Article Text

Original research
Pediatric brain arteriovenous malformation recurrence: a cohort study, systematic review and meta-analysis
  1. Jean-Francois Hak1,2,
  2. Gregoire Boulouis1,2,
  3. Basile Kerleroux1,2,
  4. Sandro Benichi3,
  5. Sarah Stricker3,
  6. Florent Gariel1,4,
  7. Lorenzo Garzelli1,
  8. Philippe Meyer5,
  9. Manoelle Kossorotoff6,7,
  10. Nathalie Boddaert1,
  11. Vincent Vidal8,
  12. Nadine Girard9,
  13. Volodia Dangouloff-Ros1,
  14. Francis Brunelle1,
  15. Heather Fullerton10,11,
  16. Steven W Hetts12,
  17. Thomas Blauwblomme3,7,
  18. Olivier Naggara1,2,7
  1. 1 Department of Pediatric Radiology UMR 1163, Institut Imagine, INSERM U1000, APHP, Necker Sick Children Hospital, Paris, Paris, France
  2. 2 Department of Neuroradiology, INSERM UMR 1266 IMA-BRAIN, GHU Paris, Paris, France
  3. 3 Department of Pediatric Neurosurgery, Institut Imagine, INSERM UMR 1163, APHP, Necker Sick Children Hospital, Paris, France
  4. 4 Department of Neuroradiology, CHU Bordeaux GH Pellegrin, Bordeaux, France
  5. 5 Department of Pediatric Neuro ICU, APHP, Necker Sick Children Hospital, Paris, France
  6. 6 Department of Pediatric Neurology, APHP University Necker Children Hospital, Paris, France
  7. 7 French Center for Pediatric Stroke, INSERM U894, APHP, Necker Sick Children Hospital, Paris, France
  8. 8 Department of Radiology, University Hospital La Timone, AP-HM, Marseille, France
  9. 9 Department of Neuroradiology, University Hospital La Timone, AP-HM, Marseille, France
  10. 10 Department of Neurology, University of California San Francisco, San Francisco, California, USA
  11. 11 Department of Pediatrics, University of California San Francisco, San Francisco, California, USA
  12. 12 Department of Radiology and Biomedical Imaging, University of California San Francisco, San Francisco, California, USA
  1. Correspondence to Dr Jean-Francois Hak, Department of Radiology, Necker-Enfants Malades Hospitals, Paris, Île-de-France, France; jeanfrancois.hak{at}


Background Recurrence following obliteration of brain arteriovenous malformations (AVMs) is common in children surgically treated, but recurrences following endovascular (EVT) and radiosurgical approaches are scantily reported.

Objective To analyze the rates and risk factors for AVM recurrence after obliteration in a single-center cohort of children with ruptured AVMs treated with multimodal approaches, and to carry out a comprehensive review and meta-analysis of current data.

Methods Children with ruptured AVMs between 2000 and 2019 enrolled in a prospective registry were retrospectively screened and included after angiographically determined obliteration to differentiate children with/without recurrence. A complementary systematic review and meta-analysis of studies investigating AVM recurrence in children between 2000 and 2020 was aggregated to explore the overall recurrence rates across treatment modalities by analyzing surgery versus other treatments.

Results Seventy children with obliterated AVMs were included. AVM recurrences (n=10) were more commonly treated with EVT as final treatment (60% in the recurrence vs 13.3% in the no-recurrence group, p=0.018). Infratentorial locations were associated with earlier and more frequent recurrences (adjusted relative risk=4.62, 95% CI 1.08 to 19.04; p=0.04).

In the aggregate analysis, the pooled rate of AVM recurrence was 10.9% (95% CI 8.7% to 13.5%). Younger age at presentation was associated with more frequent recurrences (RR per year increase, 0.97, 95% CI 0.93 to 0.99; p=0.046).

Conclusion Location of infratentorial AVMs and younger age at presentation may be associated with earlier and more frequent recurrences. The higher rates of recurrence in patients with AVMs obliterated with EVT questions its role in an intent-to-cure approach and reinforces its position as an adjunct to surgery and/or radiosurgery.

  • arteriovenous malformation
  • brain
  • pediatrics
  • stroke
  • hemorrhage

Data availability statement

Data are available upon reasonable request.

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


Brain pediatric arteriovenous malformations (AVMs) are the most common underlying etiology of pediatric intracerebral hemorrhage, and hemorrhage is the most frequent mode of AVMs in this age group.1 2 After a rupture, prevention of hemorrhage recurrence can be achieved only by complete angiographically determined AVM obliteration.3 An AVM cure can be obtained through microsurgery, stereotactic radiosurgery (SRS), endovascular treatment (EVT), or a combination of these approaches.

AVM recurrence, is an increasingly recognized evolution of angiographically determined, cured AVMs,4 5 especially in the pediatric population.5–7 A systematic review recognized that children with angiographically determined cured AVMs were subject to re-ruptures after late or unrecognized AVM recurrences due to lack of follow-up.7 Yet, most previous studies have focused on the risk of recurrence in unselected age groups of patients with AVMs,4 or in children cured only through microsurgery. The roles of SRS and EVT are central in the care of children with AVM, but the risk of recurrence after either of these approaches or their combination is seldom reported. Precise estimates of AVM recurrence after surgery, SRS, EVT, or multimodal treatment remain to be assessed, and may help guide imaging follow-up after treatment,8 9 provide further insight into the risk factors underlying this phenomenon, and modify communication and expectation management with families and caregivers after angiographically determined cure.

We conducted a single-center retrospective cohort study of children with ruptured AVMs shown angiographically to be healed, and performed a comprehensive review of data reporting recurrent cerebral pediatric AVM,1 aiming to analyze the rates of, and risk factors for, AVM recurrence in children.

Patients and methods


The manuscript was prepared following the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines and the Preferred Reporting Items for Systematic-Reviews and Meta-Analyses (PRISMA) guidelines.10 11 For the cohort study, as this was a non-interventional retrospective study of routinely acquired data, written informed consent for this study was waived. A commitment to compliance (MR04#3618210420 and CNIL #2217698) was filed to the French National Information Science and Liberties Commission. Patients were informed they could oppose the use of their health-related data for research purposes. The study is registered in the national health data institute repository at Anonymized data will be made available on reasonable request to the corresponding author by a qualified investigator, after institutional review board approval.

Patients’ selection and cohort study

Consecutive patients referred to our institution, a pediatric quaternary care center and coordinator for the French Center for Pediatric Stroke, were prospectively enrolled in an inception registry initiated in 2008 and retrospectively inclusive to 2000. The registry has been described elsewhere in detail.12 All patients meeting the following criteria were screened for inclusion: (1) pediatric, non-neonatal, patients defined as aged 28 days to 18 years at the time of symptoms onset; (2) with ruptured AVM; (3) between October 2019 and January 2000; (4) treated unimodally or multimodally; (5) leading to a complete exclusion of the AVM, as demonstrated by a dedicated digital subtraction angiography (DSA). Patients were excluded for the following reasons: (1) AVM without complete exclusion or (2) with complete exclusion but not proved with DSA.

Included children were then divided into two groups, those with or without subsequent angiographically determined AVM recurrence.

Literature search, study selection, and eligibility criteria

The systematic review and meta-analysis was conducted with reference to the PRISMA11 and MOOSE (Meta-analysis Of Observational Studies in Epidemiology) guidelines.13 A comprehensive literature search of PubMed Central was designed and conducted separately by two authors (J-FH and GB). The key words ‘recurrent’, ‘recurrence’, ‘reappearance’, ‘recur*’, ‘arteriovenous malformations’, were used in combination (using the Boolean operators OR and AND). The search retained all articles published on this topic from January 1990 to October 2020. We reviewed bibliographies for citations until no further citations were found. The references of the publications were checked again for additional studies. Only references reporting AVM were included, and therefore all references reporting vascular shunt as vein of Galen aneurysmal malformation, cerebral proliferative angiopathy, hereditary hemorrhagic telangiectasia, and RASA1 associated AVMs/fistulas were excluded. This method of cross-checking was continued until no further publications were found.

All studies with longitudinal follow-up of treated pediatric AVM reporting recurrent AVM were included if they met the following criteria: (1) series or reports of ≥1 patient aged 28 days to 18 years; (2) with available data for analysis of AVM recurrence according to the definition: obliterated pediatric AVM, proved with DSA, later developing a DSA-proved recurrence.

We excluded studies in a language other than English, editorials, technical notes, review articles, and conference abstracts. Two reviewers selected the included studies (J-FH, GB).

For each study, quality based on the Newcastle-Ottawa Scale,14 and the risk of bias according to the Cochran Collaboration tool were assessed, independently by two authors (J-FH, GB).15 Inconsistencies were resolved by consensus.

From included studies we extracted the following information about risk factors for recurrent pediatric AVM: overall proportion of recurrent AVM, age, ethnicity, sex, AVM location (deep/superficial, supratentorial/infratentorial), AVM characteristics (Spetzler-Martin grade (SM), presence of deep venous drainage), initial presentation of the AVM, treatment modality that obtained obliteration and clinical manifestation of the recurrence, mean overall follow-up, or patient-specific follow-up, when available.

Statistical analyses

In the cohort study, we analyzed the differences between children with or without recurrence, using univariable analyses with the appropriate test. Data are presented as absolute numbers (percentage), mean (SD), or median (IQR).

For the pooled analysis, data from our cohort were aggregated in a multicategory meta-analysis through the systematic review using the comprehensive meta-analysis version 3 (CMA 3, Borenstein M, Englewood, New Jersey, USA 2013) software. The extent of heterogeneity in multicategory distribution estimates was quantified with the I 2 statistic and visually through inspection of the Forest plot. For high levels of heterogeneity (I 2 >50%), random effect models were used. We explored the recurrence rates across treatment modalities by analyzing surgical excision versus other treatments to obtain obliteration. The significance level was set at 0.05 for the cohort study analyses, and at 0.01 for the aggregate estimates.


Cohort study

Among 127 patients with ruptured AVM in the registry, at last follow-up 3 had died, 7 were lost to follow-up, and 47 were still receiving treatment. A total of 70 patients (mean age 9.9 years; 57% boys) with angiographically confirmed obliterated DSA were included (figure 1). Among them, 10 children (14.3%) developed an angiographically determined recurrence after a mean follow-up of 36.6 (SD 37,3) months. There was no significant difference in clinical, intracerebral hemorrhage or AVM-related characteristics in patients who later developed AVM recurrence versus those who did not, using univariable analyses (table 1).

Figure 1

Flowchart of patients inclusions. AVM, arteriovenous malformation; DSA, digital subtraction angiography; SRS, stereotactic radiosurgery.

Table 1

Baseline characteristics of included patients stratified by recurrence status

Seventy pediatric AVMs were shown, by DSA, to be obliterated. Among them, 46 with unimodal treatment (27 surgery, 12 EVT, 7 SRS), 23 with multimodal treatment (13 EVT + surgery, 1 surgery + SRS, 6 EVT + SRS, 2 EVT + surgery + EVT, 1 EVT + surgery + SRS), and 1 patient without any treatment had a ruptured AVM that spontaneously completely obliterated during follow-up at 3 months and onwards (6 months, 1 year, 3 years, and 5 year).

Children with AVM recurrences were more commonly treated with EVT as final treatment (60% in the recurrence vs 13.3% in the no recurrence group) and less so with microsurgery (40% vs 60%) or SRS (0% vs 25%); p=0.018). Overall, this translated to recurrence rates following EVT, microsurgery, and SRS of 43%, 10%, and 0%, respectively.

Using Cox proportional hazard models, we found infratentorial location to be associated with earlier and more frequent recurrences in the study sample (adjusted rate ratio=4.62, 95% CI 1.08 to 19.04; p=0.04). No association was found between event age, sex, or SM grade and the risk of recurrence (all p>0.10) in our sample (online supplemental table I).

Literature review and study selection

Among 1574 studies screened for inclusion between January 1990 and October 2020, a total of 29 records with 1083 children were included in our review. See flowchart of studies inclusions (figure 2). Characteristics of included studies are shown in table 2. None of the studies were prospective. Twenty-two studies (encompassing 70 patients, 80.5% patients) were retrospective series,5 16–34 and nine (encompassing 17 patients, 19.5% patients) were case reports.6 35–42

Figure 2

Flowchart of studies inclusions. AVM, arteriovenous malformation

Table 2

Summary of included studies

None of the studies were rated as high-quality based on the Newcastle-Ottawa Scale.

Risk of bias was assessed as intermediary in 7,5 17 28–31 34 and high in 1516 18–27 32 33 35 36 studies according to the Cochrane Collaboration tool (details in online supplemental table II). As clinical cases seven reports6 37–42 were not assessed.

In the included studies,5 16 20 22–25 27–31 33 34 36 the pooled rate of AVM recurrence was 10.9% (95% CI 8.7% to 13.5%), p value for heterogeneity 0.00, I2=65.90, Q value 41.054. When analyzing studies according to AVM obliteration modality (only surgery, or multiple treatment: surgery/EVT/SRS), patients treated with microsurgical excision had significantly lower recurrence rates 7.0% (95% CI 5.0% to 9.8%), p value for heterogeneity 0.09, I 2 =40.12, Q value 15.03 vs 15.9% (95% CI 11.8% to 21.2%), p value for heterogeneity 0.01, Q value: 13.19, I 2 =69.67 (p value for heterogeneity across study groups 0.008) (figure 3).

Figure 3

Recurrence rates in included studies.

Individual patient data were available for 69/82 recurrent pediatric AVMs (online supplemental table III) in 27 studies.6 16–32 35–43 In these studies, the median age of children experiencing AVM recurrence was 8 years (1–17), with 45.8% boys. Most of the pediatric AVM (n=46; 85.2%) showed hemorrhage, seven pediatric AVM (13.0%) showed symptomatic manifestation but without hemorrhage, and one (1.9%) was asymptomatic (data available for 54 patients).

SM grade was I–II for 40.0% (18/45) of children and was II–V for 60.0% (27/45). Any deep venous drainage was present in 24 patients (50.0%) (data available for 48 patients). Pediatric AVMs were in an eloquent area for 34 patients (64.2%) (data available for 53 patients). Mean nidus diameter was 3.14±1.68 cm (data available for 14 patients). Complete AVM obliteration was obtained with initial surgical management for 51 patients (81.0%), with EVT associated with surgery for five patients (7.9%), with EVT associated with SRS for one patient (1.6%) and with SRS alone for four patients (6.3%) (data available for 63 patients).

The recurrent pediatric AVM was hemorrhagic for 17/51 patients (33.3%)

After angiographically confirmed complete pediatric AVM obliteration, the mean time to AVM recurrence was 52.27±52.65 months (data available for 64 patients). The major risk of bias, as well as heterogeneity of available data precluded conduction of a meta-regression.

In exploratory analyses, we ran a proportional hazards Cox model in the pooled individual patients data population, to identify factors associated with recurrences. After adjustment for age, sex, SM grade, infratentorial versus supratentorial location, and treatment to obtain obliteration, we found that younger age at presentation was associated with more frequent recurrences (RR per year increase=0.97, 95% CI 0.93 to 0.99; p=0.046).


In this large retrospective series of children with hemorrhagic AVMs shown by angiography to be cured, the rate of recurrence was around 15%, increasing to 43% if the treatment to obtain AVM obliteration was embolization. The recurrence rate in the aggregate analysis was of similar magnitude with 10.9% (95% CI 8.7% to 13.5%) of children presenting with a recurrence, with a similar higher recurrence rate if treatment to obtain obliteration was not surgery.

Our results come as confirmation of previous studies of the relatively high recurrence rates of pediatric brain AVMs after complete obliteration.23 31 36 The duration and modality of follow-up is likely to influence the rate of AVM reappearance. Lang et al 23 presented, for instance, a cohort in which patients who underwent postoperative DSA at 1 year demonstrated a recurrence rate of zero, while four recurrent pediatric AVMs were found within the following 5 years. Morgenstern et al 31 noted that long-term follow-up in their series identified one patient with a recurrence 6 years after complete resection with a previously negative follow-up DSA at 4 years. Conversely, Tagaki et al 36 reported three early recurrent pediatric AVMs at 3 months. The recurrent AVM was hemorrhagic for 17/51 patients (33.3%), further highlighting the key importance of imaging follow-up to avoid re-hemorrhage for undetected recurrence. Together, these data strongly reinforce the need for prolonged follow-up protocols after angiographically determined, cure of a pediatric AVM. At our institution when obliteration, shown by angiography, is obtained, the patient is subsequently followed-up by MRI until the age of 18 years and/or after a minimum of 5 years, whichever comes last, by repeat DSA.

Sorenson et al 4 reported, in a systematic review (including adults and children) that assessed risk factors for the recurrence of surgically resected AVMs, that hemorrhage was the most common initial manifestation of recurrent AVMs. This finding, substantiated by recent data in a large cohort of children treated with surgery,5 suggests this might result from (1) a false-negative DSA due to acute hematoma mass effect masking portions of the AVM, or alternatively, a sectorized nidus after being split by the hemorrhage making the nidus resection more difficult. Recently, Copelan et al 5 retrospectively studied 115 patients (aged <25 years) initially presenting with brain AVM completely treated with microsurgery. They conclude that intracranial hemorrhage was a significant predictor of recurrence (log rank p=0.037) with a 5-year recurrence rate of 17.8% (95% CI 8.3% to 35.7%), and all the recurrences occurred in patients who were children.

Unfortunately, owing to the lack of available data, we could not comparatively assess pooled recurrence rates across hemorrhagic versus non-hemorrhagic presentations.

The higher rate of recurrence found in the cohort study of children last treated with EVT questions the position of EVT in the curative treatment of ruptured pediatric AVMs. A recent report analyzed the characteristics and complications of embolization of AVMs with intent to cure.44 Interestingly, the authors showed that the complication rates were lower when the treatment goal was adjunctive, and concluded that complete obliteration should be seen as an unexpected benefit rather than a strategic goal in unselected patients. In low SM grade adult AVMs, with lesions typically considered for EVT curative embolization, Wang et al further demonstrated that EVT had a key role as an adjunct therapy,45 in patients subsequently cured with microsurgery or SRS. Together, our data and previous studies agree that EVT has a key adjunctive role in the treatment strategy of AVMs, but should not expose the patient to disproportionate risks of complications to obtain a complete obliteration. This is especially true in children where the recurrence rate following cure through embolization appeared to be significantly higher than with SRS or surgery.

The underlying cause of the lower recurrence reported rate in adults (around 3%)35 and recurrent AVMs in children is unclear. Some studies evaluated the theory of angiogenesis dysregulation by various growth factors, such as vascular endothelial growth factor (VEGF)35 or endothelial progenitor markers (CD31, CD34, CD105).36 In contrast Sonstein et al 35 also found VEGF-positive staining in the population with no recurrence. Further work is needed to better understand the phenomenon.

The important risk of bias and available data did not allow us to perform meta-regressions to identify clinical imaging factors related to higher risk of AVM recurrence in the pediatric population. Yet, the presented data are of key interest for physicians caring for children with cerebrovascular disorders. Indeed, most of the existing systematic reviews mixed adult and pediatric populations, preventing age-specific conclusions. Furthermore, older studies reported data for children treated mostly with exclusive nidal microsurgery, a strategy that is now obsolete with the advent of modern embolization material and techniques. Our data come as a complement, in a pediatric population with multimodal management.

Of note, in addition to younger patient age, deep venous drainage and diffuse nidus have been identified as potential risk factors for recurrence in previous works,17 23 24 a trend that was not present in our data, or from analysis of crude numbers of the aggregate patient level.

Our study has several limitations. It was not possible to make formal aggregated statistical comparisons because of the small number of patients and studies included. Approximately one-third of included studies were case reports, which introduces important bias. Furthermore, the sample in the cohort study included only ruptured pediatric AVMs, and therefore the findings cannot be reliably extrapolated to unruptured pediatric AVMs It is also important to note that angiographic disappearance is common in patients with non-distal obliteration during EVT, even in the absence of true nidal and venous obliteration. Given that the data acquisition pattern was retrospective, and despite a protocol for angiographic follow-up, it might be that some children did not experience true recurrence, but a delayed re-feeding of an incompletely treated nidus using EVT. Finally, we acknowledge that the inclusion period spans nearly two decades, and that endovascular techniques have dramatically evolved over the study period, which might introduce some bias that could have disadvantaged EVT over microsurgical or radiosurgical techniques approaches, with possibly different results if the study were to be repeated with children treated only using most advanced EVT techniques.


After angiographically determined obliteration, AVM recurrences are frequent in the pediatric population, and predictive factors are poorly understood. The higher rates of recurrence in patients cured with EVT reinforces the role of embolization as an adjunct to surgery or radiosurgery. Long-term protocolized imaging follow-up as well as information for patients and families of the risks of recurrence are important after AVM cure, as shown by angiography, in pediatric patients.

Data availability statement

Data are available upon reasonable request.

Ethics statements

Patient consent for publication


Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.


  • Twitter @gboulouis, @gboulouis

  • J-FH and GB contributed equally.

  • Contributors J-FH, ON, GB, TB were responsible for the conception of the study; J-FH, ON, GB, TB, BK, SB, SS, FG, LG, PM, MK, NB collected the data, had full access to data, and take responsibility for accuracy of data analysis. J-FH, GB, ON drafted the initial version of the manuscript. All authors, contributed to data acquisition, analysis and interpretation, revised and approved the final version of this manuscript.

  • Funding J-FH was supported by a grant provided by the Société Française de Radiologie - French Society of Radiology - (SFR) together with the Collège des Enseignants en Radiologie de France - French Academic College of Radiology (CERF)

  • Competing interests None declared.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.

Linked Articles