Low Conspicuity of Ventral Spinal CSF Leaks in the Thoracolumbar Junction in Dynamic CT Myelography: Influence of Trendelenburg Position and Pressure

BACKGROUND AND PURPOSE: Spontaneous intracranial hypotension can be caused by CSF leakage into the epidural space. Dynamic CTM in the prone Trendelenburg position is highly sensitive for the detection of ventral dural tears, but we have observed reduced effectiveness at the thoracolumbar junction with steep pelvic elevation, which might be related to a reduced subarachnoid pressure in lower levels in this position. The purpose of this study was to compare the conspicuity of ventral leaks across spinal levels and to assess the relationship between leak conspicuity and the degree of pelvic elevation. METHODS: We conducted a single-center retrospective cohort of patients with SIH and confirmed ventral spinal CSF leaks who underwent dCTM. All patients underwent CTM in prone Trendelenburg position. An additional acquisition in the neutral position and without additional contrast was obtained selectively when the initial Trendelenburg acquisition was negative or equivocal. Examinations in which the leak became definite only on imaging obtained in the neutral position were classified as Trendelenburg false-negative. TFN frequency was analyzed according to spinal level and the relative height of the leak site with respect to the foramen magnum in the examination position. RESULTS: Of 47 examinations, 43 met the inclusion criteria and were confirmed to have well-localized ventral CSF leaks. TFN occurred in 5/43 examinations and was observed exclusively at lower thoracic levels and the thoracolumbar junction (T11-T12 and T12-L1 levels), with no TFN identified at more cranial levels. When leak levels were dichotomized as above or at/below T11-T12, TFN occurred in 0/35 versus 5/8 examinations, respectively (Fisher test p < .001). The vertical height of the leak site during the examination was associated with TFN occurrence, even within the subgroup of low-level leaks. CONCLUSIONS: Trendelenburg prone dCTM demonstrated a reproducible, level-dependent visualization pattern, with false-negatives clustering at lower thoracic levels and the thoracolumbar junction, particularly at steep hip elevations. These findings support the role of subarachnoid pressure in ventral leak conspicuity on CTM. When a ventral leak is suspected and the Trendelenburg acquisition is negative or equivocal, adding a neutral acquisition may improve leak visualization and localization at the thoracolumbar junction.