Neuro-Ophthalmology & Neuro-Otology
Aug. 17, 2022
At vero eos et accusamus et iusto odio dignissimos ducimus qui blanditiis praesentium voluptatum deleniti atque corrupti quos dolores et quas.
In this article, the author reviews new concepts in spontaneous spinal CSF leak. She discusses the role of CSF volume depletion and addresses the substantial variability in the clinical aspects, imaging findings, and CSF abnormalities related to this disorder. Preexisting dural defects, likely related to disorders of the connective tissue matrix, are sometimes seen and contribute to the often complex anatomy of these spontaneous leaks. These “spontaneous” defects present therapeutic challenges far beyond the expected management of post-lumbar puncture or even post-epidural catheterization CSF leaks. An update with special emphasis on recent theories on orthostatic headache pathogenesis and advances of the imaging techniques to localize CSF leaks and treatment options is provided. The author also briefly discusses posttraumatic CSF leak.
• Spontaneous spinal CSF leak is characterized by an orthostatic headache often accompanied by neck stiffness, tinnitus, hypacusia, photophobia, or nausea. However, leaks can also present with a “second-half-of-the-day” headache and, rarely, no headache.
• The spinal location of spontaneous spinal CSF leak is important. Its pathophysiology is thought to be linked to reduced spinal dural compliance, resulting in a downward shift of the hydrostatic indifferent point and intracranial hypotension.
• The 2 main imaging tools used to localize spinal CSF leaks are fat suppressed heavily T2-weighted magnetic resonance myelography and computed tomographic (CT) myelography.
• The treatment of choice is an autologous epidural blood patch, preferably delivered at the level of the spinal CSF leak. Fibrin sealant injection and surgical repair are reserved for intractable cases.
The history of how spontaneous spinal CSF leaks have been described over the years reflects clinical controversy regarding the underlying pathophysiology of this disorder.
Schaltenbrand, a German neurologist, introduced the term “aliquorrhea,” meaning “lacking” or “absence of” to describe a disorder associated with low, unobtainable, or even negative CSF pressures. He noted that clinically, the disorder was marked by orthostatic headache and other features (105). This description provided the first understanding of what is now known as spontaneous spinal CSF leak.
From the 1960s to the 1990s, with the advent of radionuclide cisternography (44; 64), conventional myelography, and MRI, CSF leaks and CSF dynamics could finally be studied (103; 53; 43; 93). Imaging findings, such as venous sinus engorgement, evoked the Monroe-Kellie doctrine, suggesting that changes in these structures occurred to compensate for lack of CSF in the brain space. For this reason, in the late 1990s, spontaneous spinal CSF leak was called “CSF hypovolemia” or “CSF volume depletion” (80). However, patients were also noted to have low CSF pressure, and a competing term “spontaneous intracranial hypotension” emerged (25; 77). No data has ultimately supported the theory of low volume or even consistently low CSF pressure. Even the concept of the Monroe-Kellie doctrine, or the constancy of brain volume, has been challenged; a study revealed that there was a decrease in brain tissue volume in patients with spontaneous intracranial hypotension (149).
Therefore, the current preferred descriptive term is “spontaneous spinal CSF leak” (108). This term captures the importance of the CSF leak being spontaneous, or occurring with minimal or no clear precipitant. The word “spontaneous” is important because patients with spontaneous leaks can have a very different prognosis and workup than patients with iatrogenic and traumatic CSF leaks. The term also specifies that the CSF leak should be from a spinal source. This term is important because the majority of people with leaks from the cranium, ie, skull-based leaks, rarely, if ever, develop symptoms or brain imaging findings consistent with spontaneous spinal CSF leak (108; 118).
An orthostatic headache that develops while sitting or standing and is relieved by assuming a horizontal position is the cardinal symptom of intracranial hypotension (108; 28).
However, it is important to note that there is marked variability in headache characteristics. The onset of headache ranges from occurring immediately, within seconds to minutes of assuming a vertical position, to a delayed worsening after minutes or hours of being upright (28). The duration of the symptoms can affect how the symptoms present. A study showed that 93.1% of patients with less than 10 weeks of symptoms displayed typical orthostatic headache, whereas only 62.5% with more than 10 weeks of symptoms did (p = 0.004) (52).
The headache onset may be acute, thunderclap-like and may mimic subarachnoid hemorrhage (Schievink et al 2001; 38). In other patients, a “second-half-of-the-day” headache may be seen. These patients are typically headache free in the morning, but by late morning or early afternoon, an increasing headache develops if the patient continues to be up and about. These headaches may or may not have clear orthostatic features (84; 68).
Similarly, the onset of improvement after recumbency varies. Onset of symptom improvement, although not necessarily symptom resolution, ranges from minutes to hours of assuming a horizontal position (28).
The quality of head pain can be described as either throbbing or non-throbbing. The location may be frontal, fronto-occipital, occipital, or holocephalic. It is often aggravated by Valsalva-type maneuvers and is typically bilateral but sometimes unilateral.
Rarely, a paradoxical postural headache may occur and is present in recumbency and relieved in an upright position (84). Occasionally, the headaches are primarily or entirely exertional (82; Wang and Fuh 2005). Of note, the emergence of a daily, persistent, nonpositional pattern of head pain in the chronic stage of spontaneous CSF hypotension is concerning, as this change can suggest the development of a subdural effusion, subdural hematoma, or cerebral venous thrombosis (66; 108; 138).
Sometimes, despite documented CSF leaks, low CSF opening pressures, and the presence of typical MRI abnormalities of the disorder, patients have no headaches at all (86). These patients tend to have ventriculoatrial or ventriculoperitoneal shunts (70). The shunt valve opening pressure is low enough to reduce CSF filling pressure, but an “anti-siphon” shunt valve closes on standing, thus, eliminating the development of orthostatic headache (70).
• Orthostatic headache
Variations in presentation
• Nonorthostatic headache (20% to 25% of the time)
Variations in orthostatic onset
• Minutes to hours
Complications from the disorder include the development of the following comorbidities (16).
Brain sag, venous outflow obstruction, and swelling of the diencephalon can lead to frontotemporal dementia, which manifests as personality changes. Compensatory venous engorgement and stasis that results from decrease in intracranial CSF volume can lead to cerebral venous sinus thrombosis. Similarly, compensatory enlargement and congestion of hypophyseal veins can cause pituitary engorgement and predispose patients to pituitary apoplexy. In patients who have a significant orthostatic component, prolonged supine positioning can lead to deconditioning and the development of postural orthostatic tachycardia syndrome (POTS). Venous traction at the skull base may cause superficial siderosis and manifest as microhemorrhages, or bleeding may occur at the site of the dural defect.
A 37-year-old woman presented with acute orthostatic headache. She had no known systemic disease and had not been taking any medication on a regular basis. She experienced an acute onset of severe headache after she got up from bed one morning 2 weeks prior to her presentation. It was described as a persistent dull ache in the occipital region and was associated with neck stiffness and muffled hearing. The symptoms were relieved soon after she lay supine and recurred within 5 minutes after she sat erect. There was no photophobia, phonophobia, nausea, vomiting, tinnitus, nasal congestion, or other upper respiratory tract infection symptoms, and she denied having trauma, surgery, or lumbar puncture before the onset. After unsuccessful treatment as migraine or tension-type headache by her family physician, she went to the neurology service of another hospital for help. Spontaneous intracranial hypotension was suspected, although the initial brain MRI did not show characteristic features, including no diffuse pachymeningeal enhancement. She was treated with intravenous fluids, and generous caffeine intake was recommended. There was little improvement, and she was, thus, referred to a specialist. The neurologic examination was unremarkable except for mild neck stiffness. A second brain MRI with contrast on admission revealed characteristic findings suggestive of intracranial hypotension.
Heavily T2-weighted MR myelography demonstrated multiple CSF leaks in the cervico-thoracic junction and the upper thoracic regions. An epidural blood patch of 20 mL was delivered at the level of T1-2 and resulted in substantial improvement of her symptoms. She still had a mild to moderate orthostatic headache after the procedure. A follow-up heavily T2-weighted MR myelography demonstrated residual CSF leaks at the level C7-T1, and another targeted epidural blood patch was carried out, which resulted in complete and sustained resolution of her symptoms.
This case was selected to demonstrate the following:
• The characteristic presentation of an orthostatic headache relieved with recumbency in the absence of previous trauma or lumbar puncture should alert the physician to the possibility of spontaneous intracranial hypotension and prompt a search for spinal CSF leaks.
• Brain MRI findings that are characteristic of spontaneous intracranial hypotension can be trivial and easily overlooked. If the initial brain MRI does not provide objective evidence supporting the clinical diagnosis, imaging procedures to demonstrate spinal CSF leaks, such as a fat suppressed heavily T2-weighted MR myelography and CT myelography should be considered.
• Heavily T2-weighted MR myelography is useful in localizing spinal CSF leaks. It is non-invasive and radiation-free, and it may be a good alternative to CT myelography to guide placement of targeted epidural blood patches.
Spontaneous intracranial hypotension results from spontaneous spinal CSF leaks. The etiology of spontaneous spinal CSF leaks is not yet fully understood and is likely multifactorial. It has been linked to structural weakness of the spinal dural sac, disorders of the connective tissue matrix, CSF venous fistulas, dural tears from spinal osseous lesions, fluid-filled perineural (Tarlov) cysts, and trivial trauma (such as coughing, pushing, trivial falls and sports activities, and lifting).
The pathogenesis of orthostatic headache secondary to spontaneous intracranial hypotension is also unclear, with theories regarding this concept evolving over time. Ultimately, the thought is that spinal CSF leaks cause orthostatic headache by disrupting craniospinal CSF space elasticity, or compliance (70). Specifically, when the lumbar spinal CSF space compliance abnormally increases, the hydrostatic indifferent point, or the point at which venous pressure is not affected by posture, is displaced caudally. This caudal displacement of the hydrostatic indifferent point leads to acute orthostatic venous distention on position shift from recumbency to sitting or standing, thereby producing an orthostatic headache.
Of note, the theory that depletion of CSF volume is the sole cause of spontaneous intracranial hypotension or orthostatic headache is controversial. Patients with CSF rhinorrhea or otorrhea, conditions that also cause CSF volume depletion, rarely, if ever, develop symptoms or brain imaging findings typical of spontaneous intracranial hypotension (108). Therefore, even in the presence of cranial CSF leaks, a spinal source should be sought in patients with orthostatic headaches thought to be secondary to spontaneous intracranial hypotension (118).
Orthostatic headaches are secondary to traction on pain-sensitive cranial nerves (V, IX, X) and acute orthostatic venous distention on position shift from recumbency to sitting or standing (70).
Diplopia results from stretching or compression of the related cranial nerves (41; 51).
(3) Visual blurring and visual field cuts
Visual changes result from compression or vascular congestion of the intracranial portion of the optic nerve (55).
(4) Dizziness and change in hearing, decreased hearing, deafness, tinnitus, orthostatic tinnitus
Hearing and vestibular changes are due to stretching of cranial nerve VIII or pressure changes in the perilymphatic fluid of the inner ear (97; 91).
(5) Galactorrhea and increased serum prolactin level
Increased prolactin can happen due to distortion of the pituitary stalk as a result of sinking of the brain (150)
(6) Radicular upper limb symptoms
Radicular upper limb symptoms occur due to stretching of the cervical nerve roots as the result of sinking and downward displacement of the brain or irritation of the nerve root by dilated epidural venous plexus (03; 84).
(7) Encephalopathy, stupor, and coma
Mental status changes are attributed to diencephalic compression (09; 95; 37).
(8) Cerebellar ataxia and parkinsonism and bulbar manifestations
Cerebellar and motor symptoms are attributed to compression of posterior fossa and deep mid-line structures (92).
(9) Frontotemporal dementia
Frontotemporal-like dementia is attributed to the compression of frontal and temporal lobes (54; 145).
(10) Gait disorder
Gait disturbances are attributed to spinal cord venous congestion (90), cord distortion, or deformation (78; 146).
Spontaneous spinal CSF leak leads to the following:
(1) Collapse of the ventricles, which may be obvious or subtle.
(2) Sinking of the brain that, on MR imaging, is manifested by descent of the cerebellar tonsils (sometimes mimicking Chiari I malformation), a decrease in the size of the prepontine and perichiasmatic cisterns, flattening of the optic chiasm, and crowding of the posterior fossa.
(3) Intracranial venous hypervolemia and subdural fluid collections. According to the Monro-Kellie Doctrine (81), given the fact that an intact skull is not compressible and that the brain is not expected to expand, loss of CSF volume has to be somehow compensated. This is accomplished by engorgement of cerebral venous sinuses and dilation of meningeal veins. The latter is the cause of diffuse pachymeningeal enhancement. Engorgement of the pituitary vessels is responsible for pituitary enlargement.
(4) Another consequence of CSF volume depletion is partial collapse of the spinal dura, which in turn leads to a compensatory dilation of epidural venous plexus (70).
The actual incidence and the prevalence of the disorder have not been determined, as spontaneous spinal CSF leak is an underdiagnosed cause of chronic headache (106). According to a community-based study, the prevalence was estimated at 1 per 50,000 (116). An emergency department-based study reported an estimated annual incidence of 5 per 100,000 (113). The disorder can occur at any age but is rare in childhood (112). The vast majority of patients are adults (peak incidence at about age 40), and there is a female preponderance (female to male ratio of about 2 to 1) (107).
Because the pathogenesis can vary, little is known about prevention.
Differential diagnosis for orthostatic symptoms. A patient presenting with a chief complaint of an orthostatic headache relieved with recumbency should alert the physician to the possibility of spontaneous intracranial hypotension. The differential for orthostatic headache or symptoms includes postural orthostatic tachycardia syndrome (POTS) (87), orthostatic hypotension (16), cervicogenic headache (16), and migraine (16). POTS usually manifests as orthostatic tachycardia with minimal orthostatic blood pressure change and can be either comorbid or separate from spontaneous spinal CSF leak. Orthostatic hypotension presents with orthostatic symptoms; this disease manifests with fall in systolic (20 mmHg) and/or diastolic (10 mmHg) blood pressure on standing from a seated or supine position. These diseases are distinguished from spontaneous spinal CSF leak in patients with stable heart rate or blood pressure during transition from supine to sitting to standing (16). Cervicogenic headache manifests with neck pain that worsens with cervical motion and can be improved with medication or facet blocks. Migraine can also present with orthostatic vertiginous symptoms, but patients typically have a history of headache or migraine, and symptoms can be alleviated with antimigraine medications.
In the setting of recent spinal procedures or trauma, the differential diagnosis also includes postdural puncture headache and traumatic CSF fistula. A careful patient history should elicit causal events.
Differential diagnosis for new daily headache. Spontaneous intracranial hypotension should be part of the differential diagnosis of any new-onset headaches in young and middle-aged individuals (107), or new daily persistent headache (106).
Chiari type I malformation. Similar to spontaneous spinal CSF leak, imaging findings of Chiari type I malformation show descent of the cerebellar tonsils, and patients can have headache (typically cough-induced) and other neurologic symptoms (16). Patients who have spontaneous spinal CSF leak and who are mistakenly diagnosed with and treated for Chiari type I malformation with surgical posterior fossa decompression can have worsening of their symptoms. Distinguishing these disorders is therefore important.
A Chiari type I malformation occurs in less than 1% of the general population, it is typically asymptomatic and is associated with abnormal morphology at the craniocervical junction (16). In patients with spontaneous spinal CSF leak, cerebellar tonsils do not typically descend more than 5 mm below the foramen magnum and will also typically be accompanied by descent of the midbrain, fall in the cerebral aqueduct (iter) beneath the incisural line, and other signs of brain sag (16). The presence of a syrinx is more likely in Chiari type I malformation. The descended tonsils in Chiari type I malformation typically take on a peg-shaped appearance.
Associated and underlying diseases. As the development of spontaneous intracranial hypotension has been linked to disorders of the connective tissue matrix, CSF venous fistulas, dural tears from spinal osseous lesions, and perineural (Tarlov) cysts, investigation for these concomitant disorders may be warranted.
Underlying disorders of connective tissue matrix are thought to cause dural weakness, and thereby, may be risk factors for spontaneous CSF leaks. Although the presence of underlying well-characterized connective tissue syndromes are rare (less than 5%), physical skeletal characteristics suggestive of systemic connective tissue matrix disorders, such as Marfanoid features, hypermobile joints, and hyper-extensible skin, have been found in up to two-thirds of patients (82; 111). Specifically, in comparing clinical features of patients with connective tissue disorders and spontaneous intracranial hypotension with controls with connective tissue disorders but without spontaneous intracranial hypotension, dolichostenomelia (disproportionately long limbs), but not the other above-mentioned stigmata, was more common (73). Although patients with these skeletal manifestations and spontaneous intracranial hypotension have been found to have abnormalities in fibrillin-1 metabolism, a genetic link to Marfan syndrome has yet to be demonstrated (119). It has been shown that the majority of patients do not harbor mutations in FBN1 gene, encoding fibrillin 1, or in TGFBR2 gene, encoding transforming growth factor-beta receptor 2 (119; 24; 110).
New investigations have shed light on spontaneous CSF-venous fistulas leading to CSF leaks. Dilated epidural veins and arachnoid granulations can contribute to the development of a spontaneous CSF-venous fistula, which serves as a direct conduit for outflow from the subarachnoid space into systemic circulatory system via spinal epidural veins, thereby causing spontaneous intracranial hypotension (112). CSF-venous fistulas should be suspected in patients with refractory spontaneous intracranial hypotension and unremarkable conventional spinal imaging. Digital subtraction myelography, and even CT myelography, may be useful in diagnosing this underlying pathology (112; 62).
Dural tear from a spondylotic spur (136; 36; Binder et 2005) or disc herniation (147; 100) may cause a dural defect and CSF leak. Microsurgical exploration has led to the discovery of discogenic microspurs as underlying pathology for CSF leaks from primarily ventral dural tears. Up to 71% of cases with intractable spontaneous intracranial hypotension were found to be secondary to a CSF leak from a circumscribed vertical longitudinal slit in the dura caused by a calcified microspur originating from a intervertebral disc (10).
Perineural (Tarlov) cysts, or cysts filled with CSF located between the nerve root and dorsal ganglion, can occur in multiple locations in the spine and cause CSF leaks (99). Tarlov cysts should be suspected in patients with concomitant symptoms of back pain, radicular pain, or bowel or bladder dysfunction (99). CT myelogram is diagnostic, and treatment of the cysts with blood patching, surgical excision, or percutaneous drainage may resolve symptoms (99).
Low CSF pressure (less than 6 cm H20) is diagnostic of spontaneous intracranial hypotension. However, patients with spontaneous intracranial hypotension can commonly have normal CSF pressures, suggesting that a lack of low CSF pressure should not exclude this condition (63). In addition, obtaining a CSF pressure with lumbar puncture may worsen symptoms of someone with spontaneous spinal CSF leak. Fortunately, a correct diagnosis can usually be made based on characteristic clinical presentation and typical findings on noninvasive MRI techniques, and the need for a spinal tap has been greatly reduced.
Overall, a head CT scan is of little help in the diagnosis of this disorder, as it typically will not reveal abnormalities characteristic of spontaneous intracranial hypotension. Sometimes subdural fluid collections or increased tentorial enhancement may be seen (94; 123).
Common MRI brain imaging abnormalities. Magnetic resonance imaging has truly revolutionized the diagnosis and follow-up of patients with spontaneous intracranial hypotension. Typical brain MRI findings in spontaneous intracranial hypotension include:
(1) Diffuse pachymeningeal enhancement
• Descent of the cerebellar tonsils
(3) Engorgement of cerebral venous sinuses
In a retrospective analysis, pachymeningeal enhancement, signs of brain sag, and venous distension sign were the most common MRI brain abnormalities correlating with low spinal CSP pressure, present in 83%, 61%, and 75% of subjects, respectively (63).
Diffuse pachymeningeal enhancement (sparing the leptomeninges), thought to be secondary to increased transmural venous pressure causing dilation of inner dural veins (70), is the most common MRI abnormality. It is diffuse, uninterrupted, and non-nodular, and it involves the supratentorial and intratentorial pachymeninges. Typically, it appears thick and obvious on imaging but sometimes is thin (53; 43; 93).
Sinking of the brain, “sagging,” or “descent of the brain” is manifested by descent of the cerebellar tonsils mimicking a type I Chiari malformation, as well as sinking of the opening of the third ventricle aqueduct (“iter”) to a level below the incisural line (59; 71). Furthermore, descent of the brain may lead to a decrease in the size of the prepontine and perichiasmatic cisterns, crowding of the posterior fossa, inferior displacement, and flattening of the optic chiasm. It is proposed that decreased mamillopontine distance (< 5.5 mm) and reduced pontomesencephalic angle (< 50˚) may provide supportive clues for diagnosis (122).
Engorgement of cerebral venous sinuses is frequently noted (08). The venous distension sign, distension of the midportion of the dominant transverse sinus with a convex appearance on T1-weighted sagittal MRI, is useful in the detection of intracranial hypotension (39). A study showed that convex margins of the transverse sinuses, but not concave margins, predicted an association between midbrain pons angle (within brain descent cluster) and spinal cerebrospinal fluid leak severity (148). Enlargement of the pituitary gland can be obvious and may mimic pituitary adenoma or pituitary hyperplasia (04). This enlargement is due to hyperemia (increase in blood flow) in the pituitary gland and is linked to hyperemia of the dural and epidural venous sinuses. Increased height of the pituitary gland (mean ± SD, 6.9 ± 2.3 mm) has a sensitivity of 63% and specificity of 97% of spontaneous spinal CSF leak. This finding is reversible and, with resolution or management of the disease, resolves earlier than meningeal enhancement.
Subdural fluid collections are typically bilateral but may be unilateral and appear over the cerebral convexities. These fluid collections are usually hygromas that may reveal variable signal intensity depending on the concentration of protein in the fluid, although subdural hematoma may sometimes develop and cause significant mass effects (66).
Ophthalmic findings. There are additional ophthalmic findings that correlate with spontaneous spinal CSF leak. The diameter of the superior ophthalmic vein on contrast-enhanced T1-weighted coronal MRI was reported to be correlated with intracranial pressure (72), and collapsed superior ophthalmic veins might provide additional clues for intracranial hypotension (19). Decreased intersheath space of the optic nerve has been reported as well and probably also results from reduced CSF content (102), and a similar finding has been demonstrated with sonography (34).
Common spinal MRI imaging abnormalities. Conventional spinal MRI abnormalities include the following:
(1) Extra-arachnoid or epidural fluid collections
Extra-arachnoid or epidural fluid collections, when present, indicate the presence of CSF leakage (22). However, such fluid collections often extend across several levels and, thus, do not reveal the exact site of the CSF leakage (33).
Extravasation and extension of fluid into the paraspinal soft tissues are infrequently seen and may represent the actual location of CSF leakage. However, when these are seen in the high retro-cervical region, it is claimed that they may not represent the actual site of the leak and may be a false localizing sign (115).
Spinal pachymeningeal enhancement may also be seen, although not as frequently as intracranial pachymeningeal enhancement (79).
Engorgement of epidural venous plexus may be seen at any level of the spine, but typically it is more prominent in mid-thoracic and low thoracic as well as lumbar levels (18; 22). Generally, although a conventional spinal MRI is helpful in revealing abnormalities that might suggest a CSF leak, it is only occasionally that it reveals the actual site of the leakage of the CSF.
Localization of the CSF leak is important as patients refractory to nontargeted therapy may need to undergo targeted epidural injections or surgical repair. CSF leak localization, however, can be difficult and is somewhat dependent on rate of leakage. Therefore, there continues to be development of imaging modalities and algorithms to aid in targeting the location of CSF leak.
Heavily T2-weighted, fat suppressed, MR myelography is a noninvasive MRI technique useful in localizing spinal CSF leaks and has emerged as a good alternative to invasive imaging techniques, such as CT myelography, in the diagnosis and follow-up of patients with spontaneous intracranial hypotension (139; 60; 129; 61). This imaging requires neither lumbar puncture nor contrast medium administration. The principle of heavily T2-weighted MR myelography is to exaggerate the contrast between the signal of CSF, which appears bright, and those from other tissues, which are either invisible or barely visible. Unlike that on conventional spinal MRI, extravasated CSF in heavily T2-weighted MR myelography is readily discerned from the background. Axial slices throughout the entire spine can provide excellent spatial resolution comparable to CT myelography. It is also a time-efficient technique, for a single-shot fast spin-echo pulse sequence is used, and the entire spine can be imaged in both axial and longitudinal planes within 15 minutes (129). As in CT myelography, heavily T2-weighted MR myelography can demonstrate 3 major types of CSF leakages.
(1) CSF leaks along the nerve roots: These are the presumed location of dural defects. Extravasated CSF leaks out of the spinal canal along the nerve roots and at times extends into the paraspinal soft tissues. The leaks appear as bright signals extending from the flanks of “sunny side up,” which represent signals from the intradural CSF and the cord, through the neuroforamina. They assume a band- or thread-like appearance and sometimes look like a fimbria, spreading out at the end. They are most commonly seen in the cervicothoracic junction or the upper thoracic spine, and multiple leaks are not uncommon.
(2) Epidural CSF collections: Some of the extravasated CSF stays within the spinal column and appears as bright signals alongside the periphery of the “sunny side up.” Epidural CSF collections usually extend for several spinal segments and are not necessarily located in the vicinity of CSF leaks along the nerve roots. The distribution might represent compliance with gravity and can be misleading in localizing the actual location of dural tears.
(3) High-cervical extraspinal CSF collections: They appear as patchy bright signals, which may be confluent or scattered, outside the spinal canal in the high cervical region. They are mostly on the dorsal side of the “sunny side up,” but occasionally extension to the lateral or even ventral side can be seen. These are well-known false localizing signs and have nothing to do with the actual leakage sites.
A majority of periradicular postlumbar puncture CSF leaks found on heavily T2-weighted MR myelography have been demonstrated to occur within 3 segments of the lumbar puncture. This suggests that dural defects associated with spontaneous intracranial hypotension (SIH) may be in close proximity to periradicular leaks found on heavily T2-weighted MR myelography (142).
It has been demonstrated that heavily T2-weighted MR myelography was comparable to CT myelography in localizing spinal CSF leaks and was a good alternative to CT myelography prior to targeted epidural blood patching. However, its applicability for other targeted treatments, such as injection of fibrin sealant or surgical repair, is yet to be determined (130; 143).
CT myelography findings. CT myelography has been the gold standard in localizing spinal CSF leaks (33). This test may show the following:
• Extra-arachnoid leakage of fluid
Conventional CT myelography is carried out by performing a myelogram with water-soluble contrast, followed by CT scanning. Slices are typically obtained at each spinal level or at a more selected region if the myelogram itself or a previous cisternography or spinal MRI has revealed clues for potential leakage sites. Under typical circumstances, one would expect to locate the site of the CSF egress and CSF leakage. However, the rate of leakage of CSF may provide special challenges.
Delayed CT myelography may be helpful for slow-flow leaks. Dynamic CT myelography may allow detection of high-flow leaks (69; 74). Ultrafast dynamic CT myelography was found to be especially helpful in identifying CSF leaks caused by spinal osteophytes (127). In an effort to reduce multiple and unnecessary tests, Verdoon and colleagues found success in reducing the need for repeat CT myelograms by using the presence of extradural fluid on spinal MRI to direct whether a patient should have a dynamic CT myelography for CSF leak localization, rather than first undergoing a conventional CT (135).
Radionuclide cisternography findings. Radionuclide cisternography, involving intrathecal injection of indium-111, was previously frequently used, but is now rarely used, in establishing a diagnosis of CSF leak. The dynamics of injected radionuclide are followed by subsequent scanning at various intervals of up to 24 or 48 hours. Normally by 24 hours, but often earlier, abundant radioactivity is detected over the cerebral convexities. When there is CSF leakage, the radioactivity often does not extend much beyond the basal cisterns. Therefore, on 24-hour or 48-hour images, there is the absence or paucity of activity over the cerebral convexities (89; 11; 07). A more desirable but much less common abnormality is detection of “parathecal” or “paradural” activity, pointing to the site or approximate level of CSF leak. Meningeal diverticula may assume a similar appearance, and they can be confused with actual leakage sites. Multiple parathecal radioactivities do not necessarily correspond to multiple spinal CSF leaks (84). Another cisternographic finding in CSF leaks is the early appearance of radioactivity in the kidneys and urinary bladder (56). Normally, such activity is noted at the 6 to 24 hours after the intrathecal introduction of radioisotope. When there is a CSF leak, activity in the kidneys and urinary bladder may be seen in less than 4 hours.
Digital subtraction myelography. Digital subtraction myelogram is a myelogram done under fluoroscopy; the precontrasted image is digitally subtracted to enhance the visualization of the contrast. This imaging technique is useful to detect rapid leaks, ventral leaks, and leaks not associated with an obvious extrathecal CSF collection, such as a CSF venous fistulas (117). Consecutive-day right and left lateral decubitus digital subtraction myelography may be beneficial in detecting CSF venous fistulas (96).
Contrast-enhanced MR myelography. Contrast-enhanced MR myelography or gadolinium-enhanced MR cisternography involves obtaining MRI of the spine after intrathecal gadolinium administration; immediate and delayed images are obtained. This technique is debated; by some it is considered be helpful in the detection of the so-called “slow-flow leaks” (126; 02). It has been considered by some to be sensitive and accurate enough to be an alternative to CT myelography (134). However, in a retrospective study, results showed that contrast-enhanced MR myelography does not improve the diagnostic accuracy for the detection of epidural CSF and should not be included in the workup (33).
It should be noted that no gadolinium-containing contrast medium has been approved for intrathecal use, but the use of intrathecal normal saline followed by intrathecal gadolinium infusion has been studied. Intrathecal preservative-free normal saline challenge followed by contrast-enhanced MR myelography was shown to be a safe technique with the potential to increase detection of slow-flow CSF leaks (50).
Complications of invasive imaging. Challenges to invasive imaging include radiation exposure and the development of iatrogenic CSF leak. Exposure to radiation is a concern for radionuclide cisternography and CT myelography, especially for the latter. For patients receiving dynamic CT myelography, which involves multiple scans, the dose of radiation is even higher. The cumulative risk of oncogenicity should be carefully weighed against the benefits because most of these patients are young or middle-aged adults who have a considerable life expectancy (12; 124). One study reported evidence of iatrogenic lumbosacral CSF leakage on MR myelography after radionuclide cisternography in a substantial proportion of patients (104). The results indicated that iatrogenic CSF leaks could be a potential pitfall for imaging studies involving a lumbar puncture, such as radionuclide cisternography and, perhaps, CT myelography and gadolinium-enhanced MR myelography/cisternography.
Conservative treatment. Various treatment modalities have been advocated for patients with spontaneous CSF leaks. Some are based on prior experience with post-lumbar puncture headaches, rather than direct experience with spontaneous CSF leaks. These include the following:
There is an intuitive and reasonable tendency to treat the patient initially with conservative measures. Bed rest has been traditionally advocated. Hydration or overhydration, recommended in some of the older studies, is frequently practiced, but its effectiveness has not been established (128).
Evidence for pharmacological treatment is lacking. Caffeine and theophylline are used by some experts, but most of the evidence derived from studies in patients after lumbar puncture (42; 57; 17), and the effectiveness is often lacking.
Although many patients show no improvement with corticosteroids, some do (45). Considering the potential serious side effects of long-term corticosteroid therapy, this would hardly seem to be a solution to the patient’s problem. The rate of patients recovering spontaneously or with conservative measures only is unknown; 1 study showed that up to 81% of patients required an epidural blood patch due to the failure of symptom management with supportive measures only (26).
Epidural blood patches. Autologous epidural blood patch has emerged as the treatment of choice for those patients who fail an initial conservative management (49; 32; 29; 125; 120; 13), and it has been suggested that early epidural blood patching is helpful in the majority of patients (13; 75). The effect of epidural blood patch is essentially two-fold. The immediate effect may be related to increased pressure in the epidural space, leading to decrease in its compliance (70). The latent effect is related to the sealing of the dural defect by triggering a focal tissue reaction (35). Sometimes, the patient may obtain a near-immediate and lasting effect soon after the procedure. On the other hand, some patients note an almost immediate improvement followed by the recurrence of symptoms and then a latent improvement after a few days or weeks. Moreover, repeated large-volume epidural blood patches may be necessary to achieve symptomatic relief on rare occasions (76). Overall, the efficacy of epidural blood patch in spontaneous CSF leaks is approximately 30% to 35% (121), which is less satisfactory than in intracranial hypotension syndrome following lumbar puncture or epidural or spinal anesthesia (133). This discrepancy is likely attributed to the fact that epidural blood patches are delivered exactly to the site or the vicinity of the dural defects in treating post-lumbar puncture headaches, whereas in spontaneous intracranial hypotension, the blood patches are usually delivered “blind” at the lumbar region instead of the CSF leakage sites, especially when the CSF leaks could not be localized. In addition, the anatomy of dural defects in spontaneous intracranial hypotension is more complex.
Current evidence suggests that targeted epidural blood patches, or patches placed directly at the level of identified spinal CSF leaks, may double the response rates as compared to traditional, nontargeted lumbar epidural blood patches (121; 143; 23). Moreover, improvements have been made in needle placement technique to target the ventral epidural space (05).
With more understanding as to concomitant conditions leading to CSF leaks, the advent of noninvasive heavily T2-weighted MR myelography, and the development of new placement techniques, targeted epidural blood patches may be considered as the first-line treatment, directed at the identified spinal CSF leaks in the hope of hastening recovery and limiting development of complications; however, more studies are needed before the spectrum of applicability is determined.
Reports on epidural injections of fibrin glue and fibrin sealant are encouraging (46; 30; 114), especially for patients who failed epidural blood patching. However, there have been 2 patients experiencing anaphylactic reactions after fibrin sealant injection in treating spontaneous intracranial hypotension (109).
Epidural infusion of saline has produced various results (101; 132; 48). One might consider this with limited expectations in some of the patients who have failed epidural blood patches and when other measures such as epidural injection of fibrin glue or surgery are not viable options.
Similarly, the experience with intrathecal infusion of fluid in spontaneous CSF leaks is limited. However, on the rare occasions that patients show obtundation or impending coma, this technique may prove helpful in improving the level of consciousness and allowing time to search for the site of the leak and establish a more definitive treatment if no lasting effect is obtained (14; 144). One would be concerned about potential complications of continuous epidural or intrathecal infusions such as infection.
Surgery in well-selected cases often proves helpful. It may be considered in those patients who have failed less invasive treatment modalities. It is essential that the actual site of the CSF egress be determined by neuroradiological studies before surgery is undertaken. Because the anatomy of the spontaneous leak may be complex, the surgery may not always be straightforward. Sometimes, a surgeon may encounter the CSF that has leaked but may not be able to locate the exact site of the leak. In this type of case, he or she may pack the area with blood-soaked gel foam, muscle, etc. and hope for the best. Sometimes, dural defects are encountered with so markedly attenuated a border that it may not yield to suturing. Other times, meningeal diverticula or dural defects are encountered that surround 1 or more nerve roots and create technical challenges (116). It has been reported that minimally invasive surgery to correct the leaking meningeal diverticulum may be considered an alternative to conventional open surgical repair, although it is yet to be determined whether this approach is feasible for most patients (40).
Cerebrospinal fluid-venous fistula has been described as a refractory cause of spontaneous intracranial hypotension. A prospective study found that surgical ligation is highly effective for the treatment of spontaneous intracranial hypotension due to cerebrospinal fluid-venous fistula (140).
Patients with spontaneous CSF leaks can make a complete recovery either spontaneously or with conservative management. Many, however, require more invasive therapeutic approaches such as epidural blood patch, epidural injection of fibrin glue, or even surgery. Recurrences may occur in a minority.
Predictors of autologous epidural blood patch success have been studied. Structural factors such as brain sagging on imaging, severe diencephalic-mesencephalic deformity, and intracranial structural dislocation, as measured by the angle between the vein of Galen and sagittal sinus, have been found to be negative predictors of clinical response from the first epidural blood patch (58; 27; 148). Early visualization of bladder activity in radioisotope cisternography was also found to be a negative predictor of success (58). This is thought to be a marker of increased CSF leakage, leading to increased release of the radioactive tracer into systemic circulation, causing early appearance in the bladder.
A major complication of spontaneous CSF leak is the development of unilateral or bilateral subdural hematomas. As many as 20% of patients may develop subdural hematomas, which may be asymptomatic, or they can increase in size, becoming symptomatic and creating significant therapeutic challenges (06; 31; 98; 131; 66). Size of the subdural hematoma is important (20).
Patients with spontaneous intracranial hypotension and subdural hematomas that are less than 10 mm can be treated conservatively or with an epidural blood patch with good outcomes. However, patients with spontaneous intracranial hypotension and subdural hematomas more than 10 mm in size may be at risk for uncal herniation (20), which can result in bilateral posterior circulation infarcts and Duret hemorrhage (31; 21). Patients with spontaneous intracranial hypotension and subdural hematomas greater than 10 mm in size should have their Glasgow Coma Scale (GCS) scores closely monitored. In patients with decreased GCS scores, early surgical evacuation might prevent uncal herniation and prevent poor outcome (20).
Drainage or decompressive surgery is not always helpful as paradoxical herniation may occur, and treatment for spinal CSF leaks should be considered before surgical interventions are undertaken (137; 47).
Cerebral venous thromboses reportedly occur in about 2% of patients with spontaneous intracranial hypotension, and the majority (about 85%) involve dural venous sinuses (13; 108). Isolated cortical vein thrombosis has also been reported in spontaneous intracranial hypotension (65; 67; 141), as well as in intracranial hypotension syndrome following unsuccessful epidural anesthesia (01). These patients may have a change in headache pattern (40%), venous infarction, seizure, dural arteriovenous fistula, or even cortical subarachnoid hemorrhage.
Sometimes following treatment of spontaneous CSF leaks, whether by surgery or by epidural blood patch, a symptomatic syndrome of intracranial hypertension may develop (83). This is usually a self-limiting syndrome that resolves within several weeks or months.
Simy Parikh MD
Dr. Parikh of The Jefferson Headache Center at Thomas Jefferson University received an honorarium from AbbVie for service on a speakers bureau.See Profile
Shuu-Jiun Wang MD
Dr. Wang of the Brain Research Center, National Yang-Ming University, and the Neurological Institute, Taipei Veterans General Hospital, has no relevant financial relationships to disclose.See Profile
Nearly 3,000 illustrations, including video clips of neurologic disorders.
Every article is reviewed by our esteemed Editorial Board for accuracy and currency.
Full spectrum of neurology in 1,200 comprehensive articles.
Neuro-Ophthalmology & Neuro-Otology
Aug. 17, 2022
Headache & Pain
Jul. 13, 2022
Headache & Pain
Jul. 11, 2022
Headache & Pain
Jul. 01, 2022
Jun. 11, 2022
Jun. 06, 2022
Jun. 04, 2022
Jun. 04, 2022