CSF-venous fistula

Last revised by Frank Gaillard on 9 Apr 2024

CSF-venous fistulas are an underdiagnosed cause of spontaneous intracranial hypotension. They are direct communication between the spinal subarachnoid space and epidural veins allowing for the loss of CSF directly into the circulation and can be either iatrogenic or spontaneous in aetiology. 

As the number of reported cases is relatively low, precise epidemiological data is lacking. However, there does appear to be female predilection in spontaneous cases 1-3. Individuals are typically diagnosed in middle age (33-72 years of age) 3

The clinical presentation is that of intracranial hypotension and is therefore discussed in that article. The diagnosis should be suspected when clinical symptoms and imaging features of intracranial hypotension are present but no epidural CSF accumulation is identified 2.

Importantly, low opening pressures (<6 cm H2O) are only seen in approximately a third of cases 3. Furthermore, not all patients with subsequently identified CSF-venous fistulae had identifiable changes of intracranial hypotension on cranial imaging 2. Thus, in selected cases, seeking out a fistula may be warranted based on symptoms alone.

Although CSF-venous fistulas have been a recognised complication of lumbar puncture and myelography, they have also been identified in individuals without antecedent intervention 1. It is believed that in at least some individuals, these represent rupture of a perineural cyst/arachnoid granulation into an adjacent vein 1-3. This is supported by the identification of a nerve-root sleeve arachnoid diverticulum (perineural cyst) in the majority of cases (82%) 3

Unlike other cases of intracranial hypotension due to CSF leaks, no accumulation of CSF outside of the dura can be identified (i.e. no epidural fluid, no paranasal sinus fluid, no accumulation of radiotracer of contrast in an extra-dural location). The key to the diagnosis is identifying myelographic contrast opacifying an epidural vein 1. This can be achieved either with cross-sectional imaging (CT myelography or MR myelography) 2, or fluoroscopy (dynamic fluoroscopic myelography or digital subtraction myelography) 1,2.  

The site of fistula is usually in the thoracolumbar spine, most commonly the lower half of the thoracic spine (T7-T12, 68%) 3

Both conventional (non-dynamic) and dynamic CT myelography can be performed, with dynamic techniques preferred (see CT myelography).

The diagnosis hinges on identifying a hyperdense paraspinal vein that represents intrathecal contrast entering the circulation 2. A threshold of 70 HU has been suggested as a cut-off 3, although clearly, what is more important is the identification of a single paraspinal vein that is substantially more dense than veins at other levels. This is facilitated by performing dynamic CT myelography with the patient in lateral decubitus position, fistula side down 7.

The diagnostic efficacy of MR myelography in this context is likely inferior to CT or fluoroscopy-based studies 6. Furthermore, it requires off-label intrathecal injection of gadolinium-based contrast agents with the potential hitherto unknown toxic complications. As such, it should probably not be considered a first-line investigation 2,6.

Off-label warning: The use of gadolinium-based contrast agents for intrathecal administration is not approved by regulatory agencies such as the FDA. Read more: intrathecal gadolinium

Contrast is introduced into the intrathecal space. Using gradual patient tilting the contrast column is allowed to run along the region of suspected fistula, ideally with the patient in a leteral decubitus position 7. The diagnosis is established by visualising contrast passing directly from the subarachnoid space into a paraspinal vein 2. Because the technique requires substraction, once contrast is mixed in the CSF further views cannot be readily obtained and multiple sessions may be required 2.

Unlike digital subtraction myelography, as subtraction is not performed, many views can be performed at the one sitting with a tilt-table C-arm fluoroscopy unit. However, due to lack of substraction, identifying small fistulas can be difficult due to overlying bony structures 2.

Although the number of cases reported is low, at least anecdotally, an epidural blood patch may not be effective, presumably not adequately compressing or disrupting the fistula 2. Surgical exploration and disconnection of the fistula is curative 2

Endovascular selective embolisation of the fistula site, using liquid embolisation agents, is a minimally invasive and a novel approach which can be performed by navigation through the azygos vein then to the paraspinal venous system (the outflow of the CSF-venous fistula) 4.

ADVERTISEMENT: Supporters see fewer/no ads

Updating… Please wait.

 Unable to process the form. Check for errors and try again.

 Thank you for updating your details.