Outcomes of surgical revascularization for pediatric moyamoya disease and syndrome

This was an Institutional Review Board-approved retrospective review of all moyamoya surgeries performed by the senior author (SNM) between 2009 and 2019. The electronic medical records were searched for the following data: demographic data, prior treatment history, presenting symptoms, radiographic data, neurological and functional status pre- and post-operatively, surgical details, complications, and follow-up data. Ischemic events were considered transient ischemic attacks (TIAs) if the symptoms lasted less than 24 h, and no sequelae were observed on MRI. Otherwise, all events were considered ischemic strokes. Clinical follow-up data, including modified Rankin Scale score (mRS) and incidence of clinical strokes, were obtained from the last available neurosurgical, neurological, or hematological note in a patient’s record. Post-operative strokes were identified through both review of these notes and interpretation of all post-operative MRIs each patient had within the hospital system. Angiographic evaluation of revascularization was determined by the Matsushima grade (defined as the percent of the ischemic area revascularized where A =  > 66%, B = 33–66%, and C < 33%) on each patient’s most recent angiogram.

Data were segregated for analysis in the following ways. Ischemic events were divided into those occurring peri-operatively (within 30 days of surgery) or post-operatively (after 30 days). Pre- and post-operative stroke rates were calculated by dividing the number of strokes a patient had in the time from presentation to our clinic or emergency department to surgery and the time from surgery until last follow-up, respectively. Each rate was then standardized over 365 days. Post-operative ischemic events were defined as ischemic events occurring in a surgically revascularized hemisphere. Additionally, the decision to preclude ischemic events prior to presentation to our team was made to bias the results towards the null rather than including events from patient-reported histories.

The majority of patients were revascularized with superficial temporal artery (STA)-pial synangiosis as follows. Patients were positioned supine with their head rotated so the operative site was facing up. The STA course was mapped with a Doppler, and the vessel was subsequently dissected with a combination of blunt and sharp dissection under the operative microscope. Branches from the parent vessel were coagulated and cut as needed to free the artery along with a small cuff of galea. Attention was then turned back to the scalp, and the temporalis muscle was incised. Self-retaining retractors were placed, and a craniotomy was performed. The dura was opened in a stellate fashion, and the arachnoid was opened in multiple places. Using 10–0 nylon suture, the galeal cuff of the STA was sewn to the pia in multiple locations to fix it to the surface of the brain. The dural leaflets were laid back onto the surface of the brain, and the craniotomy burr holes were widened to allow for free passage of the STA. Finally, the bone flap was plated to the skull over the STA, the temporalis muscle was reapproximated, and the scalp was closed in layers. Generally, in patients with bilateral disease, the two sides were completed in a staged fashion separated by several weeks.

For all patients with moyamoya, extra precautions are taken to mitigate stroke risk. All patients are admitted the day before surgery for IV hydration with 1.5×maintenance IV fluids, and all patients remain on aspirin up until the day of surgery. During surgery, EEG monitoring is performed, and care is taken to maintain good blood pressure and avoid any episodes of hypotension. After surgery, patients are monitored in the pediatric ICU, and patients remain on IV hydration for several days after surgery to maintain good blood pressure and avoid hypotension.

Review of the electronic medical record identified 22 patients (40.9% male) with a mean age of 7.47 years (SD = 5.65). Of these, 16 patients carried a diagnosis known to predispose to MMS: six sickle cell anemia (SCA), five trisomy-21, and five neurofibromatosis type 1 (NF1). At the time of presentation, the mean modified Rankin Scale (mRS) was 1.05 (SD = 1.59). The majority of patients had clinical (63.6%) or radiographic (68.2%) strokes. Of note, both clinical and radiographic strokes were less frequently seen in NF1 patients (p < 0.01) on presentation. Children with trisomy 21 tended to present at an older age compared to other underlying causes. Additionally, 27.3% of patients were already on an antiplatelet agent at presentation to the neurosurgeon, with the remaining patients started on aspirin at the time of presentation (see Table 1).

Thirty revascularization procedures were performed in the 22 patients. All but one patient (96.7%) underwent an indirect revascularization through either a STA-to-pial synangiosis (N = 28) or a frontal pericranial-to-pial synangiosis (N = 1); one patient underwent a direct STA-to-MCA bypass. Patients requiring bilateral revascularization were more likely to have SCA than trisomy 21 or NF1 (mean number of sides 1.8 (0.8) vs. 1.0 (0.0) and 1.0 (0.0), respectively, p = 0.04). Intraoperatively, there were no ischemic complications and only 13.6% of patients had changes on electroencephalography, all of which normalized with an increase in blood pressure prior to completion of surgery. There were no intraoperative differences based on underlying etiology of MMD or MMS (see Table 2).

Among the 30 operations, seven (23.3%) ischemic events occurred in the peri-operative period within 30 days of surgery. Six (20.0%) of these were transient ischemic attacks (TIAs) while one (3.3%) resulted in a permanent neurological deficit (see Table 2). Additionally, non-perfusion-related complications occurred in five operations: one wound infection, one case of wound breakdown, one seizure, one cerebrospinal fluid leak, and one acute hydrocephalus in a patient with a history of an endoscopic third ventriculostomy (ETV).

Additional stratification of patients was performed by unilateral or bilateral disease and by surgical intervention (index or second procedure). There were 15 patients with unilateral disease and seven patients with bilateral disease. No patient progressed from unilateral to bilateral disease during the study period (Table 3). Demographic analysis demonstrated that the only significant difference between patients with unilateral or bilateral disease was in the rate of stroke as a presenting symptom (46.7% for unilateral disease and 100% for bilateral disease, p = 0.02; see Table 5). There was no difference in the incidence of peri- or post-operative ischemic events based on unilateral or bilateral disease. In the seven patients with bilateral disease requiring contralateral revascularization, there was an increased rate of peri-operative TIAs (9.1% for index surgery and 50.0% for second surgery, p = 0.03) despite there being no difference in intra-operative surgical variables or rate of peri-operative strokes (see Table 4). All four TIAs were ipsilateral to the operative side and resolved without permanent symptoms.

Patients were followed for a mean of 3.5 years clinically (SD = 2.8 years) and 3.9 years radiographically (SD = 2.6). During this time, there were five post-operative strokes (22.3%) on a revascularized side, two post-operative strokes (9.1%) on a non-revascularized side, and one mortality (4.5%) with no differences between underlying disease groups. There was also no difference in functional status based on etiology and the entire cohort’s functional status remained excellent throughout the follow-up period (pre-operative mRS = 1.05, SD 1.59 and post-operative mRS = 0.91, SD = 1.41, respectively, p = 0.38). Angiographic evaluation demonstrated successful revascularization of the ischemic territory with 90.5% of all patients achieving Matsushima A revascularization (see Table 5). A representative angiogram is demonstrated in Fig. 1.

Evaluation of the change in stroke rate was conducted for the total cohort and each underlying cause. The stroke rate decreased in the total cohort (2.45 strokes/year (SD = 3.63) pre-operatively and 0.23 strokes/year (SD = 0.77) post-operatively, p < 0.001) and for patients in each cohort (see Table 6). There was one patient who died in long-term follow-up due to severe progressive moyamoya disease.

Of all the associated conditions assessed here, patients with SCA-induced MMS appeared to have the most aggressive type of MMS. Some prior studies have shown that patients with SCA-associated MMS have high preoperative stroke risk. Newman et al. documented a stroke risk of 1 per 3.43 patient years in patients with SCA-associated MMS before surgical revascularization [14]. The prevalence of moyamoya syndrome in SCA patients has been estimated at 20–35% by cerebral angiography [15, 16]. In our study, patients with SCA-associated MMS were significantly more likely to present with prior ischemic strokes (radiographic or clinical). Additionally, they were more likely than patients in the other groups to present with bilateral involvement. Despite these issues, they responded well to indirect revascularization with a significant post-operative reduction in stroke rate.

Pediatric MMD/MMS classically presents the following ischemic events with published rates of stroke ranging from 67.8 to 92% [7, 11, 17]. The NF1 cohort presented here, however, was significantly less likely to have suffered a clinical or radiographic stroke prior to presentation. This is similar to more recently published reports of low pre-operative stroke rates in this population with clinical stroke rates between 0 and 43.7% and radiographic stroke rates between 6.7 and 56.3% [9, 10, 18]. Additionally, the cited higher rates of infarctions are seen predominantly in NF1 patients who have undergone prior cranial radiation [10]. While it is clear that earlier and incidental detection of MMS can be accomplished by the intensive intracranial screening that NF1 patients receive, our population of NF1 patients demonstrated no difference in age or pre-operative Suzuki score. This suggests that these patients’ MMS was not necessarily detected at an earlier state but rather was detected before signs of ischemic events. When this finding, which is supported by previous literature [10], is combined with the lack of ischemic complications seen in the peri- and post-operative period in both this cohort and previous reports [9, 18], it raises the possibility that the MMS variant associated with NF1 is a more benign pathology than MMS associated with other disease processes such as SCA. Because previous work has demonstrated that NF1-associated MMS is progressive in nature [10] and that earlier surgery is associated with improved outcomes [3, 10], surgical revascularization was still performed in our cohort without evidence of stroke, although most patients had evidence of “ivy sign” changes on MRI.

The differences between patients with unilateral MMD/MMS and those with bilateral MMD/MMS remain a clinically important distinction. While it is clear that there is a group of patients who progress from unilateral MMD/MMS to bilateral disease, determining the precise demographic [3], radiologic [4], and/or biochemical [5] factors predisposing this transition remains a subject of active research. When comparing the cohorts included here, the only difference in presentation between the groups was that patients with bilateral disease were more likely to present with a stroke than those with unilateral disease. Ultimately, this difference did not impact treatment and both groups were able to be safely revascularized with good functional outcomes, a finding was in agreement with previous literature [3, 8, 11].

It is worth noting, however, that despite no difference in patient outcome, more patients with bilateral disease had perioperative TIA’s after the second surgery than the first surgery. As we elected to stage bilateral surgeries on separate days, it is unclear if performing bilateral revascularization on the same day would have led to fewer TIA’s, or if those patients would have had TIA’s regardless of the timing of the second operation. Further research into this finding could better delineate the ideal time for subsequent revascularizations.

It is important to note that, as with many retrospective reviews on MMD/MMS, the small population of included patients limits the power of statistical testing. This lack of power was accepted here rather than clustering patients with MMS together to avoid obscuring differences that may exist among different etiologies of MMS as well as to avoid confounding by an underlying disease.