1. 706 www.thelancet.com/infection Vol 14 August 2014
Articles
Efficacy and safety of celgosivir in patients with dengue
fever (CELADEN): a phase 1b, randomised, double-blind,
placebo-controlled, proof-of-concept trial
Jenny G Low*, Cynthia Sung*, LiminWijaya*, YuanWei, Abhay P S Rathore, SatoruWatanabe, Boon HianTan, LiyingToh, LianTee Chua,
Yan’an Hou, Angelia Chow, Shiqin Howe,Wing Ki Chan, Kah HinTan, Jasmine S Chung, Benjamin P Cherng, David C Lye, Paul ATambayah,
Lee Ching Ng, John Connolly, Martin L Hibberd,Yee Sin Leo,Yin Bun Cheung, Eng Eong Ooi*, Subhash GVasudevan
Summary
Background Dengue infection is the most common mosquito-borne viral disease worldwide, but no suitable antiviral
drugs are available. We tested the α-glucosidase inhibitor celgosivir as a treatment for acute dengue fever.
Methods To establish eligibility for inclusion in a phase 1b, randomised, double-blind, placebo-controlled, proof-of-
concept trial, individuals aged 21–65 years who had had a fever (≥38°C) for less than 48 h, met at least two criteria
indicating probable dengue infection, and had a positive result on a dengue point-of-care test kit or PCR assay were
referred for screening at a centre in Singapore between July 30, 2012, and March 4, 2013. Using a web-based system,
we randomly assigned patients who met full inclusion criteria after screening (1:1; random permuted block length
four) to celgosivir (initial 400 mg loading dose within 6 h of randomisation, followed by 200 mg every 12 h for a total
of nine doses) or matched placebo. Patients and the entire study team were masked to group assignment. The primary
endpoints were mean virological log reduction (VLR) from baseline for days 2, 3, and 4, and area under the fever
curve (AUC) for a temperature above 37°C from 0 h to 96 h. Efficacy analyses were by intention to treat. This study is
registered with ClinicalTrials.gov, number NCT01619969.
Findings We screened 69 patients and randomly assigned 50 (24 to celgosivir, 26 to placebo). Mean VLR was greater in
the celgosivir group (–1·86, SD 1·07) than in the placebo group (–1·64, 0·75), but the difference was non-significant
(–0·22, 90% CI –0·65 to 0·22; one-sided p=0·203). The mean AUC was also higher in the celgosivir group (54·92,
SD 31·04) than in the placebo group (40·72, 18·69), but again the difference was non-significant (14·20, 90% CI
2·16–26·25; one-sided p=0·973). We noted similar incidences of adverse events between groups.
Interpretation Although generally safe and well tolerated, celgosivir does not seem to reduce viral load or fever burden
in patients with dengue.
Funding STOP Dengue Translational Clinical Research.
Introduction
Dengue infection is the most common mosquito-borne
viral disease worldwide. In 2013, Bhatt and colleagues1
estimated that 390 million infections occur per year,
with 96 million becoming symptomatic, and that 70% of
cases occur in Asia. The number of infections is expected
to rise with rapid urbanisation, international travel, and
global warming.2
Infection with any one of the four
antigenically distinct dengue virus serotypes (DENV1–4)
leads to a range of problems, from self-limiting febrile
illness to life-threatening severe dengue, which encom-
passes hypovolaemic shock from vascular leakage, internal
haemorrhage, and organ dysfunction. Mild dengue is
debilitating and contributes to substantial morbidity and
loss of economic productivity.2–8
No antiviral drugs against dengue are available. The
lifecycle of the virus offers many targets for drug
development and much focus has been placed on the
multifunctional enzyme NS3 (necessary for viral
polyprotein processing) and NS5 (necessary for RNA
replication).9–15
Clinical trials of chloroquine16
(a potential
inhibitor of dengue virus entry into host cells) and
balapiravir17
(a nucleoside inhibitor) did not show any
efficacy. Vaccine development has also been challenging:
the most advanced tetravalent dengue vaccine candidate
has only 30% efficacy,18
and researchers have called for
new approaches to vaccine development.19
Effective
interventions that either reduce risk of severe dengue or
halt transmission are urgently needed.
Celgosivir (or Bu-Cast) is a 6-O butanoyl prodrug of
castanospermine, a naturally occurring iminosugar
derived from the seeds of Castanospermum australe.20
It
exerts antiviral activity by inhibiting the endoplasmic-
reticulum-resident α-glucosidase I enzyme that, together
with α-glucosidase II, is needed for the trimming of
three terminal glucose residues attached to N-glycans of
newly synthesised glycoproteins.21
Iminosugars, such as
castanospermine and N-butyl deoxynojiramycin, can
exert broad antiviral properties by interfering with virus
morphogenesis through misfolding of glycosylated
proteins, including those encoded by the dengue virus
genome, such as E, prM, and NS1.21–23
In-vitro dengue
Lancet Infect Dis 2014;
14: 706–15
Published Online
May 28, 2014
http://dx.doi.org/10.1016/
S1473-3099(14)70730-3
See Comments page 661
*Contributed equally
Program in Emerging
Infectious Diseases
(J G Low MPH, C Sung PhD,
A P S Rathore MSc,
S Watanabe PhD, B H Tan BSc,
L Toh BSc, L T Chua BSc,
Y Hou BSc, A Chow Dip Sc,
S Howe BSc, W K Chan BSc,
K H Tan BSc, E E Ooi FRCPath,
Prof S G Vasudevan PhD) and
Center for Quantitative
Medicine
(ProfY B Cheung PhD), Duke-
NUS Graduate Medical School,
Singapore, Singapore;
Department of Infectious
Diseases, Singapore General
Hospital, Singapore, Singapore
(L Wijaya MRCP, J S Chung MRCP,
B P Cherng MRCP); Singapore
Clinical Research Institute,
Singapore, Singapore
(Y Wei MSc); Communicable
Disease Centre,TanTock Seng
Hospital, Singapore, Singapore
(D C Lye FRACP,Y S Leo FRCP);
National University Hospital
of Singapore, Singapore,
Singapore
(Prof P A Tambayah MD);
Environmental Health
Institute, National
Environment Agency,
Singapore, Singapore
(L C Ng PhD); Program in
Translational Immunology,
Singapore Immunology
Network, Singapore,
Singapore (J Connolly PhD);
Genome Institute of
Singapore, Singapore,
Singapore
(Prof M L Hibberd PhD); and
Department of International
Health, University ofTampere,
Tampere, Finland
(ProfY B Cheung)

2. Articles
www.thelancet.com/infection Vol 14 August 2014 707
infection assays have shown that celgosivir inhibits all
four serotypes of the virus at submicromolar
concentrations.23
In the case of DENV2, NS1 misfolds
and accumulates in the endoplasmic reticulum of
infected cells. A dose-dependent reduction of luciferase
reporter activity in celgosivir-treated DENV2 replicon
cells suggests that NS1 is necessary for viral RNA
replication.23
In vivo, celgosivir has activity in mice even
when treatment is delayed for 48 h after infection.23–25
A
study25
has shown that the drug’s effectiveness is
dependent on dosing frequency—50 mg/kg twice daily
yielded 100% survival in an otherwise lethal dengue
infection model, but 100 mg/kg once a day conferred no
protection.
Celgosivir has been tested in phase 1 and 2 trials as a
possible treatment for infection with HIV and infection
with hepatitis C virus,20,26
which, like dengue, is a
flavivirius. However, its efficacy was not superior to that
of existing treatments, so testing was not continued.
Nevertheless, no cardiovascular or neurological
complications were reported; adverse events were
typically osmotic diarrhoea and flatulence, as for
acarbose—an approved α-glucosidase inhibitor used for
treatment of type 2 diabetes.20
Unlike antiviral drug
treatment of hepatitis C and HIV infections—chronic
diseases that require long-term regimens of months to
years—an acute disease like dengue would only need a
short course. We aimed to assess the activity and safety of
celgosivir as treatment for acute dengue.
Methods
Study design and participants
CELADEN was a phase 1b, randomised, double-blind,
placebo-controlled, proof-of-concept trial done in one
centre in Singapore, where dengue is endemic. Between
July 30, 2012, and March 4, 2013, patients were referred
for screening at the SingHealth Investigational Medicine
Unit at Singapore General Hospital, Singapore, from
private and public primary-care clinics and collaborating
hospitals in Singapore. Patients aged 21–65 years were
referred when they had had a fever (≥38°C) for less
than 48 h, met at least two criteria indicating probable
dengue infection (on the basis of the 2009 WHO dengue
classification scheme;27
appendix), and had a positive
result on a dengue point-of-care test kit or PCR assay.
During screening, we obtained detailed medical histories
from every patient and did physical examinations, a
repeat Dengue Duo test to confirm dengue infection
(with a commercially available point-of-care SD Dengue
Duo rapid test kit [Standard Diagnostic, Yongin, South
Korea] or a previously described RT-PCR assay28
), chest
radiographs, urinalysis, and liver and renal function
tests. Additionally, we recorded full blood counts and
obtained electrocardiograms. Patients meeting full
inclusion criteria (appendix) were enrolled into the trial.
The study was approved by the Singapore Health
Services’ Centralised Institutional Review Board
(reference 2012/025/E) and was monitored independently
by the Singapore Clinical Research Institute, a publicly
funded clinical research organisation. The Health
Sciences Authority granted approval for the trial protocol.
The clinical data obtained in the CELADEN study have
been independently audited by Clinical Network Services
(Toowong, QLD, Australia; audit certificate issued on
Sept 12, 2013) at the request of 60° Pharmaceuticals
(Washington, DC, USA). Neither party had any role in
the design, conduct, or funding of this study.
Randomisation and masking
We randomly assigned enrolled patients (1:1) to receive
celgosivir or placebo. We used random permuted blocks
of length four to ensure balance over time. The block
length was established by the randomisation statistician
and was not made known to the clinical investigators and
site personnel before completion of the study.
Randomisation was done via a web-based system hosted
by the Singapore Clinical Research Institute. Packages of
study drug and placebo were coded at the manufacturing
site with codes sent from the randomisation statistician.
Patients, principal investigators, the research pharmacist,
research nurses, and the rest of the study team were fully
masked until completion of the study and locking of the
database. We used structured case report forms to obtain
all clinical data, which were then entered into a secure,
web-based database maintained by the Singapore Clinical
Research Institute. Investigators had no access to the
database until completion of the study.
Procedures
Participants were admitted to hospital for 5 days, during
which time they received capsules of placebo or
celgosivir. Patients in the celgosivir group received an
initial loading dose of 400 mg celgosivir within 6 h of
randomisation. They then received maintenance doses
of 200 mg every 12 h for a total of nine doses (total
dose 2·0 g). We decided what doses to use on the basis of
previous clinical trials of celgosivir in more
than 600 patients with HIV or hepatitis C infection,20
human pharmacokinetic simulations of various dosing
regimens,29
and pharmacology and pharmacokinetic
results.25,26
Dalton Pharma Services (Toronto, ON, Canada)
manufactured celgosivir, produced capsules, and packaged
them in blister packs for the study. Patients assigned to
placebo received capsules of pregelatinised maize starch
that were packaged in blister packs identical to those for
celgosivir. The chemistry, manufacturing, and control
package for the investigational product was assessed by the
Health Sciences Authority of Singapore and was approved
as a drug product for CELADEN.
On day 5, we discharged patients who had satisfactory
clinical status. At any time, patients at risk for or who had
severe dengue, or who had adverse events requiring
continued admission were transferred to a medical ward
of the hospital and treated accordingly until they were
See Online for appendix
Correspondence to:
Prof Subhash GVasudevan,
Program in Emerging Infectious
Diseases, DUKE-NUS Graduate
Medical School, 8 College Road,
Singapore 169857, Singapore
subhash.vasudevan@duke-
nus.edu.sg

3. Articles
708 www.thelancet.com/infection Vol 14 August 2014
deemed to be fit for discharge. Patients attended follow-
up visits on days 7, 10, and 15, in which we obtained
clinical histories and blood, and did physical
examinations. We followed up all patients who had
adverse events until complete resolution of the problem,
with unscheduled visits after 15 days.
We synthesised plasmid standards for DENV1–4 for real-
time PCR (rtPCR) and quantified the plasmid DNA using
spectrophotometric methods and converted it to copy
number, as previously described.30
Using these calculated
values, we obtained standard curves for dilutions ranging
from 10¹⁰ to 10¹ copies of the virus. We extracted viral RNA
from participants’ serum samples using QIAamp Viral
RNA Mini Kit (Qiagen, Hilden, Germany) according to
manufacturer’s instructions. For quantification, we used a
protocol for rtPCR31
with some modifications: every
reaction contained 50 pmol of both DENV1-specific and
DENV3-specific primers, 25 pmol of both DENV2-specific
and DENV4-specific primers, and 9 pmol of each probe in
a mastermix of 20 μL per well. 5 μL of a serum sample was
added to give a total reaction mixture of 25 μL per well.
We did reverse transcription at 50°C for 30 min and then
at 95°C for 2 min, followed by 45 cycles of amplification
in a CFX96TM Real-Time PCR Detection System
(Biorad, Berkeley, CA, USA) with an annealing
temperature of 60°C. We did all quantifications on the
same day as extractions and in duplicate to ascertain
reproducibility. We established dengue serotype with the
same multiplex rtPCR protocol and confirmed it with
immunofluorescence assays. We identified primary or
secondary infections with immunoglobulin M antibody-
capture ELISA32
or dengue IgG indirect ELISA (Panbio
Diagnostics, Providence, RI, USA).
We did quantitative serum NS1 assays using a Platelia
Dengue NS1 Antigen kit (Biorad, Marnes-la-Coquette,
France), according to the manufacturer’s instructions.
We generated a linear standard curve by serially diluting
a purified recombinant baculovirus-derived DENV2
NS1 protein33
from 10 ng/mL to 0·15 ng/mL. For the
assay, we serially diluted serum samples up to 1/5000
and quantified the concentrations of NS1 antigen by the
optical density of the standard curve on the same plate.
With this method, the NS1 concentration in the samples
ranged from 0·002 μg/mL to 40 μg/mL.
The clinical investigators (JGL and LW) rated the severity
of all adverse events, their relatedness to the study drug,
and whether they were expected. Grading was based on
the National Cancer Institute Common Terminology
Criteria for Adverse Events (version 4.0), with
grade 1 recorded as mild, 2–3 as moderate, and 4–5 as
severe. Adverse events could be deemed to be definitely
related, probably related, possibly related, probably not
related, and not related to the investigational product. We
reported serious adverse events to the study monitor
(Singapore Clinical Research Institute), the data safety
monitoring board, the Centralised Institutional Review
Board, and the Health Sciences Authority, and recorded
their outcomes. We recorded adverse events in the relevant
source document and transcribed them into the electronic
case report form. The principal clinical investigator (JGL)
verified the adverse events, which were then monitored by
the Singapore Clinical Research Institute. All adverse
events and serious adverse events were coded and grouped
into preferred terms by system organ class, with the
Medical Dictionary for Regulatory Activities (version 12.1).
As an additional safety measure, an independent data
safety monitoring board (consisting of a clinician, a
clinical pharmacologist, and a biostatistician) did an
interim safety analysis after data for 12 patients (six in
each group) were available. No issues were raised. Data
were aggregated by dummy treatment group, and
members of the monitoring board were masked to
treatment allocation. Members were given information
about the patients’ prognosis and the nature of treatment.
We established a procedure for emergency unmasking of
individual participants, but it did not need to be
implemented during this trial. In addition to the interim
review, we notified members of the monitoring board of
serious adverse events that occurred during the trial.
Outcomes
The primary virological endpoint was mean virological
log reduction (VLR) from baseline. We applied log
transformation to the concentration of the virus in
participants’ serum, which was measured by rtPCR at
baseline and after each daily dose during hospital stay.
We computed VLR from baseline at days 2, 3, and 4, and
used the mean value for analyses. We imputed any
missing values for a specific day from the mean viraemia
of patients in the same treatment group on the same day.
The primary clinical endpoint was the area under the
fever curve (AUC) for a temperature of more than 37°C
from after the first dose on day 1 to the last measurement
Figure 1:Trial profile
*Due to intramuscular injections of diclofenac.
69 patients screened
24 assigned to and received
celgosivir
26 assigned to and received
placebo
50 randomly assigned
19 excluded
1 had had fever for >48 h
1 unwilling to stay in hospital as
inpatient
4 platelet count <80000 per μL
2 serum creatinine kinase
>130 µmol/L*
3 haemoglobin <110 g/L
8 creatinine kinase >600 U/L

4. Articles
www.thelancet.com/infection Vol 14 August 2014 709
before discharge from the Investigational Medical Unit.
We monitored the temperature of all patients every 2 h
from after the dose on day 1 to after the dose on
day 5 (96 h). We calculated AUC on the basis of the
trapezoidal rule for a 96 h period, and imputed missing
values with linear interpolation.
One secondary endpoint was time to clearance of
serum NS1 (measured until day 15). We defined time to
NS1 clearance as the first of two consecutive timepoints
when the concentration was less than 0·01 μg/mL, on
the basis of a previous report.34
Patients without evidence
of clearance were censored on their last day of assessment
(day 15). Other secondary endpoints included maximum
negative change from baseline in white blood cell count
from day 2 to day 5, maximum negative change from
baseline in platelet counts from day 2 to day 5, maximum
haemoconcentration (ie, haematocrit on each day
compared with haematocrit at day 15) from day 2 to day 5,
pharmacokinetics, and frequency of adverse events and
serious adverse events. Exploratory endpoints included
the total number of occasions that a patient took analgesic
drugs between day 1 and day 5, and the total use of
analgesic drugs between day 1 and day 5 (defined as the
sum of the total amount of drugs taken divided by the
WHO-defined daily dose). Data for two further secondary
endpoints (maximum change from baseline in serum
NS1 between day 2 and day 5, and time to viral clearance)
and two exploratory endpoints (immunomonitoring of a
panel of cytokines and intensity of joint and muscle pain)
will be analysed elsewhere.
Statistical analysis
We calculated sample size on the basis of viraemia data
from Libraty and colleagues’ report35
and the EDEN
prospective study in Singapore.36,37
We estimated that a
sample size of 25 patients in each treatment group would
have 80% power to detect a mean VLR of at least 0·7 for
the celgosivir group compared with the placebo group,
assuming a SD of 1·0 and with a 5% one-sided type I
error. Efficacy analyses were by intention to treat. We
designed all prespecified one-sided tests for the
hypothesis that celgosivir was superior to placebo. We
did post-hoc analyses with two-sided tests when data for
AUC suggested that celgosivir was inferior to placebo.
We compared the primary endpoints between the
treatment groups with one-sided two-sample t tests. We
calculated the difference between means and 90% CIs.
To avoid multiplicity in the primary objectives, efficacy
in the clinical endpoint (AUC) cannot be claimed if the
null hypothesis of virological endpoint (VLR) cannot be
rejected.38
We applied a finite mixture model to study the
distribution of mean VLR between day 2 and day 4 for
patients given placebo or celgosivir.39
In a sensitivity
analyses, we compared the medians of virological
endpoints with a randomisation test.40
For subgroup
analyses, we stratified the mean and median VLRs by
primary or secondary dengue infection, and dengue
serotypes.
For the secondary and exploratory endpoints, we applied
log-rank tests and independent-sample t tests when
appropriate to assess differences between treatment
groups, with two-sided p values and 95% CIs. All
Celgosivir group (n=24) Placebo group (n=26)
Men 18 (75%) 22 (85%)
Women 6 (25%) 4 (15%)
Ethnic origin
Chinese 13 (54%) 13 (50%)
Malay 0 2 (8%)
Indian 5 (21%) 4 (15%)
Others 6 (25%) 7 (27%)
Age (years) 35·1 (8·7) 34·7 (11·6)
Weight (kg) 64·15 (8·95) 67·00 (13·94)
Body-mass index (kg/m2) 23·63 (3·24) 24·10 (4·29)
Temperature (°C)
Mean (SD) 38·14 (1·13) 38·35 (0·74)
Median (IQR) 37·90 (37·30–39·10) 38·25 (37·80–38·90)
Systolic blood pressure (mm Hg) 128·5 (13·3) 127·2 (14·4)
Diastolic blood pressure (mm Hg) 76·3 (9·3) 75·1 (12·7)
Pulse (beats per minute) 92 (20) 92 (17)
Oxygen saturation (%) 98·5 (1·6) 98·3 (1·4)
History of specific procedures or
medical conditions*
12 (50%) 11 (42%)
Data are n (%)or mean (SD),unlessotherwise stated. *Appendicectomy, caesarean section, asthma, chronic back pain,
insomnia, chronic ear pain, hernia repair, glucose-6-phosphatedehydrogenasedeficiency,drug allergy, gallbladder surgery,
haemorrhoidectomy, hyperlipidaemia, hypertension, industrial accident, prostate surgery,or right-shoulderoperation.
Table 1: Demographic and baseline clinical characteristics
Celgosivir group (n=24) Placebo group (n=26)
Haemoglobin (g/L) 143·8 (11·2) 141·9 (13·5)
White blood cell count (×10⁹ per L) 5·18 (2·02) 4·79 (1·88)
Haematocrit (%) 42·54 (3·30) 41·83 (3·74)
Platelet count (×10⁹ per L) 180·1 (38·1) 176·1 (36·1)
Neutrophils (×10⁹ per L) 3·87 (1·71) 3·60 (1·68)
Lymphocytes (×10⁹ per L) 0·73 (0·50) 0·58 (0·25)
Sodium (mmol/L) 134·9 (1·4) 135·3 (2·1)
Potassium (mmol/L) 3·5 (0·3) 3·5 (0·3)
Aspartate aminotransferase (IU/L) 37·6 (20·6) 38·8 (20·5)
Alanine transaminase (IU/L) 38·2 (38·0) 34·9 (19·5)
Serum bilirubin (μmol/L) 11·7 (4·0) 13·2 (3·7)
Serum albumin (g/L) 41·6 (3·0) 40·7 (2·7)
Blood urea (mmol/L) 4·19 (0·87) 4·09 (0·97)
Serum creatinine (μmol/L) 86·5 (18·9) 87·0 (16·3)
Serum uric acid (μmol/L) 330·4 (61·8) 352·2 (72·1)
Serum creatine kinase (U/L) 134·0 (79·0) 157·3 (114·0)
Prothrombin time (s) 11·51 (1·09) 11·61 (0·70)
Partial thromboplastin time (s) 30·55 (3·05) 32·33 (2·10)
Data are mean (SD).
Table 2: Baseline laboratory examinations

5. Articles
710 www.thelancet.com/infection Vol 14 August 2014
participants who received the assigned drug were included
in safety analyses. We computed the duration of an
adverse event as the difference between the end date and
start date of the event. Additionally, we compared the
median VLR between the two groups with a randomisation
test in a sensitivity analysis, because the median is a
robust measure of central tendency when data are skewed.
All analyses were done in SAS (version 9.3) or Stata/SE
(version 10.0).
This study is registered with ClinicalTrials.gov, number
NCT01619969.
Role of the funding source
The funder of the study had no role in study design, data
collection, data analysis, data interpretation, or writing of
the report. The corresponding author had full access to
all the data in the study and had final responsibility for
the decision to submit for publication.
Results
69 patients were screened and 50 were randomly
assigned to receive placebo or celgosivir (figure 1). One
patient was inadvertently given two randomisation codes,
and the second code was used. One patient in the placebo
group was lost to follow-up on day 7 (patient 042; no
samples from day 4 onwards). Except for one participant
in the placebo group, all patients completed at least 80%
of their schedule of doses.
The demographic and clinical characteristics (table 1)
and baseline laboratory characteristics (table 2) were
similar in both groups. No patients had a positive IgG
result with the Dengue Duo assay at time of enrolment,
which would indicate a secondary infection. However,
subsequent laboratory analysis of stored baseline serum
samples showed that 18 (36%) were positive for dengue
IgG and therefore had a secondary infection (13 [54%]
of 24 in the celgosivir group, five [19%] of 26 in the
placebo group). DENV2 accounted for most infections
(32 patients, 64%), followed by DENV1 (14 patients, 28%)
and DENV3 (four patients, 8%).
Mean VLR was greater in patients given celgosivir
than in those given placebo, but the difference was
non-significant (table 3). Mean VLR for patients with
primary or secondary dengue also did not differ
significantly between groups (table 3). The mean AUC
above 37°C for patients in the celgosivir group was
higher than for those in the placebo group, but the
difference was non-significant in a one-sided test
(figure 2, table 3). Because fever is usually higher
during secondary infection than primary infection, we
analysed whether this factor accounted for the increased
fever in the celgosivir group. The difference between
groups in mean AUC was greater for primary infection
than secondary infection, although both were non-
significant (table 3). We recorded no significant
interaction between group and primary or secondary
dengue infection (p=0·592).
Time to clearance of serum NS1 differed substantially
between groups until day 7 (log-rank p=0·026; figure 3).
Time to clearance of serum NS1 until day 7 stratified by
primary versus secondary infection was not significantly
different between the two groups (figure 3).
The changes in haematological markers did not differ
significantly between groups (table 4). Likewise, the
changes in biochemistry and liver transaminases, and
the coagulation profile were similar in both groups (data
not shown). The number of occasions that analgesics
were requested and total use of analgesics were higher in
the celgosivir group than in the placebo group, but the
differences were non-significant (table 4).
All patients received their allocated drug and so were
included in the safety analyses. 481 adverse events were
reported (table 5). Overall, the number of adverse events
per patient was slightly lower in the celgosivir group than
the placebo group (mean 9·2 events vs 10·0 events;
appendix). Most adverse events were consistent with
expected clinical and laboratory changes in dengue
fever,36,37,41
and occurred with similar frequency between
the groups (appendix). Diarrhoea—a common important
early clinical feature in dengue fever37
—was more
frequent in the celgosivir group (11 mild and three
moderate events; mean duration 1·72 days) than in the
placebo group (four mild events; mean duration 1·75 days;
appendix). None of the three patients with moderate
Celgosivir group (n=24) Placebo group (n=26) Difference (90% CI) p value
VLR* of all patients –1·86 (1·07) –1·64 (0·75) –0·22 (–0·65 to 0·22) 0·203
VLR* of patients with primary infection –1·37 (0·78)† –1·51 (0·66)‡ 0·13 (–0·31 to 0·58) 0·690
VLR* of patients with secondary infection –2·27 (1·13)§ –2·22 (0·88)¶ –0·05 (–1·04 to 0·93) 0·463
AUC above 37°C (0–96 h) for all patients 54·92 (31·04) 40·72 (18·69) 14·20 (2·16 to 26·25) 0·973 (0·054||)
AUC above 37°C (0–96 h) for patients with
primary infection
54·41 (34·75)† 38·82 (18·62)‡ 15·60 (–0·30 to 31·50) 0·947 (0·106||)
AUC (0–96 h) for patients with secondary
infection
55·34 (28·98)§ 48·69 (18·74)¶ 6·65 (–17·97 to 31·27) 0·678 (0·644||)
Data are mean (SD) unless otherwise stated.VLR=virological log reduction. AUC=area under the curve for fever burden. *Mean value ofVLR from day 1 at days 2, 3, and 4.
†n=11. ‡n=21. §n=13. ¶n=5. ||Two-sided p value from two-sample t test.
Table 3: MeanVLR and AUC

6. Articles
www.thelancet.com/infection Vol 14 August 2014 711
diarrhoea had any clinical or biochemical evidence of
dehydration that required any intravenous fluid
replacement. However, two patients given celgosivir who
had mild diarrhoea received intravenous normal saline
infusion for less than 24 h for indications other than
diarrhoea: one had poor appetite with nausea and
vomiting consistent with acute dengue, and the second
had mild per vaginal bleeding. The mild-to-moderate
diarrhoea in the celgosivir group was expected, because
this side-effect was reported in the trials of celgosivir for
hepatitis C infection.20
The frequency of flatulence, a
class effect of α-glucosidase inhibitors, was lower in the
celgosivir group than in the placebo group (appendix).
Other adverse events that are also expected clinical events
in dengue fever were less frequent in the celgosivir group
than in the placebo group—eg, one patient (4%) in the
celgosivir group had oropharyngeal pain compared with
seven (27%) in the placebo group (appendix).
We recorded three serious adverse events. Two were
deemed to be possibly related to investigational product:
one patient in the celgosivir group had persistent low
back pain, for which their hospital stay was extended by
3 days; and one patient in the placebo group had
hypotension secondary to poor oral intake and
dehydration, for which their hospital stay was extended
by 4 days. The serious adverse event that was probably
not related to investigational product was in the
celgosivir group: the patient had acute haematemesis
secondary to an acute gastric ulceration, for which their
hospital stay was extended by 2 days. All patients were
followed up after adverse and serious adverse events
until full recovery.
The distribution of the mean VLR, particularly in the
placebo group, is possibly bimodal (appendix) despite
logarithmic transformation. Analysis of the data
distribution with a finite mixture model indicated
marginally significant support for this bimodal
distribution (likelihood ratio test, p=0·056). We did an
exploratory analysis of the median VLR from day 2 to
day 4. Median VLR for patients in the celgosivir group was
significantly higher than for those in the placebo group
(–1·89 [IQR –2·47 to –1·19] vs –1·56 [–1·87 to –1·23];
difference –0·33; one-sided p=0·040). The difference in
median VLR was larger in patients with secondary (–0·69)
than in those with primary dengue (–0·14), although
these differences were non-significant (appendix). The
Figure 2: Mean body temperature
Temperature was measured every 2 h from 0 h to 96 h.
0 6 12 18 24 30 36 42 48 54 60 66 72 78 84 90 96
36·5
37·0
37·5
38·0
38·5
Time from trial start (h)
Temperature(˚C)
Celgosivir
Placebo
Figure 3: Kaplan-Meier plots of viral NS1 antigen clearance
*Two patients in the celgosivir treatment group had <10 ng/mL NS1 antigen
at day 0.
Patientswithnoevidenceof
serumNS1clearance(%)
Celgosivir
Placebo
0
20
40
60
80
100
All patients
Log-rank p=0·129
Patientswithnoevidenceof
serumNS1clearance(%)
0
20
40
60
80
100
Patients with primary infection
Log-rank p=0·632
Time from trial start (days)
Patientswithnoevidenceof
serumNS1clearance(%)
0 5 10 15
0
20
40
60
80
100
Patients with secondary infection
Log-rank p=0·201
Number at risk
Celgosivir
Placebo
22*
26
17
23
11
20
6
9
Number at risk
Celgosivir
Placebo
11
21
11
20
10
18
6
9
Number at risk
Celgosivir
Placebo
11
5
6
3
1
2
0
0

7. Articles
712 www.thelancet.com/infection Vol 14 August 2014
difference in median VLR for each of the three virus
serotypes with which participants were infected were also
not significant (data not shown).
Discussion
Celgosivir does not significantly reduce viral load or
fever burden in patients with dengue (panel). The drug
is generally safe and well tolerated. The adverse events
reported were typical of dengue infection37,41
or known
side-effects of celgosivir.20
Despite the increased
number of patients with mild-to-moderate diarrhoea in
the celgosivir group, no severe cases of diarrhoea were
reported.
Previous antiviral clinical trials in dengue16,17
have
shown large variation in viraemia and viraemia clearance,
which are both dependent on the immune status of the
patients and virus serotype. Notably, the viraemia
reduction in secondary dengue has been reported to be
significantly greater than that for primary dengue.16,17
To
identify patients with dengue, we used the point-of-care
SD Bioline Dengue Duo kit to simultaneously detect viral
NS1, IgM, and IgG. Previous studies34,42,43
indicated that
the sensitivity of such a bedside NS1 test is significantly
greater for patients with primary than secondary dengue,
which is thought to be caused by interference from pre-
existing cross-reactive antibodies in secondary infection.
Celgosivir group (n=24) Placebo group (n=26) Difference (95% CI) p value*
White blood cell count (×10⁹ per L)
Mean –2·71 (1·74) –2·42 (1·51) –0·28 (–1·21 to 0·64) 0·540
Median –2·42 (–3·42 to –1·53) –2·24 (–3·30 to –1·27) ·· ··
Platelet count (×10⁹ per L)
Mean –85·50 (44·99) –70·46 (37·10) –15·04 (–39·41 to 8·33) 0·202
Median –85·00 (–106·50 to –54·50) –64·50 (–101·00 to –48·00) ·· ··
Haemoconcentration
Mean 4·63 (2·55) 4·67 (2·45) –0·04 (–1·48 to 1·40) 0·957
Median 5·05 (2·70 to 5·90) 4·30 (3·00 to 6·00) ·· ··
Number of occasions analgesic drugs prescribed
Mean 9·00 (3·75) 7·73 (3·28) 1·27 (–0·73 to 3·27) 0·208
Median 9 (7 to 12) 8 (5 to 10) ·· ··
Total use of analgesic drugs†
Mean 3·19 (1·41) 2·61 (1·12) 0·59 (–0·13 to 1·30) 0·108
Median 3·08 (2·33 to 4·49) 2·67 (1·67 to 3·67) ·· ··
Data in parentheses are SD or IQR. *Two-sided p value from two-sample t test. †Sum of the total amount of drugs taken divided by theWHO-defined daily dose.
Table 4: Maximum changes from baseline in white blood cell and platelet counts, and haemoconcentration and analgesic drug use from day 2 to day 5
Number of patients Number of events
Celgosivir group
(n=24)
Placebo group
(n=26)
Overall (n=50) Celgosivir group Placebo group Overall
Any adverse event 24 (100%) 26 (100%) 50 (100%) 221 260 481
Adverse drug reaction 20 (83%) 16 (62%) 36 (72%) 39 32 71
Severity
Mild 24 (100%) 26 (100%) 50 (100%) 162 202 365
Moderate 20 (83%) 21 (81%) 41 (82%) 53 54 107
Severe 5 (21%) 4 (15%) 9 (18%) 5 4 9
Related to drug
Possibly related 20 (83%) 16 (62%) 36 (72%) 39 32 71
Probably not related 24 (100%) 26 (100%) 50 (100%) 178 221 399
Not related 4 (17%) 7 (27%) 11 (22%) 4 7 11
Expectedness
Expected 24 (100%) 26 (100%) 50 (100%) 212 252 464
Unexpected 7 (29%) 6 (23%) 13 (26%) 9 8 17
Recovered or resolved 24 (100%) 26 (100%) 50 (100%) 221 260 481
Data in parentheses are %.
Table 5: Summary of adverse events

8. Articles
www.thelancet.com/infection Vol 14 August 2014 713
Therefore, we expected to recruit mainly patients with
primary dengue. However, 36% of patients in our trial
had secondary dengue. We noted wide variation in
viraemia and VLR, as in previous reports.16,17
Our analysis showed that a small group of patients
given placebo had, on average, 1 log greater reduction in
VLR between days 2 and 4 of illness compared with
others in that group, suggesting that the VLR data could
have been skewed (appendix). This finding is potentially
exciting because it suggests that there are outliers in
viraemia clearance. An understanding of how a few
individuals can clear dengue virus from their blood more
rapidly than others can could shed new light on dengue
pathogenesis. However, in our clinical trial, these outliers
led to the bimodal distribution of VLR, which made the
comparison of mean VLR as an endpoint (as defined a
priori in our statistical analysis plan) problematic. To
overcome this possible non-normal distribution in our
data, we compared median VLR and recorded a
significant difference in virus clearance. From another
perspective, two-thirds of patients given celgosivir had
a VLR of at least 1·56 between study days 2 and 4,
compared with half of patients given placebo.
VLR in our study was lower than we had anticipated
on the basis of preclinical studies,23,25,44
the reason for
which is unknown. The preclinical studies23,25,44
showed
that early reduction in viraemia was greater in
AG129 mice infected with enhancing antibodies than in
animals infected with DENV2 alone. However, AG129
mice infected with DENV2 with or without enhancing
antibodies given celgosivir had similar reductions in
mortality. Because of this finding, we had not anticipated
a therapeutic difference between patients with primary
or secondary infections and had therefore not powered
our study to allow analyses stratified by presence or
absence of dengue antibodies at enrolment. Additionally,
the point-of-care kit that we used to identify patients
with dengue could not differentiate between primary
and secondary infections. To account for such a
difference, future studies will need to be designed to
address the question of greater therapeutic efficacy with
celgosivir in patients with secondary dengue than in
those with primary dengue.
Our NS1 clearance data also provide some signal of
pharmacological activity in patients with secondary
dengue. Clearance of NS1 antigens was faster in the
celgosivir group than in the placebo group. This
difference was prominent in patients with secondary,
although the result was non-significant, which could
again be a result of the small population.
Although celgosivir inhibits the glycosylation processes
of prM, E, and NS1 proteins of dengue virus, it could also
affect other pathways, such as those involving endoplasmic
reticulum stress proteins.22–24,43
The indirect effects of
celgosivir on endoplasmic reticulum stress could modulate
viral replication, which might have contributed to the
lower-than-expected reduction in viraemia. A preclinical
study23
showed greater differences in the expression of
endoplasmic reticulum stress proteins in DENV2-infected
THP-1 cells treated with celgosivir than in untreated cells.
Transcription of pro-apoptotic endoplasmic reticulum
stress protein DDIT3 is significantly lower and that of the
prosurvival protein EDEM1 is significantly greater when
celgosivir is used for antibody-enhanced dengue infection
than when it is used for infection with only DENV2.23
These differences could be a result of an underlying
difference in pathophysiology of primary and secondary
dengue infection established in a previous study45
and
apparent in our study. Alternatively, celgosivir could have
other effects on the host’s inflammatory or innate immune
responses that compromise replication of dengue virus.
The detailed immunomonitoring of 41 cytokines and
chemokines that is underway for our samples could
provide some explanations. Whatever the case, studies in
which these possible mechanistic explanations are
investigated are urgently needed and should lead to
improved understanding of dengue pathogenesis.
Our study draws attention to important shortcomings in
point-of-care diagnostic tests. A decreased sensitivity of
NS1 antigen tests in patients with pre-existing dengue
antibodies has been reported.42
We were surprised that the
diagnostic test did not detect any secondary dengue cases
(by detection of dengue IgG antibodies) at enrolment.
Secondary dengue is a well recognised risk factor of severe
dengue, and presence of pre-existing dengue antibodies
could be useful for triage of patients with dengue.3,12
Without a point-of-care test to discriminate between
primary and secondary dengue, a clinical trial in which
patients are randomly assigned after stratification by
primary or secondary dengue would be logistically
difficult, because all laboratory-based serological tests
have a long turnaround time. Improved point-of-care
diagnostic tests are urgently needed for routine use as well
as future clinical trials.
Panel: Research in context
Systematic review
We searched PubMed for reports published in any language
before Feb 5, 2014, with the terms “dengue virus”, “‘dengue
virus’ AND ‘clinical trial’”, “‘dengue virus’ AND ‘celgosivir’”,
“‘dengue virus’ AND ‘castanospermine’”, and “‘dengue virus’
AND ‘alpha-glucosidase inhibitor’”.We identified no previous
clinical trials of celgosivir in dengue, its active metabolite
castanospermine, or any other α-glucosidase inhibitors.
Interpretation
We have shown that twice-daily doses of celgosivir for 5 days
(2 g overall) is generally safe and well tolerated in patients
with dengue. Our data do not show any protective efficacy
against dengue virus infection as measured by viral load
reduction or reduction in clinical symptoms such as fever.
Safe and efficacious antivirals that can block replication of all
four dengue virus serotypes are urgently needed.

9. Articles
714 www.thelancet.com/infection Vol 14 August 2014
The results of our study do not exclude celgosivir or
other host α-glucosidase inhibitors as future dengue
therapeutics. Our study was underpowered to establish
whether the modest change in VLR and the trend towards
faster NS1 clearance in the patients with secondary
dengue were significant. Because other clinical trials16,17
have not shown a significant change in VLR, which
clinical endpoints correlate with an antiviral effect is
unknown. Additional clinical studies are necessary to
resolve these issues.
Overall, the drug regimen in our trial was well tolerated.
Because dengue virus is cleared from the blood
within 3–5 days and fever subsides in the same period, a
regimen with an increased dose of celgosivir or more
frequent dosing with a shorter treatment course could be
tested in future trials to see if the pharmacological effect
could be increased.
Contributors
JGL, CS, EEO, and SGV conceived and designed the study. JGL, CS, LW,
BHT, LT, LTC, DCL, PAT, LCN, YSL, and SGV set up recruitment sites.
JGL, LW, JSC, and BPC enrolled patients. APSR, SW, BHT, YH, AC, SH,
WKC, KHT, JC, and MLH did the experiments. JGL, CS, YW, APSR, SW,
BHT, YBC, EEO, and SGV analysed data. LCN, JC, MLH, EEO, and SGV
contributed reagents, materials, or analysis methods. JGL, CS, YW, BHT,
LT, YBC, EEO, and SGV wrote the first version of the report. JGL, CS,
YW, YBC, EEO, and SGV reviewed and revised the report. All authors
approved the final version of the report.
Declaration of interests
JGL, CS, APSR, SW, and SGV have a patent pending (Dosing Regimen
for Use of Celgosivir as an Antiviral Therapeutic For Dengue Virus
Infection; filing date July 26, 2013). JGL, CS, APSR, SW, EEO, and SGV
have a patent pending (Novel Dosing Regimens of Celgosivir For The
Treatment Of Dengue; filing date Dec 4, 2013). CS reports personal fees
from 60 Degrees Pharmaceuticals for examining pharmacokinetic
modelling and dosing regimen for dengue antiviral drugs. PAT received
grants (separate to the funding for this trial) from the National Medical
Research Council STOP Dengue Translational Clinical Research during
this trial; has received grants from GlaxoSmithKline, Sanofi-Pasteur,
Fabentech, Adamas, and Inviragen; has been a paid consultant for
AstraZeneca, Johnson and Johnson, and Teleflex; has received speaker
fees from Novartis; and has received travel support from Biomerieux and
3M. MLH has received grants, honoraria, and travel support from
Roche, Janssen, and Novartis. EEO is a paid member of the Scientific
Advisory Board on Dengue Vaccine (Sanofi Pasteur).
Acknowledgments
The preclinical studies to initiate the project were partly funded by a
DUKE-NUS Signature Research Program start-up grant to SGV. The
clinical trial was funded by STOP Dengue Translational Clinical Research
grant through the National Medical Research Council of Singapore. YBC
and EEO were supported by the Singapore Ministry of Health’s National
Medical Research Council under its Clinician Scientist Award. We thank
the study participants; the family physicians who actively referred
patients to our centre; Wan-Teck Lim, Aziah Ahmad, and the staff of the
SingHealth Investigational Medicine Unit for their dedication and
cooperation throughout the study; Sam Lim for his contribution to
discussions about study design; the clinical research coordinators
Sheena Ng, Huizhen Sam, and Nidhi Chlebicka for their administrative
support; Helen Isaac and Janice Ng for clinical monitoring of the trial
and the rest of the Singapore Clinical Research Institute CELADEN team
for database management; Lee How Sung at the National University of
Singapore for measurement of drug levels; the members of the data
safety monitoring board (Blaisé Genton, Edmund J D Lee, and
Denis H Y Leung); and Ranga Krishna, Duane Gubler, Marcel Tanner,
Annelies Wilder-Smith, Cameron Simmons, and Alex Matter for their
support and encouragement throughout this project.
References
1 Bhatt S, Gething PW, Brady OJ, et al. The global distribution and
burden of dengue. Nature 2013; 496: 504–07.
2 Racloz V, Ramsey R, Tong S, Hu W. Surveillance of dengue fever
virus: a review of epidemiological models and early warning
systems. PLoS Negl Trop Dis 2012; 6: e1648.
3 Simmons CP, Farrar JJ, Nguyen van Vinh C, Wills B. Dengue.
N Engl J Med 2012; 366: 1423–32.
4 Gubler DJ. Dengue/dengue haemorrhagic fever: history and current
status. Novartis Found Symp 2006; 277: 3–16.
5 Halstead SB. Antibodies determine virulence in dengue.
Ann N Y Acad Sci 2009; 1171 (suppl 1): E48–56.
6 Wilder-Smith A, Ooi EE, Vasudevan SG, Gubler DJ. Update on
dengue: epidemiology, virus evolution, antiviral drugs, and vaccine
development. Curr Infect Dis Rep 2010; 12: 157–64.
7 Guzman MG, Halstead SB, Artsob H, et al. Dengue: a continuing
global threat. Nat Rev Microbiol 2010; 8 (suppl 12): S7–16.
8 Suaya JA, Shepard DS, Siqueira JB, et al. Cost of dengue cases in
eight countries in the Americas and Asia: a prospective study.
Am J Trop Med Hyg 2009; 80: 846–55.
9 Keller TH, Chen YL, Knox JE, et al. Finding new medicines for
flaviviral targets. Novartis Found Symp 2006; 277: 102–14.
10 Noble CG, Chen YL, Dong H, et al. Strategies for development of
dengue virus inhibitors. Antiviral Res 2010; 85: 450–62.
11 Botting C, Kuhn RJ. Novel approaches to flavivirus drug discovery.
Expert Opin Drug Discov 2011; 7: 417–28.
12 Lescar J, Luo D, Xu T, et al. Towards the design of antiviral
inhibitors against flaviviruses: the case for the multifunctional
NS3 protein from dengue virus as a target. Antiviral Res 2008;
80: 94–101.
13 Sampath A, Padmanabhan R. Molecular targets for flavivirus drug
discovery. Antiviral Res 2009; 81: 6–15.
14 Sessions OM, Barrows NJ, Souza-Neto JA, et al. Discovery of
insect and human dengue virus host factors. Nature 2009;
458: 1047–50.
15 Fink J, Gu F, Ling L, et al. Host gene expression profiling of dengue
virus infection in cell lines and patients. PLoS Negl Trop Dis 2007;
1: e86.
16 Tricou V, Minh NN, Van TP, et al. A randomized controlled trial of
chloroquine for the treatment of dengue in Vietnamese adults.
PLoS Negl Trop Dis 2010; 4: e785.
17 Nguyen NM, Tran CN, Phung LK, et al. A randomized,
double-blind placebo controlled trial of balapiravir, a polymerase
inhibitor, in adult dengue patients. J Infect Dis 2013; 207: 1442–50.
18 Sabchareon A, Wallace D, Sirivichayakul C, et al. Protective efficacy
of the recombinant, live-attenuated, CYD tetravalent dengue vaccine
in Thai schoolchildren: a randomised, controlled phase 2b trial.
Lancet 2012; 380: 1559–67.
19 Halstead SB. Identifying protective dengue vaccines: guide to
mastering an empirical process. Vaccine 2013; 31: 4501–07.
20 Durantel D. Celgosivir, an alpha-glucosidase I inhibitor for the
potential treatment of HCV infection.
Curr Opin Investig Drugs 2009; 10: 860–70.
21 Mehta A, Zitzmann N, Rudd PM, Block TM, Dwek RA.
Alpha-glucosidase inhibitors as potential broad based anti-viral
agents. FEBS Lett 1998; 430: 17–22.
22 Courageot MP, Frenkiel MP, Dos Santos CD, Deubel V,
Despres P. Alpha-glucosidase inhibitors reduce dengue
virus production by affecting the initial steps of virion
morphogenesis in the endoplasmic reticulum. J Virol 2000;
74: 564–72
23 Rathore AP, Paradkar PN, Watanabe S, et al. Celgosivir treatment
misfolds dengue virus NS1 protein, induces cellular pro-survival
genes and protects against lethal challenge mouse model.
Antiviral Res 2011; 92: 453–60.
24 Schul W, Liu W, Xu HY, Flamand M, Vasudevan SG. A dengue fever
viremia model in mice shows reduction in viral replication and
suppression of the inflammatory response after treatment with
antiviral drugs. J Infect Dis 2007; 195: 665–74.
25 Watanabe S, Rathore AP, Sung C, et al. Dose- and schedule-
dependent protective efficacy of celgosivir in a lethal mouse model
for dengue virus infection informs dosing regimen for a proof of
concept clinical trial. Antiviral Res 2012; 96: 32–35.

10. Articles
www.thelancet.com/infection Vol 14 August 2014 715
26 Kaita K, Yoshida E, Kunimoto D, et al. Phase II proof of concept study
of celgosivir in combination with peginterferon alfa-2b and ribavirin
in chronic hepatitis c genotype-1 non responder patients. Digestive
Diseases Week; Washington, DC, USA; May 19–24, 2007. 442.
27 WHO. Dengue: guidelines for diagnosis, treatment, prevention and
control. Geneva: World Health Organization, 2009.
28 Lai YL, Chung YK, Tan HC, et al. Cost-effective real-time reverse
transcriptase PCR (RT-PCR) to screen for dengue virus followed by
rapid single-tube multiplex RT-PCR for serotyping of the virus.
J Clin Microbiol 2007; 45: 935–41.
29 Sorbera LA, Castaner J, Garcia-Capdevila L. Celgosivir. Drugs Future
2005; 30: 545–52.
30 Wang WK, Sung TL, Tsai YC, Kao CL, Chang SM, King CC. Detection
of dengue virus replication in peripheral blood mononuclear cells
from dengue virus type 2-infected patients by a reverse
transcription-real-time PCR assay. J Clin Microbiol 2002; 40: 4472–78.
31 Johnson BW, Russell BJ, Lanciotti RS. Serotype-specific detection of
dengue viruses in a fourplex real-time reverse transcriptase PCR
assay. J Clin Microbiol 2005; 43: 4977–83.
32 Rivetz B, Siman-Tov D, Ambal E, et al. New dengue antibody assay
with unique differential detection of IgG and IgM antibodies.
Clin Biochem 2009; 42: 180–84.
33 Rozen-Gagnon K, Moreland NJ, Ruedl C, Vasudevan SG.
Expression and immunoaffinity purification of recombinant dengue
virus 2 NS1 protein as a cleavable SUMOstar fusion.
Protein Expr Purif 2012; 82: 20–25.
34 Alcon S, Talarmin A, Debruyne M, Falconar A, Deubel V, Flamand M.
Enzyme-linked immunosorbent assay specific to dengue virus
type 1 nonstructural protein NS1 reveals circulation of the antigen in
the blood during the acute phase of disease in patients experiencing
primary or secondary infections. J Clin Microbiol 2002; 40: 376–81.
35 Libraty DH, Young PR, Pickering D, et al. High circulating levels of
the dengue virus nonstructural protein NS1 early in dengue illness
correlate with the development of dengue hemorrhagic fever.
J Infect Dis 2002; 186: 1165–68.
36 Low JG, Ooi EE, Tolfvenstam T, et al. Early dengue infection and
outcome study (EDEN)—study design and preliminary findings.
Ann Acad Med Singapore 2006; 35: 783–89.
37 Low JG, Ong A, Tan LK, et al. The early clinical features of dengue
in adults: challenges for early clinical diagnosis.
PLoS Negl Trop Dis 2011; 5: e1191.
38 The European Agency for the Evaluation of Medicinal Products.
Points to consider on multiplicity issues in clinical trials.
Sept 19, 2002. www.ema.europa.eu/ema/pages/includes/document/
open_document.jsp?webContentId=WC500003655 (accessed
April 28, 2014).
39 Basford KE, McLachlan GJ. Likelihood estimation with normal
mixture models. Appl Stat 1985: 34: 282–89.
40 Kaiser J, Michael GL. A general-purpose method for two-group
randomization tests. Stata J 2009; 9: 70-85.
41 Tanner L, Schreiber M, Low JG, et al. Decision tree algorithms
predict the diagnosis and outcome of dengue fever in the early
phase of illness. PLoS Negl Trop Dis 2008; 2: e196.
42 Chaterji S, Allen JC Jr, Chow A, Leo YS, Ooi EE. Evaluation of the
NS1 rapid test and the WHO dengue classification schemes for use
as bedside diagnosis of acute dengue fever in adults.
Am J Trop Med Hyg 2011; 84: 224–28.
43 Andries AC, Duong V, Ngan C, et al. Field evaluation and impact on
clinical management of a rapid diagnostic kit that detects dengue
NS1, IgM and IgG. PLoS Negl Trop Dis 2012; 6: e1993.
44 Watanabe S, Tan KH, Rathore AP, et al. The magnitude of dengue
virus NS1 protein secretion is strain dependent and does not
correlate with severe pathologies in the mouse infection model.
J Virol 2012; 86: 5508–14.
45 Kliks SC, Nimmanitya S, Nisalak A, Burke DS. Evidence that
maternal dengue antibodies are important in the development of
dengue hemorrhagic fever in infants. Am J Trop Med Hyg 1988;
38: 411–19.