Keywords intracranial stent - in-stent stenosis - magnetic resonance angiography - MRA - digital
subtraction angiography
Introduction
In-stent stenosis is a well-described delayed complication of angioplasty and stenting
for atheromatous disease.[1 ]
[2 ] Detection of high-risk in-stent restenosis would identify a subgroup of patients
who may benefit from close observation/early treatment. Digital subtraction angiography
(DSA) is the gold standard imaging modality for follow-up of stent patency, but is
invasive in nature and associated complication rate (range: 1.8–2.1%) would limit
its use for repeated and long-term evaluation. Various noninvasive imaging modalities
are described for in-stent stenosis evaluation. However, exact accuracy of each modality
is not widely studied.
Based on previous evidence of SSYLVIA trial,[1 ] in which stenosis > 50% was grouped as high-risk group, the delayed in-stent stenosis
in our study also defined > 50% delayed in-stent stenosis as “significant stenosis”
requiring at least clinical attention/close follow-up/intervention. Hence, we aimed
to study accuracy of contrast-enhanced magnetic resonance angiography (CE-MRA) and
time-of-flight (TOF)-MRA in detection of significant delayed in-stent stenosis as
compared with DSA.
Material and Methods
Patient Selection
The study was a retrospective–prospective evaluation of patients who underwent endovascular
stent placement in intracranial circulation (defined as cavernous internal carotid
artery and beyond, V4 vertebral artery and beyond), who had follow-up MRA and DSA
during January 2009 to December 2015. Patients treated with covered stent and stented
for atherosclerotic stenosis were excluded from the study. In addition, interval of
more than 1 month between MRA and DSA were excluded. Medical records were reviewed
for patient demographic details such as age, sex, clinical presentation, DSA and MRA
study, treatment given including medical and endovascular therapy, and follow-up.
A few patients' MRA was retrospectively evaluated, while a few patients were prospectively
followed with DSA/MRA.
Image Acquisition
MR imaging of the brain was performed using a 12-channel matrix coil on a 1.5 T clinical
scanner (Avanto, SQ Engine; Siemens, Erlangen, Germany). Digital subtraction angiogram
was obtained using a biplane neuro angiographic suit (Innova 3131, GE Medical systems,
LLC, Milwaukee, Wisconsin, United States) with standard protocols, by transfemoral
route and selective acquisition of vessels of interest in standard angiographic projections
by iodinated contrast injection (Omnipaque 320; GE Healthcare, Princeton, New Jersey,
United States).
Image Analysis
Four sets of MR images (TOF-MRA volume-rendered images, TOF-MRA source images, CE-MRA
volume-rendered images, CE-MRA source images) were evaluated independently and in
a randomized fashion by two experienced neuroradiologists (experience of reader 1:
10 years and reader 2: 5.5 years) and any deviation was resolved by consensus. Percentage
stenosis was measured by using following formula, percent stenosis = (1 − smallest
in-stent diameter/diameter at the proximal/distal normal vessel segment) × 100. This
calculation is a one-dimensional view. If there are poor quality images in any one
of the sets, the entire patient image sets were excluded from the study. We defined
“significant in-stent stenosis” as ≥ 50% stenosis requiring clinical attention and
further evaluation/intervention. “Delayed” in-stent stenosis was defined as in-stent
stenosis 3 months after stent deployment. The stenosis in five sets of images (1 DSA + 4
MRA) was grouped into < 50% and ≥ 50% stenosis for statistical analysis.
Statistical Analysis
A commercially available statistical software package (Statistical Package for the
Social Sciences, Version 17; SPSS, Chicago, Illinois, United States) was used for
analysis. Fisher's exact test was used for statistical significance evaluation. The
p value of < 0.05 was regarded as a statistically significant result. Sensitivity,
specificity, accuracy, and positive and negative predictive values of different imaging
sets against DSA were calculated. Statistical significance of CE-MRA source images
was also calculated between Leo stent (Balt, Montmorency, France) and Neuroform stent
(Stryker, Kalamazoo, Michigan, United States).
Results
Overall, 38 patients with intracranial stent deployment for various etiologies were
identified. A total of 18 stented patients for nonatherosclerotic etiology had follow-up
DSA; all the 18 patients had at least one follow-up MRA, 6 patients had two follow-up
MRA, 1 each had three and four follow-up MRAs. [Table 1 ] summarizes the demographics of stented patients, [Fig. 1 ] describes in-stent stenosis detection in different imaging modalities as compared
with DSA.
Fig. 1 Comparison of various MRA images with DSA. Preprocedure DSA—saccular aneurysm in
the communicating segment of ICA (A ), Immediate postprocedure DSA after stent-assisted coiling (B ), follow-up TOF-MRA reformatted image—interruption of ICA at the level of stent (C ), TOF-MRA source image—no definite evidence of flow in the ICA (D ), CE-MRA reformatted image—interruption of ICA at the stented segment (E ), CE-MRA source image—more than 50% stenosis of ICA (F ), postprocedure DSA—stenosis of ICA at the proximal part of stent (G ). CE, contrast-enhanced; DSA, digital subtraction angiography; ICA, internal carotid
artery; MRA, magnetic resonance angiography; TOF, time-of-flight.
Table 1
Demographics of stented patients
Age distribution (N = 18)
Mean age was 41.5 y (range 12–80 y)
Sex distribution of patients (N = 18)
M:F = 1:1
Duration of follow-up (N = 18)
Mean follow-up 17.5 mo (range 4–62 mo)
Etiology for stenting (
N
= 18)
Stent-assisted coiling
16 (88.9%)
Direct CCF
1 (5.5%)
Nonatherosclerotic stenosis
1 (5.5%)
Location of stents (
N
= 18)
Anterior circulation
11 (60%)
Posterior circulation
7 (40%)
Distribution of stents in various intracranial vessels
Cavernous ICA
3 (16%)
Communicating segment ICA
6 (33%)
MCA
2 (11%)
BA
3 (16%)
BA-PCA
2 (11%)
VA-basilar artery
1 (6%)
Type of stents used (
N
= 18)
Neuroform 3
8 (44%)
Leo
6 (33%)
Solitaire AB
3 (17%)
Enterprise
1 (6%)
Delayed in-stent stenosis in DSA
Neuroform 3
0
Leo
0
Solitaire AB
1
Enterprise
0
Delayed stent occlusion in DSA
Neuroform 3
0
Leo
1
Solitaire AB
0
Enterprise
0
Abbreviations: BA, basilar artery; CCF, carotid cavernous fistula; DSA, digital subtraction
angiography; ICA, internal carotid artery; MCA, carotid cavernous fistula; PCA, posterior
cerebral artery; VA, vertebral artery.
Overall delayed in-stent stenosis during follow-up DSA was 11%, which is slightly
high as compared with 7.8% of previously published literature.[3 ] CE-MRA source image are 100% sensitivity, specificity, positive predictive value,
and negative predictive value in detecting significant delayed in-stent stenosis (≥
50% stenosis). However, specificity and positive predictive value of all four MRA
image sets was 100%, the sensitivity and negative predictive value was significantly
different in four image sets. The sensitivity of TOF reformatted image, TOF source
image, CE-MRA reformatted image, CE-MRA source image are 33% (6/18), 55.6% (10/18),
77.8% (14/18), and 100% (18/18), respectively, while negative predictive value are
14.3% (2/14), 20% (2/10), 33% (2/6), and 100% (2/2), respectively ([Table 2 ]). Patient compliance of antiplatelet therapy was 100% in all 18 cases, indicating
the lesser role of platelet activation by stent as a direct cause for delayed in-stent
stenosis.
Table 2
Comparison of sensitivity, specificity, PPV, NPV compared against DSA for detection
of significant (≥ 50%) delayed in-stent stenosis
p Value
Sensitivity
Specificity
PPV
NPV
CE MRA SI
1.000
100%(18/18)
100% (2/2)
100% (18/18)
100% (2/2)
CE reformatted image
0.125
77.8% (14/18)
100% (2/2)
100% (14/14)
33% (2/6)
TOF MRA SI
0.008
55.6% (10/18)
100% (2/2)
100% (10/10)
20% (2/10)
TOF reformatted image
0.000
33%(6/18)
100% (2/2)
100% (6/6)
14.3% (2/14)
Abbreviations: CE, contrast-enhanced; DSA, digital subtraction angiography; MRA, magnetic
resonance angiography; NPV, negative predictive value; PPV, positive predictive value;
SI, source image; TOF, time-of-flight.
CE-MRA reformatted image, though less sensitive as compared with CE-MRA source image,
both these image sets showed no statistically significant difference. In addition,
the reformatted images were also statistically insignificant as compared with DSA
in detection of > 50% stenosis.
Discussion
Delayed in-stent stenosis is increasingly identified in different stents, though exact
treatment remains controversial. The reported incidence of Neuroform in-stent stenosis
is 5.8% in patients with symptomatic intracranial stenosis.[4 ] Intracranially stented patients need frequent and long-term follow-up for stent
patency evaluation. Hence, different noninvasive modalities are being evaluated for
delayed in-stent stenosis. Modification of regular computed tomography angiogram and
contrast enhanced trans cranial Doppler (CE-TCD) is found to be useful for intracranial
stenosis evaluation,[5 ]
[6 ] a case report of utility of CE-TCD in intracranial stent evaluation also been described.[7 ]
Though the utility of TOF-MRA and CE-MRA in in-stent stenosis has been described,
accuracy of each modality was not been evaluated in detail. In a recent study of various
MRA techniques for the evaluation of in-stent flow, CE-MRA showed better flow signals.[8 ] In our study, CE-MRA identified significant stenosis (≥ 50% stenosis + occlusion)
correctly in all cases. However, as compared with DSA, CE-MRA showed 7 (35%) cases
of less than 50% stenosis which is significantly less than previously reported 64%.
One of asymptomatic patient revealed 50% stenosis of Neuroform stent during follow-up
angiography; he was treated with tirofiban infusion. Following tirofiban infusion,
the stenosis reduced significantly. The other patient with Leo stent occlusion was
managed conservatively due to his asymptomatic clinical status and presence of other
comorbidities. There was no statistically significant difference in the involvement
between Neuroform and Leo stents.
In a series of 14 patients with symptomatic intracranial stenosis, quantitative MRA
(QMRA) showed low specificity and positive predictive value in detecting in-stent
stenosis compared with our series.[8 ] In this study, CE-MRA reformatted images showed better specificity and positive
predictive value, though sensitivity and negative predictive value is low ([Table 2 ]). As against QMRA, the CE-MRA technique is easy to perform as well for interpretation,
less subjected to artifacts from variable velocities.[9 ] Our study show high sensitivity and specificity of CE-MRA source image for detection
of significant delayed in-stent stenosis.
Limitations of the study were (1) small number of cases in the cohort, the findings
needs to support by larger study, (2) the study was retrospective–prospective study.
We suggest need for prospective blinded study. (3) CE-MRA showed 39% false positivity
for less than 50% stenosis. (4) The number of Solitaire AB (Covidien) and Enterprise
stents (Codman) were very less, hence direct comparison between these two stents against
other stents were not possible. (5) Larger field of view (including neck + brain)
used for CE-MRA image acquisition likely result in spread of acquired data over larger
pixel. This limitation could be overcome by using small FOV for intracranial circulation
only. This modified technique may further decrease false positive cases of delayed
in-stent stenosis, though this aspect was not studied in the current study.
Conclusion
Development of delayed in-stent stenosis in nonatherosclerotic intracranial stenting
is 11%. Even though CE-MRA source image analysis give false positive result in stenosis < 50%,
it is a good imaging tool for detection and follow-up of significant (> 50%) intracranial
in-stent stenosis.
