Outcome prediction of pediatric moyamoya disease using midterm cerebral blood flow measured between staged anastomoses

Cognitive outcomes of pediatric moyamoya disease are variable and difficult to predict on the basis of initial neurological signs and examinations. To determine the best early time point for outcome prediction, we retrospectively analyzed the correlation between cognitive outcomes and the cerebrovascular reserve capacity (CRC) measured before, between, and after staged bilateral anastomoses. Twenty-two patients aged 4–15 years were included in this study. CRC was measured before the first hemispheric surgery (preoperative CRC), 1 year after the first surgery (midterm CRC), and 1 year after the surgery on the other side (final CRC). The cognitive outcome was the Pediatric Cerebral Performance Category Scale (PCPCS) grade more than 2 years after the final surgery. The 17 patients with favorable outcomes (PCPCS grades 1 or 2) showed a preoperative CRC of 4.9% ± 11.2%, which was not better than that of the five patients with unfavorable outcomes (grade 3; 0.3% ± 8.5%, p = 0.5). The 17 patients with favorable outcomes showed a midterm CRC of 23.8% ± 15.3%, which was significantly better than that of the five patients with unfavorable outcomes (–2.5% ± 12.1%, p = 0.004). The difference was much more significant for the final CRC, which was 24.8% ± 13.1% in the patients with favorable outcomes and –11.3% ± 6.7% in those with unfavorable outcomes (p = 0.00004). Cognitive outcomes were first clearly discriminated by the CRC after the first-side unilateral anastomosis, which is the optimal early timing for the prediction of individual prognosis.


Introduction
Cerebrovascular moyamoya disease mainly affects the pediatric population and presents with progressive stenosis of the major cerebral arteries with unknown etiologies [1][2][3][4]. Patients initially experience acute ischemic symptoms. The efficacy of direct and indirect anastomosis surgeries in preventing ischemic attack and cerebral infarction has been established [5][6][7][8]. The development of surgical procedures, medical therapy, and daily life guidance for patients, parents, and schoolteachers helps increase the population of patients with physical independence. However, some patients still suffer from chronic ischemic symptoms or poststroke sequelae, such as difficulty with social independence accompanied by cognitive impairment, especially adolescents and adults [9]. It is difficult to predict the cognitive outcomes of an individual patient based on their preoperative neurological signs and examinations. So et al. [10] described the prediction of the clinical outcome of a study of cerebral blood flow (CBF) taken 6 to 12 months after the operation. According to our impression, the prognosis becomes clear around the midterm of the course of treatment, and we may predict the outcome earlier than in the report by So et al. [10].
To assess the preoperative ischemic state and determine the surgical indication, neurological findings and magnetic resonance imaging (MRI) are essential. CBF measurement is an objective option [11,12]. Kazumata et al. [13] indicated an association of cognitive function with CBF. CBF 1 3 data consisted of resting CBF and cerebrovascular reserve capacity (CRC) [14]. CRC is the increased ratio of CBF from resting CBF to CBF activated by acetazolamide, a vessel dilator agent. In patients with cerebrovascular steno-occlusive disease, the arterioles dilate, and the resting CBF is preserved. Therefore, resting CBF in ischemic patients is not inferior to that in nonischemic patients. When acetazolamide is administered, the arterioles in nonischemic patients dilate, and CRC is positive. Meanwhile, arterioles already dilate in ischemic patients and can no longer dilate to increase CBF, resulting in a misdiagnosis of CRC. This is why CRC better demonstrated the ischemic state of the patients than resting CBF. We operated on the patient in two stages, on the right and left sides of each, as in most institutes do [15], and we measured resting CBF and CRC before and after each surgery. However, the efficacy versus safety, effort requirement, time consumption, and cost performance in examining CRC have been debated [15,16].
In this study, we retrospectively analyzed whether midterm CBF and CBF after first-side anastomosis surgery could predict the final outcome and inform us of decisions regarding second-side anastomosis. For this purpose, we divided the patients into four groups according to the midterm CRC and compared the characteristics of the patients and outcomes. The necessity and limitations of CBF measurements were also discussed.

Patients
Twenty-two pediatric patients with moyamoya disease, nine boys and 13 girls, aged 7.6 ± 2.9 years (range, 4-15 years), underwent surgery at Yamaguchi University Hospital between 2007 and 2020. Patients with quasi-moyamoya disease or unilateral moyamoya disease were excluded. The patients were treated according to our protocol and guidelines for the diagnosis and treatment of moyamoya disease issued by the Research Committee of the Ministry of Health, Labor and Welfare, Japan [17]. Data, clinical findings, and neuroradiological findings of the patients were retrospectively analyzed.

Surgeries
Surgery was performed on patients with ischemic symptoms. Superficial temporal artery to middle cerebral artery (STA-MCA) anastomosis combined with split duro-encephalosynangiosis (split DES) [18] and cranial burr holing [19] was performed. When the MCA branches of the recipient were too small for anastomosis, encephalo-duro-arterio-synangiosis (EDAS) [20] combined with split DES and cranial burr holing was performed. The first operative side was determined based on the frequency of transient ischemic attack (TIA). If it occurred evenly bilaterally, the decision of the surgical side was made in the order of a smaller CRC side, more progression side of MCA stenosis, and then a language-dominant hemisphere. The second-side surgery was performed when TIA appeared. The interval between the two surgeries was 6 months at the shortest and 3 years at the longest. Patients who had been asymptomatic for 3 years since the first surgery did not undergo second-side surgery.

Examinations
Before first-side surgery, all patients were examined, the Wechsler Intelligence Scale for children (WISC) III or IV was used for patients aged > 5 years, and the Tanaka − Binet Intelligence scale IV or V was used for patients aged < 5 years. The neurological outcome at the final follow-up, 1.5 − 13 years after the final surgery, was assessed using the Pediatric Cerebral Performance Category Scale (PCPCS) [21] for school-age patients. We modified the PCPCS for patients aged > 18 years (Table 1). Grades 1 and 2 were regarded as favorable outcomes, and grades 3 to 6 were unfavorable outcomes. MRI and angiography were performed as necessary.

CBF measurement and patient groups
CBF was measured using 123 I-iodoamphetamine singlephoton emission computed tomography (IMP-SPECT). Dependent on others for daily support Dependent on others for daily support 5 Any degree of coma without brain death Any degree of coma without brain death 6 Brain death Brain death 1 3 The range of interest (ROI) was set in the bilateral MCA territories. Using the dual table autoradiography method [22], resting CBF and acetazolamide challenge CBF were measured on the same day. CRC was calculated as the percentage increase in CBF from resting CBF to acetazolamide-challenged CBF. Patients aged < 13 years were sedated during the measurement using CBF by intravenous administration of thiopental.
Resting CBF and CRC were measured three times: before the first-side surgery (preoperative CBF), before the second-side surgery (midterm CBF), and 1 year after the second-side surgery (final CBF) (Fig. 1). Patients who underwent surgery only on one side were examined twice for CBF measurement, and the second measurement was performed 3 years after surgery.
No standard threshold of sufficient CRC for the pediatric patient with moyamoya disease has been defined. CRC of the normal volunteers was reported as 58.8 ± 18.9% [23]. Indication of extracranial-intracranial bypass surgery for patients with ICA or MCA stenosis is CRC less than 10% [24]. So, et al. [10] classified moyamoya disease patients into the preserved and the decreased CRC groups using a threshold of 0% of CRC. From the existing literature, we chose 7% as sufficient CRC after the firstside anastomosis, which is the median value between -3% and 17%. The values -3% and 17% are the CRCs before and after the bilateral anastomoses in pediatric moyamoya disease reported by Nakagawara et al. [11]. Therefore, the patients were classified into four groups according to the results of the first-side surgery.
The number of patients included in the excellent, good, fair, and poor groups was six, six, four, and six, respectively.

Statistics
Differences between groups in the patient's age, IQ, and CBF were analyzed with analysis of variance, and sex, paralysis, infarction, stenosis of the cerebral artery, surgical method, and outcomes were analyzed using the chi-square test. A p-value of less than 0.05 was considered statistically significant.
This retrospective study was approved by the Yamaguchi University Hospital Institutional Review Boards (H2020-095 and H2021-044).

Timing of outcome prediction
The 17 patients with favorable outcomes in PCPCS or modified PCPCS of grades 1 and 2 showed 4.9 ± 11.2% of preoperative CRC, which was not better than the five patients with unfavorable outcomes of grade 3 (0.3 ± 8.5%, p = 0.5). Meanwhile, the 17 patients with grades 1 and 2 showed 23.8 ± 15.3% in midterm CRC, significantly better than the five patients with grade 3 (-2.5 ± 12.1%, p = 0.004). The difference became much more significant in the final CRC, with 24.8 ± 13.1% of grades 1 and 2 and -11.3 ± 6.7% of grade 3 (p = 0.00004). Cognitive outcomes were first clearly discriminated by the CRC after the first-side unilateral anastomosis, which proved that the classification of the four groups was appropriate.

Patient characteristics in the four groups
Age and sex differences among the four groups were not observed. The preoperative IQ was lower in the poor group (82.2 ± 13.8) than in other groups (91.5 − 95.3), however, the difference was not significant. A cortical infarction was observed in the preoperative examination in five patients. The poor group included four of them, which was a significantly high rate between the groups (p = 0.03). Cortical infarctions appeared during the course of treatment in one patient in the good group and three patients in the poor group. The infarctions in the poor group patients were not significantly larger than those in the good group (26.7 ml vs. 13.0 ml, p = 0.1), which may have influenced the cognitive outcome. Stenosis in the anterior cerebral artery (ACA) and MCA was observed in all patients. The posterior cerebral artery (PCA) was involved in stenosis in two patients in the initial, and five patients in the final examination, all of whom belonged to the poor group (p = 0.02 in Table 2 and p = 0.002 in Table 3). Although the presence of PCA stenosis correlated with the presence of infarction Fig. 1 Planning of staged anastomoses and cerebral blood flow (CBF) measurements. Hemispheric anastomosis is performed when the patient had frequent ischemic attacks. CBF was measured three times: before, after the first surgery, and after the second surgery. The interval from the first surgery to the midterm measurement was determined by symptom progression, and the interval from midterm to the final measurement was 1 year (p = 0.009), infarction located in the PCA territory was seen in only one patient. The stenosis of PCA had a greater impact on the collateral circulation for ACA and MCA territories than on PCA territory.

The outcome of the four groups
Indirect anastomosis was performed in all patients. Direct anastomosis was performed in 13 and 11 patients during the first and second surgeries, respectively. These surgical methods did not influence clinical outcomes or postoperative resting CBF or CRC. In all patients, TIA was discontinued within years after the final surgery. During treatment, cortical infarction developed in one patient in the excellent group and three patients in the poor group. Developmental outcomes assessed using PCPCS were significantly worse in the poor group than in the other three groups (p = 0.01) ( Table 3). No differences in outcomes were observed between the excellent, good, and fair groups.

Chronological change of resting CBF
A cerebral angiogram revealed neovascularity of the anastomosed arteries and regression of the moyamoya vessels. The remaining internal carotid artery (ICA), ACA, and MCA varied regardless of neurological or CBF outcomes. The preoperative, midterm, and final resting CBF values were not significantly different between the four groups. In particular, the final resting CBF of the poor group, although it showed the worst neurological symptoms, was not significantly inferior to that of the other groups (Table 3). Despite the favorable course, no increase in resting CBF was observed throughout the treatment course in the excellent, good, and fair groups (Tables 2 and 3; Fig. 2).

Chronological change of CRC
In the excellent and good groups, bilateral CRC increased after first-side surgery (Fig. 3), and midterm CRC was good not only in the operated hemisphere (31.8 ± 17.5% and 29.6 ± 15.6%) but also in the unoperated hemisphere (33.3 ± 15.9% and 21.5 ± 12.1%) ( Table 2). No further increase in CRC was observed after the second-side surgery in the good group (Table 3; Fig. 3), although TIA was alleviated. In the fair group, the CRC increased to 18.3 ± 19.1% on the first surgery side, but not on the unoperated side (11.0 ± 12.5%) ( Table 2; Fig. 3). The CRC in the unoperated hemisphere was significantly lower (p = 0.009) between the groups. The development of the circle of Willis was compared to evaluate why patients in the excellent and good groups showed a bilaterally parallel increase in CRC and the fair group showed an increase in CRC only in the surgical hemisphere. However, no significant differences were observed in the anterior communicating artery (AcomA) or posterior communicating artery (PcomA) ( Table 2). Patients in the fair group underwent second-side surgery, and the final CRC of the side increased to 23.8 ± 7.7%, which was the same level as the final CRC of the first surgical side (23.0 ± 6.1%) and those of the excellent (33.3 ± 15.9%) and good (24.6 ± 10.3%) groups. Any side of CRC in the poor group continuously deteriorated from preoperative (2.6% and 12.4%) to midterm (1.7% and 2.3%) and final (-6.9% and -4.1%) measurements, and the difference from other groups increased chronologically (Tables 2 and 3; Fig. 3), despite the clinical stabilization of ischemic symptoms and the appearance of neovascularity on postoperative angiography.

Discussion
Two-stage surgeries were performed on patients with moyamoya disease. The outcome was not correlated with preoperative CRC, but with midterm CRC. The excellent group, whose bilateral CRC increased and TIA was discontinued by unilateral surgery, obtained favorable outcomes without the requirement of contralateral anastomosis. The good and fair groups whose bilateral and ipsilateral midterm CRCs increased with TIA remaining in the contralateral hemisphere, also achieved favorable outcomes after the second surgery. The poor group, whose midterm CRC did not increase, developed unfavorable outcomes even after additional second-side anastomosis. Resting CBF did not increase in any group, regardless of the outcome. The incidence of bilateral CRC increased in the excellent and good groups after the first unilateral anastomosis. Deckers et al. [25] also reported that unilateral surgery improved bilateral CRC and TIA frequency in seven patients with moyamoya disease. The reason unilateral anastomosis increases the incidence of bilateral CRC is discussed. Angiography after the first surgery revealed neovascularization of the ipsilateral STA and middle meningeal artery (MMA) to the ipsilateral cortex. In some patients, neovascularization of the contralateral STA and MMA develops beyond the midline and enters the ipsilateral cortex. However, none of the patients showed neovascularization from the operative side to the contralateral hemisphere. The development of cross-flow through AcomA was also poor, not only in the fair group but also in other groups. In other words, the crossed CBF from the surgical side to the contralateral side cannot explain why bilateral CRCs remained symmetrical. One mechanism that balances the laterality of CRC is the contribution of PCA. PCA provides CBF to the bilateral ACA and MCA territories. Preoperatively, CBF supporting the bilateral hemispheres would be beyond the capability of the PCA. Postoperatively, the operative side would be independent of PCA support, and PCA can provide sufficient CBF to the unoperated side. As a result, the contralateral CRC increases equally with the ipsilateral CRC. Another reason is that CRC asymmetry is observed only in patients showing an asymmetric progression phase of stenosis. Although not significant, the contralateral CRC before the first-side surgery in the fair group was 11.6 ± 18.8%, higher than that in the good group (1.8 ± 10.0%, Table 2); therefore, CRC would not need to be increased by surgery. The interval between the two surgeries was not significantly longer in the fair group (3.0 years) than in other groups (1.0 − 1.6 years, Table 2) also suggested the slow progression of stenosis on the contralateral side in the fair group. After the first surgery, the stenosis gradually progressed and the CRC was inferior on the contralateral side. There is no strong evidence for these theories; however, the results indicate that most patients show balanced laterality in CRC, and asymmetric CRC is rarely seen.
The preoperative neurological and neuroradiological findings did not differ between the four groups. The ranking of the excellent to the poor group was obvious after the first surgery. The individual prognosis could not be predicted from the initial findings but from the results of the first surgery. During the same period of a course of treatment, an indication of an additional contralateral anastomosis could also be decided. Anastomosis surgery is performed bilaterally in some institutes [15] or selected patients [26]; however, we recommend operating one side after another. Children with excellent midterm results can avoid additional surgery. Children with good or fair Fig. 3 Preoperative, midterm, and final cerebrovascular reserve capacity (CRC) in the four groups. CRC is measured twice in the excellent group and three times in other groups. Solid and dotted lines indicate the first and second operation sides. The error bars indicate the standard error of the mean. Significant increases and decreases are indicated. *p < 0.01, # p < 0.05 midterm results should undergo contralateral surgery to achieve favorable outcomes. Children with poor midterm results require contralateral surgery, special education, and social support.
In the poor group, CRC decreased in a stepwise manner despite neovascularization revealed by angiography. Four of the six patients in the group showed PCA stenosis. Pediatric moyamoya disease presents with progressive stenosis of the ACA and MCA and occasionally involves PCA. Patients with PCA stenosis involvement present poor outcome [27]. Ohkura et al. [28] emphasized the importance of the collateral circulation of the PCA in preventing cortical infarction. Surgical anastomosis helps increase CBF, but may be less relevant than spontaneous collateral circulation from PCA. Another idea is that CRC decreases with multiple infarctions; however, the area of infarction would not have been large enough to influence CRC. The preservation of the resting CBF in the poor group proved that the infarction area was small.
Compared with resting CBF, CRC provides more information on postoperative improvement and outcomes. Meanwhile, CRC measurement has the potential risk of cerebral infarction caused by the cerebrovascular steal phenomenon [29,30], pulmonary edema, nausea and vomiting caused by acetazolamide [31,32], and respiratory suppression by extended sedation time. We obtained written informed consent from the parents in advance and prepared for adverse effects during and after the examination. Radioisotopes were used within the allowed limits in guideline [33].
This study had some limitations. First, CBF in patients aged < 13 years may be underestimated because it decreases with thiopental administration. For children aged > 13 years, examinations were performed in the awake state, and visual stimulation was blocked with an eye mask instead of sedation. The difference in the influence of sedation between the groups was minimal because the age distributions were the same. Second, the poor group was not younger than the others, which is different from the previous report that small children progressed quickly and presented worse outcomes [34]. The discrepancy is because a small number of small children were included in this study and that older children suffering from a longer period of illness due to misdiagnosis were included in the poor group. Third, we did not perform encephalomyosynangiosis, which is an admired indirect anastomosis method that provides a large amount of CBF [35], although it is unclear [36][37][38]. Fourth, the CBF in the ROI was calculated automatically; therefore, the area of infarction was included in the ROI. To observe a real CRC, the infarcted area should be excluded. CRC in patients with infarction may have been underestimated; however, the areas were small, as described above.

Conclusions
Staged anastomoses were performed in children with bilateral moyamoya disease. The response of CRC to the first unilateral anastomosis, not the initial CRC, was correlated with the final cognitive outcome; patients who showed increased CRC at the midterm examination resulted in a favorable outcome, especially for those with bilaterally increased CRC and discontinuation of TIA, contralateral anastomosis would not be required. The unfavorable outcomes were mainly caused by PCA stenosis, which was also predicted at the midterm examination. Unlike CRC, resting CBF was not informative in observing the patient's ischemic condition.