Research

International Journal of Stroke

0(0) 1–12
Diffusion-weighted imaging or ! 2017 World Stroke Organization
Reprints and permissions:
computerized tomography perfusion sagepub.co.uk/journalsPermissions.nav
DOI: 10.1177/1747493017710341

assessment with clinical mismatch in the journals.sagepub.com/home/wso

triage of wake up and late presenting

strokes undergoing neurointervention
with Trevo (DAWN) trial methods

Tudor G Jovin1, Jeffrey L Saver2, Marc Ribo3, Vitor Perreira4,

Anthony Furlan5, Alain Bonafe6, Blaise Baxter7, Rishi Gupta8,
Demetrius Lopes9, Olav Jansen10, Wade Smith11, Darryl Gress12,
Steven Hetts13, Roger J Lewis14, Ryan Shields15, Scott M Berry16,
Todd L Graves16, Tim Malisch17, Ansaar Rai18, Kevin N Sheth19,
David S Liebeskind2 and Raul G Nogueira20

Abstract
Rationale: Efficacy of mechanical thrombectomy for acute stroke due to large vessel occlusion initiated bfeyond 6 h of
time last seen well has not been demonstrated in randomized trials.
Aim: To establish whether subjects considered to have substantial areas of salvageable brain based on age-adjusted
clinical core mismatch who can undergo endovascular treatment within 6–24 h from time last seen (TLSW) well have
better outcomes at three months compared to subjects treated with standard medical therapy alone. Age-adjusted
clinical core mismatch is defined by age (80 or >80 years), baseline National Institutes of Health Stroke Scale (NIHSS)
(10–20 or 21), and core size (0–20 cm3 in subjects older than 80 and, in subjects younger than 80, 0–30 cm3 with
NIHSS 10–20 and 31–50 cm3 with NIHSS 21).
Design: Prospective, randomized, multicenter, Bayesian adaptive-enrichment, open label trial with blinded endpoint
assessment. For the purpose of enrolment, ischemic core size will be evaluated by CT perfusion or magnetic resonance
imaging-diffusion-weighted imaging measured by automated software (RAPID).
Procedures: Subjects with acute ischemic stroke due to computed tomography angiography- or magnetic resonance
angiogram-proven arterial occlusion of the intracranial internal carotid and/or proximal middle cerebral artery (M1) with
age-adjusted clinical core mismatch in whom treatment can be initiated between 6 and 24 h from TSLW are randomized
in a 1:1 ratio to receive mechanical embolectomy with the Trevo device or medical management alone. Sequential interim
analyses allowing adaptation of enrolment criteria or stopping new enrolment for futility or predicted success will occur
in every 50 randomized patients starting at 150 to a maximum of 500 patients.

1
UPMC, Presbyterian University Hospital, Pittsburgh, PA, USA 14
2 Los Angeles County Harbor, UCLA Medical Center, Torrance, CA,
UCLA Stroke Center, Los Angeles, CA, USA
3 USA
hOSPITAL VALL D’hEBRON, Barcelona, Spain 15
4 Stryker Neurovascular, Fremont, CA, USA
Toronto Western Hospital, Toronto, Ontario, Canada 16
5 Berry Consultants, LLC, Austin, TX, USA
University Hospitals Cleveland Medical Center, Cleveland, OH, USA 17
6 Alexian Brothers Health System, Elk Grove Village, IL, USA
Hôpital Gui-de-Chauliac, Montpellier, France 18
7 West-Virginia-University-Hospitals, Ruby-Memorial-Hospital,
Erlanger Health System, Chattanooga, TN, USA
8 Morgantown, WV, USA
WellStar Kennestone Hospital, Marietta, GA, USA 19
9 Yale New Haven University Hospital, Yale, New Haven, CT, USA
Rush University Medical Center, Chicago, IL, USA 20
10 Massachussetts General Hospital, Boston, MA, USA
Universitätsklinikum Schleswig-Holstein, Kiel, Germany
11
University of California, San Francisco, Medical Center at Parnassus, The first and last authors contributed equally to this work
San Francisco, CA, USA Corresponding author:
12
University of Nebraska Medical Center, Omaha, NE, USA Raul G Nogueira, Department of Neurology, Grady Memorial Hospital
13
Radiology and Biomedical Imaging, University of California, San and Emory University, School of Medicine, Atlanta, GA, USA.
Francisco, Medical Center, San Francisco, CA, USA Email: raul.g.nogueira@emory.edu

International Journal of Stroke, 0(0)

2 International Journal of Stroke 0(0)

Study outcomes: The primary endpoint is the modified Rankin Scale score at 90 days. The primary safety outcome is
stroke-related mortality at 90 days.
Analysis: The primary endpoint, expressed as a utility-weighted modified Rankin Scale score is analyzed using a Bayesian
posterior probability with adjustment for ischemic core size. For regulatory reasons, a nested co-primary endpoint
analysis was added consisting of the proportion of subjects with modified Rankin Scale 0–2 between the active and
control groups also analyzed using a Bayesian model.

Keywords
Reperfusion, intervention, clinical trial, acute stroke therapy, ischemic stroke, protocols

Received: 31 March 2017; accepted: 11 April 2017

progressors’’ who are less sensitive to time to reperfu-

Introduction and rationale
The efficacy and safety of mechanical thrombectomy
for acute large vessel occlusion (LVO) stroke have
been demonstrated by six completed randomized
trials.1–6 Individual patient level meta-analytic data
drawing from five of the six trials7 revealed that while
mechanical embolectomy, performed in the vast major-
ity of cases with stent retrievers and within 6 h of time
last seen well (TLSW), is associated with a strong over-
all treatment effect, the benefit of this approach declines
with increasing time from stroke onset to substantial
reperfusion. Indeed, the Highly Effective Reperfusion
evaluated in Multiple Endovascular Stroke Trials
(HERMES) collaborators reported that at 7.3 h from
TLSW to achievement of reperfusion significant evi-
dence of benefit from endovascular therapy can no
longer be demonstrated.8
The dramatic time dependency of clinical benefit
from thrombectomy on time to reperfusion shown in
the HERMES study was observed in a patient popula-
tion largely unselected with respect to collaterals, the
main driver of tissue viability in LVO stroke.9 It has
been shown that in patients with small infarcts prior to
embolectomy characterized by high ASPECTS scores10
or small lesions on diffusion-weighted imaging- mag-
netic resonance imaging (DWI MRI),11 both indirect
measures of good collaterals, the likelihood of obtain-
ing a good clinical outcome is substantially less influ-
enced by the time from TLSW to reperfusion compared
to patients with poor collaterals. Following LVO,
growth of the irreversibly damaged brain (ischemic
core) at the expense of the penumbra (hypoperfused
but still viable brain that will ultimately undergo infarc-
tion in the presence of continued vessel occlusion)
occurs with different speeds in different individuals.
Among the factors that drive this individual variability
in core growth rates, collaterals play a critical role.
Therefore, patients with poor collaterals are ‘‘fast pro-
gressors’’ and exquisitely dependent on time to reperfu-
sion while those with good collaterals are ‘‘slow
sion and may benefit from embolectomy beyond the
time windows that apply to ‘‘fast progressors.’’
The hallmark pathophysiological feature predicting
a favourable clinical response to reperfusion therapy is
the presence of mismatch between the volume of infarct
and the total volume of critically hypoperfused brain.12
It is believed that the larger the mismatch the greater
the benefit of reperfusion therapy. It is unknown13,14
whether the presence of mismatch is best measured by
using highly sophisticated imaging criteria that charac-
terize the core (by DWI MRI or computerized tomog-
raphy perfusion (CTP), computed tomography
angiography (CTA) or CT) and the territory at risk
(by MR perfusion or CT perfusion),12,15 or by using
moderately sophisticated imaging criteria to assess
core and clinical exam (NIHSS) to assess the
tissue.13,14 It is well recognized, however, that a sizeable
proportion16 of patients with LVO are ‘‘slow progres-
sors’’ and when they present in later time windows
(beyond 6 h) they may still have substantial mismatch.
While endovascular treatment in late-presenting
‘‘slow progressors’’ has been found to be feasible and
safe,17 the clinical benefit is unknown. Since these
patients presumably have good collaterals in order to
sustain tissue viability within such extended time win-
dows, it is not established that they will experience
infarct growth and consequent clinical deterioration
without reperfusion therapy. Therefore a priority for
the field of acute stroke treatment research, recognized
at the recent STAIR IX meeting,18 has been to deter-
mine whether there is benefit from endovascular treat-
ment in patients with LVO stroke and substantial
mismatch presenting beyond the 6-h time window.
Methods
Objective
To evaluate the hypothesis that Trevo thrombectomy
plus medical management leads to superior clinical

International Journal of Stroke, 0(0)

Jovin et al.
outcomes at 90 days as compared to medical manage-
ment alone in subjects experiencing an acute ischemic
stroke due to proximal anterior circulation LVO when
treatment is initiated within 6–24 h after TLSW.
Subjects must have substantial mismatch defined as
NIHSS  10 and imaging evidence of a small core on
CT perfusion or DWI MRI.
Design
Prospective, randomized, multicenter, Bayesian adap-
tive-enrichment, open label clinical trial with blinded
endpoint assessment. Sequential interim analyses allow-
ing adaptation of enrolment criteria or stopping new
enrolment for futility will occur in every 50 randomized
patients starting at 150 to a maximum of 500 patients.
Beginning with the interim analysis at 200 randomized
patients, the trial may be stopped with respect to new
enrolment because of predicted success when all
patients are followed and final outcomes are known.
An overview of study events is provided in Figure 1.
Patient population—inclusion and exclusion criteria
Inclusion and exclusion criteria are outlined in Table 1.
Patients presenting with acute stroke beyond 6 h of
TLSW are evaluated and screened for potential diffu-
sion-weighted imaging or computerized tomography per-
fusion assessment with clinical mismatch in the triage of
wake up and late presenting strokes undergoing neuroin-
tervention (DAWN) participation. This includes both
patients who present directly to the endovascular center
and those transferred from other hospitals. Screening
paradigms are site specific but in general patients with
TLSW beyond 6 h who have an NIHSS of 10 are typ-
ically screened for the absence of a large hypodensity on
plain CT (ASPECTS  7) before undergoing a CTA or
magnetic resonance angiogram (MRA) confirming anter-
ior circulation LVO. In case proximal LVO is confirmed,
patients undergo a CTP or MRI with DWI imaging based
upon which imaging eligibility criteria are established.
According to local protocols the CTP and/or MRI may
be performed as part of routine clinical care. In case both
MRI and CTP are performed, core size information for
the purposes of trial enrolment is considered on the most
recent imaging study. In subjects meeting imaging and
clinical criteria for DAWN, three distinct circumstances
of TSLW are recognized which may be associated with a
different response to reperfusion therapy. In order to
explore whether this is indeed the case, all subjects
enrolled/randomized into the trial will be categorized as
one of the following:
. Wake-up stroke: subject known to have symptoms
first detected on awakening from sleep.
3
. Witnessed stroke: subject TLSW and symptoms first
observed time known to be the same.
. Un-witnessed stroke: subject TLSW and symptoms
first observed time known to be different, but symp-
toms not first detected upon awakening from sleep.
For the purpose of trial enrollment subjects must
have a thrombus identified within the intracranial
ICA, and/or middle cerebral artery (MCA)-M1 arteries
by pre-procedure MRA or CTA. The MCA-M1 seg-
ment is defined as the first branch of the intracranial
ICA which courses horizontally from its branching
point off the ICA, through the Sylvian fissure up to
the first bifurcation distal to the lenticulostriate arteries,
in the most lateral aspect of Sylvian fissure.
Written informed consent by patient or legal repre-
sentative must be obtained for all subjects who are
screened and meet the general inclusion/exclusion cri-
teria prior to randomization/enrollment.
Randomization
A ‘‘real-time’’ central randomization procedure is
implemented via a web-based randomization system
accessible via the DAWN Trial Website Interactive
Web Response System Tool (IRT). If the subject’s eli-
gibility status is confirmed, the server allocates the
treatment on the basis of a minimization process to
balance in a 1:1 ratio the two arms both overall as
well as stratified by: age adjusted clinical core mismatch
(CCM) (Figure 1). In addition to CCM, therapeutic
window (6–12 h and 12–24 h) and occlusion site
(intracranial ICA or M1 segment) are also considered
in the stratification process.
Treatment
A. Thrombectomy: In subjects randomized to the
Trevo thrombectomy plus medical management
arm (thrombectomy arm) the procedure is per-
formed using the FDA-approved Trevo Retriever
(ProVue and XP ProVue). If for any reason (no
target occlusion at the time of angiography, exces-
sive tortuosity precluding deployment of device,
etc.) the Trevo Retriever cannot be used, the subject
will still be analyzed in the embolectomy arm as per
the intent-to-treat (ITT) analysis. The use of other
endovascular reperfusion methods (other stentrie-
vers, primary aspiration, intracranial stenting/
angioplasty, thrombolytic drugs) is not allowed
and constitute a major protocol deviation. The pro-
cedure must be started (defined as the time of arter-
ial access) no earlier than 6 h, but before 24 h, from
International Journal of Stroke, 0(0)
4 International Journal of Stroke 0(0)

Figure 1. DAWN trial flow chart.

International Journal of Stroke, 0(0)

Jovin et al. 5

Table 1. Inclusion and exclusion criteria

General inclusion 1. Clinical signs and symptoms consistent with the diagnosis of an acute ischemic stroke, and subject
criteria belongs to one of the following subgroups:
a. Subject has failed IV t-PA therapy (defined as a confirmed persistent occlusion 60 min after
administration)
b. Subject is contraindicated for IV t-PA administration
2. Age 18
3. Baseline NIHSS  10 (assessed within 1 h of measuring core infarct volume)
4. Subject can be randomized between with 6 to 24 h after time last known well
5. No significant pre-stroke disability (pre-stroke mRS must be 0 or 1)
6. Anticipated life expectancy of at least 6 months
7. Subject willing/able to return for protocol required follow-up visits
8. Subject or subject’s legally authorized representative (LAR) has signed the study informed consent
forma

General inclusion 1. Subjects receiving heparin or low-molecular weight (LMW) heparin, e.g. FragminÕ (Dalteparin
criteria Sodium) or an intravenous direct thrombin inhibitor such as AngiomaxÕ (Bivalirudin), or
(additional Argatroban within the last 24 h from screening are eligible for participation if their coagulation
information) profile remains acceptable.
2. Subjects on factor Xa inhibitors (e.g. apixaban) or direct thrombin inhibitors are eligible for
participation

Imaging inclusion 1. <1/3 MCA territory involved, as evidenced by CT or MRI

criteria 2. Occlusion of the intracranial ICA and/or MCA-M1 as evidenced by MRA or CTA
3. Clinical imaging mismatch (CIM) defined as one of the following on MR-DWI or CTP-rCBF maps:
a. 0 < 21 cm3 core infarct and NIHSS  10 (and age  80 years old)
b. 0 < 31 cm3 core infarct and NIHSS  10 (and age < 80 years old)
c. 31 cm3 to 51 cm3 core infarct and NIHSS  20 (and age < 80 years old)

General exclusion 1. History of severe head injury within past 90 days with residual neurological deficit, as determined
criteria by medical history
2. Rapid improvement in neurological status to an NIHSS < 10 or evidence of vessel recanalization
prior to randomization
3. Pre-existing neurological or psychiatric disease that would confound the neurological or functional
evaluations, e.g. dementia with prescribed anti-cholinesterase inhibitor (e.g. Aricept)
4. Seizures at stroke onset if it makes the diagnosis of stroke doubtful and precludes obtaining an
accurate baseline NIHSS assessment
5. Baseline blood glucose of <50 mg/dL (2.78 mmol) or >400 mg/dL (22.20 mmol)
6. Baseline hemoglobin counts of <7 mmol/L
7. Baseline platelet count <50,000/uL
8. Abnormal baseline electrolyte parameters as defined by sodium concentration <130 mmol/L,
potassium concentration <3 mEq/L or >6 mEq/L
9. Renal failure as defined by a serum creatinine >3.0 mg/dL (264 mmol/L)
Note: subjects on renal dialysis may be treated regardless of serum creatinine levels
10. Known hemorrhagic diathesis, coagulation factor deficiency, or on anticoagulant therapy with
INR > 3.0 or PTT > 3 times normal. Patients on factor Xa inhibitor for 24–48 h ago must have a
normal PTT.
11. Any active or recent hemorrhage within the past 30 days
12. History of severe allergy (more than rash) to contrast medium
13. Severe, sustained hypertension (systolic blood pressure >185 mm Hg or diastolic blood pressure
>110 mmHg)
Note: If the blood pressure can be successfully reduced and maintained at the acceptable level
using medication the subject can be enrolled
14. Female who is pregnant or lactating at time of admission
15. Current participation in another investigational drug or device study
16. Presumed septic embolus, or suspicion of bacterial endocarditis
17. Treatment with any cleared thrombectomy devices or other intra-arterial (neurovascular) thera-
pies prior to randomization
(continued)

International Journal of Stroke, 0(0)

6 International Journal of Stroke 0(0)

Table 1. Continued

Exclusion criteria 1. The ‘‘correction’’ of baseline glucose or coagulation laboratory values to meet inclusion criteria will
(additional not be allowed.
information) 2. Subjects who have taken Clopidogrel, aspirin, or both within the last 24 h from screening for the
trial should not be excluded if their coagulation profile remains acceptable.
3. Subjects with a questionable seizure at onset of stroke should not be excluded if CTA/MRA con-
firms the presence of intracranial ICA and/or M1 occlusion, and accurate NIHSS can be obtained.

Imaging exclusion 1. Evidence of intracranial hemorrhage on CT/MRI

criteria 2. CTA or MRA evidence of flow limiting carotid dissection, high-grade stenosis, or complete cervical
carotid occlusion requiring stenting at the time of the index procedure (i.e. mechanical thromb-
ectomy)
3. Excessive tortuosity of cervical vessels on CTA/MRA that would likely preclude device delivery/
deployment
4. Suspected cerebral vasculitis based on medical history and CTA/MRA
5. Suspected aortic dissection based on medical history and CTA/MRA
6. Intracranial stent implanted in the same vascular territory that would preclude the safe deployment/
removal of the Trevo device
7. Occlusions in multiple vascular territories (e.g. bilateral anterior circulation, or anterior circulation/
vertebrobasilar system) as confirmed on CTA/MRA, or clinical evidence of bilateral strokes or
strokes in multiple territories
8. Significant mass effect with midline shift as confirmed on CT/MRI
9. Evidence of intracranial tumor (except small meningioma) as confirmed on CT/MRI
a
If approved by local ethics committee and country regulations, the investigator is allowed to enroll a patient utilizing emergency informed consent
procedures if neither the patient nor the representative or person of trust is available to sign the informed consent form. However, as soon as possible,
the patient is informed and his/her consent is requested for the possible continuation of this research. (Not applicable to US Sites.)

Table 2. Modified Rankin Scale score and corresponding weights

mRS 0 1 2 3 4 5 6
Weight 10 9.1 7.6 6.5 3.3 0 0

TLSW and is performed according to the most cur-
rent instructions for use (IFU) of the device. It is rec-
ommended that the interventional procedure starts
within 60 min of randomization and is completed
within 2 h of arterial access. No more than six (6)
retrieval attempts in the same vessel using any of the
available Trevo devices and no more than three (3)
passes per Trevo device are allowed. In cases of
tandem extracranial high grade stenosis or occlusion
with accompanying intracranial occlusion, stenting of
the extracranial lesion is not allowed. However,
angioplasty may be performed if considered necessary
in order to access the intracranial lesion.
B. Medical therapy: All subjects enrolled into this study
are admitted to acute stroke units or ICU if needed
and medically managed according to the 2013 AHA
guidelines,19 according to the 2008 European Stroke
Organization ESO Guidelines,20 according to the
2007 Australian Clinical Guidelines for Acute
Stroke Management,21 and according to the 2015
Canadian Acute Stroke Best Practice recommenda-
tions22 depending on geographical location and hos-
pital specific policies. In the first 24 h after
randomization antiplatelet therapy (aspirin or clopi-
dogrel) or dual antiplatelet therapy as per local man-
agement protocols is allowed but full heparinization
(except low doses intra-procedurally) is not allowed.
In subjects who received IV tissue plasminogen acti-
vator (tPA), blood pressure should be managed
according to post IV tPA management guidelines
within the first 24 h. In subjects who are reperfused
after mechanical thrombectomy (defined as achieving
greater than 2/3 MCA territory) it is recommended
to maintain systolic blood pressure below 140 mm
Hg in the first 24 h to minimize the risk of reperfu-
sion related intracranial hemorrhage (ICH). In sub-
jects who do not achieve recanalization after
thrombectomy similar BP management orders are
applied as for the control subjects within each center.
Clinical and imaging evaluations (also
see Table 3)
In addition to baseline clinical and imaging data that
qualify the patient for enrolment and procedural data

International Journal of Stroke, 0(0)

 Table 3. DAWNTM trial time and events schedule.

Procedure
Jovin et al.

Screening/ (treatment 24 h (6/þ24) (post Day 5–7 / Discharge

Event baseline arm only) randomization) (whichever is earlier) Discharge Day 30  14 Day 90  14
Inclusion/Exclusion criteria 3

Demographics/Medical his- 3
tory/baseline medications

Baseline characteristics 3

Baseline labs 3

Informed consent 3

Randomization (¼time zero) 3a

Angiography procedure 3
details (treatment arm
only)b

mRSc 3 (pre stroke) 3c 3c 3c

NIHSS 3d 3e 3 3 3

Neuro imaging (to assess for 3 MRI/MRA or 3MRI/MRA or 3 MRI or

hemorrhage, occlusion CT/CTA/CTP CT/CTA CT (optional)
location/vessel patency &
infarct volume)b

AEs/SAEs (from time of 3 3 3 3 3 3

randomization)

Concomitant medications 3 3 3 3 3

In hospital med management 3

Intubation details 3

UB04 / Health economics 3

a
Randomization should occur within 1 h of obtaining neuro-imaging used to determine core infarct measurement.
b
CT/MR and angiographic images should be de-identified before being submitted to Stryker NV or core lab.
c
mRS must be conducted by an individual blinded to the treatment arm.
d
NIHSS within 1 h of corresponding core infarct measurement.
e

International Journal of Stroke, 0(0)

7

NIHSS should be obtained within approximately 2 h of the 24 (6/þ24) h neuro-imaging to determine presence/absence of hemorrhage.
8 International Journal of Stroke 0(0)
in patients randomized to thrombectomy, clinical and
imaging evaluations will be collected at 24 (6/þ24) h
post randomization (see Table 2) to include the NIHSS
and MRI/MRA or CT/CTA in order to assess for hem-
orrhage, vessel patency, and infarct volume. Ideally,
same imaging modality should be used at 24 (6/
þ24) h to measure vessel patency as was used at base-
line to identify occlusion location. Depending on the
imaging modality chosen, MRI T2 Flair or CT may
be used to assess core infarct volume. Between day 5
and 7 (if subject remains in the hospital) or prior to
discharge, whichever is earlier, evaluations include
NIHSS, modified Rankin Scale (mRS) (performed in
blinded fashion) and at the investigator’s discretion
and as per local protocols repeat imaging—MRI
(including DWI, T2 and Flair) with or without MRA
to re-assess the follow-up infarct volume and vessel
patency status. CT may be performed if MRI is contra-
indicated. For all subjects who expire prior to the day 5–7
assessment, available information regarding the primary
cause of death and date/time of death will be recorded, as
well as whether the subject was made ‘‘do not resuscitate’’
(DNR) or ‘‘comfort care only’’ prior to death.
All brain imaging studies (including baseline CT or
MRI) will be forwarded to an independent core lab for
analysis which will be performed blinded to treatment
allocation. Imaging parameters evaluated by the core
lab include ASPECTS scores, ischemic core volume,
and total volume of tissue at risk (when available),
follow-up infarct volume at 24 h and whenever avail-
able subsequently, recanalization on the 24-h follow-up
CTA or MRA scan and ICH evaluated using European
Cooperative Acute Stroke Study (ECASS) II criteria.23
In patients randomized to thrombectomy, angiographic
data will also be evaluated by an independent core lab
and the degree of reperfusion achieved will be measured
using both the original (o) TICI (Thrombolysis in
Cerebral Ischemia) score and the modified (m) TICI
score.24
An independent clinical events committee (CEC)
will adjudicate all major adverse events including
those comprising safety outcomes (neurological mor-
tality, sICH) in blinded fashion as well as procedural
adverse events (unblinded). A data safety monitoring
board will analyze the data in an unblinded manner
for safety reasons on an ongoing basis if needed and at
every 50 patients enrolled up to the first 150 patients.
Subsequent to enrolment of 150 patients, the Data and
Safety Monitoring Board (DSMB) is tasked with
making enrichment or stopping recommendations
based on futility, safety or efficacy according to the
pre-specified criteria defined by the adaptive design
outlined in the statistical analysis plan (for details of
the adaptive design please see Supplementary
Appendix).
International Journal of Stroke, 0(0)
Primary outcome
The primary endpoint for this trial is the 90-day mRS
score. DAWN analyzes this standard endpoint by
converting the mRS scores into weights that directly
reflect patient and society valuation of outcome
health states.25–27 We then model a subject’s weighted
mRS score as normally distributed with expected
value depending on infarct size and treatment
assigned.
The weights assigned to the possible mRS scores are
shown in Table 2. The primary endpoint analysis is a
comparison of the difference between the average
weighted mRS score at 90 days post randomization
between the active and control groups.
For regulatory purposes, as a nested co-primary out-
come analysis we also planned a priori to analyze the
primary endpoint in a more conventional fashion using
a dichotomized functional independence (mRS 0–2 vs.
3–6) analysis.
Secondary outcomes
Efficacy
Both arms:
1. Comparison of the proportion of subjects with a
good functional outcome at 90 days, defined as
mRS 0–2, between the active and control groups
analyzed as a traditional Fisher exact test adjusted
for stratification variables.
2. Comparison of the proportion of subjects with
‘‘early response’’ at day 5–7/discharge (whichever is
earlier), defined as a NIHSS drop of 10 from base-
line or NIHSS score 0 or 1, between the active and
control groups.
3. Difference in all cause mortality rates at 90 days
between the two groups.
4. Comparison of the median 24 h infarct size at 24
(6/þ24) h from randomization, by MRI T2/Flair
or CT (if MR is contraindicated), between
the treatment and control groups and comparison
of median infarct growth expressed as median 24 h
infarct size minus median baseline infarct size
between the treatment and control groups. Both
baseline infarct size and 24 h infarct size will be
core lab adjudicated. Difference in revasculariza-
tion rates at 24 (6/þ24) h from randomization,
by CT-MR core lab assessment of vessel patency
on CTA/MRA.
5. Difference in treatment effect across the wake up
strokes vs. witnessed stroke vs. unwitnessed stroke.
6. Differences in treatment effect across the CCM
categories.
Jovin et al.
Treatment arm only:
7. Analysis of vessel reperfusion rates (percentages)
post device and post procedure, by angiography
core lab measurement of mTICI  2 b.
Primary safety outcome
Both arms:
1. Incidence of stroke-related mortality at 90 days.
Secondary safety outcomes
Both arms:
1. Incidence of SICH, by ECASS III definition23 within
24 (6/þ24) h post randomization (time zero).
2. Incidence of neurological deterioration from baseline
NIHSS score through day 5–7/discharge (whichever
is earlier) post randomization (time zero).
Neurological deterioration is defined as 4 point
increase in the NIHSS score from the baseline score.
Treatment arm only:
3. Incidence of procedure-related and device-related
serious adverse events (PRSAEs and DRSAEs)
through 24 (6/þ24) h post randomization (time
zero), as adjudicated by the CEC, and defined as:
. vascular perforation
. intramural arterial dissection
. embolization to new territory
. access site complication requiring surgical repair
or blood transfusion
. intra-procedural mortality
. device failure (in vivo breakage)
. any other complications adjudicated by the CEC
to be related to the procedure.
Subgroup analyses
Pre-specified subgroups were developed a priori for sci-
entific interest and exploratory analyses. Multivariate
and subgroup analyses are considered supportive with-
out control of alpha and the additional endpoint ana-
lyses are likewise supportive without control of alpha.
The following differences in treatment effect for pre-
specified subgroups will be evaluated:
1. A. Age (<80 vs. 80)
B. Sex
9
C. Admission NIHSS (median split)
D. TLSW between 6 and  12 h vs. >12 to 24 h
E. Baseline occlusion location (ICA vs. M1)
Adaptive design (for details of the adaptive design
see Supplementary Appendix)
Sample size
The maximum trial size is 500 subjects, randomized
equally between the two arms. Because of the adaptive
nature of the design, the actual sample size may be less,
with the minimum being 200 subjects in a successful trial.
We investigated treatment effects that increased the
expected weighted mRS by 0, 0.5, 0.75, 1.0, 1.25, and 1.5
units above control for all infarct sizes. The effect sizes of
1.25 and 1.5 are very large and the trial offers better than
95% power to detect such improvements. The design pro-
vides 86% power to detect an effect size of 1 unit. The type
I error probability is controlled to be no more than 2.5%.
Enrichment
The trial includes a pre-specified rule for restricting the
future enrolled patients to those with smaller core
infarct sizes, ‘‘enrichment,’’ with the goal of
focusing the trial on those most likely to benefit,
should there be evidence that the treatment benefit
is restricted to those with smaller infarct sizes.
Enrichment decisions can occur starting at 150 sub-
jects enrolled and the last opportunity to enrich is at
400 subjects. The element driving enrichment is base-
line infarct size. The five possible subpopulations are
defined by infarct size as measured by RAPID using
MR-DWI or CTP-regional cerebral blood flow
(rCBF) maps:
1. The full population of infarct sizes of 0 to 50 cm3;
2. Infarct sizes of 0 to 45 cm3;
3. Infarct sizes of 0 to 40 cm3;
4. Infarct sizes of 0 to 35 cm3;
5. Infarct sizes of 0 to 30 cm3.
If the population is enriched, subjects with larger
infarct sizes are no longer enrolled, and the final effi-
cacy analysis omits subjects with larger infarct sizes
from consideration. Enrichment decisions are irrevers-
ible, but the trial can enrich the population further
after it has already been enriched. The design will
enrich if one of the following conditions is met:
If the highest currently open group of five (5) infarct
sizes has less than 40% posterior probability of an
average positive treatment effect, then this group of
infarct sizes will no longer be enrolled in the trial.
This rule is applied before the second enrichment
International Journal of Stroke, 0(0)
10
rule, and may only be applied once per interim
analysis.
If the predictive probability of a positive trial
increases by at least 10% by enriching to a smaller
subpopulation, then the trial will enrich to the smallest
subpopulation that satisfies this criterion.
Early success
The trial may only stop enrollment for expected suc-
cess if at least 100 subjects have been enrolled since
the last enrichment. The decision to stop further
enrollment is based on the predictive probability of
final trial success, once all patients are followed to
their 90-day mRS outcome, for both the weighted
mRS analysis and dichotomous mRS analysis, if no
further subjects are enrolled. The threshold for this
predictive probability is 95% for the 200 and 250
subject interim analyses, 90% for the 300 and 350
subject interim analyses, 85% for the 400 subject
interim analysis, and 80% for the 450 and 500 sub-
ject analyses. If the predictive probability exceeds the
predictive threshold at an interim analysis, for both
the planned weighted and dichotomous analyses, then
enrollment stops for expected success.
Final analysis
The final analysis using the weighted mRS is Bayesian
and includes a flexible normal dynamic linear model
(NDLM) to account for different expected outcomes
as a function of infarct size. This is a flexible spline-
like model that will capture that the average-weighted
mRS score in the control group is a (possibly non-
linear) function of infarct size. Meanwhile the average
treatment effect is assumed to be equal over the identi-
fied range of infarct sizes.
The final dichotomous analysis is the proportion of
subjects with functional independence (mRS 0–2)
between the active and control groups analyzed in a
one-sided Bayesian dynamic linear model. Of note,
from a regulatory perspective, the trial will only be con-
sidered successful, including beneficial impact on func-
tional independence, if both the primary (weighted
analysis) and the nested co-primary (dichotomous ana-
lysis) are positive according to pre-specified criteria.
The overall treatment effect is given a vague prior
distribution, and if there is a high posterior probability
that the overall treatment effect is positive, the device is
declared to be efficacious. The final analysis will be per-
formed only on the enriched population, and assumes a
constant treatment effect over all infarct sizes that are
in the population at the end of the trial. The threshold
for declaring success depends on the degree to which
the population has been enriched, with the thresholds
International Journal of Stroke, 0(0)
International Journal of Stroke 0(0)
adjusted to control type I error probability to be not
more than 0.025. If no enrichment occurs, the threshold
is 0.986.
Discussion
There are several novel features with regard to the
design of DAWN. A centerpiece of the trial is inclusion
of patients based on proof of substantial mismatch,
which necessitates assessment of core and total tissue
at risk. There is consensus that the optimal way to
assess the former is through imaging. In order to
increase precision in core measurements and to avoid
variability in its assessment across sites, we have used
an automated method detection method (RAPID)
(iSchemaView, Menlo Park, CA, USA) which has
been validated in several prospective trials12,15 and
received FDA clearance for clinical use. There is how-
ever considerable debate with regard to the optimal
method for estimating the total tissue at risk. The per-
fusion imaging method (PIM), as described in DEFUSE
2,12 SWIFT PRIME,15 and EXTEND-IA5 utilizes ima-
ging based (CTP or PWI MRI) delineation of total
tissue at risk using perfusion (Tmax) data. However
there are several disadvantages related to this approach:
it has not been sufficiently validated in the extended time
window, it is prone to artifact and it represents a snap-
shot in time of a dynamic process. Thus, the use of
NIHSS as an indicator of tissue at risk has been pro-
posed in lieu of perfusion studies, giving rise to the con-
cept of CCM whose presence was shown to predict a
favorable clinical response to reperfusion.13 In a head-
to-head comparison between PIM and CCM definition
in patients enrolled in SWIFT PRIME it has been
shown that CCM as defined in DAWN is at least as
predictive of a favorable response to thrombectomy as
PIM as defined by DEFUSE 2 criteria.14
Another novel feature in the design of DAWN is the
age-adjusted core inclusion criteria. It has been shown
that the upper end of final infarct volume associated
with functionally independent outcomes is age depend-
ent.28,29 The higher the age, the lower the infarct
volume likely to be associated with functional inde-
pendence. By restricting baseline core volume in octo-
genarians to less than 20 cm3, DAWN is designed to
maximize the chance of obtaining functional independ-
ence in octogenarians following reperfusion and thus
increase the chance of obtaining an unequivocally
worthwhile treatment effect while not utilizing an age
limit for enrollment.
The DAWN trial uses a Bayesian adaptive enrich-
ment design to address the several uncertainties at trial
launch regarding the thresholds at which patient benefit
from intervention may lie. The natural history of acute
stroke due to LVO in patients who, due to robust
Jovin et al.
collateral flow have evidence of substantial areas of
reversible ischemia, in the time window beyond 6 h,
has not been established by prospective studies.
Furthermore, variation in degree of treatment effect
based on baseline core size within the 0–50 cm3 range
is also not possible to know in advance due to a paucity
of observational natural history data and thrombec-
tomy case series data. For that reason we used an adap-
tive design, which not only allows stopping enrollment
after as early as 200 patients out of a maximum sample
size of 500 patients for predicted success but also allows
removal from the trial of patient populations whose
inclusion based on core sizes situated at the large end
of the infarct volume spectrum, may show early signs of
futility.
Finally, while the primary endpoint of DAWN is the
widely used mRS assessing global disability, the pri-
mary analysis method is different than that used in all
randomized acute stroke reperfusion trials to date. We
believe that this analysis confers advantages over the
more conventional methods of analyzing the primary
outcome in acute stroke trials due to its stronger
patient-centered nature. First, compared with crude
dichotomized analyses of the mRS, the approach in
DAWN analyzes all health state transitions that
matter to patients and families, rather than just a
single transition. For treatments that produce unidirec-
tional shifts in benefit, this approach better captures the
full magnitude of treatment effect.30 In addition, the
utility-weighted mRS analysis allows capture and ana-
lysis of a bidirectional effect of embolectomy (i.e. a
beneficial effect at one end of the outcome scale and a
harmful effect at the other) that would be completely
missed in a dichotomous analysis. Second, rather than
using arbitrarily chosen weights as in the case of classic
ordinal analysis, the utility weighted mRS analysis
takes into consideration patient perception of the
impact on overall level of functioning experienced at
each level of the mRS as assessed by previously con-
ducted surveys in different populations which have
yielded remarkably similar values.25,26 Ultimately, the
purpose of any analysis method in endovascular acute
stroke trials is to capture the value of outcomes to
patients and other stakeholders (e.g. caregivers) to the
fullest extent. We believe that the utility weighted mRS
analysis accomplishes this goal better compared to
other methods of primary endpoint analysis.27
Nonetheless, based on regulatory input, we also
added the dichotomized analysis, a more conventional
form of primary endpoint analysis, as a nested co-pri-
mary analysis with the specification that from a regu-
latory standpoint, both the weighted analysis and the
dichotomized analysis (analyzed in Bayesian fashion)
need to be positive in order for the trial to be declared
successful. Thus, in case of a positive trial by both
11
analysis methods, the rather novel approach of
weighted analysis may be validated by the more trad-
itional method of endpoint analysis which may justify
the use of the weighted analysis method in future trials.
The DAWN trial aims to generate level one evidence
in support of the concept that patients with acute stroke
due to anterior circulation occlusion and substantial
mismatch, benefit from endovascular reperfusion with
Trevo when treatment is initiated beyond 6 h from
TLSW, a time window in which level one evidence of
benefit has not conclusively been established. DAWN
builds upon several single arm studies or randomized
trials that have included patients believed to have mis-
match who were treated beyond 6 h, revealing not only
that this approach has a reasonable safety profile but
also that trends towards clinical benefit exist.7,8,11
However, no randomized, prospective study specifically
seeking to demonstrated benefit in this patient popula-
tion has been completed to date. Therefore, if positive,
DAWN may have profound implications for treatment
of stroke due to LVO, because it would validate the
physiological (rather than chronological) approach to
patient selection for endovascular therapy. It will also
allow many more patients with LVO stroke to be trea-
ted with mechanical embolectomy, especially in coun-
tries outside of the US, Australia, Canada, and Western
Europe, where due to inadequate development for
stroke pre-hospital systems of care, a large proportion
of patients with LVO stroke present to endovascular
centers outside 6 h from TLSW.
Declaration of conflicting interests
The author(s) declared the following potential conflicts of
interest with respect to the research, authorship, and/or pub-
lication of this article: The principal investigators, Drs. Jovin
and Nogueira’s DAWN-related travel expenses were covered
by Stryker Neurovascular for the duration of trial Other
steering committee members: Jeffrey L Saver, Marc Ribo,
Vitor Perreira, Anthony Furlan, Alain Bonafe, Blaise
Baxter, Rishi Gupta, Demetrius Lopes, Olav Jansen),
DSMB members (Wade Smith, Darryl Gress,Steven Hetts,
Roger J Lewis), CEC members (Tim Malisch, Ansaar Rai,
Kevin N Sheth, core lab (David Liebeskind) and Statistical
Consultants Scott M Berry and Todd L Graves) report con-
sulting fees for their work in this trial.Ryan Shields is an
employee of Stryker Neurovascular.
Funding
The author(s) received no financial support for the research,
authorship, and/or publication of this article.
References
1. Berkhemer OA, Fransen PS, Beumer D, et al. A rando-
mized trial of intraarterial treatment for acute ischemic
stroke. N Engl J Med 2015; 372: 11–20.
International Journal of Stroke, 0(0)
12
2. Saver JL, Goyal M, Bonafe A, et al. Stent-retriever
thrombectomy after intravenous t-PA vs. t-PA alone in
stroke. N Engl J Med 2015; 372: 2285–2295.
3. Goyal M, Demchuk AM, Menon BK, et al. Randomized
assessment of rapid endovascular treatment of ischemic
stroke. N Engl J Med 2015; 372: 1019–1030.
4. Jovin TG, Chamorro A, Cobo E, et al. Thrombectomy
within 8 hours after symptom onset in ischemic stroke. N
Engl J Med 2015; 372: 2296–2306.
5. Campbell BC, Mitchell PJ, Kleinig TJ, et al.
Endovascular therapy for ischemic stroke with perfu-
sion-imaging selection. N Engl J Med 2015; 372:
1009–1018.
6. Bracard S, Ducrocq X, Mas JL, et al. Mechanical
thrombectomy after intravenous alteplase versus alte-
plase alone after stroke (THRACE): a randomised con-
trolled trial. Lancet Neurol 2016; 15: 1138–1147.
7. Goyal M, Menon BK, van Zwam WH, et al.
Endovascular thrombectomy after large-vessel ischaemic
stroke: a meta-analysis of individual patient data from
five randomised trials. Lancet 2016; 387: 1723–1731.
8. Saver JL, Goyal M, van der Lugt A, et al. Time to treat-
ment with endovascular thrombectomy and outcomes
from ischemic stroke: a meta-analysis. JAMA 2016;
316: 1279–1288.
9. Sheth SA and Liebeskind DS. Collaterals in endovascular
therapy for stroke. Curr Opin Neurol 2015; 28: 10–15.
10. Ribo M, Molina CA, Cobo E, et al. Association between
time to reperfusion and outcome is primarily driven by
the time from imaging to reperfusion. Stroke 2016; 47:
999–1004.
11. Lansberg MG, Cereda CW, Mlynash M, et al. Response
to endovascular reperfusion is not time-dependent in
patients with salvageable tissue. Neurology 2015; 85:
708–714.
12. Lansberg MG, Straka M, Kemp S, et al. MRI profile and
response to endovascular reperfusion after stroke
(DEFUSE 2): a prospective cohort study. Lancet Neurol
2012; 11: 860–867.
13. Davalos A, Blanco M, Pedraza S, et al. The clinical-DWI
mismatch: a new diagnostic approach to the brain tissue
at risk of infarction. Neurology 2004; 62: 2187–2192.
14. Nogueira RGAGW, Haussen DC, Liebeskind DS,
Yavagal D, Saver J, Jovin TG and Investigators ftSP.
Clinical-imaging mismatch versus perfusion imaging mis-
match selection for mechanical thrombectomy: a post-hoc
analysis of the SWIFT PRIME Trial. SAGE, 2016, p.639.
15. Albers GW, Goyal M, Jahan R, et al. Ischemic core and
hypoperfusion volumes predict infarct size in SWIFT
PRIME. Ann Neurol 2016; 79: 76–89.
16. Copen WA, Rezai Gharai L, Barak ER, et al. Existence
of the diffusion-perfusion mismatch within 24 hours after
onset of acute stroke: dependence on proximal arterial
occlusion. Radiology 2009; 250: 878–886.
17. Jovin TG, Liebeskind DS, Gupta R, et al. Imaging-based
endovascular therapy for acute ischemic stroke due to
proximal intracranial anterior circulation occlusion
International Journal of Stroke, 0(0)
International Journal of Stroke 0(0)
treated beyond 8 hours from time last seen well: retro-
spective multicenter analysis of 237 consecutive patients.
Stroke 2011; 42: 2206–2211.
18. Jovin TG, Albers GW, Liebeskind DS and Consortium
SI. Stroke treatment academic industry roundtable: the
next generation of endovascular trials. Stroke 2016; 47:
2656–2665.
19. Jauch EC, Saver JL, Adams Jr. HP, et al. Guidelines for
the early management of patients with acute ischemic
stroke: a guideline for healthcare professionals from the
American Heart Association/American Stroke
Association. Stroke 2013; 44: 870–947.
20. European Stroke Organisation Executive C and
Committee ESOW. Guidelines for management of ischae-
mic stroke and transient ischaemic attack 2008.
Cerebrovasc Dis 2008; 25: 457–507.
21. Hill K; Acute Stroke Management Expert Working G.
Australian clinical guidelines for acute stroke manage-
ment 2007. Int J Stroke 2008; 3: 120–129.
22. Casaubon LK, Boulanger JM, Blacquiere D, et al.
Canadian stroke best practice recommendations:
Hyperacute stroke care guidelines, update 2015. Int J
Stroke 2015; 10: 924–940.
23. Hacke W, Kaste M, Fieschi C, et al. Randomised double-
blind placebo-controlled trial of thrombolytic therapy
with intravenous alteplase in acute ischaemic stroke
(ECASS II). Second European-Australasian Acute
Stroke Study Investigators. Lancet 1998; 352: 1245–1251.
24. Higashida RT, Furlan AJ, Roberts H, et al. Trial design
and reporting standards for intra-arterial cerebral
thrombolysis for acute ischemic stroke. Stroke 2003; 34:
e109–e137.
25. Hong KS and Saver JL. Quantifying the value of stroke
disability outcomes: WHO global burden of disease pro-
ject disability weights for each level of the modified
Rankin Scale. Stroke 2009; 40: 3828–3833.
26. Rivero-Arias O, Ouellet M, Gray A, Wolstenholme J,
Rothwell PM and Luengo-Fernandez R. Mapping the
modified Rankin scale (mRS) measurement into the gen-
eric EuroQol (EQ-5D) health outcome. Med Decis
Making 2010; 30: 341–354.
27. Chaisinanunkul N, Adeoye O, Lewis RJ, et al. Adopting
a patient-centered approach to primary outcome analysis
of acute stroke trials using a utility-weighted modified
Rankin scale. Stroke 2015; 46: 2238–2243.
28. Rangaraju S, Liggins JT, Aghaebrahim A, et al.
Pittsburgh outcomes after stroke thrombectomy score
predicts outcomes after endovascular therapy for anterior
circulation large vessel occlusions. Stroke 2014; 45:
2298–2304.
29. Ribo M, Flores A, Mansilla E, et al. Age-adjusted infarct
volume threshold for good outcome after endovascular
treatment. J Neurointerv Surg 2014; 6: 418–422.
30. Saver JL. Optimal end points for acute stroke therapy
trials: best ways to measure treatment effects of drugs
and devices. Stroke 2011; 42: 2356–2362.