Discussion

Arteriovenous malformations are aberrations of vascular anatomy with abnormal connections arising between the veins and arteries. Increased pressure within the venous system results in tortuosity of the vessels and venous hypertension.1 Lumbar arteriovenous malformations may be congenital or acquired, such as following trauma. The presentation is usually insidious with back pain and progressive myelopathy, giving leg weakness and sensory disturbance. There is often bladder and bowel dysfunction.2 A minority of cases have a rapid onset and deterioration of symptoms known as subacute necrotising myelopathy (Foix–Alajouanine syndrome) with pathological findings of spinal cord necrosis.3 Sudden onset headache and other symptoms of subarachnoid haemorrhage are rare and mostly occur with cervical rather than lumbar arteriovenous malformations.4–9

In this case, the headache that prompted initial presentation was severe and associated with signs of meningism. It may have represented bleed from the malformation, similar to that associated with intracranial aneurysms. However, another possibility is that the original presentation was a benign headache and the LP itself provoked a decompensation in pressure in the intrathecal system, which later produced a myelopathy. The original presentation with a purely occipital headache without back pain might support this possibility. Decompensation of arteriovenous malformations following LP can occur, whereby reduced intrathecal pressure interrupts blood flow dynamics within the malformation to accentuate cord ischaemia.3 There is also the possibility of direct trauma to the arteriovenous malformation or associated engorged vasculature.

Spinal MRI and angiography are the primary imaging modalities for arteriovenous malformations.10 MRI may show perimedullary vein enlargement, flow voids and cord oedema. There may be T2 signal change within the cord, although it is not of prognostic value.11 ,10 Angiography can highlight early venous filling and congestion of tortuous dilated veins and provide more detailed information about the location of the shunt.

Our patient underwent cord imaging after developing a myelopathy, 3 weeks after his initial presentation. This 3-week delay is much shorter than the average time to diagnosis of 11–18 months reported in the literature.10 ,12 However, given the initial bloodstained cerebrospinal fluid, should things have been done differently at his initial presentation? Most would agree that with a negative intracranial angiogram and features suggesting a traumatic LP, no further imaging was needed at that point; it was not until new lower limb symptoms developed that spinal imaging was required.13 However, this case highlights the very important lesson that bloodstained cerebrospinal fluid should not necessarily be dismissed as traumatic in the presence of a normal intracranial angiogram; rather, the clinician should consider the possibility that the blood originated from a spinal source; if there are clinical clues to support this, the patient should have spinal imaging. The ‘three-tube test’, that looks for a falling cell count in consecutive CSF samples, does not reliably differentiate a traumatic tap from subarachnoid haemorrhage.13 ,14 Furthermore, particularly bloodstained CSF contains increased oxyhaemoglobin, which interferes with CSF bilirubin detection and is a confounding element in interpretation.15

Delay in diagnosis often results from initial incorrect diagnoses, including spinal stenosis, demyelination, spinal cord tumours or peripheral nerve disorder.2 ,10 In spite of this, there are several reports of a favourable outcome in patients after significant diagnostic delay and accordingly these lesions should be treated regardless of prolonged symptom duration or pronounced neurological deficit.12