Vascular mechanism · Spinal cord imaging

Spinal dural arteriovenous fistula

what 3D SPACE can show—and what it cannot prove

Heavily T2-weighted 3D SPACE can make serpiginous perimedullary vessels conspicuous and may help narrow a suspected shunt level. It remains a localization aid: catheter angiography is needed to define the arterial feeder, dural shunt and venous drainage for treatment planning.

1. The disease is venous-hypertensive myelopathy

In the usual thoracolumbar SDAVF, a radiculomeningeal arterial branch drains directly into a radicular vein within the dura at a spinal nerve-root sleeve. Arterial pressure is transmitted into the intradural venous system. The resulting venous hypertension reduces the pressure gradient available for normal cord drainage, producing venous congestive myelopathy and—if untreated—progressive neurological injury.

Routine MRI may show longitudinal cord T2 hyperintensity, cord swelling, enhancement and serpiginous perimedullary flow voids. None is independently diagnostic. The pattern should raise suspicion and prompt a vascular work-up rather than be treated as a direct map of the fistula.

2. Why heavily T2-weighted 3D SPACE can make abnormal veins visible

A heavily T2-weighted volumetric acquisition gives bright CSF a high-signal background. Rapid or complex blood flow may lose signal and appear as a dark tubular or serpiginous structure. Thin, near-isotropic source data can then be reviewed in multiple planes or with minimum-intensity projection, helping the reader follow a candidate draining vessel through the thecal sac.

This contrast is useful but conditional. Vessel conspicuity depends on flow velocity and direction, sequence parameters, motion, spatial resolution and reconstruction. A dark line is not automatically a vein, and an absent flow void does not exclude an SDAVF.

1 2 3 4 5

Original imageFigure 1 link

3. “Suggested shunt level” is the defensible wording

The original 3D SPACE study was a retrospective series of 16 patients with angiographically diagnosed SDAVF. Two blinded observers predicted the fistula site correctly in 81% and 88% of cases; accuracy reached 94% when a prediction within one vertebral level was accepted. This is a promising targeting result from a small selected cohort—not evidence that 3D SPACE can independently diagnose or exclude SDAVF.

Venous anatomy explains the limitation. In a separate 31-patient series, enlarged intrathecal draining veins extended a mean of 10 ± 7.7 spinal levels on angiography. A candidate vein approaching a root sleeve may narrow the search, but the longest, darkest or most congested segment does not necessarily mark the shunt.

1 2 3 4

Original imageFigure 2 link

4. MRA can guide the angiogram; it does not replace the angioarchitecture

Time-resolved or contrast-enhanced spinal MRA adds temporal and vascular information that noncontrast 3D SPACE does not provide. Recent reviews report high sensitivity and useful level prediction, but exact-level performance varies across small, heterogeneous studies. Those results cannot be transferred directly to a heavily T2-weighted structural sequence.

Complete spinal digital-subtraction angiography remains the reference examination for defining the arterial feeder, precise shunt, intradural venous drainage, adjacent critical arteries and a safe treatment strategy. It is not infallible: selected retrospective series document false-negative examinations when angiography was incomplete or technically limited. The practical model is integration—MRI raises suspicion, 3D SPACE or MRA may narrow the search, and angiography resolves the treatment anatomy.