CHAPTER 341 Traumatic Cerebrospinal Fluid Fistulas
CSF fistulas may also be iatrogenic, such as after pituitary transsphenoidal surgery or sinus surgery, or may be spontaneous. In addition, CSF leakage can occur after compound vault wounds. These injuries are treated by repair of the dural tear and careful wound closure (see Chapter 339). Fistulas associated with skull base fractures form the subject of this chapter.
The most common sign of a fistula is leakage of CSF from the nose or ear. A fistula may also be indicated by intracranial air (pneumocephalus), with or without leakage of CSF, and be implied by a cranial infection (meningitis or abscess) occurring at any time after a skull base fracture. The frequency of skull base fractures increases with the force applied to the cranium.1 Consequently, CSF fistulas are more common after severe head injuries, and for this reason they may be overlooked initially. It is therefore important to have a high degree of suspicion with head injuries associated with skull base fractures, whether closed or penetrating. Fistulas may also occur with midface fractures of the Le Fort III pattern, without an associated head injury. Most fistulas heal spontaneously, particularly when the primary impact is to the facial skeleton.2 The advice of earlier neurosurgeons that “dural repair should be considered in all cases of paranasal sinus fracture with rhinorrhea, whether this is of early onset, or brief or long duration”3 no longer holds true.
Epidemiology
The causes of CSF fistula reflect the causes of neurotrauma in the community. In general, the most common causes in order of frequency are motor vehicle accidents, falls, and assaults.4,5 In series in which facial fractures dominate, there is a higher frequency of assaults and motor bike accidents. The reported incidence of skull base fractures after nonpenetrating head injury ranges from 7% to 24% and that of associated CSF fistulas from 2% to 20.8% after head injury.6,7 The incidence of skull base fractures increases with the severity of head injury; however, rhinorrhea can develop after minor head injury or primary facial impact in which there has been little or no loss of consciousness.8 Cranionasal fistulas are more common than cranioaural fistulas and less likely to cease spontaneously.9
Pathophysiology
Blunt Injury
Anterior Fossa
Rhinorrhea is most often caused by a fracture of the frontal, ethmoid, or sphenoid bones. The dura is firmly adherent to the thin bone of the anterior fossa floor and is readily torn by fractured bone edges. The most frequent site of rhinorrhea is the cribriform/ethmoid junction and the ethmoid bone itself.10 The anterior ethmoidal artery penetrates the skull base at the lateral margin of the cribriform plate, thereby creating a natural weakness at that point (Fig. 341-1).11 Fistulas in this region communicate with the nasal cavity directly or via the ethmoid air cells.
Middle and Posterior Fossa
Between 70% and 90% of temporal bone fractures are parallel to the long axis of the petrous ridge. These longitudinal fractures may damage the ossicular bones and result in conductive hearing loss, as well as damage to the seventh nerve. The tympanic membrane is often torn. A transverse fracture (10% to 30%) is more commonly associated with eighth nerve deficits, sensory neural hearing loss, and facial palsy.12 The tympanic membrane is usually intact and CSF will leak via the nose (otorhinorrhea).
CSF fistulas occur with equal frequency with either type of fracture.
A middle fossa fracture that extends to the greater wing of the sphenoid may enter a lateral extension of the sphenoid sinus, which is present in nearly a third of skulls, and result in rhinorrhea.13
Oculorhinorrhea
Rarely, a cranio-orbital fracture together with a laceration of the conjunctival sac may allow CSF to leak from the eye.14,15
Penetrating Injury and Gunshot Wounds
Traumatic CSF leaks may also result from penetrating injuries, including gunshot wounds. Penetrating missile injuries of the skull vault have a high incidence of skull base fractures, 30% of which are discontinuous (i.e., not in continuity with the vault fractures).10 Either rhinorrhea or otorrhea may develop in approximately 50% of these injuries.16 High-velocity missile injuries can cause substantial bony and soft tissue loss and disruption, and repair and reconstruction are often complex (see Chapter 339).
Cerebrospinal Fluid Fistulas in Children
CSF fistulas are less common in childhood, with only 15% occurring in children younger than 15 years. The low frequency in children is due partly to a lower frequency of frontal impact but also to the greater flexibility of the cartilaginous components of the skull base and underdevelopment of the sinuses.17 The frontal sinus is not developed until the age of 4 years or older. The ethmoid sinuses are present at birth and enlarge rapidly, but the ethmoid component of the anterior fossa is cartilaginous and therefore flexible at birth. By the age of 3 years, the nasoethmoid cavities are proportionately equivalent to their size in adults. The sphenoid sinus is very small at birth and becomes related to the anterior fossa between 5 and 10 years of age. The tegmen tympani is thin and rigid at birth, and a fistula to the middle ear is possible. Mastoid air cells are very small at birth but increase up to the age of 5 years (Fig. 341-2).
Time of Onset of Leakage of Cerebrospinal Fluid after Trauma
Early Onset
After nonpenetrating injuries, CSF rhinorrhea usually begins within 48 hours.4,5,7,18 If the defect is small, the bone may heal, and 60% to 70% of cases will cease spontaneously within 1 week.8
Delayed Onset or Recurrence
Very-Late-Onset Cerebrospinal Fluid Leakage or Infection
CSF rhinorrhea may first develop after a considerable delay, or infection alone may be the first sign of a fistula. Delayed meningitis without a history of CSF rhinorrhea has been reported up to 48 years after the original head injury, which might have been quite minor.22–24
Cranioaural Fistula
Most cases of traumatic otorrhea and otorhinorrhea cease spontaneously. In a large series, less than 5% persisted more than 14 days.26 Hence, most can be treated conservatively.27 Nonetheless, the incidence of meningitis while awaiting spontaneous healing has been reported to be as high as 18%.9
High- and Low-Pressure Leaks
In the early stages after severe head injury, potentially increased ICP may be partly “controlled” by a CSF leak, but in most instances persistent fistulas are not associated with increased ICP. High-pressure leaks are more common with spontaneous, nontraumatic CSF leakage; however, if the leak is accompanied by posttraumatic hydrocephalus, it may be maintained by the high CSF pressure, and the hydrocephalus should be managed by insertion of a lumbar peritoneal shunt as the initial treatment.19
Intracranial Air (Pneumocephalus)
Intracranial air is present in 20% to 30% of patients with posttraumatic CSF fistulas, but it may also occur without rhinorrhea.28 Alone, it carries the same risk for meningitis as a CSF leak does.29
The mechanism of pneumocephalus is debated.30,31 It may be caused by a ball-valve effect combined with episodic increased pressure within the nasopharynx on coughing or by an “inverted bottle” effect whereby there is a discontinuous exchange of air and CSF when change in posture reduces ICP below atmospheric.
Tension pneumocephalus is an uncommon but serious complication.32
Clinical Features
History
The history usually consists of a head or facial injury from a frontal impact. A profuse leak is readily identified, but small and intermittent leaks are often overlooked, particularly if the CSF is mixed with blood and mucus. In an unconscious patient with signs of a skull base fracture, evidence of CSF leakage should be sought actively. A conscious patient may complain of a nasal discharge or a salty taste in the back of the throat (because of the sodium content of CSF) or fullness of the ear with some hearing loss. A large volume of fluid leaking from the nose or ear with change in head position indicates that a CSF-filled sinus has drained (reservoir sign).33 Patients may complain of nasal dripping in the morning after sitting up, leaning forward, or straining.
Examination
Neurological examination may provide valuable localizing information. Anosmia suggests a fracture of the anterior cranial fossa at or near the cribriform plate; however, intact olfaction does not exclude the possibility of a cribriform fracture.32 Seventh or eighth nerve palsies and impaired balance suggest a fracture of the temporal bone involving the labyrinth. Defects in vision or visual fields indicating optic nerve injury and sensory loss of the first division of the trigeminal nerve suggest fracture of the anterior fossa.
The side of rhinorrhea cannot be relied on to locate the fistula. Although usually ipsilateral to the fracture site, it is frequently contralateral or bilateral.3,32
Investigation
Identifying Cerebrospinal Fluid
Bedside Tests
Glucose oxidized test strips have been used for many years to test for the presence of CSF. The test strips are positive at a relatively low level of glucose—greater than 20 mg/100 mL. Nasal secretion contains approximately 10 mg of glucose; however, nasal mucus and lachrymal gland secretions have reducing substances that may cause a positive reaction with glucose concentrations as low as 5 mg/dL. Hence, a negative glucose strip test eliminates the likelihood of a CSF leak, but a positive result cannot be interpreted.34 Furthermore, because blood contains glucose, the test must be conducted on clear discharge.
Laboratory Tests
β2-Transferrin
β2-Transferrin is a polypeptide involved in ferrous iron transport. β1-Transferrin is found in serum, tears, nasal secretions, and saliva. β2-Transferrin is found only in CSF, perilymph, and vitreous humor. It accounts for 15% of the total transferrin in CSF. It is important to determine the presence of penetrating eye injury before interpreting the β2-transferrin results. Quantitative detection of β2-transferrin can be performed on a small volume of fluid (<1 mL). The results of agarose gel electrophoresis are available in 3 hours, and it is the most sensitive and specific test to date.35,36 One study has indicated that there can be false-positive results.37 If this test is not available, glucose and chlorine concentrations help identify CSF.
Glucose Concentration
The glucose concentration can be assessed with 0.5 to 1 mL of fluid. If the glucose content is 0.5 to 0.67 that of the serum glucose concentration, the fluid is most likely CSF. It is important to take into account clinical conditions that alter the level of glucose in CSF and serum.34
Locating the Fistula
Computed Tomography
Fine-Cut Bone Window Scans
For examining the skull base, the standard format consists of fine cuts (0.6- to 1-mm intervals) performed on a helical multislice CT scanner at bone and soft tissue window levels. This protocol results in a box of data that can be sliced by the scanner in any plane. For best definition, the axis of scanning should be at right angles to the plane of the bone being examined. Thus, axial bone scans show the walls of the frontal sinuses, whereas coronal scans show the ethmoid complex, the roof of the sphenoid sinus, and the tegmen of the middle ear. The pattern of fracture lines and the presence of comminuted fractures, spikes, and wide fractures should be noted (Figs. 341-3 and 341-4).
Magnetic Resonance Imaging
Magnetic resonance imaging (MRI) has been reported to be a useful, noninvasive adjunct in the investigation of CSF rhinorrhea, particularly in the presence of inflammatory sinus disease.39 It may distinguish between mucosal disease with mucopurulent discharge and CSF, which may show the same radiopacity on a CT scan. On T2-weighted images, CSF will appear white and perimucosal discharge and nasal disease will be darker. Mucosal disease can be highlighted by the administration of gadolinium. MRI signs indicative of a CSF fistula include brain arachnoid hernia through the bone defect and a CSF signal in the perinasal sinuses that is continuous with intracranial CSF. A fistula may also be suspected if there is fluid in only one of the paranasal sinuses even though the fluid is not continuous with intracranial CSF. One study found that MRI was 100% successful and superior to CT cisternography in detecting active CSF leaks.40