Management of Pediatric Intracranial Arteriovenous Malformations: Experience With Multimodality Therapy

Abstract

BACKGROUND:

Successful management of pediatric arteriovenous malformations (AVMs) often requires a balanced application of embolization, surgery, and radiosurgery.

OBJECTIVE:

To describe our experience treating pediatric AVMs.

METHODS:

We analyzed 120 pediatric patients (< 18 years of age) with AVMs treated with various combinations of radiosurgery, surgery, and endovascular techniques.

RESULTS:

Between 1985 and 2009, 76 children with low Spetzler-Martin grade (1–3) and 44 with high-grade (4–5) AVMs were treated. Annual risk of hemorrhage from presentation to initial treatment was 4.0%, decreasing to 3.2% after treatment initiation until confirmed obliteration. Results for AVM obliteration were available in 101 patients. Initial single-modality therapy led to AVM obliteration in 51 of 67 low-grade (76%) and 3 of 34 high-grade (9%) AVMs, improving to 58 of 67 (87%) and 9 of 34 (26%), respectively, with further treatment. Mean time to obliteration was 1.8 years for low-grade and 6.4 years for high-grade AVMs. Disabling neurological complications occurred in 4 of 77 low-grade (5%) and 12 of 43 high-grade (28%) AVMs. At the final clinical follow-up (mean, 9.2 years), 48 of 67 patients (72%) with low-grade lesions had a modified Rankin Scale score (mRS) of 0 to 1 compared with 12 of 34 patients (35%) with high-grade AVMs. On multivariate analysis, significant risk factors for poor final clinical outcome (mRS ≥ 2) included baseline mRS ≥ 2 (odds ratio, 9.51; 95% confidence interval, 3.31-27.37; P < .01), left-sided location (odds ratio, 3.03; 95% confidence interval, 1.11-8.33; P = .03), and high AVM grade (odds ratio, 4.35; 95% confidence interval, 1.28-14.28; P = .02).

CONCLUSION:

Treatment of pediatric AVMs with multimodality therapy can substantially improve obliteration rates and may decrease AVM hemorrhage rates. The poor natural history and risks of intervention must be carefully considered when deciding to treat high-grade pediatric AVMs.

Arteriovenous malformations (AVMs) rarely present in children and account for only 3% to 20% of all AVMs. However, these lesions are responsible for 30% to 50% of spontaneous intracranial hemorrhages in children, leading to high rates of morbidity and mortality.^1–4 The annual risk of AVM hemorrhage, estimated to be 2%/y to 4%/y,^5–7 leads to a greater cumulative lifetime risk for children.⁸ Furthermore, the superior clinical outcomes after treatment in children compared with adults bolster early aggressive efforts to eradicate the AVM and to protect children from future hemorrhage.⁹

Current AVM treatment options include microsurgery, endovascular embolization, and stereotactic radiosurgery, alone or in combination.^8,10 Choice of the most appropriate treatment modality depends on the results of treatment in terms of AVM obliteration and complication rates, as well as final clinical outcomes. The Spetzler-Martin grading scale, developed in adults to predict the results of surgical resection, has also been used successfully to predict outcomes for lesions treated with multimodality therapy.¹¹

Accessible, low-grade (Spetzler-Martin grade 1–3) AVMs are often best managed with surgery, which offers immediate protection from rebleeding with a low risk of complications (0%-12%) and a rate of cure exceeding 80%.^1,8,10,12,13 Low-grade but inaccessible AVMs or those in eloquent locations are often treated with stereotactic radiosurgery, which, for appropriately selected lesions, offers a cure rate of 81% with a 5% risk of complications.¹⁴ However, radiosurgery requires up to 3 years for AVM obliteration, during which time the child is at risk for AVM hemorrhage, and many are reluctant to use radiation in a developing nervous system. Endovascular AVM embolization, rarely curative in isolation, can be a useful adjunct to decrease operative blood losses, an especially important factor in this patient population. Embolization can also effectively decrease AVM size to increase radiosurgical cure rates, although this remains controversial.¹⁵

The best management of high-grade (Spetzler-Martin grade 4–5) pediatric AVMs is less clear. By virtue of their greater size, high-grade lesions are more likely to involve deep or eloquent brain tissue, often rendering safe surgical resection difficult or impossible. The increased size (and volume) of high-grade lesions also contributes to lower radiosurgical obliteration rates.¹⁶ High-grade AVMs carry a greater risk of treatment-related complications,^11,17 and some have suggested that partial treatment may worsen the natural history.^18–20 On the other hand, the poor natural history of AVMs in children,^8,21,22 in conjunction with the known resiliency and ability of the pediatric nervous system to recover from injury,^23,24 supports an aggressive management philosophy.

We present our 23-year experience with pediatric AVMs using all treatment modalities at our disposal to better define the natural history and results of treatment for children with low- and high-grade AVMs.

METHODS

A prospectively maintained database (MD Analyze, Medtamic Inc) of vascular malformation patients treated at Stanford University Medical Center was used to identify all patients whose first AVM treatment occurred before 18 years of age between 1985 and 2009. Patients referred from other centers after partial treatment were not excluded. A total of 120 patients with pial AVMs were identified. Patients with spinal AVMs, vein of Galen or cavernous malformations, and dural arteriovenous fistulas were not included in this series.

Natural history data were collected on all patients. The number of hemorrhages was recorded during 2 separate intervals: time of presentation to initial treatment and initial treatment to angiographic confirmation of AVM obliteration or final clinical follow-up, as detailed elsewhere.²⁵ Hemorrhage was defined as a clinical event with acute onset of ≥ 1 symptoms, including headache, loss of consciousness, or focal neurological deficit, and confirmed with computed tomography (CT) or magnetic resonance imaging (MRI) demonstrating acute intracranial hemorrhage.

We evaluated the patient's sex, age at initial treatment, baseline modified Rankin Scale score (mRS), presentation type, AVM Spetzler-Martin grade (size, eloquence, venous drainage), location, associated aneurysms, treatment modality and number of treatments, and treatment-related complications. Each patient was evaluated by a multidisciplinary team comprising neurosurgeons, interventional neuroradiologists, radiation oncologists, and in some cases a neurologist. The decision of which treatment modalities to use was based on the team's assessment of the child's history and clinical status and the AVM location and characteristics.

Patients were treated with single-modality therapy (surgery, radiosurgery, or embolization alone), dual-modality therapy (2 of radiosurgery, embolization, or surgery) or triple-modality therapy. Treatment results were calculated per patient and per treatment session.

Endovascular Therapy

During the period of time covered by this review, various embolic materials were used, including thrombogenic coils, silk threads, polyvinyl alcohol particles, n-butyl cyanoacrylate glue, and Onyx liquid embolic. Superselective arterial catheterization with provocative amytal testing was performed in all cases when eloquent brain was considered to be at risk. Typically, the goal of embolization was not to cure the AVM but rather to reduce operative blood loss or to reduce the size of the AVM. Mean arterial blood pressure was strictly controlled between 65 and 75 mm Hg (or age-appropriate range) for 18 to 24 hours after embolization, and a head CT scan was obtained after each embolization to rule out hemorrhage. Associated aneurysms, when present, were embolized before the AVM. When treatment required multiple sessions of embolization, they were typically spaced 1 to 2 weeks apart.

Microsurgical Treatment

Stereotactic localization, functional testing, cortical localization, and intraoperative angiography were used as surgical adjuncts as necessary. Patients undergoing microsurgery were monitored with somatosensory evoked potential, motor evoked potential, electroencephalogram, and/or brainstem auditory evoked potential as clinically indicated. Surgery for clipping of AVM-associated aneurysms, when necessary, was not counted as a surgical procedure. As with embolization procedures, postoperative mean arterial blood pressure was strictly maintained within the 65- to 75-mm Hg (or age-appropriate) range for the first 24 hours and then gradually relaxed over the following 48 hours, especially when surgery was staged and there was known residual AVM.

Stereotactic Radiosurgery

Radiosurgical treatment consisted of charged-particle radiation, linear accelerator, CyberKnife, or Gamma knife treatments, all of which were performed on an outpatient basis. General anesthesia was typically necessary for children < 12 years of age. Patients were followed up with an MRI at 6 months and then on an annual basis, with confirmatory angiography performed at 3 to 4 years if the MRI suggested obliteration.

Analysis of Results and Statistical Methods

Patients receiving their first radiosurgical treatment within the last 3 years and those without angiographic or MRI results were excluded from the treatment analysis. Angiographic outcome was obtained by reviewing the available electronic, paper, and film records. Catheter angiography results were used to determine AVM obliteration, with MRI used only when catheter angiography results were not available. Patients initially treated with endovascular or surgical means with residual AVM were subsequently treated with radiosurgery. The initial treatment period was considered to be completed either when the AVM was cured with surgical or endovascular treatment (without radiosurgery) or at 3 years after radiosurgical treatment. The incidence of permanent neurological complications was recorded and classified as major or minor. A complication was counted as major (disabling) under 2 situations: either when a patient with a pretreatment baseline mRS of 0 to 1 progressed to mRS ≥ 2 or when a patient with a pretreatment baseline of ≥ 2 had any increase in mRS. Complications were considered minor and nondisabling when they did not lead to a permanent change in mRS or only increased the mRS from 0 to 1. Clinical outcomes were graded with the mRS at discharge and at final clinical follow-up. When patients had multiple admissions for AVM treatment, the discharge mRS was recorded at the time of discharge after the final treatment. A telephone interview was also conducted with patients, parents, or primary caregivers to optimize the clinical follow-up.

Hemorrhage rates before and after the first treatment were estimated by the number of hemorrhages before and after the first treatment divided by the corresponding total person-years. The 95% confidence interval (CI) of the difference in hemorrhage rates before and after first treatment was constructed with the bootstrap method. We also used the proportional hazard model with time to the first hemorrhage event as the response and treatment status as a time-dependent covariate for testing the association between treatment and hemorrhage risk. Separate multiple logistic regressions were performed to evaluate the association between sex, age (≤ 12 vs > 12 years), baseline mRS (0–1 vs ≥ 2), AVM location (left vs right), deep or superficial AVM location, deep or superficial venous drainage, eloquent cortex or not, Spetzler-Martin grade (low [1-3] vs high [4 and 5]), AVM size, prior hemorrhage or not, and AVM-associated aneurysms or not with hemorrhage risk after adjustment for the time of final follow-up. Furthermore, similar logistic regressions were performed to examine the association between the aforementioned risk factors and poor long-term clinical outcomes (mRS ≥ 2) after adjustment for time of final follow-up. Lastly, multivariate logistic regressions were performed using time of the final follow-up, with the independent variables selected from all factors significant to a level of 0.1 in the individual analyses. All statistical analyses were performed with R (2.8.0). Significance was set at 0.05.

RESULTS

Demographics and AVM Characteristics

Presented in Table 1 are patient demographics, type of presentation, baseline mRS, Spetzler-Martin AVM grade, location, and associated aneurysms.

Natural History and the Effects of Treatment

Natural history data were calculated for all 120 patients with the person-years method and are presented in Table 2. Five children had 9 hemorrhages after presentation but before first treatment, a period of time that comprised 124.6 patient-years, for an annual per-patient hemorrhage rate of 4.0%/y. After initial treatment but before angiographically confirmed obliteration, 18 children had a total of 28 hemorrhages over a combined 566 patient-years, for a posttreatment per-patient hemorrhage rate of 3.2% (Figure 1A). The difference in hemorrhage rate before and after initial treatment did not reach statistical significance (95% CI, −5.88-2.20; P = .46). The difference in hemorrhage rate between partially treated and occluded AVMs did not reach statistical significance (95% CI, 0.19-1.25; P = .13; Figure 1B). The difference in hemorrhage rate before treatment (natural history) compared with the AVM hemorrhage rate after treatment (with obliterated AVMs censored once cure was confirmed) was significant (95% CI, 1.74-58.4; P = .01; Figure 1C). In the proportional hazards analysis, the hazard ratio of hemorrhage after first treatment vs hemorrhage before first treatment was 0.695 (95% CI, 0.08-5.76; P = .75). Univariate analyses of risk factors for hemorrhage, adjusted for follow-up time, are presented in Table 3. No factor reached statistical significance.

AVM Obliteration Rates After Single-Modality, Dual-Modality, and Multimodality Therapy

The patients were divided into low-grade (1–3) and high-grade (4–5) AVM groups. Fifteen children were excluded from analysis because they were within 36 months of initial radiosurgical treatment; 4 others were excluded because no angiographic or MRI results were available. The AVM obliteration results after endovascular embolization, surgery, and radiosurgery, alone or in combination, for initial and additional AVM treatments are presented in Table 4. Catheter angiography was used to confirm AVM obliteration in 61 of 67 cure determinations (91%), with MRI used for the remainder.

Single-Modality Therapy

Endovascular Embolization

There were 195 embolization procedures in 120 patients (range, 1-15 procedures; mean, 2.6 embolizations per patient). There were 15 endovascular complications, leading to a per-session complication rate of 7.7%. The sum total of the observed complications for each modality differs from that calculated in the per-patient analysis because some children had > 1 complication. Of the 101 children included in the AVM obliteration analysis, 49 (48%) underwent embolization as initial therapy, with an initial cure rate of 2/49 (4%). Both lesions cured with embolization were low-grade AVMs: a frontal grade 1 lesion in a 13-year-old boy and a cerebellar grade 2 lesion in an 8-year-old boy.

Surgery

There were 104 surgical procedures performed in 120 children (range, 1-4 surgeries per patient). Overall, there were 10 of 104 neurological complications after surgery, for a per-session complication rate of 9.6%. Of the 101 children included in the AVM obliteration analysis, 27 of 101 children (26%) had surgery as their initial treatment, with an obliteration rate of 18 of 27 (67%). All AVMs treated with surgery alone as initial therapy were low-grade AVMs, with a mean Spetzler-Martin grade of 2.22.

Radiosurgery

There were 106 sessions of stereotactic radiosurgery (range, 1-4 sessions per patient), as well as 6 sessions of staged-volume radiosurgery. The doses of radiation used ranged from 11.5 to 89 GyE, with a mean of 21.2 GyE. The volume of tissue irradiated ranged from 200 to 113 000 mm³, with a mean of 27 400 mm³. There were 8 neurological complications after radiosurgery, leading to a per-session complication rate of 7.5%. In the analysis of AVM obliteration rates, stereotactic radiosurgery was the initial treatment for 25 of 101 children (24%), who harbored 12 low-grade and 13 high-grade AVMs, with an obliteration rate of 5 of 25 (20%). All 5 AVMs successfully obliterated with radiosurgery were low-grade lesions (mean grade, 2.6), whereas the entire group initially treated with radiosurgery had a mean grade of 3.4. An illustrative case is shown in Figure 2.

Dual-Modality Therapy

After embolization, of 47 children with residual AVMs, 30 went on to have surgery, leading to cure in 19 of 30 (63%). The other 17 patients were treated with radiosurgery, with cure obtained in 4 (24%).

Eight patients (all with low-grade lesions) had residual AVMs after their initial surgical treatment. The family of 1 patient who had presented in a coma from a cerebellar AVM refused further treatment owing to the poor clinical condition of the child. The remaining 7 children underwent radiosurgery to the residual, which led to cure in 3 of 7 (43%). Illustrative cases are shown in Figures 3 and 4.

Triple-Modality Therapy

The 11 children (3 with low-grade and 8 with high-grade AVMs) with residual AVMs after embolization and surgery then underwent radiosurgery. Triple-modality therapy resulted in cure in 2 of 11 (18%): in a grade 3 frontal lesion in a 16-year-old girl and a grade 4 frontoparietal AVM in a 5-year-old girl. Illustrative cases are shown in Figures 5 through 7.

Overall AVM Obliteration Rates After Initial Treatment

After initial treatment, 54 of 101 children (54%) with AVMs were cured; of these, 51 of 67 (76%) had low-grade AVMs, and 3 of 34 (9%) had high-grade AVMs. After initial treatment, the 47 children with residual AVM had the following dispositions: 30 underwent further treatment for the AVM, 2 relocated overseas, 5 were lost to follow-up, and 10 declined further treatment.

Results of Further Treatment for Residual AVMs

After a 3-year latency period, 6 children previously treated with radiosurgery, all with high-grade lesions, had residual AVMs though to be amenable only to repeat radiosurgery. This course of treatment led to 1 additional cure, for a total cure rate of 6 of 25 (27%) for patients treated with radiosurgery alone. Twenty-four other children with residual AVMs were treated with either repeat dual- or triple-modality therapy, leading to 12 additional cures. The overall pediatric AVM cure rate with all treatment modalities available was 67 of 101 (66%): 58 of 67 (87%) for patients with low-grade AVMs and 9 of 34 (26%) for patients with high-grade AVMs. These obliteration rates may change because 5 children either are continuing with staged embolizations or surgery or are within the latency period of radiosurgery. Four other children are due for repeat imaging, having completed 3 years since radiosurgery.

AVM Obliteration Results per AVM Grade

The overall AVM obliteration rates and time to achieve obliteration per Spetzler-Martin grade are presented in Table 5. Rates of AVM cure are presented in terms of best- and worst-case scenarios. The best-case scenario removes the 8 patients with indeterminate final angiographic outcomes from the denominator and assumes that MRI accurately proves obliteration²⁶ in the 6 cases for which only MRI results were available for determining cure. The worst-case scenario assumes that the 8 patients currently undergoing further therapy will not achieve obliteration and that the MRI results suggesting obliteration are falsely negative.

The time to achieve AVM obliteration increased per Spetzler-Martin grade, with low-grade lesions requiring a mean of 1.8 years and 2.2 procedures to achieve cure compared with a mean of 6.4 years and 6.1 procedures for high-grade AVMs. Only 1 grade 5 lesion was obliterated.

Recurrent AVMs

Before we instituted the practice of confirming angiographic cure in children after a period of at least 6 months, we had 3 children with immediate posttreatment negative angiograms who were found to harbor AVMs on late follow-up. A fourth child demonstrated residual/recurrent AVM 4 years after a delayed negative angiogram at almost 2 years. Two of these children had initially presented with an AVM hemorrhage.

The first patient, a 15-year-old girl at the time of initial treatment, presented with headaches and was found to harbor a grade 4 right occipital AVM. She underwent 2 staged embolization procedures followed by surgical resection of her AVM, with postoperative angiography (2 days after surgery) demonstrating cure. She then presented 16 years later with increasing headaches and intermittent visual symptoms; a small recurrent AVM was confirmed on angiography, and she underwent uneventful repeat surgical resection.

The second patient, a 16-year-old boy, was treated with endovascular therapy, had an angiographic cure after embolization, but had residual AVM found on angiographic follow-up at 4 months. He subsequently underwent successful surgical removal of the AVM. The third child, 11 years old at time of treatment, was treated with surgical resection. His immediate postoperative angiogram was negative for AVM. He suffered a minor late rehemorrhage 9 years later, and the recurrent/residual AVM was subsequently resected uneventfully. The final patient, a 7-year-old boy with a deep temporal AVM, underwent partial surgical resection with planned postoperative radiosurgery. However, a planning angiogram performed 21 months after surgery failed to demonstrate any residual nidus, and the family elected to wait for 4 years before performing repeat imaging, at which time angiography demonstrated residual AVM. This lesion was also resected successfully.

Overall Treatment-Related Complications per AVM Grade

Treatment-related complications were calculated for all 120 patients and are presented in Table 6 and Figure 8. Major disabling complications occurred in 4 of 77 patients (5%) with low-grade AVMs compared with 12 of 43 patients (28%) with high-grade AVMs. The complication rate was seen to decrease over time; of 36 treated AVMs from 1983 to 1994, there were 15 complications (42%), decreasing to 12 of 50 (24%) from 1995 to 2004, with only 4 of 34 complications (12%) occurring from 2005 to 2009.

Clinical Outcome per AVM Grade

Presented in Table 7 are the proportions of patients with mRS scores of 0 to 1 at baseline and at final clinical follow-up, the number of children with a stable or improved clinical course, and the number of late hemorrhages and deaths according to Spetzler-Martin grade. Mean follow-up was 9.2 years. In general, as the AVM grade increased, the proportion of children with mRS of 0 to 1 at final follow-up was seen to decrease. Eighteen children had late hemorrhages from their AVM, leading to a decrease in neurological function in 12 and death in 7. To determine the proportion of patients who became disabled (mRS ≥ 2) owing to late AVM hemorrhage (and not directly as a consequence of AVM treatment), the 8 patients who deteriorated neurologically after AVM hemorrhage (1 low-grade and 7 high-grade lesions) were returned to the numerator. This yielded a proportion of patients with a final mRS of 0 to 1 after treatment (excluding the effects of late hemorrhage) of 49 of 67 (73%) for low-grade and 19 of 34 (56%) for high-grade AVMs.

Risk Factors for Poor Outcome

The results of statistical analysis of risk factors for poor outcome, considered a final mRS ≥ 2, are presented in Table 8 and Figure 9. On univariate analysis, the following factors were predictive of poor final outcome: baseline mRS ≥ 2, high-grade AVMs, deep AVMs, AVMs located on the left side, AVMs involving eloquent cortex (all P < .05), and AVMs with associated aneurysms (P = .05). Sex, age < 12 years, AVM size, prior hemorrhage, and deep venous drainage were not statistically significant predictive factors. After multivariate analysis, baseline mRS ≥ 2 (odds ratio, 9.51; 95% CI 3.31-27.27; P < .01), left-sided lesions (odds ratio, 3.03; 95% CI, 1.11-8.33; P = .03), and high AVM grade (odds ratio, 4.35; 95% CI, 1.28-14.28; P = .02) remained significant predictors of poor final outcome.

DISCUSSION

This is a large, historical series of children treated with multimodality (endovascular, surgical, and radiosurgical) therapy for both low- and high-grade AVMs. The size of this series permits statistical analysis of risk factors for hemorrhage and poor outcome. Our relevant findings include the demonstration of a natural history of pediatric AVMs similar to that of adult AVMs and indicate that treatment initiation does not predispose an AVM to hemorrhage and that aggressive multimodality treatment for high-grade AVMs has a low cure rate and a high associated rate of complications. Poor pretreatment functional status, left-sided location, and high-grade lesions were found to be statistically significant predictors of poor outcome for children with AVMs. Finally, the mean time to achieve AVM obliteration was seen to increase with increasing AVM grade.

Natural History of Pediatric AVMs

In the absence of Level 1 evidence, the decision on whether to treat brain AVMs continues to be made by comparing the natural history with treatment outcomes, in terms of treatment efficacy, risk of complications, and final clinical outcomes. The natural history for high-grade AVMs is incompletely known; hemorrhage rates vary between 2%/y and 10%/y, depending on the methodology used for calculation.^19,25 Methods that calculate AVM hemorrhage rate from birth to treatment increase the size of the denominator in the person-year method, leading to a low annual hemorrhage rate, whereas methods starting at the time of presentation until treatment decrease the denominator, yielding higher annual rates. Because of the referral patterns for complex AVMs, there was often a delay between presentation and treatment initiation, creating an opportunity to study the natural history. The annual hemorrhage rate of 4.0% calculated from the time of presentation to initial treatment is in accordance with the accepted hemorrhage rate of 2% to 4% for adults harboring brain AVMs.^5–7 It is controversial whether pediatric AVMs may have a greater risk of hemorrhage^1,22,27 or the same risk of hemorrhage⁹ as adult AVMs. Our data suggest that the greater known proportion of hemorrhagic presentations in children may result from the decreased tendency of pediatric AVMs to present with nonhemorrhagic symptoms compared with adults.

Statistical analysis seeking predictors of hemorrhage did not yield any statistically significant predictors, although trends towards significance were seen for male sex (P = .06) and high AVM grade (P = .07) on univariate analysis. The AVM-associated aneurysms were not found to increase the risk of hemorrhage significantly (P = .23). The small population size and relatively small number of hemorrhagic events in this series may conceal a significant association.

Effects of Treatment Initiation on Rupture Rates

Our data suggest that treatment initiation does not destabilize the AVM and increase the risk of hemorrhage; it may even decrease the risk of subsequent hemorrhage. That partial treatment does not predispose an AVM to hemorrhage has been suggested for a cohort of mostly adult AVMs.²⁸ Given the time required to achieve cure for some of the high-grade AVMs in this series (up to 6.4 years), knowledge of safe partial treatment offers reassurance; however, the importance of strict blood pressure control after partial AVM treatment must be emphasized. No periprocedural hemorrhages occurred in the children treated at Stanford; the sole perioperative death resulting from hemorrhage that we report in this series occurred at another hospital after the family refused to accept our recommendations for conservative management after multiple palliative embolization attempts of a grade 5 holohemispheric AVM.

Treatment of Low-Grade AVMs

The greater cumulative lifetime AVM hemorrhage risk and the known resilience of children to treatment led us to pursue an aggressive management policy for all pediatric AVMs. With all treatment modalities available at our disposal, we were able to obliterate between 78% and 87% of low-grade AVMs using the best-case/worst-case scenario method.²⁹ These results are in keeping with many previous reports of low-grade pediatric AVMs, with reported cure rates around 80%.^1,8,10,12,13 In this series, we observed a 5% major (disabling, mRS ≥ 2) and 6% minor (nondisabling) complication rate for low-grade AVMs. Previous series of mainly low-grade pediatric AVMs treated with various modalities, including multimodality treatment, describe complication rates of 5% to 20%.^{10,12,14,30,31} Our center has served as a referral center for complex AVMs for many years, and our complication rate may be a reflection of the higher proportion of large AVMs and those located in eloquent cortex or in deep locations (but still classified as low-grade AVMs).

Treatment of High-Grade AVMs

The 35 pediatric patients with high-grade AVMs in this series make up the largest group for whom results of multimodality therapy have been reported. With the best-case/worst-case scenario used to report angiographic results, aggressive multimodality treatment resulted in cure in 26% to 35% of high-grade AVMs, with a 28% major and 23% minor complication rate. Our results are similar to those of Bristol et al,¹³ who were able to achieve AVM obliteration in 9 of 25 children (36%) with high-grade AVMs treated with multimodality therapy.

In general, on discovery of a high-grade AVM, lesions are assessed for suitability for embolization followed by surgery with radiosurgery to portions of residual AVM. Three years of watchful waiting for AVM involution follows. After completion of the latency period, when residual AVM is confirmed, active treatment is reinitiated with more embolization, surgery, and/or repeat radiosurgery. This process continues until the AVM is obliterated, the child or the family declines further therapy, or the patient is lost to follow-up. The persistence of patients and their families in striving for AVM obliteration, as well as their tolerance for complications, thus affects the rate of AVM cure. Patients and their families are more likely to remain committed to a treatment plan when they can anticipate the time, number of treatments, and risk of complications that they must typically accept to achieve AVM obliteration. The importance of clear communication should be emphasized. Patients, their families, and physicians must recognize that optimal management and monitoring of patients with high-grade AVMs may entail long-term involvement with the treatment team. Patients with incompletely treated AVMs who declined further treatment often chose to remain in contact with the treatment team. They typically returned for a brief clinic visit every several years to ascertain that their clinical condition had not changed, so we could inform them then regarding any new potential therapeutic options.

Recurrent AVMs

Establishing what constitutes a recurrent as opposed to a residual but initially undetected AVM requires good-quality diagnostic angiography performed in a delayed fashion after treatment. Several groups have reported recurrent AVM after angiographic demonstration of cure.³¹ Most reported cases, however, were imaged immediately after surgical treatment,^31,32 which may fail to detect residual AVM. The most commonly invoked causes for an initially undetected residual lesion, other than inadequate angiographic investigation, include compression of very small shunting veins by adjacent hematoma or edema after microsurgery and vasospasm of small feeders. Because this phenomenon is known to be more common in the pediatric age group, we now recommend an intraoperative (or immediately postoperative) angiogram to exclude residual AVM, followed by a repeat angiography 3 to 6 months after treatment for all children with AVMs. Only if the delayed angiogram shows no AVM is a cure proven. Applying this paradigm to the current series would result in only 1 true recurrent AVM (recurrence rate of 0.8%) and 3 uncertain cases of residual vs recurrent AVMs. To exclude any late recurrences, we also currently recommend 1 additional repeat angiography 5 years after the 3- to 6-month postoperative angiogram showing a cure.

Complication Rates

Improved technology, in terms of improved endovascular and surgical tools and equipment, refinements in radiosurgical dosing plans, and increasing recognition that perhaps not all AVMs should be treated, may account for the observed decreased complication rates over the course of this series. The patients treated with radiosurgery within the last several years and those patients who required multiple CT scans or embolization sessions remain at risk of developing delayed radiation-related complications, which may subsequently slightly increase this rate. Fast MR scans, when possible, are now the preferred cross-sectional imaging modality used to rule out hemorrhage in the pediatric population to minimize radiation exposure.

Risk Factors for Poor Clinical Outcome

Multivariate analysis revealed that poor baseline functional status in children was an independent predictor of poor outcome, as suggested in selected adult AVM populations.^33,34 The location of the AVM on the left side was also an independent risk factor; the proportion of dominant left-sided hemispheres is unknown, but it is difficult to postulate causes other than cerebral dominance for the significant difference observed. Finally, high-grade AVMs were a significant predictive risk factor for poor outcome, at least in part owing to the rate of complications associated with an aggressive treatment philosophy with curative intent. Risk factor analysis showed a trend (P = .07) for a higher hemorrhage rate in high-grade lesions, which could cause worse outcome. All 6 fatal late hemorrhages occurred in children with high-grade AVMs. The relationship between size and Spetzler-Martin grade may account for worse outcomes after treatment of larger lesions, as described for a cohort of surgically treated adults.³⁵

Radiosurgery for Pediatric AVMs

In this series, as anticipated, initial radiosurgery for low-grade AVMs was more effective than for high-grade AVMs, with 5 of 12 (42%) cured compared with 0%. The reported obliteration rates for pediatric AVMs vary from 35% to 87%, a wide range that likely reflects differences in radiation dose and lesion selection.^36–39 Our low success rate even for low-grade malformations suggests that when unanticipated residual AVM is discovered on postoperative angiography, the best treatment consists of microsurgical resection if the residual is safely accessible. For inaccessible residual AVM, a course of radiosurgery may be a reasonable second choice.

In this series, initial radiosurgical treatment for high-grade AVMs was ineffective. Several of these patients were treated in the era when the appropriate dosages and indications for radiosurgery were still being determined. Nonetheless, radiosurgery for high-grade AVM can be an important adjuvant in the multimodality management of these difficult lesions. Surgical resection after radiosurgery is simplified as a result of obliteration of the small vessel feeders of the malformation by the radiation.⁴⁰ Radiosurgery also plays a role after deliberate surgical staging and occasionally can obliterate portions of the AVM that would otherwise remain inaccessible.^41–43 Concerns about the use of radiation in pediatric patients, in terms of cognitive development, effects on cerebral neurogenesis, and the small chance of radiation-related neoplasia, have led our group to reserve the use of this modality for children with inaccessible AVMs who are thought to be at high risk of hemorrhage.

CONCLUSION

Multimodality therapy can improve the success rate of AVM obliteration for both low-grade and high-grade lesions. A considerable commitment in terms of years, multiple procedures, and acceptance of the risk of complication is required to achieve obliteration of high-grade AVMs. Until the natural history of these lesions in children is known, the costs of this commitment will continue to be balanced against the risk of morbidity and mortality from late AVM hemorrhage. An aggressive management philosophy for children with certain high-grade lesions can still be supported, especially for children with recurrent hemorrhage, incapacitating symptoms, or progressive neurological deficits. Families of children with high-grade AVMs must be educated about the chance of AVM cure, rate of complications, and expected clinical outcomes to enable informed treatment decisions.

Acknowledgments

We would like to acknowledge Hiep Ho, Beth Hoyte, and Cindy Samos for contributing to the preparation of the database, figures, and manuscript.

REFERENCES

COMMENTS

Given the relatively poor long-term natural history of pediatric arteriovenous malformations (AVMs), the quest in pediatric AVM management is the quest for cure. All neurosurgeons who treat these disorders initially evaluate patients with that issue in mind. However, there are those lesions that cannot be obliterated because of either their size and/or location or the limitations of our tools for treatment. This thoughtful and well-written review of 120 pediatric patients with intracranial AVMs treated from 1985 to 2009 at a single institution gives us significant insight into that dilemma. With information accumulated from a prospectively acquired database, Darsaut et al are honest and forthright about their challenges in using multimodality therapy to manage pediatric AVMs. The concept of multimodality AVM treatment is not new but has never been applied so consistently for such a long time on so many pediatric patients. With overall obliteration rates of 87% and 26% in low-grade (Spetzler-Martin I-III) and high-grade (Spetzler-Martin IV-V) lesions, respectively, using all modalities at their disposal, the authors underscore the difficulty of the task. They describe disabling deficits caused by treatment in 5% low-grade and 28% high-grade lesions. With the authors' data, approximately one third of patients in this series (34 of 101 who underwent either initial or further treatments) were therefore left to live with incompletely treated lesions. I, for one, would like to see more detail regarding the authors' handling of residual and recurrent lesions, but in this article the message is clear: Even in the best of hands, pediatric AVM management can be challenging, frustrating, and rewarding all at the same time.

Douglas Brockmeyer

Salt Lake City, Utah

This is a large series of 120 pediatric patients with cranial arteriovenous malformations (AVMs) managed with multimodality therapy over a 24-year period. As would be expected with such a review, the methods of treatment have evolved over the course of the study, including various methods of focused radiation and embolic agents. As has been reported previously, patients with lower-grade AVMs (Spetzler-Martin grade I-III) fared better with higher obliteration rates and lower complications. Poor clinical outcome was also related to clinical status at presentation and a left-sided location. Their obliteration rate with up-front radiosurgery was low (20%) but may have been a function of case selection and AVM size. As also recognized in pediatric AVMs, there were several recurrent AVMs. The authors recommend intraoperative or immediately postoperative angiography followed by repeat angiography at 3 to 6 months, which is routinely practiced. However, 1 AVM recurred following this type of surveillance. The authors' recommendation that patients should have repeat postoperative angiography after apparent obliteration to identify recurrences every 5 years until the age of 21 needs to be supported by more risk/benefit analysis because the risk of this type of protocol is not insignificant.

James M. Drake

Toronto, Ontario, Canada

This well-written article summarizes the experience of a high-volume center with the treatment of arteriovenous malformations (AVMs), adding substantial data to the growing pool of evidence supporting the use of coordinated multidisciplinary care for patients with these lesions. The authors are candid with respect to their complications, emphasizing that risk exists even in the best of hands. This article should be of interest to those clinicians who specialize in the treatment of AVMs but should also serve as a general call to all neurosurgeons to refer these complex cases to centers suited to offer the kind of multidisciplinary care described by the authors to provide the best care to affected children.

Edward R. Smith

Boston, Massachusetts

ABBREVIATIONS:

• AVM

  arteriovenous malformation

• mRS

  modified Rankin Scale score