CSF hypovolemia vs intracranial hypotension in “spontaneous intracranial hypotension syndrome”

Patients and methods.

Results.

Radiographic findings are summarized in table 2. T1-weighted brain MRI with Gd enhancement demonstrated characteristic intense pachymeningeal enhancement due to venous engorgement9 in all patients. In Patient 10, a bilateral subdural hematoma was found. No organic intra-axial brain lesions were detected in any patients.

In five patients, sagittal T1-weighted spinal MRI clearly demonstrated downward displacement of the cerebellar tonsil (figure 1A). Sagittal T2-weighted spinal MRI revealed hyperintense signals in the extradural space showing extradural fluid collection, that is, leaked CSF into the extradural space, and dilated epidural veins in all patients. The premedullary subarachnoid space was narrow (see figure 1B). In the magnified image, the subarachnoid space was compressed and obscured by the leaked CSF. The normal subarachnoid space and the leaked CSF were divided by the hypointense linear signal of the dura mater. In all patients, we could find the portion where the dura mater that delineated the outside of the subarachnoid space came into contact with the inside of the abnormal hyperintense signal, lining the leaked CSF (see figure 1C). Dura mater that draws a border between the subarachnoid space and an abnormal hyperintense signal is the hallmark of CSF leakage into the extradural space. The amount of leaked CSF varied according to the patient; for example, it existed in the entire extradural space caudally from the second cervical spine in Patient 8 and in the ventral extradural space locally from the seventh cervical spine to the second thoracic spine in Patient 5. However, leaked CSF in the lower cervical extradural space was common in all patients (see table 2). Fluid collection in the retrospinal region between C1 and C2 was detected in only three patients (Patients 4, 8, and 9) (see figure 1, A and B), although Yousry et al.7 emphasized its diagnostic importance.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1. Sagittal MRI of the cervical spine in Patient 1, a 36-year-old woman with severe orthostatic headache and stiff neck (A through C), and in a 56-year-old man with cervical spondylosis (D). (A) T1-weighted (repetition time/echo time [TR/TE] = 500/20 ms) MRI shows downward displacement of the cerebellar tonsils (large arrow), indicating the sagging of the brain. (B) T2-weighted (TR/TE = 4,000/90 ms) MRI shows ventral and dorsal extradural fluid collection (arrows) with a more hyperintense signal than that of CSF. The fluid indicates CSF leaked into the extradural space. The subarachnoid space is compressed and effaced by the extradural fluid collection. The epidural veins are dilated (arrowheads), and the premedullary subarachnoid space is narrowed. (C) Magnified image of (B). The hypointense dura mater (small arrows) separates the leaked extradural fluid collection (large arrows) from the subarachnoid space (arrowheads). The dura mater runs under the extradural fluid. (D) An example of hypointense linear artifacts made by CSF circulation (arrows) in T2-weighted (TR/TE = 4,000/90 ms) spinal MRI. In such a situation, the hypointense dura mater always runs outside the hyperintense signal, that is, in the subarachnoid space.

Axial T1-weighted cervical MRI with Gd enhancement showed dilated epidural veins as hyperintense lesions located in the high cervical portion in all four patients examined (figure 2A). The dilated epidural veins compressed the subarachnoid space (C3 level), but neither lesion compressing the spinal cord nor disk disease was present. Postmyelogram CT scans in Patient 7 also showed dilated epidural veins that compressed the subarachnoid space and extravasation of the contrast medium into the ventral extradural space at the C3-7 level (see figure 2B), and along the nerve sleeves at the C7 level (figure 3A). Postmyelogram CT scans in Patient 8 showed extrathecal collection of contrast medium at the C2 level, suggesting CSF leakage from a perineural cyst there (see figure 3B). Radionuclide cisternography in Patient 8 obtained 2.5 hours after injection demonstrated spinal CSF leakage at the upper cervical level and early accumulation of radioactivity in the kidneys. Radionuclide cisternography in Patient 3 obtained 2.5 hours after injection showed the accumulation of radioactivity into the intracranial arachnoid cyst, but it remained there in further sequential scanning. We could not detect the spinal CSF leak in this patient. Radionuclide cisternography in Patients 2 and 7 did not show any CSF leaks.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2. Dilated epidural veins in Patient 6. (A) Axial T1-weighted (repetition time/echo time [TR/TE] = 500/20 ms) spinal MRI with Gd enhancement at the C4 level shows dilated epidural veins as hyperintense signals that compress the dural sac (arrowheads). (B) Axial postmyelogram CT scan at the C3 level shows dilated epidural veins that compress the dural sac as negative shadows (arrowheads). Extravasation of the contrast medium into the ventral extradural space is also demonstrated as a higher-intensity signal than that of the subarachnoid space (arrows).

• Download figure
• Open in new tab
• Download powerpoint

Figure 3. Axial postmyelogram CT scans. (A) An axial postmyelogram CT scan in Patient 7 at the C7 level demonstrates extravasation of contrast medium into the ventral extradural space (large arrows), divided by low-density dura mater (small arrows) and along the nerve sleeves (arrowheads). (B) An axial postmyelogram CT scan in Patient 8 at the C2 level demonstrates an extrathecal collection of contrast medium, suggesting that CSF leaked from a perineural cyst there (arrow).

The largest axial area of a dilated epidural vein in seven patients was 40.84 ± 15.12 mm² (mean ± SD) and that of seven control patients was 10.81 ± 8.54 mm². The axial areas of epidural veins in so-called SIH syndrome were larger than control values (p < 0.001) (figure 4A). With a cutoff value of 22 mm², the sensitivity was 92.9% and the specificity was 85.7%. This fact indicated that the vasculature in so-called SIH syndrome is more engorged than normal. The number of vertebral bodies over which leaked CSF spread correlated with the largest axial area of a dilated epidural vein (R = 0.66, p < 0.01) (see figure 4B). This fact indicated that the amount of leaked CSF (on reflection, the degree of CSF depletion) correlated with the engorgement of the vasculature and indirectly supported the Monro–Kellie doctrine in this situation.

• Download figure
• Open in new tab
• Download powerpoint

Figure 4. Dilatation of the epidural veins. (A) The largest axial area of the epidural veins of seven patients with so-called spontaneous intracranial hypotension (SIH) syndrome was 40.84 ± 15.12 (mean ± SD) mm², whereas that of seven sex- and age-matched control patients with cervical myelitis was 10.81 ± 8.54 mm². Patients with so-called SIH syndrome have more dilatation of the epidural veins than do control patients with cervical myelitis (*p < 0.01). SIH = patients with so-called SIH syndrome; control = sex- and age-matched control patients with cervical myelitis. (B) The number of vertebral bodies over which the leaked CSF correlates with the largest axial area of the epidural veins (p < 0.01). The volume of leaked CSF reciprocally correlates with the volume of CSF in the subarachnoid space. Therefore, the number of vertebral bodies over which the leaked CSF has spread is an indirect index of the volume of the CSF in the subarachnoid space. Conversely, the largest axial area of the epidural veins is an index of venous engorgement. The significant correlation between the number of vertebral bodies over which leaked CSF has spread and the largest axial area of the epidural veins demonstrates indirectly that the Monro–Kellie doctrine is actually valid in the situation of so-called SIH syndrome.

Discussion.

SIH syndrome was first described in 1938 by Schaltenbrand,10 who considered three causes: CSF leak, reduced CSF production, and increased CSF absorption. With the advance of diagnostic technology, CSF leakage into the extradural space can be detected by radioisotope cisternography,4 and CSF leakage is now thought to be the cause of so-called SIH syndrome. However, the false-negative rate of radioisotope cisternography was about 30% and postmyelogram CT scan was more sensitive.11 In this study, spinal MRI detected CSF leakage in 10 of 10 patients examined, but conventional radioisotope cisternography detected the CSF leakage in only 1 of 4 patients. In addition, spinal MRI is noninvasive and easy to perform compared with radioisotope cisternography and postmyelogram CT.

The headache of so-called SIH syndrome is thought to originate from the downward displacement of the brain. In the upright position, the brain is kept afloat in the cranium by the buoyant action of the CSF and the anchoring effect of vascular structures in the cranium. If the buoyant action of the CSF decreases, the burden of the vascular structures increases, and these structures are then subjected to traction and distortion. Because these vascular structures in the cranium are pain sensitive,12 orthostatic headache results. Does intracranial hypotension, from which the name SIH syndrome originated, really cause orthostatic headache? It should be noted that 11 of 60 (18%) patients including 2 of our patients with so-called SIH syndrome had normal CSF pressure5,11⇓ and that some elderly patients had diffuse pachymeningeal enhancement, low CSF pressure, and subdural fluid collection but did not have orthostatic headache.8 This suggests that factors other than intracranial hypotension play a role in orthostatic headache. The CSF pressure is determined by the relationship between the CSF volume and the subarachnoid space volume. If CSF leakage causes CSF hypovolemia but a compensatory decrease of the subarachnoid space also occurs, the equilibrium may be maintained and the CSF pressure may not decrease. In cases of patients with typical orthostatic headache but normal CSF pressure, the buoyant action of the CSF is insufficient to keep the brain afloat because of the CSF leak. Inversely, in cases of elderly patients with low CSF pressure but no orthostatic headache, the CSF pressure decreases owing to the CSF leak, but the CSF volume is sufficient to maintain the buoyant action that keeps the brain afloat because of the relatively large CSF volume and relatively small brain volume in such patients.13 In the horizontal position, the lumbar, cisternal, and intracranial or vertex CSF pressures are equal. However, in the upright position, these pressures change rapidly: The vertex CSF pressure becomes negative, and the lumbar CSF pressure becomes relatively positive.8 The brain would not be caused to sag down by this CSF pressure gradient in the upright position. It has been also reported that orthostatic headache was induced when approximately 10% of the estimated total CSF volume was withdrawn.14 We must take into account the critical role of the buoyant action in preventing the brain from sinking. The buoyant action depends on the CSF volume, not the intracranial pressure. We think that the depletion of CSF, the CSF hypovolemia, is the essential cause of orthostatic headache, which is the cardinal symptom of the so-called SIH syndrome, and that intracranial hypotension is only a result of the CSF hypovolemia, as Mokri8 has postulated.

Imaging abnormalities of so-called SIH syndrome such as subdural hematoma and hygroma,4 dilatation of the cervical epidural vein,5,15⇓ diffuse pachymeningeal Gd enhancement, sagging of the brain, effacement of the prepontine cistern,9 and false pituitary tumor16 have been reported. In our patients, we detected dilatation of the cervical epidural veins and diffuse pachymeningeal Gd enhancement in all patients. As mentioned above, it is suggested that these abnormalities originate from the CSF depletion due to CSF leakage, not from the intracranial hypotension. To explain these abnormalities, the Monro–Kellie doctrine was introduced.9 According to the doctrine, with an intact skull, the sum of the volume of the brain plus the CSF volume plus the intracranial blood volume is constant.17 Therefore, an increase in one should cause a reduction in one or both of the remaining two. The reduction of CSF volume requires a compensatory increase of the brain volume and intracranial blood volume, primarily in the venous system. The venous system is chiefly affected because the brain volume changes only a little.17 We can use the dilatation of the cervical veins and diffuse pachymeningeal Gd enhancement in the brain MRI as markers for venous engorgement. We demonstrated that the cervical epidural veins of patients with so-called SIH were larger than those of control patients (p < 0.001). The cutoff value of 22 mm² of the largest axial area of the cervical epidural veins that we advocated as an index of significant dilatation was reliable, showing a sensitivity of 92.9% and specificity of 85.7%.

However, it has not been verified whether the Monro–Kellie doctrine is really valid in the situation of so-called SIH syndrome. In this study, we demonstrated that the number of vertebral bodies over which the leaked CSF spread, that is, an index of the CSF leakage and volume, significantly correlated with the largest axial area of dilated epidural veins, that is, an index of intracranial blood volume. Although indirectly, we are the first to demonstrate that the Monro–Kellie doctrine is valid in so-called SIH syndrome. Therefore, the Monro–Kellie doctrine is a convincing basis of the pathomechanism of so-called SIH syndrome.

Another important finding of the imaging studies is that the CSF leaked into the extradural space was detected in all 10 patients as a hyperintense signal outside the linear hypointense signal indicating the dura mater by spinal MRI. As we detected the CSF leak in only one of four patients by conventional radionuclide cisternography but in all four patients by spinal MRI, spinal MRI appears to be more sensitive than conventional radionuclide cisternography for detecting CSF leaks. We sometimes confuse hypointense linear artifacts caused by the CSF circulation in the subarachnoid space with the CSF leaked into the extradural space. However, we can easily distinguish these two conditions by the hypointense dura mater delineating the subarachnoid space inside the hyperintense region (see figure 1C). In contrast, the dura mater is always outside of the hyperintense region in cases of hypointense linear artifacts (see figure 1D).

With an intact skull, CSF hypovolemia is indicated by venous engorgement, which can be detected as diffuse pachymeningeal Gd enhancement and dilatation of the cervical epidural vein. Pachymeningeal enhancement on MRI can also be caused by subdural hemorrhage, dural metastasis, infection, and inflammatory diseases.18 However, these conditions do not involve dilatation of the cervical epidural veins, and CSF hypovolemia does not accompany these conditions. Detecting only diffuse pachymeningeal Gd enhancement is insufficient, and both diffuse pachymeningeal Gd enhancement and dilatation of the cervical epidural veins are required to verify CSF hypovolemia in so-called SIH syndrome. We emphasize that in cases of orthostatic headache without a previous history of dural tear, brain and spinal MRI with and without Gd enhancement gives us an important clue to diagnose so-called SIH syndrome. Although MRI cannot detect the site of the CSF leak precisely, most symptoms of so-called SIH syndrome usually resolve spontaneously or with strict bed rest. If the symptoms are persistent and refractory and surgical treatment is needed, postmyelogram CT should be employed to detect the involved site.

Recently, a relationship between meningeal anomalies and CSF leakage has been reported, but meningeal anomalies are not always detected.7,11,19⇓⇓ Where is the site of CSF leakage in the cases of patients who are not confirmed to have the meningeal anomalies? In all of our patients, a hyperintense signal in the extradural space indicated leaked CSF around the cervicothoracic junction. In most of the patients without meningeal diverticula, the sites of the CSF leaks were the low cervical portion or the cervicothoracic junction.11,19⇓ Patients with CSF leaks secondary to cervical bone spur have been reported.20,21⇓ Minor trauma to the meninges due to neck movement may cause CSF leakage, leading to so-called SIH syndrome.

Similar to post–lumbar puncture headache,22 which has the same pathomechanism, the so-called SIH syndrome is more common in women, with a female-to-male ratio of 3:1.23 All of our patients were women. There are no differences between men and women in CSF pressure, conductance of the CSF outflow pathway, or CSF production rate.24 Nevertheless, the relative CSF volume to the brain in women is smaller than that in men, although the brain volume of women is also smaller than that of men.13 Therefore, the buoyant action would be smaller in women than in men.

The issues concerning so-called SIH syndrome are easily understandable if we consider that the decrease of the CSF buoyant action due to CSF hypovolemia, not intracranial hypotension, plays a crucial role.