Articles

Recurrent venous thromboembolism in patients with

pulmonary embolism and right ventricular dysfunction:
a post-hoc analysis of the Hokusai-VTE study
Marjolein P A Brekelmans, Walter Ageno, Ludo F Beenen, Benjamin Brenner, Harry R Buller, Cathy Z Chen, Alexander T Cohen, Michael A Grosso,
Guy Meyer, Gary Raskob, Annelise Segers, Thomas Vanassche, Peter Verhamme, Philip S Wells, George Zhang, Jerey I Weitz

Summary
Background In patients with pulmonary embolism, right ventricular dysfunction is associated with early mortality. Lancet Haematol 2016;
The Hokusai-VTE study used N-terminal pro-brain natriuretic peptide (NT-proBNP) and right to left ventricular 3: e43745

diameter ratio on CT as indicators of right ventricular dysfunction and reported that recurrent venous See Comment page e405
thromboembolism rates were lower with edoxaban than warfarin. The aim of the current study was to further explore Department of Vascular
the signicance of right ventricular dysfunction and investigate potential explanations for the superiority of Medicine, Academic Medical
Center, Amsterdam,
edoxabanie, dierences in baseline clinical characteristics, duration of initial heparin treatment, bleeding rates, or Netherlands
quality of warfarin treatment. (M P A Brekelmans MD,
Prof H R Buller MD); Department
of Clinical and Experimental
Methods The Hokusai-VTE trial was a randomised, double-blind, event-driven non-inferiority trial in patients from
Medicine, University of
centres in 37 countries that compared edoxaban with warfarin in the treatment of acute venous thromboembolism. Insubria, Varese, Italy
Patients received treatment for at least 3 months and up to a maximum of 12 months. Patients were followed up for (W Ageno MD); Department of
12 months. Outcome data at 12 months was collected for all patients irrespective of treatment duration. This Radiology, Academic Medical
Center, Amsterdam,
prespecied subgroup analysis focuses on the included patients with pulmonary embolism. The primary ecacy
Netherlands (L F Beenen MD);
outcome was the incidence of adjudicated symptomatic recurrent venous thromboembolism dened as a composite Thrombosis and Hemostasis
of deep vein thrombosis or non-fatal or fatal pulmonary embolism at 12 months. Recurrence rates with edoxaban and Unit, Department of
warfarin were compared in patients with and without right ventricular dysfunction. In those with NT-proBNP Hematology and Bone Marrow
Transplantation, Rambam
concentrations of 500 pg/mL or higher, we compared baseline characteristics, duration of heparin treatment, and
Medical Center, Haifa, Israel
bleeding leading to study drug discontinuation in the edoxaban and warfarin groups. We also assessed quality of (Prof B Brenner MD); Global
warfarin treatment. All analyses were done with the modied intention-to-treat population. The Hokusai-VTE trial is Medical Affairs, Daiichi Sankyo,
registered with ClinicalTrials.gov, number NCT00986154. Parsippany, NJ, USA
(C Z Chen MD); Department of
Haematological Medicine,
Findings Between Jan 28, 2010, and Oct 5, 2012, 8292 patients were enrolled from 439 centres, of whom 8240 received Guys and St Thomas Hospitals,
at least one dose of study drug. 3319 patients had pulmonary embolism. NT-proBNP was 500 pg/mL or higher in 465 Kings College London, London,
(30%) of 1565 patients given edoxaban and in 507 (32%) of 1599 given warfarin. Recurrent venous thromboembolism UK (A T Cohen MD); Clinical
Development, Daiichi Sankyo
occurred in 14 (3%) of 465 patients in the edoxaban group and 30 (6%) of 507 in the warfarin group (hazard ratio [HR] Pharma Development, Edison,
050, 95% CI 026094; p=0033). The right to left ventricular diameter ratio was 09 or higher in 414 (44%) of NJ, USA (M A Grosso MD,
937 patients in the edoxaban group and 427 (45%) of 946 in the warfarin group. Recurrent venous thromboembolism G Zhang PhD); Division of
occurred in 11 (3%) of 414 and 20 (5%) of 427 patients in the edoxaban and warfarin groups (HR 057, 95% CI Respiratory and Intensive Care
Medicine, George Pompidou
027117; p=013). Baseline characteristics, duration of heparin treatment, and rates of bleeding leading to study European Hospital, Universit
drug discontinuation were similar in the edoxaban and warfarin groups and the quality of warfarin management was Paris Descartes, Paris, France
adequate for patients with NT-proBNP concentrations of 500 pg/mL or higher. (Prof G Meyer MD); College of
Public Health, University of
Oklahoma Health Sciences
Interpretation Findings from our analysis suggest that edoxaban is more eective than warfarin in the treatment and Center, Oklahoma City, OK, USA
prevention of recurrent venous thromboembolism in patients with pulmonary embolism and evidence of right (Prof G Raskob PhD); Itreas BV,
ventricular dysfunction. Amsterdam, Netherlands
(A Segers MD); Vascular
Medicine and Haemostasis,
Funding Daiichi Sankyo. University of Leuven, Leuven,
Belgium (T Vanassche MD,
Introduction especially in patients with more severe pulmonary P Verhamme MD); Department
of Medicine, University of
Venous thromboembolism is a common disorder that embolism.3 Right ventricular dysfunction identies
Ottawa, Ottawa Hospital
includes deep vein thrombosis and pulmonary embolism. patients with pulmonary embolism at higher risk for Research Institute, Ottawa, ON,
The direct oral anticoagulants are as eective as warfarin early complications.4,5 Indicators of right ventricular Canada (Prof P S Wells MD); and
for venous thromboembolism treatment and are dysfunction include raised concentrations of N-terminal Thrombosis and Atherosclerosis
Research Institute and
associated with a lower risk of bleeding.1,2 Although direct pro-brain natriuretic peptide (NT-proBNP) and right
McMaster University,
oral anticoagulants are widely used for deep vein ventricular dilatation as determined by an increased right Hamilton, ON, Canada
thrombosis treatment, there remains some reluctance to ventricular to left ventricular diameter ratio on CT images (Prof J I Weitz MD)
use them for treatment of pulmonary embolism, or echocardiography.6,7

www.thelancet.com/haematology Vol 3 September 2016 e437

 Articles

Correspondence to:
Dr Marjolein P A Brekelmans,
Department of Vascular
Medicine, Academic Medical
Center, 1105 AZ Amsterdam,
Netherlands
m.p.brekelmans@amc.nl
Research in context
Evidence before this study
Direct oral anticoagulants are as eective as and safer than
warfarin in the treatment and prevention of venous
thromboembolism (Van Es and colleagues). However,
physicians remain reluctant to prescribe direct oral
anticoagulants in patients with more extensive pulmonary
embolism. The Hokusai-VTE study was the rst to prospectively
assess two indicators of right ventricular dysfunction, a marker
of a worse prognosis, in patients with pulmonary embolism.
The ndings showed that recurrent rates of venous
thromboembolism were lower with edoxaban compared with
warfarin in these patients (Buller and colleagues).
Added value of this study
The results of the current study provide reassurance that a
regimen of heparin followed by edoxaban can be used to treat
all patients with pulmonary embolism eligible for
anticoagulation therapy. Additionally, edoxaban was superior to
warfarin for the prevention of recurrence in those with right
ventricular dysfunction, which probably reects the more
consistent antithrombotic eect of edoxaban.
Implications of all the available evidence
Patients with pulmonary embolism and evidence of right
ventricular dysfunction who were given edoxaban had a low rate
of recurrent venous thromboembolism. Edoxaban is a more
convenient, eective, and patient-friendly anticoagulant agent
than warfarin in the treatment and prevention of recurrent
venous thromboembolism in pulmonary embolism patients
with evidence of right ventricular dysfunction. Conrmation of
these ndings in the other published direct oral anticoagulant
trials in patients with pulmonary embolism is warranted.

After initial treatment with heparin, the Hokusai-VTE
study compared edoxaban with warfarin in a wide range
of patients with venous thromboembolism.8 The study
assessed two indicators of right ventricular dysfunction
in the pulmonary embolism patients and therefore was
the rst to provide the opportunity to examine the ecacy
of a direct oral anticoagulant in patients with pulmonary
embolism and evidence of right ventricular dysfunction.
The rate of recurrent venous thromboembolism was
lower with edoxaban than with warfarin in these
patients.8 The current study was undertaken to rst
conrm this nding in the complete set of patients and
second to investigate in depth potential explanations for
the recorded dierence.
Methods
Study design and participants
As previously described, the Hokusai-VTE study was
a large, international, multicentre, randomised trial
comparing edoxaban with warfarin in 8292 patients aged
18 years or older with a diagnosed acute symptomatic deep
vein thrombosis or pulmonary embolism (NCT00986154).8
A coordinating committee in collaboration with the funder
had responsibility for study design, protocol, and oversight.
An independent committee, unaware of study group
assignment, adjudicated all suspected outcomes. The
institutional review board at each centre approved the
protocol. All patients provided written informed consent.
Patients qualifying for thrombolytic treatment were
excluded. The full list of exclusion criteria is provided in
the original publication.8
In this subgroup analysis, we focused on patients with a
pulmonary embolism. Pulmonary embolism was dened
as symptoms of pulmonary embolism with one of the
following ndings: an intraluminal lling defect in (sub)
segmental or more proximal branches on spiral CT scan,
an intraluminal lling defect or a sudden cuto of vessels
more than 25 mm in diameter on the pulmonary
angiogram, a perfusion defect of at least 75% of a segment
with a local normal ventilation result (high probability) on
ventilation/perfusion lung scintigraphy, or a non-diagnostic
lung scan accompanied by documentation of deep vein
thrombosis by ultrasonography or venography.8
Randomisation and masking
Patients were randomly assigned (1:1) to receive either
edoxaban or warfarin using a double-blind, double-
dummy method (ie, patients receiving active warfarin
also received dummy edoxaban, and patients receiving
active edoxaban also received dummy warfarin). The
local site study physician or study coordinator did the
randomisation using an interactive web-based system,
with stratication according to the qualifying diagnosis
(deep vein thrombosis or pulmonary embolism),
presence or absence of temporary risk factors, and the
dose of edoxaban. The system then directed which
treatment kit the patient was to receive. All investigators,
coordinators, patient caregivers, and patients were
masked to treatment.
Procedures
All patients received initial treatment with enoxaparin
or unfractionated heparin for at least 5 days and were
then given either edoxaban or warfarin for a minimum
of 3 months and a maximum of 12 months. Edoxaban
(or placebo) was started after discontinuation of initial
heparin treatment. The dose of edoxaban was 60 mg
once a day, which was reduced to 30 mg in patients with
a creatinine clearance of 3050 mL per min, a
bodyweight of 60 kg or lower, or in those receiving
concomitant P-glycoprotein inhibitors (verapamil or
quinidine). Warfarin (or placebo) was started con-
currently with enoxaparin or unfractionated heparin
and the target international normalised ratio (INR) was
between 20 and 30. INR measurement was done with
a point-of-care device that provided the actual INR for

e438 www.thelancet.com/haematology Vol 3 September 2016


patients receiving warfarin and a sham INR value for
patients given edoxaban. INR measurement was done at
least once a month. Patients were followed up for
12 months.
The protocol of the Hokusai-VTE study prespecied a
subgroup comparison of edoxaban with warfarin in
patients with pulmonary embolism and indicators
of right ventricular dysfunction at baseline.8,9 Right
ventricular dysfunction was dened as a concentration of
NT-proBNP of 500 pg/mL or higher10 or CT evidence of
right ventricular dilatation based on a reformatted
chamber view right ventricular to left ventricular
diameter ratio of 09 or higher.11,12 We chose NT-proBNP
because it reects right ventricular volume and function
in patients with pulmonary embolism.13 Because the
most recent pulmonary embolism guidelines from the
European Society of Cardiology suggest a cuto for
NT-proBNP at 600 pg/mL or higher, we also did an
analysis using this threshold.4,7
We measured serum NT-proBNP concentrations in
samples collected at enrolment. Assays were done
centrally in Quintiles Laboratories (Marietta, GA, USA)
with the Elesys NT-proBNP electrochemiluminescence
kit on the Roche Cobas e411 platform. The right
ventricular to left ventricular diameter ratio was assessed
on all available digital CT images by an independent
expert radiologist masked to treatment allocation and
patient data. To obtain reformatted chamber views,12
images were imported in an image workstation with
multiplanar reformatting features using commercially
available software (eFilm Workstation for Windows
[version 3.4.0], Build 10, Merge Technologies, Milwaukee,
Wisconsin, USA). If ventilation/perfusion scanning or
direct pulmonary angiography only was used for the
diagnosis or no digital CT images could be processed in
the viewer, right ventricular to left ventricular diameter
ratio could not be assessed.
At the time of original publication,8 175 NT-proBNP
samples were not available for analysis because of
delayed overseas shipment. We included these samples
in the current subgroup analysis. Furthermore, right
ventricular to left ventricular diameter measurements
were calculated in 1002 of the patients in the original
publication,8 whereas an additional 881 measurements
were available for the present study. The dierence is
explained by the time-consuming analysis needed for
the assessment of right ventricular to left ventricular
diameter ratios.
Data for clinical parameters at moment of presentation
with pulmonary embolism were not available because of
a delay between presentation and randomisation in the
study. Therefore, scores on the simplied pulmonary
embolism severity index (sPESI) could not be calculated.
Outcomes
The primary ecacy outcome was the incidence of
symptomatic recurrent venous thromboembolism
www.thelancet.com/haematology Vol 3 September 2016
Articles
dened as a composite of deep vein thrombosis or
non-fatal or fatal pulmonary embolism at 12 months.8
The outcome was assessed by a central adjudication
committee whose members were masked to study
group assignments and evidence of right ventricular
dysfunction at inclusion. In the current analysis, we
made a further distinction between fatal pulmonary
embolism, non-fatal pulmonary embolism, and deep
vein thrombosis alone.
When a dierence was recorded in the rate of recurrent
venous thromboembolism between edoxaban and
warfarin recipients, several factors were investigated to
provide insight into potential explanations for this
dierence. The examined factors included baseline
clinical characteristics, the duration of initial heparin
therapy, the rates of bleeding that led to study drug
discontinuation, and the recurrence rates on-treatment
and o-treatment. Additionally, we compared the quality
of warfarin treatment dened as percentage of time
of INR below, in, or above the therapeutic range
in pulmonary embolism patients with NT-proBNP
concentrations of 500 pg/mL or higher with that in those
with lower concentrations. For these latter three
comparisons, we chose a priori an observation period of
the rst 120 days to better focus on early divergence in
recurrence rates in the two groups.
Statistical analysis
The Hokusai-VTE study was designed as an event-driven
trial to test the hypothesis that edoxaban would be
non-inferior to warfarin for the primary ecacy
outcome, with an upper limit of the condence interval
for the hazard ratio of 15 and a two-sided alpha level
of 005. This margin corresponds to retention of at
least 70% of the treatment eect of warfarin. Assuming
8292 patients enrolled
3343 patients with pulmonary embolism
randomly assigned
1663 assigned to receive heparinedoxaban 1680 assigned to receive heparinwarfarin
13 did not receive heparinedoxaban 11 did not receive heparinwarfarin
1650 included in modified intention-to-treat 1669 included in modified intention-to-treat
and safety analyses and safety analyses
78 did not complete the overall study 73 did not complete the overall study
period period
53 died 53 died
20 withdrew consent 17 withdrew consent
0 lost to follow-up 1 lost to follow-up
5 other reasons 2 other reasons
Figure 1: Trial prole
e439
 Articles

equal ecacy of edoxaban and warfarin, we estimated expected to be accrued, the date for concluding the
that 220 events would need to happen for the study to study was set (study closure), such that the last patient
have 85% power to show the non-inferiority of randomly assigned would complete 6 months of study
edoxaban. When the targeted number of events was treatment and follow up. Assuming a 3% incidence of
the primary ecacy outcome, we expected to enrol
at least 7500 patients. The statistical analysis plan
Edoxaban group Warfarin group HR (95% CI)
(n=1565) (n=1599) prespecied subgroup analysis of the patients with
pulmonary embolism.
NT-proBNP concentration 500 pg/mL 465 (30%) 507 (32%)
All ecacy analyses were done with the modied
Recurrent venous thromboembolism 14/465 (3%) 30/507 (6%) 050 (026094), p=0033
intention-to-treat population, dened as all randomly
Fatal pulmonary embolism 4 13
assigned patients who received at least one dose of the
Non-fatal pulmonary embolism 8 13
study drug. The primary ecacy point was analysed
Deep vein thrombosis only 2 4
using a Cox proportional hazards model with
NT-proBNP concentration <500 pg/mL 1100 (70%) 1092 (68%)
dose-adjustment (30 mg vs 60 mg) and risk factor
Recurrent venous thromboembolism 30/1100 (3%) 33/1092 (3%) 089 (05415), p=065
(temporary vs other) as covariates for all patients who
Fatal pulmonary embolism 4 2
had taken at least one dose of study drug and with
Non-fatal pulmonary embolism 16 16
adjudication committee conrmed index pulmonary
Deep vein thrombosis only 10 15 embolism. Patients without recurrent venous thrombo-
Data are n (%). HR=hazard ratio. NT-proBNP=N-terminal pro-brain natriuretic peptide. embolism were censored. We calculated time-to-event
curves with the Kaplan-Meier method. We used SAS
Table 1: Rates of recurrent venous thromboembolism in patients with pulmonary embolism as a
(version 9.3) for all analyses. This study is registered
function of NT-proBNP concentrations
with ClinicalTrials.gov, number NCT00986154.

100 Heparinedoxaban
Heparinwarfarin 8

90
6

80

70
Recurrent venous thromboembolism (%)

2
60

0
50 0 30 60 90 120 150 180 210 240 270 300 330 360

40

30

20

10

0
0 30 60 90 120 150 180 210 240 270 300 330 360
Days from randomisation
Number at risk
Heparinedoxaban 465 452 446 442 441 439 435 430 405 384 367 349 329
Heparinwarfarin 507 494 487 483 482 480 478 469 445 424 405 381 347

Figure 2: Kaplan-Meier cumulative rates of recurrent venous thromboembolism in pulmonary embolism patients with NT-proBNP concentrations 500 pg/mL

e440 www.thelancet.com/haematology Vol 3 September 2016

 Articles

Role of the funding source the study and had nal responsibility for the decision to
Daichii Sankyo provided nancial support for the study. submit for publication. All authors had access to the
A coordinating committee in collaboration with the primary clinical trial data, vouch for the accuracy of
funder was responsible for the design and oversight of the analyses, and participated in the writing of the
the study. The funder was responsible for the collection, manuscript.
maintenance, and analysis of data. The members of the
writing committee, including employees of the funder, Results
interpreted the data and prepared, reviewed, and approved Between Jan 28, 2010, and Oct 5, 2012, 8292 patients were
the manuscript; the funder was not involved in the enrolled in the Hokusai-VTE study in 439 centres in
decision to submit the manuscript for publication. 37 countries, of whom 8240 received at least one dose of
The corresponding author had full access to all the data in study drug. Of these 8240 patients, 3319 (40%) with
symptomatic pulmonary embolism were included in the
Edoxaban Warfarin
current prespecied subgroup analysis (gure 1). After an
group (N=507) initial course of enoxaparin or unfractionated heparin,
(N=465) 1650 (50%) of 3319 patients were given edoxaban and
Age (years) 65 (15) 64 (15) 1669 (50%) of 3319 patients received warfarin. The
Men 209 (45%) 233 (46%) median duration of treatment was 80 months
Women 256 (55%) 274 (54%) (IQR 60120) in the edoxaban group and 73 months
Unprovoked 309 (67%) 337 (67%) (60120) in the warfarin group. Baseline NT-proBNP
Concomitant deep vein thrombosis 99 (21%) 124 (25%) concentrations could be measured in 3164 (95%) of
Active cancer 16 (3%) 12 (2%) 3319 patients, whereas the right ventricular to left
Previous venous thromboembolism 97 (21%) 95 (19%)
ventricular diameter ratio could be calculated in
Known thrombophilia 4 (1%) 18 (4%)
1883 (57%) of 3319 patients. This represents an additional
Anatomical extent of pulmonary embolism
175 patients with baseline NT-proBNP concentrations and
Limited or intermediate 158 (34%) 167 (33%)
881 additional patients with right ventricular to left
ventricular diameter ratio measurements than were
Extensive 307 (66%) 340 (67%)
reported in the primary publication (in which
Cardiovascular disease 119 (26%) 143 (28%)
2989 NT-proBNP and 1002 right ventricular to left
Atrial brillation 82 (18%) 73 (14%)
ventricular diameter measurements were reported).8
Heart failure 29 (6%) 31 (6%)
The NT-proBNP concentration was 500 pg/mL or
Pulmonary disease 112 (24%) 151 (30%)
higher in 465 (30%) of 1565 edoxaban recipients and in
Bodyweight 60 kg or creatinine clearance 123 (27%) 123 (24%)
50 mL/min or P-glycoprotein inhibitor use
507 (32%) of 1599 patients given warfarin. The right
Concomitant aspirin use 58 (13%) 59 (12%)
ventricular to left ventricular diameter ratio was 09 or
Duration of initial heparin therapy
higher in 414 (44%) of 937 and 427 (45%) of 946 patients
Days 7 (69) 7 (69)
in the edoxaban and warfarin groups, respectively. Of the
1817 patients in whom both indicators of right ventricular
Bleeding and study drug discontinuation
Major bleeding* 13 (3%) 13 (3%)
Bleeding leading to study drug 11 (2%) 11 (2%)
discontinuation* Edoxaban Warfarin HR (95% CI)
group (n=937) group
All discontinuations of study drug* 106 (23%) 114 (23%)
(n=946)
Recurrent venous thromboembolism in the rst 120 days
Right ventricular to left ventricular diameter 414 (44%) 427 (45%)
Recurrent venous thromboembolism 5 (1%) 15 (3%)
ratio 09
overall
Recurrent venous thromboembolism 11/414 (3%) 20/427 (5%) 057 (027117), p=013
Fatal pulmonary embolism 3 7
Fatal pulmonary embolism 1 6
Non-fatal pulmonary embolism 2 6
Non-fatal pulmonary embolism 7 7
Deep vein thrombosis only 0 2
Deep vein thrombosis only 3 7
Recurrent venous thromboembolism 5 (1%) 10 (2%)
on-treatment Right ventricular to left ventricular diameter 523 (5%) 519 (55%)
ratio <09
Fatal pulmonary embolism 3 3
Recurrent venous thromboembolism 12/523 (2%) 16/519 (3%) 075 (035157), p=044
Non-fatal pulmonary embolism 2 5
Fatal pulmonary embolism 2 2
Deep vein thrombosis only 0 2
Non-fatal pulmonary embolism 6 7
Data are mean (SD), n (%), median (IQR). *In the rst 120 days. Deep vein thrombosis only 4 7

Table 2: Comparison of baseline characteristics, duration of initial Data are n (%). HR=hazard ratio.
heparin therapy, and rates of bleeding, study drug discontinuation, and
recurrent venous thromboembolism in patients with pulmonary Table 3: Rates of recurrent venous thromboembolism in pulmonary embolism patients as a function of
embolism with NT-proBNP concentration 500 pg/mL right ventricular to left ventricular diameter ratios

www.thelancet.com/haematology Vol 3 September 2016 e441

 Articles

100 Heparinedoxaban
Heparinwarfarin 8

90
6

80

70
Recurrent venous thromboembolism (%)

2
60

0
50 0 30 60 90 120 150 180 210 240 270 300 330 360

40

30

20

10

0
0 30 60 90 120 150 180 210 240 270 300 330 360
Days from randomisation
Number at risk
Heparinedoxaban 414 408 406 404 404 404 403 398 378 360 345 329 302
Heparinwarfarin 427 419 413 412 410 407 406 403 381 367 355 334 308

Figure 3: Kaplan-Meier cumulative rates of recurrent venous thromboembolism in pulmonary embolism patients with right ventricular to left ventricular
diameter ratio 09

dysfunction were available, 192 (21%) patients in the
edoxaban group and 207 (23%) patients in the warfarin
group had both an NT-proBNP concentration of
500 pg/mL or higher and a right ventricular to left
ventricular diameter ratio of 09 or higher.
In patients with pulmonary embolism with NT-proBNP
concentrations of 500 pg/mL or higher, the median
concentration was 1405 pg/mL (IQR 8282776) in the
edoxaban group and 1686 pg/mL (9373152) in the
warfarin group. In those patients, we recorded recurrent
venous thromboembolism in 14 (3%) of 465 patients in
the edoxaban group and 30 (6%) of 507 in the warfarin
group (HR 050, 95% CI 026094; p=0033; table 1).
The dierence in risk between edoxaban and warfarin
was 29% (95% CI 55 to 03). The number needed
to treat (NNT) is 34. Hence, 34 patients need to be treated
with edoxaban to prevent one recurrent venous
thromboembolism event compared to treatment with
warfarin. In contrast, in patients with NT-proBNP levels
below 500 pg/mL, recurrent venous thromboembolism
occurred in 30 (3%) of 1100 and 33 (3%) of 1092 patients
given edoxaban and warfarin, respectively. Figure 2
shows Kaplan-Meier cumulative recurrent venous
thromboembolism rates for patients with NT-proBNP
concentrations of 500 pg/mL or higher.
When the NT-proBNP cuto was set at 600 pg/mL or
higher, we recorded recurrent venous thromboembolism
in 14 (3%) of 429 patients given edoxaban and 28 (6%) of
464 given warfarin (HR 054, 95% CI 028102;
p=0056).
Investigation into potential explanations for the
dierence in outcomes in pulmonary embolism patients
with NT-proBNP concentrations of 500 pg/mL or higher
showed that baseline characteristics were similar in
patients in both the edoxaban and warfarin treatment
groups (table 2). Additionally, we noted no dierences in
durations of initial heparin treatment and rates of
bleeding leading to study drug discontinuation in the rst
120 days between the groups. Furthermore, in the rst
120 days of treatment, all recurrences in the edoxaban
group and ten (67%) of the 15 recurrences in the warfarin
group happened on-treatment. Although prevalence of

e442 www.thelancet.com/haematology Vol 3 September 2016


thrombophilia diered between the edoxaban and
warfarin groups, no recurrent venous thromboembolism
events were recorded in patients with a known
thrombophilia. Of the ve recurrences in the warfarin
group that occurred o-treatment, two patients
discontinued treatment shortly after suering a clinically
relevant non-major bleeding event. The quality of
warfarin treatment in pulmonary embolism patients with
NT-proBNP concentrations of 500 pg/mL or higher was
similar to that in patients with lower NT-proBNP
concentrations. The INR was in the therapeutic range
for 60% of the time in patients with NT-proBNP
concentrations of 500 pg/mL or higher, and for 62% of
the time in patients with NT-proBNP levels below
500 pg/mL, respectively.
In pulmonary embolism patients with NT-proBNP
concentrations of 500 pg/mL or higher, all-cause
mortality at 30 days was recorded in nine (2%) of
465 patients in the edoxaban group and in seven (1%) of
507 warfarin recipients. All-cause mortality at 365 days
was recorded in 29 (6%) of 465 patients in the edoxaban
group and 32 (6%) of 507 patients in the warfarin group.
In pulmonary embolism patients with a right
ventricular to left ventricular diameter ratio of 09 or
higher, recurrent venous thromboembolism occurred in
11 (3%) of 414 patients in the edoxaban group and
20 (5%) of 427 in the warfarin group (HR 057, 95% CI
027 to 117; p=013; table 3). The corresponding
number needed to treat is 49ie, 49 patients need to be
given edoxaban to prevent one recurrent venous
thromboembolism event compared with treatment with
warfarin. In those with right ventricular to left ventricular
diameter ratios lower than 09, recurrent venous
thromboembolism was recorded in 12 (2%) of 523 and
16 (3%) of 519 patients in the edoxaban and warfarin
groups, respectively. Figure 3 shows the Kaplan-Meier
cumulative rates of recurrent venous thromboembolism
for those with a right ventricular to left ventricular
diameter ratio of 09 or higher.
In pulmonary embolism patients with right ventricular
to left ventricular diameter ratios of 09 or higher,
all-cause mortality at 30 days was recorded in three (1%)
of 414 patients in the edoxaban group and in three (1%)
of 427 warfarin recipients. All-cause mortality at 365 days
was recorded in 11 (3%) of 414 patients in the edoxaban
group and 16 (4%) of 427 in the warfarin group.
In the 399 patients with NT-proBNP concentrations of
500 pg/mL or higher and right ventricular to left
ventricular diameter ratio of 09 or higher, recurrent
venous thromboembolism occurred in four (2%) of
192 and in ten (5%) of 207 patients in the edoxaban and
warfarin group, respectively (HR 044, 95% CI 014136;
p=017; table 4). The appendix shows Kaplan-Meier
cumulative rates of recurrent venous thromboembolism
for those with NT-proBNP concentrations of 500 pg/mL
or higher and a right ventricular to left ventricular
diameter ratio of 09 or higher.
www.thelancet.com/haematology Vol 3 September 2016
Articles
Edoxaban Warfarin HR (95% CI)
group (n=908) group (n=909)
NT-proBNP 500 pg/mL and right 192 (21%) 207 (23%)
ventricular to left ventricular diameter 09
Recurrent venous thromboembolism 4/192 (2%) 10/207 (5%) 044 (014136), p=015
Fatal pulmonary embolism 1 5
Non-fatal pulmonary embolism 2 3
Deep vein thrombosis only 1 2
NT-proBNP <500 pg/mL and right 435 (48%) 424 (47%)
ventricular to left ventricular diameter <09
Recurrent venous thromboembolism 8/435 (2%) 13/424 (3%) 061 (025145), p=026
Fatal pulmonary embolism 1 1
Non-fatal pulmonary embolism 4 5
Deep vein thrombosis only 3 7
Data are n (%). HR=hazard ratio. NT-proBNP=N-terminal pro-brain natriuretic peptide.
Table 4: Rates of recurrent venous thromboembolism in pulmonary embolism patients as a function of
NT-proBNP concentrations and right ventricular to left ventricular diameter ratios
Discussion
Our ndings show that a regimen of heparin followed
by edoxaban is more eective than heparin overlapped
with and followed by warfarin in the prevention of
recurrent venous thromboembolism in pulmonary
embolism patients with evidence of right ventricular
dysfunction, most notably with NT-proBNP concen-
trations of 500 pg/mL or higher. The time-to-event
curves diverge early, which suggests that the advantage
of edoxaban over warfarin is established in the rst
weeks and is maintained for the complete study
duration. The advantage of edoxaban does not seem to
be explained by an imbalance in baseline characteristics
relevant to the risk of recurrent venous thrombo-
embolism, poor quality of warfarin treatment, or
dierences in the duration of initial heparin treatment,
or the rates of bleeding leading to study drug
discontinuation. Therefore, the most likely explanation
is that in these patients with evidence of right
ventricular dysfunction, edoxaban provides a more
consistent antithrombotic eect than does warfarin. In
other words, edoxaban has a stable pharmacokinetic
and pharmacodynamic prole that allows for a
predictable and constant level of anticoagulation. By
contrast, the intrinsic characteristics of warfarin might
result in periods of overtreatment and undertreatment,
leading to a less stable anticoagulant eect over time.
Right ventricular dysfunction in patients with a
pulmonary embolism has been shown to predict mortality
at hospital discharge and at 30 days.4,5,14 Additionally,
persistent right ventricular dysfunction predicts recurrent
venous thromboembolism,15 and raised NT-proBNP
plasma concentrations are associated with long-term risk See Online for appendix
of venous thromboembolism recurrence.16 To our
knowledge, this is the rst time a study has shown that
right ventricular dysfunction at baseline also predicts
short-term and long-term risk of fatal and non-fatal
e443
 Articles
recurrent venous thromboembolism, especially when
right ventricular dysfunction is dened by NT-proBNP
concentrations. NT-proBNP concentration cutos of
500 or 600 pg/mL provide directionally similar results.
In comparison to the data from the Hokusai-VTE
study, the EINSTEIN and AMPLIFY studies,17,18 which
assessed rivaroxaban and apixaban, respectively, had a
signicantly lower proportion of patients with extensive
pulmonary embolism and did not prospectively assess
markers of right ventricular dysfunction and the
relation with recurrent venous thromboembolism.
Conrmation of our ndings in these trials would be of
great interest.
Strengths of the present study include the large sample
size, the double blind design, the a priori denition of
right ventricular dysfunction, the use of two indicators
of right ventricular dysfunction, the measurement of
NT-proBNP in a reference laboratory, the long-term
follow-up, and the masked adjudication of ecacy and
safety outcomes. However, our study had several
limitations. Although we used two indicators of
right ventricular dysfunction, measurements were not
available for all pulmonary embolism patients.
NT-proBNP data were available in 95% of all pulmonary
embolism patients. In the other patients, blood was not
drawn, was collected in inappropriate tubes, or no
reliable measurement was obtained. It is unlikely that
the missing NT-proBNP data would have aected our
ndings. Right ventricular to left ventricular diameter
ratios could only be determined in 57% of the patients
because the calculation depended on the availability of
sucient-quality reformatted CT images. Furthermore,
the protocol allowed for other modalities to establish the
diagnosis of pulmonary embolism. Nonetheless, both
indicators adequately identied high-risk patients who
beneted from edoxaban treatment. Echocardiography
was not done and troponin concentrations were not
measured; therefore, the utility of these tests in this
setting remains unknown.
Additionally, we were unable to risk stratify patients
using the sPESI score because the Hokusai-VTE study
included a delay between moment of presentation and
randomisation. Consequently, we do not have any
information available for the moment of presentation
because informed consent was not yet obtained.
Furthermore, it needs to be realised that we used
NT-proBNP concentration or right ventricular to left
ventricular diameter ratio as a simple, crude marker of
right ventricular dysfunction and did not investigate
whether other (baseline) characteristics aected the
predictive signicance of these tests. Right ventricular
dysfunction might not be the only predictor or
recurrence. Finally, it is well known that NT-proBNP
concentrations can be increased by other diseases such
as inammatory processes, ischaemia, heart failure, or
pulmonary diseases.19 Hence, we are not able to rule out
an eect of these diseases on the recorded associations.
e444
In conclusion, the results of this study provide
reassurance that a regimen of heparin followed by
edoxaban can be used to treat all patients with pulmonary
embolism eligible for anticoagulation therapy. Further-
more, the data suggest that edoxaban is a more
convenient and eective anticoagulant than warfarin in
the treatment and prevention of recurrence in stable
pulmonary embolism patients with evidence of right
ventricular dysfunction.
Contributors
MPAB, HRB, GM, GZ, and JIW were responsible for conception and
design, analysis and interpretation of data, and drafting of the report.
WA, LFB, BB, CZC, ATC, MAG, GR, AS, TV, PV, and PSW contributed
to revising the report critically for important intellectual content and
provided nal approval of the report submitted.
Declaration of interests
MPAB reports a travel grant from Daiichi Sankyo. WA reports grants,
personal fees, and non-nancial support from Bayer, Boehringer
Ingelheim, Daiichi Sankyo, and Stago; and personal fees from
Bristol-Myers Squibb, Pzer, and Ono, outside the submitted work.
LFB reports grants from Daiichi Sankyo during the conduct of the study.
BB reports personal fees from Bayer, Daiichi Sankyo, Pzer, ROVI
Laboratories, and Sano outside the submitted work. HRB reports
grants and personal fees from Bayer, Boehringer Ingelheim,
Bristol-Myers Squibb, Daiichi Sankyo, GlaxoSmithKline, Isis
Pharmaceuticals, Pzer, Roche, Sano, and Thrombogenics outside the
submitted work. CZC is an employee of Daiichi-Sankyo. ATC reports
grants from Daiichi-Sankyo Pharma Development during the conduct of
the study; grants and personal fees from Bayer, Bristol-Myers Squibb,
and Pzer; and personal fees from Boehringer Ingelheim, Janssen,
Johnson & Johnson, Portola, Sano, and XO1 outside the submitted
work. MAG is an employee of Daiichi-Sankyo Pharma Development.
GM reports grants and non-nancial support from Daiichi Sankyo
Pharma Development during the conduct of the study; grants and
non-nancial support from Bayer, Boehringer Ingelheim, and
Leo Pharma, and non-nancial support from Bristol-Myers Squibb and
Pzer outside the submitted work. GR reports personal fees from
Daiichi-Sankyo and Itreas during the conduct of the study; and personal
fees from Bayer, Boehringer Ingelheim, Bristol-Myers Squibb, Eli Lilly,
Isis Pharmaceuticals, Janssen, Pzer, and Portola outside the submitted
work. AS reports grants from Daiichi-Sankyo Pharma Development
during the conduct of the study; and grants from Isis Pharmaceuticals
outside the submitted work. TV reports grants from Daiichi-Sankyo
Pharma Development during the conduct of the study. PV reports grants
and personal fees from Daiichi-Sankyo Pharma Development during the
conduct of the study; and grants from Leo Pharma; grants and personal
fees from Boehringer Ingelheim, Sano, and ThromboGenics, and
personal fees from Bayer outside the submitted work. PSW reports
personal fees from Bayer and Daiichi Sankyo; and grants from
Bristol-Myers Squibb and Pzer outside the submitted work. GZ is an
employee of Daiichi-Sankyo Pharma Development. JIW reports personal
fees from Bayer, Boehringer Ingelheim, Bristol-Myers Squibb, Daiichi
Sankyo, Isis Pharmaceuticals, Janssen, Pzer, and Portola outside the
submitted work.
Acknowledgments
We thank Min Lin from Daiichi Sankyo Pharma Development, Edison, NJ,
for assistance in statistical analyses and Shannon Winters, Daiichi Sankyo,
Parsippany, NJ, for editorial assistance in the manuscript preparation.
References
1 van Es N, Coppens M, Schulman S, Middeldorp S, Buller HR.
Direct oral anticoagulants compared with vitamin K antagonists for
acute symptomatic venous thromboembolism: evidence from
phase 3 trials. Blood 2014; 124: 196875.
2 van der Hulle T, Kooiman J, den Exter PL, Dekkers OM, Klok F,
Huisman MV. Eectiveness and safety of novel oral anticoagulants
as compared with vitamin K antagonists in the treatment of acute
symptomatic venous thromboembolism: a systematic review and
meta-analysis. J Thromb Haemost 2014; 12: 32028.
www.thelancet.com/haematology Vol 3 September 2016

3 Rudd KM, Phillips EL. New oral anticoagulants in the treatment of
pulmonary embolism: ecacy, bleeding risk, and monitoring.
Thrombosis 2013; 2013: 973710.
4 Lankeit M, Jimenez D, Kostrubiec M, et al. Validation of N-terminal
pro-brain natriuretic peptide cut-o values for risk stratication of
pulmonary embolism. Eur Respir J 2014; 43: 166977.
5 Jimenez D, Kopecna D, Tapson V, et al. Derivation and validation of
multimarker prognostication for normotensive patients with acute
symptomatic pulmonary embolism. Am J Respir Crit Care Med 2014;
189: 71826.
6 Klok FA, Mos IC, Huisman MV. Brain-type natriuretic peptide
levels in the prediction of adverse outcome in patients with
pulmonary embolism: a systematic review and meta-analysis.
Am J Respir Crit Care Med 2008; 178: 42530.
7 Konstantinides SV, Torbicki A, Agnelli G, et al. 2014 ESC guidelines
on the diagnosis and management of acute pulmonary embolism.
Eur Heart J 2014; 35: 303369.
8 Bller HR, Dcousus H, Grosso M, et al. Edoxaban versus warfarin
for the treatment of symptomatic venous thromboembolism.
N Engl J Med 2013; 369: 140615.
9 Raskob G, Buller H, Prins M, et al. Edoxaban for the long-term
treatment of venous thromboembolism: rationale and design of the
Hokusai-venous thromboembolism study: methodological
implications for clinical trials. J Thromb Haemost 2013; 11: 128794.
10 Kucher N, Printzen G, Goldhaber SZ. Prognostic role of brain
natriuretic peptide in acute pulmonary embolism. Circulation 2003;
107: 254547.
11 Furlan A, Aghayev A, Chang CC, et al. Short-term mortality in acute
pulmonary embolism: clot burden and signs of right heart
dysfunction at CT pulmonary angiography. Radiology 2012;
265: 28393.
www.thelancet.com/haematology Vol 3 September 2016
Articles
12 Lu MT, Demehri S, Cai T, et al. Axial and reformatted four-chamber
right ventricle-to-left ventricle diameter ratios on pulmonary CT
angiography as predictors of death after acute pulmonary
embolism. AJR Am J Roentgenol 2012; 198: 135360.
13 Pasha SM, Klok FA, van der Bijl N, de Roos A, Kroft LJ,
Huisman MV. NT-pro-BNP levels in patients with acute pulmonary
embolism are correlated to right but not left ventricular volume and
function. Thromb Haemost 2012; 108: 36772.
14 Sanchez O, Trinquart L, Caille V, et al. Prognostic factors for
pulmonary embolism: the prep study, a prospective multicenter
cohort study. Am J Respir Crit Care Med 2010; 181: 16873.
15 Grifoni S, Vanni S, Magazzini S, et al. Association of persistent
right ventricular dysfunction at hospital discharge after acute
pulmonary embolism with recurrent thromboembolic events.
Arch Intern Med 2006; 166: 215156.
16 Wang Y, Liu ZH, Zhang HL, Luo Q, Zhao ZH, Zhao Q.
Association of elevated NTproBNP with recurrent thromboembolic
events after acute pulmonary embolism. Thromb Res 2012;
129: 68892.
17 Bller HR, Prins MH, Lensin AW, et al. Oral rivaroxaban for the
treatment of symptomatic pulmonary embolism. N Engl J Med 2012;
366: 128797.
18 Agnelli G, Buller HR, Cohen A, et al. Oral apixaban for the
treatment of acute venous thromboembolism. N Engl J Med 2013;
369: 799808.
19 Sanchez O, Trinquart L, Colombet I, et al. Prognostic value of right
ventricular dysfunction in patients with haemodynamically stable
pulmonary embolism: a systematic review. Eur Heart J 2008;
29: 156977.
e445