Spontaneous intracranial hypotension

Last revised by Calum Worsley on 17 Mar 2024

Intracranial hypotension, also known as craniospinal hypotension is a clinical entity that results from a cerebrospinal fluid (CSF) leak that almost without exception occurs from the spine, either into the epidural space or directly into veins in the setting of CSF-venous fistulas. It usually, but not invariably, presents clinically with orthostatic headaches. Diagnosis of the leak and treatment are challenging and rely on multimodality imaging.

As understanding of CSF dynamics and the pathophysiology of CSF leaks has evolved, the terminology has become somewhat confusing.

Intracranial hypotension can broadly be divided into:

  1. primary: usually referred to as spontaneous intracranial hypotension

  2. secondary: iatrogenic (lumbar puncture or surgery), over-shunting due to diversion devices, or traumatic

Importantly, not all types of CSF leak produce intracranial hypotension. In particular, with the exception of rare case reports 43,44, skull base CSF leaks almost never result in the clinical syndrome of intracranial hypotension. The reasons for this are complicated but are related to whether a leak occurs above or below the hydrostatic indifference point 25,26, which is usually located in the low cervical or upper thoracic spine.

Similarly, middle cranial fossa meningoceles seen in patients with underlying idiopathic intracranial hypertension do not usually manifest with clinical or imaging features of intracranial hypotension 26,29.

Spontaneous intracranial hypotension is typically encountered in middle age (30-50 years of age) and has a predilection for women (F:M = 2:1). Of interest, this is a similar demographic to idiopathic intracranial hypertension, which is believed to be an unrecognised predisposing factor.

Epidemiology of secondary intracranial hypotension is variable and matches that of the underlying cause.

The condition often presents as a postural/orthostatic headache which is relieved by lying in a recumbent position, usually within 15-30 minutes 12. This is, however, not always the case and the orthostatic quality of the headache may subside in the long term. Some patients have headaches that are not relieved by lying down, others have headaches that develop slowly during the course of the day (“second half of the day” headaches) 26 or are precipitated by the Valvalsa maneuver. Other symptoms include nausea, vomiting, neck pain, visual and hearing disturbances including tinnitus, and vertigo 17,26.

Occasionally, the clinical presentation is more sinister, with frontotemporal brain sagging syndrome (FBSS). In this phenotype, patients present with a progressive cognitive impairment mimicking behavioral variant frontotemporal lobar degeneration (dementia), may have decreased level of consciousness and coma, and may have ataxia or parkinsonism 3,31. Compulsive repetitive flexion (at the waist) with breath-holding may be a useful clinical sign to differentiate these patients from those with behavioral variant frontotemporal lobar degeneration 31.

Another uncommon manifestation, and potentially masking the underlying presence of hypotension is the development of cerebral venous thrombosis, said to occur in approximately 2% of cases of intracranial hypotension 23.

The International Classification of Headache Disorders, third edition (ICHD-3) criteria for headache attributed to low CSF pressure are 17:

  • A. any headache fulfilling criterion C

  • B. either or both of the following:

    • 1. low cerebrospinal fluid pressure (<6 cm H2O (CSF))

    • 2. evidence of CSF leakage on imaging (on brain or spine imaging)

  • C. headache has developed in temporal relation to the low CSF pressure or CSF leakage, or led to its discovery

  • D. not better accounted for by another ICHD-3 diagnosis

As not all patients with SIH experience headache, some authors modify these diagnostic criteria to include patients whose symptoms are best explained by SIH but who do not have headache 31.

It should be noted that most patients with SIH do not, in fact, have low opening pressures at lumbar puncture. In one study, only 34% patients had pressures <6 cm H2O, while the opening pressure was between 6-20 cm H2O in 61%, and was greater than 20 cm H2O in the remaining 5% 32. The use of lumbar puncture solely to measure opening pressure as a diagnostic tool in suspected SIH is therefore discouraged, but opening pressure can be measured at the time of LP performed during myelography. To allow an accurate measurement to be made, LP should be done in the lateral decubitus position.

Intracranial hypotension results from a CSF leak somewhere along the neuraxis, usually below the hydrostatic indifference point and leads to alterations in the equilibrium between the volumes of intracranial blood, CSF, and brain tissue (Monro-Kellie hypothesis). A decrease in CSF volume leads to compensatory dilatation of the vascular spaces, mostly the venous side due to its higher compliance

Spontaneous intracranial hypotension is usually the result of a CSF leak in the spine. Spontaneous leaks have been classified into four types 27,28 and include multiple causes 9,12,13,25,26:

  • type 1: dural tear

    • most often ventral and thoracic

    • when ventral mostly occur in the upper thoracic spine

    • typically related to calcified disc protrusions

    • responsible for 40% of cases 35

  • type 2: lateral leak

    • associated with a meningeal diverticulum that protrudes through a defect in the nerve root sleeve or its axilla

    • most occur in the lower thoracic region

    • although previously ascribed to rupture of the meningeal diverticulum, these leaks are now considered to occur around the margins of the diverticulum where it protrudes through the defect 45

    • responsible for 17% of cases 35

  • type 3: CSF-venous fistula 16,24,26,35,36

    • most are associated with a meningeal diverticulum

    • most occur in the thoracic spine

    • responsible for 23% of cases (although as these are increasingly recognized this is likely to be an underestimate) 35

  • type 4: distal nerve root sleeve leaks 28

Despite advances in imaging of patients with spontaneous intracranial hypotension, in a significant minority (19%) the cause remains elusive 35.

Additionally, CSF leaks can be secondary trauma (e.g. penetrating injury) or iatrogenic (e.g. lumbar puncture, surgery).

Imaging is crucial both for confirming the diagnosis of intracranial hypotension and identifying the location of the leak. The latter is discussed below in the "imaging strategy" section. 

Described features of intracranial hypotension include:

The most common qualitative finding is pachymeningeal thickening and enhancement followed by dural venous engorgement, tonsillar herniation, and subdural collection; however, these features are not always present, which is why quantitative findings are very helpful in making a more accurate diagnosis on MRI.

  • qualitative signs

    • pachymeningeal enhancement (most common finding): becomes less prevalent over time after the onset of symptoms; hence, in patients with a chronic duration of symptoms in whom clinically the headache pattern also changes from orthostatic to atypical constant headache, the absence of dural enhancement may hinder the diagnosis of intracranial hypotension 15

    • increased venous blood volume

      • venous distension sign: rounding of the cross-section of the dural venous sinuses

      • prominence of inferior intercavernous sinuses is not a sensitive or specific finding; however, it is important to recognize that in cases of intracranial hypotension, should not be mistaken for a pituitary lesion 14

      • intracranial venous thrombosis is a well-recognized, albeit uncommon, complication and may involve the cortical veins and/or dural venous sinuses 19

    • enlargement of the pituitary gland

    • subdural effusions and eventual subdural hematomas

    • diffuse cerebral edema 3

    • reduced CSF volume

    • layer cake skull

    • infratentorial superficial siderosis: although not a sign of intracranial hypotension per se, the presence of this finding is strongly suggestive of a spinal CSF leak from a dural defect in the spine 42

  • quantitative signs

A useful mnemonic to remember the aforementioned features is SEEPS.

Identifying the site of CSF leakage can be challenging, especially in spontaneous cases, and embarking upon imaging can be daunting, as the leak can be located anywhere along the neuraxis (although usually in the spine) and leaks can vary dramatically in their rate ranging from large volume rapid leaks to slow leaks to intermittent/inapparent leaks or those where the connection is directly into the venous system (CSF-venous fistula) which therefore have no pooling of CSF. As a result, no single examination is guaranteed to confirm and localize the abnormality. 

As understanding, technology and techniques have evolved so too have recommended imaging strategies. These will also depend on the local expertise. Generally, the approach should consist of:

  1. confirming the presence of intracranial hypotension

  2. confirming the presence of CSF leak

  3. identifying the specific location and cause of CSF leak

Depending on the cause and location, these may be all achieved on the first study or require a number of different studies.

MRI of the brain is an essential first step, to look for imaging findings that support the diagnosis of intracranial hypotension and also to rule out other unexpected pathology. 

It also enables the calculation of the Bern score, a predictive score that stratifies patients into high, intermediate or low probability of finding a spinal CSF leak/CSF-venous fistula at myelography 33,46,48. This in turn can help guide further investigations.

Ideally this can be performed at the same time as the MRI of the brain but need not consist of a full mulitparameter study; heavily T2 weighted fat saturated sequences through the spinal canal and adjacent soft tissues suffice 25. These can be performed as 3D sequences (ideally) or multiplanar 2D sequences.

The purpose of MRI is not to localize the site of the leak, this is seldom possible by non-invasive means, but rather to determine the most likely type of spinal CSF leak the patient has based on the presence or absence of epidural fluid. This then determines what type of myelography to perform in order to localize the leak. When epidural fluid is shown on MRI (a so-called 'wet leak') then a dural defect and a type 1 or type 2 leak is present. In the absence of epidural fluid (a 'dry leak'), a CSF-venous fistula is most likely to be the cause (or rarely a far lateral nerve root sleeve tear).

If further localization of a leak is required then a variety of myelographic techniques are useful. Which is chosen depends what the initial imaging of the brain and spine have shown 25 as well as on local preference, policies and available equipment and expertise.

There are five main options each with pros and cons:

  1. dynamic CT myelography (prone or decubitus)

  2. digital subtraction myelography

  3. fluoroscopic myelography

  4. MR myelography

  5. nuclear medicine myelography

Unlike traditional CT myelography where intrathecal contrast is introduced a significant time prior to the CT, often in the fluoroscopy room, allowing the contrast to mix evenly throughout the CSF, in dynamic CT myelography, contrast is introduced with the patient already positioned in the CT scanner, either prone or decubitus depending on the likely cause of CSF leak and dense contrast visualized to identify the site of leakage 24,25. This usually requires multiple acquisitions, and thus can have high radiation doses 48.

Dynamic CT myelography is highly sensitive for the detection of ventral leaks for dural tears, lateral leaks from nerve root sleeves and CSF-venous fistulas. A comparison of lateral decubitus CT myelography with digital subtraction myelography showed detection of more CSF-venous fistulas with CT myelography than digital subtraction myelography 41.

It should be noted that if the amount of leaked CSF is large, the distribution of fluid should not be interpreted as necessarily representing the site of leak 9. CSF in the epidural space can migrate over significant distances and pool depending on patient position and anatomy. Similarly, delayed myelography will show opacification of the entire epidural fluid collection and may misrepresent the site of leak.

A particular example of this is pooling of contrast/CSF at the C1/2 level posteriorly. This can be incorrectly attributed to a local leak, when in fact this is not the case. This is referred to as a false localizing sign 10-12

A recent study showed that renal excretion of contrast on CT myelogram is 100% specific for the presence of a CSF leak. Therefore, the identification of such finding should prompt a second look for a CSF leak 22.

Digital subtraction myelography has the best temporal resolution, and with newer equipment allowing Dyna-CT, can also obtain cross-sectional images. This is particularly useful for CSF-venous fistulas. The size of the detector limits the field of view that can be examined and it is typically necessary to undertake two separate digital subtraction myelography runs of the cervical/upper thoracic spine and the lower thoracic/upper lumbar spine to ensure adequate coverage. A further limitation is the decreased visibility of the upper thoracic spine, where most type 1 leaks occur, due to difficulty of beam penetration through the overlying shoulders.

Conventional fluorosopic myelography utilizes real-time fluoroscopic tracking of a bolus of contrast medium injected at lumbar puncture and moved along the spine by means of incremental head-down tilting of the table. This has the advantage of excellent temporal resolution (limited only by the frame rate of the fluoroscopic acquisition) but superimposition of bony structures can limit the sensitivity for leaks, especially in the upper thoracic spine 34.

MR myelography (with intrathecal gadolinium) is not approved by FDA and is associated with neurotoxicity, particularly if renal function or CSF hydrodynamics are compromised but it has been tried in some cases when other methods have failed to demonstrate the source of CSF leak 9. MR myelography with intrathecal gadolinium has been shown to be more sensitive than CT in that respect 6.

Nuclear medicine myelography can also be employed, using intrathecal 111Indium-DTPA, with images obtained typically at 1, 2, 4, 24, and in some instances even 48 hours post-injection 9. Localization can be seen as a focal accumulation of activity. In some cases, only indirect evidence of a leak being present somewhere is available, which is only really useful in instances where the diagnosis of CSF leak remains uncertain. Indirect evidence includes 9:

  • early accumulation of activity within the urinary tract (kidneys/bladder) at 4 hours

  • absent activity over the cerebral convexities at 24 hours

  • rapid loss of activity from within the CSF space

A non-targeted epidural blood patch is often used in cases of spontaneous intracranial hypotension, on the assumption that the leak is from the spine, with variable success 9. When successful, headaches resolve within 72 hours of intervention 12. Subdural effusions can resolve within a few days to weeks. Larger subdural collections often require far longer to resolve 12.

In cases where such speculative treatment fails, localization of the CSF leak is required, allowing for targeted treatment according to leak type 9:

  • type 1: dural tear

    • targeted epidural blood or fibrin glue patch 20

    • surgery 39

  • type 2: lateral leak

    • targeted epidural blood or fibrin glue patch

    • surgery 39

  • type 3: CSF-venous fistula

    • targeted fibrin glue patch 37, 40

    • transvenous embolization 38

    • surgery 39

Targeted epidural blood or fibrin glue patches for ventral dural leaks can either be performed using a translaminar approach, as for initial non-targeted injections, or aimed at entering the ventral epidural space using a transforaminal approach at one or more levels and potentially from both sides 20. For lateral dural leaks from nerve root sleeves or meningeal diverticula, a transforaminal approach can be used to target the dural defect. For CSF-venous fistulas, a good clinical response to fibrin glue is more likely when the spread of injectate is concordant with the location(s) of the draining veins of the fistula as shown by myelography 37.

ADVERTISEMENT: Supporters see fewer/no ads

Updating… Please wait.

 Unable to process the form. Check for errors and try again.

 Thank you for updating your details.