MR Imaging Selection of

Acute Stroke Patients with

E m e r g e n t L a r g e Ve s s e l
O c c l u s i o n s f o r T h ro m b e c t o m y
Thabele M. Leslie-Mazwi, MDa,b,c,d, Michael H. Lev, MDe,
Pamela W. Schaefer, MDf,g,
Joshua A. Hirsch, MD, FSIR, FSNISh,i,
R. Gilberto González, MD, PhDj,*

KEYWORDS
 MR Imaging  DWI  Ischemic stroke  Patient selection  Large vessel occlusion

KEY POINTS
 MR imaging offers unparalleled imaging returns for patients with acute stroke.
 Most patients with acute stroke are candidates for an MR imaging scan, and the proportion keeps
growing as prior MR imaging contraindications (eg, pacemakers) evolve with better data about risk.
 Necessary sequences can be obtained in very short time intervals, minimizing scan time.
 If the patient has a small core (<70–100 mL), the penumbra must be large, and direct penumbral
imaging with perfusion is not needed.
 Future applications include automation and decision support for acute stroke triage.

INTRODUCTION decisions for thrombectomy. With the success of

nearly a dozen prospective clinical trials demon-
Imaging of the patient with acute stroke with strating the efficacy of mechanical thrombec-
an anterior circulation large vessel occlusion tomy, the focus now is on optimizing the
(LVO) is a cornerstone of management and triage decision to proceed to thrombectomy for the

Financial Disclosure: Dr M.H. Lev reports personal fees from GE, MedyMatch, Takeda, and D-Pharm, and nonfi-
nancial support from Siemens, outside the submitted work; all other authors have no disclosure statements.
a
Neuroendovascular Program, Massachusetts General Hospital, Harvard Medical School, WAC-7-745, MGH, 15
Parkman Street, Boston, MA 02114-3117, USA; b Neurocritical Care, Massachusetts General Hospital, Harvard
Medical School, WAC-7-745, MGH, 15 Parkman Street, Boston, MA 02114-3117, USA; c Department of Neuro-
surgery, Massachusetts General Hospital, Harvard Medical School, WAC-7-745, MGH, 15 Parkman Street, Bos-
ton, MA 02114-3117, USA; d Department of Neurology, Massachusetts General Hospital, Harvard Medical
School, WAC-7-745, MGH, 15 Parkman Street, Boston, MA 02114-3117, USA; e Emergency Radiology, Massa-
chusetts General Hospital, Harvard Medical School, BLK-SB-0038 MGH, 55 Fruit Street, Boston, MA 02114,
[Link]

USA; f Neuroradiology, Massachusetts General Hospital, Harvard Medical School, Founders 228 MGH, 55
Fruit Street, Boston, MA 02114, USA; g Radiology, Massachusetts General Hospital, Harvard Medical School,
Founders 228 MGH, 55 Fruit Street, Boston, MA 02114, USA; h NeuroInterventional Radiology, Massachusetts
General Hospital, Harvard Medical School, Gray 241 MGH, 55 Fruit Street, Boston, MA 02114, USA;
i
Interventional Radiology, Massachusetts General Hospital, Harvard Medical School, Gray 241 MGH, 55 Fruit
Street, Boston, MA 02114, USA; j Neuroradiology, Massachusetts General Hospital, Harvard Medical School,
Gray 241 MGH, 55 Fruit Street, Boston, MA 02114, USA
* Corresponding author.
E-mail address: RGGONZALEZ@[Link]

Neuroimag Clin N Am 28 (2018) 573–584

[Link]
1052-5149/18/Ó 2018 Elsevier Inc. All rights reserved.
Descargado para Anonymous User (n/a) en Pontifical Xavierian University de [Link] por Elsevier en agosto 15, 2019.
Para uso personal exclusivamente. No se permiten otros usos sin autorización. Copyright ©2019. Elsevier Inc. Todos los derechos reservados.
574 Leslie-Mazwi et al

benefit of the individual patient. Computed to- blood flow and perfusion, produces ischemia, and
mography (CT) and MR imaging offer options for results in tissue injury and death. Irreversibly
acute stroke imaging. The major advantage of injured tissue is in close proximity to the occluded
MR imaging over CT is its capability to detect artery and is termed the ischemic core. The core is
and precisely estimate the volume of the infarct surrounded by tissue that may be underperfused
core. This precision leads to better individual pa- but is still viable. This is loosely designated as
tient selection. Precision in measuring the core the “penumbra.” The relative size of the core with
also makes possible its other advantage: the respect to the penumbra depends on the quality
elimination of the need to attempt to measure of the pial-pial collateral circulation. A robust
the “penumbra” with imaging. In this article, we collateral circulation results in a small core and
detail use of MR imaging for acute ischemic simultaneously large penumbra, a circumstance
stroke due to LVO for optimal triage decisions in that is ideal for intervention. If the pial-pial collat-
individual patients. We recognize that our multi- erals are weak, a large ischemic core rapidly
disciplinary stroke program and resources may emerges, and the patient is much less likely to
be uncommon and not easily replicated at every benefit from thrombectomy.
stroke center. Nonetheless, our experience may A consequence of this special physiology is that
be a useful guide for others to optimally modify if the ischemic core can be measured precisely,
their own stroke imaging protocols for cases of the need to directly image the “penumbra” is obvi-
emergent LVO (ELVO). ated. All that is necessary is evidence that a pen-
umbra is present, which may be provided by an
ANTERIOR CIRCULATION LARGE VESSEL abnormal neurologic examination. This fact has
OCCLUSION PHYSIOLOGY been validated by the success of the DAWN trial,1
wherein only core, and no penumbral imaging was
The MR imaging strategy for assessing patients
performed.
with ELVO derives from the unique physiology pro-
duced by the occlusion of a major anterior circula-
tion artery. The occlusion of the proximal middle GOALS OF ACUTE STROKE IMAGING
cerebral artery (MCA) or terminal internal carotid Imaging for the patient with acute stroke aims to
artery (ICA) by an embolus is common, deadly, accomplish 4 primary goals:
and potentially treatable. It manifests in alterations
in cerebral hemodynamics and physiology, as 1. Detect presence of hemorrhage
illustrated in Fig. 1. The sudden occlusion reduces 2. Identify a treatable arterial occlusion(s)

Fig. 1. Physiology of an LVO. Representation of a right MCA occlusion. Reduced blood flow results in irreversible
injury to part of the brain (ischemic core) just distal to the site of occlusion. There is a larger area designated as
“penumbra,” in which the blood flow is abnormal but the tissue is viable because of the pial-pial collateral cir-
culation. This part of the brain may recover if normal flow is reestablished. Because of the collateral circulation,
the sizes of the core and penumbra are dependent variables, so if one is small, the other must be large, and vice
versa. Because of this dependency, if the core is small, a large mismatch may be assumed and direct measurement
of the penumbra is not necessary.

Descargado para Anonymous User (n/a) en Pontifical Xavierian University de [Link] por Elsevier en agosto 15, 2019.
Para uso personal exclusivamente. No se permiten otros usos sin autorización. Copyright ©2019. Elsevier Inc. Todos los derechos reservados.
 MR Imaging Selection of Acute Stroke Patients 575

3. Estimate the core infarct volume time taken to obtain such information. An example
4. Establish the presence of a significant amount of a patient who underwent successful thrombec-
of penumbral tissue (the viable tissue with tomy after undergoing evaluation with MR imaging
altered perfusion) is shown in Fig. 2.

These goals are listed in order of priority.

Detect Presence of Hemorrhage
Excluding hemorrhage is the most important
goal. The remaining 3 goals take on increasing Accurate detection of hemorrhage is essential for
importance for patients considered for mechanical the patient with acute stroke. Hemorrhage may be
thrombectomy. In all cases, a balance exists be- due to a variety of causes. Sometimes there is
tween obtaining accurate information and the hemorrhage in the setting of vascular occlusion.

Fig. 2. Case of a woman who experienced sudden right hemiparesis while on a transatlantic flight. The witnessed
stroke occurred while flying over the Atlantic Ocean. The patient’s condition was radioed to the airport and the
hospital was alerted. On arrival, the patient was immediately transported to the hospital emergency department.
Noncontrast CT demonstrated a dense left MCA sign (not shown). (A) CTA of the head. An image is shown from
axial, thick slab, overlapping MIP series. The study demonstrated occlusion of the distal left ICA and proximal left
MCA and anterior cerebral artery (ACA) (arrow). (B) DWI shows the ischemic core involving the insula, left frontal,
and left temporal lobes. The volume of the DWI lesion was less than 25 mL. (C) Frontal, left internal carotid arte-
riogram confirms occlusion of the distal left ICA. (D) Post-thrombectomy left common carotid arteriogram dem-
onstrates removal of occlusion and anterograde flow within the left ICA, MCA, and ACA arteries and their
branches. (E) CT scan 1 week after procedure shows small infarct in the left MCA territory. The patient ultimately
fully recovered from her neurologic deficits. Full details may be found in case records of the MGH. Case 13 to
2016. (Data from Schwamm LH, Jaff MR, Dyer KS, et al. CASE RECORDS of the MASSACHUSETTS GENERAL
HOSPITAL. Case 13-2016. A 49-year-old woman with sudden hemiplegia and aphasia during a transatlantic flight.
N Engl J Med 2016;374(17):1671–80.)

Descargado para Anonymous User (n/a) en Pontifical Xavierian University de [Link] por Elsevier en agosto 15, 2019.
Para uso personal exclusivamente. No se permiten otros usos sin autorización. Copyright ©2019. Elsevier Inc. Todos los derechos reservados.
576 Leslie-Mazwi et al

Delivery of intravenous thrombolytic agents or neck, time-of-flight (TOF) MRA of the head may
mechanical reperfusion in the presence of active be more appropriate. This can be obtained in
hemorrhage could be devastating. Cerebral hem- 2-dimensional (2D) or 3D acquisitions. Besides
orrhage may be visualized by many MR sequences. avoiding a contrast load, TOF imaging has the
The most sensitive sequences for detection of advantage of repeatability if imaging is suboptimal
blood are those that are sensitive to local distur- (eg, an uncooperative or agitated patient). Two-
bances in the magnetic field, typically produced dimensional TOF imaging has lower spatial resolu-
by blood products. In stroke, the primary concern tion but more rapid acquisition than 3D TOF
is parenchymal hematoma. Secondary consider- imaging. Two-dimensional MRA is also acquired
ations for blood products include thrombus a single slice at a time, so it may be adequate
causing vascular occlusion and chronic microhe- even if the patient moves.
morrhages. These blood products produce hypo-
intensity on T2*-weighted images. The most
Estimate of the Core Infarct Volume
commonly used sequences for hemorrhage detec-
tion are gradient recalled echo (GRE) sequences. Various determinants of patient outcome exist, but
More rapid T2* echo planar images (EPI) are also the most significant of these are the severity of
effective, and have the advantage of freezing mo- initial neurologic deficit, the presence and site of
tion because slices are obtained very rapidly arterial occlusion, the success of attempted reper-
(<100 ms per slice). There are more sensitive se- fusion, and the final infarct core volume (a function
quences, including susceptibility-weighted imag- of both adequacy of collateral circulation and suc-
ing and T2 star weighted angiography imaging, cessful reperfusion). The impact of final infarct
but they have longer acquisition times and the core volume on outcome2 has led to the utilization
increased sensitivity is not essential for the patient of pretreatment infarct core volumes for selection
with LVO being considered for thrombectomy. for therapy. For mechanical thrombectomy pre-
treatment infarct core volume gains increasing
importance in later time windows. The ability of
Identify Vascular Occlusion
MR imaging to delineate this core volume with
A variety of MR imaging sequences exist to assess high precision is the key advantage of this modality
the vessels of the head and neck. Emergent vessel for scanning patients with stroke.
imaging is of particular importance for patients MR imaging is unparalleled for detection of
with suspected LVO. For patients with normal infarct core. MR diffusion-weighted imaging
renal function, comprehensive information is (DWI) allows the detection of acute ischemia as
provided using contrast-enhanced magnetic reso- early as 30 minutes from onset, with high sensi-
nance angiography (ceMRA). This is characterized tivity and specificity (>90%).3 The reduction of
by fast acquisition times (usually 2 minutes or less) the normal random movement (diffusion) of water
and a large volume of tissue coverage. An intrave- molecules causes reduction in the apparent diffu-
nous bolus of gadolinium contrast is administered sion coefficient of water in ischemic brain tissue
and the scan is performed with short gradient echo and increased signal intensity on DWI. This reduc-
sequences. Blood appears hyperintense on the tion in diffusion of tissue water is not entirely un-
sequence because gadolinium shortens blood derstood. Contributions are thought to come
T1, and signal from all the remaining tissue is sup- from failure of membrane ionic pumps with net
pressed by the short repetition time. Optimized im- influx of water from the extracellular to the more
aging using this sequence requires appropriate osmotic intracellular compartment, reduced extra-
timing, so data are obtained during the peak of cellular volumes, decreased cytoplasmic mobility,
arterial enhancement. This may be accomplished increased intracellular viscosity from the fragmen-
by use of a test bolus or more frequently by tation of cellular components, and increased cell
automatic bolus detection. Ideally ceMRA will be membrane permeability.
obtained from the aortic arch to the superior tem- For most patients, the initial DWI lesion volume
poral lobes, providing visualization of the origins of closely correlates with the final infarct volume if
the great vessels of the neck, the common carotid there is successful recanalization. However, in
arteries, the internal carotid and vertebral arteries, the context of very early reperfusion, partial DWI
the M1 and proximal M2 segments of the MCAs, reversal (tissue that has restricted diffusion but
and the basilar and proximal posterior cerebral appears normal on follow-up imaging) has been
arteries. observed. Beyond a certain time threshold, which
For patients who have poor renal function data suggest is 3.0 to 4.5 hours of cerebral
(glomerular filtration rate <30 ml/mn) or who have artery occlusion, the DWI abnormality represents
recently undergone CT or MR angiography of the tissue that has been irreversibly injured. If diffusion

Descargado para Anonymous User (n/a) en Pontifical Xavierian University de [Link] por Elsevier en agosto 15, 2019.
Para uso personal exclusivamente. No se permiten otros usos sin autorización. Copyright ©2019. Elsevier Inc. Todos los derechos reservados.
 MR Imaging Selection of Acute Stroke Patients 577

imaging is performed immediately after successful recovery (FLAIR) provides the most additional
reperfusion, there may be a transient reversal of information about brain pathology that may be
most or the entire lesion. The apparent temporary present.
reversal of DWI change after reperfusion is there-
fore described as “pseudonormalization” for these MASSACHUSETTS GENERAL HOSPITAL
patients. To summarize, between 1 and 3 hours, STROKE IMAGING PROTOCOL 2018
DWI reversal is variable. Beyond 4.5 hours, the vol-
ume of tissue that undergoes reversal is usually no Our approach to imaging patients with acute
more than 10% of the total DWI lesion, and for stroke is based on more than 20 years of experi-
most patients is less than 5 mL.4–6 ence in imaging and performing endovascular
arterial recanalization. Regular, critical reviews
Estimate of the Presence of a Significant of our outcomes, and published data from other
Volume of Penumbral Tissue stroke centers, provide the evidentiary basis for
the Massachusetts General Hospital (MGH)
The prospective observational study by Leslie- stroke imaging algorithm, which was first pub-
Mazwi and colleagues7 and the prospective, ran- lished in 2013.8 The major change implemented
domized DAWN trial by Nogueira and colleagues1 in the 2013 protocol was elimination of CT perfu-
have demonstrated that direct imaging of the “pen- sion because our data indicated that it could be
umbra” is not necessary for the successful selection unreliable when assessing individual patients.9
of patients for thrombectomy. Indeed, penumbral Using this protocol, we performed a prospective
imaging is beset by the fundamental conundrum observational trial that was published in 2017.7
that there is no generally accepted definition for The publication of the DAWN1 and DEFUSE 310
such an imaging measurement. The lack of defini- trials, as well as recent subgroup analyses of
tion has resulted in the adoption of measurements these trials, have added to our understanding,
of altered contrast delivery times (Tmax) or contrast and our protocol has been modified in response
transit times (MTT) and the use of a DWI/Tmax or to the new data. The current version of our
DWI/MTT mismatch for patient selection for throm- protocol reflecting the new evidence is shown
bectomy. In the presence of an LVO, there may exist in Fig. 3.
oligemic as well as normally perfused tissue beyond
the infarct core. For example, a patient with a Presentation Less Than 6 Hours from Onset
chronic occlusion of a proximal ICA may have fully
Our imaging protocol for patients who present
compensated, normal arterial blood flow but would
within 6 hours of ictus is shown in Fig. 3A. Imaging
meet the criteria of a large “mismatch.” As specif-
typically begins with CT. There are several reasons
ically addressed in the DAWN trial, a clinical “pen-
for this, but the major factors are that CT scans are
umbra,” that is, the presence of a neurologic
easily and quickly obtained, and they are reliable
deficit that is not completely explained by the infarct
for detecting the presence of hemorrhage as well
core visualized on DWI, is adequate to proceed to
as a treatable LVO. For patients within the window
thrombectomy when there is a small core.
for intravenous thrombolysis (less than 4.5 hours
Perfusion imaging with MR imaging may be use-
from onset), tissue plasminogen activator (tPA) is
ful in the assessment of the patient with ischemic
transported by pharmacy to the scanner with the
stroke that is not caused by an LVO. A patient
patient. If noncontrast head CT demonstrates no
may present with an acute stroke syndrome due
hemorrhage, tPA is reconstituted and the bolus
to occlusion of a small branch vessel that is not
is given as a parallel process to preparation for
visualized by MRA or CT angiography (CTA). In
CTA. CTA acquisition occurs from the arch
this circumstance, dynamic contrast-enhanced
through the vertex with delayed images from the
perfusion imaging or arterial spin labeling may
arch through the circle of Willis. CTA data are pro-
reveal a small volume of abnormal perfusion
cessed to produce thick slab (30 mm), overlapping
caused by the small branch occlusion. These
(5 mm between the center of each slab) axial, cor-
lesions are not amenable to mechanical thrombec-
onal, and sagittal maximum intensity projection
tomy, but other therapeutic approaches may be
(MIP) images. Treatment decisions for thrombec-
attempted.
tomy are made on the basis of vascular occlusion
and the adequacy of collaterals demonstrated on
Application of Additional MR Imaging
the CTA axial MIPs. The collateral circulation is
Sequences
adequate if they are similar or more prominent on
Additional sequences are not necessary for deci- the side of the occlusion when compared with
sions regarding thrombectomy. However, if time the contralateral side. This is the first of 2 major
permits, T2-weighted fluid-attenuated inversion changes in our new protocol. We have found that

Descargado para Anonymous User (n/a) en Pontifical Xavierian University de [Link] por Elsevier en agosto 15, 2019.
Para uso personal exclusivamente. No se permiten otros usos sin autorización. Copyright ©2019. Elsevier Inc. Todos los derechos reservados.
578 Leslie-Mazwi et al

Fig. 3. MGH stroke imaging protocol 2018. (A) Patient last seen well less than 6 hours before possible interven-
tion. Noncontrast CT is performed to assess for hemorrhage or large completed infarct. If neither is present, CTA is
performed to identify a large artery occlusion. If present, the collateral circulation is assessed on the axial, thick
slab, overlapping MIP images. If the collaterals on the occluded side are similar or hypervascular compared with
the contralateral side, the probability of a small ischemic core is high, and patient may proceed to thrombectomy.
If the ipsilateral collateral circulation is less robust, the patient proceeds to MR imaging for DWI acquisition.
A DWI lesion of less than 70 to 100 mL generally makes the patient eligible for thrombectomy. (B) Patient last
seen well more than 6 hours before possible intervention. The patient is fully evaluated using MR imaging, if
MR imaging is not contraindicated. T2* imaging is used to assess for presence of hemorrhage. LVO is identified
with MRA. Ischemic core size is determined with DWI. A DWI lesion of less than 70 to 100 mL generally makes the
patient eligible for thrombectomy.

the presence of symmetric collaterals is a reliable to 100 mL of infarct core on DWI. Patients with
marker of a small (<70 mL) core lesion as DWI volumes approaching 100 mL may be
measured by DWI. If the collateral circulation is excluded based on other factors, such as age
clearly decreased compared with the normal and baseline functional status, given that both
side, the diffusion lesion may be small or large. clinical and imaging parameters must be met to
Thus, for any patient for whom therapeutic ques- qualify for treatment.
tions persist after CT scanning, an emergent MR
imaging is performed to obtain core measurement Estimating the Size of the Core Revealed by
using DWI. Diffusion-Weighted Imaging
We use the A*B*C/2 method to estimate the vol-
Presentation 6 to 24 Hours or Unknown Time
ume of the diffusion abnormality on the DWI
from Onset
scan.11 The method is simple. At the level at which
For all patients presenting in time windows after the DWI lesion appears largest, anteroposterior (A)
6 hours, our protocol uses MR imaging as the and transverse lengths (B) are measured. The
sole modality (see Fig. 3B). This is the second superoinferior extent (C) is estimated by counting
major change to our stroke imaging protocol the number of slices in which the diffusion abnor-
and it follows from the DAWN1 and DEFUSE mality is present, taking into account the slice
310 results. T2* imaging is performed using a thickness and slice gap. The product of A times
gradient echo sequence, which takes less than B times C is divided by 2. This simple method is
1 minute; a DWI scan is acquired in less than accurate to approximately 10% and has been vali-
2 minutes; and either a 2D TOF MRA of the dated.12 An alternative is the use of commercial
head or a ceMRA of the head and neck is per- data processing software that automatically de-
formed within 2 minutes. We use a core-clinical tects and measures the DWI lesion volume. We
mismatch, based on information generated by have used such software, and have found it to
our center and most recently confirmed in the be generally reliable, but the software may pro-
DAWN trial. Patients are considered for throm- duce spurious, inaccurate DWI lesion volumes
bectomy if they have a clinical deficit producing and visual inspection of the measurements is
an National Institutes of Health Stroke Scale thus extremely important. The software is conve-
greater than or equal to 6, have an LVO on nient, but expensive, and so we use the manual
contrast-enhanced MRA, and have less than 70 method.

Descargado para Anonymous User (n/a) en Pontifical Xavierian University de [Link] por Elsevier en agosto 15, 2019.
Para uso personal exclusivamente. No se permiten otros usos sin autorización. Copyright ©2019. Elsevier Inc. Todos los derechos reservados.
 MR Imaging Selection of Acute Stroke Patients 579

Assessing the Core in Patients Who Are and from those to derive a penumbra/core
Unable to Undergo MR imaging mismatch ratio.
A very interesting subgroup analysis of the
In our experience, 10% to 15% of patients cannot
DEFUSE 3 trial was presented by Dr Maarten
undergo MR imaging. The most common reason is
Lansberg at the 2018 International Stroke Confer-
the presence of a pacemaker. The decision to pro-
ence.14 He reported that when considered alone,
ceed to thrombectomy relies on the clinical cir-
the 49 patients who were selected using MR imag-
cumstances as well as review of the noncontrast
ing inclusion criteria had a significantly higher per-
CT, parenchymal images from the CTA (CTA
centage of good outcomes compared with the 182
source images), and the CTA. We do not use CT
patients in the entire DEFUSE 3 cohort that
perfusion (CTP). Our studies and those from other
included patients selected by either MR imaging
groups have confirmed that although CTP may
or CTP criteria. Additionally, statistically significant
reliably estimate small core lesions, it is unreliable
differences between the thrombectomy and medi-
in individual patients with larger core lesions,
cally treated groups were found despite the small
although it may used for group studies.9 Because
size of the MR imaging–selected cohort. Subgroup
we do not know the size of the core a priori, it is
analysis of the CTP-selected cohort alone was not
prudent NOT to use this method.
presented.
These late window trials demonstrated the
MR IMAGING UTILIZATION IN ISCHEMIC validity of the “tissue-clock” concept, a refinement
STROKE TRIALS of the simplified “time-clock” model of stroke care.
The scientific foundation for our approach includes Although time is brain, each brain has its own
our own studies and published work from other time. The clinical superiority of MR imaging for
centers. In recent years, there has been a plethora patient selection, especially in the later time win-
of important studies. We review the most pertinent dows, leads to the conclusion that Comprehensive
studies here. Stroke Center requirements should include imme-
diate MR imaging access for all patients with acute
PROSPECTIVE OBSERVATIONAL TRIAL USING stroke.
MR IMAGING FOR PATIENT SELECTION
SLOW PROGRESSORS AND
Major changes in stroke care are occurring COMPREHENSIVE/THROMBECTOMY
following new data available for patients with CAPABLE STROKE CENTERS
LVOs. At the time of this writing, the Food and
Drug Administration (FDA) has approved at least The DAWN and DEFUSE 3 trials have opened the
1 device for thrombectomy up to 24 hours after therapeutic window for endovascular thrombec-
stroke onset. The trials evaluating early window tomy for up to 24 hours after known stroke onset
therapies were overwhelmingly CT-based trials. or for up to 24 hours after the patient was last
Advanced imaging selection was achieved using seen well (unknown stroke onset). As “late” treat-
CTA and CTP imaging in several trials. Fewer ment of patients with LVO becomes standard of
than 5% of the total patient population was care, comprehensive (or thrombectomy capable)
imaged using MR imaging. Patient selection by stroke centers must consider how they should
MR imaging in the SWIFT Prime trial (in which modify their LVO patient evaluation programs to
19.7% were imaged with MR imaging) had a nearly accommodate more patients. The individual pa-
identical treatment response to the 80.3% tient’s stroke physiology, especially the size of
selected using CTP.13 the infarct core, is of paramount importance in pa-
Trials evaluating patients in later time windows tients who have unknown stroke-onset time, or are
(more than 6 hours after or unknown time of onset) evaluated 6 to 24 hours after stroke onset. MR im-
had higher rates of MR imaging utilization. In the aging should be the primary imaging modality for
DAWN trial, 37.4% of the intervention arm and these patients because of its superior perfor-
35.4% of the control arm were imaged using MR mance for infarct core measurement.
imaging. In the DEFUSE 3 trial, 25% of the throm- In light of this new therapeutic window exten-
bectomy patients and 28.9% of the control arm sion, stroke centers need to consider the number
were selected using MR imaging. In these trials, of potential patients that may be eligible for treat-
MR imaging parameters were different. The ment. A significant number may be initially evalu-
DAWN trial used DWI to quantify core and applied ated at other institutions that are not equipped to
a core-clinical mismatch, whereas the DEFUSE 3 perform thrombectomy. Although data are scarce,
trial used DWI to quantify core and thresholded there may be a large number of “slow progres-
MR perfusion (Tmax >6) to quantify penumbra sors,” which we define here as patients who

Descargado para Anonymous User (n/a) en Pontifical Xavierian University de [Link] por Elsevier en agosto 15, 2019.
Para uso personal exclusivamente. No se permiten otros usos sin autorización. Copyright ©2019. Elsevier Inc. Todos los derechos reservados.
580 Leslie-Mazwi et al

have proximal LVOs with ischemic core lesions Statistically and by visual inspection, there was
smaller than 70 mL at 6 or more hours after stroke no significant relationship between the ischemic
onset. Copen and colleagues15 first reported an core size and time after ictus. Moreover, most
unexpectedly large proportion of “slow progres- patients had core lesions smaller than 100 mL in
sors.” A histogram derived from that publication every time period (for example 6–9 hours or 15–
is shown in Fig. 4. A total of 109 consecutive pa- 18 hours). This study confirmed the earlier work
tients who underwent MR imaging studies, by Copen and colleagues.15
including diffusion and perfusion imaging, within When considering slow progressors, it is
24 hours of acute ischemic stroke symptom onset important to consider the rate of growth of the
were evaluated. Sixty-eight had LVOs. Unexpect- core ischemic lesion. Clearly, there is a large
edly, more than 60% of the patients with LVOs variability in core growth when one considers
imaged at both less than and greater than 9 hours the data of Figs. 5 and 6. In experimental stroke
had small infarct cores and diffusion/perfusion studies that have used DWI to measure infarct
mismatches of greater than 160%. What was volumes, the ischemic core grows in a logarith-
perhaps even more surprising, there was no rela- mic fashion, with rapid early growth, which then
tionship between time from stroke symptom onset tapers to a slower rate. Although it is likely that
and the presence of the large mismatch. the ischemic core growth rates in people is simi-
The study by Copen and colleagues15 motivated larly logarithmic, data supporting this are rare
another study to assess the size of the DWI core because it requires that multiple imaging studies
lesion in patients with LVO at the time of presenta- be performed over a short period of time. Never-
tion in the emergency department. Hakimelahi and theless, in a series of 14 patients with acute
colleagues16 reported the ischemic core volumes ischemic stroke with LVOs imaged at a mean
for a series of 186 consecutive patients with LVO baseline of 7.5 hours as well as 4 hours later,
who presented up to 24 hours after stroke onset. 24 hours later, and 1 week later, we observed
A plot of the DWI lesion volume with respect relatively little growth from the baseline scan to
to time after stroke onset is shown in Fig. 5. the 1-week scan.17 Similarly, Fig. 6 shows
the DWI and MTT images of a patient with a left
MCA stem occlusion who was imaged with
MR imaging 3 times over 24 hours. Close

Fig. 4. Presence of a large mismatch in patients with

LVO up to 24 hours after onset. The criterion for a
large mismatch was a perfusion (MTT)/diffusion Fig. 5. Ischemic core size versus time in patients with
(DWI) mismatch of greater than 160%. Patients were LVO. Scatterplot of consecutive patients who pre-
grouped into those who had the presence or absence sented to the emergency department with anterior
of LVO when they underwent MR imaging. Patients in circulation large artery occlusions. The plot shows
each group were further subdivided by the time of the size of the DWI lesion abnormality with respect
imaging after stroke onset (less than or >9 hours). to time last seen well. There is no correlation between
More than 60% of all patients with LVO had large ischemic core size and time after stroke onset. Most
mismatches irrespective of the time after ictus. patient core sizes are below the 70 to 100 mL blue
(Adapted from Copen WA, Rezai Gharai L, Barak ER, bar when considered as a whole or any time interval
et al. Existence of the diffusion-perfusion mismatch from 0 to 24 hours. (Adapted from Hakimelahi R,
within 24 hours after onset of acute stroke: Vachha BA, Copen WA, et al. Time and diffusion
dependence on proximal arterial occlusion. Radiology lesion size in major anterior circulation ischemic
2009;250(3):878–86; with permission.) strokes. Stroke 2014;45(10):2936–41.)

Descargado para Anonymous User (n/a) en Pontifical Xavierian University de [Link] por Elsevier en agosto 15, 2019.
Para uso personal exclusivamente. No se permiten otros usos sin autorización. Copyright ©2019. Elsevier Inc. Todos los derechos reservados.
 MR Imaging Selection of Acute Stroke Patients 581

Fig. 6. Serial MR imaging scans in patient with persistent left MCA occlusion. The patient was initially scanned
approximately 6 hours after last seen well. The artery occlusion was documented on CTA. Repeat imaging per-
formed approximately 12 and approximately 24 hours after stroke onset showed persistent occlusion and large
perfusion deficit (MTT), but very little growth in the size of the ischemic core revealed by DWI.

examination of the DWI images shows that most Advantages of MR Imaging for Stroke
of the lesion volume is already present at base- Imaging
line with very slow growth over 24 hours, despite
MR imaging provides the undisputed advantage of
the persistence of an LVO and a large diffusion
detailed characterization of the ischemic core
perfusion mismatch.
lesion. Detection of ischemia is much improved
In summary, the available data indicate that
with MR imaging compared with CT, with a sensi-
there is wide variation in the growth of ischemic
tivity of greater than 90% for DWI compared
core in patients with anterior circulation LVOs.
with approximately 60% for noncontrast CT
The growth rate is likely similar to that observed
scans. CTP increases the detection of ischemia
in experimental animal studies, with rapid early
compared with CT, but remains inferior to DWI.
followed by slower later growth that has a loga-
Furthermore, measurement error between intra-
rithmic character. Moreover, the data suggest
rater and interrater lesion volumes is much lower
that most patients with LVOs are “slow progres-
for DWI compared with thresholded CTP. How-
sors,” defined as patients with ischemic core
ever, true volume thresholds to define excellent
volumes of less than 70 to 100 mL when
versus poor outcomes remain a shifting target. A
imaged 6 or more hours after stroke onset. If
lesion volume less than 70 mL has been associ-
true, this presents an enormous opportunity to
ated with good outcomes when the patient un-
benefit many patients with LVOs using mechani-
dergoes thrombectomy.2 For volumes greater
cal thrombectomy.

Descargado para Anonymous User (n/a) en Pontifical Xavierian University de [Link] por Elsevier en agosto 15, 2019.
Para uso personal exclusivamente. No se permiten otros usos sin autorización. Copyright ©2019. Elsevier Inc. Todos los derechos reservados.
582 Leslie-Mazwi et al

than 100 mL, over 90% will have less favorable Table 2
outcomes (modified Rankin score 3–6) even with MR imaging protocol for patients without
thrombectomy. An additional concern has been preceding CT/CTA
increased risk of intracranial hemorrhage (possible
harm, not just absence of benefits). These notions Target Sequence Time Comments
have come under recent scrutiny based on sub- Hemorrhage EPI GRE 1 min —
group analysis from randomized trials and sepa-
Occlusion 2D TOF 1 min —
rate reports of patients with larger core volumes
achieving either good or improved functional out- Core DWI 2 min —
comes. Beyond infarct volume alone, the precise Neck vessels Gd1 3D TOF 2 min Optional
anatomic localization that DWI sequences provide Other FLAIR 2 min Optional
aids in determining culprit vascular compromise pathology
as well as prognosis.
Abbreviations: CT, computed tomography; CTA,
MR imaging provides the advantage of vascular computed tomography angiography; D, dimensional;
imaging without the need for contrast media. TOF DWI, diffusion-weighted imaging; EPI, echo planar imag-
MRA evaluation of the intracranial circulation al- ing; FLAIR, fluid-attenuated inversion recovery; Gd, gado-
lows identification of vascular occlusion for linium; GRE, gradient recalled echo; TOF, time of flight.
patients with poor renal function for whom gado-
examination must be tailored to each individual
linium is contraindicated or CT contrast may be
patient. One of the most common scenarios is a
harmful. Use of a 2D as opposed to a 3D TOF
patient with an acute stroke syndrome who was
sequence allows for rapid image acquisition,
initially evaluated with CT and CTA. If CT imaging
and is probably the best choice in less coopera-
reveals an LVO and no hemorrhage, only the
tive patients. Because images of each slice are
DWI sequence is essential (see Table 1). The other
acquired serially and rapidly, individual images
common scenario is of the patient with acute
are not significantly degraded by motion and an
stroke syndrome who is imaged first by MR imag-
LVO can be readily easily identified. Imaging of
ing. Patients presenting more than 6 hours post
cervical vasculature is often challenged by
ictus or with an unknown time of onset will be typi-
patient motion and image resolution using TOF
cally imaged with MR imaging first. There are other
techniques.
scenarios. For example, a patient may be trans-
Finally, tailored imaging and the use of acceler-
ferred from a nearby care health facility where a
ated techniques that include parallel imaging and
head CT but not a CTA was performed. Such a pa-
EPI has allowed significant reductions in MR imag-
tient would go directly to MR imaging, but the se-
ing acquisition time; in fact, MR imaging se-
quences obtained would depend on the imaging
quences can now be obtained in time frames
findings, the patient transfer time, change in signs
that are comparable to the speed of CT scanning.
and symptoms, and other potential confounders.
For example, Nael and colleagues18 developed a
In general, we obtain DWI (to characterize infarct
relatively comprehensive 6-minute MR imaging
core), EPI GRE (to identify hemorrhage and LVO)
protocol for evaluation of acute ischemic stroke.
and 2D TOF through the circle of Willis (to identify
This rapid acquisition maintains a high degree of
LVO) on all patients who are imaged first by MR
image diagnostic quality and is increasingly used
imaging, with a total imaging time of less than
by centers that routinely apply MR imaging for
5 minutes.
acute stroke evaluation.
The MR imaging protocols adopted at MGH
Challenges with MR Imaging for Stroke
are listed in Tables 1 and 2. The MR imaging
Several limitations prevent more common use of
MR imaging. Scanner availability is variable in
Table 1 emergency departments worldwide, with access
MR imaging protocol for patients with to MR imaging much more limited than the avail-
preceding CT/CTA ability of CT scanning. However, all comprehensive
stroke centers (including thrombectomy-capable
Target Sequence Time Comments
stroke centers) should have MR imaging available,
Core DWI 2 min — even if it is not physically located in the emergency
Other pathology FLAIR 2 min Optional department.
Only 80% to 85% of patients with acute stroke
Abbreviations: CT, computed tomography; CTA,
computed tomography angiography; DWI, diffusion-
are candidates for MR imaging. True contraindica-
weighted imaging; FLAIR, fluid-attenuated inversion tions include the presence of retained metal (his-
recovery. torically this has been mainly because of cardiac

Descargado para Anonymous User (n/a) en Pontifical Xavierian University de [Link] por Elsevier en agosto 15, 2019.
Para uso personal exclusivamente. No se permiten otros usos sin autorización. Copyright ©2019. Elsevier Inc. Todos los derechos reservados.
 MR Imaging Selection of Acute Stroke Patients 583

pacemakers), and are present in approximately of tissue that permanently recovers is typically min-
10% of patients with stroke. The remaining 5% imal and not clinically significant. Despite these
to 10% of patients who are not candidates for drawbacks, MR imaging remains far superior to
MR imaging are excluded because they are too CT methods for core assessment.
unstable medically (eg, poor airway control, blood
pressure instability, repeated vomiting) or have se- FUTURE APPLICATIONS OF MR IMAGING
vere agitation. Unlike CT, screening is required to
ensure that a patient is appropriate to enter an Although the role of MR imaging in acute stroke
MR imaging scanner. In many centers, this is imaging continues to be a subject of debate,
accomplished most efficiently by prescreening certain developments should be anticipated in
where possible. However, for patients presenting the near future. Automation is likely to increasingly
with unknown medical histories and no family play a role in the interpretation of scans for pa-
members to provide screening data, either MR im- tients with stroke. At the moment, automation is
aging is not an imaging option, or scout radio- restricted to programs that process perfusion
graphs are required to exclude the presence of maps, but the potential for machine-learning algo-
ferromagnetic metals. rithms to be able to identify affected tissue volume,
The need for prescreening is coupled with the hemorrhage, the presence of an LVO, and other
need for patient safety in the MR imaging environ- salient aspects of MR imaging in acute stroke is
ment. This may require the removal of electrocar- large. It is anticipated that automation augmented
diogram leads placed by ambulance staff or with artificial intelligence will extend beyond
referring hospital emergency departments, chang- merely descriptive content of the scan to assis-
ing infusion pumps or priming infusion lines, appli- tance with treatment decisions and prognosti-
cation of MR imaging safe pulse oximetry, and so cation, by incorporating various other patient
forth. These requirements introduce delay. Given characteristics and using extensive comparative
the powerful interaction between time and treat- historical data.
ment effect for both intravenous and endovascular Accelerated scanning is also likely to improve.
therapies in patients with stroke presenting within There continues to be development of faster MR
the first few hours, the impact of these delays may imaging methods, such as parallel imaging. In
be a concern. However, it has been demonstrated addition, the expected increases in processing po-
that MR imaging can be sufficiently efficient to wer will shorten image reconstruction time. These
be the primary imaging method even for tPA incremental improvements will progressively
administration.6 shorten the time difference between MR imaging
Although MR imaging can reliably detect hyper- and CT imaging, and minimize the exclusion of pa-
acute intraparenchymal hemorrhage, it may not be tients from MR imaging due to medical instability
as reliable for detecting hyperacute blood prod- or agitation. Furthermore, increasing attention
ucts that contain predominantly oxyhemoglobin will be paid to streamlining the logistics of safely
within the cerebral spinal fluid. The diamagnetic obtaining MR imaging scans on patients with
properties of oxyhemoglobin give it minimal sus- acute stroke.
ceptibility effect. For parenchymal hematomas, Furthermore, MR imaging contraindications will
this is not a concern, but in the subarachnoid diminish. Concerns related to pacemakers may
space where dilution of blood products by cere- be overstated. A total of 189 brain MR imaging
brospinal fluid and high oxygen tension further scans performed in 123 patients with pacemakers
reduce the sensitivity of susceptibility-weighted or implantable cardioverters/defibrillators resulted
imaging, the possibility of missing a hemorrhage in only 1 case of loss of pacing. There were no
is a persistent concern. The use of and additional deaths or system revisions.19 These findings
FLAIR sequence can minimize this concern. were confirmed in a second, much larger study
Although MR imaging provides a detailed volu- in which 1509 patients had 2103 MR imaging
metric analysis of the pretreatment infarct core, studies. One pacemaker had to be replaced.20
the possibility of DWI reversibility discussed previ- Additionally, recent pacemakers are being
ously should be remembered. This is particularly designed specifically to be safe in an MR imaging
true for patients who are candidates for mechanical scanner.21 If pacemakers are considered accept-
thrombectomy in very early time frames (<3 hours). ably safe for emergent MR imaging in the future,
After endovascular therapy, as many as one-third the proportion of patients eligible for MR imaging
of patient MR imaging scans obtained immediately scanning will increase.
afterward may demonstrate DWI reversibility; how- Finally, the precision of the information avail-
ever, most of the brain tissue demonstrating this able by MR imaging is anticipated to provide op-
phenomenon progresses to infarction. The volume portunities for truly individualized stroke therapy.

Descargado para Anonymous User (n/a) en Pontifical Xavierian University de [Link] por Elsevier en agosto 15, 2019.
Para uso personal exclusivamente. No se permiten otros usos sin autorización. Copyright ©2019. Elsevier Inc. Todos los derechos reservados.
584 Leslie-Mazwi et al

Current decisions in acute stroke are based on based approach. J Neurointerv Surg 2013;
population-level data. The nuances and intricacies 5(Suppl 1). i7–12.
of the individual patient and their response to ther- 9. Schaefer PW, Souza L, Kamalian S, et al. Limited
apy will be best understood with detailed reliability of computed tomographic perfusion acute
information about their imaging profiles. This infor- infarct volume measurements compared with
mation is available from MR imaging more consis- diffusion-weighted imaging in anterior circulation
tently than from CT. stroke. Stroke 2015;46(2):419–24.
10. Albers GW, Marks MP, Kemp S, et al. Thrombectomy
SUMMARY for stroke at 6 to 16 hours with selection by perfusion
imaging. N Engl J Med 2018. [Link]
MR imaging in acute ischemic stroke provides 1056/NEJMoa1713973.
specific insights into the pathophysiology of the
11. Sims JR, Gharai LR, Schaefer PW, et al. ABC/2
disease in the individual patient. It is widely avail-
for rapid clinical estimate of infarct, perfusion,
able and the time needed to perform MR imaging
and mismatch volumes. Neurology 2009;72(24):
is similar to that of CT imaging. Clinical application
2104–10.
of this imaging modality should bear in mind the
12. Luby M, Hong J, Merino JG, et al. Stroke mismatch
known limitations, and further insights from future
volume with the use of ABC/2 is equivalent to plani-
randomized controlled trials will be helpful in
metric stroke mismatch volume. AJNR Am J Neuro-
this regard. However, MR imaging parameters
radiol 2013;34(10):1901–7.
for stroke treatment selection will be an impor-
13. Menjot de Champfleur N, Saver JL, Goyal M, et al.
tant component of future tissue-based decision
Efficacy of stent-retriever thrombectomy in magnetic
making and truly personalized acute stroke
resonance imaging versus computed tomographic
therapeutics.
perfusion-selected patients in SWIFT PRIME trial
(Solitaire FR with the intention for thrombectomy as
REFERENCES
primary endovascular treatment for acute ischemic
1. Nogueira RG, Jadhav AP, Haussen DC, et al. Throm- stroke). Stroke 2017;48(6):1560–6.
bectomy 6 to 24 hours after stroke with a mismatch 14. Lansberg MG, Investigators D. Subgroup analyses
between deficit and infarct. N Engl J Med 2018; of the DEFUSE 3 study International Stroke Confer-
378(1):11–21. ence. Los Angeles, CA, January 24–26, 2018.
2. Yoo AJ, Chaudhry ZA, Nogueira RG, et al. Infarct 15. Copen WA, Rezai Gharai L, Barak ER, et al. Exis-
volume is a pivotal biomarker after intra-arterial tence of the diffusion-perfusion mismatch within 24
stroke therapy. Stroke 2012;43(5):1323–30. hours after onset of acute stroke: dependence on
3. Schaefer PW, Grant PE, Gonzalez RG. Diffusion- proximal arterial occlusion. Radiology 2009;250(3):
weighted MR imaging of the brain. Radiology 878–86.
2000;217(2):331–45. 16. Hakimelahi R, Vachha BA, Copen WA, et al. Time
4. Schaefer PW, Hassankhani A, Putman C, et al. Char- and diffusion lesion size in major anterior circulation
acterization and evolution of diffusion MR imaging ischemic strokes. Stroke 2014;45(10):2936–41.
abnormalities in stroke patients undergoing intra- 17. Gonzalez RG, Hakimelahi R, Schaefer PW, et al. Sta-
arterial thrombolysis. AJNR Am J Neuroradiol bility of large diffusion/perfusion mismatch in ante-
2004;25(6):951–7. rior circulation strokes for 4 or more hours. BMC
5. Campbell BC, Purushotham A, Christensen S, et al. Neurol 2010;10:13.
The infarct core is well represented by the acute 18. Nael K, Khan R, Choudhary G, et al. Six-minute
diffusion lesion: sustained reversal is infrequent. magnetic resonance imaging protocol for evaluation
J Cereb Blood Flow Metab 2012;32(1):50–6. of acute ischemic stroke: pushing the boundaries.
6. Labeyrie MA, Turc G, Hess A, et al. Diffusion lesion Stroke 2014;45(7):1985–91.
reversal after thrombolysis: a MR correlate of early 19. Strom JB, Whelan JB, Shen C, et al. Safety and utility
neurological improvement. Stroke 2012;43(11): of magnetic resonance imaging in patients with car-
2986–91. diac implantable electronic devices. Heart Rhythm
7. Leslie-Mazwi TM, Hirsch JA, Falcone GJ, et al. En- 2017;14(8):1138–44.
dovascular stroke treatment outcomes after patient 20. Nazarian S, Hansford R, Rahsepar AA, et al. Safety
selection based on magnetic resonance imaging of magnetic resonance imaging in patients with
and clinical criteria. JAMA Neurol 2016;73(1):43–9. cardiac devices. N Engl J Med 2017;377(26):
8. Gonzalez RG, Copen WA, Schaefer PW, et al. 2555–64.
The Massachusetts General Hospital acute stroke 21. Mitka M. First MRI-safe pacemaker receives condi-
imaging algorithm: an experience and evidence tional approval from FDA. JAMA 2011;305(10):985–6.

Descargado para Anonymous User (n/a) en Pontifical Xavierian University de [Link] por Elsevier en agosto 15, 2019.
Para uso personal exclusivamente. No se permiten otros usos sin autorización. Copyright ©2019. Elsevier Inc. Todos los derechos reservados.