Orthostatic headaches caused by CSF leak but with normal CSF pressures

The normal limits of CSF pressure are approximately 65 to 195 mm of water or CSF.¹ Low CSF pressures are often associated with orthostatic headaches (headache in upright position relieved by recumbency). This may be associated with one or more of the following manifestations: nausea, emesis, pain or tight feeling of the neck, dizziness, horizontal diplopia (caused by unilateral or bilateral sixth nerve palsy), photophobia, blurred vision, binasal visual field defects, and aberrations in hearing.^2-4 The cause is CSF leak or CSF shunt overdrainage. MRI findings in this disorder include pachymeningeal gadolinium enhancement, subdural fluid collections, appearance of sagging of the brain including descent of cerebellar tonsils (sometimes resembling type I Chiari malformation), small or obliterated prepontine cisterns, small subarachnoid cisterns, flattening of the optic chiasm, and effacement or obliteration of perichiasmatic cistern.^5-10 Lumbar CSF analysis frequently shows a variable increase in protein and variable pleocytosis at some point in the illness. Sometimes an increase in erythrocyte count or even xanthochromia may also be noted. In some patients, particularly in the early stages, CSF analysis may reveal normal results. CSF cytology and bacteriology are always normal, and CSF glucose concentration is never low. The pressures are typically atmospheric or very low, usually considered to be below 30 or 40 mm of water.^1,3

We report seven patients with the typical clinical and imaging features of the syndrome of orthostatic headaches and diffuse pachymeningeal gadolinium enhancement (DPGE) resulting from CSF leak or CSF shunt overdrainage in whom CSF pressures were consistently within normal limits.

Patients. We studied 40 consecutive patients with the syndrome of orthostatic headaches and DPGE. All but seven patients had low or unmeasurable CSF opening pressures. In these seven patients the following features were present: all had orthostatic headaches relieved by recumbency; all had DPGE; all had undergone multiple CSF examinations while recumbent and relaxed; the CSF opening pressures were consistently within normal limits in all the taps of all the patients; five patients had documented evidence of CSF leak and two patients had overdraining CSF shunts; and all but one patient, who refused treatment, responded to surgical treatment or epidural blood patch with complete resolution of their symptoms and the related MRI abnormalities.

Results. There were five men and two women, ranging in age from 24 to 49 years. The clinical features, CSF findings, and MRI abnormalities are summarized in tables 1, 2, and 3.

Clinical features. Postural headache was the main complaint in all patients. In five patients, there was no headache in recumbency, whereas severe headache was present when the patients were upright. In two patients some lingering headache was present even when recumbent but was much less intense. One patient had horizontal diplopia related to unilateral sixth nerve palsy. Postural posterior neck pain was noted in two patients and change in hearing in one. One patient had noted transient right facial numbness and weakness for a period of 3 to 4 days and transient right upper-limb radicular symptoms in C-5 and C-6 root distributions.

CSF findings. All patients had undergone multiple CSF examinations ranging from two to four lumbar punctures. The overall range of opening CSF pressure was 65 to 140 mm of water. At no time in any of the patients were CSF pressures of less than 65 mm of water recorded, and in more than one-half of the patients the opening pressures were consistently higher than 90 mm of water. The spinal fluid was usually clear, but in four patients some of the taps showed xanthochromic fluids. CSF protein was elevated in at least one of the taps of all the patients except one. A primarily lymphocytic pleocytosis was noted in at least one of the CSF analyses of all but one patient.

MRI, myelography, CT-myelography, and radioisotope cisternography. The cardinal MRI findings in these patients consisted of DPGE without any evidence of leptomeningeal enhancement (all patients), imaging evidence of descent of the brain or "sagging brain" (five patients), or subdural fluid collections (one patient). Spinal MRI in one patient with upper thoracic leak showed abnormal signal in left T-1 neural foramen. CT-myelography was performed in four patients and demonstrated the location of the leak in each of them. Indium-111 radioisotope cisternography had been carried out in five patients. The study was considered abnormal in all in that the tracer never reached the convexities. Only in three of these five patients, however, did the study demonstrate the level of the CSF leak.

The etiology of the syndrome was overdraining CSF shunt in two patients and CSF leak in the remaining five. The site of the leak was at the cribriform plate in one patient, at lumbar thecal sac in one patient, and at thoracic level in three patients.

Treatment and outcome. The two patients with overdraining CSF shunts responded to shunt revisions with complete resolution of the symptoms and the meningeal abnormalities on MRI. One patient responded to epidural blood patch, but the procedure had to be performed twice. One patient refused invasive treatment and improved with bed rest. He also credited the use of steroids, which had been prescribed by his local physician. His symptoms have fluctuated. Although not asymptomatic, he remains functional and has been able to maintain his employment. The patient with leak from the cribriform plate responded to transnasal mucosal repair of roof of the ipsilateral ethmoid sinus. His postural headaches and MRI abnormalities completely resolved. There was a lingering tension-type headache for several months, but this gradually diminished. Two patients underwent open surgical repair of meningeal diverticula with complete resolution of the headaches and disappearance of meningeal enhancement.

Illustrative case. In October 1994, a 24-year-old, previously healthy man developed postural occipital headaches and posterior neck pain brought about by standing or sitting that were relieved by recumbency. These were associated with orthostatic nausea and sometimes emesis. Several days later he experienced a 3- to 4-day episode of numbness and sagging of the right side of the face. Later he had intermittent paresthesias in the right C-5 and C-6 root distributions for 2 weeks. Head CT showed excessive enhancement of the tentorium. CSF examination on November 22 showed opening pressure of 104 mm of water, no leukocytes, 3 erythrocytes/mm³, 31 mg/dL protein, 60 mg/dL glucose, and negative cytology and microbiology. Cervical spine MRI was normal. Head MRI showed DPGE.

In mid-December 1994, the patient was seen at the Mayo Clinic. His orthostatic headaches, neck pain, nausea, and occasional emesis had persisted. When recumbent, he had no complaints. The neurologic examination showed no deficits. MRI of the head showed DPGE (figure 1A). Tips of the cerebellar tonsils were at the foramen magnum(figure 1B). Spinal fluid examination revealed an opening pressure of 102 mm of CSF, clear and colorless fluid, 8 leukocytes per mm³, no erythrocytes, 65 mg/dL protein, 60 mg/dL glucose, and negative CSF cytology and bacteriology. Indium-111 radioisotope cisternography demonstrated an abnormal collection of radioisotope at about the T-12 to L-1 level on the right, which was thought to be within a nerve root sleeve(figure 2). Sequential imaging showed lack of progression of radioactivity over the convexities consistent with CSF leak. Iopamidol myelography/CT-myelography showed epidural contrast material extending from the T-4 to L-2 levels, but most densely along the right T-10 and T-12 nerve roots, suggesting a CSF leak in this region (figure 3).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1. MRI of the head when the patient was symptomatic with orthostatic headaches. (A) Contrast-enhanced coronal view shows thin but diffuse pachymeningeal gadolinium enhancement. (B) Unenhanced sagittal view shows descent of the brain with the cerebellar tonsils at the level of the foramen magnum.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2. Indium-111-DTPA cisternogram posterior view (P) demonstrating activity on the right side at approximately the T-12 to L-1 level (arrow).

• Download figure
• Open in new tab
• Download powerpoint

Figure 3. Iopamidol CT myelography demonstrating contrast material within the spinal subarachnoid space (small arrow) and the contrast material that has leaked outside the dura (large arrow).

In late December 1994, the area was surgically explored with findings of arachnoid diverticula through thinned dural sleeves of the T-11 and T-12 nerve roots. A repair was accomplished. On follow-up elsewhere, there was complete resolution of symptoms and MRI abnormalities.

Discussion. The rate of CSF formation in adults is approximately 0.35 mL/min or approximately 500 mL/d.¹¹ In horizontal position, the lumbar, cisternal, and presumably the intracranial or vertex CSF pressures are equal and normally measure 60 to 180 mm of water.^12,13 In erect posture, these pressures diverge and vertex CSF pressure becomes negative. CSF pressure and volume are exponentially related.^1,12 Orthostatic headache is shown to be induced when approximately 10% of the estimated total CSF volume is withdrawn.¹³ The buoyant action of CSF reduces the effective weight of the brain to less than 50 grams, which still must be supported within the cranium by the tethering veins that suspend the brain to the large dural sinuses, the skull base, the cerebellar tentorium, and large vessels of the base. Traction on anchoring vascular structures of the brain (which are pain-sensitive) can provoke headaches. According to the Monro-Kellie hypothesis, with CSF volume loss in the presence of intact skull a compensatory vasodilatation and venous hypervolemia occur.⁶ However, this ordinarily does not prevent sagging of the brain.

The presumed mechanism of the headaches and related imaging abnormalities have been previously discussed.^6,9,10 However, the mechanism of the typical syndrome in the presence of normal CSF pressures defies traditional explanations.

We hypothesize that all CSF leak headache syndromes are essentially "low CSF volume" syndromes rather than necessarily "low CSF pressure" syndromes. For the brain to remain afloat, the CSF volume must be adequate intracranially, otherwise the brain will sag in the erect posture and will exert traction over its suspending pain-sensitive structures. Similarly, traction on some of the cranial nerves may lead to additional symptoms. The descent of the brain seen on MRI likely would be more frequent and prominent if MRI could have been obtained with the patient in the upright position.^8,10

The flow characteristics of the leak (rate of the leak at a given pressure and volume) are important. Certain volume loss may occur without a "low" pressure. When CSF leak is free-flowing, both CSF volume and pressure are low. On the other hand, a check-valve phenomenon might occur when a pressure- or volume-dependent CSF leak allows CSF leak only at higher volumes or pressures. Therefore, with a lowered CSF volume an equilibrium may be reached when CSF pressures are maintained at levels that are still within normal limits. Variability of CSF pressure in different individuals should also be noted. A pressure that might be normal for one person could be low for another.

The immediate therapeutic effect of epidural blood patch is not "plugging the leak," but replacing some of the volume by displacing and compressing the dural sac. A more delayed effect, however, undoubtedly occurs if there is occlusion of the leak. This explains why some of the blood patches are not durable and why some patients may need more than one blood patch. This phenomenon may occur more frequently when the level of blood patch has been distant from the level of the leak. In our experience, occasional patients have required up to three or four blood patches before a "cure" is achieved.

In the syndrome of CSF leak and orthostatic headaches, loss of CSF volume is responsible for the symptoms. The pressure, like the symptoms, is a dependent variable, whereas volume is the independent variable. We believe that in the presence of clinical and imaging features of the syndrome, a normal CSF pressure should not discourage the clinician from searching for a source of CSF leak.