Typical subdural contrast effusion secondary to endovascular treatment of a pediatric pial arteriovenous fistula

Abstract

Subdural contrast effusion secondary to endovascular treatment is exceptionally rare and might be mistaken as subdural hematoma because of similar hyperattenuation on computer tomography. The authors present the case of a 13-month-old girl with a history of increased head circumference and developmental retardation. Cerebral digital subtraction angiography showed a high-flow pial arteriovenous fistula fed by multiple arteries on the right cerebellar surface, with occlusion of the right sigmoid sinus and severe stenosis of the left sigmoid sinus. Staged endovascular treatments were performed to eliminate the fistula. Follow-up head computer tomography scans performed 3 h after both procedures demonstrated typical high-density subdural effusion with computer tomography attenuation value similar to hemorrhage. These effusions did not aggravate the condition and disappeared spontaneously 32 h after the first treatment and 29 h after the second, respectively.

Introduction

Contrast extravasation in the brain tissue is common after endovascular therapy using iodinated contrast medium, but subdural contrast effusion (SDCE) is quite rare. SDCE demonstrates hyperattenuation on unenhanced head computer tomography (CT) which is usually performed after treatment because of its convenience and efficiency. With similar hyperattenuation to hemorrhage, SDCE might be mistaken as subdural hematoma (SDH), which could lead to unnecessary delay of the very important antithrombotic treatment after neurological endovascular therapy and even surgical treatment. To date, there have been only six case reports relating to SDCE secondary to endovascular procedure in the literature. All of these cases are adults, and some of them are atypical.^1–6 Here, we present a pediatric case of SDCE secondary to endovascular treatment of a pial arteriovenous fistula (pAVF) with typical clinical and imaging features.

Case presentation

A 13-month-old girl was transferred to our hospital with one-month history of increased head circumference and developmental retardation. She was born naturally at full term and had no family history of congenital disorders. Physical examination revealed slow response, head circumference of 50 cm, scalp veins dilatation, bulging anterior fontanelle, exophthalmos, language retardation, and motor developmental delay.

Magnetic resonance (MR) imaging showed many disordered and tortuous flow voids accompanying aneurysmal enlargement in the posterior cranial fossa and extensive and obvious enhancement of the dura mater with gadolinium (Figure 1(a) to (c)). Head CT revealed hydrocephalus and enlargement of the subarachnoid space and cerebral sulci (Figure 2(a) and (b)).

Figure 1.

Axial (a) and sagittal (b) T2 MR imaging showing many disordered, dilated and tortuous flow voids in the posterior cranial fossa; (c) enhanced MR imaging showing enhancement of the dura mater. Lateral projections of right internal carotid artery (d) and right vertebral artery (e) showing subtentorial high-flow pAVF supplied by multiple branches of the right superior cerebellar artery and posterior inferior cerebellar artery, and draining into a single cerebellar cortical vein, which flows into the confluence of sinus and retrogrades to the straight sinus and superior sagittal sinus; (f) posterior–anterior projection in the late arterial phase showing the flow continues to retrograde to the superficial cortical veins and deep veins, along with occlusion of the right sigmoid sinus and obvious stenosis of the left sigmoid sinus.

Figure 2.

Preoperative head CT (a and b) showing hydrocephalus and enlargement of subarachnoid space and cerebral sulci. Follow-up head CT 3 h after the first treatment (c and d) and the second (g and h) showing emerging hyperdense effusion in bilateral subdural space. Head CT 32 h after the first procedure (e and f) and 29 h after the second (i and j) showing the high-density effusion resolved spontaneously.

Cerebral digital subtraction angiography (DSA) under general anesthesia showed a high-flow pAVF fed by multiple branches of the right posterior inferior cerebellar artery and superior cerebellar artery and draining into a tortuous and dilated cerebellar cortical vein on the right cerebellar surface, with occlusion of the right sigmoid sinus and severe stenosis of the left sigmoid sinus. The drainage flow converged into torcular herophili and retrograded to superficial and deep cerebral veins via superior sagittal sinus and straight sinus (Figure 1(d) to (f)).

First, endovascular treatment was performed to occlude two obviously dilated feeding arteries adjacent to the fistula using coils and onyx-18 (6% EVOH, ev3 Neurovascular, Irvine, CA, USA). The procedure took 3 h, using 50 ml iopromide (300 mg of iodine/ml; Ultravist 300; Bayer Schering Pharma, Berlin, Germany) in total. One hour after operation, she recovered well from anesthesia. Follow-up head CT performed 3 h after the treatment demonstrated emerging high-density effusion within bilateral subdural space, which was more obvious in the right. The emerging effusion with an average Hounsfield unit (HU) value of 67.4 was diagnosed as SDH by both clinicians and radiologists initially (Figure 2(c) and (d)). No additional intervention was conducted as there was no apparent compression of brain tissue and no aggravation of the condition. Surprisingly, head CT performed 32 h after the treatment demonstrated that the high-density effusion resolved spontaneously (Figure 2(e) and (f)).

Twenty-one days after the first treatment, further endovascular therapy was conducted to occlude the fistula and the initial part of drainage vein completely using coils and onyx-18. This procedure took two and a half hours, using 45 ml iopromide in total. Head CT performed 3 h later demonstrated high-density effusion within bilateral subdural space again, which was less compared to the previous, with an average HU value of 70.2 (Figure 2(g) and (h)). Based on the previous experience, low molecular weight heparin was still given to prevent thrombus from spreading to normal veins, considering the high-density effusion was contrast medium rather than hemorrhage. Head CT performed 29 h after the second treatment also showed the effusion disappeared spontaneously (Figure 2(i) and (j)).

Follow-up MR angiography at seven months showed no evidence of fistula. A 16-month clinical follow-up showed normal motor and language development.

Discussion

Subdural high-density effusion appeared after both of the endovascular treatments of the high-flow pAVF and disappeared spontaneously and completely in a short time without aggravation of the condition, indicating that the high-density effusions contained contrast medium rather than hemorrhage. Compared with the seven cases reported, this case is more typical in imaging and clinical features and the first related to a pediatric pAVF. With a review of literatures, the mechanism of SDCE secondary to endovascular treatment of pAVF and how to differentiate SDCE from SDH are discussed below.pAVF is a rare cerebral vascular anomaly located in the subpial space and has a distinct angioarchitecture which consists of one or more cortical arteries in direct connection with a single cortical drainage vein through one or more fistulas. Due to the direct flow of high-pressure arterial blood into the draining vein, venous dilatation, hypertension, and even rupture may occur, especially under high flow conditions. Venous hypertension and arterial steal can hinder the absorption of cerebrospinal fluid and decrease the blood perfusion of brain tissue, resulting in hydrocephalus and brain development retardation and dysfunction. The key point to treatment of pAVF is to occlude the fistula and initial part of drainage vein, so as to eliminate venous hypertension and arterial steal.⁷

It has been reported that SDCE could occur after full-body enhanced CT in elderly patients with multiple injuries. But it presents less effusion and lower CT attenuation compared with SDCE secondary to endovascular procedure. The mechanism of SDCE is postulated to be the presence of intracranial hypotension and dural venous engorgement in the elderly patients, which may cause leakage of liquid components in dural veins carrying contrast medium into the subdural space under the effect of the hydrostatic pressure gradient.⁸ Compared with elderly patients, this child had dural venous engorgement but not intracranial hypotension.

The underlying mechanism of SDCE in this case might be that dural venous hypertension and engorgement secondary to high-flow pAVF lead to the exudation of contrast medium into the subdural space. In this case, multiple high-pressure arteries flowing directly into the drainage vein can cause venous hypertension. The occlusion of the right sigmoid sinus and severe stenosis of the left sigmoid sinus restrict the anterograde venous draining and lead to the retrograde flow of high-pressure blood to the superior sagittal sinus, the straight sinus, and cerebral veins. As a result, the pressure of the dural veins connected with the venous sinuses increases simultaneously.

High intracranial venous pressure could cause cerebral ischemia and hypoxia which increase the expression of angiogenic factors, such as hypoxia inducible factor-1α and vascular endothelial growth factor. These angiogenic factors can induce the proliferation and enlargement of vessels and the increase of connections between vessels in dura mater significantly.⁹

As we know, dural permeability is a simple diffusion process. The capillaries of dura mater are naturally fenestrated, allowing the contrast medium to enter the extracellular space of the dura. However, the capillaries at the outermost layer of the arachnoid membrane are closely connected and do not allow free passage of contrast medium.^10,11 The contrast medium in the hyperplastic and congested dural vessels may diffuse into the subdural space under the action of concentration gradient. In addition, the hydrostatic pressure gradient caused by dural venous hypertension may cause leakage of liquid components carrying contrast medium into the subdural space.

With the increase of contrast agent dosage and intravascular contrast agent concentration, more SDCE may appear. With the excretion of contrast agent and the decrease of contrast agent concentration in dural vasculature, SDCE will diffuse back into the blood and finally disappeared. We speculate that SDCE is a dynamic process under the combined action of exudation and absorption. For this reason, CT attenuation value of SDCE is also dynamic. According to our post-operative CT test, the concentration of contrast agent in the subdural space in this case is only about 0.7% (Figure 3).

Figure 3.

Three bottles of liquid (left) and their CT attenuation values (right): the top is saline (0 HU on average), the middle is cerebrospinal fluid after the first treatment (16.8 HU on average), and the bottom is iopromide diluted to 0.7% using saline (76 HU on average).

In the case of similar operation time and contrast agent usage, SDCE secondary to the second treatment is less, which may be due to partial relief of venous hypertension after the first embolization. In addition, cerebral DSA and enhanced MR imaging show cortical venous reflux, and meningeal enhancement is more significant in the right hemisphere. These findings suggest that the congestion of the right dural veins may be more serious, consistent with more SDCE in the right hemisphere.

CT attenuation value of SDCE is dynamic, depending on many factors such as contrast agent dosage and the time of CT examination after treatment. SDCE is usually in the range of 20–124 HU on CT and easy to be mistaken as SDH (28–82 HU).^4,6 There are some key points to distinguish SDCE from SDH. First, spontaneous SDH or operation-related SDH often leads to aggravation of the condition and requires active intervention, while SDCE generally does not cause aggravation of the condition.¹² Second, CT attenuation value of SDH changes slowly, while that of SDCE changes rapidly and dynamically. What is more, SDH generally lasts two to three weeks, while SDCE can disappear spontaneously within 24–36 h.^6,13

Administration of iodinated contrast medium in endovascular procedure may cause contrast-induced encephalopathy, which can lead to neurological disorders such as cortical blindness, aphasia, and motor dysfunction.¹⁴ When contrast-induced encephalopathy occurs simultaneously, it will be more difficult to distinguish SDCE from SDH. What is more, spontaneous resolution of acute SDH and enhancement of chronic SDH are also difficult to distinguish from SDCE on unenhanced CT. If possible, dual-energy CT can be used for early differential diagnosis.¹⁵ Dual-energy CT has high specificity and sensitivity in the differential diagnosis of contrast effusion and hemorrhage. SDCE manifests as hyperattenuation on traditional mixed images and loss of hyperattenuation on high-monochromatic images and virtual non-contrast images. However, SDH shows hyperattenuation on traditional mixed image, high-monochromatic images, and virtual non-contrast images.^8,15

Conclusions

SDCE is rare and might occur on early follow-up unenhanced CT after endovascular treatment of pAVF. It is necessary to avoid mistaking SDCE as SDH due to similar CT attenuation value. Dural venous hypertension and engorgement secondary to high-flow pAVF might lead to the exudation of contrast medium into the subdural space. Generally, SDCE can be differentiated from SDH by not aggravating the condition, dynamic change of CT attenuation value, and disappearing spontaneously within 24–36 h.