How is a cerebrospinal fluid (CSF) leak diagnosed?

Diagnosis of Cerebrospinal Fluid (CSF) Leak

Laboratory Confirmation

Begin with laboratory testing of fluid samples using β2-transferrin or β-trace protein assays to confirm the presence of CSF before proceeding to imaging. These are the most sensitive and specific laboratory tests available, with β2-transferrin being more commonly used due to availability 1.

• If sufficient fluid cannot be collected for β2-transferrin testing, radionuclide (DTPA) cisternography can confirm the presence of a leak 2, 3.

Initial Imaging Algorithm

For Traumatic or Sinonasal CSF Leaks (Laboratory-Confirmed)

High-resolution CT (HRCT) with thin-section bone algorithm images of the skull base with multiplanar reformation is the first-line imaging study, demonstrating 93% accuracy and 92% sensitivity. 2, 3

• Request maxillofacial CT for CSF rhinorrhea or temporal bone CT for CSF otorrhea 2.
• HRCT correctly identified the leak site in 100% of surgical cases (21/21 patients), outperforming radionuclide cisternography (16/21) and CT cisternography (10/21) 2, 3.
• No additional preoperative imaging is necessary when a single skull base defect is identified on HRCT 2.
• If multiple potential leak sites are present, proceed to CT cisternography for definitive localization 2.

For Spontaneous Spinal CSF Leaks (Intracranial Hypotension)

MRI of complete spine without and with IV contrast, optimized with fluid-sensitive sequences, is the gold standard initial imaging. 4

• The spine is the anatomical source of most symptomatic CSF leaks and venous fistulas, not intracranial structures 2, 4.
• Approximately 20% of initial brain MRIs and 46-67% of initial spine imaging may be normal in clinically suspected cases 2.
• Negative initial imaging should not preclude continued diagnostic workup when clinical suspicion remains high 2.

Second-Line Imaging Options

MR Cisternography

Use MR cisternography (high-resolution T2-weighted or steady-state free precession sequences) as a second-line option when HRCT shows multiple defects or when soft tissue detail is needed. 2, 3

• MR cisternography has 89% accuracy and 87% sensitivity, lower than HRCT 2, 3.
• Superior for identifying meningoceles, encephaloceles, and distinguishing these from sinus secretions due to better soft-tissue contrast 2, 3.
• Sensitivity ranges from 56-94% for leak site identification 2.
• A combination of HRCT and MRI with heavily T2-weighted sequences achieves 90-96% sensitivity 2.

Advanced/Third-Line Imaging

For Complex or Unlocalized Leaks

Dynamic CT myelography or dynamic digital subtraction myelography should be performed when initial imaging fails to localize a spinal CSF leak. 2

• Position the patient based on suspected defect location: prone for ventral dural defects, decubitus for meningeal diverticula or CSF-venous fistulas 2.
• CSF-venous fistulas and slow meningeal diverticular leaks are often subtle and may require advanced imaging with temporal resolution 2.

Contrast-Enhanced MR Cisternography

Consider intrathecal gadolinium-enhanced MR myelography for slow-leaking dural defects when other studies are negative, though this is off-label use requiring special dosing caution. 2, 3

• Sensitivity ranges from 92-100% for active leaks but approximately 70% for intermittent leaks 3.
• Risk of gadolinium-induced neurotoxicity requires careful dosing 2.

Common Pitfalls

• Active vs. inactive leaks: HRCT sensitivity drops from 85-92% in active leaks to 40% in inactive/intermittent leaks 2. The sensitivity of cisternography techniques depends heavily on leak activity at the time of imaging 3.
• Multiple defects: HRCT alone cannot determine which of multiple osseous defects is the active leak source 2.
• Spatial resolution limitations: Radionuclide cisternography has insufficient spatial resolution for preoperative planning despite confirming leak presence 2, 3.
• Timing considerations: For radionuclide cisternography, pledgets are measured after 24-48 hours, and delayed imaging up to 72 hours may be needed 2.