Abstract
CSF hypovolaemia has been well recognized as a cause of orthostatic headache (1–3). On brain MRI, it has several typical MRI features, including diffuse pachymeningeal gadolinium enhancement (DPMGE), subdural fluid collection, and descent of the brain (1, 2, 4). In addition, cases with typical clinical features and CSF leaks, but without DPMGE on brain MRI have been reported (1, 2, 5). However, the causative factors related to the development of typical MRI changes in CSF hypovolaemia, particularly DPMGE, are still unclear.
To determine these factors, we have compared the clinical and imaging features of CSF hypovolaemia with and without DPMGE on brain MRI.
Subjects and methods
We studied 38 consecutive patients with CSF hypovolaemia who were admitted to the Asan Medical Centre between April 1995 and July 2003. Of these 38 patients, 30 had been included in the previous report (2). Patients included in this study met at least two of the following three criteria: orthostatic headache relieved by recumbency, low CSF pressure (<60 mmH2O), and DPMGE on brain MRI. Patients with evidence of diseases, such as infection, malignancy, neurosarcoidosis, and rheumatoid disease were excluded from this study.
Of the 38 patients, seven did not undergo brain MRI and were excluded. The remaining 31 patients, all of whom underwent brain MRI with gadolinium-enhanced T1-weighted scan, were the subjects of this study. Patients were categorized as having CSF hypovolaemia with DPMGE on brain MRI (n = 22) or as having CSF hypovolaemia without DPMGE on brain MRI (n = 9).
Headache severity was evaluated on the basis of whether headache was severe enough to make the patient bedridden for >90% of the daytime or not. CSF opening pressure was classified as normal (≥60 mmH2O), mildly decreased (<60 mmH2O and ≥30 mmH2O), or markedly decreased (<30 mmH2O). Among the 31 patients, 29 underwent CSF examination, of whom 4 had variable pressure at different taps (2). Therefore, CSF opening pressure was analysed in the remaining 25 patients.
Of the MRI abnormalities in CSF hypovolaemia, engorged cerebral venous sinuses, pituitary enlargement, and subdural fluid collections were compared between the patients with DPMGE on brain MRI and those without. Twenty-five patients underwent radioisotope cisternography (RIC). Paradural uptake of tracer on RIC suggesting CSF leaks was categorized as involving one or multiple levels of the spine. Twenty-five patients received epidural blood patch (EDBP). The efficacy of EDBP was assessed by each patient's subjective response 4 weeks later and scored as complete resolution of headache, partial resolution, or persistent headache.
For statistics, we used the Mann–Whitney U-tests and χ2 tests for the comparison of demographic, clinical, and imaging features between the two groups. All statistical tests were performed with the use of the SPSS program (version 11.5); P-values < 0.05 were defined as statistically significant.
Results
The demographic, clinical and imaging characteristics of the patients are summarized in the table. Of the 31 patients, 22 (71%) showed DPMGE on brain MRI and 9 (29%) patients did not. Of the 8 patients who were enrolled over the last three years, 4 (50%) did not have DPMGE on brain MRI. There were no significant differences between the two groups in mean age, gender distribution, frequency of most severe headache, CSF opening pressure and time interval between headache onset and brain MRI (Table 1).
Demographic, clinical, and imaging features of the patients
RIC, radioisotope cisternography; EDBP, epidural blood patch; C, complete; P, partial.
On brain MRI, the frequency of engorged cerebral venous sinuses, pituitary enlargement, and subdural fluid collections was not significantly different between the two groups. There were also no significant between-group differences in frequency of paradural uptake of tracer on RIC and its multiplicity, and in the efficacy of EDBP.
Discussion
In this study, we found no significant differences in clinical and imaging characteristics between the patients having CSF hypovolaemia with DPMGE on brain MRI and those not having DPMGE on brain MRI. In particular, the level of CSF opening pressure, and the time interval between headache onset and brain MRI did not correlate with the appearance of DPMGE. These findings suggest that the development of DPMGE on brain MRI may depend primarily on a certain individual threshold, including the amount of CSF leakage and the degree of decrease in intracranial pressure. Our results are consistent with the previous study suggesting that the absence of DPMGE in patients with CSF leaks may be due to a loss of volume and change in hydrostatic pressure quantitatively insufficient to result in the significant venous congestion necessary for the appearance of DPMGE (5). In addition, these results suggest that some patients may have reset their CSF dynamics to maintain low pressures without ongoing CSF leakage (6).
According to the Monro-Kellie doctrine (7), the loss of CSF volume has to be compensated. This is accomplished by the subdural fluid collections and an increase in intracranial blood volume. The latter is primarily reflected on venous system, leading to DPMGE, engorgement of cerebral venous sinuses and pituitary enlargement. In this study, the frequency of engorged cerebral venous sinuses, pituitary enlargement, and subdural fluid collections was not significantly different between the two groups. These findings also suggest that the critical threshold of CSF volume depletion required for the occurrence of various MRI abnormalities is different in each patient. However, pituitary enlargement and subdural fluid collections were observed only in patients with DPMGE, although the differences between the two groups did not reach statistical significance. Therefore, further evaluations with a larger number of patients are required.
We also found that the frequency of DPMGE absence on brain MRI in patients with CSF hypovolaemia was higher in recent years than in the prior five years. This may be due to our better understanding of, and greater experience with this syndrome, which may have reduced the rate of under-diagnosis of CSF hypovolaemia due to normal brain MRI.
The DPMGE on brain MRI is a very important finding to diagnose CSF hypovolaemia. Its absence from many patients with CSF hypovolaemia suggests that additional diagnostic tools, such as RIC or spine MRI, may be needed to confirm and adequately manage this syndrome. To date, however, it is not known which of these imaging modalities is superior (8), indicating the need for additional prospective studies to settle this issue.