Introduction

Leakage of cerebrospinal fluid (CSF) can occur when a defect of the skull base and dura results in an abnormal communication between the subarachnoid space and nasal or middle ear cavities. ^, This pathologic entity was first reported by St Clair Thompson in 1899 and can occur in the nose (CSF rhinorrhea) or the ear (CSF otorrhea). CSF leaks are further classified into traumatic and nontraumatic. Traumatic cases account for 80% to 90% of CSF leaks and more commonly present with rhinorrhea (80%) than with otorrhea (20%) in adult patients. ^, Most traumatic CSF leaks are clinically evident within the first 2 days of injury and almost all within the first 3 months. ^,

CSF leaks that occur secondary to skull base fractures account for 2%–3% of all closed head injuries. The most common fractures associated with posttraumatic CSF leakage involve the anterior skull base, including the frontal sinus and lateral lamella of the cribriform plate. ^, Approximately 12%–30% of patients with anterior skull base fractures develop a CSF leak, which include those with a mild head injury as rated by the Glasgow Coma Scale. Therefore, all closed head injuries should be evaluated for CSF leak. In addition, the incidence of anterior skull base CSF leaks is five to six times higher compared with that of the lateral skull base. ^, This is attributed to the firmer attachment of the dura to the anterior skull base, which may increase the likelihood of dural lacerations.

In the setting of a lateral skull base trauma, CSF leakage typically courses through the middle cranial fossa (tegmen tympani and mastoideum) and into the epitympanic recess, antrum, and mastoid air cell tract. Temporal bone fractures along the petrous bone and middle ear associated with dural defects may lead to CSF otorrhea if there is a perforation within the tympanic membrane, or CSF rhinorrhea via the Eustachian tube if the tympanic membrane is intact. In a retrospective study of 1773 patients with posttraumatic CSF leaks from the Taiwan Traumatic Brain Injury Registry System, the temporal bone was the most common fracture site (40.3%) for those with CSF otorrhea. The mortality rate in patients with CSF otorrhea was 8.5%, and CSF rhinorrhea was 10.9% (33/302). These patients also had a higher rate of intracranial hemorrhage (64.7%) compared with those without CSF leakage (28.8%, P < .001). Furthermore, in a study of 13,861 pediatric patients admitted to the hospital with skull fractures, 1.46% of patients developed CSF leaks, of whom 58.4% presented with CSF otorrhea and 41.6% presented with rhinorrhea. Notably, patients with CSF leaks were more likely to have longer average hospitalizations (9.6 vs. 3.7 days, P < .0001) and higher rates of neurologic deficits (5.0% vs. 0.7%, P < .0001), meningitis (5.5 vs. 0.3%, P < .0001), and hospital readmission (24.7% vs. 8.5%, P < .0001) at 90 days. Taken together, these studies elucidate the importance of diagnosing and managing skull base CSF leaks following traumatic injury.

Lateral skull base CSF leaks tend to have a higher likelihood of spontaneous resolution than anterior skull base CSF leaks. ^, In a review of 81 cases of posttraumatic CSF leaks, 17/28 (60.7%) of lateral skull base CSF leaks resolved spontaneously, compared with only 14/53 (26.4%) of anterior skull base leaks. Similarly, in a retrospective study of 699 patients with 820 temporal bone fractures and 122 CSF leaks, 95/122 (77.9%) CSF leaks resolved spontaneously in less than 1 week, 21/122 (17.2%) in less than 2 weeks, and only 5/122 (4.1%) persisted. Despite the higher rates of spontaneous resolution, lateral skull base CSF leak following traumatic injury must be appropriately evaluated and managed to reduce the risk of meningitis and other complications.

Spontaneous CSF leaks represent cases that present without distinct etiologies for the leak, such as trauma, surgery, and congenital anomalies. Recent studies have noted an increase in the incidence of spontaneous CSF leaks, exploring associations with rising obesity rates, obstructive sleep apnea, idiopathic intracranial hypertension, and superior semicircular canal dehiscence. Unlike traumatic or intraoperative CSF leaks, spontaneous CSF leaks can have a more insidious presentation with nonspecific symptoms of otorrhea, hearing loss, and aural fullness. The etiology underlying spontaneous lateral CSF leaks is not entirely well elucidated, but one theory posits that arachnoid granulations within the temporal bone respond to CSF pulsations, causing erosion of the skull base over time.

History and physical exam

Unlike spontaneous CSF leaks in which the history can be vague, traumatic CSF leaks arise, by definition, from head trauma with associated skull base fractures. Nevertheless, while a CSF leak may be suspected, determining whether the otorrhea or rhinorrhea contains CSF can be challenging. When combined with blood, a CSF leak may stain as a “ring” or “halo” on an absorbent material. Caution must be taken when interpreting this finding, since a mixture of blood and other clear fluids can also present a similar stain. Patients with traumatic injuries may also present with occult and delayed CSF leaks, with an average of 13 days posttrauma (range 1–30 days). This delayed presentation could be due to herniation of dura into the bony defect, hematoma blocking CSF flow and leakage, slow resolution of edema, progressive increase in intracranial pressure, or wound contraction. ^, Patients with delayed presentation of CSF leak would also be at a risk of delayed meningitis. Signs of extensive skull base trauma including fractures of the otic capsule, facial nerve weakness, and ossicular chain discontinuity should increase suspicion for a traumatic CSF leak.

Physical exams should include a complete neurological and otologic exam, including an otoscopic exam to identify fluid. The tympanic membrane should be examined for a perforation or middle ear effusion. Nasal endoscopy can be used to identify whether there is transgression of clear rhinorrhea down the Eustachian tube into the nasopharynx. Patients may describe a salty or sweet taste in their mouth, most commonly in a position of standing or learning forward. It is important to note that CSF rhinorrhea can present intermittently and even mimic other rhinologic pathologies including allergies, creating difficulties for diagnostic testing. Furthermore, history and physical exam can be limited by the patient’s physical and mental status secondary to the traumatic injury. Patients may have severe intracranial injuries or require intubation, which present challenges to the diagnostic evaluation of CSF leak.

An audiogram is an essential component of the workup to guide management for patients with a traumatic injury of the lateral skull base. In a retrospective review of pediatric patients with temporal bone fractures, 2.9% with otic capsule-sparing fractures developed sensorineural hearing loss, while 47.1% developed conductive hearing loss. Notably, conductive hearing loss often resolves over a period of 6 weeks to 3 months. For example, a retrospective study of 173 patients with otic capsule-sparing temporal bone fractures demonstrated that the air–bone gap closed from 27.2 dB (average 22 days posttrauma) to 19.6 dB (average 80 days posttrauma) without surgical intervention. Closure of the air–bone gap may be related to resolution of a middle ear effusion and/or a fibrous reattachment of the ossicles. All patients with otic capsule-violating fractures had sensorineural hearing loss, and 20% had CSF leaks. In another study, Magliulo et al. reported that 42.5% of patients with otic capsule-violating injuries had CSF leaks. The presence of a CSF effusion in the middle ear can cause a conductive hearing loss, which should resolve with management of the CSF leak.

The diagnostic evaluation often involves confirmation of CSF leak by beta-2 transferrin analysis. Beta-2 transferrin is a glycoprotein detected in CSF but not in nasal or middle ear drainage or tissue. This allows beta-2 transferrin to be a marker for CSF rhinorrhea and otorrhea with a high sensitivity and specificity. In a prospective study on 205 patients with suspected CSF leak, 35 tested positive for beta-2 transferrin. Of those who tested positive, 34 were confirmed to be true positives through evaluating the patient’s history, using radionuclide cisternography, and intraoperative visualization. Additional advantages of beta-2 transferrin testing include its noninvasive approach and relatively low cost. However, disadvantages of this approach include the amount of fluid needed for the assay and the turnaround time, which can take up to 2 days.

High-resolution computed tomography (HRCT) scan is a noninvasive modality for diagnosing CSF leak. This modality has been reported to be 87% accurate in demonstrating the presence of a CSF leak. In a prospective study of 45 patients with suspected CSF rhinorrhea, HRCT accurately detected the presence or absence of CSF leak in 93% of patients. This is attributed to the utility of HRCT in identifying bony defects and fractures that may indicate the site of CSF leakage. In particular, thin fractures within the temporal bone can be readily visualized with HRCT and increase confidence of a suspected diagnosis of CSF otorrhea or rhinorrhea. This includes determining whether the fracture spares ( Fig. 12.1 ) or violates the otic capsule ( Fig. 12.2 ), which will be important for clinical management. For traumatic injuries of the lateral skull base, physicians may be able to more easily identify the site of leakage using HRCT compared with a spontaneous CSF leak.

Axial-cut high-resolution computed tomography scan of an otic capsule–sparing temporal bone fracture. Black arrow indicates a longitudinal fracture line.

Adapted from Brodie HA. Management of temporal bone trauma. Cummings Otolaryngol Head Neck Surg. 2010;2036–2047.

Axial-cut high-resolution computed tomography scan of an otic capsule–violating temporal bone fracture. Black arrow indicates a traverse fracture line.

Adapted from Brodie HA. Management of temporal bone trauma. Cummings Otolaryngol Head Neck Surg. 2010;2036–2047.

Magnetic resonance imaging (MRI) is another radiologic study that can be helpful in the diagnosis of CSF leak. This modality is noninvasive and involves detection of CSF on T2-weighted images with fat suppression. MRI has been associated with a sensitivity of 87%, though the combination of MRI and HRCT has been shown to have a sensitivity of 93% and specificity of 100%—all of which are slightly lower or comparable with those of HRCT. ^,

Other imaging modalities have been described, though these are not widely employed. Radionuclide cisternography involves intrathecal administration of a radiotracer isotope, indium-III diethylene-triaminepentaacetic acid, to assess CSF leaks. Notably, this technique requires patient follow-up over a period of 2–3 days following the injection. In a retrospective review of 42 patients with CSF rhinorrhea and/or otorrhea, radionuclide cisternography had a sensitivity of 76% compared to 100% for high-resolution CT. Additionally, radionuclide cisternography did not detect CSF leaks in 29% of patients who had clinical evidence of CSF leaks. Other studies report a false positive rate of 33% and accuracy of 28% for patients with inactive leaks. ^, Considering the discomfort caused by intrathecal administration, need for follow-up, and relatively low sensitivity and accuracy, radionuclide cisternography should be reserved for patients with multiple skull base fractures, or clinical scenarios where the diagnosis may be otherwise uncertain. CT cisternography with contrast presents a similar case and involves an intrathecal administration of radiopaque contrast. As observed for radionuclide cisternography, the sensitivity of CT cisternography is low (48%) compared with that of high-resolution CT (58.8%–100%). The density of the dye may also lead to difficulties in locating the bony defect. Therefore, this imaging modality should primarily be used for active CSF leaks to increase its sensitivity. When there is concordance between clinical findings and high-resolution CT imaging, CT cisternography is typically unnecessary considering its poorer sensitivity and discomfort of intrathecal contrast for patients.

Intrathecal fluorescein can be used preoperatively or intraoperatively to identify and localize CSF leaks. This technique involves placing a lumbar subarachnoid catheter and administering an intrathecal injection of fluorescein. Although its use has not been approved by the U.S. Food and Drug Administration, several studies have shown that using a low dose of intrathecal fluorescein resulted in an acceptable safety profile and allowed for successful identification of CSF leaks, ranging from 59.7% to 80.5%. For example, one study involving 419 patients undergoing endonasal endoscopic skull base surgery reported a sensitivity and specificity of 92.9% and 100%, respectively, for identification of intraoperative CSF leak using intrathecal fluorescein. However, intraoperative fluorescein visualization should not be used alone to localize CSF leaks due to the presence of false negatives, which can be as high as 26.2%. Additionally, safety concerns remain, with complications including seizure, hydrocephalus muscle paralysis, and radicular symptoms. To avoid these complications, intrathecal application of fluorescein should be limited to 2.5–50 mg (less than 5% concentration).

Management

Management of lateral skull base CSF leaks following traumatic injury should be tailored according to the fracture characteristics, leakage site, and associated intracranial defects. An algorithm for management has been summarized in Fig. 12.3 . Most lateral skull base CSF leaks spontaneously resolve. To facilitate spontaneous resolution of the leak, intracranial pressure should be reduced by elevating the head of bed at least 30 degrees, encouraging total bed rest, ordering stool softeners, and informing the patient to avoid Valsalva maneuvers, sneezing, and nose blowing. Conservative management can be employed for up to 10 days following the traumatic injury, with a goal of reducing active flow through the leak. Physicians should be aware that strict bed rest increases the risk of deep vein thrombosis. These risks can be mitigated with the use of sequential compressive devices applied to the lower extremity.

Algorithm for management of CSF leak.

CSF may also be diverted with a lumbar drain or ventriculostomy to avoid infections and decompress the leak. ^, Acetazolamide is a carbonic anhydrase inhibitor that decreases CSF production and is used as a first-line treatment in idiopathic intracranial hypertension. ^, The use of acetazolamide in combination with CSF diversion has been reported to significantly reduce intracranial pressure within 4–6 h of drug administration. Despite relatively widespread use of lumbar drains in the conservative management of CSF leaks, they are not without risk and may lead to increased hospital length of stay and costs, pneumocephalus, herniation, infection, and severe headaches if the drainage rate is too high. ^{, ,} In particular, studies have found a 5%–7% infection rate with lumboperitoneal or ventriculoperitoneal shunts. ^,

Conservative management may also include use of prophylactic antibiotics to reduce the risk of meningitis, though the timing and effectiveness of antibiotics remain controversial. Although most CSF leaks related to trauma will spontaneously resolve, 7%–30% of leaks may result in meningitis. Some studies have found that using prophylactic antibiotics for CSF leaks in the skull base does not significantly decrease the risk of meningitis. On the other hand, Friedman et al. found that only 10% of CSF leak patients treated with prophylactic antibiotics developed meningitis, compared with 21% of those who were not treated with antibiotics, though this was not statistically significant. In a large metaanalysis, Brodie reported that the incidence of meningitis was 2.5% among 237 patients who received prophylactic antibiotics, compared with 10% among those who did not receive prophylaxis ( P = .006). The conflicting data between different studies may be attributed to including patients whose leaks spontaneously resolved within 24 h. One author has suggested using antibiotics for 72 h after the active leak has resolved, though more evidence is needed prior to widespread implementation of this practice pattern.

If patients do not respond to conservative management within 7–10 days ^, or present with other complications related to the trauma, such as facial nerve paralysis and/or brain herniation, surgical intervention should be considered. The most common intervention is a middle cranial fossa approach (48%–50%), followed by a combined transmastoid and middle cranial fossa approach (24%–29%), and transmastoid approach alone (20%–22%). ^, Selection of surgical approach is based on the size, location, number of skull base defects, and surgeon preference. In general, a middle fossa craniotomy is useful for large bony defects involving the tegmen tympani or petrous ridge. A major advantage of this approach is the visualization of the entire middle cranial fossa floor when multiple defects are present (including asymptomatic defects) and the ability to place multiple layers of grafts. However, elderly patients may be more susceptible to dural tears due to adherence of dura to the middle cranial fossa floor or thinning of dura over the defects. Perioperative risks are low but include stroke, seizures, and intracranial and extracranial hematomas.

The combined middle fossa craniotomy–transmastoid approach is typically used for large defects involving the tegmen tympani and mastoideum. This combined approach has been associated with the lowest leak recurrent rates and is increasingly utilized in the recent literature. ^, This approach allows for a wide mastoidectomy to carefully inspect the middle cranial fossa floor and minimize the chance of overlooking other leak sites.

The transmastoid approach can be effective for small tegmen mastoideum dehiscences. The primary advantage of this approach is that it avoids the risks of a craniotomy. One prospective study on temporal bone CSF leaks reported that 9 out of 15 patients were treated with the transmastoid approach and no leak was observed at follow-up. Other studies report a 2%–14.5% recurrence rate of CSF leaks in transmastoid approaches. In addition, the transmastoid approach has been associated with lower morbidity and shorter hospital length of stay compared with that of a middle cranial fossa approach (1.7 vs. 6.3 days), though these findings may be confounded by the fact that craniotomy approaches could be favored in patients with greater preoperative trauma. Both middle cranial fossa and transmastoid approaches have been shown to result in acceptable hearing outcomes. ^, Nevertheless, the transmastoid approach presents with limitations for large tegmen tympani defects located anteriorly toward the petrous apex or at the level of the anterior epitympanum, which is less well visualized via this approach than an MCF approach, even if the ossicular chain is removed.

Mastoid obliteration for management of a CSF leak is a definitive option. A major disadvantage of a mastoid obliteration, however, is that the patient is left with a maximal conductive hearing loss. In patients with ipsilateral severe or profound sensorineural hearing loss, a mastoid obliteration may therefore be the preferred surgical approach, but in patients with less severe hearing loss, other approaches may be preferable. ^{, , ,} For patients with poor hearing (congenital or due to trauma), high risk of recurrent leaks (e.g., idiopathic intracranial hypertension, thinning of tegmen tympani and tegmen mastoideum, prior surgery, use of anticoagulants), medical comorbidities that contraindicate craniotomy, or bony defects with minimal chance of successful repair using the aforementioned surgical approaches, a subtotal petrosectomy with obliteration of ear with or without overclosure of external canal can be considered. ^,

Selection of reconstructive materials depends on the size and location of the defect and surgeon preference. Fascia, cartilage, pericranium, calvarial bone, synthetic materials such as hydroxyapatite bone cement and fibrin glue, rotational flaps, and free tissue transfer have all been used for repair of lateral skull base defects, both individually and in combination. ^{, ,} Notably, the choice of reconstructive material does not appear to significantly impact clinical outcomes for CSF leaks of the lateral skull base. ^, An exception is the use of artificial titanium mesh to reconstruct the middle fossa floor. In their retrospective review of 86 patients undergoing CSF leak repair, Carlson et al. report that the titanium mesh was associated with an increased risk of wound infection ( P = .039), recurrent CSF leak ( P = .004), and meningitis ( P = .014). If the CSF leak involves brain herniation, a three-layer closure technique can be employed: temporalis fascia as an extradural inlay graft, placement of intracranial bone graft, and temporalis fascia an extracranial onlay graft. In a retrospective review of 92 patients undergoing repair of CSF leaks, the success rate was 100% for surgeries involving a multilayer closure technique, whereas the success rate was 76% for those involving a single-layer closure.

Postoperative care for patients undergoing middle cranial fossa and combined approaches should involve admission to the intensive care unit with neurointensive monitoring. Patients undergoing a transmastoid only approach can be admitted for brief observational admission. Pressure dressings may also be used for 48–72 h after surgery to reduce the risk of CSF leak or pseudomeningocele formation. ^, Patients should also be evaluated for idiopathic intracranial hypertension to reduce the risk of persistent or recurrent CSF leaks.

Considering that traumatic brain injury accounts for 10%–20% of symptomatic epilepsy, seizure prophylaxis is paramount to decreasing the incidence of posttraumatic seizures. In the setting of a traumatic injury, the risk of seizure may be elevated when considering temporal lobe damage, cerebral ischemia, or intracerebral hemorrhage. A randomized double-blind study has shown that phenytoin significantly reduces the risk of seizures in patients with serious head trauma if administered during the first week after injury. However, the incidence of seizures occurring postoperatively after CSF leak repair is relatively low. In a case series of 65 patients undergoing the middle cranial fossa approach for spontaneous CSF leaks, only one patient developed a seizure postoperatively. A similar outcome was observed in a retrospective study of 50 patients with CSF leaks. For patients undergoing the middle cranial fossa approach, Stevens et al. have advocated for intraoperative seizure prophylaxis with 1000 mg of levetiracetam and additional indications for extended temporal lobe retraction. Yet, Zampella et al. report that levetiracetam does not reach prophylactic levels at standard dosing regiments, compared with phenytoin which did reach levels required for seizure prophylaxis. Further studies are needed on determining criteria for seizure prophylaxis following traumatic CSF leak in the lateral skull base.

Recurrent leaks

Recurrent CSF leaks are defined as leaks detected after postoperative day 3. These leaks can present as otorrhea in the external canal or emerge through the Eustachian tube as rhinorrhea or a postnasal drip. They can be detected with a head hanging test or Valsalva maneuver and confirmed with beta-2 transferrin testing. Recurrent CSF leaks have been found to be more common in cases where the transmastoid approach was used alone. ^, The leaks were often found anterior to the initial repair site. ^, Recurrent leaks can be managed with a middle cranial fossa approach or a combined approach. More aggressive approaches include a subtotal petrosectomy with obliteration with overclosure of the external auditory canal. CSF may also be diverted with lumbar drains and ventriculoperitoneal shunts, though this may increase risk of infection.

In addition, comorbid idiopathic intracranial hypertension should be considered in the differential diagnosis as it has been associated with increased risk of leak recurrence. ^, Due to a “pressure relief valve” effect, where the CSF leak may serve as a temporary release valve lowering the intracranial pressure, a diagnostic lumbar puncture can be performed 6–8 weeks after leak repair in patients with active leaks. ^, Furthermore, a pseudomeningocele may form as a collection of CSF deep to a closed wound. While it is not considered a leak, pseudomeningocele can increase the risk of transcutaneous leak formation, which can be managed conservatively with antibiotics and a lumbar drain.

Conclusion

CSF leak following lateral skull base trauma may present as rhinorrhea via the Eustachian tube or otorrhea if there is a perforation within the tympanic membrane. Considering the high morbidity and healthcare costs associated with CSF leaks, appropriate diagnosis and management is imperative to reduce the risk of meningitis and other complications. Patients can present with facial nerve weakness, ossicular chain discontinuity, and skull base fractures involving the otic capsule, which should be evaluated using an audiogram and high-resolution computed tomography scan. Since many lateral skull base CSF leaks have been reported to spontaneously resolve, conservative management can be initiated with or without the addition of a lumbar drain or a ventriculostomy to decompress the leak. Use of prophylactic antibiotics for meningitis remains controversial in the literature. When surgical intervention is implicated, an algorithm for surgical approach and postoperative care should be followed and tailored according to surgeon preference and the size, location, and number of skull base defects.