The utility of radioisotope cisternography in low CSF/volume syndromes compared to myelography

Abstract

Objective

The objective of this report is to compare computed tomography (CT) and magnetic resonance (MR) myelography with radioisotope cisternography (RC) for detection of spinal cerebrospinal (CSF) leaks.

Methods

We retrospectively reviewed 12 spontaneous intracranial hypotension (SIH) patients; CT and RC were performed simultaneously. Three patients had MR myelography.

Results

CT and/or MR myelography identified CSF leaks in four of 12 patients. RC detected spinal leaks in all three patients confirmed by CT myelography; RC identified the CSF leak location in two of three cases, and these were due to osteophytic spicules and/or discs. RC showed only enlarged perineural activity. Only intrathecal gadolinium MR myelography clearly identified a slow leak from a perineural cyst. In eight remaining cases, the leak site was unknown; however, two of these showed indirect signs of CSF leak on RC. CSF slow leaks from perineural cysts were the most common presumed etiology; and the cysts were best visualized on myelography.

Conclusion

RC is comparable to CT myelography but has spatial limitations and should be limited to atypical cases.

Introduction

Spontaneous intracranial hypotension (SIH), or a low cerebrospinal fluid (CSF) pressure/volume state, has clinical and radiographic variability (1). SIH is classically diagnosed as a clinical triad consisting of orthostatic headache, diffuse dural enhancement (DE) on magnetic resonance imaging (MRI) of the brain, and CSF pressure measuring 60 mm H₂O or less. The syndrome is better described as a low CSF pressure/volume state. In a subset of patients the CSF leak remains unidentified despite repetitive and costly investigations (2). Comparative diagnostic studies are warranted for costs containment, to minimize morbidity and patient discomfort, and for medical and/or surgical considerations.

For diagnostic evaluation, some studies suggest that radioisotope cisternography (RC) is comparable to myelography (2,3). MR myelography is a relatively recent development (4). While RC may identify an active leak, it is typically less precise than myelography. Indirect signs (i.e. early bladder accumulation, reduced diffusion of radioisotope over the cerebral hemispheres, abnormal visualization of the root sleeves, rapid disappearance of spinal activity) may be useful (5); however, the reliability of the empty bladder sign has been questioned because of technique or lymphatic drainage (6,7). Our objective was to determine the diagnostic value of RC compared to myelography in detecting the presence and location of CSF leaks. We hypothesized that RC may be useful only in cases where leaks remain occult following myelography.

Methods

After obtaining institutional review board approval, a keyword search for RC at the University of California, San Francisco (UCSF) Medical Center revealed 74 patients (mean age = 50 ± 4; including eight females and four males) who had 82 radioisotope cisternographies (1995–2010). Consent was not required by the IRB as encounters were distant; information was anonymized. A retrospective review of clinical evaluations and findings on computed tomography (CT) myelography and MR myelography was performed (Figure 1). We excluded indications unrelated to spontaneous CSF leaks. There were 16 cases with suspected SIH, of which 12 cases were finally diagnosed by the presence of orthostatic headache and features of SIH on gadolinium-enhanced MRI of the brain including dural enhancement or brain sagging, or detection of a CSF leak on CT or MR myelography, or both. One case meeting criteria for SIH had a course complicated by shunt placement for a posterior fossa cyst. All individuals had RC and CT and/or MR myelography. The remaining four suspected SIH cases did not have an identified CSF leak.

Figure 1.

Diagnostic evaluations of low CSF/volume syndromes with radioisotope cisternography and myelography.

RC was performed with early and delayed images over 24 hours using Indium-111-labeled diethylene triamine pentaacetic acid (DTPA) injected intrathecally. CT myelography was performed following intrathecal administration of iodinated contrast in conjunction with the radiotracer. In one case, there was a four-day delay between the RC and the CT myelography. Three patients also underwent MR myelography using 0.3 cc of intrathecal gadolinium injected at the same time as the iodinated contrast. Descriptive statistics were used for comparison.

Results

All cases presented initially with an orthostatic headache, of which nine of 12 cases had persistent postural symptoms (8). Five patients also had other symptomatology. The duration of symptoms preceding the RC and myelography ranged between two weeks and six years. In this group of selected patients referred to our center, multiple epidural blood patches (EBPs) had been performed without permanent symptomatic relief. Gadolinium-enhanced MRI of the brain showed diffuse dural enhancement in seven of 12 cases, and the displacement of posterior fossa contents in nine of 12 cases. Only one individual had normal brain imaging.

Of the 12 patients, three had active leaks identified on CT myelography and RC. Therefore, the sensitivity of initial CSF leak detection was 25% for both RC and CT myelography. The etiology of the CSF leaks detected by CT myelography included perineural cysts (n = 1) and disc/osteophytic spicules (n = 2). The precise location of the CSF leak was identified in two of three cases with CT myelography but unidentified on RC because of extensive tracer leakage.

All three patients with leaks identified on CT myelography showed extradural radiotracer activity on RC without definitive confirmation of the leak. Two cases demonstrated prominent radionuclide collections in the cervical and thoracic levels mimicking perineural cysts, yet they corresponded to epidural CSF collections on myelography that were associated with upper thoracic disc, or disc-osteophytic spicules penetrating the dura, or both (Figure 2(a)–(c)). One case demonstrated an extensive CSF leak between C2 and T2; a C1 puncture three days later showed cervical extravasation pooling from the thoracic transdural herniation site.

Figure 2.

CSF leak findings of low CSF/volume syndromes with radioisotope cisternography and myelography.

In one of the three cases of CSF leak, there was diffuse activity on RC along the nerve roots spanning the cervical, thoracic, and lumbar regions. This correlated with a rapid leak with extensive epidural contrast demonstrated soon after the myelogram was performed. A large perineural cyst was also noted at T7 on myelography. In a fourth case, the initial RC showed mid-thoracic activity characterized as a cyst or possible leak (Figure 2(d)); however, CT myelography was negative. In contrast, MR myelography identified two slowly leaking perineural cysts (Figure 2(e)); serial EBPs produced transient improvements on two occasions.

In eight patients, no leak was identified on CT, MR myelography or RC. CT myelography was repeated in four individuals. In two cases, the use of intrathecal gadolinium, in the same setting, did not detect a leak. A false positive was seen in a fourth case on a repeat CT myelogram, which showed extravasation of contrast from a right perineural cyst at the T4–T5 level. On further evaluation, gadolinium-enhanced MRI showed that this was more consistent with prominent venous structures. All eight cases had multiple prominent perineural cysts, and in two cases, numerous cysts were noted spanning the length of the spine (Figure 2(f) and (g)), without definite signs of a CSF leak.

There were no direct signs of CSF leaks on RC in any of the myelography leak-negative cases, although two of eight cases had indirect signs consisting of diminished radiotracer diffusion over the cerebral convexities on RC. One of these cases was associated with a possible leak versus a prominent nerve sheath. A needle artifact was also seen in one case.

Discussion

This study reports a diagnostically challenging subset of patients with SIH. We found that RC had comparable sensitivities with CT myelography in the detection of CSF leaks; however, the use of RC was insensitive to the anatomic site of the leak (5,9). In two cases that were negative for CSF leaks with myelography, RC provided indirect signs of slowed diffusion over the cerebral hemispheres. It has been suggested that indirect signs may be useful for atypical cases when the diagnosis of a spontaneous CSF leak is in doubt (10). However, in our series the diagnosis was already made clinically; therefore the indirect signs were not useful. Consistent with our hypothesis, we found that RC did not significantly augment the diagnostic evaluation (6).

Interestingly, foramenal outpouchings on RC, which did not correlate with perineural cysts on MR or CT myelography, were identified. These outpouchings on RC may appear larger than they are physically because of partial volume-based image blurring, epidural extravasation, and small amounts of normal perineural fluid. In the cases of perineural cysts, the normal anatomy will result in a somewhat serrated appearance on the nuclear medicine exam. However, a focal and/or prominent outpouching at a given level suggests a leak. These outpouchings are a good example of the need to correlate radionucleotide images with MRI if nuclear studies are to be used.

This study was limited by its small sample size and retrospective design. However, the simultaneous evaluations with RC and CT myelography allowed for several observations. First, rapid CSF leaks appeared easier to detect with both RC and CT myelography. Second, dynamic myelography is needed occasionally to detect more rapid CSF leaks. In one case, a slowly leaking perineural cyst was seen on MR myelography but was not confirmed on multiple RC or CT myelography exams. Intrathecal gadolinium administration provides superb soft tissue contrast and may be indicated in patients in whom slow leaks are suspected on CT (4,11). Slow leaks may benefit from delayed MRI, which can be performed even hours after gadolinium intrathecal instillation. Lastly, a recent study conducted in our center comparing CT and MR myelography suggests that greater use of MR myelography may have provided a greater yield (4).

Ultimately, occult CSF leaks may be secondary to the referral bias inherent in tertiary medical centers. Negative leak evaluations were common and consistently associated with multiple perineural cysts that were best seen on myelography. The presence of multiple perineural cysts and signs of spontaneous CSF leaks warrant genetic investigation for heritable connective tissues disorders with vascular and other consequences (12). Another possibility is that a leak may be present but not detectable with current techniques, or occurring directly between the thecal sac and perineural venous or lymphatic structures (13). RC may prove useful in these cases if rapid blood or urinary uptake can be shown. Moreover, orthostatic headache owing to altered CSF flow dynamics without a leak has been described (14). Such cases may benefit from RC; although this was not useful in one of our cases in which altered CSF flow dynamics was suspected (8). Alternatively, it is conceivable that the clinical features of a low CSF volume/pressure state were initiated by an index event without a persistent leak. Based on our findings, we suggest RC should be performed only in cases of atypical presentations or in cases where myelography is negative.

Clinical implications

Radioisotope cisternography (RC) is comparable to myelography; however, the superior anatomical delineations of computed tomography (CT) or magnetic resonance (MR) myelography limit its utility.

The identification of cerebrospinal fluid (CSF) spinal leaks in cases of low CSF/volume syndromes with extensive perineural cysts may be challenging despite repetitive tests.