CREST-H sites will be drawn from actively-enrolling CREST-2 centers. The additional
information obtained in CREST-H in this protocol will be the MR or CT imaging, which will be
done at baseline for all patients enrolled in CREST-H, and at 1 year for those with
hemodynamic impairment at baseline. The protocol will recommend an unblinded investigator at
each CREST-H site who will order and obtain the study-related MRI sequences, upload
de-identified image files to the CREST-2 central imaging site at U Maryland, and maintain
blinding of the hemodynamic imaging data for their site (see Blinding below). Image analysis
will be done at UCLA (for perfusion studies) and at Mayo Rochester for structural scans --
silent infarcts, WMH and microbleeds. Data management and statistical analysis will be done
at UAB, which serves this role for CREST-2.
Cognitive Assessment.
CREST-H will use existing CREST-2 cognitive assessment infrastructure -- the Survey Research
Unit at University of Alabama Birmingham. Added to the current CREST-2 battery is the Oral
Trail Making A & B as an additional measure of executive function, which will be administered
to every CREST-2 patient, regardless of their participation in CREST-H. Cognitive assessments
in CREST-H must take place prior to revascularization or within two weeks after assignment to
medical therapy alone. Testing in CREST-2 is repeated at 1 year, and every year thereafter up
to 4 years. At each test interval, a composite (mean) Z-score is derived from published
normative samples for each test outcome. The CREST-H primary outcome will be at 1 year in
which the change in the composite Z-score from baseline will be calculated. Covariates will
include age, education and depression. The test battery will be administered by a blinded
assessor the same way for all CREST-2 and CREST-H enrolled patients. The cognitive domains
being assessed in CREST-2/H are entirely consistent with those encompassed within the NINDS
Common Data Elements (CDE).
Imaging protocol.
Multimodal MRI or CT perfusion imaging will be performed at baseline on every study subject.
Multimodal MRI, including routine parenchymal sequences and PWI utilizing dynamic
susceptibility contrast technique, will be acquired at each participating CREST H site who
have been approved for this imaging modality. CT perfusion, using iodinated contrast, will be
used as an alternative PWI image for sites approved for this modality. Imaging will take
place within 14 days after CREST-H enrollment and prior to any CREST 2 intervention for those
randomized to CEA or CAS.
Standardized contrast agent injection protocol, appropriate preparation, and IV setup is used
to ensure good scan quality. An antecubital vein IV catheter of 18-20 gauge is required. A
test injection will be performed with approximately 10 ml of normal saline solution.
MRI image acquisition DWI/ADC (b=0, 1000 s/mm2 applied in each of three principal gradient
directions), FLAIR, high-resolution T1, and GRE sequences will be acquired on 1.5-3.0 T
scanners equipped with echo-planar imaging capability, using standard clinical protocols at
participating CREST-H sites Total scanning time will be approximately 40 minutes. PWI
acquisition protocol will be standardized across all CREST-H sites, using sequential
T2*-weighted (gradient echo) EPI time sequence scanning. A modified 2-phase contrast
injection scheme will be used to perform CEMRA and DSC perfusion imaging, without need for
additional contrast.
CT perfusion imaging will follow the protocol outlined in the updated Imaging MOP for
CREST-H. The CT perfusion study is identical to the clinical CTP protocol used for acute
stroke imaging in most institutions.
Perfusion imaging analysis.
PWI source images will be sent to the core laboratory at UCLA and processed with the
OleaSphere software platform, using deconvolution of tissue and arterial signals in an
expedited manner, yielding standardized data regardless of the acquisition system at each
site. Hemodynamic impairment is defined as TTP >1.25 sec in the middle cerebral artery and
anterior cerebral artery territories of the ipsilateral hemisphere to the carotid lesion
compared with the same territory in the opposite hemisphere. Longitudinal analyses will
investigate the change in the TTP >1.25 sec lesion at 1 year comparing the revascularization
versus the medical only arm. Continuous values for this volumetric change will be used to
calculate the correlation between degree of cognitive change and degree of perfusion change.
The continuous Tmax variable, as well as standard perfusion parameters of CBV, CBF Tmax, and
MTT will be analyzed on the serial imaging studies in each case with MR imaging.
Co-registered, voxel-based changes in serial perfusion values will also be explored with
multiparametric (e.g. CBV, CBF, Tmax, MTT) values.
Image de-identification and blinding.
All MR or CT image files will be de-identified under the supervision of an unblinded
Investigator (UI) at each institution and uploaded to the CREST-2 Imaging Core site at U
Maryland. In order to assure that the PWI scan information from CREST-H does not compromise
the integrity of the parent trial, the results of the perfusion scan will be blinded to the
investigator team.
Image transfer.
Participating sites will utilize the same file transfer protocol (ftp) to transfer images to
U Maryland for CREST-H as is already established for CREST-2. The images will be stored in a
HIPAA-compliant, firewall protected server within the U Maryland archival system. A CREST-2
dedicated ftp linkage between the VIC at U Maryland and UCLA; and the VIC at U Maryland and
Mayo-Rochester will be utilized to make each perfusion image file available for download by
UCLA (Liebeskind lab) and each structural image (DWI, FLAIR, GRE) by Mayo-Rochester (Huston
lab).
MRI structural analysis.
Structural MRI analysis at Mayo-Rochester will utilize NIH NINDS Common Data Elements
developed for the CREST-2 grant. The following definitions apply:
1. silent infarct --- non-confluent hyperintense lesion >1mm on FLAIR sequence on 1-year
MRI not present on baseline FLAIR MRI.
2. Cerebral microbleed - hypointense 1-2mm non-confluent lesion on baseline GRE sequence.
3. WMH volume -- White matter hyperintensity volume refers to confluent periventricular
high intensity lesions on FLAIR imaging, and will be derived using an automated T2 WMH
quantification at the Huston lab.
Data from image analysis (TTP delay, WMHV, silent infarct count, microbleed count) performed
at UCLA and Mayo-Rochester will be entered electronically on CREST-H data forms via the
CREST-2 SDCC website at UAB, where it will be stored on a separate webpage linked to the rest
of the CREST-2 data, including baseline and yearly cognitive assessments. The electronic data
entry system (eDES) for CREST-2 is a mature system, successfully reviewed by FDA audit in
other studies, providing standard approaches for entry-confirmation-locking of data forms,
and supporting range and validity checking for data provided.
. Analysis.
Specific Aim 1. To determine whether cognition can be improved by revascularization among a
subset of CREST-2 patients with hemodynamic impairment at baseline.
The primary hypothesis is to assess if the magnitude of the treatment differences
(revascularization versus medical management alone) differs between those with flow failure
compared to those without flow failure using the Z-scored cognitive outcomes (C0, C(1). That
is, the primary hypothesis is an interaction hypothesis that will be assessed using linear
regression, specifically: (C1 - C0) = β0 + β1T + β2F + β3TF + β4C0 + (other covariates),
where C1 is the cognitive z-score at year 1, C0 the cognitive z-score at baseline, T the
treatment indicator variable, F the flow failure indicator variable, and βi the regression
parameters to be estimated. The parameter of interest for the primary hypothesis is then β3
that would assess if the magnitude of treatment difference in the change in cognitive score
between baseline and 1-year is similar for those with versus without flow failure.
Secondary Aims: To determine if the number of silent infarcts and white matter hyperintensity
volume at 1 year is different between the revascularization and the medical-only arms.
For the secondary aims we will calculate the number of new silent cerebral infarctions
occurring over the first year, and the change in the WMH volume. The approach for analysis of
the number of new silent infarcts will depend on the average number and distribution of the
number of new infarcts. The analytic approach will be linear regression if the number of new
infarcts is large (considered more likely the case), or Poisson Regression if the number is
smaller (considered less likely the case). The analysis of the change in WMH will use a
linear regression approach.