lanka, predominantly the Colombo District, has been experiencing regular dengue outbreaks since 1989.
DENV IgG with or without anti-DENV IgM indicating that the majority of the patients would have had a secondary infection or past exposure to DENV.High rate of secondary infection might be due to the circulation of multiple DENV serotypes.Furthermore, we noted a prevalence of 34% (157/460) anti-DENV IgG in those that were < 19 years of age (Figure 2), as reported in some other dengue-endemic countries during different stages of the evolution of dengue [30,31,32].Detection of anti-DENV IgG is not useful in the detection of acute DENV infection and the presence of these antibodies can be attributed to past DENV infections.There is a possibility of crossreactivity between anti-DENV IgG and other flaviviral antibodies such as antibodies against Japanese encephalitis (JEV), yellow fever, and Zika viruses.The latter two infections have not been reported in Sri Lanka, however, the possibility for cross-reactivity with flaviviral antibodies could not be excluded.Only 9.6% of the patients were positive for NS1 antigen, anti-DENV IgM, and IgG among which 40% of the patients had severe dengue indicating a possible re-infection in this subgroup.NS1 antigen and anti-DENV IgM were detectable early due to the current infection and the anti-DENV IgG would have resulted from a past infection [27].Of the patients tested, 18.6% (n = 142) were positive for NS1 antigen and anti-DENV IgG, while 28.4% (n = 218) were positive for anti-DENV IgM and IgG.These patients may have had secondary DENV infection and therefore have a greater risk of developing severe dengue [33,34].The number of DHF cases was high in both primary and secondary infections, suggesting the risk of DHF in both [11].In this subset also, anti-DENV IgG resulting from crossreactivity to other flaviviral infections cannot be ignored and can only be excluded through viral neutralization at least in a subset of samples.Hepatomegaly and a low platelet count of < 25,000/mm 3 were significantly associated with severe dengue disease based on the current study findings and these are in agreement with the study carried out by Agrawal et al. [35].However, in contrast to the current study, Agrawal et al. [35] also show a significant association between splenomegaly and severe dengue disease.
In the current study, 45.3% (n = 347) of the patients had laboratory-confirmed DENV infection as detected by NS1 antigen and/or DENV nucleic acid detection.Combining DENV NS1 antigen and nucleic acid detection is useful in diagnosing DENV infections in the acute stage of illness [6,4,[36][37][38].The region where this study was conducted is also endemic for typhoid [39], typhus [40], and leptospirosis [41], which are clinically indistinguishable from dengue at the early stage.A total of 81 patients who were negative for NS1 antigen, DENV nucleic acid, and anti-DENV IgM/IgG in this cohort may have had infections, which were at early stages clinically indistinguishable from dengue.
Based on the current WHO recommendations [1], patients with platelet count < 1,00,000/mL and clinical symptoms suggestive of dengue require hospital admission.Based on the guidelines, 11.4% of the patients (n = 87) were positive for NS1 antigen with platelet count > 1,00,000/mL and they did not qualify for admission to the hospital (Table 2).If the platelet count of < 1,00,000/mL or the presence of anti-DENV IgM were considered for admission, 79.6% (n = 609) of the patients would qualify for hospital admission.NS1 antigen, anti-DENV IgM and platelet count < 1,00,000/mL together would increase the detection of patients who would require hospital admission.In the present study, > 90% of the patients admitted to the hospital on clinical suspicion for dengue had either positivity for NS1 antigen or anti-DENV IgM or both with platelet count < 1,00,000/mL.RT-PCR for the identification of DENV infection in the viremic phase is of value.However, in resource limited settings utilizing the platelet counts with NS1 antigen and anti-DENV IgM/IgG will improve the laboratory diagnosis of dengue.Additionally, adapting a diagnostic algorithm [35,42] will help to differentiate at the onset of infection other febrile infections that clinically mimic dengue.
In the 2009/2010 outbreak, DENV-2 and DENV-3 were the predominant cause of infection with a few cases of DENV-1 infection.In contrast, in the Western Province DENV-1 had been the predominant cause of infection in the same period in 2009/2010 [13].Dengue cases started to increase in Jaffna in mid-2009 and this period correlated with the opening of major highways connecting the Jaffna peninsula to the rest of the country after 30 years.DENV infections could have been introduced to the Jaffna peninsula from other parts of Sri Lanka where DENV-2 and DENV-3 were prevalent.However, we are unable to confirm this as phylogenetic data are not available from all the dengueendemic regions of Sri Lanka.Prior to 2007, DENV-3 was the predominant type followed by DENV-1 and DENV-2 in the Western Province of Sri Lanka [25].Moreover, DENV-2 or DENV-3 infection may have caused severe dengue in the Jaffna peninsula as most of the people in the region might not have had prior immunity against DENV-2 or DENV-3.DENV-1 was found to be the predominant serotype responsible for the infection and severe clinical presentation in the 2011/2012 outbreak.This finding is in agreement with other studies conducted in other parts of Sri Lanka during the same period [43].When considering the initial dengue outbreaks in the Jaffna peninsula, there was a shift in the DENV serotypes between the 2009/2010 and 2011/2012 outbreaks (Table 4).
In the 2009/2010 outbreak, mixed infections were caused by both DENV-2 and DENV-3, correlating with the predominant serotypes circulated in the population during 2009 and 2010.Similarly, the most predominant DENV serotype circulated during the 2011/2012 outbreak was DENV-1 and thus co-infections were detected between DENV-1 and DENV-2 or DENV-1 and DENV-3.The presence of DENV co-infections in both outbreaks reflects the hyper-endemicity of DENV in the Jaffna peninsula.DENV co-infections might contribute to the dominance and establishment of certain DENV serotypes in a region with or without future clinical implications.

Conclusions
Significant differences were noted in the clinicolaboratory characteristics between the 2009/2010 and 2011/2011 dengue outbreaks.Hepatomegaly and a low platelet count of < 25,000/mm 3 were able to predict severe dengue in the study sample.This will assist in the early detection of patients who are likely to progress to severe disease.NS1 antigen, anti-DENV IgM/IgG detection, and platelet count improved the diagnosis of DENV infection in patients suspected of having dengue.

Figure 1 .
Figure 1.A comparison of reported dengue cases from 2004 to 2017 in all government healthcare facilities of Colombo and Jaffna Districts of Sri Lanka (Adapted from the Epidemiology Unit, Ministry of Health, Sri Lanka).

Table 1 .
Differences in the clinical and non-specific laboratory characteristics of paediatric patients between dengue outbreaks 1 (2009/2010) and 2 (2011/2012) in the Jaffna Peninsula of the Northern Sri Lanka.These were the first two dengue outbreaks in the Northern region while other parts of Sri Lanka have been experiencing regular dengue outbreaks since 1989.

Table 2 .
Differences in the clinical and non-specific laboratory characteristics of adult patients between dengue outbreaks 1 (2009/2010) and 2 (2011/2012) in the Jaffna Peninsula, the Northern Sri Lanka.These were the first two dengue outbreaks in the Northern region while other parts of Sri Lanka have been experiencing regular dengue outbreaks since 1989.

Table 5 .
Clinical and laboratory characteristics of patients with severe dengue.Presence