Correlation of Skull Base CSF Leak Flow Rates with Fluid Pattern on Early, Delayed, and Subtraction Volumetric Extended Echo Train T2 MRI

Introduction: Cerebrospinal fluid (CSF) leak is a potentially fatal condition that occurs when a dural defect in the skull base permits CSF communication between the cranial vault and sinonasal cavities. Flow rate is an important property of CSF leaks that can effect surgical decision making and predispose a certain patient population to complications and inferior outcomes. However, predicting CSF flow rate preoperatively is challenging with current diagnostic tools. The present study employs early and late T2-weighted MRI and volumetric subtraction as a novel method of quantifying CSF flow rate, and correlates radiological results with intraoperative findings and clinical outcomes.

Methods: A total of 45 patients met inclusion criteria for this study and underwent 3-Tesla MRI, which included two identical T2-CUBE sequences that were 45 minutes apart. Semiautomated volumetric segmentation of CSF leak volumes was performed on early and late acquisitions using 3D-Slicer, and volumes were subtracted to obtain accumulated CSF volume. CSF volumes were compared between high- and low-flow patients and correlated with treatment outcomes including recurrence.

Results: Out of the 45 patients, 25 (55.6%) had definitive evidence of CSF leak, and 22 (88%) underwent surgical repair. High-flow patients ([Fig. 1]) had higher early (4,057.4 vs. 982.0 mm³, p = 0.04), late (4,584 vs. 1,095.6 mm³, p = 0.04), and accumulated (526.6 vs. 113.6 mm³, p = 0.01) CSF volumes than low-flow patients ([Fig. 2]). The five (22.7%) patients that exhibited postoperative CSF leak recurrence, had significantly greater early (6,296.9 vs. 1,227.8 mm³, p = 0.008) and late (6,874.4 vs. 1,447.9 mm³, p = 0.008) CSF leak volume. Accumulated CSF volume was not significantly greater in patients with leak recurrence (577.5 vs. 220.1 mm³, p = 0.07). Early, late, and accumulated CSF leak volume were significantly correlated with postoperative hospital stay as well as duration of postoperative lumbar drain placement, p < 0.05 for all (Table 1). Accumulated CSF volume was the most significant factor for predicting flow rate. For every one standard deviation increase in volume difference, the odds of high flow leak increased by 11.0 (95% CI: 1.5–238.8; p = 0.008). Early CSF volume was the most significant predictor for postoperative leak recurrence. For every one standard deviation increase in early volume, the odds of recurrence increased by 4.6 (95% CI: 1.5–22.0; p = 0.005).

Conclusion: High resolution T2-CUBE MRI coupled with precise volumetric segmentation and subtraction of CSF leak demonstrated predictive value in differentiating low-and high-flow CSF leaks, as well as identifying patients with postoperative complications, such as leak recurrence. These findings may be useful in the clinical workup and management of patients with anterior skull base CSF leak.