Catch me if you can: disappearing and reappearing posterior fossa dural arteriovenous malformation

Abstract

We report an unusual case of a dural arteriovenous fistula (dAVF) presenting as acute neck pain and quadripareis in a 55-year-old previously healthy man. Imaging was suspicious for cervicomedullary venous thrombosis and angiography failed to show evidence of arteriovenous malformation or dAVF. The patient was started on warfarin for a presumed cervicomedullary venous thrombosis and there was a significant clinical improvement. However, 3 weeks later, the symptoms recurred and repeat angiography revealed a dAVF fed by a posterior branch of the left middle meningeal artery draining into the posterior fossa vein. We postulated that warfarin caused recanalisation of the previously thrombosed venous pouch allowing for angiographic discovery and treatment of the dAVF. The dAVF was embolised with onyx resulting in the complete obliteration of the dAVF and symptomatic improvement. Although rare, some dAVF can be concealed or disappear on angiography due to thrombosis of the draining vein and warfarin can lead to recanalisation.

Background

Intracranial dural arteriovenous fistulas (dAVFs) are abnormal connections among dural arteries and dural veins or sinuses. It is an uncommon cerebrovascular disease that accounts for 10%–15% of all intracranial arteriovenous malformations (AVMs).^1–3 The presentation can occur in a diverse fashion, depending on venous drainage location and pattern. Symptoms may vary from headache, chemosis, exophthalmos, pulsatile tinnitus, cranial nerve palsies, pulse-synchronous bruits or other neurological deficits deriving from intracranial haemorrhages.^1–3

We describe an unusual presentation of a transiently disappearing and reappearing dAVF in the posterior fossa.

Case presentation

A previously healthy 55-year-old man developed the acute onset of neck pain and rapid quadriparesis. CT demonstrated a hyperdense dilated tortuous vessel in the pontomedullary region with peripheral calcification. MRI revealed a serpentine structure extending from C1 to the pons, compressing the pons, right ventral medulla and spinal cord. These findings were suggestive of an AVM or dAVF with a thrombosed vein in the brainstem. However, digital subtraction angiography (DSA) demonstrated no abnormalities. Therefore, the patient was placed on warfarin to treat a presumed cervicomedullary venous thrombosis and discharged to a rehabilitation centre. He regained strength, right greater than left, and could raise his arms overhead, sign his name and walk with ambulatory assistance. Unfortunately, 3 weeks after, he again developed worsening neck pain, quadriparesis, difficulty maintaining his head upright and bulbar symptoms with slight dysarthria.

Neurological examination at this time revealed dysarthria, right sixth nerve palsy and left greater than right weakness in all four extremities, more pronounced in the legs compared with arms. He could not hold his left arm against resistance and could tolerate only slight resistance in the right arm. He was not antigravity in any muscle group in the legs and had upgoing toes bilaterally.

Investigations

CT (figure 1A–C), MRI (figure 2A and B) and DSA (figure 3A–D) of the first presentation are shown. At the time of the second presentation, 3 weeks later, MRI (figure 2C–E) again demonstrated an abnormal tortuous structure extending from C1 to the pons, ventral to the spinal cord and medulla, and to the right of the pons. Mixed areas of T1 hyperintensity were worrisome for variable-aged intraparenchymal haemorrhages secondary to a presumed thrombosed vein. There was increased intraluminal enhancement at the pons. Resolution of previous T2 hyperintensity in the cervical cord occurred, but there was new increased hyperintensity in the brainstem and right middle cerebellar peduncle. Repeated DSA (figure 4) now demonstrated a dAVF fed by a posterior branch of the left middle meningeal artery (MMA) draining into a large posterior fossa vein.

Figure 1.

CT head without contrast at first admission demonstrated abnormal hyperdense tortuous structure extending from C1 to the pons, ventral to the spinal cord and medulla, and eccentric to the right side of the pons (A–C), with area of calcification (arrow).

Figure 2.

MRI brain T2 at initial admission demonstrated abnormal tortuous structure extending from C1 to the pons, ventral to the spinal cord and medulla, and adjacent to the right side of the pons (A), with minimal contrast enhancement in T1 postgadolinium (B). Three weeks later, repeated MRI demonstrated mixed areas of T1 hyperintensity worrisome for variable-aged intraparenchymal haemorrhages secondary to a presumed thrombosed vein (C). T2 hyperintensity was increased in the brainstem and right middle cerebellar peduncle (D) with GRE sequence confirming blood products within the lesion (E).

Figure 3.

DSA with selective left vertebral: (A) anteroposterior, (B) lateral, and left external carotid artery: (C) anteroposterior, (D) lateral demonstrated no major abnormalities. No evidence of any AVM or dAVF. AVM, arteriovenous malformation; dAVF, dural arteriovenous fistula; DSA, digital subtraction angiography.

Figure 4.

Repeated DSA with selective left external carotid artery injection demonstrated the presence of dAVF (white arrows) fed by a posterior branch of the MMA: (A,B) anteroposterior views, arterial and venous phase; (C,D,E) lateral views, arterial and venous phase. dAVF, dural arteriovenous fistula; DSA, digital subtraction angiography; MMA, middle meningeal artery.

Treatment

The suspicion was that after starting warfarin following initial presentation, the patient developed recanalisation of the venous pouch of a ‘disappearing’ dAVF that spontaneously thrombosed. Therefore, we elected to endovascularly embolise the left MMA feeder with onyx (figure 5), resulting in the complete obliteration of the dAVF.

Figure 5.

Microcatheter injections of MMA lateral views, demonstrated the presence of dAVF (white arrow) (A–C). Anteroposterior: (D) subtracted, (E) native; and lateral (F) subtracted, (G) native views after onyx injection demonstrated complete obliteration of the dAVF. dAVF, dural arteriovenous fistula; MMA, middle meningeal artery.

Outcome and follow-up

Three-month follow-up MRI (figure 6) demonstrated resolution of oedema in the pontomedullary and cervical spine parenchyma. Six-month follow-up angiography showed complete obliteration of the dAVF (figure 7). The patient demonstrated neurological improvement postembolisation.

Figure 6.

MRI repeated 3 months after treatment with resolution of T2 hyperintensity (A–C) and near complete involution of serpiginous vessel previously visualised (D–E).

Figure 7.

Follow-up DSA, anteroposterior (A) and lateral (B) views with injection into the left external carotid artery demonstrated no evidence of recurrent dAVF. dAVF, dural arteriovenous fistula; DSA, digital subtraction angiography.

Discussion

We present a unique case of a 55-year-old man who developed acute quadriparesis with suspected cervicomedullary venous thrombosis. This was treated with anticoagulation and initially, the patient clinically improved. Unfortunately, the patient developed neurological decline again and on subsequent angiography, recanalisation of the venous pouch was observed and a posterior fossa dAVF not seen on initial angiography was found. The dAVF was successfully embolised with onyx. This resulted in complete obliteration of the dAVF and clinical improvement.

Posterior fossa dAVFs are rare and can present in a myriad of ways. Kim et al reported on two cases of posterior fossa dAVF that presented with haemorrhage. Patients with posterior fossa dAVF can present with headache, brainstem symptomatology, cranial nerve dysfunction or even dementia.^4–6

Bret and colleagues reported a case of posterior fossa dAVF that drained into spinal medullary veins in a patient presenting with myelopathy.⁷ A tentorial dAVF draining intrathecally into spinal medullary veins was found and treated with open excision as embolisation via the feeding meningohypohyseal artery trunk was not feasible. Similar to our patient, following treatment, the patient had angiographic obliteration of the dAVF and clinical improvement. The authors reviewed 14 cases of similar intracranial dAVF that drained intrathecally, with the majority posing challenging diagnosis. Further, once surgery or endovascular embolisation occurred, the insult to the spinal cord from the dAVF had already generated permanent deficits leading to poor outcomes. The authors concluded that venous congestion in the spinal cord was the likely culprit for neurological decline. Thus, patients with suspected vascular myelopathy and negative spinal angiography should have six-vessel DSA to rule out a posterior fossa dAVF.⁷

Regarding our patient, we thought once the large draining posterior fossa vein thrombosed, after the initial presentation but prior to the initial DSA, venous hypertension in the brainstem and cervical cord parenchyma oedema had been reduced. Imaging and clinical improvement correlated. Spontaneous thrombosis of his dAVF resulted in angiographic disappearance. Subsequent initiation of warfarin led to recanalisation of the thrombosed venous pouch and subsequent return of the dAVF, along with bulbar and myelopathic symptoms.

Spontaneous thrombosis of dAVF is rare and the exact aetiology is unknown. Luciani and colleagues performed a literature review including 14 cases.⁸ They classified the dAVFs as post-traumatic or spontaneous. Five were post-traumatic and involved atypical feeding arteries and draining veins, not typically seen in dAVFs.^{8 9} They concluded that spontaneous closure of certain types of dAVFs was due to the single and small nature of some of them as opposed to large dAVFs with multiple feeders.^{8 10} Our patient did not have a history of trauma, but the feeding artery was a posterior branch of the left MMA and the draining vein was an atypical vein ventral to the brainstem. Therefore, spontaneous thrombosis in our patient may be related to one arterial feeder and atypical venous drainage. In contrast, spontaneous dAVFs that developed spontaneous thrombosis were thought to be either due to sinus thrombosis or haemorrhage, neither of which were found in our patient.⁸

Learning points.

• Dural arteriovenous fistulas (dAVFs) can present in a myriad of ways, and atypical presentations might be a challenging diagnosis.

• dAVFs can spontaneously thrombose and disappear resulting in an inability to detect on an initial investigation.

• High suspicion clinical scenarios should be evaluated with follow-up imaging to reassess possible dAVF recanalisation.