Characteristics of dengue virus-infected peripheral blood mononuclear cell death that correlates with the severity of illness


Sukathida Ubol, Department of Microbiology, Mahidol University, 272 Rama 6 Road, Ratchatewi, Bangkok 10400, Thailand.
Tel: 662-201-5674; fax: 662-644-5411; email:


The pathogenic mechanism of the severe form of dengue is complicated. Recent reports indicate that apoptotic death of various tissues or organs may be associated with vascular leakage, and ultimately leads to the death of DENV-infected patients. In the present study, we provide additional evidence supporting the detrimental role of apoptosis in DENV infection. A comparison of the rate of apoptosis in PBMCs isolated from patients suffering DF, a mild form of the disease, and the rate in patients with DHF, a life-threatening disease, revealed that PBMCs from DHF patients underwent apoptosis at a significantly higher rate than those suffering from DF alone. This suggests that the severity of natural DENV infection correlates with PBMC apoptosis. In addition, this cell death was induced not only by DENV itself, but also by the apoptotic activities of pro-inflammatory cytokines, such as TNF-α, and IL-1β, that were upregulated in DHF patients. The death of these mononuclear cells that function in an innate immune system may explain the higher viral load in DHF patients than in DF patients. Interestingly, a gene expression profile pattern elucidated that apoptosis occurring during natural DENV infection involved mainly the extrinsic apoptosis pathway, which is mediated via both caspase-dependent and caspase-independent mechanisms. In conclusion, our data highlight the adverse effect of apoptosis induced by DENV and by pro-inflammatory cytokines during natural DENV infection.

List of Abbreviations: 

dengue virus


dengue fever


dengue hemorrhagic fever


dengue hemorrhagic fever/dengue shock syndrome


interleukin 1 beta


peripheral blood mononuclear cell


tumor necrosis factor alpha


terminal deoxynucleotidyl transferase dUTP nick end labeling

Dengue fever and DHF/DSS, the world's most significant arthropod-borne viral diseases, are caused by DENV. DENV, a member of the Flavivirus genus, has caused life-threatening epidemics in more than 100 countries in tropical and subtropical areas (1). Each year, 50–100 million people are estimated to be infected by DENV (2–4). While the majority of infected patients experience uncomplicated DF, an acute febrile illness typically lasting 3–7 days, on less frequent occasions the infection can cause DHF or DSS, a potentially fatal plasma leakage syndrome (2, 5, 6). Although it has been shown that the severity of the disease depends upon both virus and host factors, the determinants that predispose infected patients to develop mild or severe forms of the disease are still uncertain (7–9).

Apoptosis, a programmed cell death that regulates homeostasis in multicellular organisms, is mediated via at least two signaling pathways: extrinsic and intrinsic pathways. The extrinsic pathway is induced by the binding of death ligands to death receptors, which then initiates death inducing signaling complex (DISC) formation. This results in a stimulation of the death execution pathway via a regulation of caspase function. However, the intrinsic pathway, a receptor-independent pathway, is triggered intracellularly by irregular homeostasis (10).

Apoptosis is commonly detected in a broad range of viral infections, and has been shown to play both protective and destructive roles, depending on the cell type and viral system. For example, apoptosis is a mechanism of CD4+ T lymphocyte depletion during HIV infection, of neuronal death in poliomyelitis in a mouse model, and of immunological anergy during measles virus infection; whereas it is responsible for an attenuation of rabies virus infection (11).

DENV is one of the viruses that induces apoptosis in several cell types, such as monocytes, dendritic cells, endothelial cells and hepatocytes; this is important to its pathogenesis (12–15). A recent study has shown that DENV can replicate in megakaryocytes, and drives these cells to undergo apoptosis. This effect could be an important event in the etiology of thrombocytopenia in DENV infection (16). In addition, apoptosis has been found in liver, brain and intestinal tissues, as well as in pulmonary microvascular endothelial cells, in fatal DSS cases (17). This evidence indicates that apoptosis can be induced by DENV infection, and contributes to its degree of severity.

As the most recognized target cells of DENV in humans are mononuclear phagocytes, which constitute a major part of the population of innate immune cells, any alteration in this population of cells during DENV infection might contribute significantly to the clinical outcome (18). To investigate this hypothesis, PBMCs from DF and DHF patients were examined for both cellular and genetic modifications. By using cDNA microarray analysis, we found that PBMCs from DENV-infected patients underwent apoptosis mainly through the extrinsic pathway, according to gene expression profiling. The percentages of PBMC apoptosis between DF and DHF patients were compared to determine whether the death of these PBMCs contributes to the disease severity. Indeed, we were able to show a direct correlation between the level of PBMC apoptosis and the severity of the disease. Moreover, we also determined that the death of these innate immune cells was partly induced by the pro-inflammatory cytokines that were upregulated in DHF serum.


Clinical samples

Blood samples were obtained from DENV-infected patients at the Queen Sirikit National Institute of Child Health, Bangkok, Thailand. The patients, aged 5–10 years, were enrolled in the study after written informed parental consent was given. The protocol for human subjects and samples was approved by the Committee on Human Rights Related to Human Experiment, Mahidol University, Bangkok, Thailand. Blood samples were obtained twice: on the first day of administration, or on a day when the patient had a fever (acute-phase sample); and again at 30 days after the day of defervescence (convalescent-phase sample). Severities of infection were graded into DF or DHF, according to World Health Organization criteria.

Peripheral blood mononuclear cells and plasma preparation

Blood was collected in EDTA-sterilized tubes. PBMCs and plasma were immediately separated by centrifugation at 560 ×g for 5 min. The separated plasma was aliquoted and kept at −80°C until used. PBMCs were isolated from the total blood cells using Lymphoprep™ (Fresenius Kabi Norge AS, Oslo, Norway) according to the manufacturer's instructions. Red blood cells were eliminated using red blood cell lysis buffer. PBMCs were split into two groups. Some of the PBMCs were washed twice with 1× PBS and immediately subjected to apoptosis detection using the TUNEL assay. The remainder of the PBMCs were stored in RNAlater™ (Ambion, Austin, TX, USA) and kept at −80°C for genomic study.

Detection of in situ apoptotic PBMC using the TUNEL assay

In situ cell death was carried out according to the protocol recommended by the manufacturer (Roche Diagnostics, Mannheim, Germany). Briefly, paraformaldehyde-fixed PBMCs were permeabilized using permeabilization solution (1% Triton-X 100 in 0.1% sodium citrate), and then stained with a TUNEL solution (containing terminal nucleotidyl transferase and fluorescein-labeled nucleotide) at 37°C for 1 hr. The incorporation of fluorescein-labeled nucleotide was detected using an antibody conjugated with alkaline phosphatase and specific substrate. The stained PBMCs were investigated under a light microscope. A total of 500 cells were counted and the percentage of apoptotic PBMCs was calculated.

cDNA microarray screening

Sample preparation and cDNA microarray hybridization were performed as described previously (19). Briefly, RNA was purified from DHF PBMCs using the RNeasy® Mini Kit (Qiagen, Hilden, Germany). The reverse transcription of RNA to cDNA, preparation of biotinylated cRNA, fragmentation of cRNA, and array hybridization were carried out according to the manufacturer's protocol (Amersham Bioscience and GE Healthcare, Buckinghamshire, UK). The labeled cRNA was hybridized to the CodeLink 50 K whole human-genome expression array. The intensity of hybridization was saved as 16-bit grayscale TIFF image. Apoptosis-related genes for which the alteration of expression was greater than a twofold change, and/or with a P value ≤ 0.05, were considered. In this experiment, transcripts of DF PBMCs were used as a control.

Real-time RT-PCR

The total RNA of PBMCs from DHF patients was isolated using an RNeasy® Mini Kit (Qiagen). The expression levels of ARHGDIB, FHIT and GADD45B genes were quantitated using real-time RT-PCR with specific primers. ARHGDIB gene primers were sense 5′ GGA CTG GGG TGA AAG TGG ATA A 3′ and antisense 5′ GAG TGA CAG GGT GGG AAA AGA T 3′. FHIT gene primers were sense 5′ GTC CTT CGC TCT TGT GAA TAG G 3′ and antisense 5′ CAC TGA AAG TAG ACC CGC AGA 3′. GADD45B gene primers were sense 5′ GAG ACT GGA GCT GAG CGT CT 3′ and antisense 5′ TAT GCT TCC CAT CTC GCT CT 3′. Human β-actin gene primers were sense 5′ CCT GGC ACC CAG CAC AAT 3′ and antisense 5′ GGG CCG GAC TCG TCA TAC 3′. The copy numbers of these genes were quantitated using a standard curve of a reference gene, β-actin. The standard curve was done using a 10-fold serial dilution of known copy numbers of human β-actin gene-containing plasmid.

Reverse transcription–polymerase chain reaction

The expression levels of CIDEB and IER3 genes of PBMC from DHF patients were semiquantitatively measured by RT-PCR using a specific primer set. CIDEB gene primers were sense 5′ CTG CTG AAT GGA GTG CTA ACC 3′ and antisense 5′ TAG AGT AGA GCC CGT AGA ATG TGG 3′. IER3 gene primers were sense 5′ ATC TTC ACC TTC GAC CCT CTC 3′ and antisense 5′ AGG GCT CCG AAG TCA GAT TA 3′. Human β-actin gene primers were sense 5′ ATC TGG CAC CAC ACT TCT ACA 3′ and antisense 5′ GTT TCG TGG ATG CCA CAG GAC T 3′. The level of gene expression was expressed as a percentage relative to β-actin gene expression, as an internal control.

Enzyme-linked immunosorbent assay

The plasma levels of TNF-α and IL-1β in DHF patients during the acute and convalescent phases were measured using a Quantikine® ELISA kit (R&D Systems, Minneapolis, MN, USA). The experiment was carried out according to the manufacturer's instructions. Briefly, 200 μl standard and plasma samples were pipetted onto wells pre-coated with a monoclonal antibody specific for TNF-α and IL-1β. An enzyme-linked polyclonal antibody specific for the target protein was added. Color developed once specific substrate was added. The intensity of the color was measured using an EIA multi-well reader II (Sigma Diagnostics, St Louis, MO, USA).

Apoptosis induction by TNF-α in DHF plasma

DHF plasma samples containing TNF-α at a concentration ≥ 20 pg/ml were selected for the apoptotic induction experiment, as previously described (20). Briefly, 3 × 106 PBMCs from healthy donors were pre-incubated with 100 μl DHF plasma in the presence or absence of 20 pg etanercept, an anti-TNF-α antibody (Wyeth, Philadelphia, PA, USA), at 37°C for 2 hr. After treatment, these PBMCs were then cultivated at 37°C in 5% CO2. At 24 and 48 hr of cultivation, cells were harvested and stained with TUNEL. Healthy PBMCs treated with either autologous or heterologous plasma were used as negative controls. The percentages of apoptotic cells were determined.

Statistical analysis

Statistical analysis was carried out using SPSS 12.0 for Windows software. The differences in the percentages of apoptosis and in the levels of TNF-α and IL-1β were analyzed by an independent-sample t-test. Data obtained from real-time RT-PCR and RT-PCR were statistically analyzed by one-way anova. The paired sample t-test was used for analysis of TNF-α-induced apoptosis. Results were considered statistically significant if the P value was less than 0.05. Data are shown as mean ± standard error (SE).


Quantitation of apoptotic PBMC in DF versus DHF patients

In order to study the relationship between PBMC apoptosis and severity of DENV infection, the numbers of apoptotic PBMC from DHF patients in both acute and convalescent phases were determined using TUNEL staining. Apoptotic nuclei were dark brown in color, whereas cells with colorless nuclei were normal cells (Fig. 1a and b). Comparing the percentages of apoptotic cells during the fever phase of DENV infection of DF and DHF patients revealed that PBMCs from DHF patients (n= 35) contained significantly higher numbers of cells undergoing apoptosis than PBMCs from DF patients (n= 32): 21.1 ± 1.0% and 5.2 ± 0.4%, respectively. Meanwhile, the amount of PBMC apoptosis in the convalescent phase was less than 2% in both DF and DHF patients, n= 11 and n= 19, respectively (Fig. 1c). These results indicated that DENV-induced PBMC apoptosis occurs during the acute phase of a natural DENV infection, and that patients who developed a severe form of the illness exhibited a higher degree of PBMC apoptosis than did patients with milder disease.

Figure 1.

PBMC apoptosis detected by TUNEL staining. PBMCs from DF and DHF patients in the acute phase were stained with TUNEL. PBMCs isolated from a patient showed colorless nuclei indicating (a) a normal cell, whereas (b) dark nuclei were clearly found in apoptotic cells (arrows). (c) Numbers of apoptotic PBMCs from DF and DHF patients (black and gray bars, respectively) in acute and convalescent phases are expressed as a percentage. Asterisk indicates significant difference at P < 0.01.

Expression profile of apoptotic genes involved in DHF PBMC apoptosis

To characterize the genes responsible for PBMC apoptosis in DHF, a PBMC gene expression profile was screened using cDNA array analysis (19). Table 1 lists the apoptotic genes that upregulated in DHF PBMC. These apoptotic genes were strongly stimulated during the fever phase, and then significantly downregulated during convalescence. The identified genes can be categorized into five groups: death receptors, death ligands, Bcl family proteins, caspases and apoptosis-associated proteins. This data indicated the participation of the extrinsic apoptosis pathway in natural DENV infection.

Table 1.  Apoptotic genes that were upregulated in DHF PBMC during an acute phase of natural DENV infection
Gene abbrev.Gene nameFold changeP
ARHGDIBRho-GDP dissociation inhibitor (GDI) beta4.440.050
ARHGEF6Rac/Cdc42 guanine nucleotide exchange factor (GEF) 66.760.050
BAXBCL2-associated X protein1.900.009
BTG1B-cell translocation gene 12.700.020
CASP8Caspase 82.070.020
CD2CD2 antigen (p50)3.940.030
CD38CD38 molecule1.900.030
CIDEBCell death-inducing DFFA-like effector b7.200.000
CIDECCell death-inducing DFFA-like effector c3.160.000
DDX41DEAD (Asp-Glu-Ala-Asp) box polypeptide 414.400.050
DIDO1Death inducer-obliterator 11.800.040
DNASE 1Deoxyribonuclease I2.300.050
DUSP2Dual specificity phosphatase 26.940.020
FHITFragile histidine triad gene2.110.000
FIS 1Fission 1 homology3.700.040
FXR 1Fragile X mental retardation2.100.000
GADD45BGrowth arrest and DNA-damage-inducible, beta2.400.040
GZMAGranzyme A4.620.050
GZMHGranzyme H2.200.030
IER3Immediate early response 318.000.000
IL1BInterleukin 1 beta7.200.000
MALMal, T-cell differentiation protein7.300.000
MRPL41Mitochondrial ribosomal protein L4111.250.050
NCAM1Neural cell adhesion molecule 12.160.030
PACS2Phosphofurin acidic cluster sorting protein 22.300.020
PCBP4Poly(rC) binding protein 43.400.030
PPMIFProtein phosphatase 1F4.200.030
RPS3Ribosomal protein S312.600.000
TNFSF2Tumor necrosis factor superfamily, member 2, tumor necrosis factor-alpha5.500.000
TNFSF6Tumor necrosis factor (ligand) superfamily, member 6, Fas ligand2.900.020

Surprisingly, only six apoptotic-related genes were downregulated in DHF PBMCs during the acute phase: BCL2-associated athanogene (BAG1), caspase 4 (CASP4), caspase 5 (CASP5), caspase 7 (CASP7), cisplatin resistance-associated overexpressed protein (CROP) and homeodomain interacting protein kinase 2 (HIPK2).

Validation of microarray data

In order to confirm microarray data, RNA isolated from PBMC of 15 patients in each phase were used for determination of target gene expression at a transcriptomic level. The expression levels of ARHGDIB, FHIT and GADD45B were quantitated using real-time RT-PCR, and the expression levels of CIDEB, long IER3 transcript and short IER3 transcript were semiquantitated by RT-PCR using the β-actin gene as an internal control.

As shown in Figure 2a, the expressions of ARHGDIB, FHIT and GADD45B genes in acute-phase PBMCs were significantly higher (P < 0.05) than those of PBMCs from patients in the convalescent phase. This was consistent with the previous results from microarray analysis.

Figure 2.

Validation of cDNA array data by (a) real-time RT-PCR, (b) RT-PCR and (c) ELISA. RNA was isolated from DF and DHF PBMCs in acute and convalescent phases (n= 15 in each group) and further subjected to real-time RT-PCR or RT-PCR with specific primers. The expression of target genes was compared with that of β-actin gene. ELISA was used to validate the array analysis at the translational level. The levels of TNF-α and IL-1β were quantitated in acute and convalescent DHF plasma using ELISA. Asterisk indicates significant difference at P < 0.05. ARHGDIB, Rho-GDP dissociation inhibitor (GDI) beta; CIDEB, cell death-inducing DFFA-like effector b; FHIT, fragile histidine triad gene; GADD45B, growth arrest and DNA-damage-inducible, beta; IER3, immediate early response 3; IL-1β, interleukin 1 beta; TNF-α, tumor necrosis-alpha; TNFSF2, tumor necrosis factor superfamily member 2.

Semiquantitation of the expression level of the CIDEB gene in DHF PBMCs by RT-PCR showed a significant increase in expression during the acute phase of the infection, compared to the convalescent phase (P= 0.01) (Fig. 2b). The IER3 gene is transcribed into two forms, a long and a short IER3. The long IER3 transcript serves as an apoptosis inhibitor, whereas the short IER3 (sIER3) transcript is an apoptosis enhancer (21, 22). In order to determine the expression levels of both transcript forms of the IER3 gene, RT-PCR was used. The results showed that expression of the long transcript of PBMCs from an acute day was downregulated in comparison with expression levels of PBMCs from a convalescent day (Fig. 2b). As expected, the expression of the sIER3 transcript in DHF PBMCs during the acute phase was significantly higher than that in the convalescent phase (P= 0.03).

In the present study, ELISA was used to validate microarray analysis at the proteomic level. Figure 2c shows the amount of TNF-α and IL-1β in DHF plasma (n= 16 and n= 25, for acute and convalescent phases, respectively). The mean levels of TNF-α and IL-1β in acute DHF plasma were 20.08 ± 1.36 pg/ml and 49.45 ± 1.97 pg/ml, respectively, which were significantly higher than those of convalescent plasma. These results supported the information obtained from cDNA microarray analysis.

TNF-α from DHF-plasma-induced apoptosis of PBMC from healthy donors

Because several of the upregulated apoptotic genes in DHF PBMC (Table 1) encode for secreted proteins detected in DHF plasma, including TNF-α, Fas ligand, IL-1β, granzyme A and granzyme H, we then asked whether these plasma proteins could induce apoptosis in uninfected PBMC. To answer this question, PBMCs from healthy donors were treated with DHF plasma, as previously described. Twelve samples of DHF plasma containing TNF-α > 20 pg/ml were used in this experiment. As illustrated in Figure 3, the percentage of apoptosis in cultivated PBMC treated with autologous plasma on day 1 and day 2 were 6.9 ± 1.1% and 8.6 ± 1.3%, respectively, which were not different from those PBMC cultures treated with heterologous plasma. However, the percentages of PBMC apoptosis in DHF-plasma-treated cultures gradually increased from 12.6%± 1.4% on day 1 to 22.1%± 1.8% on day 2. This was significantly greater than the amount of PBMC apoptosis in the negative control (P < 0.01). These data indicated that DHF plasma contains apoptosis-inducing agent(s), which may include TNF-α.

Figure 3.

Functional test of plasma TNF-α in apoptosis induction. Plasma from DHF patients which contained TNF-α > 20 pg/ml were selected and pre-incubated with healthy PBMCs in the presence or absence of etanercept. After cultivation, PBMCs were collected and stained with TUNEL. The numbers of apoptotic PBMC of the plasma-treated group (▪), plasma-treated in the presence of etanercept group (□), autologous plasma-treated negative control group inline image and heterologous plasma-treated negative control group inline image are presented as percentages. Asterisk indicates significant difference at P < 0.05.

Previous work has shown that levels of pro-inflammatory cytokines are elevated during the acute phase of DHF (23). It is also known that some of these cytokines, such as TNF-α and IL-1β, are apoptosis-inducing factors. To test whether TNF-α is the apoptosis-inducing factor present in DHF plasma, the percentage of PBMC apoptosis was compared with PBMCs treated with DHF plasma in the presence or absence of etanercept. The percentages of apoptotic cells in the etanercept-treated plasma group were 10.1%± 1.4% and 14.4%± 1.8% on days 1 and 2, respectively. This was significantly less than the apoptosis that occurred in the plasma-treated group (P < 0.05) (Fig. 3). This result indicated that TNF-α is one of the plasma factors that induces PBMC apoptosis during natural DENV infection. In addition, the incomplete inhibition of apoptosis in the presence of etanercept suggests that more than one apoptotic-inducing factor is present in DHF plasma.


To efficiently terminate a viral infection, the cooperation of innate and adaptive immune responses is required. Modification or dysfunction of one of these immune arms generally results in an increased disease severity (24, 25). One function of apoptosis, a fundamental regulatory mechanism of various systems in humans, is to serve as an innate immune response to eliminate viral-infected cells (26). However, apoptosis is also used by several viruses to perturb the balance of subsets of immune cells for the viruses’ own benefit, which, in turn, is detrimental to the host (27, 28). DENV, for instance, was previously reported to induce apoptosis in the pathogenesis of DHF via CD8+ T-cell depletion and original antigenic sin (5, 29).

By investigating apoptosis in PBMC from dengue patients, we demonstrated that PBMC from patients who developed DF, a self-limiting form of the disease, underwent apoptosis at a lower percentage than PBMC from patients who developed DHF, a more serious form of dengue. As PBMC are one of the critical components of the immune system, greater damage to this population of cells may, at least in part, explain the higher viral load in DHF than in DF patients, resulting in increased severity of the disease. However, whether the percentage of PBMC apoptosis can be used as an indicator of DHF progression will require further investigation.

A number of studies have indicated that DENV-induced apoptosis is mediated via both extrinsic and intrinsic pathways (30–32). In particular, an extrinsic pathway via TNF-α was found to be critical for DENV-induced hemorrhage in a mouse model (33). To obtain insight into the mechanism of DENV-mediated PBMC apoptosis, cDNA microarray was used to screen for the apoptotic gene expression profile of DENV-infected PBMC. This analysis revealed that most of the upregulated apoptotic genes in DHF PBMC from the acute phase involve an extrinsic apoptosis pathway, including TNFR-I/TNF-α and Fas/FasL. Support for a receptor-mediated pathway was demonstrated by an upregulation of caspase-8, which is a downstream molecule of this cascade, and its substrate, Rho-GDP dissociation inhibitor beta (ARHGDIB), in the same array screening. The reduction in the number of apoptotic cells after adding the TNF-α antagonist, etanercept, implicates the role of TNFR-I/TNF-α in DENV-induced PBMC apoptosis. However, PBMC apoptosis in DHF patients is not determined exclusively by TNF-α, but rather by multiple factors, as the TNF-α antagonist only partially inhibited apoptosis. In addition, the expression of TNF-α at the proteomic level, as determined by ELISA, paralleled that of the genomic level. To further support array analysis, the amount of TNF-α protein detected in acute-phase DHF plasma was significantly higher than that found in DF plasma, suggesting a detrimental role of TNF-α in DENV infection. These findings agree with previous reports in which TNF-α and its family members are important apoptosis mediators during DENV infection (30, 31, 34).

The upregulation of IL-1β which was observed in the screening suggests another possible extrinsic pathway, which signals through JNK and nuclear factor-kappa B (NF-κB) (35–37). As the activation of NF-κB can promote the expression of the sIER3 transcript, which is known to trigger apoptosis via caspase-3 and Bax, the fact that sIER3 was also upregulated in DHF PBMC may imply that DENV uses both IL-1β-mediated and TNFR-I/TNF-α-mediated apoptosis pathways (21, 38, 39). Furthermore, ribosomal protein S3 (RPS3), which is part of caspase-independent TNF-α/FasL-mediated apoptosis, and directly causes DNA degradation after translocation from the cytosol into the nucleus, was also found to be upregulated, suggesting another alternative apoptosis pathway induced by DENV (40). Collectively, these data demonstrate that natural DENV infection is able to induce PBMC apoptosis by various pathways, but mainly through the extrinsic pathway via both caspase-dependent and caspase-independent mechanisms. Interestingly, the present study did not detect upregulation of any anti-apoptotic gene, but was able to find downregulation of one anti-apoptotic gene, BAG1.

The severity of the progression of the disease is associated with the degree of viral burden, which, in turn, depends largely on the ability of the virus to evade the immune response. Our data suggest that natural DENV uses multiple pathways to induce apoptosis in the immune cells, PBMC in particular, leading to exacerbation of the disease. Whether interfering with any of these key elements can prevent disease progression is worth further examination.


This work was financially supported by the National Center for Genetic Engineering and Biotechnology, Thailand (grant BT-B-01-MG-14-4703[to SU]). We thank the nurses at the World Health Organization Collaborating Center for Case Management of Dengue/Dengue Hemorrhagic Fever/Dengue Shock Syndrome, Queen Sirikit National Institute of Child Health, Bangkok, Thailand, for patient recruitment.