Imaging, Clinical, and Demographic Differences in Patients With Type III Spinal Cerebrospinal Fluid Leak (Cerebrospinal Venous Fistulas) Compared With Patients With Types I and II Spinal Cerebrospinal Fluid Leak

Abstract

ObjectiveSpinal cerebrospinal fluid (CSF) leaks are an underdiagnosed cause of neurologic symptoms. The most common types of spinal CSF leaks are from dural tears (type I) and meningeal diverticula (type II). Cerebrospinal fluid–venous fistulas (type III) are less common and underrecognized. The purpose of this study was to evaluate the intracranial and spinal imaging findings in patients with types I/II versus type III leaks.Materials and MethodsThis was a retrospective, institutional review board–approved study performed on adult patients between January 2020 to September 2021 with surgically confirmed type I/II/III spinal CSF leak. Patients had preoperative brain magnetic resonance imaging (MRI) with contrast and medical records detailing symptoms and clinical diagnoses. Patients were excluded for nondiagnostic brain MRIs. Demographic and clinical information were recorded. The presence of extra-axial collections, pachymeningeal thickening, brain sagging, and decreased pontomamillary distance were evaluated on MRI.ResultsSeven patients had type III leaks, and 16 had type I/II leaks. Patients with type III leaks were older (P = 0.0003) and had higher rates of initial misdiagnosis (100% vs 31%) and longer times to correct diagnosis (P = 0.03) compared with type I/II leaks. Intracranial extra-axial collections were never seen with type III leaks but were seen in 50% of type I/II leaks. Pachymeningeal thickening and brainstem sagging occurred in nearly equal frequency between groups. Smaller pontomamillary distances were seen in type III leaks versus type I/II leaks (P = 0.047).ConclusionsWhen evaluating patients with suspected spinal CSF leak, findings of older age, absence of intracranial extra-axial collections, and small pontomamillary distances may raise suspicion for type III versus type I/II leak.