Noninvasive Differentiation of Meningiomas and Dural Metastases Using Intratumoral Vascularity Obtained by Arterial Spin Labeling

Meningiomas are the most common primary intracranial tumors in adults, representing approximately one third of all intracranial neoplasms [1, 2]. They arise from meningothelial cells of the arachnoid and present as avid, homogeneous, contrast-enhancing dural masses on computed tomography (CT) and magnetic resonance imaging (MRI) [3]. Dural metastases occur much less frequently, representing approximately 1% of intracranial tumors and approximately 10% of patients who suffer from advanced systemic cancer [4]. The most common primary neoplasm from which dural metastases develop are breast, prostate, kidney, lung and head and neck carcinomas [5]. A number of cases have been documented in the literature in which dural metastases were misdiagnosed as meningiomas on conventional imaging [6‐8]; however, accurate preoperative differentiation between meningiomas and dural metastases is crucial because this determines a patient’s prognosis and directly influences treatment options [9, 10]. With conventional MRI methods, the differentiation between meningiomas and dural metastases remains difficult because MRI findings overlap in these tumor entities [11, 12]. Previously published findings have already confirmed the advantages of advanced imaging methods in increasing the diagnostic accuracy of meningiomas and dural metastases. In particular, dynamic susceptibility contrast MRI, a contrast-enhanced MRI perfusion technique, has been revealed to be useful in the differentiation between these two tumor types [13, 14]. This technique is, among others, focused on the assessment of perfusion in intra-axial and extra-axial brain tumors; however, in patients with previous allergic reactions to contrast media or in patients with considerably elevated serum creatinine values, the administration of contrast media is critical.

Another MRI sequence used to investigate tumor vascularization is the arterial spin labeling (ASL) technique. This assesses cerebral blood flow (CBF) by using magnetically labeled water protons of the arterial blood that serve as an endogenous tracer. Water protons of the arterial blood are labeled in a labeling slab proximal to the tissue of interest (which is referred to as the imaging volume). Subsequently, the labeled blood reaches the imaging volume where the labeled images (including the delivered magnetization of the arterial blood) are acquired after a predefined inversion time. The perfusion contrast derived by this method comes from subtracting the labeled images from control images (depicting the same tissue of interest without prior labeling of the arterial water protons). Thus, there is no need for the administration of exogenous contrast agents. Using very low inversion times, ASL has been shown to generate normalized vascular intratumoral signal intensity values (nVITS), which correlate with glioma grade and can be used to differentiate between primary cerebral lymphoma and glioblastoma [15, 16].

The purpose of this study was to investigate the utility of nVITS values obtained from ASL to distinguish between meningiomas and dural metastases. This could result in a completely noninvasive method with which to differentiate these meningeal tumor entities.

A total of 30 meningioma patients (WHO I, n = 25; WHO II, n = 5) and 16 patients with dural metastases (breast cancer, n = 5; lung cancer, n = 6; prostate cancer, n = 2; renal cancer, n = 1; gastric cancer, n = 1; melanoma, n = 1) were consecutively included in this prospective study. Descriptive statistics about patient characteristics and tumor volume are shown in Table 1.

Examples of T1-weighted post-contrast images and rCBF maps obtained by PASL of a patient with a meningioma and a breast cancer patient with dural metastases are shown in Fig. 1a–d.

The mean nVITS value was 3.0 (SD: 1.2, range: 2–5.8) for meningiomas and 1.7 (SD: 0.6, range: 0.9–3.3) for dural metastases. Thus, meningiomas had a significantly higher mean nVITS value than dural metastases (p < 0.001). For illustration, see also the box-and-whisker plots in Fig. 2.

The ROC curve plotted in Fig. 3 demonstrates that the optimal cut-off value was 1.99 for 100% sensitivity and 81.2% specificity. The AUC value was 0.93.

There were no significant differences in sex (p = 0.78), age (p = 0.95) and normal appearing white matter vascular signal intensities (p = 0.62) between meningioma and dural metastases patients. Moreover, there was no significant difference in the tumor volume (p = 0.19); however, there was a wide range of the tumor volume within each tumor entity subgroup (125.9 cm³ in meningiomas and 77.9 cm³ in dural metastases, respectively).

To determine the most effective treatment option it is of critical importance to be able to differentiate dural metastases from other extra-axial neoplasms that may mimic the characteristic imaging features of this entity. One of the most frequently occurring differential diagnoses includes meningiomas. The features of meningiomas and dural metastases overlap on conventional imaging. Both are characterized as an extra-axial mass, which is isointense to grey matter on T1-weighted images and shows vivid contrast enhancement on T1-weighted, post-contrast images, including the dura, which is called the dural tail sign [11, 12]; however, contrast enhancement represents the disruption of the blood brain barrier rather than tumor neovascularization, which is known to be markedly intensified in meningiomas in contrast to the most common dural metastases, particularly those that derive from a primary breast, prostate or lung carcinoma [13, 21].

The results of this study show that PASL can provide useful information that can differentiate between meningiomas and dural metastases based on tumor vascularity. It was demonstrated that meningiomas show significantly higher nVITS values (p < 0.001) compared to dural metastases and a nVITS cut-off value of 1.99 was determined. A nVITS value of 1.99 or less suggested dural metastases, whereas a nVITS value greater than 1.99 suggested meningioma, with a sensitivity of 100% and a specificity of 81%. Therefore, especially in patients with a known primary neoplasm or in the case of an atypically presenting meningioma (e.g. with an adjacent osseous destruction or leptomeningeal extension) the addition of an ASL sequence to the conventional MRI protocol in the routine clinical setting is recommended. Although meningiomas are the most common primary intracranial tumors in adults, a dural lesion with low nVITS values should raise the suspicion of dural metastases; however, in the case of highly vascularized brain metastases, which were represented in this study by one renal cell carcinoma patient, one melanoma patient and one breast cancer patient, nVITS values were increased and overlapped with those of meningioma patients. For this purpose, an additional proton MR spectroscopy to assess an existing alanine peak could help to distinguish highly vascularized brain metastases from meningiomas [22].

The results of this study are consistent with the previously published literature. Kremer et al. [13] and Zimny and Sasiadek [14] revealed that relative cerebral blood volume (rCBV) values obtained by dynamic susceptibility contrast MRI can provide useful information with which to differentiate between dural metastases and meningiomas due to the highly vascular nature of meningiomas compared to the most common dural metastases; however, Kremer et al. also pointed out that rCBV values of hypervascular metastases may overlap with those of meningiomas [23].

To generate perfusion maps obtained by PASL imaging, a postlabeling delay of at least 2 s is recommended. This has been shown to be the time the labeled blood needs to completely enter the tissue space [24]. In contrast, nVITS values are obtained with a lower postlabeling delay time of 370 ms, and therefore, represent tumor vascularization rather than perfusion because the magnetically labeled water protons are primarily intravascular [24]. This technique has recently been revealed to be a useful tool for the differentiation of low-grade and high-grade astrocytomas, as well as of glioblastomas and primary central nervous system (CNS) lymphomas [15, 16]. In this study, it could also be shown that nVITS values offer the possibility to noninvasively differentiate between meningiomas and dural metastases. Therefore, the application of gadolinium-based contrast agents could be avoided, which is an advantage particularly for patients with elevated serum creatinine values or known previously allergic reaction after contrast medium administration. Another advantage of using nVITS values obtained by PASL is the low acquisition time of this MR sequence of 1 min and 20 s. This acceleration, compared to other ASL sequences, where the acquisition times usually range between 3–6 min, can be achieved due to the reduced number of measurement repetitions because of the signal intensification of intravascular blood flow compared with perfusion blood flow within the tumor tissue [24‐26]. A negative side effect of the reduced number of measurement repetition is that this is accompanied by a decreased signal to noise ratio; however, because of the extended MR protocols for brain tumors due to advanced MR techniques, the time for each MR sequence should be as short as possible in order to gain the greatest amount of imaging tumor characteristics in the shortest possible time, in terms of patient care. Moreover, a cut-off value for nVITS values (1.99) has been determined, proposing a fast and easily applicable marker in the routine clinical setting with which to differentiate between meningiomas and dural metastases.

A potential limitation of this study is that the calculation of the cut-off value was based on a small sample size. In particular, the number of patients with dural metastases is low due to the relatively rare occurrence of this tumor entity compared to meningiomas [1, 27]. Further studies with an increased sample size as well as different MRI scanner or ASL sequence parameters are needed to support these findings and to take possible variability due to different sequence parameters, MR scanners, manufacturers and postprocessing techniques into account. Moreover, in 5 (31%) brain metastasis patients no tissue specimens were available, because they did not undergo surgery or stereotactic biopsy due to the poor general condition; however, due to the known neoplastic disease, the multiple intracranial and extracranial metastases and the further anatomical imaging details (contrast-enhancing lesion with broad dural attachment that was assumed to be located at least partially extra-axially), those lesions were included in the dural metastases subgroup. Another technical limitation is the limited spatial resolution of the ASL technique. In addition, a voxel size of 3 × 3 × 6 mm and a 1.5 mm spacing were used in this study, which could lead to an underestimation of the signal intensity due to partial volume artefacts especially in small tumorous lesions. The smallest lesion included in this study was a meningioma with a volume of 2 cm³, which showed clearly increased VITS intensity values in comparison to the contralateral normal white matter tissue (NWM) (nVITS value = 2.1); however, further studies are needed to investigate also smaller tumorous lesions using this MR perfusion technique. Furthermore, the tumor vascular blood flow may be underestimated in cases of arterial stenosis, which would result in a prolonged bolus arrival time. In order to minimize this potential bias, flow-void signals obtained from T2-weighted images were used to exclude vascular occlusion or profound stenosis; however, non-contrast medium-based vessel imaging techniques, such as time-of-flight angiography, should be added to the MR imaging protocol to rule out marginal reduction of arterial blood flow, which could affect PASL data.

The nVITS values obtained from PASL sequences noninvasively differentiate between meningiomas and dural metastases on the basis of tumor vascularization with high sensitivity and specificity. Therefore, a dural lesion with low nVITS values strongly argues against the diagnosis of a meningioma and dural metastases should be considered as part of the differential diagnosis. The ASL technique represents a fast and easily applicable MRI technique and in our opinion should be included in the clinical routine MRI protocol for all intracranial neoplasms, including extra-axial tumors.