Spontaneous Spinal Cerebrospinal Fluid Leaks: A Review

, Cedars-Sinai Neurosurgical Institute, Los Angeles, and Department of Neurosurgery, University of California, Irvine, California

Neurosurg Focus. 2000;9(1) 

In This Article

Treatment Options

Fortunately, most cases of spontaneous intracranial hypotension resolve spontaneously or respond well to bed rest and a generous intake of oral fluids. However, the duration of symptoms among patients who seek medical care has ranged from weeks to many years. Medical management of spontaneous intracranial hypotension, which has included the administration of caffeine, glucocorticoid medication, mineralocorticoid agents, and theophylline,[30,32,62,70,81] usually are of limited benefit.

The injection of autologous blood into the lumbar epidural space, the so-called epidural blood patch, is the initial treatment of choice.[5,15,49,70,84] Nosik[52] and Gormley[26] were the first to report the use of an epidural blood patch for the treatment of intracranial hypotension. They injected only 1 to 3 ml of blood and achieved excellent results. Since then, however, it has been shown that the success rate of the epidural blood patch procedure increases with the volume of blood used, from approximately 80% after injection of 10 to 15 ml of blood to greater than 95% after injection of 20 ml of blood.[12,26] Relief of symptoms often is almost instantaneous, and this is likely related to the acute compression of the thecal sac resulting in an increase of CSF pressure. In addition to this acute effect, an epidural blood patch can provide long-term relief by forming a dural tamponade that permanently seals the CSF leak. Nevertheless, in patients with spontaneous intracranial hypotension the benefit of performing the initial epidural blood patch procedure often is only temporary. The epidural blood patch procedure can then be repeated, several times if necessary, and a larger volume of blood, up to 30 ml, may be considered for injection.[15] An important limitation of a large-volume blood patch is the development of back and leg pain. Epidural blood patches have been shown to spread within the epidural space over many spinal levels,[80,82] and the initial epidural blood patch generally is placed in the lumbar spine regardless of the level of the CSF leak. Therefore, it may not be unreasonable to proceed with this treatment without undertaking any spinal neuroimaging studies to localize the level of the leak. However, if lumbar epidural blood patches fail to provide relief, the next step often is the placement of an epidural blood patch directed at the exact site of the CSF leak, which usually is at least several vertebral levels higher than the lumbar spine.[70,71] Thoracic or cervical epidural blood patches are associated with a higher risk,[14] and institutional experience with such directed epidural blood patches may be limited. It has been our experience that the directed epidural blood patch is more successful than the lumbar epidural blood patch, particularly in providing long-term relief.

When the epidural blood patch procedure is ineffective, epidural or intrathecal saline infusion may be considered,[13,15,24,49,62,70] but such infusions should not be expected to seal the CSF leak. Percutaneous CT-guided placement of fibrin glue into the epidural space may also be an option. This technique has been used successfully in patients who developed postoperative spinal CSF leaks.[57]

Surgical repair of spinal CSF leaks is a safe and generally effective treatment of spontaneous intracranial hypotension in selected patients.[34,59,70,71,72,83] Surgery is reserved for patients with persistent symptoms in whom there is evidence of a structural abnormality or focal CSF leak identified on their neuroimaging studies. Different types of leaking spinal meningeal diverticula have been found at surgery, and most of these can be ligated safely.[70,71,72] However, in the majority of surgically treated patients a structural abnormality is not demonstrated on their preoperative neuroimaging studies; a focal CSF leak, however, is often observed, and when such a leak is carefully explored, a clear structural cause of the leak is found only rarely.[71] In these cases, a small dural tear or "blow out" of a small meningeal diverticulum is suspected. Nevertheless, the results of packing of the epidural space and nerve root with fibrin glue at the level of the focal CSF leak have been satisfactory.[71] I recommend using muscle and gel-foam, and usually apply fibrin glue as well, in such surgically treated cases.

We have observed transient intracranial hypertension associated with papilledema and retinal hemorrhage following successful surgical treatment of a chronic spontaneous spinal CSF leak.[70,71] In this case, papilledema resolved spontaneously over a 1-month period. This is an uncommon complication, probably related to the sudden interruption of the abnormal CSF resorption pathway that had been present for many years.

Abbreviations used in this paper: CSF = cerebrospinal fluid; CT = computerized tomography; MR = magnetic resonance.