Start here: what dengue is
Dengue is a viral infection carried between people by mosquitoes — chiefly Aedes aegypti, a day-biting, city-dwelling mosquito that breeds in the water containers around human homes. The virus, DENV, is a positive-stranded enveloped RNA flavivirus, and the disease it causes has been reported from more than 100 countries across the tropics and subtropics [1].
Most infections are mild or silent. The classic illness is an abrupt fever with a headache behind the eyes, a rash, and the deep muscle and joint pain that gave dengue its other name, breakbone fever. A small fraction of patients take a different course: as the fever breaks, plasma begins to leak out of the capillaries, the platelet count falls, and the circulation can collapse. That is severe dengue — historically split into dengue haemorrhagic fever and dengue shock syndrome — characterised by thrombocytopenia, leucopenia and increased vascular permeability [1].
The fact that organises everything else: there are four of them. DENV has four antigenically distinct serotypes, DENV-1 to DENV-4 [1]. Infection with one gives lasting immunity to that serotype and only brief cross-protection against the others. Worse, in a way developed below, prior infection with one serotype can make a later infection with a different serotype more dangerous rather than less. Everything difficult about dengue — the prognosis, the vaccines, the trial design — follows from those four numbers and the immunology between them.
Scale. Over the past fifty years dengue incidence rose roughly 30-fold, reaching approximately 390 million infections in 2010 [2]; more recent syntheses put it at around 400 million cases and 22,000 deaths each year [1]. Dengue now affects over half the world's population by geography [3], and it is the world's most important arboviral disease by number of people affected [2].
Why it is in the news. Because the map is moving. Globalisation, trade, travel, urbanisation and warming temperatures are all associated with the recent spread of Aedes aegypti and Aedes albopictus and of the virus with them [2] [4] [5]. Brazil recorded its greatest dengue epidemic [6], with high-altitude regions that had previously acted as a barrier to transmission becoming high-incidence areas [7]; Latin America's outbreak has been framed as a global health threat [8]. And transmission has begun appearing where it did not: the first two autochthonous — locally acquired, not imported — dengue infections in metropolitan France were reported in September 2010 [9], and vector surveillance is now an active concern in the United States and its territories [10].
Its siblings in this series. Dengue sits between two reviews already here. Valley fever is the other climate-expanding infection in this collection — there the climate acts on a soil fungus and its dispersal; here it acts on a mosquito's metabolism. Malaria is the other mosquito-borne disease — a parasite rather than a virus, a night-biting Anopheles rather than a day-biting Aedes, but the same thermal physics governs both, and the same modelling framework covers them together [11] [12].
Three pillars follow — measurements, medicines and prevention, and progress — with a temperature model between the first two that explains why warming does not simply mean more dengue everywhere.
Pillar 1: measurements and diagnosis
Which serotype, and have you met one before
The single most consequential piece of information about a dengue patient is not the serotype they have now but whether they have had dengue before.
Antibody-dependent enhancement (ADE) is the mechanism. Antibodies raised against one serotype bind a different serotype without neutralising it; the antibody-coated virus is then taken up more efficiently into cells bearing Fc receptors, so a partially matched immune response increases infection rather than blocking it [13] [14]. This had long been suspected but was hard to demonstrate outside the laboratory. It was shown in humans using a long-term paediatric cohort in Nicaragua, and the shape of the finding is the important part: the risk of severe dengue is highest within a narrow range of pre-existing anti-DENV antibody titres, while at high titres there is protection from all symptomatic disease. The authors' conclusion is precise and worth restating — the immune correlates of severe dengue must be evaluated separately from the correlates of protection, because enhancement, and not merely absence of protection, is what is at stake [15].
That is a genuinely unusual dose-response: too little antibody and you are unprotected; the wrong amount and you are worse off than naive; enough and you are protected. It is why secondary heterotypic infection carries elevated severe-disease risk [1] [16], why dengue and Zika cross-reactivity matters [17], and — as Pillar 2 shows — why a vaccine can harm the people it was supposed to help.
Serotype and immune status jointly shape severity; nineteen years of paediatric clinical studies in Nicaragua have now been analysed on exactly that two-way split [18].
Finding the virus, and finding the antibodies
Three test types, used at different points in the illness [19] [20].
RT-PCR detects viral RNA and is definitive during the first few days of fever, while viraemia lasts. It is what the vaccine trials used to confirm cases [21].
NS1 antigen is the practical workhorse. NS1 is a dengue non-structural protein secreted into the blood during the acute phase, and detecting it directly was first demonstrated for DENV-1 [22]. Its kinetics depend on immune status and serotype [23], rapid tests and ELISAs have been evaluated against clinical samples [24], and NS1 levels themselves have been examined as a marker of severe disease [25].
IgM and IgG serology arrives later and answers the other question — has this person been infected before? An anti-NS1 IgG ELISA is exactly the assay that was used to reconstruct participants' baseline serostatus in the Dengvaxia reanalysis [26], and serostatus testing is now a prerequisite for one of the two licensed vaccines.
Warning signs, and the WHO classification
Dengue's danger is concentrated in a narrow window. The febrile phase ends and the critical phase begins — typically around defervescence — and it is then that plasma leakage happens. The clinical task is therefore prediction, not diagnosis.
The WHO revised its case classification from the older dengue fever / dengue haemorrhagic fever / dengue shock syndrome scheme to dengue, dengue with warning signs, and severe dengue, precisely to make triage actionable. The revision has been evaluated in multicentre prospective studies across four South-East Asian and three Latin American countries [27], in an 18-country applicability study [28], and against the traditional scheme [29], with a continuing argument about whether it needs further modification [30]. Severe dengue turns on severe plasma leakage, severe bleeding, or severe organ impairment.
The measurements that matter in that window are simple and serial: haematocrit, which rises as plasma leaks out of the vasculature and concentrates the red cells; platelet count, which falls; and clinical signs of fluid accumulation. A systematic review and meta-analysis has assembled the risk predictors of progression to severe disease during the febrile phase [31]. Serum chymase — a mast-cell protease — correlates with severe dengue warning signs and clinical fluid accumulation in hospitalised paediatric patients, an example of the biomarker search running alongside the clinical signs [32].
Centerpiece: a simple simulatable model of temperature and transmission
The mosquito is an ectotherm and so is the virus inside it. Every step that has to happen for transmission to occur — the mosquito biting, surviving long enough, the virus replicating and reaching the salivary glands — has a rate that depends on temperature, and none of them increases forever. Below some temperature the virus never completes its extrinsic incubation before the mosquito dies; above some temperature the mosquito itself dies too fast. Transmission is therefore a hump.
The thermal-biology literature has quantified this across pathogens. Synthesising 11 pathogens transmitted by 15 mosquito species, transmission "varied strongly and unimodally with temperature, peaking at 23–29 °C and declining to zero below 9–23 °C and above 32–38 °C" — and among tropical pathogens, dengue and Zika had the highest thermal optima, at 29 °C [11]. For the Aedes-borne viruses specifically, mechanistic models and human case data agree that transmission occurs between 18 °C and 34 °C, with maximal transmission in a range from 26–29 °C [33].
Model that as an asymmetric thermal-performance curve pinned by three published temperatures — the lower limit, the optimum, and the upper limit:
R(T) = ((T_max − T)/(T_max − T_opt))^k · ((T − T_min)/(T_opt − T_min))^(k·(T_opt−T_min)/(T_max−T_opt))
zero at both limits, exactly 1 at the optimum, by construction.
What the model explains. Four things.
First, why "warming means more dengue" is too simple. The curve has a peak. Warming increases transmission below the optimum and decreases it above [11] — so the same two degrees that open a new season in Europe can shave the peak off transmission in the hottest already-endemic places. The right panel shows both happening at once. The projected global picture is accordingly a redistribution, not a uniform increase: poleward shifts, with substantial risk increases for most of Europe, alongside significant reductions in climate suitability expected for Ae. albopictus in parts of southeast Asia and west Africa [34].
Second, why temperate outbreaks stay small. Even where the vector is present, temperate transmission windows are short — at most about three months a year [33]. A short window limits the number of transmission generations after an introduction, which is why imported cases in temperate zones usually produce a handful of local infections rather than an epidemic. The French autochthonous cases in 2010 were exactly that shape: two [9]. The model's warning is that lengthening the window is precisely what warming does.
Third, why altitude stopped protecting people. High-altitude regions of Brazil that had previously acted as a barrier to dengue transmission became areas of high incidence, with prolonged temperature anomalies identified among the main factors [7]. On this curve, altitude is just a way of being cold; remove a couple of degrees of protection and the barrier is gone.
Fourth, why the two vectors behave differently. Ae. albopictus has a cooler and narrower thermal window than Ae. aegypti — 19.9–29.4 °C against 21.3–34.0 °C [34]. That is why albopictus has been the species colonising Europe and North America [4] [35] [36], and why the two vectors' ranges will move differently under the same warming.
What the model deliberately does not do. It is not R₀. Real vectorial capacity is a product of separately temperature-dependent traits — biting rate, vector competence, adult survival, egg production, the extrinsic incubation period — each with its own thermal response [11] [37] [38] [39]. It uses mean temperature, when daily temperature fluctuation around a mean measurably changes transmission [40]. And it contains no mosquitoes, no people and no immunity: vector abundance, urbanisation [5], human movement [41] and the local serotype history all matter, and multi-model intercomparisons exist precisely because no single model settles this [42] [3] [43].
Pillar 2: medicines and prevention
There is no antiviral, and the mainstay is fluid
This has to be said first because it is the honest position. There is no specific antiviral drug for dengue. Chloroquine was tested in a randomised controlled trial in Vietnamese adults and did not deliver [44]. Treatment is supportive.
Supportive does not mean passive. The mainstay is careful fluid management through the critical phase: enough intravenous fluid to maintain the circulation while plasma is leaking out of the capillaries, and — equally important — not more than that, because when the leak reverses the fluid comes back into the circulation and overload becomes the danger. Getting that judgement right, on serial haematocrit and clinical assessment, is what converts a potentially fatal illness into a survivable one, and it is why case-fatality in well-run settings is low despite the absence of any drug that touches the virus. Antibody-based therapeutics are in development [45] [46], and potent broadly neutralising human monoclonals against conserved envelope epitopes have been characterised [45], but none has replaced fluid management.
Vaccines, and the cautionary tale
Dengue vaccine development has to solve a problem no other vaccine faces: it must protect against all four serotypes simultaneously and evenly, because leaving someone partially immune is not neutral — it can put them into the enhancing range identified above [15] [47] [48].
Dengvaxia (CYD-TDV) is the object lesson. A recombinant live-attenuated tetravalent vaccine, it showed efficacy in phase 3 trials in Latin America and Asia [49], was licensed, and was then reanalysed after excess dengue hospitalisations appeared among younger vaccine recipients. Because baseline samples were limited, investigators developed an anti-NS1 IgG assay and inferred serostatus retrospectively. The result split cleanly by prior exposure. Among dengue-seronegative participants aged 2–16, the cumulative 5-year incidence of hospitalisation for virologically confirmed dengue was 3.06% among vaccine recipients against 1.87% among controls — a hazard ratio of 1.75 (95% CI 1.14–2.70). Among seropositive participants of the same ages it was 0.75% against 2.47%, a hazard ratio of 0.32 (0.23–0.45) [26]. The vaccine protected people who had already had dengue and behaved like a first infection in those who had not.
The consequences were immediate: WHO's Strategic Advisory Group of Experts revised its recommendation to pre-vaccination screening [50] [51], US recommendations followed the same logic [52], and modelled public-health impact and cost-effectiveness had to be recomputed against seroprevalence [53] [54]. It is also a methodological warning that outlasts the product: phase 3 dengue efficacy estimates can be biased by heterogeneous exposure and by differential detectability of primary infections across trial arms [55], and neutralising antibody titres are an imperfect correlate [56] [46].
Qdenga (TAK-003) is the second-generation answer. In the phase 3 TIDES trial, 20,071 children and adolescents aged 4–16 across endemic regions of Asia and Latin America received two doses three months apart. Overall vaccine efficacy against virologically confirmed dengue was 80.9% (95% CI 75.2–85.3), with 95.4% efficacy against dengue leading to hospitalisation — and, critically, 74.9% efficacy in the 27.7% of the per-protocol population that was seronegative at baseline [21]. Two-year [57] and 4.5-year results followed: cumulative efficacy 61.2% against virologically confirmed dengue and 84.1% against hospitalised dengue [58]. Efficacy varies by serotype, which remains the honest caveat [21] [59]. It is now WHO-recommended for endemic areas [60] [61], with practical guidance for clinicians [62] and for deployment into routine immunisation [63].
Vector control, and Wolbachia
Conventional Aedes control — insecticides, source reduction — has typically had limited effectiveness in preventing outbreaks [2], and insecticide resistance is an active problem [64].
Wolbachia is the striking exception, and it works by an unusual route: rather than killing mosquitoes, it makes them poor hosts for the virus. Wolbachia is an inherited bacterial endosymbiont of insects; introduced into Aedes aegypti, it limits infection with dengue, chikungunya and Plasmodium [65] [66] [67], through mechanisms including reactive-oxygen-dependent Toll-pathway activation [68], and it spreads through wild populations on its own because of the reproductive advantage it confers [69] [70].
The definitive test was AWED, a cluster-randomised trial in Yogyakarta, Indonesia. Twelve clusters received wMel-infected mosquito deployments and twelve did not; 8,144 participants were enrolled under a test-negative design. Virologically confirmed dengue occurred in 2.3% of participants in intervention clusters against 9.4% in control clusters — an odds ratio of 0.23 (95% CI 0.15–0.35) and a protective efficacy of 77.1% (65.3–84.9), similar against all four serotypes. Hospitalisation for dengue fell further: 86.2% protective efficacy (66.2–94.3) [71].
That result — a randomised trial, four serotypes, no drug and no vaccine — is among the strongest in the field.
Pillar 3: progress
Wolbachia at scale
The question after AWED was whether a cluster-randomised result survives being turned into a public health programme. So far it has. Large-scale releases established wMel across the Colombian cities of Bello, Medellín and Itagüí [72]; city-wide releases in Niterói, Brazil have now been followed for long-term durability and public-health impact through a dengue epidemic [73]. A different strain, wAlbB, was chosen for Malaysia specifically because it is stable at high temperatures, which matters for hot tropical climates — and it delivered an average 62.4% reduction in dengue across 20 release sites against 76 control sites, rising to an estimated 75.8% reduction at 100% Wolbachia frequency [74]. Modelling of the Indonesian scale-up found it cost-effective [75].
The temperature-stability point connects directly to the model above: an intervention deployed into the hot end of the curve has to survive the same heat that limits the mosquito.
Next-generation vaccines
A third vaccine has reached long-term readout: the single-dose tetravalent Butantan vaccine has reported long-term efficacy and safety results [76]. Because a single dose removes the multi-visit problem that limits coverage in endemic settings, this matters operationally as much as immunologically.
Beyond that, the pipeline is explicitly organised around the ADE problem: live-attenuated candidates designed to lack antibody-dependent enhancement [77], virus-like-particle strategies [78], and reviews of where the field should go next [79] [80] [81] [82]. Estimated efficacy against asymptomatic infection — which matters for transmission rather than for the individual — has been analysed for TAK-003 [83]. Vaccine hesitancy, unsurprisingly given the Dengvaxia history, has its own literature [84] [85].
Where the virus meets the host
Mechanistic work continues on both sides. Dengue pathogenesis and the host molecular machinery it exploits have been mapped [86] [87] [16], including the immune-evasion strategies flaviviruses use against interferon signalling — the same JAK/STAT antagonism whose deliberate mutation is a route to live attenuated vaccines [88]. RNA sensing by the RIG-I-like receptors is where that battle begins [89] [90] [91].
The forecasting problem
Predicting where dengue goes next is now a modelling discipline in its own right, combining climate, population and socioeconomic projections [3] [34] [42], vector distribution mapping [4] [92] [36], and the seventy-year history of how serotypes spread [93] [41] [94]. Even social-media signals have been tested as nowcasts [95]. The burden trend among working-age adults has been quantified separately [96], and the escalating burden reviewed as a policy problem [97] [98] [99] [100].
Dig deeper in lmmol
Dengue belongs to two families of disease in this collection, and reads best against both:
- Valley fever — the climate-expanding sibling. There the warming acts on a soil fungus and the dust that disperses it; here it acts on a mosquito's metabolism. Both reviews centre on a model in which climate sets the envelope and the pathogen fills it.
- Malaria — the vector-borne sibling. A parasite in a night-biting Anopheles rather than a virus in a day-biting Aedes, but the same thermal-biology framework covers both, and malaria's optimum sits several degrees cooler than dengue's [11] [12].
- Measles — the counterpoint on vaccines. Measles has one serotype and a vaccine that produces sterilising immunity; dengue has four and an immune response that can make the next infection worse. The contrast is the whole reason dengue vaccine development took so long.
- Tuberculosis — the other infectious review here where prior immunity is central and a partially effective vaccine is the live question.
- CKDu — the third climate-linked review in this series, and the one where heat acts on the patient rather than on a vector.
- The health reviews index collects the rest of the series.
Then move down into lmmol's graph:
- Chymase — the mast-cell protease whose serum levels track severe-dengue warning signs and clinical fluid accumulation, i.e. the plasma leak itself [32].
- RIG-I and MDA5 — the cytoplasmic RNA sensors that detect an incoming flavivirus genome and start the interferon response [89] [91] [90].
- JAK1 and TYK2 — the kinases of the interferon-induced JAK/STAT pathway that flaviviruses antagonise, and whose deliberate disruption is a design route for live attenuated vaccines [88].
- For entities without a linked static page here, use the graph index, all proteins, or all diseases rather than guessing an entity URL.