Articles

Efficacy of a tetravalent dengue vaccine in healthy children

aged 4–16 years: a randomised, placebo-controlled,
phase 3 trial
Shibadas Biswal*, Charissa Borja-Tabora*, Luis Martinez Vargas, Hector Velásquez, Maria Theresa Alera, Victor Sierra,
Edith Johana Rodriguez-Arenales, Delia Yu, V Pujitha Wickramasinghe, Edson Duarte Moreira Jr, Asvini D Fernando, Dulanie Gunasekera,
Pope Kosalaraksa, Felix Espinoza, Eduardo López-Medina, Lulu Bravo, Suely Tuboi, Yanee Hutagalung, Pedro Garbes, Ian Escudero,
Martina Rauscher, Svetlana Bizjajeva, Inge LeFevre, Astrid Borkowski, Xavier Saez-Llorens*, Derek Wallace*, for the TIDES study group†

Summary
Background A substantial unmet need remains for safe and effective vaccines against dengue virus disease, particularly Lancet 2020; 395: 1423–33
for individuals who are dengue-naive and those younger than 9 years. We aimed to assess the efficacy, safety, and Published Online
immunogenicity of a live attenuated tetravalent dengue vaccine (TAK-003) in healthy children aged 4–16 years. March 17, 2020
https://doi.org/10.1016/
S0140-6736(20)30414-1
Methods We present data up to 18 months post-vaccination from an ongoing phase 3, randomised, double-blind trial
See Comment page 1402
of TAK-003 in endemic regions of Asia and Latin America (26 medical and research centres across Brazil, Colombia,
See Articles page 1434
Dominican Republic, Nicaragua, Panama, Philippines, Sri Lanka, and Thailand). Healthy children aged 4–16 years
*Authors share first and last
were randomly assigned 2:1 (stratified by age and region) to receive two doses of TAK-003 or two doses of placebo, place authorship equally
3 months apart. Investigators, participants and their parents or guardians, and sponsor representatives advising on
†Collaborators are listed in the
trial conduct were masked to trial group assignments. Participants presenting with febrile illness were tested for appendix
virologically confirmed dengue (VCD) by serotype-specific RT-PCR. In timeframes beginning 30 days post-second Takeda Vaccines, Boston, MA,
dose, the primary endpoint (overall vaccine efficacy) was assessed in the first 11 months, and the secondary endpoints USA (S Biswal MD, P Garbes MD,
(efficacy by baseline serostatus, serotype, hospitalised dengue, and severe dengue) in the first 17 months. This study D Wallace MBBS); Research
Institute For Tropical Medicine,
is registered with ClinicalTrials.gov, NCT02747927.
Muntinlupa, Philippines
(C Borja-Tabora MD); Centro de
Findings 20 099 participants were randomly assigned and vaccinated between Sept 7, 2016, and Aug 18, 2017; Atención e Investigación
19 021 (94·6%) were included in the per protocol analysis, and 20 071 (99·9%) in the safety set. The primary endpoint Médica, Dominicana,
Santo Domingo, Dominican
was achieved with an overall vaccine efficacy of 80·2% (95% CI 73·3 to 85·3; 61 cases of VCD in the TAK-003 group
Republic (L Martinez Vargas MD);
vs 149 cases of VCD in the placebo group). In the secondary endpoint assessment timeframe, an overall vaccine Centro de Atención e
efficacy of 73·3% (95% CI 66·5 to 78·8) was observed. Analysis of secondary endpoints showed efficacies of 76·1% Investigación Médica, Acacias,
(95% CI 68·5 to 81·9) in individuals who were seropositive at baseline, 66·2% (49·1 to 77·5) in individuals who were Colombia (H Velásquez MD);
Philippines-Armed Forces
seronegative at baseline, 90·4% (82·6 to 94·7) against hospitalised dengue, and 85·9% (31·9 to 97·1) against dengue
Research Institute of Medical
haemorrhagic fever. Efficacy varied by individual serotypes (DENV 1, 69·8% [95% CI 54·8 to 79·9]; DENV 2, 95·1% Sciences Virology Research
[89·9 to 97·6]; DENV 3, 48·9% [27·2 to 64·1]; DENV 4, 51·0% [–69·4 to 85·8]). Cumulative rates of serious adverse Unit, Cebu City, Philippines
events were similar in TAK-003 (4·0%) and placebo (4·8%) recipients, and were consistent with expected medical (M Theresa Alera MD); Centro de
Atención e Investigación
disorders in the study population. Infection was the most frequent reason leading to serious adverse events.
Médica, Yopal, Colombia
20 participants (<0·1% of the safety set) were withdrawn from the trial due to 21 adverse events by the end of part two; (V Sierra MD); Centro de
14 of these participants received TAK-003 and six received placebo. Atención e Investigación
Médica, Aguazul, Colombia
(E Johana Rodriguez-
Interpretation TAK-003 was well tolerated and efficacious against symptomatic dengue in children regardless of Arenales MD); De La Salle
serostatus before immunisation. Vaccine efficacy varied by serotype, warranting continued follow-up to assess longer- Medical and Health Sciences
term vaccine performance. Institute, Dasmariñas,
Philippines (D Yu MD);
University of Colombo,
Funding Takeda Vaccines. Colombo, Sri Lanka
(V P Wickramasinghe MD);
Copyright © 2020 Elsevier Ltd. All rights reserved. Associação Obras Sociais
Irmã Dulce Hospital
Santo Antônio and
Introduction American countries—is characterised by periodic Oswaldo Cruz Foundation,
Almost half of the global population live in dengue- outbreaks, which have a substantial effect on human Bahia, Brazil
endemic areas, with the highest burden of disease health and on global and national economies.5 Dengue is (Prof E Duarte Moreira Jr MD);
observed in the Americas and Asia.1–4 According to one caused by four virus serotypes (DENV 1–4), which are Faculty of Medicine, University
of Kelaniya, Kelaniya, Sri Lanka
estimate, 390 million cases of dengue infection occur transmitted by mosquito vectors.6 A tetravalent dengue (Prof A D Fernando MD); Faculty
annually and 96 million of those manifest clinically.1,2 vaccine (CYD-TDV; Dengvaxia, Sanofi Pasteur, Lyon, of Medical Sciences, University
Dengue—a leading cause of hospitalisation and death France) is approved in several endemic countries for of Sri Jayawardenenpura,
among children and adults in most Asian and Latin individuals aged 9 years and older who have evidence of Gangodawila, Sri Lanka

www.thelancet.com Vol 395 May 2, 2020 1423

 Articles

(Prof D Gunasekera MD); Faculty

of Medicine, Khon Kaen Research in context
University, Khon Kaen,
Thailand (P Kosalaraksa MD); Evidence before this study data for this study, with TAK-003 showing an overall vaccine
National Autonomous Between database inception and Feb 10, 2020, we searched efficacy of 80%. Rapid onset of protection was observed after a
University of Nicaragua, León, PubMed for publications in English with terms including first dose along with encouraging data from exploratory
Nicaragua (F Espinoza MD);
Centro de Estudios en
“dengue vaccine”, “dengue vaccine clinical trial”, and “dengue analysis of secondary efficacy endpoints.
Infectología Pediátrica, vaccine phase 1, 2, and 3”. We also searched ClinicalTrials.gov,
Added value of this study
Universidad del Valle and and national and international public health agencies.
Centro Médico Imbanaco, Cali, Our data show that TAK-003 has an acceptable safety profile in
All reports of phase 1–3 dengue vaccine clinical trials were
Colombia healthy children aged 4–16 years, and is efficacious in the
included as being relevant or of interest. At the time this
(E López -Medina MD); prevention of symptomatic dengue disease in both individuals
University of the Philippines phase 3 study was first conceived and designed, no dengue
who are dengue-naive and those previously exposed. Efficacy
Manila, Ermita, Philippines vaccines were licensed for use in any country. Since then,
(Prof L Bravo MD); Takeda varied against individual serotypes, with an overall efficacy of
CYD-TDV (Dengvaxia, Sanofi Pasteur, Lyon, France) has been
Pharmaceuticals, São Paulo, 66% in individuals who were dengue-naive and 76% in those who
registered in 18 dengue-endemic countries for individuals aged
Brazil (S Tuboi MD); Takeda were pre-exposed. In addition, TAK-003 reduced the number of
Vaccines, Singapore 9 years and older; this vaccine is not recommended for use in
dengue cases that were hospitalised by 90% along with an
(Y Hutagalung MD, individuals who are dengue-naive and hence requires evidence
I Escudero MD); Takeda 86% reduction in dengue haemorrhagic fever. These data
of previous dengue exposure for its use. Two additional vaccine
Pharmaceuticals International, represent a major step forward in the development of an effective
Zurich, Switzerland
candidates are under advanced clinical development:
and safe dengue vaccine for use in people of all ages, irrespective
(M Rauscher PhD, TV003/TV005 and TDENV PIV. Several phase 1 and 2 clinical
of previous dengue exposure at the time of vaccination.
S Bizjajeva PhD, I LeFevre MD, trials have shown two doses of TAK-003 to be immunogenic
A Borkowski MD);
against all four dengue serotypes in adults and children Implications of all the available evidence
and Hospital del Niño
Dr José Renán Esquivel, Sistema regardless of previous dengue exposure. This phase 3 trial was A solution is required urgently to the major public health
Nacional de Investigación at designed to assess the efficacy and safety of TAK-003 in children challenges raised by dengue disease. The efficacy and safety
Secretaría Nacional de Ciencia y aged 4–16 years in eight dengue-endemic countries in Asia and data presented in this report suggests potential for TAK-003 as
Tecnología, Centro de
Latin America. A previous report described primary endpoint a component of the multimodal approach to control dengue.
Vacunación Internacional
(Cevaxin), Panama City,
Panama
(Prof Xavier Saez-Llorens MD) previous infection.7–10 Vaccines that can provide protection against VCD.24 Onset of efficacy was observed after one
Correspondence to: in all age groups including young children, regardless of dose, with 81·0% (64·1–90·0) efficacy during the period
Dr Shibadas Biswal, Takeda serostatus before immunisation, are still needed. between the two doses. Exploratory analyses of secondary
Vaccines, Boston, MA 02139,
The Takeda tetravalent dengue vaccine candidate efficacy endpoints during part 1 found vaccine efficacy
USA
shibadas.biswal@takeda.com (TAK-003) was originally designed and constructed by was 74·9% (57·0–85·4) in individuals who were
See Online for appendix
scientists at the Division of Vector-Borne Diseases of the seronegative at baseline and 95·4% (88·4–98·2) against
US Centers for Disease Control and Prevention using a hospitalised VCD. Here, we present the main findings
live-attenuated dengue serotype 2 virus, which provides after completion of part 2 of the study (ie, 18 months after
the genetic backbone for all four vaccine viruses.11,12 The the second dose), in which secondary vaccine efficacy
DENV 2 strain (TDV-2) is based on an attenuated endpoints by baseline serostatus, against individual
laboratory-derived virus, DEN-2 PDK-53.13 The other serotypes, hospitalised VCD, and severe VCD were
three vaccine strains (TDV-1, TDV-3, and TDV-4) are assessed formally.
chimeras generated by replacing the envelope and pre-
membrane genes of TDV-2 with those from wild-type Methods
DENV 1, DENV 3, and DENV 4 strains.12,14–16 Study design and participants
Phase 1 and 2 studies have shown TAK-003 to be well We completed part 2 (18 months post-vaccination) of
tolerated and immunogenic against serotypes 1–4.17–22 In a an ongoing phase 3, randomised, double-blind trial.
placebo-controlled study published alongside this Article, Healthy children aged 4–16 years inclusive were enrolled
in which safety and immunogenicity were assessed up to and randomly assigned at 26 medical and research
48 months after vaccination, antibodies persisted above centres in eight dengue-endemic countries: Brazil
baseline at 48 months and there was a lower relative risk (four sites); Colombia (four sites); Dominican Republic
of virologically con­firmed dengue (VCD; 0·35, 95% CI (two sites); Nicaragua (one site); Panama (four sites);
0·19–0·65).23 We did a large, phase 3, randomised clinical the Philippines (four sites); Sri Lanka (four sites); and
trial in three parts at sites in Latin America and Asia Thailand (three sites). The main exclusion criteria were
to assess the efficacy, immunogenicity, and safety of febrile illness at the time of randomisation, impaired or
two doses of TAK-003 in healthy children aged 4–16 years. altered immune function, hypersensitivity or allergy to
At the end of part 1 (ie, 12 months after the second dose) any vaccine component, pregnancy or breastfeeding,
of this study, we presented an interim report with and previous receipt of a dengue vaccine (study
assessment of the primary endpoint, showing a high inclusion and exclusion criteria are described fully in
overall vaccine efficacy of 80·2% (95% CI 73·3–85·3) the appendix p 3).

1424 www.thelancet.com Vol 395 May 2, 2020

 Articles

Informed assent or consent forms, and the study vaccination. Part 2, as reported here, lasted for a further
protocol and its amendments were reviewed and ap­ 6 months for the assessment of secondary efficacy
proved by institutional review boards, independent ethics endpoints, and is being followed by an additional 3-year
committees, or health authorities. Written informed period (part 3) for long-term efficacy and safety evalu­
assent or consent was obtained from all participants or ation. The total duration of follow-up at the end of part 2
their parents or legal guardians before enrolment. During was around 21 months after the first dose, or 18 months
the study, re-consent was obtained from participants as after second dose administration.
they legally became adults. At the time of analysis at the One 0·5 mL dose of TAK-003 contained approximately
end of part 1, adult re-consent was still in process for 3·6, 4·0, 4·6, and 5·1 log10 plaque forming units of
some participants; however, all data were included and TDV-1, TDV-2, TDV-3, and TDV-4, respectively. Placebo
analysed. In the subsequent analysis at the end of part 2 was a 0·5 mL injection of saline. Vaccine or placebo was
of the trial, any data collected after legal adult age was administered subcutaneously into the upper arm. The
reached was removed if re-consent had not been obtained lyophilised vaccine kits were kept at 2–8°C during
within a reasonable timeframe; however, this did not shipping and storage, and reconstituted before admin­
involve the censoring of any dengue cases. This trial is in istration in PBS solution.
accordance with the Declaration of Helsinki and the ICH Blood samples were taken from all participants on day 1
harmonised tripartite guidelines for good clinical practice, (pre-vaccination) and day 120 to measure amounts of
with applicable local regulations. dengue neutralising antibodies by microneutralisation
test. Additional microneutralisation test blood samples
Randomisation and masking were taken on days 30, 90, 270, and 450, and then annually
Children who met the study entry criteria were randomly from the subset participants. Micro­ neutralisation test
assigned 2:1 to receive two doses of TAK-003 or two titres are expressed as the reciprocal of the highest
doses of placebo, 3 months apart. Randomisation was dilution of test serum that shows a 50% reduction in
stratified by region and age (4–5 years, 6–11 years, plaque counts compared with that of virus controls.
and 12–16 years) using an interactive web response Subset safety assessments included diary-recorded local
system and dynamic block assignment. Random­isation reactions for 7 days post-vaccination and systemic adverse
information was gen­erated by personnel authorised by events for 14 days post-vaccination, and unsolicited
the trial sponsor, and stored in a secure area accessible adverse events for 28 days post-vaccination. Serious
only to authorised personnel. A subset of 4000 of the adverse events and adverse events leading to withdrawal
20 099 participants was randomly selected as described from the study were collected in all study participants for
for additional safety and immunogenicity assessments. the duration of the trial.
Investigators, participants, and their parents or guard­ During active surveillance, participants presenting
ians, and sponsor representatives advising on trial with febrile illness or clinically suspected dengue had
conduct were unaware of trial group assignments. blood samples taken in the acute (ie, as soon as possible
One or more designated pharmacists or vaccine admin­ and preferably within 5 days of fever onset) and
istrators were unmasked at each site, but had no role in convalescent phases (ie, 7–14 days after the acute
the collection or assessment of participant safety data. sample; appendix p 7). Testing included: quantitative
These individuals accessed randomisation infor­mation serotype-specific RT-PCR, dengue NS1/IgM/IgG ELISA,
through a web portal. To maintain masking, medical haematocrit, liver enzyme (aspartate aminotransferase
writers and some sponsor-affiliated authors had access and alanine aminotransferase), and platelet counts.
to group and anonymised individual-level study data. RT-PCR and NS1 ELISA assays were done only on the
Other authors had access only to the data presented in acute sample. Febrile illnesses were evaluated clinically,
this report. An independent data monitoring committee and additional tests could be done as per local stan­
with responsibility for safety oversight had access to dard of care. Additional details of study methods and
unmasked data on request. procedures are provided in the appendix (p 3).

Procedures Outcomes
The study consists of three parts for each participant, For efficacy objectives, VCD was defined as febrile
with active surveillance during part 1 and part 2, and illness or illness clinically suspected to be dengue by the
modified active surveillance during part 3 (appendix p 7). investigator with confirmation by positive serotype-
Participants or their parents or guardians were contacted specific RT-PCR. Only the first VCD case in a participant
at least weekly for the entire duration of the study to was included in the overall vaccine efficacy analysis.
remind them to present for evaluation of febrile illness However, vaccine efficacy analysis by serotype included
(defined as fever ≥38°C on any 2 of 3 consecutive days) to the first VCD case for a specific serotype in an individual
ensure robust identification of dengue cases. Part 1 was participant. The primary endpoint was vaccine efficacy
completed once at least 120 VCD cases were confirmed of two doses of TAK-003 in preventing VCD induced
and participants had 12 months follow-up post-second by any dengue serotype occurring from 30 days

www.thelancet.com Vol 395 May 2, 2020 1425

 Articles

hospitalisation were not defined in the study protocol;

23 401 individuals screened for eligibility participants were hospitalised according to the judgment
of individual investigators. Severity of VCD was assessed
3302 not randomly assigned
using two approaches: (1) masked review by the Dengue
616 screen failure Case Adjudication Committee (DCAC) using predefined
2391 withdrawal by participant or criteria, and (2) by a programme developed by the study
participant's parent or guardian
295 other statisticians to analyse data according to the WHO 1997
dengue haemorrhagic fever criteria.25 Details of the
DCAC criteria are provided in the appendix (p 5).
20 099 randomly assigned

Statistical analysis
Efficacy endpoint analyses were done on the per protocol
6698 assigned to placebo 13 401 assigned to TAK-003
set data (ie, all participants without any major protocol
violations; all analysis sets are defined in the appendix
p 5). Vaccine efficacy is defined as 1–(λV/λC), where λV
11 not treated* 17 not treated* and λC denote the hazard rates for the TAK-003 and
placebo groups, respectively. Hazard ratios and corres­
6687 received at least one dose and 13 380 received at least one dose ponding 95% CIs were estimated using a Cox proportional
included in safety set and included in safety set hazard model with trial group as a factor, adjusted for age,
and stratified by region. The primary vaccine efficacy
objective was considered to be met if the lower bound of
121 discontinued before 231 discontinued before
second dose in the second dose in the the 95% CI for vaccine efficacy was above 25%. The
randomisation set† randomisation set† sample size calculation was based on the assumption
5 adverse events 9 adverse events
5 lost to follow-up 17 lost to follow-up
of true vaccine efficacy of 60% and a background
6 pregnancies 11 pregnancies annual dengue incidence of 1%. Randomisation of
2 protocol violations 4 protocol violations 20 100 participants in a 2:1 ratio (TAK-003:placebo) could
101 withdrawn‡ 181 withdrawn‡
2 other reasons 9 other reasons enable identifi­cation of 120 VCD cases between 30 days
post-second vaccination and the end of part 1, providing
at least 90% power to rule out a vaccine effect of 25%
6573 received two doses in the 13 164 received two doses in the
safety set safety set or more (with a two-sided significance level of 0·05).
Secondary vaccine efficacy endpoints were evaluated on
the per protocol set using the same methods as the
6366 completed part 2 in the 12 757 completed part 2 in the
safety set safety set
primary endpoint analysis with the aim to rule out vaccine
efficacy of 0% in the assessment period 30 days post-
second vaccination until the end of part 2. Additional
6317 evaluated for illness and 12 704 evaluated for illness and
included in per protocol set included in per protocol set
analyses were done on the per protocol set, safety set, full
1726 seronegative at baseline 3533 seronegative at baseline analysis set, and safety and immunogenicity subsets.
4588 seropositive at baseline 9167 seropositive at baseline Statistical analyses were done using SAS version 9.3
software. An independent data monitoring committee
1247 included in immunogenicity 2518 included in immunogenicity was used in the study. This study is registered with
analysis per protocol analysis per protocol ClinicalTrials.gov, NCT02747927.

Figure 1: Trial profile Role of the funding source

Participants who did not receive a vaccine dose are included in the total numbers of participants who discontinued the The sponsor was responsible for overall study design
trial before the second dose. Three participants in the vaccine group and two in the placebo group did not receive a
second dose, but continued in the study. Four participants (three assigned to the vaccine group and one assigned to the (taking into consideration investigators’ input), study site
placebo group) received both vaccine and placebo due to an administrative error; these participants were consequently selection, and data analysis. The study investigators were
excluded from the vaccine and placebo groups in the safety population. One participant assigned to the vaccine group responsible for data collection and day-to-day study site
received placebo; this participant was consequently included in the placebo group in the safety population. Participants management. Employees and subcontractors of the
had 12 months of follow-up after second dose at the time of completing part 2 of the trial. Some data might differ from
what has been previously published24 due to the inclusion of updated datasets.*Reason not listed to preserve masking. sponsor had a role in study design, data collection, data
†Includes non-vaccinated participants. ‡Withdrawn by participant or parent or guardian. analysis, data interpretation, and writing and critical
review of the report. Based on a manuscript outline
post-second vaccination until the end of part 1.24 agreed on by the authors, a medical writer prepared a first
Secondary endpoints of vaccine efficacy against indivi­ draft manuscript (funded by the sponsor). All authors had
dual sero­types, by baseline serostatus, and efficacy in full access to the presented data, provided critical input
prevention of hospitalisation and severe dengue were during manuscript preparation, and approved the final
assessed in the timeframe of 30 days post-second version for submission. All authors had final responsi­
vaccination to the end of part 2. Specific criteria for bility for the decision to submit for publication.

1426 www.thelancet.com Vol 395 May 2, 2020

 Articles

TAK-003 dengue TAK-003 cases per Placebo dengue Placebo cases per Vaccine efficacy
cases 100 person-years cases 100 person-years (95% CI)
Secondary efficacy endpoints
Seropositive, 4–16 years 75/9167 (0·8%) 0·6 150/4589 (3·3%) 2·4 76·1% (68·5 to 81·9)
Seronegative, 4–16 years 39/3531 (1·1%) 0·8 56/1726 (3·2%) 2·4 66·2% (49·1 to 77·5)
DENV 1 38/12 700 (0·3%) 0·2 62/6316 (1·0%) 0·7 69·8% (54·8 to 79·9)
DENV 2 8/12 700 (<0·1%) <0·1 80/6316 (1·3%) 0·9 95·1% (89·9 to 97·6)
DENV 3 63/12 700 (0·5%) 0·4 60/6316 (0·9%) 0·7 48·9% (27·2 to 64·1)
DENV 4 5/12 700 (<0·1%) <0·1 5/6316 (<0·1%) <0·1 51·0% (–69·4 to 85·8)
Exploratory efficacy endpoints
Overall 114/12 700 (0·9%) 0·6 206/6316 (3·3%) 2·4 73·3% (66·5 to 78·8)
Seropositive
4–5 years 18/957 (1·9%) 1·3 26/464 (5·6%) 4·1 67·7% (41·1 to 82·3)
6–11 years 40/4807 (0·8%) 0·6 80/2425 (3·3%) 2·4 76·2% (65·2 to 83·7)
12–16 years 17/3403 (0·5%) 0·4 44/1700 (2·6%) 1·9 81·2% (67·0 to 89·2)
Seronegative
4–5 years 14/662 (2·1%) 1·5 9/337 (2·7%) 1·9 22·9% (–78·1 to 66·7)
6–11 years 22/2200 (1·0%) 0·7 37/1065 (3·5%) 2·5 71·2% (51·2 to 83·0)
12–16 years 3/669 (0·4%) 0·3 10/324 (3·1%) 2·3 85·7% (47·9 to 96·1)
Seropositive
DENV 1 21/9167 (0·2%) 0·2 37/4589 (0·8%) 0·6 72·0% (52·2 to 83·6)
DENV 2 7/9167 (<0·1%) <0·1 54/4589 (1·2%) 0·9 93·7% (86·1 to 97·1)
DENV 3 43/9167 (0·5%) 0·3 54/4589 (1·2%) 0·9 61·8% (43·0 to 74·4)
DENV 4 4/9167 (<0·1%) <0·1 5/4589 (0·1%) <0·1 61·2% (–44·3 to 89·6)
Seronegative
DENV 1 17/3531 (0·5%) 0·3 25/1726 (1·4%) 1·0 67·8% (40·3 to 82·6)
DENV 2 1/3531 (<0·1%) <0·1 26/1726 (1·5%) 1·1 98·1% (85·8 to 99·7)
DENV 3 20/3531 (0·6%) 0·4 6/1726 (0·3%) 0·3 –68·2% (–318·9 to 32·4)
DENV 4 1/3531 (<0·1%) <0·1 0/1726 (0%) 0·0 ··
Seropositive
Asia Pacific 62/4391 (1·4%) 1·0 125/2169 (5·8%) 4·3 76·6% (68·3 to 82·7)
Latin America 13/4776 (0·3%) 0·2 25/2420 (1·0%) 0·7 73·7% (48·5 to 86·5)
Seronegative
Asia Pacific 27/1503 (1·8%) 1·3 35/773 (4·5%) 3·4 61·8% (36·8 to 76·9)
Latin America 12/2028 (0·6%) 0·4 21/953 (2·2%) 1·6 73·4% (45·9 to 86·9)
Yellow fever vaccine received 10/2719 (0·4%) 0·3 18/1355 (1·3%) 1·0 72·4% (40·1 to 87·2)
Yellow fever vaccine not received 104/9981 (1·0%) 0·7 188/4961 (3·8%) 2·8 73·4% (66·2 to 79·1)
Japanese encephalitis vaccine received 16/3157 (0·5%) 0·4 82/1552 (5·3%) 4·0 90·8% (84·3 to 94·6)
Japanese encephalitis vaccine not received 98/9543 (1·0%) 0·7 124/4764 (2·6%) 1·9 61·5% (49·8 to 70·4)
Yellow fever and Japanese encephalitis 88/6825 (1·3%) 0·9 106/3409 (3·1%) 2·3 59·6% (46·4 to 69·6)
vaccine not received
Data are n/N (%) or % (95% CI) unless otherwise specified. Per protocol set data for 30 days post-second vaccination until end of part 2 (duration 17 months). Percentages
were calculated on the basis of the number of participants who underwent evaluation for VCD. In the per protocol population, 12 700 of 12 704 participants in the vaccine
group and 6316 of 6317 in the placebo group were included in the evaluation of endpoints. The per protocol set was determined after exclusion of participants in a masked
manner before database lock in accordance with prespecified criteria (see appendix p 5). For analyses involving the per protocol population, data from participants who
discontinued were censored at the day of discontinuation. Two instances of VCD occurred in two participants during parts 1 and 2 of the study (only the first instance was
included in the efficacy calculation, except when calculating efficacy by serotype). For serotype-specific vaccine efficacy calculations, only the first instance of VCD due to the
individual serotype in question was included, regardless of previous instances of VCD due to other serotypes. Participants were classified as seronegative when testing
seronegative for all dengue serotypes at baseline. Participants were classified as seropositive when showing a reciprocal neutralising antibody titre of 10 or more against at
least one dengue serotype at baseline. VCD=virologically confirmed dengue.

Table 1: Efficacy of TAK-003 in preventing VCD fever

Results screening 23 401 individuals, 20 099 participants were

We completed part 2 of this trial between March 1, 2018, randomly assigned and 20 071 (99·9%) received a first
and Jan 7, 2019. The trial commenced on Sept 7, 2016, injection. 6698 (33·3%) of 20 099 were assigned to the
and is planned to be completed by December, 2021. After placebo group, 11 (0·2%) of whom were not treated.

www.thelancet.com Vol 395 May 2, 2020 1427

 Articles

TAK-003 cases TAK-003 cases per Placebo cases Placebo cases per Vaccine efficacy
100 person-years 100 person-years (95% CI)
Secondary efficacy endpoints
Hospitalised, 4–16 years 13/12 700 (0·1%) <0·1 66/6316 (1·0%) 0·8 90·4% (82·6 to 94·7)
Dengue haemorrhagic fever* 2/12 700 (<0·1%) <0·1 7/6316 (0·1%) <0·1 85·9% (31·9 to 97·1)
Severe VCD† 2/12 700 (<0·1%) <0·1 1/6316 (<0·1%) <0·1 2·3% (–977·5 to 91·1)
Exploratory efficacy endpoints
Hospitalised, 4–5 years 5/1619 (0·3%) 0·2 6/801 (0·7%) 0·5 59·1% (–33·9 to 87·5)
Hospitalised, 6–11 years 6/7009 (<0·1%) <0·1 41/3491 (1·2%) 0·8 92·9% (83·4 to 97·0)
Hospitalised, 12–16 years 2/4072 (<0·1%) <0·1 19/2024 (0·9%) 0·7 94·8% (77·8 to 98·8)
Dengue haemorrhagic fever*
Seropositive 1/9167 (<0·1%) <0·1 6/4589 (0·1%) <0·1 91·7% (30·9 to 99·0)
Seronegative 1/3531 (<0·1%) <0·1 1/1726 (<0·1%) <0·1 49·4% (–709·2 to 96·8)
Severe VCD†
Seropositive 0/9167 0 1/4589 (<0·1%) <0·1 100% (NA)
Seronegative 2/3531 (<0·1%) <0·1 0/1726 0 ··
Hospitalised, seropositive
4–16 years 8/9167 (<0·1%) <0·1 45/4589 (1·0%) 0·7 91·4% (81·7 to 95·9)
4–5 years 3/957 (0·3%) 0·2 3/464 (0·6%) 0·5 51·6% (–139·7 to 90·2)
6–11 years 3/4807 (<0·1%) <0·1 26/2425 (1·1%) 0·8 94·5% (81·9 to 98·3)
12–16 years 2/3403 (<0·1%) <0·1 16/1700 (0·9%) 0·7 93·8% (73·0 to 98·6)
Hospitalised, seronegative
4–16 years 5/3531 (0·1%) 0·1 21/1726 (1·2%) 0·9 88·1% (68·5 to 95·5)
4–5 years 2/662 (0·3%) 0·2 3/337 (0·9%) 0·6 65·3% (–108·0 to 94·2)
6–11 years 3/2200 (0·1%) <0·1 15/1065 (1·4%) 1·0 90·0% (65·6 to 97·1)
12–16 years 0/669 0 3/324 (0·9%) 0·7 100% (NA)
Hospitalised, Asia Pacific 12/5894 (0·2%) 0·1 61/2942 (2·1%) 1·5 90·4% (82·2 to 94·8)
Hospitalised, Latin America 1/6806 (<0·1%) <0·1 5/3374 (0·1%) 0·1 90·1% (15·3 to 98·8)
Data are n/N (%) unless otherwise specified. Per protocol set data for 30 days post-second vaccination until end of part 2 (duration 17 months). Percentages were calculated
on the basis of the number of participants who underwent evaluation for VCD. In the per protocol population, 12 700 of 12 704 participants in the vaccine group and 6316 of
6317 in the placebo group were included in the evaluation of endpoints. The per protocol set was determined after exclusion of participants in a masked manner before
database lock in accordance with prespecified criteria (see appendix p 5). For analyses involving the per protocol set, data from participants who discontinued were censored
at the day of discontinuation. Participants were classified as seronegative when testing seronegative for all dengue serotypes at baseline. Participants were classified as
seropositive when showing a reciprocal neutralising antibody titre of 10 or more against at least one dengue serotype at baseline. One case in the TAK-003 group met the
criteria defining both dengue haemorrhagic fever and severe VCD. VCD=virologically confirmed dengue. NA=not applicable. *VCD cases meeting WHO 1997 dengue
haemorrhagic fever criteria. †Determined by dengue case adjudication committee.

Table 2: Efficacy of TAK-003 in preventing hospitalisations due to VCD, severe VCD, and dengue haemorrhagic fever

13 401 (66·7) of 20 099 were assigned to the TAK-003 10 or more to at least one dengue virus serotype at
group, 17 (0·1%) of whom were not treated (figure 1). baseline and were considered to be seropositive.
First injections were administered between Sept 7, 2016, During parts 1 and 2 of the study period of around
and March 31, 2017. 19 741 (98·2%) participants received 21 months, 13 881 cases of febrile illness were reported,
both injections, and 19 126 (95·2%) completed part 2 of and acute samples were collected from 13 657 (98·4%) of
the study in the randomisation set, which corresponded these cases (93·6% within 5 days of fever onset).
to 18 months post-second dose. Enrolment was broadly 390 cases were VCD by serotype-specific RT-PCR,
balanced across Latin America (53·5%) and Asia Pacific including second episodes in two participants (safety
(46·5%) and baseline characteristics were similar across set data; appendix p 15); 98 of these cases (approxi­
both treat­ment groups (appendix p 8). The mean age of mately 25%) required hospitalisation. The greatest
participants was 9·6 years (SD 3·4) and 27·7% were number of VCD cases occurred in the Philippines,
seronegative for all four serotypes at baseline, as assessed where 131 (74·0%) of 177 cases were due to DENV 3. In
by microneutralisation test. The proportion of partici­ Sri Lanka, 63 (66%) of the 96 reported VCD cases were
pants who were sero­ negative at baseline varied by hospitalised and 57 (91%) of the 63 hospitalised cases
country (Panama 62·2%, Sri Lanka 38·5%, Thailand were due to DENV 2. No VCD cases were reported in the
34·4%, Brazil 28·8%, Nicaragua 22·3%, Colombia 15·4%, Dominican Republic during this period of observation.
Philippines 12·4%, and Dominican Republic 2·8%). Although all four serotypes were reported in Asia,
Other participants had a microneutralisation test titre of almost all the reported VCD cases in Latin America were

1428 www.thelancet.com Vol 395 May 2, 2020

 Articles

due to DENV 1 and DENV 2, except for one DENV 4 case

A Incidence of VCD cases
in Colombia. Of the 390 VCD cases during parts 1 and 2,
4·0 TAK-003, seropositive
259 cases were in the placebo group (78 [30·1%] DENV 1, TAK-003, seronegative
109 [42·1%] DENV 2, 65 [25·1%] DENV 3, and seven Placebo, seropositive

Percentage of participants
[2·7%] DENV 4). 3·0 Placebo, seronegative

Data describing vaccine efficacy against VCD in the

17-month observation period starting 30 days post-second 2·0
vaccination is presented in table 1. Vaccine efficacy
against VCD of any serotype was 76·1% (95% CI
1·0
68·5 to 81·9) in the baseline seropositive population and
66·2% (49·1 to 77·5) in the baseline seronegative popu­
lation. The overall vaccine efficacy during this time period 0
0 3 6 9 12 15 18 21
was 73·3% (66·5 to 78·8). Secondary vaccine efficacy Number at risk
endpoints by serotype were met for DENV 1–3 but TAK-003, seropositive 9667 9604 9509 9466 9425 9346 9242 6074
TAK-003, seronegative 3714 3690 3662 3643 3625 3608 3581 2663
continued to be variable; vaccine efficacy against DENV 2 Placebo, seropositive 4853 4808 4728 4674 4640 4581 4515 2934
was 95·1% (89·9 to 97·6), against DENV 1 was 69·8% Placebo, seronegative 1833 1818 1776 1764 1756 1738 1719 1293
(54·8 to 79·7), and against DENV 3 was 48·9% (27·2 to
64·1). Vaccine efficacy remained inconclusive against B Incidence of hospitalised VCD cases

DENV 4 at 51·0% (–69·4 to 85·8). 1·50

Vaccine efficacy in the prevention of hospitalisations
1·25
due to VCD, dengue haemorrhagic fever as per WHO
Percentage of participants

1997 criteria, and severe VCD as per DCAC criteria are 1·00
presented in table 2. Overall, 13 individuals with VCD
0·75
required hospitalisation in the TAK-003 group compared
with 66 individuals with VCD in the placebo group, 0·50
with a vaccine efficacy of 90·4% (95% CI 82·6–94·7);
overall vaccine efficacy was similar regardless of baseline 0·25

serostatus (91·4% in individuals who were seropositive vs 0

88·1% in individuals who were seronegative). Vaccine 0 3 6 9 12 15 18 21
efficacy against dengue haemorrhagic fever was 85·9% Time since first vaccination (months)
Number at risk
(31·9–97·1). Only three cases of severe VCD (all due to TAK-003, seropositive 9667 9612 9520 9485 9462 9391 9300 6127
TAK-003, seronegative 3714 3693 3667 3650 3640 3630 3608 2689
DENV 3) were reported with two cases occurring in the Placebo, seropositive 4853 4826 4759 4726 4713 4678 4624 3036
TAK-003 group and one case in the placebo group. There Placebo, seronegative 1833 1828 1792 1786 1781 1770 1758 1327
was little overlap in cases meeting the 1997 WHO and
Figure 2: Cumulative incidence of VCD and hospitalised cases of VCD by baseline serostatus (safety set data)
DCAC severity criteria; only one case in a vaccine recipient
VCD=virologically confirmed dengue.
met both the criteria. There were three cases in the
TAK-003 group and eight cases in the placebo group that
met either criteria. 0·66 to 4·05). Among the three hospitalised VCD cases in
Exploratory analysis by serostatus and serotype found vaccinees, two were classified as severe according to the
vaccine efficacy against DENV 1 of 72·0% (95% CI DCAC criteria. One of these two severe VCD cases in
52·2–83·6) in the seropositive populations and 67·8% vaccinees and the only hospitalised VCD case in the pla­
(40·3–82·6) in the seronegative populations, and vaccine cebo group also met the WHO 1997 dengue haemorrhagic
efficacy against DENV 2 of 93·7% (86·1–97·1) in the fever criteria.
seropositive populations and 98·1% (85·8–99·7) in the The cumulative incidence of VCD and hospitalised
seronegative populations; thus showing similar vaccine cases of VCD by baseline serostatus over a 21-month
efficacy regardless of baseline serostatus for DENV 1 and period following first dose are presented in figures 2, 3.
DENV 2. Vaccine efficacy analysis by serostatus was not Additional exploratory analyses of vaccine efficacy are
possible for DENV 4, with only one case reported in presented in table 1 and the appendix (pp 13–15). In
individuals who were seronegative at baseline. general, vaccine efficacy in the subgroups could be
Vaccine efficacy against DENV 3 varied by baseline largely explained by the variation of efficacy by serotype
serostatus. In individuals who were seropositive, vaccine and the relative distribution of serotypes in the analysis
efficacy was 61·8% (95% CI 43·0 to 74·4) whereas vaccine subpopulations (appendix pp 12, 15).
efficacy was not shown in individuals who were sero­ The clinical features of all VCD cases after first dose in
negative (–68·2%, –318·9 to 32·4). A numerical imbalance the safety set by serostatus are presented in the appendix
of VCD cases was seen (20 [three hospitalised] in vaccine (p 16). Bleeding, plasma leakage, and low platelet counts
recipients vs six [one hospitalised] in the placebo group), were reported in a small proportion of VCD cases, in both
with an inconclusive relative risk of 1·63 (95% CI the vaccine and placebo groups. Bleeding was observed in

www.thelancet.com Vol 395 May 2, 2020 1429

 Articles

DENV 1
four (5%) of 89 participants in the TAK-003 seropositive
2·0 TAK-003, seropositive group versus 12 (6%) of 188 in the placebo seropositive
TAK-003, seronegative group; and two (5%) of 42 in TAK-003 versus five (7%) of
Percentage of participants

1·6 Placebo, seropositive

Placebo, seronegative
71 in placebo seronegative groups. Plasma leakage was
1·2 observed in one (1%) of 89 in TAK-003 versus 13 (7%) of
188 in placebo seropositive groups; and two (5%) of 42 in
0·8
TAK-003 versus three (4%) of 71 in placebo seronegative
0·4 groups. Platelet counts of 100 × 10⁹ per L or less were
0 observed in five (6%) of 89 in TAK-003 versus 45 (24%) of
0 3 6 9 12 15 18 21 188 in placebo seropositive groups; and three (7%) of 42 in
Number at risk
TAK-003, seropositive 9667 9613 9523 9488 9462 9391 9293 6116 TAK-003 versus ten (14%) of 71 in the placebo seronegative
TAK-003, seronegative 3714 3692 3664 3646 3633 3621 3595 2679 groups. Further analysis of DENV 3 cases in individuals
Placebo, seropositive 4853 4826 4770 4748 4736 4698 4638 3046
Placebo, seronegative 1833 1824 1800 1795 1788 1772 1756 1325 who were seronegative at baseline found bleeding in
one (5%) of 22 TAK-003 group cases versus one (17%) of
DENV 2 six placebo group cases; plasma leakage in two (9%) of
2·0 22 TAK-003 group cases versus one (17%) of six placebo
group cases; and platelet counts of 100 × 10⁹ per L or less
Percentage of participants

1·6
in two (9%) of 22 TAK-003 group cases versus one (17%) of
1·2
six placebo group cases.
0·8 Cumulative rates of serious adverse events (parts 1
0·4
and 2) were similar between groups at 4·0% in the
vaccine group and 4·8% in the placebo group (table 3).
0 No additional cases of related serious adverse events
0 3 6 9 12 15 18 21
Number at risk during part 2 of the study were observed. One TAK-003
TAK-003, seropositive 9667 9609 9518 9482 9460 9390 9299 6124
TAK-003, seronegative 3714 3693 3667 3650 3640 3631 3610 2691
and four placebo recipients had serious adverse events
Placebo, seropositive 4853 4819 4747 4710 4698 4663 4611 3020 during part 1, which were considered by the investigators
Placebo, seronegative 1833 1823 1783 1775 1771 1761 1747 1316 to be related to receiving masked investigational product
(two had hypersensitivity, two were diagnosed with
DENV 3
2·0
dengue, and one with dengue haemorrhagic fever).
Additional serious adverse event data are provided in the
Percentage of participants

1·6 appendix (pp 9–11). 20 participants (<0·1% of safety set)

1·2 were withdrawn from the trial due to 21 adverse events
by the end of part 2; 14 of these participants received
0·8
TAK-003 and six received placebo. The six deaths that
0·4 occurred in parts 1 and 2 of the study (table 3; aseptic
0
meningitis, arteriovenous malformation of the cerebral
0 3 6 9 12 15 18 21 vessels, anaplastic ependymoma, gunshot wound, suffo­
Number at risk
TAK-003, seropositive 9667 9612 9520 9478 9443 9368 9273 6100 cation, and road traffic accident) were all considered to
TAK-003, seronegative 3714 3691 3665 3647 3632 3618 3596 2678 be unrelated to the investigational product or study
Placebo, seropositive 4853 4828 4774 4746 4721 4673 4623 3039
Placebo, seronegative 1833 1829 1807 1800 1797 1787 1775 1341
procedures by the investigators, sponsor, and data
monitoring committee.
DENV 4 Among those who presented with a febrile illness within
2·0 30 days of vaccination, vaccine viraemia was detected in
34 (7%) of 479 participants after the first dose (31 of 34 with
Percentage of participants

1·6
serotype 2), and in one (<1%; serotype 2) of 503 partici­
1·2 pants after the second dose (30 [88·2%] of 34 cases had
0·8 onset of fever within 7–13 days of first dose administration).
This included 15 participants with replication-competent
0·4
vaccine virus (all serotype 2), with four isolates showing a
0 single reversion at the 5′NCR attenuation locus. None of
0 3 6 9 12 15 18 21
these four febrile illnesses were clinically diagnosed as
Time since first vaccination (months)
Number at risk dengue or had evidence of bleeding, positive tourniquet
TAK-003, seropositive 9667 9612 9523 9488 9464 9394 9304 6130
TAK-003, seronegative 3714 3693 3667 3650 3640 3631 3610 2691 test, low platelet count, or plasma leakage. Two of these
Placebo, seropositive 4853 4830 4779 4761 4752 4718 4670 3077 cases presented with rash, which resolved within 7 days.
Placebo, seronegative 1833 1829 1807 1803 1800 1791 1780 1345
Seropositivity rates (defined as group proportions with
Figure 3: Cumulative incidence of virologically confirmed dengue for each serotype by baseline serostatus reciprocal neutralising titres ≥10) against each serotype
(safety set data) over time are presented in the appendix (p 17). On day 120

1430 www.thelancet.com Vol 395 May 2, 2020

 Articles

(ie, 1 month after second dose), tetravalent seropositivity

TAK-003 (n=13 380) Placebo (n=6687)
rates in individuals who were seropositive at baseline
were 99·8% versus 85·2% in TAK-003 versus placebo Serious adverse events 538 (4·0%) 324 (4·8%)

recipients, and in individuals who were seronegative at Non-investigational product-related* serious adverse 537 (4·0%) 320 (4·8%)
events
baseline, 99·5% versus 3·5%.
Investigational product-related* serious adverse events 1 (<0·1%) 4 (<0·1%)
Geometric mean titres of neutralising antibodies against
Serious adverse events leading to investigational product 18 (0·1%) 8 (0·1%)
each serotype are presented in table 4. Generally, geometric withdrawal or trial discontinuation
mean titres were of similar magnitude to those observed Deaths 5 (<0·1%) 1 (<0·1%)
in previous studies of this vaccine.20 On day 120 (1 month
Data are number of participants with events (%). Participants with at least one adverse event after any injection
after second dose), TAK-003 induced geometric mean (vaccine or placebo administration). *As assessed by investigator.
titres of 2115 (DENV 1), 4897 (DENV 2), 1761 (DENV 3),
and 1129 (DENV 4) in individuals who were seropositive at Table 3: Overview of safety set data until end of part 2
baseline, and 184 (DENV 1), 1730 (DENV 2), 228 (DENV 3),
and 144 (DENV 4) in individuals who were seronegative at
baseline. On day 270, TAK-003 induced geometric mean TAK-003 seropositive Placebo seropositive TAK-003 seronegative Placebo seronegative
(n=1816) (n=902) (n=702) (n=345)
titres of 1446 (DENV 1), 3691 (DENV 2), 1088 (DENV 3),
and 778 (DENV 4) in individuals who were seropositive DENV 1
at baseline, and 87 (DENV 1), 929 (DENV 2), 72 (DENV 3), Day 1 410 (365–461) 445 (377–524) 5 (5–5) 5 (5–5)
and 64 (DENV 4) in individuals who were seronegative Day 30 2404 (2204–2622) 430 (361–512) 118 (106–131) 6 (5–6)
at baseline. Antibody persistence was observed on Day 90 1945 (1791–2112) 410 (349–481) 91 (82–102) 6 (5–6)
day 450 with TAK-003-induced geometric mean titres of Day 120 2115 (1957–2286) 451 (381–534) 184 (169–201) 6 (6–7)
1243 (DENV 1), 2993 (DENV 2), 799 (DENV 3), and Day 270 1446 (1328–1573) 415 (350–492) 87 (79–97) 6 (6–7)
817 (DENV 4) in individuals who were seropositive at Day 450 1243 (1139–1357) 451 (380–536) 77 (68–86) 7 (6–8)
baseline, and 77 (DENV 1), 656 (DENV 2), 53 (DENV 3), DENV 2
and 64 (DENV 4) in individuals who were seronegative at Day 1 745 (674–825) 802 (697–924) 5 (5–5) 5 (5–5)
baseline. Day 30 6696 (6300–7116) 744 (635–870) 6277 (5648–6977) 7 (6–7)
Day 90 4826 (4571–5095) 724 (624–839) 1682 (1543–1833) 7 (6–8)
Discussion Day 120 4897 (4646–5163) 766 (654–896) 1730 (1614–1855) 8 (7–9)
We previously reported the initial vaccine efficacy Day 270 3691 (3496–3898) 776 (665–906) 929 (855–1009) 9 (7–10)
analyses from part 1 of this 3-part study.24 The 6 months Day 450 2993 (2831–3165) 747 (637–875) 656 (601–717) 8 (7–10)
of additional follow-up in part 2 have allowed us a more DENV 3
precise look into the efficacy measures, particularly the Day 1 357 (321–398) 356 (305–415) 5 (5–5) 5 (5–5)
final analyses of secondary objectives. It also enabled Day 30 2254 (2093–2428) 349 (298–409) 194 (173–218) 6 (5–6)
observation of how the vaccine performed beyond the Day 90 1563 (1453–1682) 321 (277–374) 94 (85–104) 6 (5–6)
first year after completion of the vaccine course, when Day 120 1761 (1646–1884) 353 (301–414) 228 (212–246) 6 (5–7)
potential cross-protection from the dominant serotype 2 Day 270 1088 (1008–1174) 307 (261–360) 72 (66–78) 6 (6–7)
component might have declined. Broadly, the results
Day 450 799 (737–865) 282 (240–331) 53 (49–59) 6 (6–7)
from part 2 were similar to those of part 1.24
DENV 4
The trial met all of the secondary objectives for which
Day 1 218 (198–241) 234 (203–270) 5 (5–5) 5 (5–5)
there were sufficient VCD cases to enable assessment.
Day 30 1306 (1224–1393) 222 (191–258) 111 (98–125) 5 (5–6)
It confirmed the earlier exploratory findings of overall
Day 90 1002 (940–1069) 215 (187–248) 63 (57–70) 6 (5–6)
efficacy regardless of baseline serostatus (66·2% in
Day 120 1129 (1066–1196) 241 (208–280) 144 (134–155) 6 (5–6)
individuals who were seronegative vs 76·1% in indi­
Day 270 778 (730–830) 229 (197–266) 64 (59–70) 6 (6–7)
viduals who were seropositive), a high overall efficacy
Day 450 817 (765–873) 293 (253–341) 64 (58–71) 6 (6–7)
against hospitalised dengue (90·4%), and variable
efficacy by serotype (48·9–95·1%). The highest vaccine Data are geometric mean titres (95% CI). Per protocol set for immunogenicity data. Data rounded to nearest whole
number. Baseline seronegative is defined as seronegative to all serotypes. Baseline seropositive is defined as having
efficacy was seen against DENV 2, which provides the reciprocal neutralising antibody titres of 10 or more to one or more serotype. n refers to number of participants in the
genetic backbone of TAK-003, and it was likely to be analysis set (number of participants evaluated at each timepoint might vary).
contributed by anti-NS1 antibodies as well as cell-medi­
Table 4: Geometric mean titres by dengue serotype
ated responses.26,27 Overall vaccine efficacy at the end of
part 2 was 73·3%, which is lower than the 80·2% reported
at the end of part 1. This finding can be explained largely in the secondary versus primary endpoint analyses
by two factors: (1) a relatively higher proportion of DENV 1 timeframes; and (2), some decline in vaccine efficacy
(30·1% vs 20·1% in the placebo group and 33·3% vs primarily against DENV 3 (62·6% vaccine efficacy at the
26·2% in the TAK-003 group) and lower proportion of end of part 1 vs 48·9% at the end of part 2).
DENV 2 (38·3% vs 43·0% in the placebo group and 7·0% A comprehensive subanalysis was planned to fully
vs 4·9% in the TAK-003 group) VCD cases being included understand the underlying determinants of efficacy

www.thelancet.com Vol 395 May 2, 2020 1431

 Articles

beyond secondary endpoints. Vaccine efficacy in these functional impairment or substantial intervention to
subanalyses was dependent on the relative distribution manage complications (details of the DCAC criteria are
of serotypes identified in the particular subpopulation. provided in the appendix p 5). These three cases were all
For example, high efficacy against hospitalised dengue due to DENV 3 (one case in the placebo group and two
was influenced by a high proportion of DENV 2, and the cases in the TAK-003 group) and were associated with low
low efficacy observed in the 4–5-year age group was blood pressure, hypotensive shock, or respiratory distress.
influenced by a high proportion of DENV 3 along with a One of the limitations of this study is that it is not large
low proportion of DENV 2. Similarly, the higher efficacy enough to enable the identification of dengue severe
estimates in participants previously immunised with enough to be associated with functional impair­ment. In
Japanese encephalitis vaccine was influenced by the addition to DCAC-defined severe dengue, nine cases of
relatively high proportion of DENV 1 and DENV 2 cases dengue that met the WHO 1997 criteria25 of dengue
in Sri Lanka and Thailand, where vaccination against haemorrhagic fever were identified. These were due to
Japanese encephalitis is routine. Further analysis to DENV 1 (n=2), DENV 2 (n=4), and DENV 3 (n=3) and
understand vaccine efficacy better in these subpopu­ occurred frequently enough to enable the demonstration
lations is planned. of vaccine efficacy against dengue haemorrhagic fever
A key finding during the analysis of part 1 was the (85·9%, 95% CI 31·9–97·1). There was little overlap
absence of efficacy against DENV 3 in individuals who between these definitions with only one case (in the
were seronegative at baseline, and this observation con­ TAK-003 group) meeting both the dengue haemorrhagic
tinued in part 2. This finding was only made possible by fever and the DCAC definitions. The low incidence of
important features of the trial design that enabled such a severe cases (dengue haemorrhagic fever-defined or
detailed assessment of vaccine efficacy subgroups. These DCAC-defined) and the observation that the efficacy of
included the large sample size, the broad geographical TAK-003 varied by serotype suggests the need for cautious
distribution of trial sites, baseline sampling in all par­ interpretation of these data.
ticipants to enable full assessment of serostatus, an age We have previously reported that TAK-003 was well
range that ensured inclusion of sufficient participants who tolerated, and that no important safety risks were
were seronegative at baseline, and the continuation of identified in part 1.24 There was no change to this
long-term active surveillance. Without these features of the conclusion after cumulative analysis of adverse events in
trial design, the efficacy observed against DENV 3 overall part 2.
and in individuals who were seronegative generally would With the protocol-defined primary and secondary
have masked this important finding. It is important to efficacy objectives assessed up to 18 months after last
continue monitoring this trend, particularly to see if the vaccine dose, it is reasonable to consider the potential
imbalance of febrile illness translates into an increase in utility of this vaccine candidate. Dengue is a growing
hospitalisation. Currently, there are too few cases of public health threat for more than half of the global
hospitalised dengue due to DENV 3 in participants who population,4,28,29 and there are considerable challenges
were seronegative (n=4) for any conclusions to be drawn. in developing an effective vaccine.30 The only licensed
Nearly all DENV 3 cases were reported from the Philippines vaccine is not indicated in those younger than 9 years
where the propor­tion of participants who were seronegative and requires the identification of individuals who are
was low (n=480 [12·4%]). During the long-term follow-up seropositive.7–10 Other public health measures have
period, additional cases of DENV 3 outside the Philippines traditionally failed to eliminate the risk of dengue.31,32 In
and in older children who are seronegative might help to this context, TAK-003 has shown high overall efficacy
provide a clearer picture. against both symptomatic and hospitalised dengue
The trial was not able to conclude on vaccine efficacy irrespective of baseline serostatus, with rapid onset of
against two secondary endpoints, VCD due to DENV 4, protection after one dose. However, efficacy differs by
and severe dengue as defined by the DCAC criteria. The serotype and more data are needed to understand the
trial design anticipated difficulties in identifying all safety and efficacy profile of this candidate against
serotypes by including 26 sites across eight countries to DENV 3 and DENV 4 in individuals who are seronegative.
provide epidemiological heterogeneity and an additional These nuances will require careful balancing by regu­
6 months of surveillance for secondary efficacy endpoints. lators and public health officials in determining the
Despite these measures, there were too few DENV 4 potential utility of this vaccine as a component of a
cases to enable a conclusion. Encouragingly, the point multimodal approach to reducing the global burden of
estimate of efficacy against DENV 4 was positive and dengue.
similar to the efficacy estimates for DENV 1 and DENV 3. Contributors
Comparison of RT-PCR and NS1 antigen ELISA results CB-T, LMV, HV, MTA, VS, EJR-A, DY, VPW, EDM, ADF, DG, PK, FE,
in the acute samples ruled out the possibility of missing EL-M, LB, and XS-L were study investigators. ShB, AB, PG, ST, MR,
and DW were responsible for study design. ST, YH, IE, and PG were
cases with PCR (data not shown). responsible for medical monitoring and data review. ShB, AB, MR, IL,
Only three cases of dengue met the DCAC criteria of SvB, and DW were responsible for data analysis and interpretation.
severity, which required cases to be associated with either ShB was responsible for publication management.

1432 www.thelancet.com Vol 395 May 2, 2020

 Articles

Declaration of interests 15 Osorio JE, Huang CY, Kinney RM, Stinchcomb DT. Development of
ShB, ST, YH, PG, IE, MR, SvB, IL, AB, and DW are permanent employees DENVax: a chimeric dengue-2 PDK-53-based tetravalent vaccine for
of the Takeda group of companies. DW has a patent WO2017/179017 protection against dengue fever. Vaccine 2011; 29: 7251–60.
pending. XS-L, PK, and LB have served as advisory board members for 16 Osorio JE, Wallace D, Stinchcomb DT. A recombinant, chimeric
Takeda. All other authors declare no competing interests. tetravalent dengue vaccine candidate based on a dengue virus
serotype 2 backbone. Expert Rev Vaccines 2016; 15: 497–508.
Data sharing 17 George SL, Wong MA, Dube TJ, et al. Safety and immunogenicity
Access to the study protocol, statistical analysis plan, and de-identified of a live attenuated tetravalent dengue vaccine candidate in
patient data will be provided to suitably qualified researchers at the flavivirus-naive adults: a randomized, double-blinded phase 1
discretion of an independent committee using a data sharing platform clinical trial. J Infect Dis 2015; 212: 1032–41. For data sharing platform see
on written request when appropriate marketing authorisation has been 18 Osorio JE, Velez ID, Thomson C, et al. Safety and immunogenicity https://vivli.org/ourmember/
received. of a recombinant live attenuated tetravalent dengue vaccine takeda/
(DENVax) in flavivirus-naive healthy adults in Colombia:
Acknowledgments a randomised, placebo-controlled, phase 1 study. Lancet Infect Dis
We thank the trial participants and their parents or guardians for agreeing 2014; 14: 830–38.
to take part in the trial, the data monitoring committee, the dengue case 19 Rupp R, Luckasen GJ, Kirstein JL, et al. Safety and immunogenicity
adjudication committee members, the Takeda expanded study team, of different doses and schedules of a live attenuated tetravalent
the study team at Pharmaceutical Product Development, and all the trial dengue vaccine (TDV) in healthy adults: a phase 1b randomized
staff in each of the countries. The authors would also like to acknowledge study. Vaccine 2015; 33: 6351–59.
the contribution of the late Dan Stinchcomb, and employees of Inviragen 20 Sáez-Llorens X, Tricou V, Yu D, et al. Immunogenicity and safety of
(Fort Collins, CO, USA) in the initial developmental work of TAK-003 one versus two doses of tetravalent dengue vaccine in healthy
(Inviragen was subsequently acquired by Takeda). The authors are children aged 2–17 years in Asia and Latin America: 18-month
grateful to Jamie Stirling (OLC Bioscience, London, UK) and Keith Veitch interim data from a phase 2, randomised, placebo-controlled study.
(keithveitch communications, Amsterdam, Netherlands) for editorial Lancet Infect Dis 2018; 18: 162–70.
assistance in the preparation of this manuscript. 21 Sáez-Llorens X, Tricou V, Yu D, et al. Safety and immunogenicity of
one versus two doses of Takeda’s tetravalent dengue vaccine in
References children in Asia and Latin America: interim results from a phase 2,
1 WHO. Dengue guidelines for diagnosis, treatment, prevention and randomised, placebo-controlled study. Lancet Infect Dis 2017;
control: new edition. World Health Organization Press, 2009. 17: 615–25.
https://apps.who.int/iris/handle/10665/44188 (accessed 22 Sirivichayakul C, Barranco-Santana EA, Esquilin-Rivera I, et al.
Feb 10, 2020). Safety and immunogenicity of a tetravalent dengue vaccine
2 Bhatt S, Gething PW, Brady OJ, et al. The global distribution and candidate in healthy children and adults in dengue-endemic
burden of dengue. Nature 2013; 496: 504–07. regions: a randomized, placebo-controlled phase 2 study. J Infect Dis
3 Brady OJ, Gething PW, Bhatt S, et al. Refining the global spatial 2016; 213: 1562–72.
limits of dengue virus transmission by evidence-based consensus. 23 Tricou V, Sáez-Llorens X, Yu D, et al. Safety and immunogenicity of a
PLoS Negl Trop Dis 2012; 6: e1760. tetravalent dengue vaccine in children aged 2–17 years: a randomised,
4 Simmons CP, Farrar JJ, Nguyen V, Wills B. Dengue. N Engl J Med placebo-controlled, phase 2 trial. Lancet 2020; published online
2012; 366: 1423–32. March 17. https://doi.org/10.1016/S0140-6736(20)30556-0.
5 WHO. Fact sheets. Dengue and severe dengue. 2019. http://www. 24 Biswal S, Reynales H, Saez-Llorens X, et al. Efficacy of a tetravalent
who.int/news-room/fact-sheets/detail/dengue-and-severe-dengue dengue vaccine in healthy children and adolescents. N Engl J Med
(accessed Feb 10, 2020). 2019; 381: 2009–19.
6 Guzman MG, Halstead SB, Artsob H, et al. Dengue: a continuing 25 WHO. Dengue haemorrhagic fever: diagnosis, treatment,
global threat. Nat Rev Microbiol 2010; 8 (suppl): S7–16. prevention, and control. Geneva: World Health Organization, 1997.
7 Capeding MR, Tran NH, Hadinegoro SR, et al. Clinical efficacy and https://www.who.int/csr/resources/publications/dengue/
safety of a novel tetravalent dengue vaccine in healthy children in Denguepublication/en/ (accessed Feb 10, 2020).
Asia: a phase 3, randomised, observer-masked, placebo-controlled 26 Waickman AT, Friberg H, Gargulak M, et al. Assessing the diversity
trial. Lancet 2014; 384: 1358–65. and stability of cellular immunity generated in response to the
8 Sridhar S, Luedtke A, Langevin E, et al. Effect of dengue serostatus on candidate live-attenuated dengue virus vaccine TAK-003.
dengue vaccine safety and efficacy. N Engl J Med 2018; 379: 327–40. Front Immunol 2019; 10: 1778.
9 Villar LA, Rivera-Medina DM, Arredondo-García JL, et al. Safety and 27 Sharma M, Glasner DR, Watkins H, et al. Magnitude and
immunogenicity of a recombinant tetravalent dengue vaccine in functionality of the NS1-specific antibody response elicited by a live-
9–16 year olds: a randomized, controlled, phase II trial in attenuated tetravalent dengue vaccine candidate. J Infect Dis 2019;
Latin America. Pediatr Infect Dis J 2013; 32: 1102–09. published online Feb 19. DOI:10.1093/infdis/jiz081.
10 WHO. Dengue vaccine: WHO position paper—September 2018. 28 WHO. Emergencies. Top ten threats to global health in 2019. 2019.
Wkly Epidemiol Rec 2018; 93: 457–76. https://www.who.int/emergencies/ten-threats-to-global-health-
11 Butrapet S, Huang CY, Pierro DJ, Bhamarapravati N, Gubler DJ, in-2019 (accessed Feb 10, 2020).
Kinney RM. Attenuation markers of a candidate dengue type 2 29 Bos S, Gadea G, Despres P. Dengue: a growing threat requiring
vaccine virus, strain 16681 (PDK-53), are defined by mutations in vaccine development for disease prevention. Pathog Glob Health
the 5′ noncoding region and nonstructural proteins 1 and 3. J Virol 2018; 112: 294–305.
2000; 74: 3011–19. 30 Wilder-Smith A, Ooi EE, Horstick O, Wills B. Dengue. Lancet 2019;
12 Huang CY, Kinney RM, Livengood JA, et al. Genetic and phenotypic 393: 350–63.
characterization of manufacturing seeds for a tetravalent dengue 31 Alvarado-Castro V, Paredes-Solís S, Nava-Aguilera E, et al.
vaccine (DENVax). PLoS Negl Trop Dis 2013; 7: e2243. Assessing the effects of interventions for Aedes aegypti control:
13 Yoksan S, Bhamarapravati N, Halstead SB. Dengue virus vaccine systematic review and meta-analysis of cluster randomised
development: study on biological markers of uncloned dengue 1–4 controlled trials. BMC Public Health 2017; 17 (suppl 1): 384.
viruses serially passaged in primary kidney cells. In: TD St George 32 Bowman LR, Donegan S, McCall PJ. Is dengue vector control
BKaJBe, ed. Arbovirus research in Australia Proceedings of the deficient in effectiveness or evidence?: systematic review and
4th Symposium, CSIRO/QIMR; 1986; Brisbane, Australia; 1986. meta-analysis. PLoS Negl Trop Dis 2016; 10: e0004551.
35–38.
14 Huang CY, Butrapet S, Tsuchiya KR, Bhamarapravati N, Gubler DJ,
Kinney RM. Dengue 2 PDK-53 virus as a chimeric carrier for
tetravalent dengue vaccine development. J Virol 2003; 77: 11436–47.

www.thelancet.com Vol 395 May 2, 2020 1433