Temporal Characteristics of Type 2 Lateral Spinal CSF Leaks on Digital Subtraction Myelography: Fast, Medium or Slow Leaks?

Abstract

Background and purpose: Type 2 leaks occur in up to 20% of spontaneous intracranial hypotension (SIH) due to a spinal lateral dural tear, typically accompanied by arachnoid hernia. Their CSF-outflow dynamics are unclear, but could have implications on performing myelography for best possible detection. This cross-sectional study analyzed temporal characteristics of type 2 leaks using digital subtraction myelography (DSM).

Materials and methods: Between February 2020 and April 2025, 63 consecutive patients with type 2 leaks were retrospectively identified. Patients undergoing sufficient decubitus DSM (comprising additional fluoroscopy and X-ray images) were included. We assessed the time for the contrast agent to first appear in the epidural space after reaching the level of the leak intrathecally at 1-2 frames-per-second (fps), and categorized them as fast (0-9 sec), medium (10-90 sec), and slow (>90 sec) leaks. Furthermore, effects of intrathecal pressurization, arachnoid hernia size, opening pressure and symptom-duration on CSF-outflow were studied.

Results: Forty-five patients (36 women) were included. Mean age was 39.0 years (SD ± 11.4 years), mean BMI 23.2 (SD ± 3.9) and median Bern score 6 (IQR 5). Type 2 leaks most commonly occurred at the T10/11 level (12/45; 26.7%) ranging between T7/8 -L1/2. During DSM, contrast appeared in the epidural space within 0-9 sec in 3/45 (6.7%), 10-90 sec in 24/45 (53.3%) and >90 sec in 5/45 (11,1%) of cases (range: 4 to 473 sec). If DSM (or fluoroscopy/X-ray) missed the leak, subsequent cone-beam or CT myelography detected it (13/45; 28.9%); total slow leaks were 18/45 (40%). All patients undergoing surgery (40/45) had the leak confirmed intraoperatively. In a subgroup of patients undergoing pressurization during DSM (12/45), there were significantly more leaks detected within 90s (p=0.02), while arachnoid hernia size, opening pressure and symptom duration did not affect CSF-outflow significantly.

Conclusions: Type 2 leaks show a wide range of CSF-outflow characteristics, with most being medium and slow. For DSM, we propose using a 90-second run with intrathecal pressurization and cone-beam CT standby for effective leak detection, whereas less than 1 fps (e.g., 0.5 fps) seems feasible to minimize radiation. Alternatively, dynamic CT myelography can be considered -although timing of CT scans has yet to be evaluated.

Abbreviations: SIH=spontaneous intracranial hypotension; DSM (digital subtraction myelography); CB-CTM (Cone-beam CT myelography); EID-CTM (energy-integrating detector CT myelography).