Dengue is a viral infection emerging as a significant threat to global health, caused by the dengue virus (DENV). DENV is primarily transmitted through a human-to-mosquito-to-human cycle by specific mosquito vectors known as Aedes aegypti and Aedes albopictus. In the year 2023, the number of reported dengue cases reached a historical high of 6.5 million (1); as of 30 April 2024, the total number of cases had reached a staggering 7.6 million (2) for the year 2024. It is estimated that 400 million yearly infections occur, and currently, half of the world’s population is at the risk of contracting dengue (1, 3). Previously endemic to the tropics and subtropics of Asia, the Americas, the Western Pacific, and Africa; it has now been reported to be rapidly expanding into previously non-endemic regions, such as temperate and even high-altitude regions due to climate change (4, 5). This growing burden of dengue infections highlights the importance of a more comprehensive understanding of the virus, its role in disease progression, and its interaction with the human immune system.

Dengue is an acute febrile illness caused by four types of single-stranded DENVs from the Flavivirus genus that elicit distinct serological responses with the antibodies in human blood (6). The four DENV serotypes share about 65% of their genome, with subvariants within each serotype (7). The DENV genome comprises 10 genes coding for 3 structural and 7 non-structural proteins. The three structural proteins are capsid (C), membrane (M), and envelope (E), while the seven non-structural proteins are NS1, NS2A, NSB, NS3, NS4A, NS4B and NS5 (8, 9). The C protein forms the nucleocapsid of the virus, and the M protein plays a crucial role in viral maturation (10, 11). The E protein is essential for viral entry into the host cell by receptor binding and subsequent fusion (11). The non-structural proteins NS1, NS2A, NS4A, and NS4B are involved in the RNA replication process, with NS1 and NS5 also serving as antigens to initiate immune responses in the host cell (9, 12–14). The NS3 protein has helicase and protease functions, while NS4A is involved in autophagy (9, 14).

The replication cycle of the DENV begins with viral binding to the host cell receptors through clathrin-dependent receptor-mediated endocytosis (15). Inside the host cells, the virus undergoes endosomal processing facilitated by Rab5 and Rab7 proteins. Subsequently, the viral genome is released into the cytoplasm and translated into the endoplasmic reticulum (ER). The newly synthesised components are assembled into immature viral particles in the ER undergo maturation in the Trans-Golgi Network via furin-mediated cleavage and are eventually released from the cell through exocytosis (16).

Four serotypes of DENV have been reported to date (DENV1-4), and each serotype can cause dengue fever (DF), dengue haemorrhagic fever (DHF) (Grades 1-2), and dengue shock syndrome (DSS) (DHF Grades 3 and 4). DENV1-4 are the primary dengue-causing serotypes circulating in humans, with different regions reporting varying profiles of circulating serotypes ( Figure 1 ). Common symptoms of DHF include high fever, muscle and joint pain, severe headache, vomiting, bleeding from gums, and skin rashes. Without immediate medical treatment, DHF (Grades 1-2) might develop into DSS, the severe form of DHF, characterized by shortness of breath, rapid pulse rate, low blood pressure and severe abdominal pain. 60-80% of individuals with a primary DENV infection are asymptomatic; however, the risk of severe disease increases significantly during a second infection, especially among those with longer intervals since the primary DENV infection (17). DENV-2 and DENV-4 are more commonly identified in secondary dengue infections, while DENV-1 and DENV-3 often cause primary infections (7, 18).

There is no consensus on the pathogenesis of DENV infection in the scientific community. The three major factors related to the severity of DENV infection include secondary infection, host genetics, and viral virulence (19–23), with the majority view that secondary infection is the main risk factor for DHF. Primary DENV infection involves the stimulation of interferon gamma (IFNγ) (24). During a secondary infection, non-neutralizing antibodies (25, 26) from the primary infection can bind to the serotype of the second infection. Instead of neutralizing the virus, these antibodies facilitate its entry into host cells, leading to increased viral replication and a more severe immune response. The phenomenon is commonly known as Antibody-Dependent Enhancement (ADE) (27, 28). This heightened immune response can cause increased vascular permeability, plasma leakage, and other severe symptoms characteristic of DHF. The detection of DENV by RIG-I (retinoic acid-inducible gene I) and MDA5 (melanoma differentiation-associated protein 5) triggers the phosphorylation of IRF3 (interferon regulatory factors 3) and IRF7 (interferon regulatory factors 7), leading to the production of type I and III interferons (IFNs), the activation of the JAK-STAT pathways, and the upregulation of interferon-stimulated genes (ISGs) (29). During severe infection, immature DENV particles are recognized by TLR2 (Toll-like Receptor 2) and DC-SIGN (Dendritic Cell-Specific Intercellular adhesion molecule-3-Grabbing Non-integrin) on monocytes and immature dendritic cells, resulting in the release of inflammatory mediators like IL-1β (interleukin 1β) and TNF-α (tumour necrosis factor α), which increase endothelial cell permeability or dysfunction, contributing towards severe complications (30, 31). Moreover, excess secretion of anti-DENV antibodies can exacerbate ADE ( Figure 2 ), further increasing autoantibody production and its potential glycosylation, a signature seen in severe DHF and DSS (32, 33).

Endothelial dysfunction could lead to complications such as hypotension, ascites, pleural effusions, shock, and organ dysfunction (34). NS1 protein and inflammatory mediators from monocytes, macrophages, dendritic cells, and mast cells contribute to this dysfunction (34, 35). Elevated cytokines and chemokines, such as IL-1β, IL-6, TNF-α, and MMP-9, directly cause endothelial dysfunction, while others like IL-10 may reduce it (36). Lipid mediators, including PAF, leukotrienes, prostaglandins, and sPLA2 enzymes, also play a role in vascular permeability (36). PAF and prostaglandin metabolites are elevated in severe cases and cause endothelial dysfunction (34, 36, 37). Mast cell products such as histamine, tryptase, chymase, VEGF, and serotonin are linked to disease severity and vascular leakage (38–40). Recent studies from Nicaragua and India showed that both primary and secondary DENV infections can range from subclinical to severe implications (41, 42).

DSS also involves severe cases of hepatomegaly and bleeding, especially in gum, nose, and stool. A typical signature of DSS, hepatomegaly and other complications regarding the liver, high levels of aspartate transaminase (AST) levels are inferred to have played an influential role (43). Secondary hemophagocytic lymphohistiocytosis (HLH), characterized by macrophage activation and cytokine storms, is a significant cause of severe liver dysfunction and high mortality in dengue patients (44, 45). Dengue can cause various bleeding manifestations, including petechiae, gum bleeding, and hematemesis (36). Bleeding results from thrombocytopenia, dysfunctional platelets, coagulation pathway abnormalities, and prolonged shock (36, 46). Over 50% of hospitalized dengue patients experience severe thrombocytopenia, which correlates with disease severity (36). Platelets can be directly infected by DENV, leading to activation through multiple pathways, including serotonin release from mast cells and direct activation by the virus (40, 47).

Since DENV is a foreign body infecting host cells, the genes influencing host defence mechanisms against such foreign infecting agents could play a crucial role in influencing effective response against DENV. One such large complex of genes are the HLA genes. HLA genes are known to be the most polymorphic regions in the human genome and are the part of Major Histocompatibility Complex (MHC) region located on the short arm of chromosome 6. Class I and class II HLA genes encode antigen-recognising sites crucial for adaptive immune responses, recognising between self and non-self. HLA class I proteins present foreign peptides to T-cell receptors on CD8+ T cells (cytotoxic T cells) to launch immune responses. HLA class II proteins are expressed on active immune cells like antigen-presenting cells (APCs) and B cells, where they detect antigens and present them to CD4+ T cells (helper T cells) to initiate immune responses. During Antibody-Dependent Enhancement (ADE) where Fcγ receptors (FcγR) facilitate the increased entry of DENV into immune cells, it could lead to the upregulation of HLA class I molecules, suppressing the activity of NK cells (48). Such reduced activity could potentially contribute to disease pathogenesis. HLA class I alleles further influence the magnitude of CD8+ T cell responses against DENV (49–51). Therefore, the host’s HLA allele profile is paramount in determining the immunogenic response to DENV infection. This leads to varied immunopathological outcomes, including susceptibility or protection against DENV infection or more severe outcomes ( Figure 2 ) (52, 53).

Given the crucial role of HLA genes in determining the pathophysiology of DENV infection, numerous case-control studies have investigated how HLA polymorphisms influence susceptibility to and protection against severe outcomes of DENV infection. By late 1970s, the idea that HLA proteins could determine the susceptibility towards viral infections started gaining ground (54). HLA responses against DENV can also show serotype specificity. A study on the Sri Lankan population has demonstrated that DENV-2-specific responses, likely due to the population’s previous history of DENV-2 infection, signifying that epidemiological history could influence serotype-specific immune responses (51). However, HLA allele associations with DENV infection outcomes could be complicated by factors such as HLA alleles typing resolution, the population studied, predominant DENV serotypes, HLA allele composition, and how the different stages/categories of dengue are defined in the study design. Therefore, usage of standardised pathophysiological classifications for dengue, clear definition of populations studied, and large enough case & control sizes are imperative to derive statistically significant inferences.

The earliest case-control study suggesting the influence of HLA alleles in determining susceptibility or protection against DENV infection was performed on Thai children in 1981, identifying several HLA class I antigens as potential crucial regulators in the development of severe forms of dengue (55). Throughout the years, several case-control studies, primarily across Southeast Asia, Central and South America, have further cemented the crucial role of different HLA class I alleles in determining dengue susceptibility (55–73).

Dengue cases have been rising exponentially over the past five decades with the Americas (Caribbean and Latin American), Southeast Asia, and South Asia, bearing the highest burden of dengue cases ( Figure 3A ). From 2022 to 2023, dengue cases more than doubled in 16 Caribbean and 9 Latin American countries, with St Kitts and Nevis and Argentina reporting increases of 28500% and 17364.4%, respectively. Similarly, cases in Bangladesh in South Asia and Thailand in Southeast Asia showed increases of 414.6% and 241.1% ( Figure 3B ), respectively. In Europe, the locally transmitted cases increased significantly in 2023, with Spain, Italy, and France reporting 3, 82, and 45 cases, respectively (2).

The global increase in dengue cases can be attributed to several factors, including viral evolution (74, 75), human settlement patterns (76), socioeconomic drivers (77), human mobility (78), age (79) and climate change (80). A complex interplay of these factors determines the epidemiological fate of dengue for any given population or geographic setting. Specifically, DENV vectors belonging to Aedes sp have rapidly expanded into new habitats, driven by warmer and more humid conditions due to climate change (81–85). Such climatic conditions could result in a shorter extrinsic incubation period (EIP), the period required for a mosquito to become dengue-transmittable after having an infectious meal, increasing dengue risk (86). Furthermore, rising temperatures have been found to increase the biting incidence of dengue vectors, while extended periods of warmer conditions lengthen the suitable period for dengue transmission, heightening the risk of severe epidemics (87).

It is projected that within the next six decades, close to two-thirds of the world’s population will be vulnerable to dengue infection (82). Without adequate preparations, the potential surge of dengue cases could overwhelm medical infrastructures, particularly considering emerging infectious diseases like COVID-19. This growing threat underscores the importance of understanding genetic factors that influence dengue susceptibility and severity.

In the context of the rising dengue cases due to climate change and other anthropogenic factors, we review and perform a meta-analysis of the known associations of HLA alleles with dengue. As key regulators of adaptive immunity, HLA molecules present viral peptides to T cells, influencing immune activation and viral clearance. Variability in HLA alleles has been linked to differences in dengue susceptibility, severity, and immune response efficiency (88–90). Understanding these associations is crucial to gaining deeper insights into how our adaptive immune system functions and influences DENV infection outcomes.

Global DENV infection cases and endemicity data were compiled and collected from WHO’s 2024 reports (1, 2) and ECDC (91), latest until 24 June 2024. Missing data from DENV infection cases was collected from the official reported figures mentioned governmental health ministry and infectious disease surveillance website for each respective country (92). We had a total of 75 countries reporting endemic dengue cases until the start of 2024.

To look for candidate papers for HLA-Dengue association meta-analysis, we searched two databases, PUBMED and Semantic Scholar. In order to make the best use of the two datasets, custom search strategies were used for each, mentioned below:

With the studies obtained from the above search strategies, we followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines (94) to screen and finalise studies which would be included in the meta-analysis ( Figure 4 ). During the selection of the studies to be included for the meta-analyses, the investigators decided on the final set of candidates without any conflicting viewpoints or opinions.

A total of 19 studies have been included in the meta-analysis ( Table 1 ), DENV infection phenotypes such as DF, DHF and DSS were comprehensively extracted. Symptomatic DENV infection cases defined as DEN and the sum of DF and DHF was considered as DEN in our meta-analysis. Cochrane RevMan Web (99) was used for calculation of effect sizes of the studies, confidence intervals (CIs), heterogeneity across studies (I² statistics) and generation of forest plots. Fixed-effects model was used to calculate the pooled effect sizes and corresponding 95% CIs. Statistical significance was set at Mantel-Haenszel P < 0.05. Forest plots were plotted based on dengue phenotypes (DF, DHF and DEN) and further stratified by populations.

The HLA-A*02 allele shows a significant amount of heterogeneity (P = 1.0E-03) across the five tested studies when DF is compared with healthy controls. Despite such heterogeneity, HLA-A*02 [P = 2.0E-03, OR = 1.37 (1.13–1.67)] is inferred to have a predisposing effect towards DF development ( Figure 5 ).

When comparing DHF with healthy controls, the data from seven selected studies also show significant heterogeneity [P = 1.0E-02]. The combined odds ratio for HLA-A*02 for DHF against healthy controls is insignificant [P = 0.07, OR = 1.18 (0.99–1.40)], with a tendency towards being more predisposing to DHF ( Figure 5 ).

When comparing DEN with healthy controls based on 13 studies, the combined odds ratio [P = 2.0E-02, OR = 1.16 (1.02–1.32)] suggests a predisposing influence of HLA-A*02 despite the significant heterogeneity in the observation [P = 1.0E-04] ( Figure 5 ).

To infer the regional tendencies of the HLA-A*02 allele influencing the susceptibility to any forms of DEN, we calculated the odds ratio after grouping the studies based on their region of origin ( Figure 6 ). Based on a single study conducted in Jamaica (72), the allele showed a tendency towards being a risk allele with the Caribbean region with the statistics remaining insignificant [P = 0.06, OR = 2.02 (0.97–4.22)]. Meanwhile, in Latin America, no significant trend was observed [P = 0.93, OR = 1.01 (0.81–1.25)].

In South Asia, the two studies based on Sri Lankan (71) and Indian (61) cohorts reported an insignificant association with dengue with respect to HLA-A*02. The combined odds ratio remained insignificant [P = 0.19, OR = 0.78 (0.53–1.14)], with a tendency towards being protective ( Figure 6 ). In Southeast Asia, despite significant heterogeneity across studies [P = 2.0E-03] driven primarily by the Vietnamese study done in 2001 (66) and the study on Filipino children in 2015 (56), a clear trend of risk emerged. Based on the six selected studies across Thailand, Vietnam, Malaysia, and Philippines, HLA-A*02 is inferred to be a statistically significant risk allele for DEN [P = 2.0E-03, OR = 1.31 (1.10–1.57)] ( Figure 6 ).

When combined across different regions and accounting for significant heterogeneity (P = 8.0E-04), the risk predisposition characteristic of HLA-A*02 stands [P = 4.0E-02, OR = 1.14 (1.01–1.30)] ( Figure 6 ). It is very likely that this inference is driven primarily by the Southeast Asian studies, given their higher weight in the analysis conducted. Moreover, regardless of significant associations with either being a risk or protective against DEN within each region, the tendency of odds ratio differs significantly between each regional group (P = 2.0E-02). This difference among different region groups suggests a population-specific variability of the susceptibility profile of HLA-A*02.

HLA-A*03 do not show any significant effect on DF susceptibility when compared to healthy controls across four studies [P = 0.72, OR = 0.95 (0.70–1.28)] ( Figure 7 ). When DHF is compared with healthy samples, significant heterogeneity between the four included studies was observed (P = 1.0E-03). Despite this heterogeneity, the combined odds ratio [P = 1.0E-04, OR = 1.83 (1.34–2.50)] suggests a predisposing effect of HLA-A*03 towards DHF, primarily driven by the Thai study conducted in 2005 (64).

When DEN was compared with healthy controls, no significant associations towards risk or protection were observed [P = 0.81, OR = 1.03 (0.82–1.28)]. Overall, HLA-A*03 shows a statistically insignificant [P = 0.07, OR = 1.15 (0.99–1.34)] tendency to be a risk allele. Within the three tested outcomes, DF, DHF, and DEN, a significant difference between each group was observed [P = 4.0E-03] ( Figure 7 ).

Within the Caribbean [P = 0.07, OR = 2.22 (0.95–5.23)] and Latin American [P = 0.21, OR = 1.12 (0.81–1.56)] populations, statistically insignificant influence of HLA-A*03 on DEN susceptibility was observed ( Figure 8 ). In South Asia [P = 0.80, OR = 0.92 (0.50–1.71)] and Southeast Asia [P = 0.34, OR = 0.84 (0.58–1.21)], no significant influence of HLA-A*03 on DEN was inferred ( Figure 8 ). Overall, when all regions are compared, HLA-A*03, does not show any significant effect on DEN susceptibility [P = 0.81, OR = 1.03 (0.82–1.28)].

HLA-A*24 ( Figure 9 ) shows no statistically significant protective or predisposing effect for DF [P = 0.85, OR = 1.04 (0.83–1.30)]. When we compare DHF with healthy controls across eight studies, the significant combined odds ratio [P = 3.0E-04, OR = 1.38 (1.16–1.64)] suggests a predisposing influence of HLA-A*24 towards DHF.

When we compare DEN with the healthy control across 12 studies ( Figure 9 ), the predisposing influence of HLA-A*24 [P < 1.0E-05, OR = 1.40 (1.23–1.60)] stands despite significant heterogeneity [P = 1.0E-03] across different studies.

In terms of combined influence of HLA-A*24 across the three different subtypes of dengue infection, the statistics [P < 1.0E-05, OR = 1.32 (1.20–1.45)] ( Figure 9 ) suggest an overall predisposing influence towards dengue.

In terms of population specific tendencies ( Figure 10 ), HLA-A*24 shows significant predisposition towards DEN in the Caribbean population [P = 1.0E-03, OR = 14.29 (2.90–70.35)]. It is noted that the wide confidence interval for the study is due to its small sample size and should be interpreted with caution. In the three Latin American studies (59, 68, 69), HLA-A*24 does not show statistically significant associations with being a risk or a protective factor for DEN [P = 0.57, OR = 1.10 (0.78–1.55)]. Having said that, the tendency of the allele could be considered towards being predisposed to DEN.

Among South Asians, HLA-A*24 shows a similar trend to that of Latin Americans, with no statistically significant associations towards being a risk or protective factor for DEN [P = 0.33, OR = 1.19 (0.84–1.69)], while having a tendency towards being a risk allele for DEN ( Figure 10 ). However, in Southeast Asia, HLA-A*24 is inferred to be a risk allele for DEN based on the combined odds ratio [P = 3.0E-04, OR = 1.33 (1.14–1.56)] ( Figure 10 ).

When all the regional populations are considered together, HLA-A*24 is inferred to be a predisposing allele for DEN [P < 1.0E-05, OR = 1.30 (1.14–1.49)] ( Figure 10 ). This inference is significantly driven by the outcomes from the studies on Southeast Asian populations, which contributed to a higher number of sample counts.

For HLA-A*33 ( Figure 11 ), when comparing DF with healthy controls based on four studies, no significant influence of HLA-A*33 on DF was inferred [P = 0.71, OR = 0.94 (0.68–1.30)]. Whereas, when we compared DHF with healthy controls, HLA-A*33 shows significant protective influence against DHF [P = 6.0E-04, OR = 0.63 (0.48–0.82)], despite a significant degree of heterogeneity [P = 5.0E-02].

Despite significant heterogeneity observed across studies [P = 3.0E-04], HLA-A*33 was found to confer a protective effect against DEN [P = 3.0E-02, OR = 0.80 (0.66–0.98)] ( Figure 11 ). When observed across all the three different tested scenarios for dengue (DF, DHF, DEN), HLA-A*33 was inferred to have a protective effect [P = 4.0E-04, OR = 0.77 (0.67–0.89)] ( Figure 11 ). Such an inference could have been driven by HLA-A*33’s stronger protective influence against DHF.

When compared across worldwide populations ( Figure 12 ), HLA-A*33 is observed to be a risk factor for DEN in the Caribbean population [P = 3.0E-03, OR = 4.51 (1.65–12.33)], a trend completely disparate from the general trend observed for HLA-A*33 in the previous set of inference. Furthermore, the small sample size could have played a significant role in the wide confidence interval observed for the Caribbeans. Interestingly, in Latin America, a study on the Brazilian population (68) showed a similar predisposing tendency for HLA-A*33 as the Caribbeans (72), whereas another study on Cubans (69) demonstrated a protective tendency. Overall, HLA-A*33 does not demonstrate any significant protective or predisposing effect with respect to DEN to the Caribbeans [P = 0.76, OR = 0.92 (0.52–1.61)].

Similar to Latin America, HLA-A*33 does not show an association with a protective or predisposing effect towards DEN in South Asians [P = 0.53, OR = 1.13 (0.77–1.65)] ( Figure 12 ). In contrast, in Southeast Asia, HLA-A*33 is inferred to confer protection against DEN [P < 1.0E-4, OR = 0.57 (0.43–0.75)] ( Figure 12 ). When examined individually across the four Southeast Asian studies included in the analysis, the protective trend remains consistent throughout (56, 66, 67, 97).

Considering the four regional population groups together, HLA-A*33 is inferred to be a protective allele against DEN [P = 3.0E-02, OR = 0.80 (0.66–0.98)] ( Figure 12 ). This inference is primarily driven by the associations inferred from the analysis of the Southeast Asian populations, having a higher sample size. The statistically significant difference in the combined odds ratio across the different regions further highlights the population-specific difference in the influence of HLA-A*33 on DEN.

HLA-B*44 ( Figure 13 ) showed no significant protective or predisposing association with DF when compared to healthy controls [P = 0.28, OR = 0.88 (0.70–1.11)]. Nonetheless, the observed tendency was towards being protective against DF. When we compare DHF with healthy controls, HLA-B*44 was found to have a protective effect against DHF [P = 1.0E-03, OR = 0.60 (0.44–0.82)]. However, this observation does not stand true when comparing DEN with healthy controls [P = 0.77, OR = 0.97 (0.81–1.17)], as no significant protective or predisposing effect is observed.

The overall tendency of HLA-B*44 allele suggests a protective effect against dengue [P = 2.0E-02, OR = 0.86 (0.75–0.98)] ( Figure 13 ). The significant difference between each subgroup highlights HLA-B*44’s increased protective influence on DHF likely have influenced the overall role for the allele with respect to dengue.

Comparing regional population tendencies ( Figure 14 ), in the Caribbean, no protective or predisposing influence of HLA-B*44 was observed [P = 0.38, OR = 1.56 (0.58–4.22)], although the observed trend was towards being predisposing. In Latin America, despite showing significant heterogeneity [P = 3.0E-02], HLA-B*44 is inferred to have predisposing influence towards DEN [P = 4.0E-02, OR = 1.37 (1.02–1.85)].

In South Asia, HLA-B*44 shows a tendency towards being predisposing; however, no significant association was observed [P = 0.75, OR = 1.07 (0.69–1.67)] ( Figure 14 ). In contrast, from the two Thai-based study in Southeast Asia, we can infer a protective effect of HLA-B*44 against DEN [P = 4.0E-03, OR = 0.65 (0.49–0.87] ( Figure 14 ).

HLA-B*44 shows no significant association with DEN when all four regions are taken together ( Figure 14 ). However, significant heterozygosity [P = 4.0E-03] is observed across each region, highlighting the population-specific influence of HLA-B*44 towards DEN.