Epidemiology

Cerebrospinal fluid (CSF) leaks are increasingly considered an important cause of low-pressure headaches, particularly among younger patients [1]. While rigorous epidemiological evidence is lacking, several series have documented the relative incidence of CSF leak from varying etiologies. Approximately 80% of CSF fistulae with rhinorrhea have been reported to be caused by nonsurgical trauma and 16% by surgical trauma, with the remaining 4% having nontraumatic etiology [2]. Of course, for patients with spontaneous intracranial hypotension, the underlying CSF leak should be looked for in the spine and not at the level of the skull base [3]. Spontaneous intracranial hypotension has been estimated to range between 1 per 50 000 to 5 per 100 000 persons in community-based and emergency department based settings, respectively [4,5]. Spontaneous intracranial hypotension affects women twice as frequently as men and typically presents in the fourth or fifth decade of life, though the specific pathophysiological reason for this demographic trend is currently unknown [6–16].

Pathogenesis

The specific etiology underlying spontaneous CSF leaks remains largely unknown, but underlying structural weakness of the spinal meninges is suspected, though a history of trauma may be elicited in up to a third of patients [14,17–19]. This dural weakness, ranging from simple dural tears to meningeal diverticula, allows CSF to leak into the epidural [20,21]. Ventral spinal CSF leaks are often seen at the level of the disk space and may be associated with degenerative disk disease.

Varying connective tissue disorders may play a role in the development of spontaneous spinal CSF leaks, with two-thirds of patients having some physical exam features consistent with such disorders, inclusive of joint hypermobility or subtle skeletal manifestations of Marfan syndrome [22]. These patients with some skeletal manifestations consistent with Marfan syndrome have been found to have a defect in microfibrils, an important component of the extracellular matrix associated with fibrillin, rather than a mutation in fibrillin-1 itself that is characteristic of Marfan syndrome [23]. Less frequently, spontaneous intracranial hypotension has occurred in well-described connective tissue disorders including Ehlers–Danlos syndrome type II, autosomal dominant polycystic kidney disease, neurofibromatosis, Lehman syndrome, and true Marfan syndrome [24–28].

Some authors have suggested that spontaneous intracranial hypotension results from decreased CSF secretion or generalized CSF hyperabsorption, though no significant data support such alternate mechanisms. It seems that the final common pathway is likely not CSF hypovolemia but rather altered distribution of craniospinal elasticity due to spinal loss of CSF [29].

Clinical presentation

Diagnosis

Cranial magnetic resonance imaging

While some findings of CSF leak in MRI can be variable, there are generally five characteristic imaging features most often associated with spontaneous intracranial hypotension: subdural fluid collections, enhancement of the meninges, engorgement of venous structures, pituitary hyperemia, and sagging of the brain (mnemonic, SEEPS) [1] (Figure 22.1).

Figure 22.1 Pretreatment and posttreatment magnetic resonance imaging. (a) Axial FLAIR imaging shows resolution of bilateral subdural hematomas (black arrowheads) and engorgement of the superior sagittal sinus (white arrow). (b) Axial post-gadolinium contrast, T1-weighted imaging shows resolution of meningeal enhancement (black arrowheads). (c) Sagittal T1-weighted imaging without contrast shows resolution of brain sag inclusive of less flattening of the pons/effacement of the prepontine cistern (black arrowheads), less displacement of the cerebellar tonsils within the posterior fossa (black arrow), and less hyperemia of the pituitary gland (white arrows). Used with permission from Schievink WI. Spontaneous spinal cerebrospinal fluid leaks and intracranial hypotension. JAMA 2006;295:2286–96.

Subdural fluid collections are common in spontaneous intracranial hypotension, are thought to be caused by tearing of subdural bridging veins, and occur in approximately 50% of such patients [41]. Most of these collections are bilateral hygromas located over convexities and do not cause any significant mass effect or midline shift. True subdural hematomas are less common, affecting approximately one-quarter of all patients [41]. Most subdural hematomas in this setting can be managed by primarily treating the underlying spinal CSF leak, either with treatment via interventional radiology or with open spinal surgery; only a minority of patients required craniotomy for direct surgical evacuation of blood [41]. Improvement in this MRI finding can be expected within days to weeks of successful treatment for small subdural collections, while larger subdural hematomas may require up to 3 months to fully resolve. Clinical improvement generally precedes radiographic improvement in most spontaneous intracranial hypotension patients.

Diffuse enhancement of both the supra- and infratentorial meninges is the best-known imaging finding in spontaneous intracranial hypotension, with the dilation of thin-walled blood vessels of the subdural zone thought to be the mechanism for contrast enhancement. However, up to 20% of patients do not develop enhancement on MRI [42].

Engorgement of venous structures is also a common finding in spontaneous intracranial hypotension, with findings most evident in the large cerebral veins and/or dural sinus. Such findings may be relatively subtle and may require the comparison of post- versus pre-gadolinium images to help identify them.

Pituitary hyperemia has additionally been described as an often cardinal feature of spontaneous intracranial hypotension. The relative hyperemia in such cases may at times mimic a pituitary adenoma or hyperplasia, and requires some clinical correlation of radiographic findings.

Sagging of the brain is thought to be due to loss of CSF buoyancy and may be accompanied by ventricular collapse. This brain sag is associated with several key features including effacement of perichiasmatic cisterns, bowing of the optic chiasm, flatting of the pituitary stalk, effacement of the prepontine cisterns, flattening of the pons, and descent of the cerebellar tonsils within the posterior fossa. This latter feature may be mistaken for a Chiari malformation. As noted above, subdural fluid collection may be seen, though the degree of brain sagging is generally out of proportion to any mass effect caused by these subdural collections.

Spinal MRI

Although spinal MRI has not been as effective at localizing CSF leaks as CT myelography, more recent work has demonstrated some specific spinal manifestations of spontaneous intracranial imaging in spinal MR imaging. Such findings include dilated epidural and intradural veins, dural enhancement, and meningeal diverticula [46,47]. MR myelography is particularly well-suited to demonstrating spinal meningeal diverticula and dural ectasia. Also, plain MRI is an excellent method of demonstrating the extent of ventral spinal CSF leaks (Figure 22.2).

Figure 22.2 MR myelography demonstrating various examples of dural ectasia and meningeal diverticula in patients with spontaneous intracranial hypotension (normal scan is on the left).

Diagnostic criteria

Several iterations of diagnostic criteria for spontaneous intracranial hypotension have been developed over the past decade, with some variations in inclusion and exclusion criteria used including varying clinical symptoms and radiographic evidence. Perhaps the most widely cited set of diagnostic criteria are those established by the Headache Classification Subcommittee of the International Headache Society [48]. This set of criteria included a positional headache with at least one additional clinical symptom, at least one radiographic or diagnostic criterion, no history of dural puncture or other cause of CSF fistula, and a headache that resolves within 72 hours of epidural blood patch, as outlined in Table 22.1.

Table 22.1 Diagnostic criteria for headache due to spontaneous spinal CSF leak and intracranial hypotension as defined by the International Classification of Headache Disorders, 2nd Edition (2004) [48]

A more recently proposed classification scheme allows for a three-step approach to the diagnosis. For instance, patients may either have a demonstrated spinal CSF leak (for instance, via CT or MR myelography) or may fulfill one of two separate sets of criteria. The first of these two sets of criteria include having MR brain imaging findings consistent with CSF leak in addition to having either a low opening pressure (less than or equal to 60 mmH₂O), the presence of spinal meningeal diverticula, or improvement in symptoms following epidural blood patch. The second of these two sets of criteria are for those without a demonstrated spinal CSF leak or MR brain findings consistent with a leak; such patients may meet the criteria for diagnosis by meeting all three of the additional criteria noted above (low opening pressure, spinal meningeal diverticula, and improvement in symptoms after blood patch) or two of such criteria in addition to orthostatic headaches [49]. Other groups have generally seemed to agree with this newer set of diagnostic criteria, while pointing out the potential need to define specific criteria for measuring spinal meningeal diverticula and the potential need to define a 24- to 72- hour time window to allow for symptomatic improvement following a blood patch [50].

Most recently, a multi-institutional group of experts have proposed a simplified algorithm for diagnosis in which any patient with orthostatic headaches, no recent history of dural puncture, and headaches not attributable to another disorder can meet criteria for spontaneous intracranial hypotension with the addition of only one of the following additional criteria: low opening pressure (less than or equal to 60 mmH₂O), sustained improvement following blood patch, demonstration of active spinal CSF leak, or cranial MR changes consistent with intracranial hypotension [51]. Rigorous work analyzing the relative efficacy of these varying diagnostic criteria has yet to be undertaken.

Treatment, outcomes, and future directions

Although data are lacking, it is often stated that many cases of spontaneous intracranial hypotension resolve spontaneously without any intervention and with only conservative management at most. Conservative management has traditionally consisted of bed rest, oral hydration, and use of an abdominal binder. While this conservative management is anecdotally felt to help in most, a subset of patients may have persistent debilitating symptoms or may desire more timely resolution of symptoms. For such patients, additional interventions may include the use of steroids, intravenous caffeine, or theophylline, though the efficacy of such interventions seems limited at best.

For those patients who continue to have significant symptoms despite the conservative or relatively noninvasive approaches noted above, the mainstay of first-line interventional treatment has been the injection of 10–20 ml of autologous blood into the spinal epidural space, otherwise known as an epidural blood patch [52,53]. For those undergoing this procedure, the relief of symptoms often is instantaneous, thought to be due to a combination of replacing CSF volume loss with blood volume within the spinal canal, due to the blood patch forming a type of dural tamponade thus sealing the leak, and due to the blood restricting CSF flow within the spinal epidural space thereby interfering with CSF absorption. Although such a blood patch is effective in relieving symptoms in up to one-third of patients, those with residual significant symptoms may either have the 10–20 ml blood patch repeated or may alternatively be offered a high-volume (20–100 ml) blood patch. The largest volume we have ever given was 130 ml. Because patients with spontaneous intracranial hypotension have such a large epidural space due to the loss of intrathecal CSF, such a large volume can be accommodated. Given the high volume of such patches, a minimum of 5 days is usually advised between a regular and high-volume blood patch.

For those patients in which epidural blood patch fails, either percutaneous placement of fibrin sealant or an open/direct epidural blood patch may be advised. Generally, either of these additional approaches would necessitate that the exact site of the CSF leak be known in order to maximize the odds of successful intervention. Attempting percutaneous fibrin sealant placement remains a viable non-operative therapeutic option and may be beneficial in up to one-third of patients in whom an epidural blood patch has not been effective [54].

Surgical treatment is reserved for those in whom these nonsurgical interventions such as blood patch or fibrin glue have failed and remains a viable option for those in whom a focal structural abnormality leading to CSF leak can be identified. Leaking meningeal diverticula can be ligated with either suture or aneurysm clips, while dural tears can be repaired either directly with suture or with placement of a muscle pledget and fibrin sealant. Ventral dural tears (Figure 22.3) are best treated with suturing although placement of a dural substitute along with fibrin sealant or placement of a sling may also be effective. Headache following successful percutaneous or open treatment may indicate either a recurrent leak or rebound transient intracranial hypertension. The pattern of headaches post-intervention may help elucidate the etiology of such headaches. For instance, if the character of the headaches remains similar to initial presentation, including orthostatic headaches, then likelihood of recurrent leak may be relatively increased. However, if the patient experiences worse headaches when supine, so-called pressure headaches, the likelihood of rebound transient intracranial hypertension may be relatively increased and such patients may benefit from at least a short course of acetazolamide [55].

Figure 22.3 Intraoperative picture of a ventral dural tear in a patient with spontaneous intracranial hypotension.

Overall, outcomes are good for the great majority of patients with spontaneous intracranial hypotension, but some patients are refractory to any treatment and innovative but unproven therapies should be explored [56]. Also, orthostatic headaches can be due to various other ailments and there are some patients who insist that they have a CSF leak even though extensive investigations fail to support such a diagnosis.

Much needs to be learned about the enigma of spontaneous intracranial hypotension. Radiologic methods to define a CSF leak in patients with multiple spinal meningeal diverticula but no demonstrable CSF leak are needed, a noninvasive therapy for closure of ventral dural tears is a priority, and methods of strengthening the dura need to be investigated.