Chapter 138 Management of Cerebrospinal Fluid Leaks
Cerebrospinal fluid (CSF) fistula is a serious and potentially fatal condition whose successful management requires a fundamental multidisciplinary approach. Until recently, the management of this condition was almost exclusively neurosurgical. Over the past decade, however, otolaryngologists skilled in functional endoscopic sinus surgery have contributed considerably to the surgical armamentarium, and the endoscopic approach is considered by some of those surgeons to have become the standard of care most specially in CSF leaks of the anterior and middle skull base.1 Although the endoscopic approach may be well-suited to the majority of cases encountered in otolaryngologic practice, it does not exhaust the techniques with which the neurosurgeon must be conversant in view of the full range of anatomic and pathophysiologic conditions confronted in neurosurgical practice. This chapter, therefore, addresses both the traditional neurosurgical approaches and the more recent endoscopic techniques. In addition, it reviews the use of glues, engineered tissues, and tissue substitutes in the management of CSF fistulas.
The correlation of posttraumatic rhinorrhea with leakage of CSF was made in the 17th century by a Dutch surgeon, Bidloo the Elder.2,3 Cases in which apparently nontraumatic CSF rhinorrhea resulted from increased intracranial pressure were then reported by Miller in 18264 and by King in 1834.5 The full significance of CSF fistulas was not elucidated until 1884, however, when Chiari6 demonstrated a fistulous connection between a pneumatocele in the frontal lobes and the ethmoid sinuses of a patient who died of meningitis following rhinorrhea. The introduction of roentgenography enabled the diagnosis of a fistula to be made in vivo through the detection of intracranial air,7 ultimately leading to the development of pneumoencephalography as a diagnostic procedure8 and, less directly, to the refinement of surgical techniques for the repair of CSF fistulas.9,10
Despite a number of early attempts, successful repair was not consistently achieved until the mid-1930s. In 1937, Cairns11 published a series of cases demonstrating that CSF leaks could be repaired by the extradural application of fascia lata. The actual need for surgical intervention was regarded as unproven, however, and the indications remained controversial until the latter part of World War II.
By 1944, Dandy12 advocated surgical repair of any CSF leak within 2 weeks of onset to prevent meningitis. In British neurosurgery, Lewin’s review of the British combat experience and of a large series of basilar skull fractures13,14 became the basis for the adoption of aggressive operative management as the standard of care. Lewin demonstrated that the cessation of an active leak did not itself eliminate the risk of meningitis, because the possibility of intermittent communication with the contaminated extracranial space persisted unless repaired. By the mid-1950s, it became virtually axiomatic to operate on all CSF fistulas, except for some that had a well-understood cause and that closed spontaneously within several days of onset.
The recent enthusiasm for extradural endoscopic approaches to the skull base reflects both the advancement of endoscopic technology and the relative safety of extradural techniques. In fact, Cairns and Dandy both advocated the extradural approach because of its safety. As intradural surgery became safer in the years following World War II, the extradural approach was supplanted, at least among neurosurgeons.15 Evidence upon which to base a consensus regarding the more successful operative route has yet to be determined.
The introduction of minimally invasive approaches to CSF fistulas, whether microscopic or endoscopic, has simultaneously facilitated the extradural approach and blurred the distinction between truly extradural and intradural repair.1,16–18
For purposes of this discussion, endoscopic repair is considered extradural when it is intended to patch or plug a fistula from outside in, and is considered intradural when it is intended to facilitate the localization or repair of a fistula by placing a plug or graft intradurally.19,20
The primary danger of CSF leakage has been framed in terms of the potential for meningitis, and the primary indication for treatment has been driven by the rarity of spontaneous and permanent cessation of the leak. Why should some leaks stop and others not? Why should some recur? Why should 20% or more of repairs that are done come to failure? With these questions in mind, the principle promoted by Lewin13 (i.e., the idea that all cranial CSF fistulas, without exception, be repaired surgically unless they resolve spontaneously within 5 days or a week) has come under increasing scrutiny. The observations that have led to a reconsideration of the Lewin principle are as follows:
Rhinorrhea, fluid dripping from the nose, and otorrhea, fluid dripping from the ear, are terms are intended to indicate the site of the drip, rather than the site of the fistula or the leak. CSF leaking through temporal bone fractures, for example, can easily reach the nasopharynx through the eustachian tube and mimic a leak through the cribriform plate. There is no specific term to describe leaks from the spinal subarachnoid space.
Transcranial CSF leaks fall into two major categories: traumatic leaks and the so-called spontaneous or nontraumatic leaks. The traumatic group, in turn, is divided into two groups: acute or early leaks that present within 1 week of injury and delayed leaks that occur months or years later. The nontraumatic group is also divided into subsets, including leaks associated with intracranial mass lesions, congenital defects of the skull base, osteomyelitis, osteonecrosis (and other causes of bony erosion), and focal cerebral atrophy5; and leaks associated with an ill-defined group of acquired hernias, meningoceles, and meningeal diverticula perforating pneumatized bone in the anteromedial middle fossa.30 Nontraumatic fistulas are divided again into high-pressure and low-pressure categories.5,31 There is good reason to apply an analogous nosology to the traumatic group as well.25 Iatrogenic or postoperative leaks are usually included in the category of traumatic fistulas.
Spinal CSF leaks can be classified similarly. Most spinal CSF leaks are postoperative and therefore traumatic. A number of rare congenital anomalies can give rise to meningopleural or meningoperitoneal fistulas.32 The distinction between high-pressure and low-pressure fistulas is particularly important in the management of spinal leaks. In children with spinal dysraphism or other anomalies, the leak may be the first expression of hydrocephalus or shunt failure.33,34
The most common cause of CSF leaks is head trauma, particularly basilar skull fracture.35 In Lewin’s series of 100 patients with head injury,13 7% had basal skull fractures, and 2% had CSF leaks. A CSF leak was detected in 2.8% of 1250 head injuries and in 11.5% of the basilar fractures studied by Brawley.22 In another study of 1077 skull fractures, including a particularly large proportion of high-speed road traffic accidents, 20.8% of 168 basilar skull fractures had an acute CSF leak.36 The association of incidence with high speed, although not rising to the level of a correlation, is certainly suggestive. The incidence in cases of penetrating missile injuries is comparable: in 1133 cases, 101 (8.9%) developed a CSF fistula. The proportion is somewhat higher with transventricular penetration.37 Traumatic CSF leaks typically begin within 48 hours, and it is estimated that 95% of them will be evident within 3 months of injury.38,39
In childhood, the incidence of traumatic CSF leaks is far lower at 1% or less of closed head injuries.40 This disparity may be caused by differences in fragility between the adult and the pediatric skull, as well as by the lack of development of the air sinuses in children. As a rule, the frontal sinuses become visible between the 4th and 12th year, and they are always detected by the 15th year. These sinuses are often asymmetric until age 20 years. The ethmoids are present at birth, enlarge by age 3 years, and are fully formed by age 16 or 17 years. The cavity of the sphenoid sinus is usually recognizable by age 4 years and is fully developed by puberty. In the pediatric age range, the interpretation of sinus x-ray studies is often difficult because of small size, variations in development, and normal calcification and clouding.35
This term should be restricted to leaks explained neither by trauma nor by any other cause and because there have been few, if any, collected series of nontraumatic leaks rigorously studied, there are insufficient data to extrapolate quantitative estimates of incidence or cause.31 Tumors and increased intracranial pressure (ICP) are highly correlated with nontraumatic leaks. Anecdotal series suggest that pituitary tumors are the most common neoplastic cause of spontaneous CSF leaks. Increased ICP may be present or absent. Because of the structures eroded by sellar masses, such leaks generally present as rhinorrhea.33,41,42 Other presentations, including a serous otitis media, have also been reported.43–45 On the other hand, there is an intriguing report of pituitary hyperemia in the context of nontraumatic CSF leak masquerading as pituitary adenoma in three patients. After surgical repair of the leak, the magnetic resonance imaging (MRI) abnormalities, including an enlarged pituitary resembling pituitary tumor, reverted to normal.45
Large and potentially dangerous subgaleal collections of CSF were common before the modern era of neurosurgery. In retrospect, they probably represented the most visible manifestations of altered postoperative CSF flow characteristics, increases in intracranial pressure, or unrecognized or untreated hydrocephalus, in some combination. These collections often leaked through the incision. In consequence, postoperative incisional CSF leaks were understood to be common neurosurgical complications. Attempts to prevent this complication were focused on careful dural closure, buttressing the suture line, multiple-layer incisional repair, and other techniques for reconstruction and reinforcement of violated tissue planes and compartments, and especially the air sinuses after frontal and posterior fossa surgery.
Although more radical approaches to cerebellopontine angle lesions, to tumors straddling the nasopharynx and the anterior and middle fossae, and to the skull base as a whole appear to have increased the prevalence of CSF leaks of all types, an accurate estimate of the incidence of incisional CSF leaks is difficult to provide. A recent study reports a 12% incidence (10 patients) in 85 posterior fossa procedures, but this figure may not pertain to other types of craniotomy.46
In large series CSF fistula occurs in 1.4% to 22% of operations for cerebellopontine angle tumor. The wide range may reflect the fact that it often proves necessary to report results encompassing many years and spanning important variations in technique in order to achieve significance.46–50 The incidence of leakage has been reduced by careful technique, including waxing and plugging mastoid air cells as they are opened and placing a graft of adipose tissue in the opened porus acusticus.51–54 The use of endoscopy to inspect the craniectomy site for unsealed air sinuses has also been advocated.55 Several recent series demonstrate that the incidence can be reduced below 11%, and that much lower rates are achievable, but that the incidence of leak seems relatively consistent irrespective of surgical approach (e.g., posterior fossa, transmastoid, or middle fossa).46,48–51,54 It is noteworthy that rhinorrhea, a classic false localizing sign in this setting, may be the presenting sign in up to 50% of leaks.46
In trans-sphenoidal approaches to the pituitary, leaks occur in 1.4% to 6.4%.56–58 In a small series reporting outcomes after the endoscopic transnasal trans-sphenoidal approach, the incidence was 14% (1/7).59
Endoscopic techniques were developed to improve visualization in the hope of reducing complications associated with blind instrumentation. It is not immediately obvious that the prevalence of CSF fistula has changed substantially, however. Estimates of incidence range from 0.002% to 2.9%.60,61 The latter figure was derived from a study explicitly designed to maximize the likelihood of diagnosing occult leaks using β-trace protein (prostaglandin D synthase) analysis and a 6-month follow-up.61
Intracranial air, a pathognomonic sign of CSF fistula after trauma or spontaneous rhinorrhea (but not pathognomonic after surgery), is demonstrable in approximately 20% of patients with CSF leaks.62 Pneumocephalus is post-traumatic in 75% of these patients and is spontaneous, or otherwise unexplained, in 10%.63
Meningitis occurs in approximately 20% of acute post-traumatic leaks and in 57% of delayed leaks.35 The incidence of meningitis in nontraumatic CSF leaks has not been well documented. Anecdotally, copious, continuous leakage of the high-pressure type is less likely to be associated with meningitis than is intermittent leakage.31 The overall risk of meningitis associated with traumatic CSF leaks of all types is on the order of 25%.1,14,64–67 In neurosurgical postoperative leaks, the incidence of meningitis can be calculated to be on the order of 20%, but this figure is admittedly complicated by the problem of distinguishing aseptic from bacterial meningitis, and by the complexity introduced by factors such as steroid administration, chronic illness, and immunosuppression that might impair wound healing and predispose to both leaks and meningitis. The high incidence of delayed infection in military penetrating head injuries is demonstrably correlated to CSF fistulas, actively leaking or not.68,69
The presence of glucose in clear leaking fluid has been used historically to differentiate CSF from nasal secretions and other sources of serous or serosanguinous drainage. The concentration of glucose in CSF equals or exceeds 50% of the serum concentration except as follows1: during meningitis,2 after subarachnoid hemorrhage, or under other unusual circumstances.3 The glucose concentration in nasal secretions, in contrast, is 10 mg/dl or less.70
Quantitative measurements of glucose concentration are diagnostic. Qualitative spot tests, such as those provided by chemical testing strips (e.g., Clinistix, Dextrostix, Uristix, or Tes-Tape), are not definitive for two reasons.1 First, the glucose oxidase test on which they are based is too sensitive, turning positive at values less than 20 mg/100 ml of glucose.2 Second, normal nasopharyngeal secretions often elicit false-positive reactions even in the absence of glucose.71,72 Thus, although a negative glucose oxidase reaction effectively eliminates the possibility of CSF rhinorrhea, a positive result does not diagnose it unequivocally.
It is widely held that true CSF leaks produce quantities of fluid sufficient for collection and quantitative analysis at some time in their course. The reservoir sign, the ability of a patient to produce CSF at will by positioning the head in a certain way, is generally taken to be quite specific for a fistula with pooling in the sphenoid sinus.30 Although Dandy12 believed that this sign would differentiate leakage through the frontal sinus from ethmoidal and sphenoidal leaks, it is not reliably localizing.
The target sign refers to the pseudochromatographic pattern produced by the differential diffusion of CSF admixed with blood or other serosanguinous fluid on filter paper or bedclothes. CSF migrates further, creating a bull’s-eye stain with blood in the center. This is a convenient but unreliable sign, because whenever watery nasal secretions and blood are mixed, the same phenomenon occurs.
CSF leaks can be accompanied by high-pressure or low-pressure headaches. Intermittent high-pressure leaks are characterized by high CSF pressure headaches that are relieved by the sudden discharge of fluid and that build up again over time. Normal-pressure leaks, in contrast, are characterized by postural low CSF pressure headaches, relieved by reclining or otherwise allowing pressure in the subarachnoid space to rise to normal levels.
The finding of unusually low opening pressure in the lumbar subarachnoid space is corroborative evidence for CSF leak. Unilateral or bilateral anosmia is associated with defects or leaks in the region of the cribriform plate and the fovea ethmoidalis. Olfaction may be preserved, however, in cases of spontaneous CSF rhinorrhea with congenital defects of the cribriform fossa.31,64 Optic nerve lesions point to the tuberculum sella, the sphenoid sinus, and the posterior ethmoids as the likely site of injury. Impaired vestibular function, facial nerve palsy, and cochlear damage accompany fractures in the temporal bone.
Imaging techniques are used to detect intracranial air, fractures and defects in the skull base, mass lesions, and hydrocephalus, and to demonstrate flow through fistulas or skull defects.73 Plain films, multiplanar tomography, computed tomography (CT), and MRI have been used to delineate the anatomy and pathology of the skull base, sinuses, and calvaria. Radiographic data must be interpreted in the context of clinical findings. As always, positive data obtained from radiography are helpful, but negative data are often meaningless.
Plain films and CT are examined for evidence of fracture; air/fluid levels in the frontal, ethmoidal, and sphenoid sinuses; intracranial air; chronic increased intracranial pressure; erosion of bone by tumor or infection; congenital anomalies; and penetrating objects. Although multiplanar tomography provided exquisite detail of bony anatomy, it is of historical interest only, having been supplanted by high-resolution thin-slice CT with overlapping cuts.74 Contrast cisternography in conjunction with CT provides dynamic information about flow patterns of CSF.75 Similar information is also obtainable from MRI in addition to providing superb detail of soft tissue pathology at the skull base and in the nasopharynx.76
The categorical proof of CSF fistula is the ability to retrieve extracranially a tracer substance injected into the CSF. The nonradioactive substances injected into the CSF historically have included methylene blue, phenolsulfonphthalein, indigo carmine, and fluorescein.77,78 Only indigo carmine and fluorescein remain in common use for intraoperative visualization: the others proved unacceptably toxic.5
In the presence of an active leak, cotton pledgets placed along the anterior roof of the nose, the posterior roof and the sphenoethmoid recess, and the middle meatus and below the posterior end of the inferior turbinate can be used to confirm that a leak exists. When differentially stained or contaminated by radioactive isotopes, they indicate the location of the leak.29 The following procedure has been recommended to localize a leak with fluorescein:
|Location of Stain||Probable Site of Fistula|
|Anterior nasal||Cribriform plate or anterior ethmoidal roof|
|Posterior nasal or sphenoethmoidal||Posterior ethmoid or sphenoid sinus|
|Middle meatus||Frontal sinus|
|Below posterior end of inferior turbinate||Eustachian tube (middle fossa)|
|Behind tympanic membrane||Not accurately predicted|
Caveat: Fluorescein has been associated with severe adverse reactions in this application; see subsequent discussion.
The use of intrathecal fluorescein injection is quite controversial. Although otolaryngologists continue to recommend and utilize it routinely, neurosurgeons have become wary of its use because of reports of transverse myelitis and other serious adverse events such as seizures.79 Indigo carmine may be preferred by some neurosurgeons, not only because of its safety record but also because it is more visible than fluorescein to the unaided eye. This characteristic makes it useful to check for the presence of CSF fistulas intraoperatively, when ultraviolet illumination may not be readily available.30
Radioactive tracers such as 131I (RISA) were widely used for cisternography in the past but have most recently been replaced by 111In DTPA, an isobaric tracer that combines improved physical properties, fewer adverse reactions, better imaging quality, and shorter half-life (2.8 days). 169Yb DTPA and 99mTc albumin have also been approved for CSF imaging, but these tracers suffer from suboptimal imaging characteristics and half-lives (32 days and 6 hours).
Isotope cisternography is an effective method by which to demonstrate the existence of a CSF leak, but it becomes inaccurate for purposes of localization when “flooded” by a high-volume leak. The tracer saturates the pledgets and contaminates surrounding tissues with radioactivity so that differential localization becomes impossible.35,80,81
The importance of the timing of radioactive contamination is insufficiently appreciated. Active leaks can be documented by contamination of accurately placed pledgets within 0.5 to 2 hours. Slow or intermittent leaks can sometimes be detected by leaving the pledgets in longer or by replacing them over 6 to 48 hours. The danger lies in overinterpretation or misinterpretation of the data. There are two confounding mechanisms. First, the isotope can be absorbed into the bloodstream from the CSF and undergo secondary secretion into the nasopharynx.82 Second, there exists an alternate pathway for isotope secretion, first recognized in normal dogs and subsequently documented in normal human volunteers, involving active transport from the CSF and passage via the olfactory nerves, or passive lymphatic drainage leading to contamination of nasopharyngeal secretions.83
The accuracy of faintly positive tests can be improved slightly by calculating a radioactivity index (RI) ratio. This ratio compares the radioactivity in counts per minute of an exposed pledget with that of 1 ml of blood, as follows:
An RI ratio less than 0.3 is normal. Canine studies suggest that the ratio is at least five times greater in the presence of a leak. In the canine model, the RI ratio of nasal activity to CSF is 1:14 when measured 2 to 5 hours after cisternal injection of radioisotope. The RI ratio of nasal activity to blood is 1:2 to 3 when measured 2 to 5 hours after intravenous injection of radioisotope. Thus the RI ratio of CSF to blood is 4.6:7. The significance of tracer substances that appear in low concentrations only 5 hours or more after injection should be carefully weighed.
Two additional points deserve mention. First, it has been suggested that tracer injected into the cervical subarachnoid cistern can be forced through a slow, intermittent, or low-pressure leak by raising the CSF pressure with saline or artificial CSF delivered via a constant infusion pump.84 Second, scans carried out 24 and 48 hours after injection of isotopes can help define the mechanism of a leak by detecting defects in CSF absorption and circulation.
Early attempts at demonstrating CSF leaks by injecting air or iophendylate (Pantopaque) into the subarachnoid space failed to produce consistently satisfactory images. The combination of high-resolution CT and metrizamide, or of other form of cisternography with water-soluble contrast agents, yields excellent visualization of active leaks. Smaller fistulas have been demonstrated through stressing the barriers to CSF flow by having the patient cough or carry out a Valsalva maneuver.35 For maximal contrast enhancement, cisternal injections can be performed via C1–C2 punctures. Overlapping views in both the coronal and the axial planes are required.74 Direct coronal studies are preferable to reconstructed images. The risk of provoking seizures and aseptic or chemical meningitis following cisternography should be discussed with the patient.
A newer and safer technique utilizes MRI technologies to visualize CSF flow.76,85–88 A full discussion of the technique is outside the scope of this chapter, but heavily T2-weighted fast spin–echo studies with fat suppression and video reversal of the images are reported to yield sensitivity, specificity, and accuracy of 0.87, 0.57, and 0.78, respectively, in a study in which 65% of patients eventually underwent surgical exploration.86
Immunologic methods differentiate between proteins in CSF and those in nasopharyngeal secretions.95,96 Irjala and colleagues97 have described the use of an immunofixation technique for the identification of microaliquots (100 μl) of CSF by demonstrating two electrophoretically characteristic bands of transferrin. The B1 fraction consists of normal transferrin and sometimes two variant fractions. The B2 fraction characteristic of CSF contains smaller amounts of neuraminic acid. This method is not subject to contamination from other body fluids (e.g., tears or nasal secretions). The immunofixation method could theoretically be used for localization of the leak by differential suction techniques in the nasopharynx, but large-scale clinical trials of this method have yet to be reported. Another surrogate marker for CSF is β-trace protein (prostaglandin D synthase), as reported by Arrer and colleagues.98 The β-trace protein test is reported to offer an overall accuracy of 95.7%, a specificity of near 100%, and a sensitivity of 91.2%.99
CSF leak can occur wherever the dura is lacerated during an injury. It is more likely to persist or recur, rather than close spontaneously, where a meningeal hiatus is maintained by bony spicules, by dura entrapped in the edges of a fracture, or by herniating brain and leptomeninges. Avulsion of the olfactory fibrils can result in a dural fistula through the cribriform plate even without a fracture, particularly in the elderly or where the tissue around the ethmoidal roof has been thinned.
Nontraumatic leaks are usually confined to a single region where an anatomic defect is demonstrable, but exceptions have been noted.100 It is usually easier to demonstrate the defect than the leak. High-pressure leaks that act as safety valves for hydrocephalus occur where the skull is thinnest, usually the cribriform fossa and the sellar region. This is the case, for example, in Crouzon’s disease and osteopetrosis (Albers-Schönberg disease).
The middle fossa can be the site of CSF leaks that are direct in that they do not cross the inner ear. Such fistulas have been described mainly in conjunction with a pneumatized temporal fossa. Pulsatile CSF forces induce additional thinning of the bone and enlargement of the pits and small bony defects that are normally present. The leptomeninges and brain herniate, thinning the dura and leading to rupture of the arachnoid. The leak may be constant or intermittent depending on several factors, such as the underlying intracranial pressure, whether an arachnoid diverticulum is created, and whether brain tissue temporarily obliterates the leak. A similar sequence of events has been postulated to explain CSF leaks in the empty sella syndrome30 and in focal atrophy.31
Indirect fistulas through the temporal bone are the most elusive. In extralabyrinthine fistulas the defect is in the middle fossa in the region of the tegmen tympani. In intralabyrinthine fistulas, CSF escapes into the labyrinth through the subarachnoid space of the posterior fossa. In either case, the leak can present as otorrhea or, when the tympanic membrane is intact, as rhinorrhea.62 The possibility of temporal bone dysplasia should be investigated whenever a patient with severe hearing loss develops unexplained or recurrent meningitis.30,35,101 For example, in the Mondini malformation (i.e., unreactive ear with a shortened cochlear coil, dilated semicircular canal system, and widened inner ear vestibule), it is hypothesized that a widened, patent cochlear aqueduct allows CSF to pass from the subarachnoid space to the inner ear via a leak in the oval window. Other proposed routes include defects of the scala tympani; of the footplate of the stapes; or of the thin, bony plate separating the internal auditory canal and the inner ear vestibule and perforated by nerve fibers innervating the utricular and saccular maculas.102–104
When a CSF leak is the manifestation of increased intracranial pressure from mass effect or from hydrocephalus, the underlying cause must be treated before the leak can be effectively repaired. The existence of increased intracranial pressure can be inferred from several sources. Skull films and CT scans disclose signs of pressure, mass effect, and tumors. Radiographic signs suggestive of defective circulation and absorption of CSF include periventricular lucencies, enlarged temporal horns, a disproportionately plump third ventricle narrowed at the massa intermedia, and an empty sella. Extracerebral collections of CSF can represent the so-called fifth ventricle phenomenon, a hydrocephalus variant. Other concomitants of high-pressure leakage include papilledema and optic atrophy; pallor of the optic disc; enlarged central scotoma or subtle binasal visual field cuts; a history of headaches worse in the morning or while recumbent, relieved by a gush of fluid; variable or intermittent diplopia; intermittent clonus or pyramidal tract signs reversing spontaneously after leakage; and a history of granulomatous meningitis, subarachnoid hemorrhage, or head trauma.
Infants with incisional leaks after fresh meningomyelocele repair can be assumed to have hydrocephalus, especially if there is an accompanying Dandy-Walker malformation. In contrast to other cases of postoperative leak, in which the temporizing maneuvers discussed subsequently are often effective, CSF shunting is generally required.
In traumatic leaks, antibiotics have not been proven effective in changing the incidence of meningitis and are no longer recommended routinely.15–17,21,36,105–109 For postoperative leaks, however, prophylactic antibiotics are commonly, if not universally, employed. There is some theoretical justification for distinguishing between the two situations. Several principles are useful to keep in mind when prescribing prophylactic antibiotics108,110:
3. Patients of different ages and in different locations harbor different vulnerabilities because of changes in nasopharyngeal and environmental flora; thus, Haemophilus influenzae is a common cause of meningitis in children and in the elderly, whereas diplococcus is more common in healthy adults.
4. Patients can develop meningitis even while on prophylactic antibiotics; after the usual investigations are carried out, the antibiotics are changed to cover the appropriate organisms and sensitivities.
External drainage of CSF has been used in various forms for many years. External ventricular drainage111,112 has been replaced in most centers by continuous lumbar drainage, first described in 1963.113 Since then, lumbar drainage has been found useful in controlling and sometimes curing CSF leak of every cause.23–25 McCoy114 provided theoretical justification for the initial management of CSF fistulas with CSF diversion by demonstrating that granulation can seal the fistulas, provided that the leakage has stopped. Lumbar drainage should be considered, therefore, whenever positioning alone does not eliminate, or at least significantly diminish, a leak within 24 hours.
A 19-gauge catheter is threaded percutaneously through a 17-gauge Touhy needle inserted into the lumbar subarachnoid space between L4–L5 and L2–L3. Aside from increased attention to sterile technique, there is no difference from standard lumbar puncture procedure. After 10 to 20 cm has been threaded rostrally, the needle is removed over the catheter. If there is any significant resistance to passage, the needle and the catheter should be removed as a unit. Under no circumstances should the catheter be withdrawn through the needle once the tip has protruded, because the needle tip may shear the catheter, leaving the tip irretrievably lost in the subarachnoid space or in the subcutaneous tissue. The proximal end of the catheter is connected via the appropriate adapters to a closed, sterile drainage system. A waterproof, occlusive dressing should be applied, with the catheter coiled and taped to relieve strain and to prevent disconnection.
The infection rate with indwelling catheters ranges in some series as high as 10% or more.115 The risk is lower with lumbar catheters.26 Infection can be reduced by prophylactic antibiotics (potentially, by two thirds115) and by externalization of the catheter through an extended subcutaneous tunnel.116
Prophylaxis with antibiotics is continued for 8 to 24 hours after the catheter is withdrawn. Daily samples of CSF are obtained; cultured; examined by Gram stain; and analyzed for cell count and differential, glucose, and protein. The presence of a catheter does not, of its own accord, lower the CSF glucose or evoke a major leukocytotic reaction; the cell count and glucose concentration remain quite stable in uninfected CSF over 4 to 9 days. Any persisting variation of 2 standard deviations or more from the cumulative average cell count and glucose concentration over several days is cause for concern and careful reexamination of the CSF for signs of opportunistic infection.117 So too would be the emergence of any clinical signs or symptoms of meningeal irritation.
External drainage has been maintained in large series for 10 days without infection. Longer durations have been reported in exceptional cases.25 By analogy with central venous access lines, it may be wise to change catheters if drainage is continued beyond 7 days.
The data on which these recommendations are made are empirically derived. Drainage should be continued for 3 to 5 days after stoppage of the leak to allow healing. If leakage recurs, operative repair is indicated. If the underlying problem is increased intracranial pressure or hydrocephalus, implantation of a drain acts purely as a temporizing maneuver: no “cure” is effected. Similarly, the patient whose leak is not controlled by external drainage should be considered for early operation. In Findler’s series of 50 patients,29 drainage of 350 to 420 ml daily was continued for an average of 10 days, with a leak recurrence rate of 14%. There was an additional 8% incidence of delayed leak at the site of lumbar puncture.
As a rule, acute posttraumatic and postoperative normal-pressure leaks respond to external drainage. Transitory high-pressure leaks also respond, so long as the pressure elevation recedes over the duration of the drainage. Delayed and recurring leaks cannot be definitively managed by drainage.
High CSF protein concentrations predispose against a successful drainage. If, for technical reasons, patency of the drainage catheter cannot be maintained, repetitive lumbar punctures through a large needle (18 gauge) often afford almost the same benefit.
Calcaterra18 records one case of fatal postoperative suboccipital hemorrhage attributed to overdrainage of CSF in an elderly patient. Similar complications have followed spinal anesthesia.119–121 Overdrainage of CSF can also cause life-threatening pneumocephalus.64,122–124 The CSF pressure should be lowered, and may even be lowered substantially, but should not be reduced to less than 0 through negative pressure.
The acute reduction of CSF pressure can also precipitate headache, nausea, and vomiting. This reaction can be prevented, according to Findler and colleagues,29 by gradually lowering the pressure and increasing the drainage over several days. An accidental siphon effect can be avoided by relating the height of the drainage valve or the drainage bag to the level of the ventricular system rather than the bed. In this way, the pressure column remains constant as the bed is raised and lowered or as the patient is moved. Most commercial systems incorporate a micropore-filtered air port to prevent siphonage. Improvised systems are generally unable to include such a port, and positioning becomes critical.
There has been long-standing concern regarding the possibility of inducing meningitis by retrograde migration of bacteria into an open fistula under the influence of negative CSF pressure induced by an external drain.5 This seems to be an extremely rare complication, avoidable by maintaining a low but steady positive pressure in the CSF.
Catheters should not be removed forcibly. After removal of the catheter, the tip is examined to ensure that no part of it has been left behind. If the catheter is not intact, an effort should be made to locate and identify the retained segment with an x-ray or a CT.
There are two strong indications for retrieval of a broken catheter tip: (1) infection in the region of the tip, and (2) radicular pain or paresis associated with juxtaposition of the retained catheter to an appropriate root. In most cases, the retained fragment can be ignored safely.
Dural-cutaneous fistulas can occur at the site of catheter insertion, particularly in the setting of a high-pressure leak. Most such fistulas stop spontaneously or seal with a single stitch. Low-pressure or normal-pressure leaks can also be sealed by an injection of 10 to 20 ml of autologous blood as an epidural blood patch. The technique has been shown by a number of studies to improve over natural history125–129 with a success rate of 93%.130,131
Epidural blood patching has resulted in symptomatic mass effect, hemorrhagic complications,132 and infection. Surgical repair of the dura may still be needed. External lumbar drainage is contraindicated in the context of increased intracranial pressure from a mass in the posterior fossa because of the danger of precipitating herniation through the foramen magnum.