The clinical management of sinonasal CSF leaks has evolved over the last decade. Current comprehensive management reflects an improved understanding of spontaneous CSF rhinorrhea and the potential importance of intracranial hypertension. Novel surgical methods have also been developed to repair the large defects that accompany endoscopic skull base resections. This update will review the clinical presentation and management strategy for spontaneous CSF rhinorrhea, as well as common methods of skull base repair.

Spontaneous CSF Rhinorrhea

Most CSF leaks have a clear etiology, such as those secondary to acute traumatic accidents, congenital skull base dehiscence, or expected outcome of a tumor resection. Those leaks without an identifiable inciting event or cause are usually considered "spontaneous." In early case series, spontaneous CSF leaks accounted for only 3-5% of all leaks. However, more recent case series report much higher rates for spontaneous leaks, ranging from 14-46%. In comparison to a 10% failure rate for all leaks, initial studies estimated failure rates after repair of spontaneous CSF leaks between 25-87%. The apparent relative increase in spontaneous CSF leaks, together with their increased proclivity toward recurrence, generated a renewed interest into their pathophysiology and clinical management.

Pathophysiology: Recent series focusing solely on cases of spontaneous leakage have consistently documented increased intracranial pressure (ICP) in those undergoing surgical repair, with average ICP ranging from 24-26.5 cm H20 (normal 0-15 cm H20). The current hypothesis is that dural pulsations generated from elevated ICP exert direct pressure on the bony skull base. Over time, the thinnest areas of the skull base would completely erode, allowing brain and meninges to herniate and CSF to leak. This hypothesis presumes both elevated ICP and a skull base predisposed to erosion, often secondary to hyperpneumatization.

Figure1: Epidemiology of idiopathic intracranial hypertension.

Epidemiology: Most case series are overwhelmingly comprised of women (77-85%), of middle age (50-61 yrs), with an elevated BMI (average=36.2 kg/m2). With regard to age, gender, and BMI, the clinical characteristics of spontaneous CSF leak patients closely mirror those seen in idiopathic intracranial hypertension (IIH), a condition of unexplained elevated ICP formerly known as psuedotumor cerebri or benign intracranial hypertension (Figure 1). In an MUSC series, 72% of patients met modified Dandy criteria for IIH-leading to speculation that many spontaneous CSF leaks may be an end result of IIH.

Figure 2: Left: CT scan of meningoencephalocele of the lateral sphenoid. Right: MRI scan with "empty" sella turcica.

Clinical Presentation and Workup: Because spontaneous CSF leaks lack a clear inciting event, symptoms are often initially attributed to more common conditions such as allergic rhinitis and correct diagnosis may be delayed. Headaches are seen in roughly 44-91% of cases, with patients also commonly reporting secondary symptoms such as imbalance (13.2-27%), tinnitus (13.2-45%), and visual disturbance (10.5-18%). Correct diagnosis is most commonly performed by testing sinonasal fluid for beta-2-transferrin, a test which is highly specific (>95%). Imaging with CT scan and MRI can give clues as to site of leakage, presence of an associated encephalocele (85-96%), and signs of elevated ICP (Figure 2). Radiographic signs of increased ICP include an empty sella turcica, dilated optic nerve sheaths, posterior globe flattening, or stenosis of cerebral venous sinuses.

Operative Considerations: The most common location for a spontaneous leak varies by case series, but generally involves the lateral aspect of the sphenoid sinus, cribriform plate, or ethmoid roof. Intrathecal fluorescein can be used to localize the leak intraoperatively and ensure a watertight seal following repair. Considering that elevated ICP is likely to persist in the postoperative period, we lean towards using underlay grafting techniques and rigid or semi-rigid materials. Ideally, bone or cartilage can be shaped to span the defect and placed in an underlay fashion. This material is meant to buttress the brain and meninges, minimizing the effect of dural pulations. Mucosal overlay grafts are usually placed over the rigid graft, and these can be free or pedicled.

Post-operative Considerations: In patients with spontaneous CSF leaks suspected of ICP and IHH, we recommend weight loss via diet modification and exercise. Sinclair et al. report outcomes from a prospective cohort study of 25 women with IIH placed on a low calorie (425kcal/day) diet for 3 months. Findings 3 months after completion of the diet showed an average weight loss of 16 kg and a mean decrease in ICP of 8.0 cm H20. There was also significant improvement in headache scores, optic disk elevation (papilledema), diplopia, and tinnitus. Those patients with objective evidence of elevated ICP (either on imaging, lumbar puncture, or fundoscopic exam) are begun on acetazolamide and referred for a formal ophthalmologic examination. Long-term compliance with medication can be problematic, as paresthesias, altered taste, and lethargy may be attributed to the medication.

Outcomes after endoscopic CSF leak repair: Most outcomes data from surgical repair of CSF leaks comes from small and medium-sized retrospective case series. We recently performed a systematic review with meta-analysis of studies reporting success rates after endoscopic repair. A total of 49 studies were examined detailing 1,326 leak repairs. The summary success rate for all types of CSF leaks was 90.6% (95% CI: 80-92.5%).   Interestingly, the success rate for spontaneous CSF leak repairs appears to have increased over time. When examining data from 7 series published within the last 5 years, the summary success rate was 91.1% (95% CI: 86.7-94.6%). The increase in success over time may reflect an improved understanding of the importance of elevated ICP and adoption of comprehensive management strategies, such as those outlined above.

Figure 3: Left: Frontal lobe exposed after endoscopic excision of an esthesioneuroblastoma. Right: Endoscopic appearance after repair with a nasoseptal flap. 

Standard surgical techniques for CSF leak repair may be inadequate to address the challenging defects created with endoscopic skull base resections. Novel multi-layer repair methods have been developed, utilizing local and regional pedicled flaps as overlay grafts. The nasoseptal flap has become the current workhorse for skull base reconstruction (Figure 3). This pedicled flap is based on branches off the sphenopalatine artery and can cover a surface area between 22.6-27.7 cm2. The size and flexibility of this flap makes it ideal for defects of the sella turcica, planum sphenoidale, and ethmoid skull base. Centers utilizing this flap have seen post-operative CSF leak rates drop significantly. The major downside of the flap is prolonged healing at the septal donor site, which can take several months to fully mucosalize.

Videos of endoscopic CSF repairs can be seen at  www.sinusvideos.com