Imaging Technique

The selection of MRI has a significant impact on the conspicuity of normal meningeal enhancement and sensitivity in detecting meningeal disease (see Figs. 18-6 and 18-7 ). Numerous factors influence the appearance of the meninges on MRI, including (1) the presence and amount of MR contrast agent administered, (2) the type of pulse sequence performed, and (3) the exact pulse sequence parameters.

For partial saturation (spin echo or gradient echo), short echo time (TE) sequences, crucial imaging parameters include the repetition time (TR) and excitation flip angle. The shorter the TR, the greater the degree of saturation of magnetization (i.e., decreased signal intensity) from all imaged tissues. As a result, contrast-enhancing tissue will be more conspicuous against the more saturated (i.e., hypointense) background tissues on a typical short TR, large flip-angle gradient echo study than on a spin echo study. Therefore the normal meninges usually exhibit diffuse enhancement on such ultrashort TR, large flip-angle gradient echo imaging sequences (see Fig. 18-7 ). Imaging plane also affects visualization of meningeal enhancement; the coronal plane is preferred to axial MRI data for evaluating meningeal enhancement over the cerebral convexity.

Similarly, factors that either increase signal from the meninges or decrease signal from background tissues enhance the contrast-to-noise ratio (CNR) and therefore the conspicuity of the enhancing meninges. Double-dosing and triple-dosing of paramagnetic contrast agents have demonstrated improved detection of parenchymal lesions with MRI, and there is evidence to support a similar dose effect for enhanced MRI of meningeal disease. More recently, cases of leptomeningeal metastases that had been diagnosed by MRI and high-dose (0.3 mmol/kg) gadolinium that were not visualized with standard dosing (0.1 mmol/kg) technique have been reported ( Fig. 18-8 ). Also, the use of fat saturation or magnetization transfer options increases the CNR by decreasing the signal from the background tissues.

Effect of contrast dose on meningeal enhancement. Axial T1-weighted images obtained after administration of 0.1 mmol/kg ( A ) and 0.2 mmol/kg ( B ) doses of gadolinium show increased enhancement in a pia–subarachnoid space pattern at the higher dose in this patient with carcinomatosis of the meninges from gastric carcinoma.

Sensitivity of MRI

In the past, nonenhanced spin echo MRI was insensitive in detecting meningeal disease. The advent of the FLAIR sequence has greatly improved the sensitivity of MRI performed in the absence of gadolinium to detect meningeal and SAS abnormalities. Singer and coworkers evaluated the use of FLAIR MRI in 62 patients (21 with proven SAS or meningeal disease) and 41 control patients. The sensitivity, specificity, and accuracy of FLAIR for SAS disease were 85%, 93%, and 90%, respectively. Although all six cases of acute SAH were interpreted as abnormal on FLAIR images in the Singer series, this was a source of false-negative interpretation in the series by Williams and associates.

Poor detection of SAH has been attributed to the relatively high oxygen tension in the SAS, which does not permit evolution to deoxyhemoglobin, and to the diluting effect of CSF. In 24 patients who underwent both FLAIR and Gd-enhanced T1-weighted MRI, FLAIR imaging (sensitivity, specificity, and accuracy of 86%, 91%, and 89%, respectively) was superior to Gd-enhanced T1-weighted imaging (43%, 88%, and 74%).

Williams and coauthors prospectively evaluated 376 consecutive cases performed with FLAIR imaging and showed that FLAIR may result in false-negative diagnoses of meningeal or SAS disease in cases of infectious meningitis, carcinomatosis, and SAH. In this series a false-positive diagnosis of meningeal or SAS pathology was made in the presence of normal and hyperintense cortex, susceptibility artifact, prominent pial vessels, concatenated saturation pulse, or flow artifact. Neoplastic or inflammatory processes involving the SAS are best evaluated with paramagnetic contrast agents for visualization of enhancement of the accompanying meningeal involvement ( Fig. 18-9 ).

Axial FLAIR image in a patient with leptomeningeal spread of breast carcinoma ( A ) shows signal abnormality in a pia–subarachnoid space pattern (arrowheads) that enhances on the T1-weighted image after administration of gadolinium ( B ).

Role of Imaging

Before and after gadolinium administration, MRI plays an important role in the diagnosis of meningeal neoplasm in the asymptomatic patient when CSF cytologic examination results are equivocal or when lumbar puncture is contraindicated. Imaging, however, does not replace CSF examination in diagnosis of meningeal neoplasm.

The limited sensitivity of CSF cytologic examination continues to be a significant obstacle and reinforces the complementary role of imaging. The percentage of positive spinal CSF cytology in cases of primary CNS tumors with histologically confirmed meningeal involvement varies from 12% to 63% and is increased in symptomatic patients. With meningeal metastases from non-CNS neoplasms, CSF cytology was positive in 45% to 80% of cases, with higher yields after multiple spinal taps. CSF flow cytometry increases the detection of CSF spread of hematologic malignancies over cytologic examination alone. The modest sensitivity of CSF cytology has fueled the search for CSF tumor markers to detect dissemination of neoplasm. Recent literature suggests that markers such as CA 15-3 may have value in detecting breast carcinoma in CSF, and nuclear magnetic spectroscopy may have value in detecting metabolites in CSF indicating disseminated lung adenocarcinoma. Additional tumor markers that may be detected in CSF to indicate leptomeningeal neoplasm include transforming growth factor (TGF)-β1, vascular endothelial growth factor (VEGF), urokinase-type plasminogen activator (uPA), and tissue-type plasminogen activator (tPA).

Like CSF cytology, tumor detection with CSF markers may also be hampered by suboptimal sensitivity and specificity.

The sensitivity of MRI in detecting meningeal neoplasm was originally reported to be lower than that of CSF cytology, with high false-negative rates: 30% to 33% for MRI and 58% for CT. The early studies, however, may have unwittingly introduced a selection bias by using positive cytology as an inclusion criterion. Later series comparing CT myelography with MRI and cytology showed a higher detection rate of CSF metastases with MRI (range, 65%-72%) than with CT myelography (range, 45%-47%) and cytology (29%).

A Memorial Sloan-Kettering Cancer Center study showed that when the patient cohort was not restricted to patients with positive cytologic findings, the rate of detection of meningeal carcinomatosis was increased. This study examined 137 patients with signs and symptoms of meningeal disease. CT and MRI were assessed for signs of meningeal or SAS neoplasm, including hydrocephalus, enhancement of the dura, leptomeninges, and cranial nerves. Leptomeningeal metastases were identified in 77 of 137 patients. The diagnosis of leptomeningeal metastases was based on the clinical and imaging findings alone in 31% of those cases. Abnormal imaging findings were reported much more frequently in cases of solid tumor primaries (90%) than in hematologic neoplasms (55%). More recent studies confirm these findings. A caveat in using MRI for the diagnosis of leptomeningeal tumor involvement, therefore, is that MRI is less sensitive in detecting involvement of the meninges resulting from hematologic malignancies than in detecting detecting solid tumors.

When the distinction between meningeal neoplasm and inflammatory meningeal disease cannot be established by CSF and clinical data, MRI can support the diagnosis of neoplasm and guide meningeal biopsy. Although infectious meningitides are often uncovered by CSF analysis, MRI may be used to target meningeal biopsies when needed.

Cheng and coworkers reported an improved yield from meningeal biopsy specimens in cases of chronic meningitis when tissue specimens were obtained from enhancing regions identified on MRI study ( Fig. 18-10 ). MRI has an advantage over CSF examination in its ability to characterize bulky neoplastic disease that may be more responsive to radiation therapy. Because CSF cytologic examination produces a higher false-negative rate in focal than in diffuse disease, imaging may be of particular value in detecting focal spread of neoplasm to the meninges or SAS. MRI also has the potential to allow noninvasive monitoring of treatment response, although it is conceivable that the imaging abnormalities may persist beyond the eradication of neoplastic cells in the CSF. In a series of patients with coccidioidal meningitis, diminution of meningeal enhancement was seen in treated patients with improving CSF profiles. Therefore MRI also may be useful as a means of monitoring response to therapy in fungal meningeal disease.

Chronic meningitis. A, Thick slightly irregular dura-arachnoid enhancement on axial T1-weighted imaging proved to be idiopathic hypertrophic pachymeningitis on biopsy. B and C, In another patient, thin slightly irregular dura-arachnoid enhancement on axial ( B ) and coronal ( C ) T1-weighted imaging also proved to be idiopathic hypertrophic pachymeningitis on biopsy.

Imaging is a useful adjunct for the diagnosis of neoplastic meningeal disease in the appropriate clinical setting, along with CSF examination to exclude infectious or noninfectious processes of the meninges.

Finally, imaging may allow the clinician to assess outcome, in that diffuse leptomeningeal involvement confers a poor prognosis.

Cardinal Imaging Signs of Meningeal Pathology

The main imaging findings that have been associated with meningeal pathology are as follows :

• •
  

  Hydrocephalus ( Fig. 18-11 )

  
  

  
  

  
  

  

  
  

  
  

  Hydrocephalus with ependymal enhancement and pia–subarachnoid space abnormality. A, Axial T1-weighted image obtained after gadolinium administration shows enlargement of the lateral ventricles and thin linear ependymal enhancement. B, There is a pia-subarachnoid enhancement pattern between the cerebellar folia and along the third cranial nerves, indicating meningeal spread of small cell lung carcinoma in a patient with multiple cranial neuropathies.

  
  
  
  
• •
  

  Dura and arachnoid enhancement or signal abnormality (see Fig. 18-10 )

  
  
• •
  

  Pia and SAS enhancement or signal abnormality (see Fig. 18-11 )

  
  
• •
  

  Subependymal enhancement or signal abnormality ( Fig. 18-12 )

  
  

  
  

  
  

  

  
  

  
  

  Focal nodular ependymal enhancement. A, Nodular enhancement of the ependyma of the frontal horns was the probable entry point for subarachnoid spread of melanoma. B, Sagittal T1-weighted image of the lumbar spine shows nodular enhancing drop metastases to the cauda equina (arrowheads).

  
  

Hydrocephalus may or may not be associated with enhancement of the meninges or ependyma. In this setting, hydrocephalus alone implies a resorptive block to CSF flow. Hydrocephalus may occur in the setting of SAH, infectious or noninfectious meningitis, and neoplasm. In the patient with neoplastic meningeal disease, hydrocephalus is more likely when leptomeningeal invasion or SAS has occurred rather than when neoplasm is limited to the dura.

Enhancement of the meninges may occur in the spine or brain. Meningeal or subependymal enhancement may be focal ( Fig. 18-13 ) or diffuse (see Fig. 18-12 ) and may have either a smooth or nodular contour ( Fig. 18-14 ; see Fig. 18-13 ). Diffuse leptomeningeal involvement is a harbinger of a worse prognosis than focal disease. Although a nodular pattern of enhancement may suggest neoplasm, it is not specific; nonneoplastic entities (e.g., sarcoidosis) may also produce nodular thickening of the meninges (see Fig. 18-14A and B ). Infiltration of the leptomeninges overlying the convexities or in the basal cisterns may result in sulcal or cisternal obliteration on a noncontrast T1-weighted image ( Fig. 18-15 ) or a FLAIR sequence (see Fig. 18-9 ). Minimal shortening of T1 and T2 relaxation times in the cisternal CSF (“dirty CSF sign”) may be seen on nonenhanced MRI (see Fig. 18-15 ). In rare instances, subarachnoid tumor or inflammatory exudates may result in distention of the SAS or perivascular (subpial) spaces ( Fig. 18-16 ).

Diffuse nodular pia–subarachnoid space enhancement. A and B, Multiple pial-based enhancing nodules are present in this patient with disseminated lung carcinoma.

Nonneoplastic focal nodular and linear meningeal enhancement. A and B, Nodular enhancement in a pia-subarachnoid pattern (arrows) mimics neoplasm in a case of sarcoidosis. C, Focal linear enhancement in a linear dura-arachnoid pattern (arrows) in a different case of sarcoidosis.

“Dirty CSF sign” and ventricular debris. T1-weighted image ( A ) demonstrating obliteration of bilateral sulci, and FLAIR image ( B ) showing material that is hyperintense to CSF within the subarachnoid space (arrows) in a patient with gram-positive meningitis. C, After gadolinium administration there is extensive pia–subarachnoid space enhancement.

Perivascular space pattern. Axial T2-weighted image ( A ) and enhanced T1-weighted image ( B ) show distention of the perivascular spaces at the base of the brain by the gelatinous nonenhancing pseudocysts of cryptococcosis.

Imaging Patterns

Two distinct patterns of meningeal enhancement or signal abnormality may be observed with MRI. The dura-arachnoid pattern follows the inner contour of the calvaria ( Fig. 18-17 ), whereas the pia mater–SAS pattern extends into the depths of the sulci ( Fig. 18-18 ). Enhancement or signal abnormality surrounding the brainstem is always of the pia mater–SAS type, in that the arachnoid is clearly separated from the pia mater by the intervening basal cisterns in this region. Although pia mater–SAS enhancement does occur more commonly in the setting of meningitis than with tumor involvement, both inflammatory and neoplastic processes may demonstrate either pattern. A diffuse appearance favors an inflammatory etiologic mechanism, whereas nodular meningeal enhancement suggests a neoplasm ( Boxes 18-2 to 18-6 ).

Diagram illustrating the dura-arachnoid pattern of meningeal abnormality that follows the inner table of the skull and the dural reflections (arrows).

(From Meltzer CC, et al: MR imaging of the meninges: Part 1. Normal anatomic features and nonneoplastic disease. Radiology 201:297–308, 1996.)

Diagram illustrating the pia–subarachnoid space pattern of meningeal abnormality that invaginates into sulci (arrows).

(From Meltzer CC, et al: MR imaging of the meninges: Part 1. Normal anatomic features and nonneoplastic disease. Radiology 201:297–308, 1996.)

Dura-arachnoid pattern.

Because the dura and arachnoid are closely approximated, the distribution of meningeal enhancement does not reliably distinguish purely pachymeningeal (dural) from leptomeningeal (arachnoidal and pial) involvement. As a result the dura-arachnoid pattern of enhancement does not imply that the leptomeninges or the SAS is spared. This fact has practical importance because some authors report a lower incidence of positive CSF cytologic findings with the dural pattern, in contrast to the pial pattern, implying a lack of SAS neoplasm.

A surprisingly moderate rate (55%) of positive CSF cytologic results was reported in a subset of 11 patients with a dura-arachnoid enhancement pattern on MRI. The presence of malignant cells in the CSF of patients in this subgroup implies neoplastic involvement of the arachnoid and SAS. This confirms the limitation of MRI in distinguishing abnormal enhancement of the dura from that of the arachnoid. When restricted to the dura, however, meningeal carcinomatosis often results in negative CSF cytology; in this setting, MRI can play an important role in disease detection.

The dural tail sign, once considered specific for meningioma, may be observed in a wide variety of other extraaxial lesions, including schwannoma ( Fig. 18-19 ), dural metastases ( Fig. 18-20 ), lymphoma, tuberculoma, and sarcoidosis. Occasionally a dural tail may be seen in association with intraaxial mass lesions such as gliomas or non-CNS metastases. Although this imaging feature has been useful in suggesting the diagnosis of meningioma, its lack of specificity may occasionally cause misleading interpretation of MRI. Similar signal, shape, and enhancement characteristics found with several of these lesions may further confound the distinction between meningiomas and other entities.

Focal dura-arachnoid enhancement. A dural tail (arrowheads) adjacent to a left acoustic schwannoma illustrates the nonspecificity of this sign as an indicator of meningioma on an enhanced axial T1-weighted image.

Focal dura-arachnoid mass. CT scan ( A, B ) shows colon adenocarcinoma metastases to the calvaria (arrow) extending to the dura (arrows) on enhanced axial and coronal T1-weighted images ( C, D ).