Dengue virus infection: Pathogenesis

INTRODUCTION — Substantial gaps remain in the comprehensive understanding of the pathogenesis of dengue virus (DENV) infections. In large part, this limitation is related to the lack of a suitable animal model of disease [1]. Rhesus monkeys develop RNAemia and viremia similar in a pattern to humans after DENV challenge but do not develop clinical disease. Careful epidemiologic and experimental challenge studies in humans have provided valuable information on DENV infection, but detailed data on virus distribution in vivo are available only from small numbers of patients with more severe disease, unusual manifestations, or the later stages of infection. Little pathogenetic information is available concerning milder infections, which constitute the vast majority of cases.

THE DENGUE VIRUS REPLICATION CYCLE — DENVs are members of the family Flaviviridae genus Flavivirus. They are small, enveloped viruses containing a single-strand RNA genome of positive polarity [2]. DENVs infect a wide range of human and nonhuman cell types in vitro. Viral replication involves the following steps:

●Attachment to the cell surface

●Entry into the cytoplasm

●Translation of viral proteins

●Replication of the viral RNA genome

●Formation of virions (encapsidation)

●Release from the cell

Binding of dengue virions to cells, which is mediated by the major viral envelope (E) glycoprotein, is critical for infectivity. The determination of the three-dimensional structures of the dengue E glycoprotein and the intact virion has facilitated the understanding of this process [3-5]. DENVs bind via the E glycoprotein to viral receptors on the cell surface, which may include heparan sulfate or lectins such as DC-SIGN and CLEC5A [6-8]; they can also bind to cell surface immunoglobulin receptors in the presence of antibodies to the E glycoprotein or precursor membrane (pre-M) protein, as described further below [9].

Following fusion of viral and cell membranes in acidified endocytic vesicles, the viral RNA enters the cytoplasm. The viral proteins are then translated directly from the viral RNA as a single polyprotein, which is cleaved to yield the three structural and seven nonstructural proteins [2]. Cleavage of several of the viral proteins requires a functional viral protease encoded in the nonstructural protein NS3. The nonstructural protein NS5 is the viral RNA-dependent RNA polymerase, which assembles with several other viral proteins and several host proteins to form the replication complex. This complex transcribes the viral RNA to produce negative-strand viral RNA, which serves as the template for the production of the viral genomic RNA.

The assembly and budding of progeny virions is still poorly understood. The pre-M structural protein is cleaved by a cellular enzyme, furin, as one of the final steps in maturation of progeny virions [2]. Cleavage of the pre-M protein enhances the infectivity of the virions 100-fold.

COURSE OF INFECTION — The course of DENV infection is characterized by early events, dissemination, and the immune response and subsequent viral clearance (figure 1).

Early events — DENV is introduced into the skin when an infected mosquito, most commonly Aedes aegypti, takes a blood meal from a host. The spread of virus early after injection has been studied in rhesus monkeys [10]. During the first 24 hours, virus could only be isolated from the injection site. The major cell type infected was not defined; in humans, Langerhans cells, myeloid dendritic cells, mast cells, and dermal fibroblasts have all been proposed to be target cells for DENV infection in the skin based on their permissiveness to infection in vitro [11]. In rhesus monkeys, virus was detected in regional lymph nodes 24 hours after infection [10]. Macrophages and dendritic cells were demonstrated to be early cellular targets for DENV infection in a mouse model deficient in both type I and type II interferon (IFN) receptors [12].

Dissemination — In rhesus monkeys, viremia begins two to six days after subcutaneous injection and lasts for three to six days. In early studies of humans infected with near wild type DENV strains, viremia began approximately one day later than in monkeys, but the duration of viremia was similar [13]. In a subsequent experimental human infection study, an under-attenuated DENV-1 strain had a mean incubation period was 5.9 days (range five to nine days), mean time of viremia was 6.8 days (range three to nine days), and mean peak RNAemia of 1.6 x 10⁷ genome equivalents (GE)/mL (range 4.6 x 10³ to 5 x 10⁷ GE/mL) [14]. In a study of natural human infection, viremia was detectable 6 to 18 hours before the onset of symptoms and ended around the time fever resolved [15].

In rhesus monkeys during the period of viremia, virus was frequently detected in lymph nodes distant from the site of inoculation and less commonly from spleen, thymus, lung, and bone marrow [10]. Virus was also isolated from peripheral blood leukocytes at the end of the viremic period and sometimes for one day after.

The distribution of virus in humans has been studied in blood, biopsy, and autopsy specimens from patients with natural DENV infection. Infection of peripheral blood mononuclear cells persists beyond the period of detectable viremia [16-18]. Conflicting data have been published regarding the principal infected cell type in the peripheral blood. An older study reported more frequent isolation of infectious virus from the adherent cell population than the nonadherent population, suggesting that monocytes are the primary target cell for infection [16]. A similar conclusion was reached in a study using flow cytometry, which reported the detection of dengue viral antigen in a very high percentage of circulating monocytes [18]. However, an earlier study using flow cytometry reported that the majority of cell-associated virus was contained in the CD20+ (B lymphocyte) fraction [17]. Another study using polymerase chain reaction found the highest levels of dengue viral RNA in B cells but could not exclude passive binding as an explanation [19].

The yield of DENV from tissues obtained at autopsy has generally been low. However, in one study using the most sensitive techniques for virus isolation, virus was isolated most often (4 of 16 cases) from liver tissue [20]. Antigen staining has suggested that the predominant cell types infected are macrophages in the skin [21] and Kupffer cells in the liver [22,23]; DENV antigens have also been detected in hepatocytes in some cases [24].

Immune response and viral clearance — Both innate and adaptive immune responses induced by DENV infection are likely to play a role in the clearance of infection (figure 2) [25]. Infection of human cells in vitro induces antiviral responses, including the production of interferons [26]. Consistent with these observations, elevated serum levels of IFN-alpha have been demonstrated in children with dengue [27].

The role of these cytokine responses is uncertain. Interferon inhibits DENV infection in vitro. In addition, DENV–infected cells are susceptible to lysis by natural killer cells in vitro [28]. However, DENV proteins are able to inhibit both the production of interferons and their antiviral function in infected cells [29,30]. In several studies, the expression of genes associated with type I interferon signaling was significantly lower in patients with dengue shock syndrome (DSS) than in patients without DSS [31,32]. Whether attenuated interferon responses are the result or cause of severe dengue disease is unknown.

The antibody response to DENV infection is primarily directed at serotype-specific determinants, but there is a substantial level of serotype-cross-reactive antibodies (figure 3). E, precursor membrane (pre-M), and NS1 are the principal viral proteins that are targeted. In vitro, E protein-specific antibodies can mediate neutralization of infection, direct complement-mediated lysis or antibody-dependent cellular cytotoxicity of DENV-infected cells, and block virus attachment to cell receptors [2]. Pre-M–specific antibodies only bind to virions that have not fully matured and have remaining uncleaved pre-M protein. NS1 is not found in the virion; NS1-specific antibodies are therefore incapable of neutralization of virus infection but can direct complement-mediated lysis of infected cells [33].

The basis of neutralization of virus by antibody is not well understood. Neutralization clearly requires a threshold level of antibodies; when the concentration of antibodies is below this threshold, the uptake of antibody-bound virus by cells that express immunoglobulin (Ig) receptors is paradoxically increased, a process termed antibody-dependent enhancement (ADE) of infection [34,35]. Since monocytes, the putative cellular targets of DENV infection in vivo, express immunoglobulin receptors and manifest ADE in vitro, this phenomenon is thought to be highly relevant in natural DENV infections (see below). In rhesus monkeys, passive transfer of low levels of dengue-immune human sera or a humanized chimpanzee DENV–specific monoclonal antibody resulted in a 2- to 100-fold increase in dengue-2 or dengue-4 viremia titers as compared with control animals [36,37]. An increase in viral titers in blood and tissues and enhanced disease were also observed after passive transfer of low levels of DENV-specific antibody in mice lacking interferon receptors [38]. DENV entry via ADE has also been found to suppress innate immune responses in infected monocytes in vitro [9].

One study characterized 301 human DENV-specific monoclonal antibodies [39]. Pre-M-specific antibodies represented a larger fraction of the monoclonal antibodies detected than antibodies directed at E or NS1. Pre-M-specific antibodies showed poor neutralization of infection in vitro but could mediate ADE.

The T lymphocyte response to DENV infection also includes both serotype-specific and serotype-cross-reactive responses [40]. DENV-specific CD4+ and CD8+ T cells can lyse DENV-infected cells in vitro and produce cytokines such as IFN-gamma, tumor necrosis factor (TNF)-alpha, and lymphotoxin. In vitro, IFN-gamma can inhibit DENV infection of monocytes. However, IFN-gamma also enhances the expression of Ig receptors, which can augment the antibody-dependent enhancement of infection [41].

Primary versus secondary infection — Infection with one of the four serotypes of DENV (primary infection) generally provides long-lasting immunity to infection with a virus of the same serotype [13]. In contrast, immunity to the other dengue serotypes is transient, and individuals can subsequently be infected with another dengue serotype (secondary infection). Two prospective cohort studies found that the interval between primary and secondary DENV infections was significantly longer among children who experienced a symptomatic secondary infection than those who had a subclinical secondary infection, suggesting that heterotypic protective immunity wanes gradually over one to two years [42,43].

In one report, the distribution of DENV in secondary infections was evaluated in eight rhesus monkeys [10]. The onset and duration of viremia were similar to primary infections. Autopsy specimens from six monkeys yielded virus somewhat more frequently from various tissues than specimens from primary infections. Another study found higher plasma virus titers in secondary than primary dengue-2 virus infections but not in secondary infections with DENVs of the other serotypes [44].

There is little information from human studies to allow comparisons of virus distribution or titer in primary and secondary infections. Several studies have reported that higher peak plasma virus titers in secondary dengue infections were associated with more severe illness [45-47]. Two studies failed to demonstrate higher viremia titers in patients with secondary dengue infections than in patients with primary dengue infections [48,49], but a study using quantitative reverse-transcription polymerase chain reaction reported higher viral RNA levels in CD14+ monocytes among dengue fever patients with secondary infections compared with dengue fever patients with primary infections [19].

The kinetics of DENV-specific antibodies in secondary dengue infections differ from those of primary dengue infections in several ways.

●Low concentrations of antibodies to the virus serotype causing the secondary infection are present before exposure to the virus. As a result, antibody-dependent enhancement of infection could occur early in secondary DENV infections. Consistent with this hypothesis, two separate analyses of data from longitudinal cohort studies of children in Nicaragua and Thailand found that the risk of more severe dengue illness was highest within a narrow range of pre-existing anti-DENV antibody titers [50,51].

●Concentrations of DENV-specific antibodies increase earlier in secondary infection, reach higher peak titers, and have a lower IgM:IgG ratio, suggestive of an anamnestic response. Thus, the levels of DENV-specific antibodies are much higher during the late stage of viremia in secondary infections, with greater potential for forming immune complexes of dengue virions and activating complement.

The kinetics of the T lymphocyte response in secondary infections also would be expected to differ from those of primary infections, with an earlier onset and higher level of DENV-specific T lymphocyte proliferation and cytokine production in secondary infections. Studies of circulating T lymphocytes during acute secondary infections have shown a high percentage of cells expressing markers of activation and high frequencies of dengue antigen-specific cells, consistent with this hypothesis [52-55]. However, a study that compared the frequencies of T cells specific for an immunodominant dengue epitope between symptomatic primary and secondary DENV infections found no significant differences [56].

The severity of dengue disease has been correlated with the level and quality of the DENV-specific T lymphocyte responses in some studies [53,54] but not in others [56,57].

Some serotype-cross-reactive T cells present after primary infection display qualitatively altered functional responses to other dengue serotypes [40]. In one prospective cohort study, specific T cell responses prior to secondary DENV infection were associated with the subsequent occurrence of dengue hemorrhagic fever, such as production of TNF-alpha in response to stimulation with dengue antigens [58]. In contrast, higher frequencies of CD4+ T cells producing IFN-gamma or interleukin (IL)-2 in response to stimulation with dengue antigens were associated with subclinical dengue infection, suggesting a protective effect as well [59]. In a separate study, the secretion of IL-6, IL-15, and macrophage chemotactic protein 1 in response to in vitro stimulation with DENV was associated with symptomatic dengue infection whereas the secretion of other cytokines, including IL-12 and CCL5/RANTES (Regulated on Activation, Normal T cell Expressed, and Secreted) was associated with subclinical infection, pointing to the potential for either protective or pathologic immune response profiles [60].

FACTORS INFLUENCING DISEASE SEVERITY — Most DENV infections produce mild, nonspecific symptoms or classic dengue fever (DF). Severe dengue, dengue hemorrhagic fever (DHF)/dengue shock syndrome (DSS), occurs in less than 1 percent of all DENV infections. Thus, considerable attention has been focused upon understanding the risk factors for DHF (table 1).

Viral factors — DHF can occur during infection with any of the four dengue serotypes; several prospective studies have suggested that the risk is highest with dengue-2 viruses [15,61-63]. Genetic analyses of DENV isolates from the Western hemisphere strongly suggest that DHF only occurs during infection with viruses that fall into specific genotypes within each dengue serotype [64,65]. These "virulent" genotypes were originally detected in Southeast Asia but are now widespread. Several studies have suggested that "virulent" and "avirulent" genotypes differ in their ability to replicate in monocytic cells [66,67], but it is not clear that this difference in in vitro replication is the factor responsible for virulence.

Prior dengue exposure — Multiple epidemiologic studies have shown that the risk of severe disease (DHF and DSS) is significantly higher during a secondary DENV infection than during a primary infection [9,61,68,69].

The increased risk of DHF in secondary DENV infections is felt to reflect the differences in immune responses between primary and secondary DENV infections described above: antibody-dependent enhancement of infection, enhanced immune complex formation, and/or accelerated T lymphocyte responses.

The increased risk for DHF associated with secondary DENV infections appears not to apply to infections with "avirulent" genotypes (see above). A prospective study in Peru found no cases of DHF or DSS during an outbreak of dengue-2 virus infections that was estimated to involve over 49,000 secondary infections in children [65]. At least 880 cases of DHF would have been expected based upon previous studies in Thailand. Furthermore, there are numerous documented cases of DHF occurring during primary infection, suggesting that differences in viral virulence, as discussed above, are also important [1,15].

Age — The risk for DHF appears to decline with age, especially after age 11 years. During the 1981 epidemic of DHF in Cuba, the modal age of DHF cases and deaths was 4 years, although the frequency of secondary dengue-2 infections was similar in those 4 to 40 years of age [70,71].

A specific population at higher risk for DHF in endemic areas is infants, particularly those between 6 and 12 months of age. These children acquire DENV-specific antibodies transplacentally and become susceptible to primary DENV infection when antibody levels decline below the neutralization threshold [72]. This observation is taken to support the hypothesis of antibody-dependent enhancement of infection as a primary factor in determining the risk for DHF. A direct correlation between antibody-dependent enhancement (ADE) activity of preinfection serum and the severity of infection has not been demonstrated, however [73].

Nutritional status — Some studies have reported that, unlike other infectious diseases, DHF and DSS are less common in malnourished children than in well-nourished children [74], and this has been taken to reflect the role of the immune response in disease pathogenesis. A systematic review found no consistent association with nutritional status, however [75].

Genetic factors — Epidemiologic studies in Cuba showed that DHF occurred more often in White persons than in Black persons [71], and a similar genetic resistance to DHF in Black individuals has been reported from Haiti [76]. Racial differences have been described in viral replication in primary monocytes and in the level of dengue serotype-cross-reactive T cell responses [77], but it is unclear if either of these explains the genetic association.

Genome-wide association studies conducted in Vietnam [78,79] and Thailand [80] have found significant genetic associations of two single-nucleotide polymorphisms, one in the major histocompatibility complex class I polypeptide-related sequence B (MICB) gene and one in the phospholipase C epsilon 1 (PLCE1) genes, with both dengue shock syndrome and less severe dengue. The mechanisms for these associations have not been defined.

In focused studies, DHF has been associated with specific human leukocyte antigen genes [81,82], with blood group [83], and with polymorphisms of tumor necrosis factor-alpha, vitamin D, Fc gamma IIa, and DC-SIGN genes [81].

PATHOPHYSIOLOGY OF DISEASE MANIFESTATIONS

Capillary leak syndrome — Plasma leakage, due to an increase in capillary permeability, is a cardinal feature of dengue hemorrhagic fever (DHF) but is absent in dengue fever (DF). The enhanced capillary permeability appears to be due to endothelial cell dysfunction rather than injury, as electron microscopy demonstrated a widening of the endothelial tight junctions [84]. DENV infects human endothelial cells in vitro and causes cellular activation [85]. Additionally, soluble NS1 protein, which can be detected in the serum during acute infection, has been reported to bind to endothelial cells, to activate cells through toll-like receptor 4 signaling, to induce endothelial permeability and disrupt the glycocalyx, and to serve as a target for antibody binding and complement activation [86-88]. However, the effects on endothelial cell function during infection are thought more likely to be indirectly caused by DENV infection for the following reasons:

●Histologic studies show little structural damage to capillaries [89].

●Infection of endothelial cells by DENV is not apparent in tissues obtained at autopsy [23].

●Increased capillary permeability is transient, with rapid resolution and no residual pathology.

●Maximal plasma leakage occurs several days after the peak of circulating levels of viral RNA or NS1 protein, during the stage of rapid immune clearance.

Most investigations have focused on the hypothesis that circulating factors induce the transient increase in capillary permeability. Multiple mediators are likely to be involved in vivo, and interactions between these different factors have been demonstrated in experimental animals. Nitric oxide has been associated with increased vascular permeability and severe dengue in two prospective studies in Asia [90,91]. However, the most important mediators are still thought to include tumor necrosis factor (TNF)-alpha, interferon (IFN)-gamma, interleukin (IL)-2, IL-8, vascular endothelial growth factor (VEGF), and complement [92]. The sources of these cytokines have been proposed to include virus-infected monocytes, dendritic cells and mast cells, activated platelets, and DENV-specific CD4 and CD8 T lymphocytes.

Elevated serum levels of TNF-alpha, IL-8, IFN-gamma, IL-2, and free VEGF have been observed in patients with DHF [92]. Other studies have found reduced serum levels of the complement proteins C3 and C5 in patients with DHF, with a corresponding increase in the serum concentrations of anaphylatoxins C3a and C5a [93,94].

It is difficult to detect elevated cytokine levels in the circulation, because of the short half-life of these molecules. Analysis of more stable markers of immune activation has provided additional, although indirect, support for the immunopathogenesis model of plasma leakage. Several studies have shown that children with DHF have elevated circulating levels of the soluble forms of CD8, CD4, IL-2 receptors, and TNF receptors [92]. Increased plasma concentrations of soluble TNF receptor II were found to correlate with the subsequent development of and with the magnitude of plasma leakage into the pleural space. The intensity of the immune response may ultimately be determined by the level of viral replication, however, as one study found that the plasma viremia titer was the strongest independent factor that correlated with plasma leakage [27].

Blood and bone marrow — Leukopenia, thrombocytopenia, and a hemorrhagic diathesis are the typical hematologic findings in DENV infections. Leukopenia is apparent early in illness and is of similar degree in DHF and dengue fever [95]. It is thought to represent a direct effect of DENV on the bone marrow. Bone marrow biopsies of children in Thailand with DHF revealed suppression of hematopoiesis early in the illness, with marrow recovery and hypercellularity in the late stage and during early clinical recovery [96]. In vitro studies have shown that DENV infects human bone marrow stromal cells and hematopoietic progenitor cells [97,98] and inhibits progenitor cell growth [99].

Some degree of thrombocytopenia is common in both dengue fever and DHF, but marked thrombocytopenia (<100,000 platelets/mm³) is one of the criteria used to define DHF. Multiple factors are thought to contribute to the fall in platelet count, which is most severe late in the illness [95]. Bone marrow suppression may play a role, but platelet destruction is probably more important. In one study, 10 of 11 Thai children with DHF had a shortened platelet survival time, ranging from 6.5 to 53 hours [100]. Adsorption of dengue virions or virus-antibody immune complexes to the platelet surface, with subsequent activation of complement, are thought to be responsible for the platelet destruction.

Manifestations of the hemorrhagic diathesis in DENV infections range from a positive tourniquet test to life-threatening hemorrhage. Fatal DHF may be associated with diffuse petechial hemorrhages involving the stomach, skin, heart, intestine, and lungs [89]. (See "Dengue virus infection: Clinical manifestations and diagnosis".)

Despite the nomenclature, however, the occurrence of hemorrhage does not define DHF as compared with dengue fever since a positive tourniquet test may occur with equal frequency in the two disorders [95]. Several different mechanisms, possibly acting synergistically, contribute to bleeding tendency of DENV infections. Both the vasculopathy and thrombocytopenia described above create a predisposition to bleeding.

Endothelial cell activation and injury and activation of coagulation and fibrinolysis have been reported in dengue, particularly in severe infections. Abnormalities that have been described include increased numbers of circulating endothelial cells [101], elevated levels of von Willebrand factor, tissue factor, tissue plasminogen activator, and platelet activator inhibitor [102], and an increased fractional catabolic rate of fibrinogen [103]. However, most of these findings are based on small studies and comparison with non-dengue controls. Frank coagulopathy is uncommon except in patients with shock.

A final etiologic factor may be molecular mimicry between dengue viral proteins and coagulation factors. One study of 88 Tahitian children with DENV infection found that antibody responses to homologous peptides derived from the DENV E protein cross-reacted with plasminogen; these antibodies correlated with the occurrence of hemorrhagic signs (including petechiae) but not with thrombocytopenia or shock [104]. Another study reported that monoclonal antibodies directed at the DENV NS1 protein bound in vitro to human fibrinogen, platelets, and endothelial cells and induced hemorrhage in mice [105].

Liver — Elevations of serum aminotransferases that are usually mild are common in DENV infections [95]. Typical pathologic findings in the livers of fatal cases of dengue include hepatocellular necrosis and Councilman bodies with relatively little inflammatory cell infiltration, similar to the findings in early yellow fever virus infection [89]. The pathologic similarities between these two diseases and the relatively frequent isolation of DENV from liver tissues of fatal cases suggest that liver injury is directly mediated by DENV infection of hepatocytes and Kupffer cells. DENV has been shown to infect and induce apoptosis in a human hepatoma cell line in vitro [106]. However, immune-mediated hepatocyte injury is a potential alternative mechanism.

Central nervous system — Rare cases of encephalopathy have been attributed to DENV infections. True encephalitis has been reported, with detection of DENV in brain tissue [107], but this is clearly the exception in humans. In one series of 100 fatal cases of dengue, no evidence of central nervous system inflammation was found [89].

INFORMATION FOR PATIENTS — UpToDate offers two types of patient education materials, "The Basics" and "Beyond the Basics." The Basics patient education pieces are written in plain language, at the 5^th to 6^th grade reading level, and they answer the four or five key questions a patient might have about a given condition. These articles are best for patients who want a general overview and who prefer short, easy-to-read materials. Beyond the Basics patient education pieces are longer, more sophisticated, and more detailed. These articles are written at the 10^th to 12^th grade reading level and are best for patients who want in-depth information and are comfortable with some medical jargon.

Here are the patient education articles that are relevant to this topic. We encourage you to print or email these topics to your patients. (You can also locate patient education articles on a variety of subjects by searching on "patient info" and the keyword(s) of interest.)

●Basics topic (see "Patient education: Dengue fever (The Basics)")

SUMMARY AND RECOMMENDATIONS

●Vector – DENV is introduced into the skin when an infected mosquito, most commonly Aedes aegypti, takes a blood meal from a susceptible host. (See 'Early events' above.)

●Replication cycle – Dengue viruses (DENVs) are small, enveloped viruses that are members of the family Flaviviridae genus Flavivirus. Viral replication involves the following steps: attachment to the cell surface, cellular entry, translation of viral proteins, replication of the viral RNA genome, formation of virions by encapsidation, and cellular release. (See 'The dengue virus replication cycle' above.)

●Dissemination – Viremia is detectable in humans 6 to 18 hours before the onset of symptoms and ends as the fever resolves. (See 'Dissemination' above.)

●Immune response – Both innate and adaptive immune responses induced by DENV infection are likely to play a role in the clearance of infection. (See 'Immune response and viral clearance' above.)

•Serotype immunity – Infection with one of the four serotypes of DENV (primary infection) provides long-lasting immunity to infection with a virus of the same serotype [13]. However, immunity to the other dengue serotypes is transient, and individuals can subsequently be infected with another dengue serotype (secondary infection). (See 'Primary versus secondary infection' above.)

•Enhanced infection – Antibodies to proteins on the DENV surface can cause increased infection of cells bearing immunoglobulin receptors, a phenomenon known as antibody-dependent enhancement of infection. (See 'Immune response and viral clearance' above.)

●Determinants of severe illness

•Lymphocytic response – The severity of dengue disease has been correlated with both the level and quality of the DENV-specific T lymphocyte responses. (See 'Primary versus secondary infection' above.)

•Serotype – Although dengue hemorrhagic fever (DHF) can occur during infection with any of the four dengue serotypes, several prospective studies have suggested that the risk is highest with dengue-2 viruses. (See 'Factors influencing disease severity' above.)

•Secondary infection – Epidemiologic studies have shown that the risk of severe disease is significantly higher during a secondary DENV infection than during a primary infection. (See 'Prior dengue exposure' above.)

●Pathogenesis of severe illness – Plasma leakage, due to an increase in capillary permeability, is a cardinal feature of DHF but is absent in dengue fever. The enhanced capillary permeability appears to be due to endothelial cell dysfunction rather than injury. (See 'Pathophysiology of disease manifestations' above.)