Development of Collateral Vessels after Anterior Circulation Large Vessel Occlusion in Pediatric Arterial Ischemic Stroke Relates to Stroke Etiology: A Longitudinal Study

DISCUSSION

This study addressing anterior circulation LVO in pediatric AIS (excluding neonates) provides novel data concerning a specific radiologic evolution pattern of intracranial LVO in children relating to stroke etiology.

In our series, intracranial LVO was observed in 24.6% of children with acute anterior circulation AIS. Though our study focused only on anterior circulation LVO, these findings are in line with the reported prevalence of LVO in childhood stroke of 23.5% in a retrospective population-based cohort study by Bhatia et al² and 22.4% in a single-center retrospective study by Bonnet et al.¹

Concerning angiographic outcomes, we described a specific pattern of U-PS anastomotic bridge, characterized by the development in the first year after stroke of a collateral network forming a bridge in bypass of the residual stenosis or occlusion, with visible distal flow in the still or formerly occluded main arterial trunk, in patients without visible collaterals at the acute phase. This pattern is specific as it does not meet the criteria for other described approaching vascular patterns (Fig 4).

• Download figure
• Open in new tab
• Download powerpoint

FIG 4.

Distinctive features of U-PS anastomotic bridge compared with vascular patterns with close appearance. (i)-ICA = (intracranial)-internal carotid artery; ECA = external carotid artery; VA = vertebral artery; BA = basilar artery. Illustration is from Lin et al²² for rete mirabile. Drawings by F.B. and M.K., with courtesy.

First, as collaterals were not present at stroke onset, this precludes congenital variations or anomaly, such as fenestration, duplication, aplastic or twig-like artery aspects. In these congenital variations, the regression of anastomoses from the embryonic primitive mesh of intracranial arterial networks did not occur, leading to the persistence of remnants of embryologic development.^13⇓⇓-16 They are sometimes reported as a potential risk factor for ischemic or hemorrhagic stroke.^17,18 Of note, no such variation was observed in our patients. Second, as U-PS anastomotic bridge developed after LVO, a specific intracranial arterial compensatory mechanism of downstream chronic ischemia can be hypothesized. Indeed, a similar compensatory mechanism has been described in cervical or systemic arteries, occurring during the embryonic phase, or acquired hypoxic conditions. Cervical carotid rete mirabile is an embryonic compensatory phenomenon, visible as a meshwork of multiple, freely intercommunicating arterioles fed by external carotid artery branches, which reconstitute the absent or hypoplastic segments of the internal carotid artery. It may be bilateral and associated with other vascular aspects (aorta malformation, posterior circulation rete mirabile) or a general condition (pseudoxanthoma elasticum).^19⇓⇓-22 It may represent a risk factor for ischemic or hemorrhagic cerebrovascular diseases.²³ Postocclusive neovascularization has been described in systemic arteries, in which vasa vasorum, adventitial vessels form plexus in the wall of large blood vessels, with a mainly nutritive role; development would be stimulated by local subacute hypoxic conditions (atherosclerosis, diabetes, vasculitis, etc), figuring a phenomenon of compensatory neovascularization. This has been notably reported in coronary arterys, aorta, and cervical internal carotid artery occlusion.^24⇓-26 To our knowledge, these mechanisms have not been reported affecting the circle of Willis, except in a handful of case reports.²⁷ The existence of intracranial vasa vasorum has been debated, and they are suggested to be present only in the proximal parts of MCA, ACA, and intracranial ICA.^28⇓-30 U-PS anastomotic bridge could thus represent a similar mechanism of postocclusive neovascularization, with the development of intracranial vasa vasorum located in the proximal segments of the circle of Willis arteries.

Interestingly, not every patient with LVO developed U-PS anastomotic bridge. No association with age at stroke onset or duration of vessel occlusion was found. U-PS anastomotic bridge development was strongly associated with stroke etiology: this pattern was only observed in patients with unilateral FCA, and 66% of children with FCA and LVO developed U-PS anastomotic bridge. This suggests that, rather than a nonspecific mode of vascular healing after local ischemic changes related to LVO, it could be related to subacute local conditions and modifications associated with FCA, for instance, subacute hypoxia and/or vessel wall inflammation, stimulating angiogenic factors.^30,31

Finally, it seems important to differentiate this vascular aspect from Moyamoya Angiopathy MMA. MMA is a progressive steno-occlusive disease involving the distal internal cerebral artery and its bifurcation, and the adjacent proximal ACA and MCA. It often shows bilateral, symmetric, or asymmetric segmental stenoses of the involved arteries. Stenoses are associated with several types of developing compensatory vessels: basal collateral vessels, leptomeningeal collaterals, and anastomotic internal-external carotid systems collaterals. Basal collaterals are abnormally dilated lenticulostriate and thalamo-perforating arteries, arising from the MCA to the basal ganglia and thalamus.^12,27,32 Key points differentiate MMA and U-PS anastomotic bridge: 1) in MMA dilated perforating arteries supplying the basal ganglia arise perpendicularly from the MCA trunk, whereas in U-PS anastomotic bridge collaterals go parallel with the MCA main trunk (the difference is well observed on coronal view) (Figs 3 and 4); and 2) MMA is classically bilateral, sometimes asymmetrical but rarely unilateral, whereas U-PS anastomotic bridge is unilateral without contralateral arterial anomaly (stenosis/occlusion). The latter point is of utmost importance as both an aspect of anastomotic bridge and abnormal perforating lenticulostriate (Moyamoya network) may be observed in patients with Moyamoya. It seems important to identify isolated U-PS anastomotic bridge because, though these patients have developing collaterals, they did not experience any stroke recurrence nor vascular disease progression, contrary to patients with Moyamoya angiopathy. In our study, recurrent strokes only occurred in non-FCA patients. This emphasizes the fact that patients with FCA usually do not experience recurrent stroke regardless of whether they had collaterals or not. Thus, the observation of a U-PS anastomotic bridge after stroke with an isolated strictly unilateral LVO, and no contralateral vascular anomaly should not be considered as a risk for poor outcome or stroke recurrence.

Our study limitations mainly relate to the small sample and to its retrospective nature. Different angiographic imaging modalities may also induce biases: TOF-MRA may over-call occlusion in a stenosis with minimal flow, and 3T MRA may have a better resolution than 1.5T MRA for the anastomotic bridge observation. However, the homogeneous technique in our study may limit the bias for intrasample angiographic comparisons: 25/26 patients had TOF-MRA in the acute phase and during follow-up, mostly on 1.5T scan. Only 1 patient had initial MRA and CTA at 48 h poststroke, but he had a persistent right proximal M2 segment occlusion, which precludes a bias of patency related to imaging technique.