Skip to main content
European Commission logo
español español
CORDIS - Resultados de investigaciones de la UE
CORDIS
CORDIS Web 30th anniversary CORDIS Web 30th anniversary
Contenido archivado el 2024-06-18

International Research Consortium on Dengue Risk Assessment, Management, and Surveillance

Final Report Summary - IDAMS (International Research Consortium on Dengue Risk Assessment, Management, and Surveillance)

Executive Summary:
The research activities in the IDAMS project respond to the need of new tools in both clinical research, biological sciences as well as population research and disease control. The key areas of research were i) improving diagnosis and clinical management of dengue, and ii) assessing the risk of dengue spread and developing effective and affordable early warning and outbreak response systems.
A large prospective observational multicentre study was designed a) to differentiate between dengue and other common febrile illnesses within 72 hours of fever onset, and b) among patients subsequently confirmed to have dengue, to identify markers that are predictive of likely evolution to a more severe disease course (Jaenisch T et al: Clinical evaluation of risk factors for severe disease: protocol for a multicentre study in 8 countries. BMC Infectious Diseases 2016; ClinicalTrials.gov registration number NCT01550016). 7428 patients with undifferentiated febrile illness were recruited in 8 countries, among them 2696 with laboratory-confirmed dengue. Complex modelling techniques were applied and parameters that differentiate between dengue and other febrile illness as well as parameters associated with a severe course of disease were analysed. In addition, we have identified virological and immunological correlates of disease and immunity. This represents important enabling science to enhance development and validation of vaccines and therapeutic drugs.
Modern mapping techniques were used to map the global distribution and burden of dengue, its future global distribution based on projected climatic and socioeconomic data, and the past distribution of the different dengue virus types. The total number of dengue infections was estimated at approximately 390 million per year (as of 2010) with ~100 Million symptomatic infections (Bhatt S, et al: The global distribution and burden of dengue. Nature 2013). Dengue-related health care costs per square kilometre globally were produced alongside a range of work on cost of illness, cost of dengue outbreaks, and the cost effectiveness of early warning systems for dengue. When Zika was declared a Public Health Emergency of International Concern by the WHO, the inclusion criteria in the observational prospective study were adapted to capture Zika cases and species distribution modelling was carried out to create a global map of environmental suitability for Zika virus transmission to humans. Much of this work involved the compilation of extremely comprehensive databases, which were made available in the public domain.
For the development of outbreak response systems, systematic reviews critically examining current practices in dengue surveillance systems were completed and published suggesting that sensitivity and specificity of Early Warning through alert signals for dengue outbreaks can be increased by combining different surveillance approaches, and by quality controlled laboratory support, respectively. Hence, the collection and validation of retrospective data from five countries (Dominican Republic, Mexico, Brazil, Malaysia, and Viet Nam) was undertaken. In each country, climatic, entomological and epidemiological ‘alarm’ variable data were explored for evidence of quantitative associations with subsequent probable and hospitalized dengue cases. The statistical model was developed and tested. Results showed that the most common reliable ‘alarm’ indicators were mean temperature, the presence of probable dengue cases and changes in mean age of dengue-infected individuals. In the next step, the model was evaluated in a prospective study. The Early Warning and Response System focuses on the use of a ‘staged response’, where specific interventions (comprising improved use of surveillance information, timely application of response activities as per national guidelines, particularly vector population management strategies, rapid vector control intervention) are deployed in response to the presence of specific alarm signals.
The final dissemination meeting was jointly organized with the Institut Pasteur-funded DENFREE consortium in December 2016 in Paris and provided a platform for the presentation of the scientific work of the two consortia - as well as mapping out the way forward towards an integrated arbovirus research agenda.
Project Context and Objectives:
In the twenty-first century, dengue incidence, outbreak frequency and global distribution have reached unprecedented levels, causing a pandemic of unprecedented extent. All WHO regions are now endemic for dengue (after the outbreak in Madeira 2012/13 in the EURO region) and an increasing number of countries is reporting dengue cases to the WHO each year (WHO: Global strategy for dengue prevention and control 2012 – 2020. WHO, 2012). The total number of dengue infections was estimated at approximately 390 million per year as of 2010 with ~100 Million symptomatic infections (Bhatt S et al: The global distribution and burden of dengue. Nature 2013, 596(7446): 405-7).
Clinical management of the disease relies on careful monitoring and the cautious use of intravenous fluid therapy. Clinical or laboratory-based warning signs as tools to evaluate likely progression to more severe disease are based on expert opinion or were validated on small number of patients. On the population level, the available tools and approaches for detecting, predicting and responding to dengue outbreaks all derive from a previous era when the burden of dengue was a significantly lesser challenge. The research activities in the IDAMS project respond to the need of new tools in both the clinical research area as well as on the population level (Jaenisch T et al: Dengue Research Funded by the European Commission – Scientific Strategies of Three European Dengue Research Consortia. PLoS Neglected Tropical Diseases 2013) The key research areas are aimed at:
• improving diagnosis and clinical management of dengue through two linked work packages designed a) to identify readily available clinical and laboratory parameters and/or viral and immunological markers, that differentiate between dengue and other common febrile illnesses within 3 days of fever onset, and b) to identify any of the available markers that are predictive of the likelihood of evolving to a more severe disease course
• assessing the risk of dengue spread though linked work packages focused on a) mapping and modelling techniques to define the current extent of dengue disease globally and to evaluate possible scenarios of spread or risk to previously uninfected regions in the future, and b) developing effective and affordable early warning and outbreak response systems.
These research areas are represented in four work packages which are supported by a fifth work package dedicated to networking and translational activities to ensure that outputs from the various research activities are used to maximal advantage (Figure A.1).
The primary objective of Work Package 1 (WP1) was to carry out a large multinational observational study aiming to improve diagnosis and clinical management of dengue through approaches designed a) to differentiate between dengue and other common febrile illnesses within 72 hours of fever onset, and b) among patients subsequently confirmed to have dengue, to identify markers that are predictive of likely evolution to a more severe disease course.
The goal of Work Package 2 (WP2) was to identify virological and serological correlates of severe dengue. In addition, WP2 provides the key diagnostic outcomes for WP1. Throughout the project, a close collaboration has been maintained between WP2 and WP1.
The principal aim of Work Package 3 (WP3) was to develop an evidence based dengue outbreak warning and response system based on locally validated and applicable alarm indicators which trigger an initial, early or late response. As a precondition for this alert tool a well-developed surveillance system is required. Hence the essential components of different elements of dengue surveillance were explored and made available to countries through WHO and other partners.

The principal aim of Work Package 4 was to develop and generate a contemporary map of dengue occurrence. Modelling techniques were used to evaluate possible scenarios of spread or risk to previously uninfected regions including Europe and Africa.
The principal aim of Work Package 5 was to involve stakeholders and end-users in the research process from the beginning to ensure ownership and use of the research products of all workpackage findings.
Project Results:
Work Package 1

Work Package 1: Clinical evaluation of dengue and identification of risk factors for severe disease
(Lead investigators. Bridget Wills, Thomas Jaenisch)
• Aim 1: To identify factors which differentiate between dengue and non-dengue illness during the early febrile phase
• Aim 2: To identify factors among dengue infected patients that predict likely progression to a more severe disease course.
• Aim 3: To update the guidelines for “Integrated Management of Childhood Illness” used in dengue-endemic countries.
• Aim 4: To evaluate practical application of the original and the new dengue classification scheme (WHO 2009) across a series of clinical sites.

In Work Package 1 (WP1), patients with undifferentiated fever and possible dengue were recruited and prospectively followed daily throughout the course of their illness using a structured case report form (CRF) and a standardized protocol across all sites. The study is registered with ClinicalTrials.gov under the registration number NCT01550016 and the study design was published in 2016 (Jaenisch T et al: Clinical Evaluation of dengue and identification of risk factors for severe disease: protocol for a multicentre study in 8 countries. BMC Inf Dis, 2016).
Between October 2011 and June 2016 a total of 7428 patients were enrolled at 26 sites in eight countries across Asia and Latin America (Figure 1.1). Final monitoring/close-out visits were carried out at the various sites during the first half of 2016 and data entry was completed at all sites by August 2016. Comprehensive data cleaning and processing was carried out over the next 3 months, alongside derivation of definitions for all outcomes. For the purposes of this study we consider severe disease in two ways: first the development of severe dengue as defined in the WHO 2009 guidelines; and second in terms of the need for close medical attention, defined either by the need for hospitalization or requirement for parenteral fluid therapy. Since severe disease is a rare event we also defined intermediate severity endpoints for the three subcategories of severe dengue included in WHO 2009.

WP1, AIM 1: To identify factors which differentiate between dengue and non-dengue illness during the early febrile phase
The analysis population for this aim comprises 5177 patients with a) laboratory confirmed dengue (PCR/NS1 positive) or b) laboratory-confirmed non-dengue (other febrile illness), all of whom were enrolled within the first 84 hours from fever onset, with fever (>37.5 °C) measured or reported at enrolment or on study day 2.
The proportion of laboratory confirmed dengue infections varied considerably between the study sites and over time during the 5-year study period. This heterogeneity is due to different seasonality patterns and local outbreak situations. For the analysis presented here, 5110 patients with data from DOI 3 were included. In total, 32 candidate predictors were considered. Given the marked heterogeneity noted in dengue/other febrile illness (OFI) diagnosis that we noted, we included variables representing the continent (Asia/Latin America), and month of study (starting with month 1 in October 2011 and ending in May 2016 with month 56) in the diagnostic model in order to adjust for these differences.
In univariable analysis most of the variables were significantly associated with the differentiation between dengue and OFI. For the multivariable analysis a full model was built including all variables that were significantly associated with the outcome in the univariable analysis, plus two-way interactions for continent, sex and age with all other variables. Backward stepwise selection was applied in order to identify variables for a parsimonious model.
In order to take into account the variability of the selection process, the dataset was split into training data (2/3 of dataset) and test data (1/3 of dataset). The variable selection was performed on the training dataset. This step was repeated 500 times and only variables that were selected in at least 40% of the selection runs were kept in the final model.
The following variables were selected as predictors for dengue:
• month of study: starting in October 2011 (month 1) and ending in May 2016 (month 56) (modeled as spline term)
• absence of cough
• higher body temperature (modeled as spline term)
• lower platelet count (modeled as spline term)
• lower white blood cell count

plus an interaction term between
• continent and month of study

The final model was adjusted for continent, sex and age (modeled as spline). These variables were added to the model in addition to the predictors listed above.
The ORs for variables which are part of an interaction term (like continent and month of study) require careful interpretation. In order to estimate the effect for one of the interacting variables the other variable has to be set to a fixed value. The OR for continent was estimated for two fixed values of the interacting variable month of study: month 21 (corresponding to June 2013) and month 41 (corresponding to February 2015). For example, the OR of 0.54 indicates that in February 2015 the odds for being infected with dengue was 0.54 times lower in Latin America than in Asia. Month of study was modeled as a spline term. The OR is the estimated effect for an increase in the variable from month 21 (June 2013) to month 41 (February 2015). The estimated ORs are given separately for Asia and Latin America.
The area under the curve (AUC) is a measure for the performance of a diagnostic model. The highest AUC value of 1 would be reached for a model with 100% sensitivity and specificity. The multivariable model presented in Table 2 has a mean AUC value of 0.92. The mean AUC was obtained by repeatedly splitting the dataset into training and test data (as explained above). In each run the final model was fitted to the training dataset and the AUC was estimated in the test dataset.



WP1, AIM 2: To identify factors among dengue infected patients that predict likely progression to a more severe disease course.
In this analysis we aimed to identify risk factors among patients with dengue that predict likely progression to a more severe disease course, and also to develop prediction models for severe dengue which can incorporate longitudinal data. The analysis population included 2450 patients with laboratory-confirmed dengue, who were enrolled within the first 84 hours from fever onset and had sufficient data available to assess the occurrence of the clinical outcomes of interest.
There was considerable heterogeneity in the prevalence of clinical outcomes (for both hospitalization OR IV fluid, and severe OR intermediate dengue) between study sites, which can be explained partly by heterogeneity in immune status. In Bangladesh, the prevalence of all the outcomes of interest was very low compared to all other countries. Asian countries, except for Bangladesh, generally had higher prevalence of these outcomes compared to the Latin American countries. The proportion of secondary infection was also higher in the Asian countries, compared to Latin America. Within Latin America, the outcome hospitalization OR IV fluid was higher in Brazil compared to El Salvador and Venezuela. These heterogeneities still existed after adjusting for all potential explanatory factors, including immune status. However, there is no evidence that there is heterogeneity in the effects of predictors on outcome between sites.
To preserve the generalizability of results derived from this analysis, a pragmatic approach to incorporating these heterogeneities into the prognostic models was adopted, by including continent as a covariate in the model development step. However, re-calibration of the final prognostic models will be necessary to correct for the heterogeneity in outcome prevalence. Other sources of heterogeneity in this database are age and DOI of information. Such differences can be incorporated into prognostic models by including them and their interactions with other factors as covariates in the model development steps.
We started the analysis for AIM2 by developing prediction models for severe OR intermediate dengue based on data on DOI 3, using information at this single-time point only, with the outcome occurring at any subsequent point during the illness evolution. 2181 patients, of whom 367 [17%] were classified as severe OR intermediate dengue, had information on DOI 3 and were included in this analysis. In total, 31 candidate predictors were considered, of which 17 were identified as risk factors in the univariable analysis (Table 4).
The 7 factors that were identified as independent predictors in the multivariable analysis were:
• absence of chills
• vomiting
• lower total white blood count
• lower lymphocyte percentage
• lower monocyte percentage
• lower platelet count when the platelet count was below 200,000 cells/mm3
• younger age in Asia

Ongoing work includes development of the following models:
• prediction models for severe OR intermediate dengue based on data from DOI 3, using information at two time-points (information on the current day and the previous day)
• prediction models for severe OR intermediate dengue based on the data from DOI 1-2-3, using information at single and two time-points
• prediction models for short-term prediction of severe OR intermediate dengue after DOI 3 – i.e. prediction of outcome occurrence within the next 24 hours

Work Package 2

Work Package 2: Viral and immunological determinants of severe dengue
(Lead investigators. Cameron Simmons, Maria Guzman, Federica Sallusto, Adrian Hill)
Work Package Aims
• Aim 1: To identify virological correlates of severe dengue. Using validated assays, we will measure plasma viremia and NS1 antigenemia in samples obtained within the first 72 hours of fever from dengue patients enrolled in WP1.
• Aim 2: To identify early serological correlates of severe dengue. We will use a battery of assays to quantify serum antibodies specific for neutralizing and non-neutralising DENV epitopes and characterize differences between severe and uncomplicated disease evolutions.
• Aim 3: To validate and characterize the association between MICB and dengue. Using DNA samples collected from patients in WP1, we will examine whether the SNPs recently discovered to be associated with DSS among Vietnamese children are also associated with a) clinically apparent dengue and b) with the magnitude of viral infection as shown by viremia and NS1 levels.
• Aim 4: To develop the concept of a T cell-based vaccine against dengue. A T cell-based vaccine could protect against all dengue serotypes and avoid the risk of deleterious antibody-mediated enhancement

Aim 1: To identify virological correlates of severe dengue.
This aim is designed to identify early virological correlates of severe dengue. The platform for this work are the laboratory and clinical data of WP1. To this end, WP2 has performed laboratory diagnostic testing of all study participants enrolled in WP1 who had eligible clinical specimens. RT-PCR diagnostic tests were performed to identify dengue, chikungunya or Zika virus RNAemia. NS1 ELISAs (Platelia) were also performed to diagnose dengue. Serology was performed to enable classification of virologically confirmed dengue cases as either primary or secondary. The diagnostic test algorithm is shown in Figure 2.1.
The outcome of dengue diagnostic testing in Latin America is shown in Figure 2.2. In brief, the diagnostic yield, as a percentage and absolute number of virologically confirmed dengue cases, was highest in El Salvador. In Venezuela and Brazil, enrolment of participants occurred during the emergence of the Zika virus epidemic.
Amongst the lab-confirmed dengue cases, there was a strong geographical basis to serotype prevalence, with DENV-2 most prevalent in Venezuela, DENV-3 in El Salvador and DENV-1 in Brazil (Figure 2.3). Essentially monotypic transmission is typical of the epidemic cycle of dengue in much of Latin America.
In Asia, the outcome of dengue diagnostic testing is shown in Figure 2.4. The diagnostic yield was highest in Vietnam, followed by Cambodia, Malaysia, Indonesia and Bangladesh. Unlike in LA, there were multiple DENV serotypes in circulation in each Asian country (Figure 2.5). Only Bangladesh had a predominantly single serotype, DENV-1, accounting for most cases.
In line with the stated aims of this work package, we examined using multivariate logistic regression whether basic laboratory and virological features of DENV infection recorded at the time of enrolment (within 72 hrs of illness onset) were associated with clinical outcome. The outcomes of interest were hospitalisation or intermediate/severe plasma leakage. Preliminary results from the Vietnamese cohort of patients indicate the presence of NS1 antigenemia, a 1log10 increase in plasma viremia and a serological profile that was classified as secondary dengue were all significantly associated with hospitalisation amongst the population of dengue cases (Figure 2.6). Only secondary infection and plasma viremia at enrolment were associated with the outcome “intermediate/severe plasma leakage”. The platelet count at the time of enrolment was significantly associated with both hospitalisation or intermediate/ severe plasma leakage outcomes. Collectively these results reveal the role, the relative contributions and the estimated effect size on clinical outcomes of measurable virological variables. They reinforce the evidence base suggesting that the early magnitude of DENV viremia is associated with eventual disease severity – this has important implications for arguing the case for therapeutics and for the role of vaccines in eliciting immune responses that can limit viral replication. Further analyses of this dataset will explore the relationship between daily viremia levels measured in the Vietnamese cohort and clinical outcomes.

Aim 2: To identify early serological correlates of severe dengue.
Antibodies are central to the concepts of dengue pathogenesis and naturally-acquired or vaccine-elicited immunity. For example, the most prominent hypothesis to explain increased risk of severe disease during secondary infections is a process termed antibody dependent enhancement (ADE). The ADE hypothesis suggests that antibodies generated to a primary infection will not be of sufficient concentration or avidity to neutralize a secondary infecting DENV but can enhance viral entry to Fc-bearing cells leading to higher virus concentrations in infected tissues. ADE can be linked to the clinical epidemiology of dengue in infants and readily demonstrated in vitro and in vivo in animal models. In contrast, virus neutralizing antibodies are seen as important correlates of immunity and their induction by vaccination is the goal of all candidate vaccines in clinical development (REF). The only licensed dengue vaccine, Dengvaxia, was developed partly on the basis that it could elicit antibodies that in vitro neutralized DENV-1-4. Despite the centrality of antibodies to pathogenesis and immunity, laboratory correlates of immunity (or disease enhancement risk) have not been defined or agreed upon beyond the general consensus that higher concentrations of serum neutralizing Ab are associated with reduced risk of symptomatic infection. This uncertainty risks misleading assessments of a vaccinated populations’ immune status. The availability of evidence-based and agreed upon correlates of immunity (or disease risk) could enable the critical bridging studies that will fast track development of new dengue vaccines.
WP2 developed a novel viremic blood neutralization assay that allows identification and potency ranking of a large panel of human mAbs generated in the field. An example of one such potent mAb is 14c10, a DENV-1 specific mAb from a human donor. Partner 2 has shown that 14c10 is able to potently neutralize the infectiousness of DENV virions in the blood of viremic dengue cases (Figure 2.7). Antibodies such as 14c10 are of major interest as therapeutic candidates and as correlates of immunity in vaccine trials. Hence measurement of antibodies in human sera that can compete with 14c10 is of interest.

Serological assays optimised and characterised
We previously isolated antibodies against DENV from memory B cells of donors with primary or secondary DENV infection and characterized them for their neutralizing activity, breadth and specificity (Beltramello et al, PLoS Pathogen 2010). We are using these well-characterized antibodies as probes to determine the presence in serum of antibodies targeting the same sites. For this purpose, we developed a competitive ELISA (“blocking of binding” or BOB assay) to measure the titre of plasma that results in 80% inhibition of binding of an antibody of defined specificity. The assay is quantitative and highly specific for a limited region of a given antigen and it can detect all antibodies whose epitopes overlap with the epitope recognized by the probe antibody.
First, we selected DENV-specific IgG from the panel of monoclonal antibodies and cloned their genes into a vector encoding the murine Fc constant region. The “murinized” antibodies (mu-Abs) were produced in transiently transfected 293F cells and used to optimize the competitive ELISA. In addition, the genes encoding the variable region of IgG antibodies specific for the E protein were cloned into an expression vector encoding the human µ constant region. Table 2.1 shows the panel and amount of antibodies produced.
The recombinant antibodies were used to standardize a viral neutralization assay using different DENV strains and VERO cells as targets, and to determine the level of IgM in DENV patients' sera in competitive ELISA. In preliminary experiments, we used a mu-Ab (mu-DV82.11 that binds to DI, II of all 4 DENV serotypes) as probe to measure epitope-specific antibodies in the serum of DENV infected donors. As shown in Figure 2.8A the capacity of mu-DV82.11 to bind DENV was confirmed by intracellular staining of DENV-infected VERO cells. The concentration corresponding to 80% of the maximal OD level (dotted lines) was determined and used in competitive ELISA. The capacity of sera from two DENV-infected donors to inhibit binding of mu-DV82.11 was measured by intracellular staining of DENV1-infected VERO cells (Figure 2.8B). The inverse serum dilution that blocked mAb binding by 80% was 1121 for DON92 and 29.9 for DON33, indicating that antibodies binding the epitope recognized by DV82.11 can be readily detected in serum and are present in much higher concentrations in DON92 than in DON33.
The DENV and Zika E protein is formed by three domains: EDI, which is involved in the conformational changes required for viral entry, EDII, containing the fusion loop, and EDIII, which may be involved in binding to cellular receptors (30). DI, DII and DIII of ZIKV and DENV share 35%, 51% and 29% amino acid identity, respectively. According to the structure of the ZIKV E protein dimer the majority of the ZIKV and DENV conserved solvent-accessible residues are located in EDII, particularly in the fusion loop and the neighbouring region. To understand the role of antibodies in ZIKV neutralization and heterologous enhancement of flavivirus infection, we isolated a panel of 119 monoclonal antibodies (mAbs) from four ZIKV-infected donors, of which two were DENV-naïve (ZA, ZD) and two had serological records of DENV infection (ZC, ZB). These mAbs were selected from EBV-immortalized memory B cells based on their binding to ZIKV NS1 or E proteins and for their ability to neutralize ZIKV infection, and were compared, side by side, with a panel of mAbs previously isolated from DENV-infected donor. The mAbs were primarily IgG, were highly polyclonal and carried a lower level of somatic mutations compared to that of other acute or recurrent infections. To address the biological properties of ZIKV and DENV antibodies, we used a selected mAb panel and measured in parallel their neutralizing and ADE activity (Figure 2.9). This panel also contained Fc mutant versions of mAbs that do not bind to Fc and complement (LALA mutants). The EDIII-specific mAb ZKA64 was highly potent in ZIKV neutralization. Furthermore, this mAb enhanced infection of ZIKV in the non-permissive K562 cells at a broad range of concentrations, including those that fully neutralized ZIKV infection on Vero cells. Consistent with their cross-reactivity, these EDI/II-specific mAbs also enhanced DENV1 infection.

AIM 3: To validate and characterise the association between MICB and dengue.
Partner 2 has identified a single nucleotide substitution in the MICB gene that results in a stop codon mutation and hence a non-functional MICB gene. This stop codon mutation is associated with a doubling of risk (OR=2.01 P=0.001) for severe dengue in Vietnamese children.
In the current period, 2742 dengue cases were successfully genotyped at MICB rs3132468 and PLCE1 rs3740360. 573 cases were in adults (age > 15) and 2180 were in children. There were 109 cases of severe dengue across the cohorts (6 in adults and 103 in children). The mean viremia was compared between patients with the spectrum of MICB genotypes. The mean viremia in patients carrying the C/C allele was 7.16 (95%CI: 6.84-7.47) log10-copies/mL. In those with the C/T allele it was 7.05 (95%CI: 6.92-7.18) log10-copies/mL, and in those with the T/T allele it was 7.08 (95%CI: 7.01-7.15) log10-copies/mL. The difference between these values was not statistically significant. The comparison in viremia levels between MICB genotypes was repeated with data stratified by DENV serotype. Again, no significant differences were demonstrated. Collectively, these data indicate some variation in the viremia between the serotypes but no significant difference between the MICB genotypes.
The mean viremia was compared between the PLCE1 genotypes. The mean viremia in patients carrying the A/A allele was 7.08 (95%CI: 7.00-7.15) log10-copies/mL. In those carrying the A/C allele it was 7.09 (95%CI: 6.99-7.19) log10-copies/mL, and in those with the C/C allele it was 7.02 (95%CI: 6.78-7.26) log10-copies/mL. The difference between these values was not statistically significant. The comparison in viremia levels between PLCE genotypes was repeated with data stratified by DENV serotype. Again, no significant differences were demonstrated. The data indicate some variation in the viremia level between the serotypes but no significant difference between the PLCE genotypes.
The platelet nadir, the maximum haematocrit and the minimum white cell count were compared between the MICB genotypes in both an overall analysis and then stratified by serotype. The median platelet nadir in patients with the C/C allele was 82.5 (IQR: 47-101) x 109/L, in those with the C/T allele was 80 (IQR: 48-115) x 109/L, and in those with the T/T allele was 70 (41-113) x 109/L. These differences were not statistically significant. The minimum white cell count in those with the C/C allele was 2.6 (IQR: 2.3-3.1) x 109/L, in those with the C/T allele was 2.9 (IQR: 2.3-3.8) x 109/L and in those with the T/T allele was 2.7 (IQR: 2.1-3.4) x 109/L. These differences were not statistically significant. The median maximum haematocrit in patients with the C/C allele was 45.4% (IQR: 44-49.2%), in those with the C/T allele was 45% (IQR: 42-48.3%), and in those with the T/T allele was 44.7% (IQR: 41.8-48.4%). These differences were not statistically significant. When these data were stratified by DENV serotype again no significant differences were demonstrated. These data suggest no measurable association between MICB genotype and the laboratory variables explored in this analysis.
The platelet nadir, the maximum haematocrit and the minimum white cell count were compared between the PLCE1 genotypes in both an overall analysis and then stratified by serotype. The median platelet nadir in patients with the A/A allele was 71 (IQR: 41-109) x 109/L, in those with the A/C allele was 76.5 (IQR: 45-118) x 109/L, and in those with the C/C allele was 66 (37-114) x 109/L. These differences were not statistically significant. The minimum white cell count in those with the A/A allele was 2.7 (IQR: 2.1-3.4) x 109/L, in those with the A/C allele was 2.8 (IQR: 2.2-3.6) x 109/L and in those with the C/C allele was 2.9 (IQR: 2.2-3.6) x 109/L. These differences were not statistically significant. The median maximum haematocrit in patients with the A/A allele was 44.6% (IQR: 41.9-48.1%), in those with the A/C allele was 45.1% (IQR: 42-48.3%), and in those with the C/C allele was 45% (IQR: 42.4-50.3%). These differences were not statistically significant. When these data were stratified by DENV, serotype again no significant differences were demonstrated. These data suggest no measurable association between PLCE genotype and the laboratory variables explored in this analysis.

AIM 4: To develop the concept of a T cell-based vaccine against dengue
One challenge in developing a tetravalent DENV vaccine is the viral diversity, where low intra-type and high inter-type variability results in type-specific and type cross-reactive T-cell determinants. To define conserved regions within a sequence dataset, WP2 has generated a generic computer algorithm defining variability using pairwise comparisons within given window sizes. Subsequently, conserved segments under a set threshold were selected and combined using different immunogen design strategies; based on segment length, sequence conservation or number of conserved segments. We have used this algorithm to analyse the sequence diversity and to identify conserved regions in DENV (Figure 2.10). Vaccine immunogens were assessed using HLA (human leukocyte antigen) prediction programs to define potential strong binders to common HLA supertypes. The algorithm was used to design a new immunogen using the conserved regions of Dengue virus, which was cloned into recombinant viruses to assess immunogenicity.
The use of conserved viral segments from the non-structural proteins should provide protection against all serotypes and because our approach uses non-replicating vectors to induce T cells this poses no risk of antibody-mediated enhancement. This approach for dengue follows a strategy successfully followed at the Jenner Institute (UK) in other diseases.
Four arrays of Dengue-conserved sequences retrieved by our analyses were synthesised and cloned by Geneart (Invitrogen). Geneart plasmidic DNA, as well as the destination plasmid vector (pMono 2489) were expanded and digested with KpnI and NotI restriction enzymes. Size-specific DNA fragments were purified and ligated using T7 DNA Ligase (New England Biolabs). The resulting plasmids were expanded and verified by DNA sequencing.
i) Development of adenoviral and MVA vectors. To generate the adenoviral vectored vaccine, Attl sites in pMono-containing Dengue sequences and AttR sites in the destination adenoviral plasmid were recombined by using the Gateway LR Clonase (Life Technologies) and verified by PCR and DNA sequencing. To produce the MVA vectors, pMono-Dengue sequences plasmid, as well as an MVA-p434 plasmid were expanded and digested with KpnI and Xho I restriction enzymes. Size-specific DNA fragments were purified and ligated using T7 DNA Ligase (New England Biolabs). In total, we have developed 4 adenoviral and 4 MVA vaccines.
ii) Development of DNA vaccines to assess immunogenicity of the dengue sequences. Production of the viral vectors makes it possible to produce DNA vaccines to assess immunogenicity in mouse models. With this aim, plasmids pMono-expressing dengue antigens were expanded and DNA purified by using the Endo-free DNA isolation Mega prep kit (Qiagen).
We developed a novel insert comprised of conserved segments of the internal NS3-NS5 genes and expressed this in the simian adenoviral vector AdChOx1 (a group E simian adenovirus, like the AdCh63 vector used safely in malaria trials) and in the established boosting vector MVA. We undertook immunogenicity trials in mice.
Assessing immunogenicity of the vaccines by ex vivo IFN-gamma ELISPOT assay. Immunisation and tissue collection were performed following licensed procedures for animal handling. Groups of 6 outbred CD1 mice were immunised intramuscularly with 100μg/mouse of the DNA vaccines diluted in sterile PBS. A control, mock-vaccinated group received an empty pMono plasmid vector at a similar dose of 100ug/mouse. 15 days after immunisation, blood was taken by tail puncture. Peripheral Blood Mononuclear Cells (PBMCs) were isolated and cultured in the presence of peptide pools in ELISPOT plates previously incubated with an anti-IFN-gamma antibody (MAbTech). Peptide pool stimulation was performed by duplicate for each mouse. Positive controls for stimulation were included by exposing PBMCs to Ionomycin, PMA or both. Upon development, plates were scanned and number of spots forming cells (SFC) per million PBMCs were calculated and background of wells without peptides was substracted. Our results indicated that our construct was immunogenic in CD1 mice upon a single vaccination with the DNA vaccine expressing the conserved Dengue antigen sequences (Figure 2.11). These promising results demonstrate proof of concept that immunogenic vaccine constructs can be generated using conserved candidate T cell epitopes. Future studies will test whether T cell responses alone can provide protection in the murine model.

Work Package 3

In Work Package 3 (WP3), retrospective data were gathered by participating countries (Dominican Republic, Mexico, Brazil, Malaysia, Vietnam) using a collaborative data matrix, compiled by members of WP3 and WP5, to inform the data collection. These data were verified and validated at country meetings to ensure that known limitations were documented and that source data precisely corresponded to data present in each country data matrix. Concurrently, a statistical model, based on the Shewart Method, was constructed using STATA 13.0 and used for the analysis of these data. The primary aim of the model was to use ‘alarm’ variables to predict dengue outbreaks, based on historical mean values of the number of probable or hospitalized cases recorded during previous outbreaks. Once country data collection was complete, datasets were cleaned and organised by epidemiological week before they were input into the model. Many iterations of the model were devised, partly using test data alongside the retrospective country datasets. This provided the statistical team insight into the quality of the datasets, which was sometimes less than adequate, and interpret/understand the results of the model accordingly. Alarm variables were correlated during each historic period to an outcome variable using logistic regression. The resulting outbreak probabilities defined the thresholds above which an alarm was triggered during the evaluation phase. The final analysis revealed to what extent an alarm variable was able to predict an outbreak, and the accompanying sensitivity and positive predictive value of each alarm variable. Across countries where the outcome was either probable or hospitalised dengue cases, the most promising alarm variables included, but were not limited to: a relative change in the mean age; an absolute increase in mean temperature; an absolute increase in the number of reported probable cases.
Subsequently, an epidemiological study design was defined during IDAMS/WHO collaborative meetings in Freiburg, Germany and Liverpool, UK. These meetings sought to devise a study design to evaluate the model originally developed during the retrospective study, and hence test the predictive capacity of early warning or ‘alarm’ signals generated by that retrospective analysis. A cluster controlled study design was chosen to evaluate the model in three countries (Brazil, Mexico and Malaysia); in each country, 10 intervention districts and 10 control districts should utilise the new staged (intervention arm) or standard (control arm) dengue outbreak responses accordingly. Staged responses to alarm signals were determined using the time-dependent impact of a given intervention. Interventions that impacted with a lag time of >2-3 weeks were deemed appropriate in the presence of weak alarm signals (i.e. probability of outbreak = 20% or less). Equally, interventions that had a greater impact within a shorter time period (lower lag time) were deemed appropriate when the sensitivity and positive predictive values were known, as this would both reduce the probability of false alarms and increase the chance of successfully halting an outbreak. This method applied to vector control measures that impacted upon the circulating mosquito population, and also to the consequences of health system alerts for the operating capacity of health facilities.
To facilitate integration and ease of use in each participating district, an intelligible statistical tool used to evaluate the importance of each alarm signal were provided at a number of scheduled ‘kick-off’ meetings. During these meetings, standardized presentations and protocols were delivered to ensure that each country had the necessary training and capacity to use the statistical tool, and to respond adequately, proportionately and in a timely manner to each alarm signal. The prospective study was started in Malaysia, Brazil and Mexico in 2015 and the findings were discussed in August 2016 showing a high acceptance of the Early Warning and Response System (EWARS) by users and its ability to predict outbreaks in a timely manner.
The International Committee of the Red Cross collaborated with the WP3 team to produce a guidance document for use within the prospective study outlined above. This document drew on existing literature and increasing evidence to better define the role communities can play in combating dengue outbreaks and best practices for involving various stakeholders in the process. IDAMS/WHO meetings involving a number of national program managers (stakeholders) held at Liverpool, UK and Freiburg, Germany were used to determine the factors that influence community involvement. The overarching recommendations derived from the document built on the existing social mobilization approach and focused considerably on the following: building community trust within the pre-existing societal structures, developing and liaising with community working groups to increase participation, sustaining behaviour change, evaluating communication strategies, confirming existing and novel preventative strategies available to communities to limit the increase of mosquito abundance and transmission of dengue. This document formed the foundation of standardized protocols to be implemented at the community level in response to predicted imminent dengue outbreaks within the prospective study.
Work on evidence based surveillance systems started with systematic reviews critically examining current practices in dengue surveillance systems and the contribution of vector surveillance to dengue control and outbreak prediction; these were completed and made available to a wider audience, particularly country control programmes, through a web publication by WHO (Technical Handbook for Dengue Surveillance, Dengue Outbreak Prediction/Detection and Outbreak Response (“Model Contingency Plan”).

Work Package 4

In Work Package 4 (WP4), we first estimated the global distribution and burden of dengue (Bhatt S et al (2013): The global distribtion and burden of dengue. Nature 2013), after which we published a summary of the database used in making these estimates and an open-access link for its download (Messina, J. P. et al: A global compendium of human dengue occurrence: 1960-2012. Sci Data, 2014).). The results of the work are summarized in the form of 3 maps in Figure 4.1: (1) a map of evidence-based consensus on the definitive extents of dengue presence or absence at a national (and in some cases subnational) level; (2) a contemporary baseline map of the probability of dengue occurrence (from 0 to 100%) based upon known instances of human infection with dengue virus; and (3) a map of estimated burden from the disease, showing the number of yearly apparent infections at a national level.
This was followed by a study establishing the contemporary global patterns and history of spread for each of the four dengue virus types since first isolation of the virus in 1943 (Messina, J. P. et al: Global spread of dengue virus types: mapping the 70 year history. Trends Microbiol, 2014.). In this work, we provided spatially and temporally detailed maps of global dengue serotype patterns that were previously absent in the published literature. In so doing, we added to the understanding of how dengue has spread and will continue to do so in the face of urbanization, population growth, and changes in international travel patterns. We published work presenting maps of the global distributions of Aedes aegypti and Ae. albopictus vectors (Figure 4.2) (Kraemer, M.U.G. et al: The global distribution of the arbovirus vectors Aedes aegypti and Ae. Albopictus. eLife, 2015). Taking advantage of the results of this work, we contributed to another study looking into dengue distribution and expansion in Africa, a phenomenon, till recently, very much overlooked (Jaenisch et al., Dengue Expansion in Africa – Not Recognized or Not Happening, Emerging Infectious Diseases, 2014).

In order to plan mitigation strategies, it is essential that public health policymakers, vaccine developers, and vector control agencies be provided with estimates for the future global distribution of dengue. Therefore, these models were extended to account for urbanization and human movement. The maps, along with contemporary and projected climatic and socioeconomic data, were used to generate new high-resolution projections of dengue burden for the years 2020, 2050, and 2080, using the most exhaustive collection of dengue occurrence locations to date, a comprehensive set of current and projected socioeconomic and environmental covariates, and an ensemble species distribution modelling technique that allows for the quantification of uncertainty surrounding predictions. The results of this work, soon to be submitted (Messina, J.P. et al: The future global distribution and population at risk of dengue.) (such as illustrated in Figure 4.3) represent an important contribution to the evidence base available for tackling the growing threat of dengue and as an exemplar of other emerging Aedes-borne viruses.

ERGO (with personnel from WP4) continued to revise and update the covariate databases used for the spatial modelling of both dengue and the Aedes vectors – including the production of a time series of MODIS covariates and their Fourier transformed derivatives to provide a series running from 2000-2015 in collaboration with VMERGE (another FP7 project) (see main graphic in Figure 4.3). These were made available in the public domain to all registered members of the vmergedata.com website. In addition, following discussions with WP3 colleagues, ERGO provided datasets of covariate values used in the WP3 retrospective analyses of dengue outbreaks. These datasets comprised averages of several climatic parameters for administrative units in parts of Vietnam, the Dominican Republic, and Malaysia. The parameters included total rainfall (derived from daily global data layers) and mean daytime temperature and minimum relative humidity (derived from MODIS satellite imagery) for every 8-day period from 2007 to 2015 (see inset in Figure 4.4).

To aid in the understanding of the economic costs of dengue, a geographical analysis of disease treatment costs was done by combining (also with the help of ERGO expertise) county-level healthcare costs per person with the WP4-derived global distributions of dengue suitability, and public domain human population distributions to produce a first approximation of dengue-related costs per square kilometer, as presented at the final project meeting in Paris. Recognizing the critical importance of the quality of health systems and health-seeking behavior – especially in generating severity of, and mortality from, dengue – a cost adjustment was incorporated for health system and health-seeking ‘performance’ (Figure 4.5). The results pointed to important equity issues, and a link to WP1 (via the impact of improvements in case management on costs of dengue).

This work paralleled other work, involving informal collaboration with colleagues at the IVI (Dengue Vaccine Initiative), on new cost of illness evidence (both public and private costs, measured from onset of illness till recovery, covering inpatients and outpatients, children and adults) for Vietnam, Thailand, and Columbia (Lee, J-S. et al: A multi-country study of the economic burden of dengue fever, forthcoming in PLoS NTD). This too illustrated a significant equity issue (Figure 4.6).

Currently, there is no vaccine or causal treatment for dengue fever, and methods to reduce burden of dengue rely, in part, on early warning responses. Several papers have been produced on the cost effectiveness of early warning systems for dengue, starting with a background literature review, and country case studies, of the cost of dengue outbreaks (Stahl, H-C. et al: Cost of dengue outbreaks: literature review and country case studies, BMC Public Health 2013). Then, in the context of Colombia, and involving further collaboration with colleagues at IVI, work was performed Identifying Populations at High Risk (IPHR) and an early warning system for Colombia based on climate and non-climate datasets (Lee, J-S. et al: Early Warning Signal for Dengue Outbreaks and Identification of High Risk Areas for Dengue Fever in Colombia using climate and non-climate datasets, forthcoming in BMC Infectious Diseases). Figure 4.7 shows a map of the Climate Risk Factor (CRF) for Colombia. The climate and non-climate datasets have been cleaned up to expand the model into Thailand and Vietnam, with preliminary outcomes currently being generated.

The cost-effectiveness of early warnings and surveillance systems (EWS) for dengue in Brazil was analysed, using the SINAN database (national notifiable diseases information system), WHO unit costs adjusted to international dollars, I$, and the Oxford dengue risk map incidence number (Stahl, H-C. et al: Signalling disease outbreaks: cost-effectiveness analysis of early warnings and response systems in the case of dengue control, Antimicrobial Resistance & Infection Control, 2016). A decision-tree model including 3 response efficacies was constructed to assess the cost-effectiveness of an early warning system at a state-level for Brazil, and it was found that implementing Early Warning and Response Systems (EWARS) with a medium or a high efficacy showed efficiency gains, which is cost-effectiveness. On route, the per capita cost and total costs of dengue in Brazil were shown (Figure 4.8).

Several papers have been submitted, or are in progress, that involve transmission modeling. The first (Fitzpatrick, C. et al: An economic evaluation of vector control in the age of a dengue vaccine, WHO Neglected Tropical Diseases Investment for Impact Working Group, forthcoming in PLoS NTD), involving the WHO Neglected Tropical Diseases Investment for Impact Working Group, models the impact of sustained vector control in the context of vaccines. Just for illustration, we can see the response to no immunization (Figure 4.9) or to immunization on its own (Figure 4.10). The second, involving other Oxford colleagues, also involves modeling dengue transmission dynamics in the context of vaccines. This body of work is especially pertinent at the moment given huge uncertainty over dengue vaccines. The first dengue vaccine, by Sanofi Pasteur, is not proving as popular as had been initially suggested. Some of the countries that licensed the vaccine have stopped vaccinating people, and other vaccine candidates by Takeda and NIH are conducting their phase 3 trials. The highly heterogeneous spatial distribution of dengue—and now, we also know, of costs—greatly complicates vaccine decision making for both developers and policy makers. Careful future work, built on the foundations of the IDAMS results, is likely to be especially valuable in trying to work out if, when, and indeed ever, a particular vaccine should be used.

In 2015, the World Health Organization declared Zika a Public Health Emergency of International Concern. Being a closely related disease to dengue (both biologically and ecologically), we felt it was important to study its global distribution alongside that of dengue. As a result, we conducted species distribution modelling to create a global map of environmental suitability for Zika virus transmission to humans. The main publications (Messina, J. P. et al: Mapping global environmental suitability for Zika virus. eLife, 2016) showed a large portion of tropical and sub-tropical regions globally to have suitable environmental conditions for Zika, with over 2.7 billion people inhabiting these areas. The map from this paper (Figure 4.11) provides geographical information for the optimization of surveillance and the tailoring of public health guidelines and intervention strategies. The current status of Zika globally is also illustrated (Figure 4.12).


Work Package 5

Aim 1: To enable the scientific work packages (WP1 to WP4) through close interaction and coordination with network partners, Ministries of Health, International Organizations (WHO-TDR) and regional networks (ECDC, INDEPTH, International Red Cross) to receive INPUT, support and feedback for their research from potential users and interested parties and to achieve consensus on research design and interpretation of results.
To achieve aim 1, extensive interaction and networking with IDAMS partners and with major stakeholders in relation to dengue control and interventions were conducted. To these belonged:
• Information sharing among international organizations, particularly WHO headquarters in Geneva (NTD department, Emergency Response Group) and WHO regional offices (AMRO, SEARO, WPRO), Red Cross International in the Netherlands, INDEPTH network in Africa
• Close interaction with other WPs, particularly WP3, WP4 and WP1 through meetings in Geneva and elsewhere, electronic communication and teleconferences
• Information sharing with Ministries of Health in 10 selected countries of Asia and Latin America;
• Intensive face to face interaction with national dengue programs in Brazil, Mexico, Dominican Republic, Vietnam, Malaysia and later in Peru;
• Organization of discussion groups on a) cost assessment of dengue outbreaks; b) surveillance systems; c) climate and dengue;
• Transferring the in-country analyses regarding experiences with surveillance, dengue outbreak detection and response into a framework for stakeholder consideration and interpretation;
• Organization of international stakeholder meetings to a) discuss country experiences, b) develop recommendations regarding standardized definitions of key terms such as dengue outbreak, alarm signals, early and late response, monitoring and evaluation tools; c) to outline key elements of a model contingency plan for dengue outbreaks and d) consider the findings of the retrospective study for the validation of alarm signals and feasibility of a model contingency plan; e) support the prospective analysis of improved dengue surveillance;
• Post-meetings interactive work through Skype and e-mail on joint reports and subsequent publications.
In a number of countries, these activities led to a re-assessment of existing surveillance and outbreak response plans and to the development of better tools. The cooperation with WHO regions and countries was excellent and led to a strong cooperation towards improved and standardized dengue surveillance and a growing interest in validated alarm signals for dengue outbreaks to be later on used in national plans. The interest was high, to use IDAMS methodologies for better mapping dengue (WP4), for taking into account the evidence collected in WP1 for validated clinical warning signs for severe dengue and in WP3 on the characteristics of adequate dengue surveillance, the costs of dengue outbreaks, the importance of climatic factors, and the best options for dengue vector management before and during outbreaks.

Aim 2: To facilitate the OUTPUT transfer of scientific work package findings into policy and practice of end-users by enabling:
- The development of cost-effective dengue surveillance and outbreak response adapted to specific country needs in cooperation with national and international partners
- The process of updating the current dengue guidelines with material relevant to the management of dengue in children, adolescents and adults (WP1).
To achieve aim 2, the main activities were an extension of those described for aim 1:
• Intensive networking with major stakeholders;
• Participation of WP coordinators in the international stakeholder meetings to provide their inputs
• Preliminary recommendations for dengue surveillance were developed together with stakeholders and preliminary recommendations for in-country implementation were discussed (particularly in Brazil, Mexico and Malaysia)
• Use of individual WP findings and recommendations (particularly WP1, 3 and 4) by international organizations were prepared through meetings in Geneva, Freiburg, Liverpool, Ghana and Paris
• The Paris meeting in December 2016 was the final stakeholder/dissemination meeting held together with other EU-funded dengue consortia (DENFRAME and DENTOOLS). It was organized by the WP5 team together with other WP leaders. (Publication in preparation)
• Substantial contribution to the “Global Dengue Strategy 2012 to 2020” designed by expert groups under the coordination of WHO (NTD Department)

This work enabled the WP5 team to submit the consensus documents on a) the research process and b) policy recommendations (Deliverable 5.1).
The main achievements of WP5 were a) to make substantial inputs into the global WHO strategy for dengue prevention and control; b) to motivate countries (particularly Brazil, Mexico, Malaysia and Vietnam) to re-think their dengue control strategies and prepare corrective actions; c) to bring the different WP teams together and share information and lessons learned; d) to engage WHO and other stakeholders in standardizing procedures and taking advantage of IDAMS findings.

Aim 3: To assure high quality research through trained external monitors, review meetings and corrective actions.
In order to reach aim 3 individual work package managers (WPs 1-4) were asked to document their quality assurance measures over the research process in a standardized way, with the aim to document the wealth of quality assurance measures tailor made for the requirements of the different research packages. The QA mechanisms included SOPs (Standard Operational Procedures), application of standardized research tools, monitoring visits, supervised data management processes and others. WP5 undertook those activities including the monitoring visits for WP3 and summarized all quality assurance activities in a joint summary report across WP 1-4.

Potential Impact:
The IDAMS observational prospective multicentre study (Work Package 1) is the most comprehensive clinical assessment of dengue ever performed. The protocol and data collection format were identical at all sites across 8 dengue-endemic countries making the various outputs broadly generalizable globally. Work is ongoing to refine the existing diagnostic and prognostic algorithms, in particular to develop novel prediction models that incorporate longitudinal data. We anticipate that these algorithms based on signs, symptoms and readily available laboratory tests will be of great practical utility in endemic settings, resulting both in improved case management and more appropriate use of limited resources. The availability of such a large well-characterised patient dataset accompanied by serial plasma samples, provides a unique and valuable resource for additional studies, including assessment of potential biomarker panels, and evaluation of proposed clinical endpoint definitions for use in research studies. In a parallel and highly synergistic effort (sponsored by the National Institutes of Health [NIH, USA] and the Partnership for Dengue Control) WP1 researchers (Bridget Wills & Thomas Jaenisch) are involved to define research endpoints for severe dengue.
Within WP1, two expert meetings were organised. The first one was organised together with the IDAMS partner INDEPTH in Accra, Ghana, in 2013. Stakeholders from WHO, CDC, ECDC, Pasteur Institute, and African INDEPTH sites were invited to discuss the situation of ‘Dengue in Africa‘ and the results including a recommendation on research priorities were published in 2014 (Jaenisch et al., Dengue Expansion in Africa – Not Happening or Not Recognized, Emerging Infectious Diseases, 2014).
A second expert meeting (September 2015) was organised in Ho Chi Minh City, Vietnam, in order to appraise key stakeholders from Asian countries of the activities being undertaken and the potential results/expected timeline of outputs from the IDAMS study. The meeting was attended by more than 40 participants from 13 countries in the region. At this meeting the usefulness of the “Integrated Management of Childhood Illness” (IMCI) guidelines and the potential need to revise and update these guidelines was discussed extensively. Experienced clinicians from the dengue-endemic countries in the region presented on their current diagnostic algorithms and the lessons to be learned for an update of the IMCI guidelines. This lead to a publication where the need of an update of the IMCI guidelines was emphasized (Deen J. et al: Dengue in the Context of the Integrated Management of Childhood Illness. PLoS Negl Trop Dis, 2016).
The virological and serological investigations conducted (Work Package 2) will be regarded as major advances in the field because of their large sample sizes and multi-country nature. There will also be practical outcomes. For example, the confirmatory discovery that early viremia levels are associated with disease severity is already being used as underpinning science to support the development of antiviral drugs that can be used therapeutically or prophylactically to improve patient outcomes. Another key discovery was the development of a viremic blood neutralisation assay that enabled us to score the potency of a panel of anti-DENV human monoclonal antibodies with different epitope binding characteristics. The results have implications for vaccine development, therapeutics and the advancement of a new generation of immune correlate assays.

The systematic reviews conducted in WP3 are now widely quoted in the literature and referenced in WHO guidelines. They are often the starting point for new research questions leading to a broader evidence base for vector control, surveillance and early outbreak warning.
The Early Warning and Response System for dengue outbreaks is the first of its kind and more advanced than outbreak alert systems developed for influenza, malaria, rift valley fever, cholera and others. It is particularly attractive because it is based on local (district) data and can be managed by district medical officers. It overcomes the confusion between an “outbreak” and a “seasonal peak” and it provides a measurable threshold for the start and end of an outbreak. Its particular value is in the outbreak prediction but also in bringing staff from different sectors (Ministries) together to deal with dengue outbreaks. This has been repeatedly mentioned by country control programme officers who have used the EWARS tool.
Due to the appearance of Chikungunya and Zika outbreaks during the prospective test phase of the EWARS tool, the control districts and intervention districts were “contaminated” with emergency control measures so that the reduction of case numbers, staff efforts and costs could not be determined. However, the before–after comparison (before and after the EWARS implementation) showed clearly the increased efficiency of control staff in responding to outbreak alert and to strengthened team work among different actors. Ongoing promotion through WHO-HQ and Regional Offices will help to disseminate the new tool quickly.
The numbers generated by WP4 have already found uses across the spectrum, including an update of the global burden of disease numbers for dengue. GAVI, the Vaccine Alliance, also requested maps of dengue burden numbers for the countries it operates. Currently, there is no great likelihood that GAVI, and significant funders such as the Gates Foundation, will have an appetite to subsidise the first generation of dengue vaccines. Indeed, given the risks, many countries where the first-generation vaccine has been licensed have eased off on their enthusiasm for its wide application. This all suggests that for some time to come, great care will be needed to work out the best way to precisely target interventions, and that geographically-detailed burden data will be extremely useful in thinking about strategies for using any new interventions, such as vaccines and vector control. There has been much recent modeling of dengue vaccine strategies, and an increasing realisation that there is no immediate technological panacea, given, for example, the challenges of using vaccines on seropositive individuals, and yet the extreme difficulty of tracking the sero-status of potential vaccinees.
Those working on WP4 have found the IVI, and WHO is keen to use the numbers to help formulate strategies across interventions, including vector control, as well as vaccines. With the horizons for potential future interventions stretching, it is likely the results will be especially valuable for long-term planning; this could also include guiding investments, and new public-sector and foundation funding, into second-generation vaccine concepts. It is also likely that the goal of interest will be a multiple-disease approach, such as vector control that impacts a range of diseases and not just dengue, for which more timely and precise data will be essential. With efforts to keep climate change on the agenda and to push for action, the importance of the work on dengue projections and future modelling can only increase.
Meanwhile, the global push for a new set of Sustainable Development Goals, coupled with the interest of key players, such as the World Bank, in universal health care coverage, has put renewed emphasis on issues of health equity, which, it turns out is especially important in the case of dengue. It is also highly likely that the results of the project will find a use in the embryonic work developing early warning systems and financial insurance instruments to protect against health emergencies.


The main outcome of the stakeholder-work by the WP5 team was to draw the attention of country control programs to the different aspects of dengue: clinical, virological-immunological, burden and surveillance/outbreak alert. All aspects are important for a comprehensive dengue control programme at country level and beyond. But also other international agencies benefitted from the work, particularly UNICEF and Interamerican Development Bank (now joining into the outbreak detection work), NIH and WHO interested in dengue classification, WHO taking advantage of the burden of disease studies and vaccine and diagnostic developers taking notice of the virological-immunological work.
This was achieved through intensive stakeholder- and networking activities: The WP5 team organized several international stakeholder meetings (2011, 2012, 2013, 2014, 2015) in a central location (Freiburg, Germany), bringing together participants from Ministries of Health, in-country academic institutions and international organizations such as WHO (including their Regional Offices), the International Red Cross, ECDC-Stockholm/Sweden and partners across the IDAMS network. The interim results of WP1, 3 and 4 were presented and discussed. Additionally, several satellite meetings were organized during the reporting period: Vietnam Malaysia, Brazil, Dominican Republic, UK (Liverpool) and Cuba. The joint analysis of country experiences involving major stakeholders made clear that country information on dengue is based primarily on obligatory notification and reporting (“passive surveillance”), with laboratory confirmation (in all Latin American countries and some Asian countries) or by using a syndromic definition. Few countries had sentinel sites with active dengue reporting, though some used virological surveillance. The subsequent analysis of long term surveillance data in a large number of districts per country (see WP3 report) underlined the deficiencies of national surveillance systems, highlighting many problematic areas: missing data, excessive data aggregation, reporting key indicators (e.g. changes in serotype) only once per month or even per year, all of which limited the usefulness of surveillance data for predicting dengue outbreaks. This meant that only districts with a more complete data set could be included in the analysis of alarm signals for dengue outbreaks. The deficiencies of surveillance data were extensively discussed with Ministries of Health and the country programme officers. It is imperative that the information describing the quality of data that is needed for outbreak prediction is disseminated to other dengue endemic countries. During the stakeholder meetings, country representatives noticed a considerable variation between countries with regard to surveillance, outbreak detection, and response. Through discussion and consensus building, suggestions were made for developing a more standardized approach using the model contingency plan developed by the IDAMS team.
As a view to the near future: The global response to the imminent dengue pandemic requires close co-operation between all parties involved, from local healthcare, surveillance and public health personnel, to regional and international strategic bodies, if timely and effective control is to be achieved with the resources available. As part of this programme of work, new alliances must be formed between series of hospitals in endemic countries to establish a clinical network that will work together with WHO/TDR and established surveillance and public health agencies such as RCCC, INDEPTH and ECDC. Development and strengthening of these links will stimulate ideas for future networking and collaboration. As one example, we expect that interactions and discussions between senior clinicians will lead to recommendations for standardization of care across the network of clinical sites, a move that is likely to result in better overall quality of care, and also preparing the participating sites for clinical trials of future therapeutic interventions for dengue.


List of Websites:
www.idams.eu;