CSF Rhinorrhea

From the first intracranial repair in the 1900s to the use of endoscopes and image-guidance systems, the management of cerebrospinal fluid (CSF) rhinorrhea has greatly evolved. Dandy is credited with the first surgical repair of a CSF leak via a frontal craniotomy approach in 1926. Various other authors, including Dohlman (1948), Hirsch (1952), and Hallberg (1964), subsequently reported successful repair of CSF rhinorrhea through different external approaches. In 1981, Wigand reported on the use of the endoscope to assist with the repair of a skull base defect. Since then, endoscopic repair has become the preferred method of addressing CSF rhinorrhea, given the high success rate of 90-95% and the decreased morbidity associated with this approach.

The underlying defect responsible for cerebrospinal fluid (CSF) leaks, regardless of the etiology, is the same: disruption in the arachnoid and dura mater coupled with an osseous defect and a CSF pressure gradient that is continuously or intermittently greater than the tensile strength of the disrupted tissue.

Frequency

The frequency of cerebrospinal fluid (CSF) rhinorrhea is determined by the underlying etiology. Please refer to Etiology for further details.

Cerebrospinal fluid (CSF) leaks are generally classified as traumatic, iatrogenic, and spontaneous/idiopathic. Traumatic causes include both blunt and penetrating facial injuries. Iatrogenic causes include neurosurgical and otolaryngologic approaches to neoplastic disease, as well as functional endoscopic sinus surgery (FESS). Most spontaneous, or primary, causes of CSF rhinorrhea are now thought actually to be secondary to elevations in intracranial pressure (ICP) that might be seen in patients with idiopathic intracranial hypertension (IIH). Congenital skull base defects and certain tumors can also lead to CSF rhinorrhea.[7]

A literature review by Lobo et al indicated that in addition to increased ICP, risk factors for spontaneous CSF leaks include obesity, female gender, and obstructive sleep apnea. In the study, about 72% of patients with spontaneous CSF leaks were female, and about 45% had obstructive sleep apnea.[8]

Traumatic CSF rhinorrhea

Penetrating and closed-head trauma are responsible for 90% of all cases of CSF leaks. CSF rhinorrhea following a traumatic injury is classified as immediate (within 48 hours) or delayed. The majority of patients with a CSF leak due to accidental trauma (eg, motor vehicle accident) present immediately. Most of the patients (95%) with a delayed CSF leak present within 3 months after the injury.

Iatrogenic CSF rhinorrhea

In contrast to traumatic leaks, only 50% of patients with iatrogenic CSF leaks present within the first week after the insult. In most cases, the patient will have been discharged when the leak presents itself. Hence, educating the patient regarding the common symptoms associated with a CSF leak such as salty or metallic taste is of paramount importance.

Any surgical manipulation near the skull base can result in an iatrogenic CSF leak. Skull base injuries can vary from simple cracks in the bony architecture to large (>1 cm) defects with disruption of the dura and potentially brain parenchyma.

Otolaryngology procedures, including FESS and septoplasty, can lead to a skull base defect and CSF rhinorrhea. Certain neurosurgical procedures such as craniotomy and transsphenoidal pituitary resections are most commonly associated with an increased risk of CSF rhinorrhea.

In patients undergoing endoscopic sinus surgery, the site of injury is most frequently the lateral lamella of the cribriform plate, where the bone of the anterior skull base is thinnest. Other common locations include the posterior fovea ethmoidalis and the posterior aspect of the frontal recess.

Tumor-related CSF rhinorrhea

The growth of benign tumors does not commonly result in CSF rhinorrhea. However, locally aggressive lesions such as inverted papilloma and malignant neoplasms can erode the bone of the anterior cranial fossa. The enzymatic breakdown or destruction of the bony architecture results in inflammation and potential violation of the dura. Even if the tumor itself does not lead to CSF rhinorrhea, the resection typically results in immediate leakage. Hence, the surgical team should be prepared to repair the resulting CSF leak at the time of the resection, either transcranially or endoscopically.

Congenital CSF rhinorrhea

Defects in the closure of the anterior neuropore can result in the herniation of central nervous tissue through anterior cranial fossa. These are infrequently associated with CSF rhinorrhea. The embryologic defect is typically a patent fonticulus frontalis or foramen cecum. Meningoencephaloceles usually present in childhood as an intranasal/extranasal mass that transilluminates and expands with crying (Furstenberg sign). A high index of suspicion should be maintained with all pediatric intranasal masses, particularly those occurring at the midline. A biopsy should never be obtained unless a complete imaging workup has been conducted.

Spontaneous CSF rhinorrhea

Spontaneous CSF rhinorrhea occurs in patients without antecedent causes. This terminology seems to imply that spontaneous CSF leaks are idiopathic in nature; however, recent evidence has led us to realize that spontaneous CSF rhinorrhea may in reality be secondary to an intracranial process, namely elevated intracranial pressure (ICP). There are several causes of elevated ICP; however, the proposed mechanism underlying spontaneous CSF rhinorrhea is idiopathic intracranial hypertension (IIH). Obstructive sleep apnea (OSA) has also been linked to elevated ICP.[9]

Despite the multifactorial causes of elevated ICP, once this problem ensues, the pressure exerted on areas of the anterior skull base such as the lateral lamella of the cribriform or lateral recess of the sphenoid sinus results in bone remodeling and thinning. Ultimately, a defect is formed. At this point, the dura herniates through the defect (meningocele). If the defect is large, brain parenchyma may also herniate through the defect (encephalocele).

Immediate traumatic leaks result from a bony defect or fracture in conjunction with a dural tear. A possible cause of a delayed traumatic leak is a previously intact dural layer that has slowly herniated through a bony defect, finally tearing and allowing the cerebrospinal fluid (CSF) to leak. According to another theory, the tear and bony defect are present from the time of the original injury, but the leak occurs only after the masking hematoma dissolves.

Spontaneous CSF rhinorrhea usually manifests in adulthood, coinciding with a developmental rise in CSF pressure with maturity. The dura of the anterior cranial base is subject to wide variations in CSF pressure because of several factors, including normal arterial and respiratory fluctuations. Other stresses include Valsalva-like maneuvers during nose blowing or straining. This can lead to dural tears in areas of abnormalities of the bony floor.

A study by Lieberman et al found evidence of a significant incidence of multiple simultaneous skull base defects in cases of spontaneous CSF rhinorrhea, reporting the existence of such defects in eight out of 44 patients (18.2%) in the study. The investigators suggested that intracranial hypertension may put patients at risk for developing these defects.[10]

However, increased intracranial pressure is not always present in the case of spontaneous CSF rhinorrhea. Other proposed mechanisms for nontraumatic CSF leaks include focal atrophy, rupture of arachnoid projections that accompany the fibers of the olfactory nerve, and persistence of an embryonic olfactory lumen.

Iatrogenic CSF rhinorrhea results from surgical disruption of the skull base and dura as previously discussed.

History

A thorough history is the first step toward accurate diagnosis. The typical history of a cerebropsinal fluid (CSF) leak is that of clear, watery discharge, usually unilateral. Diagnosis is made more easily in patients with recent trauma or surgery than in others. Delayed fistulas are difficult to diagnose and can occur years after the trauma or operation. These cases often lead to a misdiagnosis of allergic and vasomotor rhinitis. On occasion, the patient has a history of headache relieved by drainage of CSF. Drainage may be intermittent as the fluid accumulates in one of the paranasal sinuses and drains externally with changes in head position (ie, reservoir sign).

A history of headache and visual disturbances suggests increased intracranial pressure. Sometimes, associated symptoms can assist in localizing the leak. For example, anosmia (present in 60% of individuals with post-traumatic rhinorrhea), indicates an injury in the olfactory area and anterior fossa, especially when it is unilateral. Optic nerve deficits suggest a lesion in the region of tuberculum sellae, sphenoid sinus, or posterior ethmoid cells. Patients with recurrent meningitis, especially pneumococcal meningitis, should be evaluated for a defect that exposes the intracranial space to the upper airway, regardless of the presence or absence of CSF rhinorrhea.

Physical examination

Physical examination should include complete rhinologic (including endoscopic), otologic, head and neck, and neurologic evaluations. Endoscopy may reveal an encephalocele or meningocele. Drainage of CSF in some cases may often be elicited on endoscopy by having the patient perform a Valsalva maneuver or by compressing both jugular veins (Queckenstedt-Stookey test). However, most of the time physical examination is unrevealing, especially in patients with intermittent CSF rhinorrhea.

In patients with head trauma, a mixture of blood and CSF may make the diagnosis difficult. CSF separates from blood when it is placed on filter paper, and it produces a clinically detectable sign: the ring sign, double-ring sign, or halo sign. However, the presence of a ring sign is not exclusive to CSF and can lead to false-positive results.[11] In contrast to unilateral rhinorrhea, bilateral rhinorrhea gives no clue of the laterality of the defect. However, even in this situation, exceptions can occur. Paradoxical rhinorrhea occurs when midline structures that act as separating barriers (eg, crista galli, vomer) are dislocated. This dislocation allows CSF to flow to the opposite side and manifest at the contralateral naris. The clinical findings most frequently associated with CSF rhinorrhea are meningitis (30%) and pneumocephalus (30%).

Unless medical or surgical contraindications exist, surgical repair is recommended in all patients with spontaneous or iatrogenic cerebropsinal fluid (CSF) rhinorrhea in order to prevent ascending meningitis.

In patients with nonsurgical trauma, waiting a period of 5-7 days to allow conservative measures (bed rest, stool softeners, and lumbar drainage) to assist with secondary closure of the traumatic defect is reasonable. However, if CSF rhinorrhea persists beyond this point, or if a large skull base defect is observed at the time of injury, surgical repair is warranted.

If an iatrogenic leak is detected intraoperatively, it should be repaired at the time of the original surgery. In most cases of iatrogenic injury presenting in a delayed fashion, surgical repair is necessary. A lumbar drain placed at the time of repair has not been shown to decrease recurrence of the CSF leak.

The most common anatomic sites of spontaneous cerebrospinal fluid (CSF) leaks are the areas of congenital weakness of the anterior cranial fossa and areas related to the type of surgery performed. The lateral lamella of the cribriform plate appears to be involved in approximately 40% of the cases, whereas a defect in the region of the fontal sinus is detected 15% of the time. The sella turcica and sphenoid sinus are involved in 15% of the cases as well.

Common sites of injury secondary to endoscopic sinus surgery include the lateral lamella of the cribriform plate and the posterior ethmoid roof near the anterior and medial sphenoid wall. Rarely, the leak can originate in the middle or posterior cranial fossa and can reach the nasal cavity by way of the middle ear and eustachian tube. These patients typically present with aural fullness due to a serous middle ear effusion.

Surgical repair of skull base defects resulting in cerebrospinal fluid (CSF) rhinorrhea is contraindicated in any patient who is not medically stable to undergo a general anesthetic or comply with postoperative care.

The management of CSF rhinorrhea depends on the cause, location, and severity of the leak. When trauma is the cause, the interval between trauma and the onset of the leak is important. The natural history of CSF rhinorrhea is highly dependent on the underlying etiology.

Traumatic leaks stop spontaneously in the majority of cases, thus a conservative approach is best. The leakage stops within 1 week in 70% of patients, within 3 months in 20-30%, and within 6 months in most patients. The leak almost never recurs. The opposite is true for nontraumatic leaks, as only one third stop spontaneously. Intermittent leakage over several years is characteristic.

Glucose content

A rapid but highly unreliable test is glucose-content determination with the use of glucose oxidase paper. This method of detecting cerebrospinal fluid (CSF) rhinorrhea is not recommended as a screening or confirmatory laboratory test to detect the presence of CSF in the nasal cavity for the following reasons:

• Reducing substances present in the lacrimal-gland secretions and nasal mucus may cause false-positive results.

• Glucose, at a concentration of 5 mg/dL, can lead to a positive result with this test.

• Active meningitis can lower the glucose level in the CSF and may lead to false-negative readings.

• This test is not specific for the side or site of leak.

Beta-trace protein [12]

Also known as prostaglandin D synthase, this protein is synthesized primarily in arachnoid cells, oligodendrocytes, and the choroids plexus within the CNS. Beta-trace protein is also present in the human testes, heart, and serum. It is altered by the presence of renal failure, multiple sclerosis, cerebral infarction, and certain CNS tumors. This test has been used to diagnose CSF rhinorrhea in multiple studies, with a sensitivity of 92% and specificity of 100%. This test is not specific for side or site of leak and can be difficult to collect if the leak is intermittent.

Beta2-transferrin

Beta2-transferrin is produced by neuraminidase activity within the central nervous system. Therefore, beta2-transferrin is located only within the CSF, perilymph, and aqueous humor.

The assay has a high sensitivity and specificity, it is performed rapidly, and it is noninvasive.[13] A minimum of 0.5 mL of fluid is necessary for electrophoresis, but difficulties in collection of this fluid have been noted, especially in intermittent, low-volume leaks.

Beta2-transferrin is stable at room temperature for approximately 4 hours; therefore, immediate refrigeration following collection is recommended. Specimens should not be frozen.

This is currently single best laboratory test for identifying the presence of CSF in sinonasal fluid. It should be kept in mind, however, that this test does not provide information regarding the site or laterality of the defect. Not all centers are capable of testing fluid for beta2-transferrin; therefore, sending the laboratory specimen out for processing may delay diagnosis.

Computed tomography (CT) scanning

High-resolution CT scanning is the imaging modality of choice for identifying a skull base defect associated with CSF rhinorrhea. CT scans may demonstrate skull base defects resulting from accidental or iatrogenic trauma, an underlying anatomic or developmental abnormality, or an erosive lesion such as a neoplasm.

CT scans should be performed in the axial plane with 1 mm (or less) slice thickness and reformatted into coronal and sagittal planes. The evaluation of congenital defects or spontaneous defects may be aided by 3-dimensional reconstruction of the bone to permit in-depth analysis of the floor of the anterior or middle cranial fossa.

Pneumocephalus on a CT scan may indicate a dural tear. A deviated crista galli is a radiologic sign in patients presenting with primary CSF rhinorrhea; this finding supports a congenital bony dehiscence as the etiologic basis for this condition. In some circumstances, an air-fluid level is present in one or more of the sinuses. This is not diagnostic of CSF and may be the result of acute or chronic inflammation.

High-resolution CT imaging may reveal defects in the skull base that do not leak or are not sites of active leaking, making the diagnosis more difficult.

CT cisternography[14]

CT cisternography improves the diagnostic yield of plain CT by injecting intrathecal contrast to better localize the site of the CSF leak. As opposed to conventional CT imaging, only one study is typically necessary. CT cisternography depicts the precise location of CSF rhinorrhea in most patients with active leaks. Patients with intermittent CSF rhinorrhea may have false-negative CT cisternograms. Another disadvantage of this technique is that it may miss cribriform or ethmoid sinus defects.

This is an invasive procedure and is not very frequently used. Despite its low morbidity, it can be associated with nausea, headaches, and acute organic psychosyndromes.

Magnetic resonance imaging (MRI)

MRI typically is not recommended as a first-line imaging modality in the evaluation of CSF rhinorrhea unless an encephalocele is demonstrated on examination or is suspected. Unlike CT imaging, MRI does not delineate well bony defects within the anterior or middle cranial fossa. In addition, MRI is more costly and more time consuming. In many instances, the injection of a contrast agent may be necessary. Similar to CT imaging, MRI may not be localizing.

MR cisternography[15]

Avoidance of intrathecal injection of contrast is a key benefit of MR cisternography. T2-weighted imaging can be used to detect the presence of CSF in the sinonasal cavity without the invasiveness of contrast injection. Pulse sequences performed during MRI can be designed so as to enhance the probability of detecting CSF within the sinonasal cavity. As with CT cisternography, false-negative studies may result when CSF rhinorrhea is intermittent.

Nuclear medicine studies

Radioactive isotopes can be introduced into the CSF by means of a lumbar or suboccipital puncture. Serial scanning or scintiphotography can then be used to determine the distribution of these agents.

A commonly used adjunct is the placement of nasal pledgets in various high-risk areas. These pledgets then can be analyzed for the presence of the tracer. Different tracers, including radioactive iodine-131, radioactive iodinated serum albumin (RISA), ytterbium-169, diethylenetriamine pentaacetic acid (DTPA), indium-111 DTPA, technetium-99m human serum albumin, and99m Tc pertechnetate can be used. Despite their relative safety, studies based on these tracers have several limitations, including the following:

• Precise localization of the defect site is difficult

• The isotope is absorbed into the circulatory system and can contaminate extracranial tissue.

• Patient positioning can cause distal pledgets to incorrectly absorb the isotope.

• False-positive results are present in as many as 33% patients.

• Borderline readings are not reliable. A high reading of radioactivity is necessary to diagnose a true leak.

For further information, please see Cerebrospinal Fluid Leak Imaging in the Radiology section.

The injection of intrathecal fluorescein has been used to diagnose and localize the site(s) of cerebrospinal fluid (CSF) rhinorrhea.

The injection of intrathecal fluorescein is commonly used to diagnose and localize the site(s) of CSF rhinorrhea. However, the US Food and Drug Administration has not approved the use of fluorescein for this purpose.

A lumbar puncture and/or placement of a subarachnoid lumbar drain is used to facilitate the injection. After puncture or drain placement, 10 mL of CSF is withdrawn in a sterile fashion. Precisely 0.1 mL of 10% nonophthalmic fluorescein solution is diluted in the 10 mL of CSF. The mixture is then reinjected into the subarachnoid space over a period of 10 minutes. The use of this dilution and the slow injection technique help minimize central potential complications (eg, seizures) that have been previously reported with intrathecal fluorescein.

In most instances, fluorescein is visible with standard xenon light sources used during endoscopic sinus surgery. However, minute amounts of fluorescein resulting from small bony defects may be difficulty to detect using a rigid endoscope.

Since the peak absorption of fluorescein occurs at 494 nm, a blue-light filter (440-490 nm wavelength) can help enhance visualization. This is particularly useful when fluorescein is filling an encephalocele or in cases of very small leaks that cannot be observed with standard xenon light sources. See the image below.

Conservative management

Conservative treatment has been advocated in cases of immediate-onset cerebrospinal fluid (CSF) rhinorrhea following accidental trauma, given the high likelihood of spontaneous resolution of the leak. Conservative management consists of a 7-10 day trial of bed rest with the head of the bed elevated approximately 15-30°. This angle of inclination is sufficient to reduce the CSF pressure at the basal cisterns. Coughing, sneezing, nose blowing, and heavy lifting should be avoided as much as possible. Stool softeners should be used to decrease the strain and increased ICP associated with bowel movements.

A subarachnoid lumbar drain may be placed to drain approximately 5-10 mL of CSF per hour. Continuous drainage is recommended over intermittent drainage to avoid spikes in CSF pressure. The utility of a lumbar drain is limited in cases of a large skull base defect or iatrogenic CSF leaks. The long-term consequences of a persistent defect in the anterior cranial fossa dissuade many physicians from using this method of treatment.

A study by Albu et al indicated that in patients with CSF rhinorrhea caused by closed head trauma, leakage time can be significantly shortened by early placement of a lumbar drain. In the study, patients treated with early lumbar drain placement had a CSF leakage time of 4.83 days, compared with 7.03 days for those treated conservatively with bed rest and head elevation.[16]

Antibiotics

It is logical to assume that the communication between a sterile environment (intracranial vault) and a nonsterile environment (sinonasal cavity) will ultimately result in infection of the sterile compartment. This has led to the use of prophylactic antibiotics in patients with CSF rhinorrhea. However, no conclusive evidence suggests this practice decreases the risk of ascending meningitis.

Prior studies assessing the benefits of prophylactic antibiotic use in cases of traumatic CSF rhinorrhea have yielded mixed results. Two large meta-analyses of patients presenting with nonsurgical traumatic CSF leaks revealed no difference in the rates of ascending meningitis in patients treated with prophylactic antibiotics compared with patients treated with conservative measures alone.

Similarly, a literature review by Ratilal et al did not find evidence for the usefulness of antibiotic prophylaxis in patients with basilar skull fractures, with or without indication of CSF leakage. Evaluation of five randomized, controlled trials involving patients with CSF leakage found that when those treated with antibiotic prophylaxis were compared with controls, no significant difference existed with regard to the frequency of meningitis, all-cause mortality, meningitis-related mortality, and the need for surgical correction. However, the investigators found the studies to be flawed by biases, determining that no conclusion could be reached on the effectiveness of prophylactic antibiotics in cases of basilar skull fracture.[17]

The use of prophylactic antibiotics in patients incurring skull base injuries during endoscopic sinus surgery has not been studied in a randomized controlled fashion. The administration of antibiotics in this setting is reasonable because patients undergoing sinus surgery have underlying inflammatory or infectious pathology. Invasion of the sterile intracranial compartment with resulting meningitis is a feared complication, which leads to the commonplace use of antibiotics under these circumstances.

Diuretics

Acetazolamide can be a useful adjunct in the treatment of patients with spontaneous CSF rhinorrhea associated with elevated intracranial pressure. Acetazolamide is a nonbacteriocidal sulfonamide that is used primarily as a diuretic, given its ability to inhibit carbonic anhydrase. It inhibits the reversible conversion of water and CO2 to bicarbonate and hydrogen ions.

The relative deficiency of hydrogen ions within epithelial cells results in decreased Na/K ATPase activity, which leads to a decreased efflux of water into the CSF. Ultimately, this reduces the volume of CSF.

A randomized, prospective study by Gosal et al, however, suggested that acetazolamide may not aid in resolving traumatic CSF rhinorrhea and may instead cause harmful metabolic and electrolyte disturbances. The study involved 44 patients with head trauma-related CSF rhinorrhea, 21 of whom received acetazolamide and 23 of whom did not. CSF leaks had a median duration of 5 days before resolving in the acetazolamide group, compared with 4 days in the other patients, with the acetazolamide patients demonstrating decreased levels of serum pH, bicarbonate, and potassium.[18]

The side effects of acetazolamide include weight loss, diarrhea, nausea, metabolic acidosis, polyuria, and paresthesias, any of which may result in the cessation of therapy. Metabolic profiles should be monitored on a regular basis to ascertain the effect on serum electrolytes.

Several surgical options for repair of CSF leaks arising from the anterior skull base exist. There has been a paradigm shift over the last 30 years while choosing the best approach given the advancements made in endoscopic techniques.

Intracranial repair

Intracranial repair was frequently used (and is still used in select cases) for the routine repair of anterior cranial fossa CSF leaks. These leaks were typically approached via a frontal craniotomy. In rare situations, a middle fossa or posterior fossa craniotomy was required. Different repair techniques have been used, including the use of free or pedicled periosteal or dural flaps, muscle plugs, mobilized portions of the falx cerebri, fascia grafts, and flaps in conjunction with fibrin glue. Leaks arising from the sphenoid sinus are difficult to reach by means of an intracranial approach.

Advantages of the intracranial approach include the ability to inspect the adjacent cerebral cortex, directly visualize the dural defect and seal a leak in the presence of increased ICP with a larger graft. When preoperative localization attempts fail to reveal the site of a leak, intracranial approach with blind repair has been successful. In these situations, the cribriform and the sphenoid area, if necessary, are covered with the repair material.

Disadvantages of the intracranial approach include increased morbidity, increased risk of permanent anosmia, and trauma related to brain retraction, including hematoma, cognitive dysfunction, seizures, edema, and hemorrhage. In addition, the postoperative hospital stay is longer, adding to the overall cost of the procedure. Failure rates for this approach are 40% for the first attempt and 10% overall.

External approaches

Defects in the posterior table of the frontal sinus may be approached externally via a coronal incision and osteoplastic flap. The osteoplastic flap provides the surgeon with a view of the entire posterior table of the frontal sinus and is especially useful for defects more than 2 cm above the floor and lateral to the lamina papyracea. In select cases, these defects may also be approached with a simpler eyebrow incision and an extended trephination of the frontal sinus in combination with an extended endoscopic frontal sinusotomy. Care must be taken to avoid unnecessary trauma to the surrounding mucosa and the frontal recess entirely.

External approaches to the skull base can also be obtained through various incisions or through nasal approaches for access to the ethmoid sinuses and sphenoid sinus. These include external ethmoidectomy, transethmoidal sphenoidotomy, transseptal sphenoidotomy, and the transantral approach to the skull base. These procedures are infrequently chosen in current practice, given the high success rates and low morbidity associated with the endoscopic approach. However, they should be part of every skull base surgeon’s armamentarium.

External ethmoidectomy

An external ethmoidectomy begins with a tarsorrhaphy on the ipsilateral eye in order to prevent corneal injury. The incision is made halfway between the medial canthus and the midline of the nose down to bone. Lateral elevation of the periosteum exposes the anterior lacrimal ridge and the lacrimal fossa. The lacrimal sac is elevated and retracted out of the fossa.

As the periosteum is elevated posteriorly along the lamina papyracea, the anterior ethmoidal artery will be encountered 2-2.5 cm posterior to the lacrimal crest. This artery needs to be ligated to increase exposure. The frontoethmoid suture line marks the level of the fovea ethmoidalis, thus dissection should never be superior to this line. The posterior ethmoidal artery is found approximately 1.2 cm posterior to the anterior ethmoidal artery in the frontoethmoid suture line. The optic nerve lies 5 mm posterior to the posterior ethmoidal artery.

The ethmoidal cells are then entered in the area of the lacrimal fossa, and the anterior two thirds of the lamina are removed. A complete dissection of the ethmoid labyrinth is performed. The skull base is then identified in the posterior ethmoids, and the anterior wall of the sphenoid is exposed.

Transethmoidal sphenoidotomy

To perform a transethmoidal sphenoidotomy, an external ethmoidectomy is carried out first as described above. The sphenoid sinus ostium is identified and opened first with a small curette or a beaded probe. A Kerrison punch can then be used to enlarge the opening. The anterior wall of the sphenoid is removed in a meticulous fashion to gain access to the sellar region.

Transseptal sphenoidotomy

The transseptal approach to the sphenoid can be carried out using a sublabial or transnasal incision. An external rhinoplasty incision is preferred by the authors.

The sublabial approach requires the use of a gingivobuccal sulcus incision to expose the pyriform aperture and free the nasal spine. The caudal septal cartilage is then identified, and a left (or right) septal mucoperichondrial flap is elevated. This mucoperichondrial flap is elevated laterally and inferiorly along the nasal floor in the subperiosteal plane. The cartilaginous septum is dislocated from the maxillary crest, and the contralateral nasal floor mucoperiosteal flap is elevated. The contralateral nasal septum is, therefore, not elevated off the cartilage. Once the bony-cartilaginous junction is reached, it is disarticulated and the contralateral posterior flap is elevated. The bony septum is removed to expose the sphenoid rostrum, which is widely removed via osteotomies or a drill to expose the entire sphenoid sinus.

Transantral approach

A transantral approach to the skull base offers wider access to the anterior sphenoid, ethmoids, pterygopalatine fossa, and maxilla. An open anterior maxillary sinus antrostomy is known as the Caldwell-Luc procedure. A gingivobuccal sulcus incision is made, and the anterior wall of the maxilla is exposed. The periosteum is elevated superiorly as far as the infraorbital nerve, exercising extreme care to avoid injuring the nerve as it exits via the infraorbital foramen. A canine fossa osteotomy is performed to enter the maxillary sinus. Kerrison rongeurs are then used to extend the opening into the maxillary sinus. The ethmoidal bone can then be approached medially and superiorly through the maxilloethmoidal angle. A more posterior route is taken to expose the sphenoid sinus. When needed, exposure of the pterygopalatine fossa is achieved by creating an opening into the posterior wall of the maxillary sinus.

Endoscopic approaches

Compared with external techniques, endoscopic techniques have several advantages, including better field visualization with enhanced illumination and magnified as well as angled visualization. Another advantage is the ability to more accurately position the underlay or overlay grafts. Multiple studies demonstrate a 90-95% success rate with closure of skull base defects using the endoscopic approach.[1, 2, 3, 4, 5, 6]

General endoscopic concepts

As previously mentioned, the role of antibiotic prophylaxis has not been studied in a controlled fashion for iatrogenic and spontaneous CSF rhinorrhea. However, the authors believe that given the previously published rates of ascending meningitis in untreated CSF leaks, the administration of perioperative intravenous antibiotics is warranted.

Decongestion of the nasal cavity with topical 1:1000 epinephrine or 4% cocaine solution is recommended in order to maximize endoscopic visualization. Injection of 1% lidocaine with 1:100,000 epinephrine at the axilla of the middle turbinate and region of the sphenopalatine artery via a transoral or transnasal route causes vasoconstriction of the blood vessels and helps to minimize bleeding. The use of intravenous anesthesia with propofol and remifentanil has also been demonstrated to reduce intraoperative blood loss when compared with inhalational anesthesia. This is related to a decreased heart rate, which translates into decreased cardiac output, thus reducing the amount of peripheral circulatory volume.

Placement of a lumbar drain has not been demonstrated to decrease recurrence rates of CSF rhinorrhea after endoscopic repair.[19] In theory, lumbar drain placement decreases the pressure exerted by the CSF at the site of the repair, thus allowing the tissues to heal. However, this theory has not been validated. In fact, a recent study found no difference in leak recurrence when patients who had a lumbar drain were compared to those who did not. This finding remained true when the patients were subdivided according to the etiology of the leak.

In general, lumbar drain placement remains institution and surgeon dependent. One must take into account that a lumbar drain can lead to headaches related to overzealous CSF drainage and limits patient mobility postoperatively. One of the benefits of lumbar drain placement is the ability to administer fluorescein to guide in the localization of the leak.

When a lumbar drain is used, fluorescein mixed with autologous CSF is injected slowly over several minutes. As previously discussed, fluorescein is not approved by the FDA for the diagnosis and treatment of CSF leaks. Precisely 0.1 mL of 10% fluorescein is mixed with 10 mL of CSF or bacteriostatic saline. The authors have found that injecting this mixture over 10 minutes has resulted in significantly fewer adverse events such as seizures when compared with early reports in the literature.

A study by Elmorsy and Khafagy of 31 patients with spontaneous CSF rhinorrhea indicated that skull base defects can be successfully closed endoscopically using a septal graft and a middle turbinate rotational flap. In a retrospective chart review, the investigators found that defect closure was obtained in 27 patients after one surgery, with closure achieved in two more after a second surgery, giving the procedure an overall success rate of 93.5%. Closure was unsuccessful in two of the 31 patients even after a third surgery, leading to referral for a shunt procedure.[20]

Similarly, a retrospective study by Kreatsoulas et al indicated that patients with spontaneous CSF rhinorrhea can safely and effectively be treated with endoscopic endonasal repair, with lumbar puncture performed 24-48 hours postoperatively to determine whether undiagnosed idiopathic intracranial hypertension is present. The investigators also found evidence that patients with morbid obesity undergoing this treatment have a greater likelihood of requiring postoperative CSF diversion (via shunt placement), with the odds ratio of those with a body mass index (BMI) of over 40 kg/m2 being 4.35.[21]

A study by Lemonnier et al indicated that endoscopic endonasal eustachian tube closure is an effective management technique for refractory CSF rhinorrhea occurring after lateral skull base surgery. The surgery was successful in seven out of nine patients in the study, although one of the seven patients required a revision procedure.[22]

A literature review by Makary et al indicated that endoscopic CSF rhinorrhea repair is effective in children, finding a pooled, weighted success rate of 94% on first attempt.[23]

Specific endoscopic approaches

Several different endoscopic approaches have been developed. Each is designed to gain access to the area of interest in the most efficient fashion. The transfrontal, transcribriform, transplanum, transsellar, transclival, and transpterygoid have all been well described.

Transfrontal approach

The transfrontal approach allows access to the floor and posterior wall of the frontal sinus. Leaks originating from this area can be successfully repaired using this approach in the majority of the cases. The frontal sinus outflow tract must be carefully preserved in order to prevent mucocele formation in the long term. The main advantage of the transfrontal approach is that it avoids obliteration of the frontal sinus with an osteoplastic flap. This approach, however, may not effectively manage defects originating in the most lateral or superior aspects of the frontal sinus, since these regions may exceed the limitations of current instrumentation when the technique is performed endoscopically.

The approach begins by performing a complete ethmoidectomy. This is followed identification and dissection of the frontal recess. This area is then widened via a modified endoscopic Lothrop or Draf III procedure, which provides a panoramic exposure of the posterior table of the frontal sinus.

See the image below.

Transcribriform approach

The transcribriform approach exposes the medial anterior cranial fossa from the medial aspect of the middle turbinate to the olfactory groove. Posteriorly, it extends to the anterior aspect of the planum sphenoidale. Removing the perpendicular plate of the ethmoid allows access to the crista galli. Extreme care must be used when dissecting near the area of the olfactory groove as damage to the olfactory fibers will cause anosmia.

Transfovea approach

Access to the lateral aspect of the anterior cranial fossa can be achieved by using the transfovea approach. The dissection extends from the middle turbinate laterally to the lamina papyracea. The frontal sinus marks the anterior limit, and the anterior wall of the sphenoid sinus defines the posterior limit. In some cases, the middle turbinate is removed and the transfovea and transcribriform approaches are combined.

Transplanum approach

The transplanum approach allows exposure of skull base defects along the planum sphenoidale and those with significant involvement of the suprasellar region. An anterior ethmoidectomy is performed first. This is followed by a posterior ethmoidectomy, which provides access to the most anterior aspect of the planum. The anterior wall of the sella is taken down to provide posterior exposure.

Transsellar approach[24]

The transsellar approach is the route of choice for defects on the sella turcica with minimal suprasellar extension. It begins with a complete ethmoidectomy followed by identification and opening of the sphenoid ostia. The opening is then generously enlarged to provide wide exposure to the sella. If bilateral access is needed, the posterior bony septum and the intersinus septum can be removed.

Transclival approach[25]

The first steps to perform a transclival approach include a bilateral complete ethmoidectomy and a wide sphenoidotomy. The intersinus septum and rostrum are taken down. The dissection extends from carotid to carotid bilaterally and exposes the floor of the sella, the optic canals, and the upper clivus. Drilling the posterior wall of the sphenoid sinus permits exposure of the upper one third of the clivus. The abducens nerves define the lateral limit of the dissection. If access to the lower two thirds of the clivus is required, the nasopharynx is exposed via a transnasal route. The basopharyngeal fascia and prevertebral muscles are incised. The clivus is drilled down until the dura is exposed. The eustachian tubes mark the vertical segments of the carotid arteries and define the lateral extension of the dissection.

Transpterygoid approach

The transpterygoid approach begins by performing an endoscopic modified medial maxillectomy. This permits a wide view of the lateral extent of the maxilla and the posterior wall of the maxillary sinus. The infraorbital nerve is then identified and its trajectory followed. A complete sphenoethmoidectomy is then performed. The crista ethmoidalis is isolated, and the main branch of the sphenopalatine artery is identified.

At this point, the surgeon should decide whether a vascularized nasal-septal flap is going to be used to close the defect. If so, every effort to preserve the sphenopalatine artery and its more proximal supply is made. If free mucosal grafts are going to be used, the artery may be cauterized. In either situation, the bone of the posterior wall of the maxillary sinus is removed so the sphenopalatine artery can be dissected proximally to identify the (internal) maxillary artery and its ascending and descending branches. The sphenopalatine artery is also an important landmark since the pterygopalatine ganglion is situated directly posterior to the artery. Care must be taken to preserve the ganglion and its parasympathetic fibers, which contribute to lacrimation.

After the infraorbital nerve, maxillary artery and parasympathetic fibers are identified, the fat within the pterygopalatine fossa may be dissected or cauterized with bipolar cautery until the anterior wall of the lateral recess of the sphenoid sinus is identified. This bone is removed with a drill, thus exposing the contents of the lateral recess of the sphenoid sinus. Typically, any defect in the middle fossa floor occurs in this vicinity, lateral to the Sternberg canal and the foramen rotundum.

See the image below.

Preoperative CT scans in both the axial and coronal planes should be thoroughly reviewed prior to the start of the case. All critical anatomic structures should be analyzed in detail. This includes identifying areas of the skull base prone to injury and spontaneous defects such as the posterior table of the frontal sinus near the frontal recess, the cribriform plate and fovea ethmoidalis, the planum sphenoidale, and if present, the lateral recess of the sphenoid sinus.

The use of image-guidance systems is strongly encouraged for skull base procedures. When available, stereotactic image-guided equipment can be calibrated and used intraoperatively to improve navigation and localization during surgery.

See the image below.

The surgeon and anesthesiologist should communicate a plan prior to the initiation of surgery so as to avoid untoward events during and after the procedure. A multidisciplinary approach involving otolaryngology, anesthesiology and neurosurgery is often helpful for the comprehensive care of the patient.

As wide an exposure of the defect is recommended prior to resecting an encephalocele and repairing a skull base injury. This includes performing an adequate maxillary antrostomy, ethmoidectomy, sphenoidotomy, and, if necessary, a frontal sinusotomy. Widely opening the paranasal sinuses can help with visualization and can help prevent iatrogenic sinusitis postoperatively when the nasal cavity is packed with graft material.

For CSF leaks and encephaloceles occurring in the region of the cribriform plate, removing the middle turbinate is sometimes needed to gain adequate exposure. The removed middle turbinate can then be used as grafting material. Large defects in the sphenoid sinus may require a posterior septectomy for exposure.

See the image below.

Once the defect is isolated, the surgeon must ascertain whether an encephalocele is present. If one is detected, it should be resected using bipolar electrocautery (or cold ablation) until the stalk can be reduced into the anterior cranial fossa. Resection of the encephalocele is a time-consuming process and must be done in a meticulous manner to ensure that all bleeding is controlled so as to avoid intracranial hemorrhage. Once the encephalocele is resected, the mucosa surrounding the defect must be cleared. This is performed by elevating the mucosa away from the defect so as to achieve a margin of 2-5 mm of exposed bone. Bipolar electrocautery should also be used to eliminate nests of mucosa that may remain after elevation. Ensuring that mucosa is not retained within the defect is key to prevent future mucocele formation.

After the defect is completely exposed, its dimensions should be measured using a flexible ruler. Defect size is an important factor that should be taken into account when choosing the number of layers used to repair the CSF leak.

See the image below.

The type of graft has not been shown to affect the success of the endoscopic repair, particularly if the defect is less than 2 mm. The repair of defects 2-5 mm in size is generally successful with a simple onlay graft of mucosa or fascia. If comminution of the surrounding bone or a significant dural tear is found, the placement of a composite graft is warranted. Composite grafts typically involve a multilayer closure in which separate underlay and overlay grafts are used. Defects greater than 5 mm also require composite grafting.

Another important factor that must be considered is whether the patient has elevated ICP. In patients with normal CSF pressure, the size of the defect is less of an issue when it comes to graft selection and method of repair. The repair of encephaloceles and defects resulting from elevated ICP require multilayered grafting.

Several different types of grafting materials exist. In addition to defect size, location, and elevated ICP, the type of graft chosen is also influenced by surgeon’s preference, surgeon’s level of expertise, and tissue availability.

Bone is typically used as the underlay graft. An epidural pocket is created first, and the bone graft is then placed between the dura and the skull base defect. An adequate amount of overlap around the edges of the opening helps prevent dislocation of the underlay graft.

There are a multitude of potential donor sites if bone is required for the repair. Septal bone is commonly used, given that it is easily accessible during an endoscopic approach. If the amount of septal bone available is insufficient, bone can be obtained from other donor sites. These include calvarial bone, iliac crest, and mastoid tip. In addition to requiring an external incision, some of these approaches can cause a significant amount of postoperative pain, especially when bone is harvested from the iliac crest.

Once the underlay graft has been placed and secured with fibrin glue, autologous fat can be placed to further assist with closure of the defect. The abdominal fat graft is usually harvested at the beginning of the procedure via a periumbilical incision. Using this incision rather than a tangential unilateral incision in the right lower abdomen results in better cosmesis and avoids confusion with an appendectomy incision.

Lastly, a mucosal graft is placed as an overlay graft. Mucosal grafts, whether free standing or pedicled, can be obtained from several different sources. A free graft provides a scaffold along which reepithelialization can occur.[26] Septal mucosa and middle turbinate free grafts are commonly used. The amount of mucosa that can be harvested from the septum is significantly larger and thicker than what the middle turbinate provides. A free septal graft should be harvested in a posterior-to-anterior direction to avoid obstructing the operative view with bleeding from the mucosal edges. In order to harvest a pedicled septal flap, 3 incisions are made: vertical, superior horizontal, and inferior horizontal. The flap remains attached posterolaterally. The pedicle contains the sphenopalatine neurovascular bundle. Several different modifications can be made to tailor the length and width to the specific shape of the skull base defect.

In recent years, AlloDerm has been used as an alternative to a mucosal graft. This has proven particularly useful in cases in which there is a lack of native tissue due to prior surgery or involvement of the potential mucosal graft by neoplasm. If a stronger outermost later is needed, such as in clival defects, a temporalis fascia or tensor fascia lata graft can provide the necessary support and has been used with acceptable success.

Grafts and flaps may be anchored to the skull base with the administration of fibrin sealant. The amount of fibrin sealant should be sufficient to anchor the graft yet not too abundant as it can actually prevent the graft from adhering to the underlying tissue. Once all the grafts are in place, the repair is reinforced with Gelfoam and nonabsorbable packing to help apply pressure to the site.

It is of critical importance to ensure adequate placement of the grafts and use fibrin sealant conservatively to avoid obstruction of adjacent paranasal sinuses. The importance of widely opening the sinuses prior to skull base repair cannot be overemphasized. If necessary, a frontal sinus stent may be placed if graft material is used in the frontal recess or adjacent to it.

See the image below.

At the end of the procedure, antiemetics should be administered and the stomach should be aspirated of blood and fluid to minimize postoperative nausea and vomiting. If safe, a deep extubation should be attempted and nasal positive pressure is to be avoided.

If the repair of the skull base immediately followed an inadvertent injury to the skull base during routine surgery (eg, endoscopic sinus surgery), a head CT scan should be obtained to ascertain the extent of injury to the brain.

Lumbar drainage is performed at 5-10 mL per hour for 48 hours. In patients with known or suspected elevated ICP, the drain is clamped after 48 hours for 6 hours. At this point, an opening pressure is measured. If it is above 20 mm Hg, adjunctive medical therapy is advised (see Diuretics in the Treatment section).

Nonabsorbable packing should be removed 7-10 days postoperatively. Regular endoscopic examination with minimal debridement of the surgical site should be performed to monitor for encephalocele and cerebrospinal fluid (CSF) leak recurrence.

If the patient is found to have elevated ICP, close follow-up by a multidisciplinary team involving an internist, ophthalmologist, neurologist, and neurosurgeon is invaluable to monitor for other potential complications of increased ICP, such as papilledema, diplopia, vision loss, and neurologic deficits.

Close monitoring of serum electrolytes should be performed if adjunctive acetazolamide is used. In those patients who develop intolerable adverse effects, placement of a ventriculoperitoneal shunt should be considered. However, patients should be counseled extensively regarding the morbidity associated with a ventriculoperitoneal shunt. This procedure should be done in a timely fashion after the CSF leak repair to prevent recurrence of the leak due to pressure on the repair site by the elevated ICP.

Meningitis is the most feared and severe complication of a CSF leak. Bacterial meningitis is typically due to Streptococcus pneumoniae and Haemophilus influenzae. The risk of meningitis during the first 3 weeks after trauma is estimated to be 10%. The rate increases to 40% in nontraumatic CSF rhinorrhea.

Some studies have reported that conservative management using bed rest and lumbar drains is associated with a high incidence of ascending meningitis. Thus, prompt surgical closure of a CSF leak is advocated by some authors. Meningitis caused by a persistent CSF leak is associated with a high mortality rate.

Only a small percentage (< 1%) of patients develop new meningitis after surgical closure. Not all cases of postoperative meningitis are due to the aforementioned bacteria. Some patients may develop aseptic meningitis due to meningeal irritation as a result of manipulation during surgical repair.

The surgical mortality rate is 1-3% for intracranial procedures and is negligible for external procedures. Anosmia is the main contributor to morbidity for intracranial approaches, occurring in 20-25% of the cases.

Long-term outcomes following endoscopic repair have been well described. Various authors have concluded that most recurrent leaks manifest within 2 years after the repair. The overall success of the repair is determined by the etiology of the leak, with higher failure rates among patients with increased intracranial pressure. Spontaneous leaks recur in an average of 7 months, while traumatic leaks recur in an average of 4 months. Half of the traumatic leaks recur within 2 weeks postoperatively. This is attributed to a technical error and is unlikely to represent a true recurrence.

A study by Teachey et al indicated that in cases of spontaneous CSF rhinorrhea, evaluation and treatment of the patients for increased intracranial pressure results in a greater success rate for primary endoscopic leak repair. In patient cohorts that underwent such evaluation and treatment (with acetazolamide or CSF shunting), the repair success rate was 92.82%, compared with 81.87% in patients who were not actively managed for increased intracranial pressure.[27]

A retrospective study by Adams et al found that in endoscopic repair of CSF rhinorrhea, outcomes were comparable between patients treated on an outpatient basis and those who were admitted to the hospital postoperatively. (However, the proportion of small defects [< 1 cm2] was greater among the outpatients, and the operative technique differed significantly between the two groups.)[28]

Although the management of cerebrospinal fluid (CSF) rhinorrhea has greatly advanced since the first repair was described in the 1920s, controversial areas still remain. The need for lumbar drain placement continues to be a topic for debate, as there are no prospective randomized control studies evaluating the effect of lumbar drain on recurrence rates after endoscopic repair.

There are also unanswered questions regarding the use of other adjunctive measures such as acetazolamide and ventriculoperitoneal shunting. The benefit-risk ratio of long-term acetazolamide use should be considered, given that long-term therapy with this medication is associated with significant adverse effects and subjects patients to a lifetime of blood tests. In addition, the timing of ventriculoperitoneal placement in a patient with spontaneous CSF rhinorrhea after endoscopic repair is not clear. Earlier placement makes the most intuitive sense, but there is no consensus regarding what “early” actually means.

Lastly, there are no conclusive studies demonstrating that prophylactic use of antibiotics after repair of iatrogenic CSF leaks actually decreases the incidence of ascending meningitis. A randomized controlled study is needed to answer this question.