Cerebrospinal Fluid (CSF) Leaks

Timothy C. Hain, MD.• Page last modified: April 3, 2023

Also see: nph and hydrocephalus

Cerebrospinal fluid depletion may be caused by a leak, a shunt, inadequate production or too rapid absorption. There are also some similar syndromes where the compliance of the intracranial compartment is too high, causing similar symptoms when the brain sags downward on standing, and floats back upward supine. It is estimated that CSF leaks have an incidence of about 5/100,000 per year.

Symptoms of CSF leak commonly include headaches, which are more severe in the upright position and are alleviated by supine or head-lowered below chest (Trendelenburg) positioning. Horizontal diplopia, change in hearing, tinnitus, blurring of vision, facial numbness, nausea, and upper limb radicular symptoms (tingling) may occur. These symptoms are nonspecific as they are commonly encountered in migraine and post-traumatic headache. Cognitive decline has also been reported (Hong et al, 2002; Pleasure et al, 1998).

The diagnostic criteria of the International Headache Society for CSF leak headache include:

  1. Orthostatic headache (headache worse upright than supine)
  2. At least one associated symptom among:
    1. neck stiffness
    2. tinnitus
    3. hypoacousia
    4. photophobia
    5. nausea
  3. Evidence of low CSF pressure or leakage by at least one of the following:
    1. brain MRI showing indirect signs of low CSF pressure
    2. lumbar puncture recording CSF pressure < 60 mm H2) in the lateral decubitus
    3. CT myelography
    4. Radionuclide cisternography showing the site of leakage

Of these symptoms, neck pain or stiffness, nausea and vomiting are the most common symptoms. If we consider which symptoms are the most specific, this is obviously #3. The associated symptoms are common human complaints -- for example, severe tinnitus is endorsed by 6% of the population, and migraine headaches which commonly include photophobia and nausea affect roughly 15% of the entire population. Regarding #3, It has been our observation that brain MRI showing indirect signs of CSF pressure (i.e. dural enhancement) is rare.

Shievink et al (2016) observed that CSF leaks at the level of the skull base -- e.g. CSF rhinorrhea, do not cause spontaneous intracranial hypotension, and presumably then, do not cause orthostatic headaches either. Thus, all CSF leaks do not cause headache.

Orthostatic headache without CSF Leak

Leep and Mokri (2008) reported that a small number of patients with typical orthostatic headaches lack radiological confirmation of CSF leak. They speculate that in these patients, there might be either an occult leak, or an abnormally distensible dura.

In the "trephine syndrome", similar symptoms are associated with a large skull defect (see image below), and are relieved by cranioplasty. (Stiver et al, 2008). One would think that there would be postural hearing fluctuations in the trephine syndrome, but to our knowledge, this has not been tested.

Postural headaches, generally on upright, may also occur in persons who are intermittently obstructing their CSF pathways such as in colloid cysts (see image below) or a cyst of the septum pellucidum.

Low CSF pressure without a leak being found, called "spontaneous intracranial hypotension", is rare, with a yearly incidence of about 5/100,000. These may be dectected from dural enhancement or lumbar puncture.

As this is an otoneurology site, we will go into the hearing related symptoms of CSF leaks in more detail. While CSF leak is generally accompanied by an orthostatic (upright only) headache, this association is not universal and in fact, hearing loss may be more common than headache (Oncel et al, 1992). Orthostatic tinnitus is also possible.(Arai, Takada et al. 2003). The hearing loss of CSF leak likely results from lowering of CSF pressure, which lowers perilymphatic (inner ear) pressure, and results in a picture similar to Meniere's disease. (Walsted et al., 1991).

trephine Colloid cyst
MRI of patient with large craniotomy defect, potentially causing trephine syndrome (see above). Colloid cyst of third ventricle -- these can intermittently obstruct, causing very severe orthostatic headaches.
shunt Thick meninges post shunt
Shunt to drain ventricles. Shunts sometimes work too well and cause low CSF pressure. MRI of patient post shunt with thickening of the meninges, possibly due to low pressure. This is a contrast study.


Locations of CSF leaks

cerebellar effusion
A cerebellar effusion that was discovered 2 weeks after a cervical epidural.

Acquired leaks can be caused by head or spine injury, surgery, infection or tumor (see above)

Leaks can occur in at dural root sleeves throughout the spine and may be a potential mechanism for orthostatic headache following motor vehicle accidents (Ishikawa et al, 2007). Ehlers-Danlos syndrome patients can develop leaks as a result of dural ectasia. Sometimes shunts put in to relieve hydrocephalus can "overshunt", resulting in a situation similar to a CSF leak.

Infections: Only 20 cases of leaks due to middle ear infection had been reported as of 2007 (Manjunath, 2007).

Congenital leaks are most often associated with the development of anomalous transcranial pathways.

Other things that can look like a leak

Other conditions to exclude when considering the diagnosis of CSF leak are orthostatic hypotension (which can cause dizziness on standing), positional vertigo due to inner ear disturbances such as BPPV, and orthostatic tremor. When considering leaks from the nose, of course other fluids than CSF can leak out of the nose. "Pseudo-CSF rhinorhea" can occur as a result of injury to the sphenopalatine ganglion (Hilinski et al, 2001).

CSF leaks can also result from high-pressure hydrocephalus. In this situation, leaks may recur until the cerebral ventricles are shunted (Tosun et al, 2003)

There are many patients with fluid in their middle ear, and exceedingly few that have CSF in their middle ear.

Diagnostic testing

Tests not requiring lumbar puncture:

Before starting the discussion, lets first say that none of these techniques are very useful. The best, at the present time, involves using MRI to find side effects of low CSF pressure -- venous engorgement or downward displacement of the brain. It has been commented that no imaging techniques are particularly useful. (Kranz et al, 2016)

Head MRI findings include diffuse meningeal gadolinium enhancement due to engorgement of the cerebral venous system (Mokri et al, 1997, see page on dural enhancement), imaging evidence of sinking of the brain resembling the Chiari malformation (Paynor, 1994), subdural fluid collection, decrease in ventricular size and prominent dural sinuses.

Spine MRI may show extra-arachnoid fluid, meningeal diverticuli, meningeal enhancement, or engorgement of epidural venous plexi (Miyazawa et al, 2003; Medina et al, 2010; Starling et al, 2013; Bonetto et al, 2011).

The engorgement of the epidural venous plexus is analogous to the findings of meningeal enhancement on cerebral MRI. Engorgement may be missed if the MRI is read by a radiologist unfamiliar with this syndrome.   Extradural fluid collections were reported in more than 85% of patients in several studies  The collections were best visualized on axial T2, generally were located epidurally, and usually extended over 5 or more spinal segments. While this finding suggests that a CSF leak is possible, it does not identify the site of leak. It does not replace myelography. The image above shows an epidural collection outside the cerebellum.

Bonetto et al (2011) suggested that post-processing image analysis with subtraction of T1 from T2 weighted spinal MRI was highly successful for identification of epidural leakage causing spontaneous intracranial hypotension. All of Bonetto's patients also had meningeal enhancement, and most had ventricular collapse. In other words, these were severe cases.

Response to abdominal binders is both used as a treatment as well as a diagnostic maneuver. (Schievink, 2008). It seems likely that compression garments, such as the "Zoot compression clothing", would work to even better effect.

c1-c2 Hygroma
C1-C2 sign from J Neurol Neurosurg Psychiatry2003;74:821-822 Spinal hygroma found in patient with orthostatic headaches.

The C1-C2 sign is a focal area of fluid between the spinous processes of C1 and C2 (image to left above).  According to Medina et al, the C1-C2 level may be prone to fluid due to mobility, lack of epidural fat, and laxity of connective tissue at this level.

Spinal fluid hygromas (see image above to right), are similar fluid collections. When found in the setting of orthostatic headaches, they are a good reason to pursue the diagnosis of CSF leaks.

Spontaneous leaks into the middle ear are usually diagnosed by high-resolution CT scans. This may reveal discrete tegmen defects, multiple pinhole fistulae, or a combination of these findings. About 6% of the population has some degree of bony defect. Most patients have an ipsilateral conductive or mixed hearing loss, the conductive component caused by CSF in the middle ear. These can be repaired via a middle fossa craniotomy (Leonetti et al, 2005).

Other tests not requiring lumbar puncture that have not been generally adopted

It has been reported that doppler or MRI of the superior opthalmic vein may reveal abnormalities in persons with low CSF pressure (Chen et al. 1999; Chen et al., 2003). Doppler may show increased flow velocity. It may be difficult to find a sonographer with expertise in this method. We have not tried to use this method in Chicago -- and it also seems unlikely to work in any case.

MRI may show collapse of the superior opthalmic vein (Chen et al., 2003). Radiologists do not generally measure the diameter of the superior opthalmic vein, and if this technique is planned, it seems prudent to either seek out an experienced radiologist or develop the expertise through a series of MRI's done in normal persons. We have not had any success in this technique to date, but this may be simply because we have not seen an appropriate patient.

Leaks of spinal fluid from the nose or external ear can be detected by assaying the fluid with a chemistry test called beta-trace protein assay (Bachman et al, 2002; Wernecke et al, 2004). This test when available seems reasonable when there is considerable drainage. If it is positive, there is no need to go on to a more sensitive test (such as injection of contrast or radioactive material using a lumbar puncture)

Tests requiring lumbar puncture:

Opening pressure on LP

In adults, CSF opening pressure is generally considered normal between 10 and 20 cm H20 (Wright et al, 2012). Some authors suggest a wider range of 6-25 is within the 95% confidence intervals (Lee and Leuck, 2014). In normal children values may be even higher (28).  A pressure less than 60 mm (6 cm) is suggestive of CSF leak.

Some practitioners use a provocative test where artificial CSF is injected, symptoms are assessed, and then withdrawn.

Radioactive labels

Use of radioactive material into the spinal fluid is logical as it can be detected with extremely high sensitivity, and also there is no natural source for radioactive secretions. Radioactive label materials may include technetium or indium 111. Indium is an agent that attaches specifically to CSF proteins. A major disadvantage of the radioactive imaging is high cost and the possibility of false-positive results. This study requires injection of material into the spinal canal, a process which could introduce a CSF leak by itself. Radioisotope cisternography characteristically shows a decrease or absence of activity over the cerebral convexities and early accumulation of radioisotope in the bladder.


Radiologic studies are also useful --

CT contrast cisternography after metrizamide injection is considered to be the most useful during a leak. This may be impractical however if one does not know the level of the leak and can also be falsely localizing (Lue and Manolidis 2004). CT also requires radiation.

Hyrtl fissure
Hrtyl fissure on Axial CT of temporal bone (Image from Oney et al, 2022).


CT scans of the temporal bone can be useful to identify defects in the skull base, such as due to high CSF pressure or arachnoid granulations. A source of CSF leak into the middle ear, the Hyrtl fissure, can be seen on axial CT. This can result in CSF filling the middle ear.

MRI scans can also show large arachnoid granulations (which may erode the skull base), an empty sella (correlated with high CSF pressure), drooping of the cerebellum (due to low CSF pressure from the leak), and of course, other potential sources of headache.

MRI myelography

Gadolinium MRI documenting location of CSF leak (arrow). Image from Jinkins et al, 2002.

A new test, injection of a small amount of gadolinium contrast into the lumbar spinal fluid followed by MRI is very promising but rarely done (Jinkins et al, 2002). Gadolinium is not as toxic as is CT contrast dye, and also while CT contrast could be confused with bony structures, gadolinium is not easily confused with other body parts. Of course, MRI can scan the entire neuroaxis if the site of leak is unclear (but at great expense).

Older methods, not recommended at this date

Although not recommended as a diagnostic test, on lumbar puncture, in approximately half of patients, the opening CSF pressure is 40 mm or less. However, if this is found, one still needs to locate the site of leak. The other half of patients may have normal CSF pressure so a lumbar puncture is not 100% reliable to diagnose low CSF pressure. Of course, there is some danger of causing a leak, per se, with a lumbar puncture, and the risks/benefits must be carefully considered. In a small minority of patients with CSF leak, CSF pressure may be normal although patients continue to have symptoms and abnormal MRI findings. Perhaps this indicates an intermittent situation. Spinal fluid testing may show a minor pleocytosis of 5 or more cells, and a modest and variable increase in protein (Mokri et al, 1997).

In the past it was thought that CSF could be distinguished from other fluids by it's glucose content. Testing of the fluid dripping from the nose for glucose is no longer thought to be useful. Testing for beta-transferrin may help determine if it is CSF or something else. While not difficult to implement, have not had much success with this method. Others report success and also that patients can be sent home with a test-tube to collect drainage.

A surer method to detect CSF leak than collecting drainage is to inject radioactive label or a fluorescent dye into the spinal fluid and test for the label or dye in the fluid. Nasal pledgets can be left in the nose for extended periods, enabling detection of intermittent rhinorhea. 0.5 ml of 5% fluorescein diluted in 9.5 cc of CSF is used (according to Hilinski et al, 2001). Smaller amounts are used in children. (Lue and Manolidis 2004). It is useful to use control pledgets under the upper lip for comparison. the test is analyzed under ultraviolet light. No complications were reported by Montgomery in more than 200 injections of fluorescein. Seth et al (2010) used this technique during surgery. However some authors have reported seizures from fluorescein can occur. For this reason, this method is not recommended.


Audiogram in CSF leak
Mechanism through which CSF leaks cause hearing changes via endolymphatic hydrops(Michel and Brusis, 1992). This diagram shows the general idea but it is inaccurate as there are other pathways for CSF flow other than the cochlear aqueduct. Michel and Brusis, 1992. Illustration of low tone hearing loss type of audiogram following myelogram. 1,2,3 are about 1 month apart.


Audiometry (hearing testing) is sometimes affected by CSF-leak. (Kilickan and Gurkan, 2003; Michel and Brusis 1992; Walsted, Salomon et al. 1991; Walsted, Salomon et al. 1991; Walsted, Salomon et al. 1993; Walsted, Salomon et al. 1993; Walsted, Nielsen et al. 1994; Walsted, Nilsson et al. 1996; Walsted, Salomon et al. 1996; Walsted 1998; Walsted 2000). The pattern of hearing loss may closely resemble that of Meniere's disease and exhibit a low-tone sensorineural pattern.

Although logically, hearing in persons with CSF leaks should improve in the supine position, as pressure goes up in the cranial compartment, this has been little studied. One would also think that tests that correlate with hearing, such as OAEs, could be postural too. This has been reported to occur in such patients (e.g. Miyazaki et al, 1997. Unfortunately, a confounding problem with both postural hearing and OAE testing, is that middle ear pressure may change with posture.

Treatment of CSF leaks

Treatment of CSF leak may include bed rest, hydration and steroids. Abdominal binders may be helpful. (Scheviak, 2008). Epidural blood patch as well as injection of fibrin glue is used in patients with spinal leaks who fail noninvasive measures. Blood patches are generally thought to be safe but occasional reports of increased CSF pressure and persistent epidural fluid collections have been reported.

The overwhelming majority of patients have a spinal level leak, although they are generally higher than the lumbar level (Mokri, 1997). Surgical repair may be performed in patients that fail blood patch if the site of the leak has been identified.

Leaks in persons with fluid in their ears. Here the leak cannot be in the spine, and blood patches are irrational.