Radiologic and histologic features of the T2 hyperintensity rim of meningiomas on magnetic resonance images

Abstract

A hyperintensity rim is often seen at the brain–tumor interface of meningiomas upon T2-weighted (T2WI) magnetic resonance imaging (MRI), and it is referred to as the cerebrospinal fluid (CSF) space; however, the true nature of the rim remains unclear. We surveyed the MRI findings and the histopathologic characteristics of such rims. Our study population consisted of 53 consecutive patients who underwent meningioma removal at our hospital. The intensity of the rim on MRI scans obtained with different imaging sequences was assessed in all patients. We used 22 tumors for histopathologic investigation: tissue samples were acquired from both the tumor surface and from a deep intratumoral site. Of the 53 meningiomas, 37 (69.8%) manifested a hyperintensity rim on T2WI (T2-rim). The other 16 showed neither a hyperintense nor a hypointense rim on their T2WI. An enhancement effect corresponding to the rim was observed in 28 of the 37 (75.7%) T2-rim positive tumors. While 9 among the 37 tumors with a T2-rim (24.3%) did not show rim enhancement, they showed low intensity on fluid-attenuated inversion recovery (FLAIR) images. The microvascular density in the tumor capsule was significantly greater in the 12 T2-rim and rim enhancement positive tumors than in 10 tumors that were T2-rim negative or T2-rim positive, but rim enhancement-negative (p < 0.001, Mann–Whitney U test). We found that 75.7% of T2 hyperintense rims that were detected at the brain–meningioma interface reflected a microvascular-rich capsule layer, rather than the CSF space.

Introduction

A thin, hyperintense rim is often encountered on T2-weighted images (T2WI) at the brain–meningioma interface. Some investigators considered the rim to reflect the cerebrovascular fluid (CSF) space, suggestive of a neuroimaging hallmark of extra-axial tumor and a good predictor of the smooth separation of the tumor from the brain surface.^1-5

As we found that these rims occasionally manifested an enhancement effect we performed a prospective survey on preoperative magnetic resonance imaging (MRI) scans and examined the histopathology of a series of meningiomas to elucidate the true nature of these rims.

Methods

Patients and MRI studies

Between April 2014 and January 2016, we had 58 consecutive patients with intracranial meningiomas who underwent their transcranial meningioma removal at Kagoshima University Hospital in Japan. From the study, we excluded two patients with tumor recurrence, two patients with hemorrhagic onset and one patient with no preoperative MRI scans, due to the presence of a pacemaker. Therefore, this prospective study included 53 patients: 15 men and 38 women, ranging in age from 36 to 86 years (mean 61.8 ± SD of 12.8 years). Of these, 45 were World Health Organization (WHO) Grade I, seven were Grade II, and one was Grade III. The tumor location was supratentorial in 45 patients and infratentorial in eight patients.

All patients underwent MRI studies on a 3T unit (Ingenia 3.0T, Philips Healthcare, Cleveland, OH, USA). The image acquisition sequences included T1-weighted imaging (T1WI), T2WI, fluid-attenuated inversion recovery (FLAIR), diffusion-weighted imaging (DWI) and contrast-enhanced (CE) T1WI. Tumor diameter was recorded as the longest diameter on axial, coronal and sagittal post-contrast images. We assessed the intensity of the brain–tumor interface on T2WI, T1WI, FLAIR, DWI and CE T1WI scans.

Histopathology studies

We used 22 tumors for histopathology analysis; these tissue samples were acquired from both the surface of the tumor (brain side) and from a deep intra-tumoral site. The samples were embedded in paraffin blocks and subjected to hematoxylin–eosin (HE) and immunohistochemical (IHC) staining.

IHC studies were performed using mouse monoclonal CD34 antibody 1:1 (Beckman Coulter, Brea, CA, USA) and rabbit polyclonal vascular endothelial growth factor (VEGF) antibody 1:50 (Santa Cruz Biotechnology, Dallas, TX, USA). For heat-induced epitope retrieval, de-paraffinized sections were placed in a microwave for 15 min in 0.01 mol/L citrate buffer. This was followed by endogenous peroxidase blocking, in methanol and 0.3% hydrogen peroxide for 30 min at room temperature (RT). Non-specific binding was blocked with 2% normal goat or horse serum (Vector Laboratories, Burlingame, CA, USA) in 0.01 mol/l phosphate-buffered saline (PBS) for 30 min at RT. Primary antibody was then added at a previously-determined optimum dilution and this was followed by overnight incubation at 4℃. Biotinylated secondary antibody (Vector Laboratories, Burlingame, CA, USA) was added and the tissue samples were incubated for 30 min at 37℃. We then applied the Vectastain Elite ABC Kit (Vector Laboratories, Burlingame, CA, USA) for 30 min at RT, followed by a 3,3′ diaminobenzidine solution (Nichirei Biosciences, Tokyo, Japan).

Evaluation of immunohistochemical staining

The microvascular density based on CD-34 positive vessels was scored as 0 (negative to scarce), 1 (moderate), and 2 (rich).⁶ The incidence of VEGF-positive cells among neoplastic cells was recorded as 0 (<5%), 1 (<25%), 2 (<50%), 3 (<75%) and 4 (>75%); the intensity of positive staining was recorded as 0 (no staining), 1 (weak staining), 2 (moderate staining) and 3 (strong staining). Multiplication of these values served as the VEGF staining score.⁷

Statistical analysis

Statistical analysis was performed with StatFlex version 6.0 (Artech Co., Osaka, Japan). We analyzed the data using the Kruskal–Wallis, chi square, Student’s t and Mann–Whitney U test, depending on the characteristics of the dataset. Differences of p < 0.05 were considered statistically significant.

Ethical considerations

This study was approved by the ethics committee of Kagoshima University Hospital (No. 26–27). The authors certify that this study involving human subjects was in accordance with the Helsinki declaration of 1975, as revised in 2000; as well as the Ethical Guidelines for Epidemiological Research (effective 1 July 2002), promulgated by the Ministry of Education, Culture, Sports, Science and Technology of Japan; and by the Ministry of Health, Labor and Welfare of Japan. To protect patient privacy, all data were collected and then analyzed after anonymization, in a linkable fashion.

Results

Radiologic and clinical features

Of the 53 tumors, 37 (69.8%) manifested a T2 hyperintensity rim (T2-rim) on T2WI scans (Table 1 and Figures 1–4). The other 16 tumors had neither a T2-rim nor a lower intensity gap between the brain and the meningioma. The thickness of the T2-rim ranged from 0.6 to 4 mm (mean: 1.5 ± 0.8). Of the 37 T2-rim-positive tumors, 28 (75.7%) revealed a post-contrast enhancement effect corresponding with the rim (Figures 1–3). Of these 28 tumors, the rim was hypointense to isointense on T1WI in all, as well as hyperintense to isointense on FLAIR images in 24 (85.7%) of them. This observation indicated that the T2-rim was not compatible with a CSF signal. We occasionally saw hypointense spots suggesting a vascular flow void in the T2-rim (Figures 2 and 3). Among the 37 tumors with a T2-rim, nine (24.3%) did not show rim enhancement (Figure 4); their rim was hypointense on T1WI, FLAIR and DWI scans. Based on this finding, we thought that this type of rim was compatible with the CSF signal.

Table 1.

Intensity of T2 hyperintense rims on T1WI, FLAIR, and DWI scans (n = 37).

	Post-contrast T1
	C/E negative (n = 9)	C/E positive (n = 28)
Intensity on T1WI	Hypo: 9	Hypo: 24
	Iso: 0	Iso: 4
	Hyper: 0	Hyper: 0
Intensity on FLAIR	Hypo: 9	Hypo: 4
	Iso: 0	Iso: 13
	Hyper: 0	Hyper: 11
Intensity on DWI	Hypo: 9	Hypo: 6
	Iso: 0	Iso: 2
	Hyper: 0	Hyper: 18

C/E: contrast enhancement of rim; DWI: diffusion-weighted image; FLAIR: fluid-attenuated inversion recovery; T1WI: T1-weighted image; T2: T2-weighted image

Figure 1.

A 58-year-old woman with a right sphenoid-ridge meningioma: (a) T2WI; (b) T1WI; (c) FLAIR image and (d) Post-contrast T1WI. A hyperintense rim is present on T2WI scan (a), as shown by arrows. The rim was hypointense on T1WI scan (b) and isointense on FLAIR (c) and well enhanced on post-contrast T1WI scans (d).

FLAIR: fluid attenuated inversion recovery; T1WI: T1-weighted image; T2WI: T2-weighted image.

Figure 2.

A 67-year-old woman with a right pyramidal meningioma: (a) T2WI; (b) T1WI; (c) FLAIR; and (d) post-contrast T1WI. The rim is hyperintense on T2WI scan (a), as shown by arrows; and hypointense on T1WI (b) scan. It is isointense to hypointense on FLAIR (c) and well enhanced on post-contrast T1WI scan (d). The hypointense dot (a, d) shown by the arrow head suggests a vascular flow void in the rim.

FLAIR: fluid attenuated inversion recovery; T1WI: T1-weighted image; T2WI: T2-weighted image.

Figure 3.

A 49-year-old woman with a right parasagittal meningioma: (a) T2WI; (b) T1WI; (c) FLAIR and (d) post-contrast T1WI. The rim is hyperintense on T2WI scan (a, arrows) and hypointense on T1WI scan (b). It is mainly isointense on FLAIR (c) and enhanced on post-contrast T1WI scan (d). The hypointense dot suggests a vascular flow void in the rim (a, c, arrow heads).

FLAIR: fluid attenuated inversion recovery; T1WI: T1-weighted image; T2WI: T2-weighted image.

Figure 4.

A 55-year-old woman with a right convexity meningioma: (a) T2WI; (b) T1WI; (c) FLAIR and (d) post-contrast T1WI. A hyperintense rim is seen on T2WI scan (a, arrows). On T1WI scan the rim is hypointense (b); on FLAIR it is mostly low intensity (c). It is not enhanced on post-contrast T1WI scan (d).

FLAIR: fluid attenuated inversion recovery; T1WI: T1-weighted image; T2WI: T2-weighted image.

Table 2 shows the clinico-radiologic features of the 53 patients. There was no significant difference with respect to the patient age and gender, the tumor site (infra- or supra-tentorial), the WHO grade, T2-rim thickness, tumor diameter and peri-tumoral edema (thickness of the hyperintensity area on FLAIR images was > 10 mm); however, compared to patients having T2-rim-negative or T2-rim-positive, but rim enhancement-negative tumors, those with T2-rim-positive and the rim enhancement-positive tumors tended to be younger (p = 0.071) and the tumor site tended to be infratentorial (p = 0.078).

Table 2.

Summary of clinico-radiologic factors according to T2-hyper rim and its enhancement effect (n = 53).

	T2 rim negative (n = 16)	T2 rim positive (n = 37)
		C/E negative (n = 9)	C/E positive (n = 28)
Age (years) mean (SD)	66.8 (11.6)	65.6 (10.3)	57.8 (13.3)	p = 0.071 Kruskal–Wallis test
Gender (M/F)	6 / 10	3 / 6	6 / 22	p = 0.489 Chi square test
Infra/ supra-tentorial	0 / 16	1 / 8	7 / 21	p = 0.078 Chi square test
WHO grade of meningioma (I/II/III)	13/3/0	7/1/1	25/3/0	p = 0.234 Chi square test
Thickness of T2 rim, mean (SD) mm	n.a.	1.1 (0.55)	1.48 (0.82)	p = 0.208 Student’s t test
Tumor diameter, mean (SD) mm	44.2 (22.3)	45.0 (16.2)	43.2 (14.7)	p = 0.958 Kruskal–Wallis test
Perifocal edema (>10 mm)	9 (56.2%)	2 (22.2%)	12 (42.9%)	p = 0.256 Chi square test

C/E: contrast enhancement of rim; F: female gender; M: male gender; mm: millimeter; n.a.: not applicable

Histological evaluation

To better understand the nature of the rim exhibiting T2 hyperintensity and contrast enhancement, we examined hematoxylin and eosin (HE)-stained tumor samples, the CD34-positive vascularity in the tumor capsule, and the vascular endothelial growth factor (VEGF) immunoreactivity of neoplastic cells. Table 3 shows the mean microvascular density in the tumor capsule was 1.55 ± 0.52 (SD) in 12 T2-rim- and rim enhancement-positive meningiomas in Group 1 (Figure 5): it was 0.33 ± 0.50 (SD) in the other 10 tumors that were T2-rim-negative (Figure 6) or T2-rim positive but rim enhancement-negative (Group 2). The difference between the two groups was statistically significant (p < 0.001, Mann–Whitney U test). There was no difference in the VEGF expression of neoplastic cells between the two groups in both the brain side (B) and intratumoral site (I). On the other hand, when we compared the expression of VEGF-A in specimens from (B) and (I) we found that the B-I was significantly higher in Group 1 than Group 2 (5.3 ± 4.0 versus 1.2 ± 2.4; p = 0.013, Mann–Whitney U test).

Table 3.

Summary of vascularity score and VEGF staining scores (n = 22).

		Group 1 (n = 12): T-2 rim positive and C/E positive	Group 2 (n = 10):	T2-rim negative or T2-rim positive; but C/E negative	p value
Microvascular density score in the tumor capsule		1.55 ± 0.52	0.33 ± 0.50		p < 0.001
VEGF-A expression score	brain side (B)	7.8 ± 3.2	6.1 ± 3.4		p = 0.25
	Intratumoral site (I)	2.5 ± 1.9	4.9 ± 3.9		p = 0.14
	B-I	5.3 ± 4.0	1.2 ± 2.4		p = 0.013

B: brain side; C/E: contrast enhancement of rim; I: intratumoral site; p value: by Mann–Whitney U test; VEGF: vascular endothelial growth factor

Figure 5.

A 54-year-old woman with a left convexity meningioma: (a) T2WI; (b) post-contrast T1WI; (c) hematoxylin–eosin staining; (d) CD34 immunostaining of tissue from the brain side of the tumor; (e) vascular endothelial growth factor (VEGF)-A immunostaining of tissue from the brain side of the tumor; (f) VEGF immunostaining of an intratumoral sample. Bars: 500 µm; circles: site of the pathologic specimen in brain side; asterisks: capsular layer of the tumor.

Note the T2-hyperintense rim (a) with positive contrast enhancement (b). H and E (c) and CD34 stains (d) show the rich vasculature in the tumor capsule. VEGF positivity was stronger on the brain side of the tumor (e), with VEGF score 12, rather than intratumorally (f), with VEGF score 4.

H and E: hematoxylin–eosin stain; T1WI: T1-weighted image; T2WI: T2-weighted image; VEGF: vascular endothelial growth factor.

Figure 6.

A 58-year-old woman with a left convexity meningioma: (a) T2WI; (b) post-contrast T1WI; (c) H and E stain; (d) CD34 immunostaining; (e) VEGF immunostaining of samples from the brain side of the tumor; and (f) VEGF immunostaining of an intratumoral sample. Bars: 100 µm; circles: site of the pathologic specimen in brain side; asterisks: capsular layer of the tumor.

There was no distinct rim at the brain–tumor interface (a) and (b). H and E (b) and CD34 stains (c) showed a very thin capsule with very scarce microvascular density. VEGF positivity on the brain side (d) with VEGF score: 1 and intratumorally was similarly scarce and weak (e), with VEGF score: 1.

H and E: hematoxylin and eosin stain; T1WI: T1-weighted image; T2WI: T2-weighted image; VEGF: vascular endothelial growth factor.

Discussion

Regarding the appearance and nature of the brain–tumor interface, Spagnoli et al.⁴, who used a 1.5 T scanner, reported that their survey indicated that the presence of a hyperintense ‘cleft’ on long TE and T2WI scans and iso- or slight hypointensity on short TE, T2WI scans reflected the CSF space. They observed this CSF cleft in 20 out of 25 meningiomas.⁴

Nakasu et al.² studied the brain–tumor interface using a 0.5T MR unit. They basically classified the rims into rims with low intensity on both T1WI and T2WI scans (Type 1), rims with low signal intensity on T1WI and high intensity on T2WI scans (Type 2), and their combinations. Their pathologic studies showed that Type 1 rims were comprised of thick collagenous tissue covering the tumor, and that Type 2 rims reflected the CSF space located at the brain–tumor interface. Among the 31 meningiomas in their survey, eight (25.8%) were surrounded by a Type 2 rim and seven (22.6%) were partially positive for a Type 2 rim; however, these surveys were conducted early in the era of clinical MRI studies, and their image quality was still suboptimal.

In 2003, Takeguchi et al.⁸ reported the results of their survey of the brain–tumor interface on FLAIR images acquired at 1.5 T. Among 50 meningiomas, 35 (70%) manifested low-intensity rims on T1WI and hyperintensity on T2WI scans.⁸ The rims tended to be isointense to hyperintense on FLAIR images, and were post-contrast enhanced. None of the 35 lesions exhibited low intensity on their FLAIR images. They asserted that the rim was not a CSF cleft, but rather representative of the capsular structure of the tumor surface; however, their study lacked histological assessment.

In their more recent survey on the brain–tumor interface on 3D-FLAIR images, Enokizono et al.⁹ assigned 4 grades to relative low-intensity rims on non-enhanced three-dimensional (3D) FLAIR scans: they ranged from not visible (Grade 0) to visible over most of the brain–tumor interface (Grade 3). They also used these four grades to classify relatively high-intensity rims on CE 3D FLAIR images. They found that high-grade rims on non-enhanced 3D FLAIR scans correlated positively with surgical brain–tumor cleavability and the amount of connective tissue at the interface; and that it negatively correlated with the degree of peritumoral brain edema. On the other hand, high-grade rims on CE 3D FLAIR images correlated with a prominent pial blood supply. They did not cite the possibility of a CSF cleft in the interface.

Our prospective analysis of 53 consecutively-operated meningiomas, using a 3T scanner, revealed that 37 of 53 (69.8%) showed a T2-hyperintensity rim. There were nine of the 37 (24.3%) that exhibited a hypointense rim on T1WI and FLAIR scans, and they lacked enhancement on post-gadolinium T1WI scans, suggesting that the rim reflected the CSF space. The other 28 rims (75.7%) were gadolinium-enhanced: this finding negated the possibility of their reflecting the CSF space.

Histologically, T2 hyperintense rims exhibiting an enhancement effect manifested rich microvascular density in the capsular layer covering the brain side of the meningioma. VEGF is a potent angiogenic factor that tends to be over-expressed in malignant neoplasms; it can also be expressed in benign tumors including meningiomas. VEGF expression in meningiomas have been investigated in view of its potent ability to elicit neo-angiogenesis, invasion into the dura mater and peritumoral edema; and its effect on the pial blood supply.^7,10-12

In our series there was no difference in the VEGF expression between tumors with a T2 hyperintense rim manifesting the enhancement effect and the other meningiomas. This observation may reflect sample to sample subtle differences in the immunostaining procedure, e.g. sample preparation, antigen retrieval, incubation time and temperature and antigen visualization, rather than tissue-to-tissue differences in the expression of VEGF; however, when we compared differences in the VEGF expression of specimens from the surface (brain side) and samples obtained from deep intratumoral sites, we found that the difference was significantly larger in tumors that were T2-rim- and enhancement-positive than in the other tumors. We think that the B-I (brain side minus intratumoral site) value reflected the actual VEGF expression on the brain side, because VEGF immunoreactivity in the samples from the intratumoral site served as the inner control.

It is notable that the rate of infratentorial tumors was higher among tumors with a T2 hyperintense rim manifesting the enhancement effect (7/21 or 33%) than among tumors with T2-rim-negative or T2-rim-positive CE-negative rims (1/24 or 4.2%) (p = 0.078, chi square test). Of the eight infratentorial meningiomas, seven (87.5%) showed a T2 rim manifesting the enhancement effect. All these seven meningiomas attached to the dura mater on the pyramidal bone. Differences in the blood supply to the tumor may contribute to differences in the manifestation of rims. We also found that patients with T2-rim CE-positive rims were younger than patients with T2-rim-negative or T2-rim-positive CE-negative rims (p = 0.071, Kruskal–Wallis test). Our observations must be confirmed in larger-scale surveys.

The tumor size was not different among our patients (p = 0.958, Kruskal–Wallis test), suggesting that the features of the brain–tumor interface on MRI scans are not affected by the tumor size, nor the degree of pressure on the arachnoid-pia-brain complex.

Conclusions

On high magnetic-field MRI scans, 37 of 53 meningiomas (69.8%) were T2-rim-positive and 28 (75.7%) of the 37 rims were also CE-positive. This denies that the rim reflects the CSF space. The role of the microvascularity, VEGF expression, tumor location, and the patient age with respect to the presence of a rim on MR studies must be further investigated.

Conflict of interest

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.