Embolization of cerebral arteriovenous shunts in infants weighing less than 5 kg

Restricted access


Endovascular treatment in children, especially neonates, can be more challenging than analogous procedures in adults. This study aimed to describe the clinical and radiological findings, type and timing of endovascular treatment, and early outcomes in children who present with neurovascular malformations, who are treated with embolization, and who weigh less than 5 kg.


The authors carried out a retrospective review of all consecutively treated children weighing less than 5 kg with neurovascular arteriovenous malformations (AVMs) at a single institution over a 10-year period.


Fifty-two patients were included in the study. Thirty-eight had a vein of Galen aneurysmal malformation, 3 a pial AVM, 6 a pial arteriovenous fistula, and 5 a dural sinus malformation. The endovascular treatment goals were control of cardiac failure or hydrocephalus in cases of nonhemorrhagic malformations or to prevent new bleeding in cases of previous hemorrhage. A hemorrhagic complication occurred in 12 procedures and an ischemic complication in 2. Both complication types were correlated with the age of the infant (age cutoff at 3 months) (p = of 0.015 and 0.049, respectively). No correlation was found with the weight of the infant or the duration of the procedure.


The embolization of AVMs in these patients prevented adverse cardiac effects, hydrovenous disorders, and rebleeding. The risk of major cerebral complications seems mainly correlated with age, with a threshold at 3 months. A multidisciplinary team involved in the treatment of these children may help to improve treatment success and management.

ABBREVIATIONS AVF = arteriovenous fistula; AVM = arteriovenous malformation; DSM = dural sinus malformation; ICA = internal carotid artery; KOSCHI = King’s Outcome Scale for Childhood Head Injury; VGAM = vein of Galen aneurysmal malformation.

Article Information

Correspondence Guillaume Saliou: Centre Hospitalier Universitaire Vaudois, Lausanne, Switzerland. guillaume.saliou@chuv.ch.

INCLUDE WHEN CITING Published online February 22, 2019; DOI: 10.3171/2018.11.PEDS1865.

Disclosures The authors report no conflict of interest concerning the materials or methods used in this study or the findings specified in this paper.

© AANS, except where prohibited by US copyright law.



  • View in gallery

    Images obtained in a 2.000-g newborn with a VGAM causing high-flow cardiac failure and pulmonary suprasystemic hypertension and leading to myocardial infarction on day 2. A–C: MR image and conventional angiograms on day 2 showed a choroidal type with multiple large shunts (A and C, arrows: dilated posteromedial and posterolateral choroidal branches that drained through a dilated embryonic precursor vein of Galen [A and B, empty arrow]). D–H: The aim of the first embolization session was to stabilize the cardiac failure, and this was achieved by using pure histoacryl glue embolization of the 4 largest shunts (D–G: selective cast of glue for each injection; H: final cast of glue). To save contrast medium and avoid contrast toxicity, no final angiogram was acquired at the end of this first embolization session. Following this first embolization, the patient dramatically improved and was discharged on day 6 from the ICU. Clinical watchful waiting showed normal psychomotor development, and repeated MRI performed at 3 and 9 months also showed good brain development with good vascular remodeling. I: Axial T2-weighted MR image at 9 months showing the VGAM almost occluded. A final embolization session will be scheduled at 3 years. AP = anteroposterior; Lat = lateral.

  • View in gallery

    Images obtained in a 3.820-g newborn with a right frontal pial fistula causing high-flow cardiac failure. A: On day 6, the right ICA angiogram demonstrated high-flow fistulas that drained through a single dilated venous collector (empty arrow) and some degree of vascular steal (star: a delay in filling of the normal pial network of the right hemisphere). A dilated pial network around the fistula attributed to high-flow aspiration was also noted (arrowheads) associated with a dilated superior longitudinal sinus (double arrows). To avoid a rupture of this weak vasculature, especially the dilated pial network, staged embolization was performed. B and C: The aim of the first session was to stabilize the cardiac failure, and this was achieved by loose-coiling embolization in one of the largest shunts to reduce the flow (B, empty arrow: coils in the foot of the draining vein and arterial feeder) followed by high concentrated glue embolization in the remaining biggest shunt (arrow: selective microcatheterization; and C, double arrows: cast of glue). D: On the final angiogram, improvement of contrast filling of the normal pial network (star). Clinically watchful waiting showed normal psychomotor development and repeated MRI performed at 3 and 6 months also showed good brain development. E and F: A second embolization was performed at 7 months with glue and Onyx (E: right ICA angiogram before embolization), which managed to progressively decrease the dilated pial network in the vicinity of the fistula, as demonstrated on the embolization session performed at 1 year (F: disappearance of the dilated pial network). G and H: Therefore, the fistula was completely occluded by a third embolization session with Onyx injection (G: cast of Onyx; H: right ICA angiogram showing the complete cure of the fistula).

  • View in gallery

    Weight versus age for postoperative hemorrhagic complications of embolization in infants.

  • View in gallery

    Weight versus age for postoperative ischemic complications of embolization in infants.

  • View in gallery

    Kaplan-Meier curves showing long-term mortality rate VGAMs (VGA) and pial AVMs and fistulas. F-MAV indicates pial fistulas and pial AVMs. DSMs are not presented since none of these patients died during the follow-up.





All Time Past Year Past 30 Days
Abstract Views 71 71 71
Full Text Views 18 18 18
PDF Downloads 26 26 26
EPUB Downloads 0 0 0


Google Scholar