Childhood Moyamoya Disease: Quantitative Evaluation of Perfusion MR Imaging—Correlation with Clinical Outcome after Revascularization Surgery

Introduction

Moyamoya disease is defined at angiography as a chronic progressive steno-occlusion of bilateral internal carotid arteries with characteristically abnormal vascular networks (so-called moyamoya vessels) at the base of the brain (^,1). Hemodynamics in moyamoya disease are complex because leptomeningeal collateral vessels from the posterior circulation and transdural collateral vessels from the external carotid artery, as well as moyamoya vessels, could develop to supply the ischemic brain (^,1,^,2). Moyamoya disease generally manifests in childhood, either as transient ischemic attacks manifesting as motor disturbances (70%–80%) or as epileptic seizures (20%–30%) (^,2). The clinical consequences of moyamoya disease are noteworthy; high incidences of recurrent ischemic attacks and major neurologic and cognitive impairments are common (^,1,^,2). Several surgical revascularization procedures have been considered to prevent recurrent ischemic attacks. These can be divided into direct procedures, which involve direct anastomosis of arteries between the extracranial and intracranial circulations, and indirect procedures, which involve the placing of the superficial temporal artery or vascularized tissues such as the temporalis muscle, dura mater, or omentum directly on the brain surface to promote collateral vessel formation (^,2).

Recently, perfusion magnetic resonance (MR) imaging has been used to noninvasively investigate cerebral hemodynamics in association with ischemic brain lesions. Furthermore, this technique has been found to be effective at estimating cerebral hemodynamics in moyamoya disease (^,3–^,7) and at revealing regions of decreased cerebral perfusion and cerebrovascular reserve, which can be improved by using revascularization surgery (^,8,^,9). However, longitudinal changes in cerebral hemodynamics depicted by using perfusion MR imaging after revascularization have not been well established in the literature. Thus, the aim of our study was to evaluate whether perfusion MR imaging can depict hemodynamic status after revascularization surgery and whether changes at perfusion MR imaging after revascularization surgery correspond with clinical outcome in moyamoya disease.

MATERIALS AND METHODS

Patients

The institutional review board of Seoul National University Hospital approved this retrospective study; the requirement for informed consent was waived. From July 2002 to April 2006, 113 consecutive children with moyamoya disease confirmed at angiography underwent revascularization surgery at our institution. Of these, 67 patients who satisfied the following criteria were enrolled in this study: (a) patients in whom pre- and postoperative perfusion MR imaging were performed (nine patients excluded); (b) patients who underwent follow-up perfusion MR imaging more than 3 months after surgery (17 patients excluded); (c) patients in whom the cerebellum was not involved and at least three cerebellar sections were available (20 patients excluded); and (d) patients who visited an outpatient clinic for the evaluation of postoperative clinical outcomes (no patient excluded). During this study period, perfusion MR imaging was performed twice in 12 patients who underwent bilateral revascularization surgery with intervals. For the 12 patients, we used perfusion MR images from the first surgery only. Finally, 134 perfusion MR imaging examinations (67 pre- and postperfusion MR image sets) were included in this study. There were 28 boys and 39 girls (mean age, 7.2 years; range, 2–13 years). The mean age of boys was 6.9 years (range, 2–13 years), and the mean age of girls was 7.5 years (range, 3–13 years). Clinical types were as follows: those who had a transient ischemic attack, 29 patients; those who had a transient ischemic attack with infarction, 35 patients; and those with no transient ischemic attack with or without infarct, three patients. The mean period between preoperative perfusion MR imaging and surgery was 3.9 days (range, 1–38 days). The mean period between surgery and postoperative perfusion MR imaging was 5.3 months (range, 4–13 months), and the mean follow-up period after postoperative perfusion MR imaging was 10.5 months (range, 4–23 months).

Perfusion MR Imaging Technique

Perfusion MR imaging was performed with a 1.5-T unit (Signa or CV/i; GE Medical Systems, Milwaukee, Wis) and gradient echo-planar imaging sequences by using the following parameters: repetition time msec/echo time msec, 2000/60; flip angle, 90°; matrix, 128 × 128; field of view, 23 or 24 cm; section thickness, 5 mm; and intersection gap, 2 mm. A series of images (10 sections, 50 images per section) was obtained before, during, and after the administration of contrast agent (0.1 mmol gadopentetate dimeglumine [Magnevist; Schering, Berlin, Germany] per kilogram of body weight at a rate of 2 mL/sec) by using an MR imaging–compatible power injector (Spectris; Medrad, Pittsburgh, Pa) in those older than 5 years or by using manual injection in those 5 years and younger. This was followed by a 15-mL bolus of saline administered at the same injection rate. Perfusion maps of time to peak (TTP) and cerebral blood volume (CBV) were generated after eliminating the effect of contrast agent recirculation by using γ-variate curve fitting (^,10,^,11).

Analysis of Perfusion MR Imaging Data

For quantitative analysis, regions of interest (ROIs) were drawn in the cerebral and cerebellar (reference region) hemispheres. The ROIs in the cerebrum were rectangular and covered the whole cerebral hemisphere, including the cerebral cortex, but did not include infarcted areas. The ROIs in the cerebellum were used to minimize potential errors when calculating TTP and CBV caused by differences in arterial input function associated with bolus delay and dispersion (^,6–^,8). These ROIs were rectangular (mean size, 200 mm²) (^,Fig 1). Regional TTP (rTTP) and regional CBV (rCBV) were calculated by adjusting cerebral TTP and CBV values by using cerebellar reference values as follows: rTTP = TTP_cb − TTP_cbl and rCBV = CBV_cb/CBV_cbl, where cb is cerebrum and cbl is cerebellum. Segmentation analysis of rTTP was evaluated by using categorizing values (>0, >2, >4, >6, and >8 seconds). Pixels in each time category were expressed as percentages of all pixels (number of pixels in cerebral ROI with a TTP greater than the specified delayed time divided by total number of pixels in the cerebral ROI) (^,5,^,12) (^,Fig 2). After cerebral and cerebellar ROIs were copied and pasted into spatially coregistered homologous sections on postoperative images, these procedures were repeated. To compare preoperative and postoperative perfusion MR imaging parameters, we calculated the following differences: ΔrTTP = rTTP_pre − rTTP_post, ΔrCBV = rCBV_pre − rCBV_post, and change in percentage of pixels of rTTP = percentage of pixels of rTTP_pre − percentage of pixels of rTTP_post, where Δ indicates change, pre indicates before surgery, and post indicates after surgery. One experienced neuroradiologist (T.J.Y., more than 5 years of experience) who was unaware of clinical outcomes performed the procedure.

Assessment of Postoperative Clinical Outcomes

Postoperative clinical outcomes were assigned to one of the following four categories by experienced neurosurgeons (K.C.W., more than 20 years of experience) at last outpatient clinic visit: 1 indicated excellent (preoperative symptoms, such as transient ischemic attacks, had totally disappeared without fixed neurologic deficits); 2, good (symptoms had totally disappeared but neurologic deficits remained); 3, fair (symptoms persisted but their frequencies had decreased); and 4, poor (symptoms remained unchanged or had worsened) (^,8). The symptoms and neurologic deficits evaluated were limited to those associated with the involved hemisphere.

Statistical Analysis

The paired t test was used to compare pre- and postoperative MR imaging perfusion parameters (rTTP, rCBV, and percentage of pixels of rTTP). One-way analysis of variance and the Scheffé post hoc test were used to compare clinical outcomes and perfusion MR findings. Preoperative and postoperative rTTP, rCBV, percentage of pixels of rTTP, change in rTTP, and change in percentage of pixels of rTTP were considered to be independent variables, and postoperative clinical outcome was considered to be a dependent variable. Statistical analysis was performed by using commercially available software (SPSS, version 12.0 for Windows; SPSS, Chicago, Ill), and P values less than .05 were considered to indicate statistically significant differences.

RESULTS

Clinical Outcomes

Postoperative clinical outcomes as determined by neurosurgeons at last outpatient visit were as follows: excellent (n = 20), good (n = 18), fair (n = 24), and poor (n = 5).

Perfusion MR Imaging Data before and after Surgery

Preoperative and postoperative mean rTTP values for total sets were 3.35 seconds ± 2.53 (standard deviation) and 1.73 seconds ± 1.57, respectively; mean postoperative rTTP was significantly shorter than mean preoperative rTTP (P < .001) (^,Fig 3a^,). Preoperative and postoperative mean rCBV values were 1.35 ± 0.34 and 1.24 ± 0.26, respectively; postoperative mean rCBV was significantly lower than preoperative mean rCBV (P = .014) (^,Fig 3b). After revascularization surgery, percentage of pixels of rTTP in each category significantly decreased (P < .001) (^,Table 1).

Relation between Perfusion MR Imaging Data and Clinical Outcomes

One-way analysis of variance testing showed that preoperative rTTP was significantly different for the clinical outcome categories (P = .010). However, there was a statistically significant difference only between excellent and fair outcomes, according to Scheffé post hoc analysis (P < .05). Postoperative rTTP was significantly different for the four clinical outcome categories (P = .012). However, there was a statistically significant difference only between fair and poor outcomes, according to Scheffé post hoc analysis (P < .05).

Means and standard deviations of change in rTTP in each clinical outcome category were as follows: excellent, 3.10 ± 1.77; good, 1.62 ± 2.09; fair, 1.04 ± 0.86; and poor, −1.52 ± 2.01. One-way analysis of variance showed that change in rTTP was significantly different for the clinical outcome categories (^,Fig 4a^,). However, there was no statistically significant difference of the change in rTTP values between the good and fair categories according to Scheffé post hoc analysis (P = .728). In terms of change in percentage of pixels of rTTP, values were significantly different for the clinical outcomes according to the following time categories: change in percentage of pixels of rTTP at more than 0 seconds (P < .001), at more than 2 seconds (P < .001), at more than 4 seconds (P < .001), and at more than 6 seconds (P = .004). The patients with higher change in percentage of pixels of rTTP values had better clinical outcome at more than 0 seconds and at more than 2 seconds according to Scheffé post hoc analysis (^,Fig 5^,).

Preoperative rCBV (P = .846), postoperative rCBV (P = .571), and change in rCBV (P = .889) were not significantly different for the clinical outcome categories (^,Fig 4b, ^,Table 2).

DISCUSSION

The goal of surgical treatment in moyamoya disease is to reestablish cerebral blood flow to ischemic regions. Encephaloduroarteriosynangiosis is most widely used in childhood moyamoya disease because of its simplicity and the low risk of ischemia due to the temporary blocking of cerebral blood flow during conventional superficial temporal artery–middle cerebral artery anastomosis (^,13–^,16). After encephaloduroarteriosynangiosis, the superficial temporal artery and/or an adjacent middle meningeal artery participate in neoangiogenesis, which can be detected by using conventional angiography or MR angiography (^,17,^,18).

Although cerebral angiography is considered the standard imaging modality for diagnosis and postoperative follow-up of moyamoya disease, its invasive nature militates against its use in children, which creates the need for a minimally invasive reliable hemodynamic imaging tool. The noninvasive character of perfusion MR imaging and its susceptibility to microvascular hemodynamic alterations, short acquisition times, and use of nonionizing radiation, in addition to the widespread availability of MR imagers, make the modality well suited for children. In addition, the faster heart rates, smaller total blood volumes, and the smaller brain sizes of children also favor the use of MR imaging (^,19). Perfusion MR imaging has been widely used for postoperative follow-up imaging of children with moyamoya disease (^,9,^,13). Lee et al (^,9) reported that all 13 of their patients showed delayed TTP in the middle cerebral artery territory before encephaloduroarteriosynangiosis compared with that in control subjects and that TTP values were significantly reduced after encephaloduroarteriosynangiosis.

In moyamoya disease, the presence of collateral vessels introduces large delays and is likely to disperse contrast agent; therefore, maps generated by using deconvolution may result in much increased preoperative rTTP and rCBV values (^,7). Tanaka et al (^,6) suggested that quantitative evaluations of CBV and mean transit time by using perfusion MR imaging can provide useful hemodynamic information regarding the presence of misery perfusion and that it does so at the same level of reliability as positron emission tomography. However, Perthen et al (^,20) concluded that mean transit time and cerebral blood flow determined by using MR imaging are biased estimates of their true values and that mean transit time is substantially overestimated and cerebral blood flow is underestimated. In the present study, we used TTP and CBV values as indicators of cerebral perfusion, and rTTP, percentage of pixels of rTTP, and rCBV values were found to be significantly reduced after revascularization surgery. It appears that these phenomena are probably because of rapid parenchymal perfusion after revascularization surgery and decrease of collateral vessel formation.

The clinical value of perfusion MR imaging results as predictor of outcome after revascularization surgery has not been well established in the literature (^,8). In our study, change in rTTP and change in percentage of pixels of rTTP after revascularization were found to reliably predict clinical outcome in moyamoya disease. With regard to change in percentage of pixels of rTTP, values were significantly different among all the clinical outcome categories; significant differences were found in change in percentage of pixels of rTTP at more than 0 seconds and at more than 2 seconds according to Scheffé post hoc analysis. With the consideration that clinical outcomes were assigned to the four categories according to the change in symptoms and neurologic deficits, the results imply that the patients with higher change in percentage of pixels of rTTP at more than 0 seconds and at more than 2 seconds after revascularization have higher possibility to experience symptomatic and neurologic benefit after surgery. However, there were substantial overlaps between groups, and that suggests limitations to the use of these results for specific case planning.

Even though our data showed statistically significant differences in pre- and postoperative rTTP for the clinical outcomes according to one-way analysis of variance, there were significant differences only between excellent and fair outcomes in preoperative rTTP and between fair and poor outcomes in postoperative rTTP according to post hoc testing.

In terms of evaluating hemodynamic status in moyamoya disease, single photon emission computed tomography (SPECT) has been considered a reference standard (^,21–^,25). SPECT with intravenous acetazolamide challenge has an advantage for the evaluation of vascular reserve. Intravenous administration of acetazolamide induces marked dilatation of cerebral vessels, and decreased reactivity to acetazolamide is thought to be due to a decrease in cerebral vascular reserve secondary to vasodilatation in the resting state. There have been several reports (^,26,^,27) of acetazolamide or arterial spin labeling challenge during MR perfusion imaging in patients with ischemic brain injury. If perfusion MR imaging with or that without acetazolamide stress (^,26) or arterial spin labeling (^,27) can provide information comparable to that of SPECT in moyamoya disease, it would be advantageous in the clinical setting, and further study is needed for clinical application of these perfusion MR imaging techniques in patients with moyamoya disease.

Our study had several limitations. First, the accurate drawing of middle cerebral artery territories was challenging; therefore, we measured perfusion changes in whole hemispheres to reflect general perfusion change in involved hemispheres. However, potential bias might have been introduced by the inclusion of portions of the temporal and occipital lobes in ROIs, as these were supplied by the posterior circulation. In addition, because we selected levels of greatest asymmetric hemispheric perfusion without obvious infarction, real TTP and CBV values may have been overestimated.

The second limitation concerns arterial input function, which is generally measured by using a large artery, because in moyamoya disease accurate determinations of true arterial input function are challenging. Thus, we considered it more reasonable to use cerebral TTP values adjusted by using cerebellar TTP values. Even though the cerebellum is supplied by the posterior circulation, which differs from the anterior circulation in terms of arterial input function and bolus delay, it appears to be the best reference area because posterior cerebral artery involvement would occur at relatively high internal carotid artery stages and because it is hardly affected by revascularization surgery (^,7).

The third limitation concerns suboptimal image quality after revascularization surgery. Gradient echo-planar imaging has advantages in terms of the compartmentalization of perfusion images. However, the accumulation of contrast agents with high magnetic susceptibility within the randomly oriented capillary network of the brain results in localized variations in tissue magnetic fields, and the resulting magnetic susceptibility effects reduce image quality, especially after revascularization surgery. Finally, a repetition time of 2000 msec is a bit too long for temporal resolution of a perfusion study, and the rate of injection (2 mL/sec) is slow, but this was necessary because our study was performed in children.

In summary, our study shows that the use of TTP and CBV perfusion maps can depict hemodynamic status after revascularization surgery in children with moyamoya disease. Changes in TTP perfusion maps after revascularization surgery correspond with clinical outcome in moyamoya disease. We postulate that changes in TTP values can be used to predict clinical outcomes after revascularization surgery.

ADVANCES IN KNOWLEDGE

IMPLICATION FOR PATIENT CARE

Figure 1:

Download as PowerPointOpen in Image Viewer

Figure 2:

Download as PowerPointOpen in Image Viewer

Figure 3a:

Download as PowerPointOpen in Image Viewer

Figure 3b:

Download as PowerPointOpen in Image Viewer

Figure 4a:

Download as PowerPointOpen in Image Viewer

Figure 4b:

Download as PowerPointOpen in Image Viewer

Figure 5a:

Download as PowerPointOpen in Image Viewer

Figure 5b:

Download as PowerPointOpen in Image Viewer

Table 1. Comparison between Preoperative and Postoperative Perfusion MR Imaging Parameters

Note.—Data are means ± standard deviations.

^* Paired t test was used for statistical analysis.

Table 2. Relation between Perfusion MR Imaging Parameters and Postoperative Clinical Outcomes

Note.—Data are means ± standard deviations.

^* Comparisons were made by using one-way analysis of variance followed by Scheffé post hoc analysis.