Abstract
Objective: To investigate clinical, MRI, and radioisotope findings and therapeutic outcome of the syndrome of CSF hypovolemia.
Methods: Retrospective review was performed of 30 consecutive patients (10 men, 20 women; mean age 37 years) with the syndrome of CSF hypovolemia.
Results: All patients had an orthostatic headache, which was alleviated to a variable extent on recumbency. Additional clinical symptoms included nausea, dizziness, neck stiffness, blurring of vision, tinnitus, plugged ear, hearing difficulties and radicular pain of the arm. Eighty-two percent of the patients had CSF opening pressure less than 60 mm H2O, 59% had CSF pleocytosis, and 95% had increased CSF protein. Brain MRI showed diffuse pachymeningeal gadolinium enhancement on T1-weighted image in 83%, which was seen as hyperintense signals on T2-weighted imaging. Other features included subdural hematoma/hygroma in 17% and descent of the brain in 48% of the patients. Radioisotope cisternographic results identified CSF leakage sites in 52%, most often at the lumbar region. Also observed were limited ascent of the tracer to the cerebral convexity (91%), early appearance of radioisotope in the bladder (65%), and early soft tissue uptake of radioisotope (43%). Epidural blood patches were performed in 23 patients, which produced complete resolution of headaches in 70%. Two patients underwent drainage of subdural hematoma. None died or were disabled during hospitalization.
Conclusions: Patients with CSF hypovolemia frequently have distinct MRI and radioisotope cisternographic abnormalities and often respond favorably to an epidural blood patch.
Spontaneous intracranial hypotension is characterized by orthostatic headache, low CSF pressure, and MRI findings of diffuse pachymeningeal gadolinium enhancement1-7⇓⇓⇓⇓⇓⇓ without previous history of head trauma or lumbar puncture. Spontaneous CSF leakage from a spinal meningeal diverticula or simple dural tear has been suggested as the underlying pathogenic mechanism.3,4,8-12⇓⇓⇓⇓⇓⇓ Recently, various atypical patients have been reported, including those without headache,13,14⇓ those with lingering chronic daily headache,15,16⇓ and those presenting with reversible encephalopathy,17 cervical myelopathy,18 or parkinsonism.19 Patients with the absence of pachymeningeal enhancement on MRI14,20⇓ also were reported. Furthermore, in disagreement with the term “intracranial hypotension,” patients with normal CSF pressure have been described who presented with otherwise typical orthostatic headache caused by CSF leakage.4,14,15,21⇓⇓⇓ Therefore, the term “CSF hypovolemia” recently has been used,14 with which researchers agree because the loss of CSF volume seems to be the key feature explaining the various clinical and imaging features of this syndrome.
Despite the increasing interest and familiarity of this syndrome, frequently in previous literature, few were subjects studied,1-3,9⇓⇓⇓ and some reports included patients who had undergone ventriculoperitoneal shunt or lumbar puncture.4 Moreover, MRI and radioisotope studies were performed only in a few patients.5,7,11,12,20,21⇓⇓⇓⇓⇓ Therefore, studies with more patients with CSF hypovolemia are needed to gain a better understanding of this syndrome. In this study, we review the clinical features, MRI findings, radioisotope results, and therapeutic outcome of 30 patients with CSF hypovolemia.
Subjects and methods.
We retrospectively studied 30 consecutive patients with the syndrome of CSF hypovolemia who were admitted to the Asan Medical Center between April 1995 and January 2000. Patients who were included met at least two of the following three criteria: orthostatic headache, low CSF pressure, and diffuse pachymeningeal gadolinium enhancement on cranial MRI without evidence of diseases such as infection, malignancy, neurosarcoidosis, and rheumatoid disease. A few patients with normal CSF pressure or normal MRI findings (without pachymeningeal enhancement) were included if they had typical orthostatic headaches and if CSF leaks were confirmed by radioisotope cisternography. Orthostatic headache was defined according to the criteria of International Headache Society (1988) as a headache that occurs less than 15 minutes after assuming an upright position and disappears or improves less than 30 minutes after resuming the recumbent position.22 We excluded patients who had a recent history of head trauma and those who underwent lumbar puncture or ventriculoperitoneal shunt.
A detailed medical history was taken on all patients, and all underwent neurologic examination by one of the authors. To collect CSF in the patients with unmeasurable CSF opening pressure, patients were asked to cough or sit up. If these measures failed, we used the negative pressure of a syringe. However, on eight occasions, we could not obtain enough CSF to assess the blood cell count and amount of protein. Brain MRI was used to support the diagnosis. The descent of the brain was assessed on the basis of the incisural line and the foramen magnum line on midsagittal MRI.7 In selected patients, radioisotope cisternography, CT myelography, and spinal MRI were performed to identify the level of CSF leak. Radioisotope cisternography was performed after a lumbar injection of 0.4 mL of 99mTc–diethylenetriamine penta-acetic acid (DTPA) containing 5 to 7 mCi. Images were obtained at 30 minutes and 1, 2, 4, 6, and 24 hours after the injection.
For the initial treatment, supportive measures (bed rest, analgesics, and fluid replacement) were tried in all patients. When these methods failed in 5 to 7 days, epidural blood patches, intrathecal saline injection, or surgical drainage of subdural hematoma/hygroma were performed with the consent of the patients. The efficacy of treatment was assessed according to the patients’ subjective response at 4 weeks after the treatment as complete resolution of headache, partial resolution, or no response.
Results.
Demographic and clinical features.
The patients’ demographic and clinical characteristics are summarized in table 1. There were 10 men (33%) and 20 women with an age range of 25 to 53 (mean 37 ± 8.4) years. Although no patients had an obvious history of head or neck trauma, history taking identified possible factors related to CSF leakage in seven patients, which included chiropractic management (Patient 1), playing golf (Patient 2), vigorous physical activity (Patients 5 and 9), swimming (Patient 18), yoga exercise (Patient 18), and severe cough associated with upper respiratory infection (Patients 16 and 21).
Demography, clinical characteristics, CSF findings, and outcome of 30 subjects
All patients presented with orthostatic headaches. In 12 patients (40%), the orthostatic headache completely resolved with recumbent position, whereas in others, a headache of milder degree persisted. In two (Patients 10 and 21), the typical orthostatic headaches later were replaced with relatively continuous, lingering headache, whereas one (Patient 18) presented with a fluctuating headache from the beginning. The mean interval between the onset of headache and hospitalization was 18 days (range 2 to 75 days). The onset of the headache was sudden in 10 patients (33%) and insidious in others. Twenty-four patients developed headache in less than 30 seconds, whereas others developed headache less than 5 minutes after assuming an upright position. Laughing, coughing, jugular venous compression, and Valsalva maneuver exacerbated their headaches. The intensity of headache varied from moderate to severe degree and tended to decrease in the morning and increase in the afternoon. Six patients (Patients 1, 3, 8, 10, 11, and 15) remained bedridden because of the severe headache. The quality of the headache was described as, “My brain is falling down like a waterfall,” in 11 patients (37%); “My head is being tugged by someone behind,” in 10 (30%); “pulsating” in 4 (13%); “pressing” in 3 (10%); and “splitting” in 2 patients (7%). The location of the headache was holocranial in 15 patients, occipital in 7, bifrontal in 3, occipital and bifrontal in 3, and bitemporal in 2 patients. In addition to the head pain, 12 patients had posterior neck pain.
Other clinical symptoms included nausea in 16 patients (53%), dizziness in 9 (30%), neck stiffness in 5 (17%), plugged ear in 6 (20%), tinnitus in 6 (20%), radicular symptoms of upper extremities in 2, blurring of vision in 1, and hearing difficulty in 1 patient. All of these symptoms were aggravated by upright posture. Patient 10 additionally presented with disorientation, inattentiveness, decreased memory, decreased verbal output, and bradykinesia late in her clinical course, which were caused by the massive bilateral subdural hematomas (see later).
CSF findings.
Of the 30 patients, 28 underwent CSF examination (see table 1). The CSF opening pressure was measured at 30 to 60 mm H2O in six patients (21%) and was unmeasurable in 17 patients (61%). Four patients (Patients 3 and 8 through 10) had variable pressure at different taps (unmeasurable, 140 mm H2O). In one patient (Patient 25) with normal CSF opening pressure, the diagnosis was made by the presence of typical orthostatic headache, diffuse pachymeningeal gadolinium enhancement on MRI, and multiple CSF leaks on radioisotope cisternography.
Seventeen patients had clear CSF in all spinal taps, and 19 patients had at least one clear CSF finding. Three patients (Patients 1, 3, and 24) had xanthochromic CSF in at least one of the taps. In four patients (Patients 1, 4, 6, and 8), the CSF was blood tinged because of traumatic tap. A variable number of erythrocytes were detected in 19 patients (86% of 22 patients who were tested for erythrocytes): up to 9,000 cells/mm3. Thirteen patients (59%) had CSF pleocytosis (>5 cells/mm3), usually of 6 to 50 cells/mm3, most of them being lymphocytes. However, two of them had an exceptionally high number of cells: 130 cells/mm3 in Patient 9, and 161 cells/mm3 in Patient 1. Twenty patients (95%) had an increased level of CSF protein (>40 mg/dL); of these, Patient 4 had the highest level (566 mg/dL). The CSF glucose concentration was normal, and all tests for cancer cells and bacteria gave negative results.
MRI findings.
Brain MRI with gadolinium enhancement was done in 23 patients. Diffuse pachymeningeal gadolinium enhancement was noted in 19 patients (83%). The enhancement was symmetric, linear, and uninterrupted (figure 1A). All enhanced pachymeninges were hyperintense on T2-weighted images (see figure 1B). In one patient (Patient 16), follow-up MRI 27 days later showed increased meningeal enhancement compared with the initial findings.
Figure 1. Patient 16. Axial T1-weighted gadolinium-enhanced MRI shows diffuse pachymeningeal enhancement (A, arrow), which was seen as a high signal intensity on T2-weighted image (B, arrow).
Subdural fluid collections associated with mass effect were detected in four patients, bilaterally in three (Patients 10, 13, and 14), and unilaterally in one (Patient 12). On T1-weighted imaging, the signal intensity of the fluid was homogeneously higher than that of the CSF and slightly lower than that of brain parenchyma (figure 2, A2 and A3), whereas it was high on T2-weighted imaging (see figure 2, A4). In Patients 10 and 13 (see figure 2, B4), there were inhomogeneous subdural fluids containing an admixture of the blood of various stages and septa. In Patient 10, the subdural fluid collections were enlarged and transformed into the signal intensity of subacute hematoma on follow-up MRI 60 days later, which was associated with severe compressive effect on the midbrain and bilateral hemispheres (see figure 2, B1 to B4).
Figure 2. Patient 10. Row A shows MRI findings on admission; row B shows MRI findings 60 days later, when the patient’s neurologic conditions deteriorated. Numbers 1 to 3 indicate T1-weighted images, and 4 indicates T2-weighted image. B1 shows more severe compression of the midbrain, which were caused by a herniated temporal lobe, than the previous one (A1). A2 to A4 show bilateral subdural fluid collections with mild mass effect, which were significantly enlarged from subdural hematomas of subacute stage (B2 to B4).
A sagging of the hindbrain was detected in 11 patients (48%), which was associated with other evidence of brain descent: effacement of prepontine cistern, obliteration of suprachiasmatic structure, and inferior displacement of optic chiasm. In most, the degree of the descent was mild. In Patient 2, there was an air bubble on the left lateral ventricle, which may have been caused by an air sucked into the subarachnoid space through the stylet during the lumbar puncture (figure 3).
Figure 3. Patient 2. T2-weighted axial MRI shows an air bubble in the left lateral ventricle (arrow).
Radioisotope cisternographic findings.
Of the 23 patients who underwent radioisotope cisternography, 21 (91%) showed abnormal findings: CSF leaks in 12 patients (52%) (figure 4, A and B), limited ascent of the tracer to the cerebral convexity in 21 (91%) (see figure 4C), early accumulation of the radioisotope in the urinary bladder in 15 (65%) (see figure 4D), and early soft tissue uptake of radioisotope in 10 patients (43%) (see figure 4E; table 2). The locations of the CSF leaks were as follows: cervicothoracic junction in three patients (Patients 6, 22, and 23), thoracic region in two (Patients 29 and 30), thoracic to lumbar region in two (Patients 24 and 25), and lumbar area in five patients (Patients 8, 17, 19, 26, and 27). Multiple leaks were detected in five patients (Patients 8 and 24 through 27). In two patients (Patients 5 and 16), there was a faint increase in the activity of radioisotope in the posterior portion of the cisterna magna. We could not conclude whether these findings indicate CSF leaks.
Figure 4. Radioisotope cisternography findings. (A) Patient 6. CSF leakage was demonstrated at the level of cervicothoracic junction (arrow). (B) Patient 25. Multiple CSF leaks were demonstrated at the thoracic and lumbar areas (arrows). (C) Patient 22. Limited ascent of the tracer to the cerebral convexity is demonstrated. (D) Patient 8. Early appearance of radioisotope in the bladder (arrow, 1 hour after injection of tracer). (E) Patient 8. Early soft tissue uptake of radioisotope is shown.
Radioisotope cisternographic findings
Other miscellaneous studies.
The CT myelography was done in only one patient (Patient 17). Neither CSF leaks nor underlying anatomic abnormalities predisposing to a CSF leak were detected. Three patients (Patients 8, 27, and 30) underwent spinal MRI, which did not show any structural abnormalities related to CSF leaks.
Treatment and outcome.
Seven patients received supportive measures only (bed rest, analgesics, and hydration). Twenty-three patients underwent epidural blood patches; in five patients, they were done twice. The procedure was accompanied by an intrathecal saline injection in one and by surgical drainage of subdural hematoma in two patients. Overall, the headache resolved completely in 18 patients (60%) and partially in 11 patients (37%). The therapeutic effect of epidural blood patch was immediate in five patients (Patients 7, 17, 19, 20, and 22), whereas it was delayed (>24 hours) in others. The rate of complete resolution of headache (70%) in patients receiving epidural blood patches was high compared with those who received supportive management only (29%). One patient (Patient 8) continued to have headache despite an epidural blood patch. In five patients with multiple CSF leaks (Patients 8 and 24 through 27), a complete resolution of headache was achieved less often (40%) than in those with a single CSF leak (86%). Of the four patients with subdural fluid collection, two received epidural blood patches (which partially alleviated the headache) followed by surgical drainage, which produced complete resolution of headache (Patients 10 and 12). In two patients who received an epidural blood patch only, one (Patient 14) had complete resolution of headache, whereas another (Patient 13) had partial improvement.
Discussion.
In this study, we analyzed a large number of patients with syndrome of CSF hypovolemia. Although patients with a history of trauma were excluded, some patients had a history of chiropractic manipulation, golf exercise, vigorous physical activity, swimming, yoga exercise, or cough, which may have precipitated the CSF leakage. Symptoms other than orthostatic headaches were similar to those from other published series.4,8⇓ Nausea, dizziness, blurring of vision, tinnitus, plugged ear, and hearing disturbances may be caused by the traction on cranial nerves secondary to the descent of brain, although the last three symptoms also may be attributed to the changes in intralabyrinthine pressure caused by alteration of the pressure gradient across the cochlear aqueduct.23,24⇓ Neck stiffness and radicular symptoms may be caused by the traction on cervical spinal nerves from the descent of the cervical cord.
We found that women are affected more often than men for an unknown reason, with a ratio of 2:1. This female predominance was observed in some of the previous studies7,8,25⇓⇓ but not in others.4 The mean age of our patients was 37 years. The absence of the elderly patients may be related to the fact that the brain weight is lower by 7% to 8% in the elderly than in young adults.26 Because a traction of pain-sensitive structures has been considered a pathogenic mechanism for an orthostatic headache,7 decreased brain weight may prevent the brain from sinking and, consequently, development of headache. In our study, most of the patients had CSF pressure less than 60 mm H2O. However, CSF pressure was found to be normal in at least one of the spinal taps in five patients (18%), which may have been caused by intermittent CSF leaks or check valve phenomenon.21 The increased protein (95% of the tested patients) and erythrocyte count (86%) may be attributed to compensatory meningeal vasodilation and subsequent diapedesis of protein and red cells,3,7⇓ whereas CSF pleocytosis (59%) may reflect a local inflammatory reaction at the site of CSF leakage or could be a response to the presence of red cells.3 We thought that CSF xanthochromia (three patients) was caused by high protein level (Patients 3 and 24) or previous traumatic tap (Patient 1).
We noticed that enhanced pachymeninges on T1-weighted MRI were shown as hyperintense signals on T2-weighted image (see figure 1B), an observation that has not been emphasized previously. This phenomenon may be related to increased interstitial fluid of pachymeninges from compensatory meningeal vasodilation or tearing of the blood vessels.7 The subsequent increase of water content in pachymeninges may have caused high signal intensity on T2-weighted image. These MRI findings can be useful in differentiating the syndrome of CSF hypovolemia from idiopathic intracranial pachymeningitis, where the signal intensity of the dura on T2-weighted image is hypointense or slightly hyperintense together with a central hypointensity.27 In our series, pachymeningeal gadolinium enhancement was not observed in four patients (Patient 5, 8, 17, 18; 17%) who otherwise had typical orthostatic headaches, CSF findings, and radioisotope cisternography results. Perhaps, in these patients, the loss of CSF volume and hydrostatic pressure changes may have not been sufficient to result in compensatory meningeal vasodilation.20 In this study, subdural fluid collections were detected in only 17% (four patients), which disagrees with previous studies (56% to 69%).4,9⇓ The fluid collected could be either subdural hematoma or high in proteinaceous material. Considering that the headache resolved only partially with epidural blood patches in three of the four patients with subdural fluid collection, the subdural fluid seems to contribute at least in part to the development of the patients’ headache. It also was the cause of the clinical deterioration in Patient 10.
The radioisotope cisternographic finding of CSF hypovolemia has been studied previously in few patients.4,9,11,12,28⇓⇓⇓⇓ Our study examined the largest number of patients, and our results are of interest. We found that 91% of the patients showed at least one abnormal finding. The most common location of CSF leaks was the lumbar region (42%), followed by the cervicothoracic junction, which is in contrast with previous reports showing that the thoracic region was the most common site of leakage.4,9⇓ Also, multiple locations of CSF leak were detected in as many as five patients (42% of the patients who showed CSF leak). In this group of patients, therapeutic response of epidural blood patch was poor compared with those having a single CSF leak. Other evidence of CSF leak included limited ascent of the tracer to the cerebral convexity and early accumulation of the radioisotope in the urinary bladder. In addition, we made a new discovery that early soft tissue uptake of radioisotope also is suggestive of CSF leak (43% in our study). These results may have been produced by unusually rapid uptake of the radioisotope into the bloodstream through the epidural venous plexus.12,28⇓
Our results demonstrate that an epidural blood patch is an effective and well-tolerated treatment. Its therapeutic effect was either immediate or delayed. In one patient (Patient 14), the intensity of headache transiently increased after the epidural blood patch, although the headache eventually improved in 2 days. This phenomenon might be related to transient overcompensation of previous intracranial hypotension after the epidural blood patch. On the other hand, the incomplete response of some patients likely results from the failure of the blood to patch the hole either because of incorrect identification of the hole or from multiple sites of CSF leakage. Two patients with subdural fluid collections underwent surgical drainage in addition to epidural blood patches, which resulted in complete headache resolution.
Regarding the prognosis, although a patient death was reported previously,29 none of our patients died or were disabled. Although one patient (Patient 10) had a progressive decline in neurologic condition associated with progressively enlarging subdural hematoma, she eventually improved with the treatment. Therefore, prognosis of the syndrome of CSF hypovolemia seems good if appropriately diagnosed and treated. Finally, the limitations of our study should be acknowledged. First, because the patients undergoing epidural blood patches did not agree to undergo further diagnostic studies, it remains unknown whether the CSF leakage resolved in these patients. Second, because the patients were not randomly assigned to the treatment modalities, we could not compare the therapeutic effects between conservative management and epidural blood patch. Last, the incidence of CSF hypovolemia cannot be known precisely with the current design. Further studies are required to answer these questions.
Acknowledgments
Acknowledgments
The authors thank the staff and residents of the Asan Medical Center who took care of the patients.
- Received March 22, 2000.
- Accepted July 20, 2000.
References
Letters: Rapid online correspondence
REQUIREMENTS
You must ensure that your Disclosures have been updated within the previous six months. Please go to our Submission Site to add or update your Disclosure information.
Your co-authors must send a completed Publishing Agreement Form to Neurology Staff (not necessary for the lead/corresponding author as the form below will suffice) before you upload your comment.
If you are responding to a comment that was written about an article you originally authored:
You (and co-authors) do not need to fill out forms or check disclosures as author forms are still valid
and apply to letter.
Submission specifications:
- Submissions must be < 200 words with < 5 references. Reference 1 must be the article on which you are commenting.
- Submissions should not have more than 5 authors. (Exception: original author replies can include all original authors of the article)
- Submit only on articles published within 6 months of issue date.
- Do not be redundant. Read any comments already posted on the article prior to submission.
- Submitted comments are subject to editing and editor review prior to posting.
