Comparative outcome of endovascular embolization with microsurgery in managing acute spontaneous cerebral hemorrhage in pediatric patients, an institutional experience

A few previous studies have reported the role of embolization with curative intent in the treatment of the early phase of a spontaneous cerebral hemorrhage in pediatric patients, and its efficacy needs to be compared with surgery at the same time risk factors for hemorrhage following early embolization in such patients need to be evaluated. From a pool of 80 pediatric (< 18 years) who had undergone treatment for ruptured AVM with hemorrhage at our center between July 2018 and July 2022, we identified 36 patients with spontaneous bleeding due to AVM. Out of which, 20 were treated solely by embolization (group 1), while the remaining patients were treated surgically (with and without adjuvant embolization) (group 2). Spetzler-Martin’s grading of the lesion suggested seven lesions < 3 and 13 lesions ≥ 3 in the embolization group. Similarly, seven lesions were < 3 and nine ≥ 3 Spetzler-Martin grade in the surgery group. Incomplete embolization was associated with hemorrhage in two patients treated with curative intent and four patients treated with embolization as adjuvant in the surgery group (p = 0.01). On follow-up, 18 patients in the embolization group and 12 in the surgery group had Glasgow outcome scores ≥ 4 (p = 0.273). In the pediatric age group, incomplete embolization is the significant risk factor for hemorrhage in AVMs treated after a hemorrhagic stroke. Embolization with curative intent is as effective as surgery in treating such lesions as adjuvant embolization with careful patient selection.


Introduction
The causes of spontaneous brain hemorrhage differ between pediatric and adult patients. In pediatric patients (1-18 years), barring neonates, arteriovenous malformations (AVMs), cardiac anomalies, deranged coagulation profile, aneurysm, cavernomas, and vasculitis are the primary causes in descending order of frequency [1,2]. AVMs constitute 70-90% of the etiology of pediatric spontaneous hemorrhage [1,2]. In most cases, multimodal therapy involving adjuvant embolization, radiosurgery, and microsurgical excision is required. Embolization is one of the treatment modalities used to treat such lesions, either alone or in combination with other therapies. Its role in the early stages of hemorrhage caused by AVMs has received little attention, particularly in the pediatric age group [3]. Spontaneous hemorrhage associated with arteriovenous malformation presents a unique set of challenges, and incomplete treatment in the presence of hematoma always carries the risk of rebleeding and hematoma expansion. According to few studies, complete obliteration of AVM following AVM bleed is one of the risk factors. At the same time, incomplete obliteration and the need for multiple shifts of embolization have been documented as risk factors for hemorrhage [4][5][6]. Embolization of AVM in an acute phase where hematoma is causing a mass effect in the form of midline shift is difficult, with little knowledge existing to treat such lesion by embolization with curative intent and its outcome following early embolization in such patients [7]. We attempted to evaluate the efficacy of embolization with curative intent with microsurgical excision (with or without prior adjuvant embolization) of the AVMs with spontaneous cerebral hematoma. In such patients, we also tried to evaluate the risk factors for hemorrhage following early embolization.

Material and methods
Between July 2018 and July 2022, we examined patients treated for arteriovenous malformation in pediatrics (< 18 years) after its rupture in the acute phase with embolization alone and surgery. From a pool of 80 patients who had undergone treatment for ruptured AVM with hemorrhage at our center, we identified 36 patients with spontaneous cerebral hemorrhage with AVM as the etiology. Twenty of these were treated solely by embolization (group 1), while the others were treated surgically (with and without adjuvant). Our study was approved by our institutional ethical committee (IEC No. 1243/21). All patients with ruptured AVM who were diagnosed after a spontaneous intracranial bleed and treated with either microsurgical excision (with and without adjuvant) or endovascular embolization (EVE) alone were included in this study. Patients with dissecting aneurysms, unruptured AVM, dural arteriovenous fistula, a vein of Galen malformation, and carotid-cavernous fistula were excluded.
We obtained the demographic information from patients, such as age and gender, as well as clinical data, such as Glasgow comma scale at admission, duration of treatment since the onset of symptoms, postoperative modified Rankin Scale (mRS) scores, Glasgow outcome score at discharge and subsequent follow-up time between the onset of symptoms and treatment, comorbid conditions, treatment type (microsurgical clipping or coiling), and the occurrence of perioperative events such as intraoperative rupture, hemorrhage, infarct, brain edema, postoperative infarct, hydrocephalus, and a transient and permanent neurological deficit from the medical records of our institution.
We used computed tomography (CT) angiography and 3D image reconstructions, followed by digital subtraction angiography (DSA), to assess blood flow, arterial feeders, feeding arteries, draining veins associated with anomalies such as aneurysms, and AVM size. We performed postoperative CT angiography and DSA on each patient after 24 h and followed up at 6 months and 1 year. The preoperative DSA report included angiographic details such as the type of feeding arteries (end artery, en-passage artery, perforators supplying it), major contributing artery, draining veins (superficial deep), venous ectasia and stenosis, concurrent aneurysm with AVM, size of the aneurysm, and Spetzler-Martin grade. We measured the extent of AVM obliteration after DSA embolization before discharge and at 6-month and 1-year follow-up. After we explained the benefits and drawbacks of each procedure, patients and relatives chose to have EVE or surgery. After obtaining their consent, we planned the treatment using either surgery or EVE. We also obtained the permission of the patients or their relatives to use their data for teaching and clinical research.

Endovascular procedure
General anesthesia was used for the endovascular procedures. Heparin was administered intravenously via catheter pressure infusion system with continuous infusion of 1000 ml normal saline mixed with 1000-unit heparin. The femoral artery was used to access all of the patients. For distal access, we used a 5F/6F long femoral sheath (Cook Medical, Bloomington, IN, USA), Navien ™ Intracranial Support Catheter (Covidien Vascular Therapies, Mansfield, MA, USA), Marathon ™ microcatheter (Medtronic, Minneapolis, Minnesota, USA) or Magic ™ microcatheter (010) (Balt Extrusion Sas, Montmorency, France), and Asahi chikai ™ wire (008) (Asahi Intecc Co., Ltd., Aichi, Japan). We used them to get to the right feeding artery. Moreover, we performed superselective angiography to look for AVM compartments supplied by the feeding artery. Superselective angiography revealed that the supplying artery is either an end artery or an en-passage artery, that the venous outflow contains ectasia, stenosis, and a draining pattern, that the nidal obliteration is complete, and that the venous outflow is obliterated/slowing down. For diffuse nidus, NBCA, Histoacryl ® (B/Braun, Tuttlingen, Germany) was mixed in a 1:3 ratio with Lipiodol ® UltraFluid (Guerbet, Villepinte, France) and for compact nidus, onyx liquid embolic agent (ev3, CA, USA).

Surgery
In cases where the patient could not afford embolization, we used adjuvant embolization before surgery and performed immediate surgery. Patients were operated on as a secondstage procedure within 24-48 h. Patients with GCS was less than 8 were kept intubated for second-stage microsurgical excision and were extubated after initial embolization. As needed in individual cases, we used a tailored approach. For AVMs in the frontal region, we used bifrontal craniotomy, fronto-temporal craniotomy in the temporal lobe, parietal craniotomy in the parietal lobe, and midline suboccipital craniotomy for lesions in the posterior fossa.
We used chi-square tests, Fisher's exact tests for categorical variables, and standard t-tests for continuous variables to conduct statistical analyses of baseline characteristics. Further, univariate analysis was used to examine all factors influencing the outcome, followed by logistic regression analysis. A p-value < 0.05 was considered statistically significant. Statistical analysis was carried out using the SPSS software for Windows, version 22 (IBM Corp., Armonk, NY, USA).

Results
In our department, we treated 80 patients with spontaneous hematomas, 36 of whom had evidence of bleeding from a ruptured arteriovenous malformation. Endovascular embolization with curative intent was used to treat 20 of the 36 patients. The remaining 16 patients in this study were treated with microsurgery, with adjuvant embolization performed in 10 patients with large AVMs with significant mass effect (> 5 mm) and microsurgical excision alone performed without adjuvant embolization in 6 patients.

Clinical and radiological features
The youngest patient in this study was 12 months old, and the oldest was 16 years old. Arteriovenous malformation was responsible for 45% of lesions responsible for spontaneous bleeding, followed by aneurysms (12.5%), cavernomas (7.5), intratumoral bleed (5%), bleeding disorder (3.75%), and vein of Galen malformation (1.25%), and the cause of bleeding could not be determined in 20 patients.
The average age of patients treated by embolization was 10.95 ± 2.98 years, with a minimum age of 1 year and a maximum age of 15 years, and the average age of patients treated by surgery was 11.93 ± 2.54 years, with a minimum age of 2.5 years and a maximum age of 16 years. The maleto-female ratio in the embolization group was 12:8 and 10:6 in the surgery group (p = 0.02). In the embolization and surgical groups, the most specific site was temporal with the perisylvian location ( Fig. 1), followed by temporo-parietal ( Fig. 2) and frontal (Fig. 3) in descending order. The duration of treatment since the onset of symptoms in the embolization group was 7 ± 6.81 days and 4.24 ± 1.89 days in the microsurgery group (p = 0.019) ( Table 1). In the embolization group, there was also subarachnoid hemorrhage in five patients and intraventricular bleed in four patients. In comparison, lobar bleed during surgery was associated with SAH in six cases and IVH in 4 cases ( Table 1).
The average midline shift in the embolization group was 2.12 ± 1.68, while it was 4.24 ± 2.89 mm in the surgery group (p = 0.019) ( Table 1).
A preoperative mRS score ≤ 2 was observed in 19 patients in the embolization group and 12 patients in the surgery group (Table 1).

Angiographic anatomy
Spetzler-Martin's grading of the lesion suggested 7 lesions < 3 and 13 lesions ≥ 3 in the embolization group. Similarly, seven lesions were < 3 and nine ≥ 3 Spetzler-Martin grade in the surgery group ( Table 2).
The middle cerebral artery (MCA) was the primary feeding artery in ten lesions ( Fig. 1A-G), the anterior cerebral artery in three, the watershed location of the ACA-MCA in three [ Fig. 2A-F], and the anterior choroidal artery ( Fig. 3A-F) and vertebrobasilar arteries in two in the embolization group. MCA was the major feeding artery in eight of the surgically treated lesions, ACA in three, watershed areas of ACA-MCA in two, and 1 each in PCA, anterior choroidal, and vertebrobasilar arteries. There were 16 end arteries in the embolization group and 12 in the microsurgery group in this feeding artery. Five feeding arteries were en-passage arteries in the embolization group, and eight were en-passage arteries in the surgery group.
The presence of feeders from perforators such as lenticulostriate, choroidal, and thalamo-perforators was ≤ 2 in four lesions each and ≥ 2 in three lesions in the embolization with curative intent group. Surgery was used to treat three AVMs supplied by > 2 perforators.
Venous drainage was superficial in 13 lesions in the embolization group and 12 lesions in the surgery group (p = 0.106).
Venous ectasia was found in 6 lesions in the embolization group and 13 lesions in the microsurgery group (p = 0.01).
Concurrent aneurysms with AVM were discovered in four lesions on the feeding artery ( Fig. 2B), five in the intranidal (Fig. 1B), and one in the extranidal on the flow-related distal artery in patients solely treated with embolization. A feeding artery aneurysm was discovered in three of the patients who underwent surgery, and an intranidal aneurysm was discovered in four.
The lesion had a high flow rate in nine lesions in the embolization group and seven in the surgery group (p = 0.22) ( Table 2).

Perioperative complications
Periprocedural hemorrhage was observed in one patient who had complete obliteration after embolization ( Fig. 4A-D) with spontaneous resolution on follow-up and in two cases where incomplete embolization was performed initially and required another shift of embolization for complete obliteration. Incomplete embolization was associated with six patients experiencing hemorrhage (p = 0.01). We observed hematoma expansion in three patients who received embolization as an adjuvant before surgery, as evidenced by an increase in midline shift. Moreover, two patients in the microsurgery group had a hematoma in the operative bed when screening NCCT head for deteriorating neurological conditions (Table 3).
Periprocedural infarction was observed in two patients who underwent embolization and four patients who underwent surgery (p = 0.33) ( Table 3). One patient in the embolization group and two patients in the surgery group had hydrocephalus (p = 0.383).
We used a logistic regression analysis to determine the factors responsible for incomplete embolization of the AVM in this subgroup, and we discovered that the lesion size was the most important factor for incomplete obliteration (p = 0.043). Other significant factors included feeding artery (p = 0.11), venous ectasia (0.18), and midline shift (p = 0.11) ( Table 4). We discovered incomplete embolization (p = 0.049) as a significant risk factor for periprocedural hemorrhage in these patients using multiple logistic regression analysis (Table 5).

Angiographic outcome
We achieved complete obliteration of AVM in 17 patients of 20 patients who underwent embolization alone. We used NBCA as a liquid embolization agent in 15 patients and onyx in 5. We were able to completely remove AVMs in 14 of 15 patients who received NBCA and 3 of 5 patients who received onyx as an embolization agent. One of the 15 patients embolized by NBCA required a second-stage embolization, which was performed 2 days later, and 2 patients in the onyx group required a second-stage embolization. Moreover, findings suggest lower risk of recanalization on longer follow-up. The rate of complete obliteration was comparable (p = 0.24). In the current study, NBCA embolization was used in 20 patients with compete obliteration of AVM in 15 patients (75%), and onyx was used in 10 patients with complete obliteration in 5 patients (50%). The complete obliteration of AVM after NBCA appears to be better than onyx, but the difference is not statistically significant (p = 0.178).
Adjuvant embolization was performed an average duration of 1.46 ± 1.12 days before surgery. In five of these cases, NBCA was used as a liquid embolization agent, and in five cases, onyx was used as an embolization agent. Follow-up DSA suggested complete obliteration in 16 patients. Moreover, two patients who underwent microsurgery rebled, developed massive infarcts, and died as a result of their illness (Table 6).

Clinical outcome
Four patients in the surgical group developed transient neurological deficits postoperatively, three of whom had persistent neurological deficits on follow-up. Two patients in the embolization group developed neurological deficits, one of which did not improve on follow-up (Table 3).
We found an mRS score of ≤ 2 in 18 patients in the embolization group and 10 in the surgery group (p = 0.302). During follow-up, 18 patients in the embolization group and 12 in the surgery group had a Glasgow outcome score ≥ 4 (p = 0.273). The average follow-up in the embolization group was 25.80 ± 7.8, and 29.18 ± 11.85 in the surgery group (p = 0.232). It suggests that both groups will have a similar outcome (Table 7).

Discussion
The cause of spontaneous hemorrhage changes with age [1]. Previous research has suggested that AVMs and cardiac and hematological disorders are the most common causes of spontaneous intracranial hemorrhage in children [1][2][3]. A prior understanding of the spectrum of etiologies responsible for spontaneous intracranial hemorrhage allows us to recommend appropriate investigation to diagnose underlying disease. Prematurity, cardiac anomalies, and hematological disorders are the most common risk factors in children under 1 year old, while AVMs, aneurysms, intratumoral bleed, and cavernoma are the most common causes in children aged 1 to 18 years old, according to various studies [1][2][3]. The metaanalysis for etiological factors responsible for spontaneous hemorrhage reported that AVMs caused bleeding in 68.5% of such patients, which is higher than what we found in this study [1,6,8]. Other etiologies, such as aneurysms and rare anomalies, such as a vein of Galen malformation, could be to blame for the spontaneous bleeding in our patients. Similar to previous studies, features of raised intracranial pressure, altered sensorium, and seizures were the leading symptoms in descending frequency [8]. Lower GCS of patients in previous studies have been reported as poor predictors of outcome following the treatment of pediatric AVMs [9]. Patients in our study have an average GCS of 13 ± 1.8 in the embolization group, which is of moderate severity on the head injury severity scale (Glasgow coma scale) compared to 10 ± 1.21 in the surgery group, which may have aided patients undergoing embolization alone in their postoperative recovery. Intraventricular bleed has been identified as a poor prognostic factor for postoperative outcome [8,9]. In an earlier study, GCS less than eight was found to be a poor predictor of outcome [10]. In the current study, the number of patients undergoing embolization and surgery was comparable, and both groups responded to external ventricular drainage of CSF without the need for a ventriculoperitoneal shunt. Intraventricular bleed has been associated with poor outcomes in a few studies, but the number of patients in the current study was lower when compared to previous studies that also responded to CSF diversion via EVD [9][10][11]. The midline shift caused by the mass effect of intracranial bleed forces us to choose between surgical treatment and embolization while keeping the patient on conservative treatment. Studies have suggested that a midline shift of > 5 mm results in unfavorable outcomes [11][12][13]. In the current study, all of the patients who underwent embolization had a midline shift of less than 5 mm, which may have resulted in a favorable outcome. Compared to embolization, surgery was preferred for midline shift > 5 mm in 37% of our patients, with similar outcomes on follow-up. None of our patients who were treated by embolization alone had a bleeding disorder or cardiac anomalies, which are associated risk factors for the expansion of hematoma, as reported in a few studies [5].
There was a similar distribution of Spetzler-Martin's grade in patients treated by embolization with curative intent and other groups treated by surgery, with 25% in grades 1 and 2, 45% in grade 3, and 20% in grades 4 and 5. The Spetzler-Martin's grade of patients in the current study is similar to the other study reported on patients having spontaneous bleeding due to AVMs [15][16][17][18][19][20][21][22]. In the present study, we included patients with hemorrhagic stroke, but in the previous studies, the ratio of ruptured and unruptured AVMs varied from 70 to 90% [15][16][17][18][19][20][21][22]. Decision-making with the intent to cure in a circumstance like in the present study requires continuous monitoring to avoid periprocedural rebleed and subsequent changes in the treatment modality. Because the size of the lesion in the current study was smaller and more lesions were in the non-eloquent location, the outcome was better than in previous studies. According to different studies, both in the embolization and surgery groups, eloquent location, deep venous drainage system, and larger size are risk factors for periprocedural rebleed and poor outcomes [14,15]. In the current study, none of the lesions is larger than 6 cm. Although more patients had lesions in prominent locations in the current study, they were predominantly draining into the superficial venous system, resulting in a lower rebleeding rate when compared to previous studies [13][14][15]. The average number of end arteries and perforators feeding the nidus was lower in the current study compared to the previously reported study, which resulted in complete obliteration of the nidus with curative intent in nearly 75% of the patients in such circumstances [14,17,19]. Similarly, in the surgery group, the size and number of feeders were lower than in the study reported by Lawton et al. [15]. We use n-butyl cyanoacrylate (NBCA) in the majority of patients (75%), and we observed permanent neurological deficit in 5% of those who had embolization alone and 11% who had it as an adjuvant to surgery. Previously reported studies found that NBCA caused 2.5-6.5% permanent disability [24][25][26]. Our findings and these reports suggest that NBCA is still useful in these situations, whereas liquid embolic agents such as onyx, squid, and PHIL are becoming more popular [14]. Periprocedural hemorrhage was found to be significantly associated with AVM incomplete obliteration in the current study. Earlier non-visualization and a slowing of venous drainage were observed in these patients. It could be due to prior to surgery with a 90% obliteration rate of brain AVM [27,28]. Steiger et al. reported early bleeding within 24 h in all the patients with hemorrhage following embolization and congested nidus even after 11 days of embolization, indicating the persistent risk of bleeding due to incomplete embolization, implying that early embolization is beneficial [29]. In the current study, we chose early embolization with an average time interval similar to the previous studies [26][27][28][29].
In their study on the comparative analysis of AVM embolization after NBCA and onyx, Crawley et al. and Izumo et al. found no difference in the outcome, but multiple sessions were required for complete embolization by onyx and lesser time taken to embolize by NBCA [30,31]. In our study, more patients, who were embolized initially by onyx, required second-stage embolization as compared to patients treated with NBCA. In their series of 73 consecutive patients who underwent embolization with curative intent for ruptured and unruptured SM 1 and 2 AVMs between 2008 and 2016, Iosif et al. achieved 95% angiographic cure rates with procedure-related morbidity and mortality rates of 2.7% and 0%, respectively [32]. Wu et al. published a systematic review in 2019 that included 15 studies that included 597 patients with 598 AVMs [33]. An angiographic cure rate    of 58.3% was reported, with overall clinical complication and procedure-related mortality rates of 24.1% and 1.5%, respectively. We were able to completely obliterate 75% of AVMs due to our careful use of NBCA and more favorable anatomy. Meisel et al. reported a significant reduction (30-40%) in recurrent hemorrhage in ruptured AVM patients [34]. We have not received a recanalization patient after embolization, despite the fact that the follow-up period is shorter in a recent study. Steiger et al. made similar findings after a 6.5-year follow-up period [29]. We found no significant difference in clinical outcome, as measured by Glasgow outcome score and modified Rankin Scale score, between patients treated with embolization with curative intent and those treated with surgery in the early stages of bleed, which is consistent with an earlier reported study [32,33].
We did not observe any mortality patients who underwent embolization alone, but two patients died as a result of their illness after receiving incomplete obliteration as an adjuvant and undergoing surgery due to rebleed and subsequently having a prolonged ventilatory requirement and multiple organ dysfunctions. Mortality and morbidity in the current studies are comparable to those reported in previous studies where embolization is used as a curative and adjuvant treatment [14][15][16][17][18][19][20].

Conclusion
AVMs treated with embolization with curative intent may have similar outcomes to surgery (with embolization as adjuvant) in carefully selected pediatric patients with hemorrhagic stroke. Incomplete embolization increases the risk of hemorrhage, emphasizing the importance of complete obliteration of AVMs in such lesions in pediatric patients. NBCA can be used with the same efficacy as other liquid embolization agents such as onyx to achieve complete obliteration of AVMs in such a scenario. Our findings, while significant and suggest the role of embolization as a curative treatment in pediatric patients with hemorrhagic stroke, require further prospective research on a larger patient population with longer follow-ups to validate our findings.
Author contribution Conceptualization, clinical work, data collection, data analysis, manuscript drafting, and revision had been done by VCJ. Data collection and analysis had been done by VCJ, MSA, and VSS. Data analysis and manuscript supervision was done by VCJ, MSA, and VSS. All the authors have read and approved the final version of the manuscript. This manuscript has neither been presented as a whole nor as part of any conference or scientific meeting. This article is neither published nor under consideration for publication anywhere else.
Availability of data and materials It will be produced at a reasonable request from the corresponding author.

Declarations
Ethics approval and consent to participate Written informed consent was taken from patients/relatives at the time of admission to use their data for teaching and research. Approval of the institutional ethical committee with approval no. IEC/2021/1243 was taken to conduct this study.

Conflict of interest
There is no conflict of interest among authors.