Professional Documents
Culture Documents
ORIGINAL ARTICLE
a
Department of Neuroradiology, CHU Grenoble, University Joseph-Fourier, BP 217, 38043 Grenoble cedex 9, France
b
Department of Anesthesiology and Critical Care, CHU Grenoble, University Joseph-Fourier, Grenoble, France
KEYWORDS Summary
Cerebral angiography; Background and purpose. — After subarachnoid hemorrhage (SAH), vasospasm is frequent and
Perfusion computed increases the risk of stroke and poor clinical outcome. The purpose of this study was to identify
tomography; the best perfusion parameters in perfusion-CT (PCT) able to predict vasospasm diagnosed by
Subarachnoid angiography after SAH.
hemorrhage; Methods. — Seventy-six patients with SAH were investigated by PCT and cerebral angiogra-
Vasospasm phy. Using regions of interest (ROI) on parametric maps of mean transit time (MTT), time to
peak (TTP), cerebral blood volume (CBV) and cerebral blood flow (CBF), PCT data were com-
pared to an arteriographic score in two categories (severe vasospasm: ≥ 50% and non-severe
vasospasm: < 50%) for each artery. Best PCT predictors of the arteriographic score were tested
using multiparametric logistic regression.
Results. — Among the 76 patients, PCT data were reliable in 65 patients. Twenty-seven patients
had a severe vasospasm. Logistic regression showed that MTT was the best predictor of the
arteriographic score. Using MTT, odds ratios having a vasospasm were superior to 3.1 and the
occurrence of a vasospasm was accurately predicted in 78.5 to 100%, depending on the artery
considered. However, no absolute value of the MTT could be identified to predict the occurrence
of vasospasm. In fact, abnormal values of MTT ranged from 123 to 221% (m = 146%) of the control
values.
Discussion and conclusions. — PCT may accurately identify severe vasospasm and might be used
as a convenient noninvasive imaging modality to monitor patients with SAH. When detected,
severe vasospasm could be confirmed and managed using angiography and endovascular treat-
ment, appropriately.
© 2010 Elsevier Masson SAS. All rights reserved.
∗ Corresponding author. Tel.: +33 4 76 76 54 99; fax: +33 4 76 76 58 92.
E-mail address: VLefournier@chu-grenoble.fr (V. Lefournier).
0150-9861/$ – see front matter © 2010 Elsevier Masson SAS. All rights reserved.
doi:10.1016/j.neurad.2010.03.003
Perfusion CT to quantify the cerebral vasospasm following subarachnoid hemorrhage 285
segments were compared with the same ones on the angiog- Imaging examinations
raphy performed at initial admission.
Angiographic sessions
PCT Twenty-seven patients had a severe vasospasm (41.5%)
Unenhanced CT were reviewed for grading the SAH accord- and 38 patients had non-severe vasospasm, among those
ing to the Fisher scale. PCT data were computed using 24 patients had no vasospasm. Notably, in the severe
Brain Perfusion® , Philips Medical Systems giving MTT, TTP, vasospam group, aneurysmal rupture concerned 16 Acoms
CBV, and CBF maps. For each parameter, quantitative val- aneurysms including one pericallosal, and 15 MCA bifurca-
ues were extracted in cerebral arterial territories, by tions, while in the non-severe vasospasm one, aneurysmal
drawing five regions of interest (ROIs) on each hemi- rupture concerned 12 posterior circulation aneurysms. Arte-
sphere symmetrically located. The ROIs were defined in rial segments demonstrating vasospasm were located on the
the anterior cerebral artery (ACA), middle cerebral artery ACA (59.1%), on the MCA (31.8%), and on the PCA (9.1%)
(MCA), and posterior cerebral artery (PCA) territories, as (Table 1).
well as in the junctional territories, the anterior junc-
tional (AJ) between MCA and ACA territories, and posterior
junctional (PJ) territories, between MCA and PCA territo- PCT sessions
ries. In patients without any vasospasm (n = 24), mean val-
The overall effective dose equivalent for the CT protocol ues ± standard deviation across ROI were for CBV:
was 1.84 mSv (1 mSv for unenhanced CT and 0.84 mSv for 3.43 ± 0.08 ml/100 g, CBF: 43.34 ± 1.1 ml/100 g/min,
PCT). MTT: 4.86 ± 0.07 s, TTP: 14.45 ± 0.24 s. For each perfusion
parameter and for each ROI, these values were compared
Statistical analysis to those measured in patients with a vasospasm below 50%
(n = 14). No difference could be detected between patients
without any vasospasm from those with a vasospasm below
Statistical analyses were conducted in two steps. First, to
50%. Thus, data from both groups were further pooled and
identify among the perfusion parameters measured across
labeled as patients without severe vasospasm (< 50%) in
ROIs those significantly different between arteriographic
order to be compared to those from patients with severe
score, ANOVA with repeated measures were performed,
vasospasm (≥ 50%).
post-hoc comparisons using Student T-tests and Mann-
Tables 1 and 2, and Fig. 1 summarize perfusion parame-
Whitney tests were conducted when appropriate. Second, to
ters across arterial territories. An illustrative case is shown
determine the reliability of PCT to predict the arteriographic
Fig. 2.
score, an unconditional logistic regression with stepwise
backward selection was used. The probability value for exit
in the logistic regression was set at 0.10. Adjusted odds
ratios with 95% confidence intervals were calculated. Sta- Comparison of PCT and angiography in the diagnosis of
tistical procedures were performed using SPSS 15.0® , with vasospasm
significance set for p < 0.05. Because of significant interaction between ‘‘arteriographic
score’’ and ‘‘ROI’’, post-hoc comparisons were conducted
for all parameters in all ROI. Perfusion parameters signifi-
Results cantly different between arteriographic score are listed in
Table 2. In case of vasospasm, between-subjects compar-
Patients population isons showed mainly significant increase of MTT in proper
arterial territory. Additionally, MTT in adjacent junctional
Among the 76 patients initially managed, 11 patients were territories were also increased in case of vasospasm of
excluded from further analysis because of unreliable PCT by MCA. TTP and CBV were increased in case of vasospasm
poor technique (mostly due to a poor bolus due to impaired of ACA and left MCA. CBF were identical, but in case of
hemodynamic), leaving 65 patients for the study, 24 men vasospasm of the right PCA where a significant decrease was
and 41 women with a median age of 50 years (range, 20—76 observed.
years). The CT-Fisher score was recorded as grade 4 (n = 45), These parameters were further used in binary logis-
grade 3 (n = 17), and grade 2 (n = 3). WFNS scores were grade tic regressions to predict the arteriographic score. Results
1 (n = 26), grade 2 (n = 25); grade 3 (n = 1), grade 4 (n = 12), of regressions are summarized in Table 1. For all arterial
grade 5 (n = 1). vasospasms but in left ACA, MTT measured in the proper
A total of 61 ruptured aneurysms were identified: one arterial territory was identified as the best predictor of the
internal carotid artery bifurcation, two carotid ophthalmic arteriographic score. In left ACA vasospasm, MTT in the ante-
arteries, 18 anterior communicating arteries (Acoms), three rior junctional territory was the best predictor. Compared to
pericallosal arteries, 16 MCA bifurcations, 14 posterior com- the control values, MTT in severe vasospasm ranged from 123
municating arteries (Pcoms), five basilar arteries tip, two to 217% (m = 149%). No clear-cut value could be identified to
posterior inferior cerebellar arteries, including 14 patients simply predict the occurrence of vasospasm. At the arterial
with multiple aneurysms (21%). The patients were treated level and using MTT, the odds ratios to have a vasospasm
either by clipping (n = 10) or coiling (n = 50), one patient ranged from 3.1 to > 106 . The occurrence of a vasospasm
died before aneurysmal treatment. Four patients had no was accurately predicted in 78.5 to 100%, depending on the
aneurysm detected by angiography. artery considered.
Perfusion CT to quantify the cerebral vasospasm following subarachnoid hemorrhage 287
Artery* Vasospasm ≥ 50%a Significant predictors (odds ratio [CI95%]) False positive/False negative
mean ± sd (control values)b (accuracy)c
In the present study, we showed that using perfusion-CT In clinical practice, vasospasm after SAH is routinely
(PCT), an increase of the mean transit time (MTT), ranging assessed using serial clinical examinations, TCD, and angiog-
from 123 to 221% (m = 146%), was the most accurate param- raphy within the first two week after the onset [1].
eter to predict severe vasospasm in 78.5 to 100%, depending TCD is the most widely used to detect vasospasm noninva-
on the artery considered. sively, although it has clear limitations. It has been reported
of each artery.
b Parameters in anterior junctional ROI were computed for MCA and ACA.
c Parameters in posterior junctional ROI were computed for MCA and PCA.
d p-value of post-hoc comparisons.
288 V. Lefournier et al.
Figure 1 Cerebral perfusion parameters for each cerebral artery. PCT parameters for each artery in proper and adjacent ROI
when severe vasospasm (red) and non-severe vasospasm (blue) (see Table 2 for significant comparisons).
as a highly specific but less sensitive test [22]. Moreover, in (SPECT) [27,28], positron emission tomography (TEP)
patients with disturbed autoregulation after SAH, induced [29,30], xenon enhanced CT [16,31], perfusion-weighted
hypertension could alter cerebral blood flow velocities [23]. MRI [32]. Diffusion-weighted MR imaging has also been
Indeed, increased blood flow velocities measured by TCD studied [33,34] and provided predictive markers of silent
do not correlate with brain perfusion values and cannot ischemic lesions and/or progression toward symptomatic
reliably distinguish between patients with vasospasm of var- ischemia in asymptomatic vasospasm [35].
ious severity [24]. Finally, although TCD may detect severe However, those methods of investigation are not easily
vasospasm, reliable conclusion could only be obtained for performed in daily clinical practice in patients with SAH
MCA [25,26]. In our series, almost 60% of vasospasm occurred requiring neurointensive care, long time procedure, and lim-
on the ACA. Thus, the use of TCD as a screening method is ited clinical access.
questionable. In such patients, CT is easy to perform in clinical prac-
In fact, many centers rely on cerebral angiography, an tice, rapid, readily available, convenient, and a less invasive
invasive technique, for the diagnosis of vasospasm [1]. alternative to catheter angiography. Combining cerebral
Although angiography is the gold standard imaging modality morphological examination, CT angiography and PCT, is a
to provide morphologic depiction of macroscopic vascula- major interest to manage vasospasm [16].
ture, it does not allow quantitative analysis on cerebral PCT has been validated to study brain perfusion [36,37].
perfusion. Moreover, it was reported to be a useful diagnostic tool to
investigate vasospasm [9,13—20].
Cerebral perfusion imaging In our study, PCT values in patients with severe
vasospasm, low CBF, high CBV and prolonged MTT, are in
Cerebral perfusion abnormalities during vasospasm have agreement with previous reports [13], especially concerning
already been studied with single-photon emission CT MTT as being the most sensitive parameter [16,19,20,38].
Perfusion CT to quantify the cerebral vasospasm following subarachnoid hemorrhage 289
Figure 2 Illustrative case. A 46-year-old woman, WFNS 3 Fisher 4, presenting with a sylvian hematoma after SAH related to a
right MCA ruptured aneurysm (clipped). Angiography performed 10 days after SAH shows vasospasm > 50% on the right MCA (black
arrows). At PCT, MTT is increased in the right AJ (dotted arrow) (7.76 s), MCA (arrow) (8.87 s), and PJ (large arrow) territories
(8.47 s), compared to the opposite AJ (5.29 s), MCA (4.68 s), and PJ territories (5.44 s). CBF is decreased on the right side with
34.54 ml/100 g/mn in AJ, 40.89 in MCA, and 28.64 in PJ, while the values are normal in the opposite territories. CBV is only
increased in the right MCA with 6.05 ml/100 g.
Unreliable values may occur when vasospasm is severe, the vasospasm and no-vasospasm groups, the latter showing
as arterial input function cannot be properly selected, thus physiological prolonged values compared to proper territory.
leading to irrelevant data; however, when it happens, this Nevertheless, close attention should be paid to those crit-
should definitely lead to angiography for prompt endovas- ical border areas since they are the first to be affected in
cular treatment. In acute stroke patients, the selection of diffuse vasospasm.
different arterial input function has been demonstrated to
have no significant effect on PCT results for an individual
patient [39]. TTP
We acknowledge that the imaging technique has some
limitations. PCT consisted of four slices located at the level TTP values were not significantly higher in the vasospasm-
of the basal ganglia with limited coverage (in this study only group. Indeed, TTP is not a discriminative parameter, as
2 cm). The brain perfusion analysis thus under diagnosed being calculated directly without any arterial input refer-
potential delayed perfusion on distal territories. However, ence, opposed to MTT. Consequently, TTP is very sensitive
currently, the use of increased detectors leads to expanded to extracerebral or precerebral variables, such as cardiac
volume coverage. function, aortic and aortic arch branch vessel stenosis and
occlusion, or even variable position of the arm with vari-
able impingement effect on the intravenous line where the
MTT iodinated contrast material bolus is injected for the PCT
examination.
The most accurate PCT parameter to predict angiographic
vasospasm was MTT. No clear-cut value could be pointed out,
contrary to a previous study, which reported a threshold at CBV and CBF
6.4 s as the most accurate parameter for the diagnosis of
angiographic vasospasm [19]. The lack of clear-cut value is Differences between CBV and CBF reflect various moments
in agreement with other studies [13—16]. Some studies were in the timecourse of stroke and highlight vascular autoreg-
conducted with visual observation only [18,20]. ulation. CBV was not discriminative between patients
Compared to the control values, MTT increased from with and without vasospasm, as reported previously
123 to 221% (m = 146%). An increase of the MTT over 20% [38]. The overall higher CBV values in patients with
than the mean (corresponding to 120% of the control MTT) vasospasm were related to preservation of the autoregu-
has been shown to indicate the progression of cerebral lation, resulting in normal or increased CBV values [40].
vasospasm, and patients with vasospasm-related infarcts Conversely, declining CBV indicates impaired autoregula-
exhibited a MTT more than 47% greater than the mean tion. Thus, low CBV is only observed in areas of infarction,
value [16]. In diffuse vasospasm, the junctional territories, which is seen in a minority of vasospasm cases at that
so-called watershed area, are the first involved. Although stage, making CBV inappropriate as a sensitive screen-
higher MTT values were recorded in those areas, they were ing parameter for vasospasm. As previously reported [16],
non-significant, because of overlapping MTT values both in higher MTT may be a prerequisite for cerebral vasospasm
290 V. Lefournier et al.
cerebral vasospasm detection after subarachnoid hemorrhage. flow in aneurysmal subarachnoid haemorrhage with 99 mTc-
Surg Neurol 2007;67(6):604—11. ECD. Nucl Med Commun 2002;23(1):5—11.
[17] Aralasmak A, Akyuz M, Ozkaynak C, Sindel T, Tuncer R. [30] Sviri GE, Mesiwala AH, Lewis DH, Britz GW, Nemecek A, Newell
CT angiography and perfusion imaging in patients with sub- DW, et al. Dynamic perfusion computerized tomography in
arachnoid hemorrhage: correlation of vasospasm to perfusion cerebral vasospasm following aneurysmal subarachnoid hem-
abnormality. Neuroradiology 2009;51(2):85—93. orrhage: a comparison with technetium-99 m-labeled ethyl
[18] Moftakhar R, Rowley HA, Turk A, Niemann DB, Kienitz BA, Van cysteinate dimer-single-photon emission computerized tomog-
Gomple J, et al. Utility of computed tomography perfusion in raphy. J Neurosurg 2006;104(3):404—10.
detection of cerebral vasospasm in patients with subarachnoid [31] Yonas H, Sekhar L, Johnson DW, Gur D. Determination of irre-
hemorrhage. Neurosurg Focus 2006;21(3):E6. versible ischemia by xenon-enhanced computed tomographic
[19] Wintermark M, Ko NU, Smith WS, Liu S, Higashida RT, Dillon WP. monitoring of cerebral blood flow in patients with symptomatic
Vasospasm after subarachnoid hemorrhage: utility of perfusion vasospasm. Neurosurgery 1989;24(3):368—72.
CT and CT angiography on diagnosis and management. AJNR [32] Rowe J, Blamire AM, Domingo Z, Moody V, Molyneux A,
Am J Neuroradiol 2006;27(1):26—34. Byrne J, et al. Discrepancies between cerebral perfusion
[20] Binaghi S, Colleoni ML, Maeder P, Uske A, Regli L, Dehdashti and metabolism after subarachnoid haemorrhage: a mag-
AR, et al. CT angiography and perfusion CT in cerebral netic resonance approach. J Neurol Neurosurg Psychiatry
vasospasm after subarachnoid hemorrhage. AJNR Am J Neu- 1998;64(1):98—103.
roradiol 2007;28(4):750—8. [33] Rordorf G, Koroshetz WJ, Copen WA, Gonzalez G, Yamada
[21] Lodi CA, Ursino M. Hemodynamic effect of cerebral K, Schaefer PW, et al. Diffusion- and perfusion-weighted
vasospasm in humans: a modeling study. Ann Biomed Eng imaging in vasospasm after subarachnoid hemorrhage. Stroke
1999;27(2):257—73. 1999;30(3):599—605.
[22] Sloan MA, Haley Jr EC, Kassell NF, Henry ML, Stewart SR, [34] Busch E, Beaulieu C, de Crespigny A, Moseley ME. Diffusion MR
Beskin RR, et al. Sensitivity and specificity of transcra- imaging during acute subarachnoid hemorrhage in rats. Stroke
nial Doppler ultrasonography in the diagnosis of vasospasm 1998;29(10):2155—61.
following subarachnoid hemorrhage. Neurology 1989;39(11): [35] Condette-Auliac S, Bracard S, Anxionnat R, Schmitt E, Lacour
1514—8. JC, Braun M, et al. Vasospasm after subarachnoid hem-
[23] Manno EM, Gress DR, Schwamm LH, Diringer MN, Ogilvy CS. orrhage: interest in diffusion-weighted MR imaging. Stroke
Effects of induced hypertension on transcranial Doppler ultra- 2001;32(8):1818—24.
sound velocities in patients after subarachnoid hemorrhage. [36] Wintermark M, Maeder P, Thiran JP, Schnyder P, Meuli R. Quanti-
Stroke 1998;29(2):422—8. tative assessment of regional cerebral blood flows by perfusion
[24] Minhas PS, Menon DK, Smielewski P, Czosnyka M, Kirkpatrick PJ, CT studies at low injection rates: a critical review of the under-
Clark JC, et al. Positron emission tomographic cerebral per- lying theoretical models. Eur Radiol 2001;11(7):1220—30.
fusion disturbances and transcranial Doppler findings among [37] Nabavi DG, Cenic A, Craen RA, Gelb AW, Bennett JD,
patients with neurological deterioration after subarachnoid Kozak R, et al. CT assessment of cerebral perfusion: exper-
hemorrhage. Neurosurgery 2003;52(5):1017—22 [Discussion imental validation and initial clinical experience. Radiology
1022—4]. 1999;213(1):141—9.
[25] Lysakowski C, Walder B, Costanza MC, Tramer MR. Transcra- [38] Wintermark M, Dillon WP, Smith WS, Lau BC, Chaudhary S,
nial Doppler versus angiography in patients with vasospasm due Liu S, et al. Visual grading system for vasospasm based on
to a ruptured cerebral aneurysm: A systematic review. Stroke perfusion CT imaging: comparisons with conventional angiogra-
2001;32(10):2292—8. phy and quantitative perfusion CT. Cerebrovasc Dis 2008;26(2):
[26] Fontanella M, Valfre W, Benech F, Carlino C, Garbossa D, Ferrio 163—70.
M, et al. Vasospasm after SAH due to aneurysm rupture of the [39] Wintermark M, Lau BC, Chien J, Arora S. The anterior cerebral
anterior circle of Willis: value of TCD monitoring. Neurol Res artery is an appropriate arterial input function for perfusion-
2008;30(3):256—61. CT processing in patients with acute stroke. Neuroradiology
[27] Lewis DH, Eskridge JM, Newell DW, Grady MS, Cohen WA, 2008;50(3):227—36.
Dalley RW, et al. Brain SPECT and the effect of cerebral angio- [40] Soehle M, Czosnyka M, Pickard JD, Kirkpatrick PJ. Contin-
plasty in delayed ischemia due to vasospasm. J Nucl Med uous assessment of cerebral autoregulation in subarachnoid
1992;33(10):1789—96. hemorrhage. Anesth Analg 2004;98(4):1133—9 [Table of
[28] Powsner RA, O’Tuama LA, Jabre A, Melhem ER. SPECT imaging contents].
in cerebral vasospasm following subarachnoid hemorrhage. J [41] Wintermark M, Maeder P, Verdun FR, Thiran JP, Valley JF,
Nucl Med 1998;39(5):765—9. Schnyder P, et al. Using 80 kVp versus 120 kVp in perfusion
[29] Hosono M, Machida K, Matsui T, Honda N, Takahashi T, Dei S, CT measurement of regional cerebral blood flow. AJNR Am J
et al. Non-invasive quantitative monitoring of cerebral blood Neuroradiol 2000;21(10):1881—4.