755

Early Immune Activation in Acute Dengue Illness Is Related to Development

of Plasma Leakage and Disease Severity
Sharone Green, David W. Vaughn,1 Center for Infectious Disease and Vaccine Research, University
Siripen Kalayanarooj, Suchitra Nimmannitya, of Massachusetts Medical School, Worcester, and Division
Saroj Suntayakorn, Ananda Nisalak, Robert Lew, of Rheumatology and Immunology, Brigham and Women’s Hospital,
Boston, Massachusetts; Department of Virology, Armed Forces
Bruce L. Innis, Ichiro Kurane,1 Alan L. Rothman, Research Institute of Medical Sciences, and Queen Sirikit National
and Francis A. Ennis Institute for Child Health (formerly Bangkok Children’s Hospital),
Bangkok, and Department of Pediatrics, Kamphaeng Phet Provincial
Hospital, Kamphaeng Phet, Thailand; Department of Virus Diseases,

Downloaded from https://academic.oup.com/jid/article/179/4/755/888053 by guest on 26 May 2023

Walter Reed Army Institute of Research, Washington, DC

T lymphocyte activation and increased cytokine levels have been described in retrospective
studies of children presenting with dengue hemorrhagic fever (DHF). Serial plasma samples
obtained in a prospective study of Thai children presenting with !72 h of fever were studied.
Plasma levels of 80-kDa soluble tumor necrosis factor receptors (sTNFRs) were higher in
children who developed DHF than in those with dengue fever (DF) or other nondengue
febrile illnesses (OFIs) and were correlated with the degree of subsequent plasma leakage.
Soluble CD8 and soluble interleukin-2 receptor levels were also elevated in children with DHF
compared with those with DF. Interferon-g and sTNFR 60-kDa levels were higher in children
with dengue than in those with OFIs. TNF-a was detectable more often in DHF than in DF
or OFIs (P ! .05). These results support the hypothesis that immune activation contributes
to the pathogenesis of DHF. Further studies evaluating the predictive value of sTNFR80 for
DHF are warranted.

Dengue viruses are arthropodborne flaviviruses that cause cence they develop plasma leakage manifested by hemo-
significant morbidity and mortality in tropical and subtropical concentration, ascites, and pleural effusion that may result in
regions of the world. There are four serotypes (dengue 1–4) of shock. It is estimated that 100 million cases of dengue fever
dengue viruses. Classical dengue fever (DF) is a self-limited and 250,000 cases of DHF occur annually [1].
illness characterized by fever, headache, myalgia, arthralgia, Over 90% of DHF cases occur during secondary infections
and abdominal pain. Since the 1950s, a more severe form of [2]. A secondary dengue infection is caused by a different se-
the disease, dengue hemorrhagic fever (DHF), has been rec- rotype of dengue virus than that which caused a given primary
ognized. Patients who develop DHF present clinically in a sim- infection. Several hypotheses have attempted to explain the
ilar fashion to DF patients, but around the time of deferves- increased incidence of severe disease after heterotypic dengue
virus infection. In vitro studies have demonstrated that cross-
Received 6 August 1998; revised 16 November 1998. reactive nonneutralizing dengue antibodies form complexes
Presented in part: 44th annual meeting, American Society of Tropical with heterologous dengue viruses. These virus antibody com-
Medicine and Hygiene, San Antonio, Texas, November 1995 (abstract 160); plexes facilitate viral uptake into monocytes via Fc receptors,
45th annual meeting, American Society of Tropical Medicine and Hygiene,
Baltimore, December 1996 (abstract 126). a process known as “antibody-dependent enhancement” (ADE)
Informed consent was obtained from the parents or guardians of all study [3]. Our laboratory has demonstrated that memory dengue vi-
participants. The study protocol was approved by the institutional review
rus–specific cytotoxic CD81CD42 and CD41CD82 T lympho-
boards of the Thai Ministry of Public Health, the Office of the US Army
Surgeon General, and the University of Massachusetts Medical School. cytes are induced after primary dengue virus infections [4–7].
The opinions contained herein are those of the authors and should not We hypothesize that a positive feedback loop exists in which
be construed as representing the official policies of the NIH, the Department
ADE increases the number of antigen-presenting cells (APCs)
of the Army, or the Department of Defense.
Financial support: NIH (AI-34533); US Army Medical Research and that stimulate dengue cross-reactive memory CD41 and CD81
Materiel Command. T cells. Activation of monocytes and T cells induces the pro-
1
Present affiliations: Department of Virus Diseases, Walter Reed Army
duction of cytokines and chemical mediators, which cause cap-
Institute of Research, Washington, DC (D.W.V.); Department of Virology
I, Tokyo National Institutes of Infectious Diseases, Tokyo (I.K.). illary leakage and may lead to shock [8]. In other viral illnesses,
Reprints or correspondence: Dr. Sharone Green, Center for Infectious such as hemorrhagic fever with renal syndrome [9], hantavirus
Disease and Vaccine Research, University of Massachusetts Medical School,
55 Lake Ave. N., Worcester, MA 01655 ([email protected]).
pulmonary syndrome [10], and atypical measles [11], immune
responses may also underlie the pathogenesis of disease.
The Journal of Infectious Diseases 1999; 179:755–62
q 1999 by the Infectious Diseases Society of America. All rights reserved.
Previously, we found higher levels of the T cell activation
0022-1899/99/7904-0001$02.00 markers, soluble (s) CD4, sCD8, and soluble interleukin (IL)-
756 Green et al. JID 1999;179 (April)

2 receptor (sIL2R), in the sera of children with DHF compared Dickinson, Franklin Lakes, NJ), immediately placed on ice, and
with DF, and elevated levels of IL-2 and interferon (IFN)-g in transported to the blood processing laboratory. Samples were main-
children with dengue infections compared with healthy controls tained at 47C throughout processing. Samples were initially cen-
[12]. Other investigators have found elevated levels of tumor trifuged at 300g for 10 min; platelet-rich plasma was transferred
to polystyrene centrifuge tubes (Becton Dickinson) and centrifuged
necrosis factor (TNF)-a, IL-1b, IL-6, and sTNF receptor p75
at 800g for 10 min. The platelet-poor plasma was divided into
in patients with severe dengue illness [13–18]. In those reports,
aliquots and frozen at 2707C until analysis.
samples were obtained late in the course of illness, patient num- Assays for IL-1b, TNF-a, IL-6, IL-4, IFN-g, sIL2R, sCD8, sCD4,
bers were usually small, serial samples were seldom studied, sTNFR60, and sTNFR80. Cytokines and receptors were mea-
and there were no concurrently enrolled comparison groups sured by commercial ELISAs (IL-1b and TNF-a: Cistron Bio-
with other febrile illnesses. technologies, Pine Brook, NJ; IL-4, IL-6, and IFN-g: Endogen,
The purpose of this study was to investigate the levels of Cambridge, MA; sCD4, sCD8, and sIL2R: T Cell Diagnostics,

Downloaded from https://academic.oup.com/jid/article/179/4/755/888053 by guest on 26 May 2023

cytokines and soluble receptors in children with dengue prior Cambridge, MA; sTNFR60 and sTNFR80 [for measurement of
to the overt manifestations of DHF in order to elucidate the sTNFR-55 and -75 kDa, respectively]: Bender MedSystems, Vi-
immunologic pathways that ultimately lead to plasma leakage. enna) according to manufacturers’ recommendations. The lower
The duration of fever in dengue is typically 3–5 days, and limits of detection were as follows: IL-1b, !2 pg/mL; TNF-a, !10
pg/mL; IL-4 and -6, !1 pg/mL; IFN-g, !5 pg/mL; sCD4, 1.1 U/
plasma leakage tends to occur at or around the time of defer-
mL; sCD8, 50 U/mL; sIL2R, 24 U/mL; sTNFR60, !80 pg/mL;
vescence. We therefore enrolled children with fever of !72 h
and sTNFR80, !150 pg/mL. Samples with levels above the max-
duration so that our observations would precede the period of imum optical density were diluted and retested.
maximal plasma leakage [19]. As a result of this design, the Sample selection. A subset of study subjects was selected from
study included a cohort of controls with other nondengue feb- each of the three diagnostic categories: DHF, DF, and OFIs for
rile illnesses (OFIs), enabling us to determine whether the ob- immunoassay testing. Because the volume of plasma obtained was
served phenomena were specific to dengue virus infection. limited, specimens from the same patients could not be assayed for
all immune response parameters. These subset populations were
selected without knowledge of clinical data other than final diag-
Methods nosis. For each immunoassay, the ranges of patients tested were
Study design. The Dengue Hemorrhagic Fever Project enrolled as follows: DHF, 17–24; DF, 13–20; OFIs, 8–20. We tested 2–6
Thai children aged 6 months to 14 years with fever of !72 h du- serial specimens for each subject in a given immunoassay. For each
ration and no obvious source of infection [19]. Children were hos- subject, we included samples from study day 1, fever days 21 and
pitalized, observed without study-specific intervention, and had a 0, and the next outpatient follow-up visit whenever available.
venous blood sample drawn daily until the day after defervescence Plasma samples from a 6-month follow-up visit from study subjects
and an outpatient blood sample obtained between study days 8 with acute dengue virus infection were tested in each immunoassay
and 11. On the day after defervescence, a right lateral decubitus as healthy controls (n 5 14 –19). All samples were tested under code.
chest radiograph was obtained. The pleural effusion index (PEI) Statistical analysis. Mean plasma levels of the measured par-
was calculated as follows: [(width of effusion/width of hemithorax) ameters were initially compared by two-factor analysis of variance
3 100]. A total of 189 children were enrolled between 25 April and (ANOVA) using final diagnosis (DHF, DF, or OFIs) and fever day.
14 December 1994. Sixty patients had dengue (28 with DHF, 32 When significant differences were noted between groups based on
with DF) and 112 children had OFIs (presumed to be viral). A final diagnosis, we tested for significant differences on each fever
subset of these enrolled subjects was evaluated in the present study day with general linear models, an extension of ANOVA, using
(see below). conservative imputation procedures. Comparisons were made be-
Study definitions. Study day 1 was the day a child was enrolled tween all subjects with acute dengue virus infection (combined DF
in the study. Fever day 0 was the day of defervescence, when the and DHF) and those with OFIs and also between subjects with
temperature dropped below 387C without a subsequent elevation. DHF and DF. We used both Pearson’s and Spearman’s correla-
Days before fever day 0 are designated fever day 21, 22, etc. A tions. Raw values were log transformed to stabilize the variances.
clinical diagnosis of DHF and severity grading (grades 1–4) were Differences between diagnostic groups in the proportion of samples
assigned according to WHO criteria [20]. All children with the with detectable levels of TNF-a were tested using Fisher’s exact
clinical diagnoses of DF or DHF had evidence of acute dengue test. Due to the small number of primary dengue cases (4/60), an
virus infection by dengue IgM ELISA [21], hemagglutination in- analysis comparing subjects with primary and secondary dengue
hibition (HAI) antibody responses [22], or dengue virus isolation infection was not done. Analyses were done on the Harvard School
in Toxorhynchites splendens mosquitoes [23–25]. Any subject with of Public Health computer using an SAS package [26, 27].
evidence of acute dengue infection who did not meet criteria for P ! .05 was considered significant.
DHF was assigned a clinical diagnosis of DF. OFIs were defined
as those patients who had no dengue virus–specific IgM or HAI
antibody responses, no dengue virus isolated from plasma, and no Results
obvious bacterial, rickettsial, or protozoal etiology for their illness
and were presumed to most likely have self-limited viral illnesses. Clinical data at study entry. Clinical findings for this study
Sample processing. Blood was drawn into EDTA tubes (Becton have been reported [19]. The subset population included 26
JID 1999;179 (April) Early Immune Activation in Dengue Illness 757

Downloaded from https://academic.oup.com/jid/article/179/4/755/888053 by guest on 26 May 2023

Figure 1. 80-kDa plasma soluble tumor necrosis factor (TNF) receptors in dengue hemorrhagic fever (DHF), dengue fever (DF), and other
febrile illnesses (OFIs). Fever day 0 represents day of defervescence. Horizontal lines indicate mean values. Outpt 5 outpatient follow-up sample
at study days 8–13. OFIs vs. dengue (DHF 1 DF): * P ! .05, *** P ! .001; DHF vs. DF: ¶ P ! .10, ‡ P ! .05, ‡‡ P ! .01.

children with DHF, 24 with DF, and 27 with OFIs. Among the DF. There were no significant differences between subjects se-
children with DHF, 5 (19%) were grade 1, 12 (46%) were grade lected for study and those not selected for study in any diag-
2, 9 (35%) were grade 3, and none were grade 4. Four children nostic group for sex, age, duration of fever before enrollment,
had primary dengue virus infection, and all 4 were classified as symptoms, platelet count, hematocrit level, absolute monocyte

Figure 2. 60-kDa plasma soluble tumor necrosis factor (TNF) receptors in dengue hemorrhagic fever (DHF), dengue fever (DF), and other
febrile illnesses (OFIs). Fever day 0 represents day of defervescence. Horizontal lines indicate mean values. Outpt 5 outpatient follow-up sample
at study days 8–13. OFIs vs. dengue (DHF 1 DF): * P ! .05, ** P ! .01.
758 Green et al. JID 1999;179 (April)

Mean plasma sTNFR80 levels were significantly higher in chil-

dren with DHF than DF from 2 days before defervescence
(P ! .01) until the day of defervescence (P ! .01 ). Mean plasma
levels of sTNFR80 and sTNFR60 were significantly elevated
in children with dengue compared with those with OFIs on the
same days (P ! .001 and ! .05, respectively) (figures 1, 2). Levels
of sTNFR80 measured 2 days before defervescence in children
with dengue virus infections correlated with the pleural effusion
indices measured 1 day after defervescence (Pearson’s r 5
0.788, P 5 .0001; Spearman’s r 5 0.620, P 5 .0003) (figure 3).

Downloaded from https://academic.oup.com/jid/article/179/4/755/888053 by guest on 26 May 2023

IFN-g levels. Mean plasma IFN-g levels were significantly
higher in children with dengue than in those with OFIs as early
as fever day 22 (P ! .05), and this difference became more
apparent on fever days 21 and 0 (P ! .05 and ! .001, respec-
tively). Mean IFN-g levels were not significantly different in
patients with DHF than in those with DF on any given fever
day. However, we noted that IFN-g levels increased and de-
Figure 3. Correlation between 80-kDa plasma soluble tumor ne- creased abruptly in patients with DHF and that by fever day
crosis factor receptor (sTNFR) levels 2 days before defervescence and
11 (after defervescence), the mean level in children with DHF
pleural effusion index (PEI) 1 day after defervescence. PEI 5
100 3 (width of pleural effusion/width of hemithorax) on decubitus (20.9 pg/mL) was similar to that of children with DF (19.0 pg/
chest radiograph. Open symbols, dengue fever; closed symbols, dengue mL) or OFIs (16.2 pg/mL). The day of peak IFN-g production
hemorrhagic fever. for all patients occurred before or on the day of defervescence
(figure 4).
count, or alanine aminotransferase. Subjects with OFIs selected TNF-a, IL-1b, IL-4, and IL-6 levels. There were no dif-
for study had slightly higher aspartate aminotransferase levels ferences between groups in mean plasma levels of TNF-a, IL-
than those not selected, and subjects with DF selected for study 1b, IL-4, or IL-6 (data not shown). However, the proportion
had slightly higher total white blood cells, absolute neutrophils, of samples with detectable TNF-a was greater in children with
and albumin than those not selected. DHF than in those with DF or OFIs (table 1). By Fisher’s
sTNFR80 and sTNFR60 levels in children with dengue. exact test, the difference in the proportion of samples with

Figure 4. Plasma interferon (IFN)-g in dengue hemorrhagic fever (DHF), dengue fever (DF), and other febrile illnesses (OFIs). Fever day 0
represents day of defervescence. Horizontal lines indicate mean values. Outpt 5 outpatient follow-up sample at study days 8–13. OFIs vs. dengue
(DHF 1 DF): * P ! .05, *** P ! .001.
JID 1999;179 (April) Early Immune Activation in Dengue Illness 759

Table 1. Proportion of samples with detectable levels of tumor ne- higher levels of sTNFR80, IFN-g, sCD8, and sIL2R and the
crosis factor (TNF)-a.
more frequent detection of TNF-a in children who developed
Fever day DHF during the course of the study (table 2). Activated T cells
Diagnosis 22 21 0 release TNF-a [28], IFN-g [29], sTNF receptors [30], sIL2R
Other febrile illnesses 0/2 (0) 1/6 (17) 1/8 (13) [31], and sCD8 [32, 33]. Our present findings of higher levels
Dengue fever (DF) 0/10 (0) 2/19 (11) 1/19 (6) of sIL2R and sCD8 in children who develop DHF are consis-
Dengue hemorrhagic fever (DHF) 3/12 (25) 4/16 (25) 5/13 (38)
tent with those of our prior study in Thai children later in the
NOTE. No. of specimens with detectable TNF-a/total specimens tested (%); course of dengue infections [12] and support a role for CD81
comparison of DHF vs. DF by Fisher’s exact test P ! .05 for fever day 0 and
!.10 for fever days 22 and 21 combined.
T cell activation in the pathogenesis of DHF. In addition, we
have found higher levels of sTNFR80 and IFN-g and were
detectable TNF-a was statistically significant on the day of more frequently able to detect TNF-a in children with DHF,

Downloaded from https://academic.oup.com/jid/article/179/4/755/888053 by guest on 26 May 2023

defervescence (P ! .05 ), and there was a trend toward statistical which indicates that immune activation precedes defervescence
significance before defervescence (P ! .10 ). Logistic regression and the onset of plasma leakage and that the increased immune
analysis showed that the proportion of samples with detectable activation is associated with the severity of dengue illness. Al-
TNF-a was significantly higher (P ! .05 ) in subjects with DHF though others have reported elevated levels of sTNF receptors
over this whole time period. [17, 18] and TNF-a in DHF [13, 14, 16], our study is the first
sIL2R, sCD8, and sCD4 levels. Levels of sIL2R and sCD8 to compare serial levels in children with DF and DHF and to
were significantly higher in children with dengue than in chil- include a febrile control group without dengue infection.
dren with OFIs on the day of defervescence (P ! .05 ) (figures IFN-g levels were elevated early in children with dengue com-
5, 6). There was a trend (P ! .10 ) toward higher levels of sIL2R pared with children with OFIs. The curve of IFN-g production
and sCD8 in children with DHF compared with those with DF in some children with dengue was very abrupt. This increase
from as early as fever day 21, and this achieved statistical did not occur in all patients, but was seen more often in children
significance on the day following defervescence (P ! .001 and with DHF than in those with DF. The most likely explanation
! .05, respectively). sCD4 levels were not significantly different for these results is the transient nature of cytokine production
between children with OFIs, DF, or DHF (data not shown). and the short plasma half-life of these compounds. Our daily
Cytokine-to-albumin ratios. In the 1998 study of Bethell et blood sampling could have missed a short-lived burst of IFN-
al. [18], cytokine concentrations were adjusted to albumin levels g activity. Although peak IFN-g levels occurred on different
in order to account for the translocation of small proteins dur- days, the peak occurred before or on the day of defervescence.
ing plasma leakage. Our data were analyzed as a ratio of cy- Presumably, dengue virus–specific cross-reactive memory T cells
tokine-to-albumin levels (data not shown). No new associations are activated by dengue-infected APCs to produce IFN-g early
were noted using these procedures, but the relationship between in infection [34]. The abrupt decline in IFN-g levels after de-
sIL2R and sCD8 and disease severity (DF vs. DHF) were fervescence coincides with the disappearance of viremia [25]
strengthened. The sIL2R:albumin ratio was significantly higher and may be due to lysis of dengue-infected APCs. Production
in children with DHF than in those with DF on fever days 23, of IL-4 and IL-10 can also down-regulate IFN-g production,
21, and 0 (P ! .05 ). Similarly, the sCD8:albumin ratio was but we found no increase in IL-4 in these patients.
significantly higher in children with DHF than in those with The elevated levels of IFN-g, TNF-a, and sTNFRs in chil-
DF 1 day before and on the day of defervescence (P ! .01). dren with DHF are of great interest. IFN-g and TNF-a have
There was no change in the relationship between sTNFR80 a synergistic effect on endothelial cell cultures in vitro by in-
and disease severity utilizing this ratio. creasing monolayer permeability, which might play a role in
capillary leakage [35, 36]. In addition, they activate endothelial
cells as demonstrated by up-regulation of both major histo-
Discussion
compatibility complex class I and expression of soluble intra-
The prospective design of the present study permitted us to cellular adhesion molecule-1 on the cell surface, thereby ena-
measure cytokines and T cell activation markers early in the bling lymphocyte binding and migration into the tissues [37].
course of dengue illness and to evaluate the kinetics of these IFN-g up-regulates expression of TNF receptors on myeloid
responses in individual patients during the course of their ill- and epithelial cells, which may render these cells more sensitive
ness. Samples from a cohort of children with nondengue febrile to the effects of TNF-a [38–40]. In vivo evidence of the inter-
illnesses were available for comparison. While circulating levels action between these cytokines also exists. Mouse models of
of cytokines may or may not reflect secretion at the tissue level, cerebral malaria and bacterial sepsis have demonstrated that
our study design enabled us to evaluate the relationship of these the TNF-a–mediated morbidity or mortality is regulated by
markers to acute dengue virus infection and to the development IFN-g [41–44].
of the plasma leakage seen in DHF. A recent study in Vietnam [18] reported that sTNFR80 155
The major findings of this study were the demonstration of pg/mL had a sensitivity of 93% and a specificity of 34% for
760 Green et al. JID 1999;179 (April)

Downloaded from https://academic.oup.com/jid/article/179/4/755/888053 by guest on 26 May 2023

Figure 5. Plasma soluble interleukin (IL)-2 receptors in dengue hemorrhagic fever (DHF), dengue fever (DF), and other febrile illnesses (OFIs).
Fever day 0 represents day of defervescence. Horizontal lines indicate mean values. Outpt 5 outpatient follow-up sample at study days 8–13.
OFIs vs. dengue (DHF 1 DF): * P ! .05; DHF vs. DF: ¶ P ! .10, ‡‡ P ! .01.

predicting the development of shock in children with suspected of defervescence only, when we utilized a cutoff value of 1.6
dengue. We were unable to assess prediction of shock as there ng/mL, we found a sensitivity of 94%, specificity of 25%, pos-
were too few shock cases; however, we analyzed the potential itive predictive value (PPV) of 53% and negative predictive
value of sTNRr80 to differentiate DF and DHF. On the day value (NPV) of 83%. However, when we examined sTNFR80

Figure 6. Plasma soluble CD8 in dengue hemorrhagic fever (DHF), dengue fever (DF), and other febrile illnesses (OFIs). Fever day 0 represents
day of defervescence. Horizontal lines indicate mean values. Outpt 5 outpatient follow-up sample at study days 8–13. OFIs vs. dengue (DHF 1
DF): * P ! .05, *** P ! .001; DHF vs. DF: ¶ P ! .10, ‡ P ! .05.
JID 1999;179 (April) Early Immune Activation in Dengue Illness 761

Table 2. Summary of changes in plasma levels of cytokines and Infection enhancement by non-neutralizing antibody. J Exp Med 1977;
soluble receptors in Thai children with dengue hemorrhagic fever 146:201–17.
(DHF), dengue fever (DF), and other febrile illnesses (OFIs) during 4. Kurane I, Meager A, Ennis FA. Dengue virus–specific human T cell clones.
the course of acute illness. Serotype crossreactive proliferation, interferon-g production, and cyto-
toxic activity. J Exp Med 1989; 170:763–75.
DHF DF OFI
5. Bukowski JF, Kurane I, Lai CJ, Bray M, Falgout B, Ennis FA. Dengue
Soluble (s) TNFr80 FF F r virus–specific cross-reactive CD81 human cytotoxic T lymphocytes. J Virol
sTNFr60 F F r 1989; 63:5086–91.
Interferon-g FF F F
6. Green S, Kurane I, Edelman R, et al. Dengue virus–specific human CD41
TNF-a F r r
T-lymphocyte responses in a recipient of an experimental live-attenuated
sCD8 FF F r
sIL2R FF F F dengue virus type-1 vaccine: bulk culture proliferation, clonal analysis,
and precursor frequency determination. J Virol 1993; 67:5962–7.
NOTE. r, little to no change in level compared to healthy controls; F, mild 7. Gagnon SJ, Zeng W, Kurane I, Ennis FA. Identification of two epitopes on
but significant increase compared to controls; FF, moderate increase compared

Downloaded from https://academic.oup.com/jid/article/179/4/755/888053 by guest on 26 May 2023

the dengue 4 virus capsid protein recognized by a serotype-specific and a
to controls. TNF, tumor necrosis factor; IL, interleukin.
panel of serotype cross-reactive human CD41 cytotoxic T-lymphocyte
clones. J Virol 1996; 70:141–7.
levels 2 days before defervescence, the same cutoff had sensi- 8. Kurane I, Ennis FA. Cytokines in dengue virus infections: role of cytokines
tivity decreased to 67%, specificity increased to 80%, a PPV of in the pathogenesis of dengue hemorrhagic fever. Semin Virol 1994; 5:
443–8.
66%, and an NPV of 69%. Regardless of the cutoff value used,
9. Krakauer T, LeDuc JW, Krakauer H. Serum levels of tumor necrosis factor-
the sensitivity for predicting DHF decreased earlier in the alpha, interleukin-1, and interleukin-6 in hemorrhagic fever with renal
course of illness. syndrome. Viral Immunol 1995; 8:75–9.
If febrile children in an endemic area can be rapidly diag- 10. Ennis FA, Cruz J, Spiropoulou CF, et al. Hantavirus pulmonary syndrome:
nosed with a dengue infection (e.g., by polymerase chain re- CD81 and CD41 cytotoxic T lymphocytes to epitopes on Sin Nombre
virus nucleocapsid protein isolated during acute illness. Virology 1997;
action) [45–47], an elevated immune marker such as sTNFR80
238:380–90.
might help identify those at highest risk for developing DHF. 11. Griffin DE, Ward BJ, Esolen LM. Pathogenesis of measles virus infection:
If children at high risk for developing DHF can be identified a hypothesis for altered immune responses. J Infect Dis 1994; 170(suppl):
early in the course of disease, clinical studies of therapeutic S24–31.
interventions to avert plasma leakage would be feasible. Fur- 12. Kurane I, Innis BL, Nimmannitya S, et al. Activation of T lymphocytes in
dengue virus infections. High levels of soluble interleukin 2 receptor, sol-
ther studies of immune activation markers should be done to
uble CD4, soluble CD8, interleukin 2, and interferon-g in sera of children
determine their predictive value for DHF. with dengue. J Clin Invest 1991; 88:1473–80.
13. Vitarana T, de Silva H, Withana N, Gunasekera C. Elevated tumour necrosis
Acknowledgments factor in dengue fever and dengue haemorrhagic fever. Ceylon Med J
1991; 36:63–5.
We thank Narongchai Kunentrasai, Atchara Teeraratkul, Wiboon 14. Hober D, Poli L, Roblin B, et al. Serum levels of tumor necrosis factor-a
Viramitrchai, and Supaporn Kiatpolpoj for excellence in carrying out (TNF-a), interleukin-6 (IL-6), and interleukin-1-b (IL-1-b) in dengue-
infected patients. Am J Trop Med Hyg 1993; 48:324–31.
the clinical studies; the Armed Forces Research Institute Medical Sci-
15. Kuno G, Bailey RE. Cytokine responses to dengue infection among Puerto
ences research nurses (Pranom Vangnai, Nathada Plavooth, Somnuk
Rican patients. Mem Inst Oswaldo Cruz 1994; 89:179–82.
Lumjiak, Somalvadee Saravasee, Sumetha Hengprasert, and Wipa
16. Kalayanarooj S, Nimmannitya S, Lortholary O, Mencia-Heuta JM, Braque
Chawachalasai) for dedication in data and blood collection; the ward P. Immune mediators in dengue and dengue hemorrhagic fever patients.
nurses at Bangkok Children’s Hospital and Kamphaeng Phet Provin- JAMA Southeast Asia 1994; 10(Suppl):400–2.
cial Hospital for excellent patient care; Kittinan Hussem, Somkiat 17. Hober D, Delannoy AS, Benyoucef S, De Groote D, Wattre P. High levels
Changnak, Choompun Manomuth, Aree Na-nongkhai, Pranee Sai- of sTNFR p75 and TNFa in dengue-infected patients. Microbiol Immunol
sang, Somsamai Tapana, Weerasak Yeepoo, and Songdej Sangsri for 1996; 40:569–73.
excellent technical assistance; Tipawan Kungvanrattana and Warinda 18. Bethell DB, Flobbe K, Phuong CXT, et al. Pathophysiologic and prognostic
Srikham for data entry; Christine Kozik for auditing the clinical data; role of cytokines in dengue hemorrhagic fever. J Infect Dis 1998; 177:
778–82.
Chitchai Hemachudha for database management; Jana Kubrin for as-
19. Kalayanarooj S, Vaughn DW, Nimmannitya S, et al. Early clinical and lab-
sistance with manuscript preparation; Pantipa Patanawin for interpre-
oratory indicators of acute dengue illness. J Infect Dis 1997; 176:313–21.
tation of chest radiographs; and Timothy P. Endy for advice during
20. Dengue haemorrhagic fever: diagnosis, treatment and control Geneva: World
manuscript preparation. Health Organization, 1986.
21. Innis BL, Nisalak A, Nimmannitya S, et al. An enzyme-linked immunosor-
bent assay to characterize dengue infections where dengue and Japanese
References encephalitis co-circulate. Am J Trop Med Hyg 1989; 40:418–27.
22. Clarke DH, Casals J. Techniques for hemagglutination and hemagglutina-
1. Monath TP. Dengue: the risk to developed and developing countries. Proc tion-inhibition with arthropod-borne viruses. Am J Trop Med Hyg 1958;7:
Natl Acad Sci USA 1994; 91:2395–400. 561–73.
2. Halstead SB, Nimmannitya S, Cohen SN. Observations related to patho- 23. Rosen L, Gubler DJ. The use of mosquitoes to detect and propagate dengue
genesis of dengue hemorrhagic fever. IV. Relation of disease severity to viruses. Am J Trop Med Hyg 1974; 23:1153–60.
antibody response and virus recovered. Yale J Biol Med 1970; 42:311–28. 24. Rosen L. The use of toxorhynchites mosquitoes to detect and propagate
3. Halstead SB, O’Rourke EJ. Dengue viruses and mononuclear phagocytes. I. dengue and other arboviruses. Am J Trop Med Hyg 1981; 30:177–83.
762 Green et al. JID 1999;179 (April)

25. Vaughn DW, Green S, Kalayanarooj S, et al. Dengue in the early febrile 37. Pober JS. Cytokine-mediated activation of vascular endothelium. Am J
phase: viremia and antibody responses. J Infect Dis 1997; 176:322–30. Pathol 1988; 133:426–33.
26. SAS/STAT user’s guide, version 6. 4th ed. Vols 1 and 2. Cary, NC: SAS 38. Aggarwal BB, Eessalu TE, Hass PE. Characterization of receptors for human
Institute, 1994. tumour necrosis factor and their regulation by g-interferon. Nature
27. Armitage P. Statistical methods in medical research. New York: Wiley, 1971. 1985; 318:665–7.
28. Sung SSJ, Bjorndahl JM, Wang CY, Kao HT, Fu SM. Production of tumor 39. Ruggiero V, Tavernier J, Fiers W, Baglioni C. Induction of the synthesis of
necrosis factor/cachectin by human T cell lines and peripheral blood T tumor necrosis factor receptors by interferon-g. J Immunol 1986; 136:
lymphocytes stimulated by phorbol myristate acetate and anti-CD3 an- 2445–50.
tibody. J Exp Med 1988; 167:937–53. 40. Pandita R, Pocsik E, Aggarwal BB. Interferon-g induces cell surface ex-
29. Trinchieri G, Perussia B. Immune interferon: a pleiotropic lymphokine with pression for both types of tumor necrosis factor receptors. FEBS Lett
multiple effects. Immunol Today 1985; 6:131–6. 1992; 312:87–90.
30. Bemelmans MHA, Abramowicz D, Gouma DJ, Goldman M, Buurman WA. 41. Grau GE, Heremans H, Piguet PF, et al. Monoclonal antibody against in-
In vivo T cell activation by anti-CD3 monoclonal antibody induces soluble terferon-g can prevent experimental cerebral malaria and its associated

Downloaded from https://academic.oup.com/jid/article/179/4/755/888053 by guest on 26 May 2023

overproduction of tumor necrosis factor. Proc Natl Acad Sci USA 1989;86:
TNF receptor release in mice. J Immunol 1994; 153:499–506.
5572–4.
31. Rubin LA, Kurman CC, Fritz ME, et al. Soluble interleukin 2 receptors are
42. Heinzel FP. The role of IFN-g in the pathology of experimental endotoxemia.
released from activated human lymphoid cells in vitro. J Immunol
J Immunol 1990; 145:2920–4.
1985; 135:3172–7.
43. Waki S, Uehara S, Kanbe K, Ono K, Suzuki M, Nariuchi H. The role of T
32. Fujimoto J, Levy S, Levy R. Spontaneous release of the leu-2 (T8) molecule
cells in pathogenesis and protective immunity to murine malaria. Im-
from human T cells. J Exp Med 1983; 158:752–66.
munology 1992; 75:646–51.
33. Tomkinson BE, Brown MC, Ip SH, Carrabis S, Sullivan JL. Soluble CD8
44. Car BD, Eng VM, Schnyder B, et al. Interferon-g receptor deficient mice are
during T cell activation. J Immunol 1989; 142:2230–6.
resistant to endotoxic shock. J Exp Med 1994; 179:1437–44.
34. Kurane I, Innis BL, Nisalak A, et al. Human T cell responses to dengue 45. Morita K, Maemoto T, Honda S, et al. Rapid detection of virus genome
virus antigens. Proliferative responses and interferon gamma production. from imported dengue fever and dengue hemorrhagic fever patients by
J Clin Invest 1989; 83:506–13. direct polymerase chain reaction. J Med Virol 1994; 44:54–8.
35. Stolpen AH, Guinan EC, Fiers W, Pober JS. Recombinant tumor necrosis 46. Seah CLK, Chow VTK, Tan HC, Chan YC. Rapid, single-step RT-PCR
factor and immune interferon act singly and in combination to reorganize typing of dengue viruses using five NS3 gene primers. J Virol Methods
human vascular endothelial cell monolayers. Am J Pathol 1986; 123:16–24. 1995; 51:193–200.
36. Burke-Gaffney A, Keenan AK. Modulation of human endothelial cell per- 47. Sudiro TM, Ishiko H, Green S, et al. Rapid diagnosis of dengue viremia by
meability by combinations of the cytokines interleukin-1 a/b, tumor ne- reverse transcriptase–polymerase chain reaction using 30-noncoding region
crosis factor-a, and interferon-g. Immunopharmacology 1993; 25:1–9. universal primers. Am J Trop Med Hyg 1997; 56:424–9.