Abstract
Certain magnetic resonance (MR) techniques are flow sensitive, and phase-contrast magnetic resonance imaging (PC-MR) has been utilized both qualitatively and quantitatively to evaluate cerebrospinal fluid (CSF) flow in a variety of clinical conditions (1). Although spontaneous intracranial hypotension (SIH) is, by definition, a disorder of CSF homeostasis typically resulting from a spinal CSF leak, PC-MR has been studied very little in this setting. An initial case report in 2004 first documented decreased systolic and diastolic CSF flow volume at the level of the cerebral aqueduct using PC-MR in a patient with SIH (2). Although confined to a single patient, the findings were compelling because normalization was documented following resolution of SIH. A number of years elapsed before a more systematic study of PC-MR in SIH was completed when Hasiloglu et al. showed a change in CSF flow dynamics in 25 SIH patients compared with controls and further related the PC-MR findings to clinical findings and opening pressure (3).
The manuscript entitled Usefulness of Phase-Contrast Magnetic Resonance Imaging for Diagnosis and Treatment Evaluation in Patients with SIH by Tung et al. is a welcome addition to the existing body of literature (4). The discriminating power of some of the measured CSF flow parameters at the level of the cerebral aqueduct rivals that of the more widely accepted anatomic imaging features of SIH. Moreover, there may be potential for PC-MR to identify SIH patients who lack characteristic MR findings such as dural enhancement. The change in CSF flow witnessed with treatment suggests that PC-MR may also be beneficial for monitoring patient progress during treatment for SIH. The authors point out that results with PC-MR are known to vary in subjects in relation to age, sex, body mass index, cardiac function, and technical parameters (i.e. velocity encoding gradient, anatomic site, MR scanner, etc.). As a result, the impact of these variables will need to be more fully defined in SIH patients, as has been conducted for healthy subjects. Appropriately, the authors also recognize multiple potential technical limitations in their study. However, moving forward, a number of considerations unique to the use of PC-MR in SIH patients will need to be considered.
Although correlation of CSF flow dynamics with patient symptoms and validated imaging findings in SIH is useful, a quantifiable relationship with opening pressure would be preferred. No lumbar punctures were performed in the current study to allow for correlation of opening pressure with CSF flow dynamics as measured by PC-MR (4). When patient presentation and imaging findings strongly suggest headache due to SIH, it is increasingly common in clinical practice to forego lumbar puncture for opening pressure measurement. Indeed, this approach is also endorsed in the International Classification of Headache Disorders, 3rd edition (beta version), which states that dural puncture to directly measure CSF pressure is not necessary in patients with positive MR findings (5). However, it would be ideal for future research studies using PC-MR in SIH to report opening pressures as much as possible. Applying PC-MR to SIH is a relatively new application of existing technology. Hasiloglu et al. demonstrated a significant positive correlation between opening pressure and multiple CSF flow parameters (CSF flow volume toward the third ventricle, CSF flow volume toward the fourth ventricle, absolute stroke volume, peak systolic velocity, and peak diastolic velocity) (3). These correlations need to be further validated through the accumulation of more data. Ultimately, PC-MR would be most useful and generalizable if it could offer a reliable estimate of the severity of intracranial hypotension based on extensive benchmark data. The ability to noninvasively estimate opening pressure in SIH would be a real step forward in the management of these patients during and following treatment.
Depending on institutional preference, many patients suspected of having SIH are initially evaluated with computed tomography (CT) myelography. Experience has shown that some of these patients have an occult CSF leak, and others have either slow or fast leakage of myelographic contrast from within the thecal sac, with the presence of extra-arachnoid fluid on spine MR images (MRI) predicting rapidity (6). Additional experience has revealed that SIH patients with longitudinally extensive extradural fluid collections often require digital subtraction myelography for CSF leak localization because of the high flow nature of CSF egress (7). The authors of the current study employed heavily T2-weighted MR myelography as a means of documenting the presence of a spinal CSF leak (4). As an area for future investigation, it would be useful to stratify patients based on the rapidity of CSF leak and then correlate with PC-MR results. Controlling for opening pressure, cranial imaging features of SIH, and symptom severity, do intracranial CSF flow dynamics vary with the location (vertebral level) and rapidity of the spinal CSF leak?
One potential confounding factor for the use of PC-MR in SIH is the impact that brain sagging may have on CSF flow dynamics. The alterations of CSF flow in the current study can largely be attributed to decreased CSF volume and/or pressure along the craniospinal axis (4). Essentially, one might envision a contained system that has a greater capacity resulting in a blunted CSF flow response during cardiac systole and diastole. However, severe cases of SIH with significant descent of the cerebellar tonsils through the foramen magnum could alter flow through the cerebral aqueduct. If the tonsillar descent resulted in effacement of the foramen magnum, the intracranial subarachnoid space could, depending on the severity of herniation, become partially or completely isolated from the spine. Intracranial pressure should then rise to a greater degree during cardiac systole. It would be interesting to learn if this scenario could push CSF flow dynamics in SIH more toward normal because the spinal subarachnoid space has been eliminated as a reservoir, thereby decreasing elasticity of the system as a whole. This would have implications not only for diagnosis but also for monitoring treatment. Comparing pre- and post-treatment PC-MR results would need to take this potential confounding factor into account, as the foramen magnum obstruction would resolve with successful treatment. However, simply reporting a distance for tonsillar ectopia would not be sufficient. Research using PC-MR in Chiari I malformation has shown that CSF flow can better predict symptomatology and response to treatment compared with measurement of tonsillar descent (8,9). Consequently, future studies of PC-MR in SIH should address patients with severe brain sagging causing tonsillar herniation as a separate cohort. In this group, correlation with CSF flow across the foramen magnum may be needed or, at minimum, quantification of the degree of effacement of the subarachnoid space at the foramen magnum.
As an additional consideration in monitoring treatment response, the impact of high volume epidural blood patch could be investigated with PC-MR. Extrinsic compression by epidural blood decreases the volume of the spinal subarachnoid space and the elasticity of the thecal sac. Examining the net result on intracranial CSF flow with PC-MR may offer some insight into which patients will not respond to an epidural blood patch.
Once the PC-MR technique has been more extensively validated, certain logistics need to be considered to allow for general use. Fortunately, PC-MR can be readily performed on most modern high field-strength MR scanners (1.5 or 3 Tesla). Although an MR physicist is not necessary for implementation, some of the commercially available systems may require an additional software site license. Many institutions already have this PC-MR technical capability because of current use in other neurologic applications such as Chiari I malformation and normal pressure hydrocephalus. The sequence used in the current study was approximately eight minutes in length, so a time penalty does exist (4). However, SIH patients constitute a very small percentage of those undergoing MR imaging, so the overall impact on scanner throughput should not be significant.
Well-defined criteria exist for the diagnosis of spontaneous intracranial hypotension, but some cases can be challenging because symptoms are atypical or characteristic imaging findings are absent (10). The study by Tung et al. adds to a limited body of literature showing promise for PC-MR in the diagnosis and monitoring of patients with SIH (4). Additional research needs to be pursued in this area in order to validate PC-MR as a clinically useful tool that can be applied to individual patients. If this occurs, PC-MR could eventually become an integral component of imaging protocols used in cases of suspected SIH.
Footnotes
Conflict of interest
None declared.