pISSN 2384-1095 iMRI 2021;25:23-34 https://doi.org/10.13104/imri.2021.25.1.23 eISSN 2384-1109

Differentiation between and Solitary : Morphologic Assessment by Conventional MR Imaging and Diffusion-Weighted Imaging

Bo Young Jung1, Eun Ja Lee1, Jong Myon Bae2, Young Jae Choi1, Eun Kyoung Lee1, Dae Bong Kim1 1Department of Radiology, Dongguk University Ilsan Hospital, Goyang-si, Korea 2Department of Preventive Medicine, Jeju National University School of Medicine, Jeju, Korea Original Article

Purpose: Differentiating between glioblastoma and solitary metastasis is very important for the planning of further workup and treatment. We assessed the ability Received: December 8, 2019 of various morphological parameters using conventional MRI and diffusion-based Revised: January 3, 2021 Accepted: January 4, 2021 techniques to distinguish between and solitary metastases in tumoral and peritumoral regions. Correspondence to: Materials and Methods: We included 38 patients with solitary brain tumors (21 Eun Ja Lee, M.D. glioblastomas, 17 solitary metastases). To find out if there were differences in the Department of Radiology, morphologic parameters of enhancing tumors, we analyzed their shape, margins, Dongguk University Ilsan Hospital, 814, Siksa-dong, Ilsandong-gu, and enhancement patterns on postcontrast T1-weighted images. During analyses of Goyang-si, Gyeonggi-do 10326, peritumoral regions, we assessed the extent of peritumoral non-enhancing lesion Korea. on T2- and postcontrast T1-weighted images. We also aimed to detect peritumoral Tel. +82-31-961-7836 neoplastic cell infiltration by visual assessment of T2-weighted and diffusion- Fax. +82-31-961-8281 based images, including DWI, ADC maps, and exponential DWI, and evaluated which E-mail: [email protected] sequence depicted peritumoral neoplastic cell infiltration most clearly. Results: The shapes, margins, and enhancement patterns of tumors all significantly differentiated glioblastomas from metastases. Glioblastomas had an irregular shape, ill-defined margins, and a heterogeneous enhancement pattern; on the other hand, metastases had an ovoid or round shape, well-defined margins, and homogeneous This is an Open Access article distributed under the terms of the Creative Commons enhancement. Metastases had significantly more extensive peritumoral T2 high signal Attribution Non-Commercial License intensity than glioblastomas had. In visual assessment of peritumoral neoplastic cell (http://creativecommons.org/licenses/ infiltration using T2-weighted and diffusion-based images, all sequences differed by-nc/4.0/) which permits unrestricted non-commercial use, distribution, and significantly between the two groups. Exponential DWI had the highest sensitivity for reproduction in any medium, provided the diagnosis of both glioblastoma (100%) and metastasis (70.6%). A combination the original work is properly cited. of exponential DWI and ADC maps was optimal for the depiction of peritumoral neoplastic cell infiltration in glioblastoma. Conclusion: In the differentiation of glioblastoma from solitary metastatic lesions, visual morphologic assessment of tumoral and peritumoral regions using conventional Copyright © 2021 Korean Society MRI and diffusion-based techniques can also offer diagnostic information. of Magnetic Resonance in Medicine (KSMRM) Keywords: Glioblastoma; Intracranial metastasis; MR imaging; Diffusion weighted imaging

www.i-mri.org 23 Morphologic Assessment for Glioblastoma and Solitary Metastasis | Bo Young Jung, et al.

INTRODUCTION of MRI sequences that are easier to implement, such as conventional MRI, diffusion-weighted imaging (DWI), ADC Glioblastomas and metastases represent the most- maps, and exponential DWI for the differentiation of these common malignant brain tumors in adults (1). When tumors. Previous study has indicated that exponential DWI, intracranial tumors are encountered, a history of primary which represents the negative exponential of ADC maps and malignancy or the presence of multifocal lesions may assist depicts diffusion effects more accurately by removing the in the diagnosis of metastasis, but differentiation is often T2 shine-through effect, is useful for distinguishing tumors difficult when patients present with a solitary enhancing (11); therefore, we included this sequence in this study, in mass of unknown primary malignancy. which we individually evaluated whether morphological Because the clinical management of these two types of parameters of enhancing tumors can distinguish between tumor is radically different, it is important to differentiate glioblastomas and solitary metastases and assessed T2- them. Patients with glioblastomas do not typically require weighted and diffusion-based images for visual analysis of systemic work-up, because tumor invasion outside of the peritumoral neoplastic cell infiltration in the peritumoral central nervous system is very rare. However, any patient region. We also identified the sequence that depicts with suspected brain metastasis and no previous history of peritumoral neoplastic cell infiltration most clearly. To the primary should undergo systemic staging to detect best of our knowledge, no previous study has considered all its origin, and any evidence of distant metastasis, before these factors. treatment. Standard treatment of glioblastoma consists of maximal surgical resection, radiotherapy, and concomitant and adjuvant . In patients with metastasis, MATERIALS AND METHODS more conservative management (usually a nonsurgical approach) should be considered (2-5). This retrospective study was approved by our Institutional Conventional magnetic resonance imaging (MRI) is of Review Board. limited value for the differentiation of glioblastoma from metastasis, because of these tumors’ similar imaging Patients appearances (6-8). Therefore, numerous studies have used We retrospectively reviewed the MRI examinations of advanced MRI techniques, such as MR spectroscopy (MRS), 38 patients with a diagnosis of glioblastoma or solitary perfusion-weighted imaging (PWI), diffusion tensor imaging metastasis between February 2006 and May 2014. The (DTI), and measurement of the absolute apparent diffusion patients ranged in age from 41 to 87 years (mean age, coefficient (ADC) in an attempt to differentiate them (2, 3, 62 ± 8.6 years); there were 20 males and 18 females. All 9-12). The key to differentiating the two lies in patients had a previously untreated solitary enhancing the peritumoral region. In glioblastomas, peritumoral non- and peritumoral non-enhancing lesion and had enhancing T2 high signal intensity is the result of both undergone conventional brain MRI and DWI before surgical neoplastic cell infiltration and vasogenic edema, whereas in intervention. Patients with infratentorial lesion were metastases it results from pure vasogenic edema (12-15). excluded, as were those with a previous history of surgery The continued development of advanced MRI techniques or whose peritumoral T2 high-signal lesion was not large has allowed for some success in the differentiation of enough to evaluate on T2-weighted imaging. Patients with peritumoral neoplastic cell infiltration from pure vasogenic obvious movement artifacts were also excluded. In total, 10 edema, in the peritumoral region using quantitative patients were excluded. methods. However, most investigators have found no Of the 38 patients, WHO grade IV glioblastoma was significant differences in intratumoral areas using these diagnosed in 21 cases and solitary metastasis in 17. new techniques (2, 3, 9-12, 16). Furthermore, advanced Metastatic brain tumors included lung carcinoma (n = MRI is not available in all centers and requires more time 8), anorectal carcinoma (n = 2), breast carcinoma (n = 1), and expense; this can put great pressure on patients. thyroid carcinoma (n = 1), hepatic cellular carcinoma (n = Quantitative evaluation is not easy, even using advanced 1), gall bladder cancer (n = 1), and carcinoma of unknown MRI techniques, and differential diagnosis remains a origin (n = 3). Diagnosis of glioblastoma was made by challenge with a reported accuracy of < 65% (8, 17). For means of pathology in all patients; diagnosis of solitary these reasons, we attempted to assess the diagnostic utility metastasis was confirmed pathologically in 13 patients and

24 www.i-mri.org https://doi.org/10.13104/imri.2021.25.1.23 was made clinically in the remaining four patients based 0.1 mmol/kg of body weight. on their history, MRI findings, and response to palliative Of the patients, 33 underwent T2* gradient-echo chemotherapy and/or . (GRE) images. For the MRI examinations of 29 patients (14 glioblastomas, 15 metastases), T2* gradient-echo MRI Techniques (GRE) images were evaluated to assess the presence of All MRI examinations were done using a 1.5 T MRI intratumoral hemorrhage. The remaining 4 patients were scanner (Avanto, Siemens Healthcare, Erlangen, Germany) excluded because of severe motion artifacts. with a standard head coil. Standard T1- and T2-weighted images, fluid attenuated inversion recovery (FLAIR) images, Image Analysis contrast-enhanced T1 weighted images and DWI were All MRIs were analyzed in consensus by a staff obtained for all patients. The following pulse sequences neuroradiologist (with 20 years of clinical experience) and were acquired. a fourth-year radiology resident; the two observers were • For precontrast and contrast-enhanced axial T1- blinded to patient information including age, sex, clinical weighted images: repetition time (TR)/echo time (TE)/ history, and histopathology. inversion time (TI), 1700/14/745 ms; slice thickness, In a first step, we evaluated the morphologic parameters 5 mm; intersection gap, 2 mm; matrix size, 320 × of enhancing tumors. All enhancing tumors on post- 210; field of view (FOV), 19.3 × 22.0 cm; number of contrast T1-weighted images were analyzed in accordance excitation (NEX), 3. with the following morphologic criteria: shape (irregular • For axial T2-weighted images: TR/TE, 4480/89 ms; slice / round or ovoid), margin (ill-defined / well-defined), and thickness, 5 mm; intersection gap, 2 mm; matrix size, enhancement pattern (heterogeneous / homogeneous). 448 × 294; FOV, 19.3 × 22.0 cm, NEX, 2. Ill-defined margin was defined as having spiculated and • For axial FLAIR images: TR/TE/TI, 4780/93/1745 ms; jagged edge with sharp demarcation, or a poorly defined slice thickness, 5 mm; intersection gap, 2 mm; matrix margin in which the tumor cannot be differentiated from size, 320 × 196; FOV, 19.3 × 22.0 cm; NEX, 2. the normal brain parenchyma; well-defined margin was Recently, three-dimensional T1 black-blood fast spin-echo defined as having a clearly distinguishable smooth edge. We imaging has been found effective for selective suppression also recorded the presence of intratumoral hemorrhage on of blood vessels and better detection of brain metastases T2* gradient echo images. (18), but no patient in our study has obtained this imaging. In the next step, we evaluated the peritumoral regions DWI was done by using a single-shot spin-echo echo- to differentiate glioblastomas from metastases. Prior to planar imaging sequence with the following parameters (TR/ analysis, we defined 'peritumoral non-enhancing lesion' TE, 6000/63-99 ms; slice thickness, 3 mm; intersection gap, as a hyperintense region on T2-weighted images without 0.3 mm; matrix size, 140 × 140; FOV, 22.0 × 22.0 cm; NEX, enhancement on post-contrast T1-weighted images, 1). DWI was done in three orthogonal directions using a commonly referred to as a non-enhancing abnormal area maximum b-value of 1000 s/mm2. A baseline image with a surrounding the enhancing tumor core. b-value of 0 s/mm2 was also obtained. We assessed the extent of peritumoral non-enhancing A suppression pulse was implemented lesions using T2-weighted and postcontrast T1-weighted to avoid diffusion falsely elevated by cerebrospinal fluid imaging. We measured the ratio between the maximum partial-volume averaging. On a pixel-by-pixel basis, ADC diameter of the peritumoral non-enhancing lesion on maps and exponential DWI were calculated from isotropic T2-weighted images and the maximum diameter of the DWI using the following formula: enhancing mass on post-contrast T1-weighted images. For visual analysis of peritumoral neoplastic cell -ADC × b SIeDWI = SIDWI(b=1000) / SIDWI(b=0) = e infiltration, we assessed T2-weighted and diffusion-based images, such as DWI, ADC maps, and exponential DWI,

(SIeDWI, signal intensity of exponential DWI; SIDWI(b=1000), with reference to the same section on post-contrast T1- signal intensity of DWI with a b-value of 1000 s/mm2 ; weighted images; each image was analyzed separately. We 2 SIDWI(b=0), signal intensity of DWI with a b-value of 0 s/mm ). hypothesized that peritumoral neoplastic cell infiltration In all patients, contrast agent (Gadovist; Bayer Schering, would have abnormal signal intensity, as in the contrast- Berlin, Germany) was administered at the standard dose of enhancing region. To evaluate the presence of peritumoral www.i-mri.org 25 Morphologic Assessment for Glioblastoma and Solitary Metastasis | Bo Young Jung, et al. neoplastic cell infiltration, we carefully selected non- the peritumoral non-enhancing lesion to the maximum enhancing peritumoral areas based on T2-weighted and diameter of the enhancing mass exhibited the optimum post-contrast T1-weighted images. We employed the combination of sensitivity (76.2%, 95% confidence interval following criteria to denote 'peritumoral neoplastic cell [CI]: 52.8, 91.8) and specificity (70.6%, 95% CI: 44.0, 89.7, infiltration' in each sequence: (1) the region is attached P = 0.008). to the enhancing tumor with encompassment > 180°, to During visual assessment of peritumoral neoplastic cell avoid contamination from the signal intensity of the normal infiltration using T2-weighted and diffusion-based images cortex; and (2) the region exhibits the same degree of signal (Fig. 2), all sequences differed significantly between the two intensity as does the . We visually assessed groups (T2-weighted imaging, P = 0.001; DWI, P = 0.006; the presence or absence of peritumoral neoplastic cell ADC map, P = 0.000; exponential DWI, P = 0.000; Table 2). infiltration. Glioblastomas exhibited significant peritumoral neoplastic cell infiltration on all sequences (Fig. 3a and Fig. 4); on Statistical Analysis the other hand, peritumoral neoplastic cell infiltration in We did statistical analyses using the SPSS software metastases was significantly absent (Fig. 3b and Fig. 5). package (ver. 20.0; SPSS, Inc., Chicago, IL, USA). To find However, metastatic brain lesions showed many false- out whether there were significant differences between positive results (35% in T2WI, 59% in DWI, 47% in ADC, 29% glioblastoma and metastasis, we analyzed the morphologic in exponential DWI) in detecting peritumoral neoplastic cell parameters of the enhancing tumor and peritumoral regions infiltration (Fig. 6). Exponential DWI showed the highest using the chi-squared test. We also evaluated which sensitivity in both the glioblastoma (100%) and metastasis sequence was best for depicting peritumoral neoplastic (70.6%) groups; DWI was the least-sensitive sequence. cell infiltration. Furthermore, we used receiver operating The combination of exponential DWI and ADC maps was characteristic (ROC) analysis to assess the extent of the optimum for the depiction of peritumoral neoplastic cell peritumoral non-enhancing lesion and to find the cutoff infiltration in glioblastomas, with a sensitivity of 100% value with the optimum combination of sensitivity and (Table 3). specificity for distinguishing glioblastoma from metastasis; P < 0.05 was taken to indicate statistical significance.

Table 1. Patient Characteristics and Morphologic Parameters of RESULTS the Enhancing Tumor Solitary Parameters Glioblastoma P value Table 1 describes the predictive values of the morphologic metastasis criteria of enhancing tumors. The shape (P = 0.006), Age (y) 60 ± 10.05 64 ± 6.16 0.1599 margin (P = 0.003), and enhancement pattern (P = 0.046) Sex 57.1 47.1 0.745 of tumors significantly differentiated glioblastoma from (% of men) metastasis; the largest difference between the two groups Irregular 18/21 7/17 Shape of the (85.7%) (41.2%) was for tumor margin. Glioblastomas were significantly 0.006 associated with an irregular shape, ill-defined margins, and tumor Round or ovoid 3/21 10/17 (14.3%) (58.8%) a heterogeneous enhancement pattern (Fig. 1a). An ovoid Ill-defined 16/21 4/17 or round shape, well-defined margins, and homogeneous Margin of (76.2%) (23.5%) 0.003 enhancement pattern were significantly related to the tumor Well-defined 5/21 13/17 metastases (Fig. 1b). There were no significant group (23.8%) (76.5%) differences in patient age (P = 0.160), sex (P = 0.745), or Heterogeneous 16/21 7/17 Enhancement (76.2%) (41.2%) presence of hemorrhage (P = 0.169). 0.046 During assessment of the extent of peritumoral non- of the tumor Homogeneous 5/21 10/17 enhancing lesion, we observed that metastases had (23.8%) (58.8%) Presence 13/14 10/15 significantly more extensive edema than glioblastomas Hemorrhage (92.9%) (66.7%) 0.169 had (Fig. 2). According to ROC curve analysis, a cutoff in the tumor Absence 1/14 5/15 value of 2.21 for the ratio of the maximum diameter of (7.1%) (33.3%)

26 www.i-mri.org https://doi.org/10.13104/imri.2021.25.1.23

a b Fig. 1. Different morphologic patterns in enhancing tumors of glioblastoma and metastasis. Post-contrast T1-weighted image (a) in a 61-year-old female with glioblastoma exhibits an irregular, ill-defined, heterogeneously enhancing mass in the left frontal lobe with peritumoral non-enhancing lesion, representing a typical enhancement pattern of glioblasma. Post-contrast T1-weighted image (b) in a 68-year-old male with solitary metastasis from exhibits a round, well- defined, homogeneously enhancing mass in the right superior frontal gyrus with more extensive peritumoral non-enhancing lesion, representing typical enhancement pattern of metastasis.

DISCUSSION

Our results indicate that visual assessment of enhancing tumors and peritumoral non-enhancing lesions can help to differentiate glioblastoma from metastasis. The key to doing that lies in the peritumoral non-enhancing lesion beyond the enhancing margin of the mass. Peritumoral

Table 2. MR Sequence in Assessment of the Peritumoral Neoplastic Cell Infiltration Peritumoral Solitary MR Glioblastoma neoplastic cell metastasis P value sequence (n = 21) infiltration (n = 17) Presence 18 (85.7%) 6 (35.3%) T2WI 0.001 Absence 3 (14.3%) 11 (64.7%) Presence 20 (95.2%) 10 (58.8%) DWI 0.006 Fig. 2. Empiric receiver operating characteristic (ROC) curve Absence 1 (4.8%) 7 (41.2%) to assess the extent of the peritumoral non-enhancing Presence 21 (100%) 8 (47.1%) lesion, used to distinguish glioblastoma from metastasis. ADC 0.000 The area under the ROC curve was 0.706 (95% CI: 0.54, 0.84). Absence 0 (0%) 9 (52.9%) At a cutoff ratio value of 2.21, the sensitivity (76.2%) and Presence 21 (100%) 5 (29.4%) e-DWI 0.000 specificity (70.6%) combination was optimal. Absence 0 (0%) 12 (70.6%) ADC = apparent diffusion coefficient; DWI = diffusion-weighted imaging; e-DWI = exponential diffusion-weighted imaging; T2WI = T2-weighted imaging www.i-mri.org 27 Morphologic Assessment for Glioblastoma and Solitary Metastasis | Bo Young Jung, et al.

a

b Fig. 3. MRI imaging templates for peritumoral neoplastic cell infiltration. (a) Positive peritumoral neoplastic cell infiltration on all imaging sequences of T2WI, DWI, ADC map, and exponential DWI (eDWI), representing a typical pattern of glioblastoma. Peritumoral non-enhancing lesion shows heterogeneous signal intensity with a gradient of measured values on each imaging sequence. (b) Negative peritumoral neoplastic cell infiltration on all imaging sequences of T2WI, DWI, ADC, and exponential DWI (eDWI), representing typical pattern of metastasis. Peritumoral non-enhancing lesion shows nearly homogeneous signal intensity with nearly homogeneous values on each imaging sequence. non-enhancing lesion is generally considered vasogenic. The have been found in the non-enhancing-T2 hyperintense term 'vasogenic edema' has been used to describe regions regions that surround glioblastoma histopathologically. of T2 high signal abnormality surrounding the enhancing Infiltrating neoplastic cells are situated behind the blood- tumor. On histopathological examination, vasogenic edema brain barrier and typically invade along white-matter tracts appears as a disruption of the blood-brain barrier, tumor- (13). Therefore, a peritumoral non-enhancing lesion of induced increases in interstitial water, and altered capillary glioblastoma consists of 'vasogenic' plus 'neoplastic cell permeability. In glioblastomas, which exhibit an aggressive infiltrative' edema. However, in metastases, peritumoral and infiltrative pattern of growth, peritumoral areas areas contain no infiltrating neoplastic cells, because of demonstrate not only altered interstitial water, but also their expansile growth pattern that causes displacement of scattered neoplastic cell infiltration. Indeed, neoplastic cells surrounding brain tissues rather than invasion; increased

28 www.i-mri.org https://doi.org/10.13104/imri.2021.25.1.23

a b c

Fig. 4. Glioblastoma in an 81-year- old female. Post-contrast T1- (a) and T2-weighted images (b) show that a heterogeneously enhancing mass has an irregular margin and ill-defined border in the right temporal lobe with peritumoral non-enhancing edema. The peritumoral region exhibits different signal intensities. On T2-weighted images, ADC maps (c) and exponential DWI (d), intermediate signal intensity lesions (arrows) are thought to indicate neoplastic cell infiltrative edema. Relatively pure vasogenic edema (arrowheads) exhibits high signal intensity on T2-weighted image and d e ADC maps, and low signal intensity on exponential DWI. DWI shows that the differentiation of tumor infiltrative edema and pure vasogenic edema are relatively indistinct on DWI (e). permeability occurs only when white-matter fibers become the site of hemorrhage, presence of necrosis, and different relatively loose around the tumor membrane (3, 6, 8, 12, 19, degrees of hypercellularity. Glioblastoma typically exhibits 20). thick, irregular, ring-like or heterogeneous enhancement We believe that the different growth patterns that with central hypo-intense necrosis (21, 22). Several characterize glioblastoma and metastasis may influence pathologists have suggested that, in all infiltrative tumor morphology. We found that glioblastomas exhibited glioblastomas, there is no clear margin microscopically an irregular shape, ill-defined margins, and heterogeneous (23). In contrast, metastatic tumors have a macroscopically enhancement, whereas metastases were ovoid or round, ovoid or round shape, well-defined margins, and exhibit with well-defined margins and homogeneous enhancement. homogeneous enhancement. Microscopically, brain The intratumoral heterogeneity of glioblastomas reflects metastases are also typically sharply demarcated and are www.i-mri.org 29 Morphologic Assessment for Glioblastoma and Solitary Metastasis | Bo Young Jung, et al.

a b c

Fig. 5. Solitary metastasis from lung cancer in a 68-year-old female. Post- contrast T1-weighted (a) and T2- weighted images (b) show that a heterogeneously enhancing mass has irregular margin and ill-defined border in the right occipital lobe with extensive peritumoral T2 high signal intensity. Peritumoral regions exhibit relatively homogeneous high signal intensity on T2-weighted image and ADC map (c), and low signal intensity on exponential DWI (d) with no significant evidence of tumor infiltrative edema. It is notable that d e the tumor and peritumoral lesion are poorly delineated on DWI (e).

Table 3. Combination of MR Sequences for the Assessment of surrounded by gliotic brain parenchyma, which is a very Peritumoral Neoplastic Cell Infiltration helpful feature in the differentiation of primary diffuse Positive peritumoral Negative peritumoral Combination of MR . However, they may infiltrate into the surrounding infiltration in infiltration in solitary sequences tissue in the later stages. Thus, metastatic lesions are often glioblastoma metastasis less circumscribed with surrounding reactive gliosis and e-DWI+ADC 21 (100%) 9 (52.9%) perivascular extension (24, 25). e-DWI+T2WI 18 (85.7%) 9 (52.9%) Our results showed that ovoid homogeneous enhancing e-DWI+ADC+T2WI 18 (85.7%) 7 (41.2%) glioblastomas may mimic metastatic tumors, and ill-defined e-DWI+ADC+DWI 20 (95.2%) 5 (29.4%) irregular heterogeneous metastatic tumors may mimic e-DWI+T2WI+DWI 17 (81%) 5 (29.4%) glioblastoma in tumor appearance. e-DWI+ADC+T2WI+DWI 17 (81%) 3 (17.6%) We hypothesized that peritumoral neoplastic cell ADC = apparent diffusion coefficient; DWI = diffusion-weighted imaging; e-DWI infiltration in glioblastoma would exhibit abnormal signal = exponential diffusion-weighted imaging; T2WI = T2-weighted imaging intensity, like that of the contrast-enhancing region. We visually assessed the presence or absence of peritumoral

30 www.i-mri.org https://doi.org/10.13104/imri.2021.25.1.23

a b c

Fig. 6. A false-positive case for peritumoral neoplastic cell infiltration in the solitary metastasis from lung cancer. Post-contrast T1-weighted image (a) shows a rim enhancing mass with extensive peritumoral non- enhancing lesion. We interpretated peritumoral non-enhancing lesion as positive peritumoral neoplastic cell infiltration on T2-weighted image (b), DWI (c) and ADC map (d). However, exponential DWI (e) was interpretated as negative peritumoral neoplastic cell d e infiltration. neoplastic cell infiltration on T2 weighted and diffusion- shine-through effects. The signal intensity of exponential based images. There was a significant difference in all DWI, which is generated on a workstation using simple sequences between the two groups. Glioblastomas exhibited image algebra, is derived by dividing the maximal B value significant peritumoral neoplastic cell infiltration on all DWI image by the B0 image and exhibits reverse signal sequences, compared to metastasis. Exponential DWI intensity on ADC maps. Therefore, exponential DWI more exhibited the highest sensitivity, for both the presence of accurately depicts diffusion effects than does DWI, with no peritumoral neoplastic cell infiltration in glioblastomas T2 shine-through effects (26, 27). Exponential DWI allowed (100%) and its absence in metastases (70.6%). DWI for increased lesion delineation during visual assessment of exhibited the least sensitivity. The signal intensity of DWI peritumoral neoplastic cell infiltration and could distinguish is influenced by both the intrinsic T2 properties of the lesions better than ADC maps could, despite being merely tissue and water diffusibility. 'T2 shine-through effects' a reverse image of these maps. The combination of refer to high signal intensities on DWIs that result not exponential DWI and ADC maps was useful for depicting from restricted diffusion, but rather from T2 hyperintensity peritumoral neoplastic cell infiltration in glioblastomas, (26). In the present study, one glioblastoma patient with a sensitivity of 100%. showed negative peritumoral neoplastic cell infiltration, Metastatic brain lesions in our study showed many and 10 metastases patients exhibited positive peritumoral false-positive results (35% in T2WI, 59% in DWI, 47% in neoplastic cell infiltration on DWI. We believe that the ADC, 29% in eDWI) in detecting peritumoral neoplastic lower sensitivity of DWI, in terms of detecting peritumoral cell infiltration. We believe that the following may have neoplastic cell infiltration, might have resulted from T2 affected this result: contamination of the normal cortex www.i-mri.org 31 Morphologic Assessment for Glioblastoma and Solitary Metastasis | Bo Young Jung, et al. when assessing peritumoral neoplastic cell infiltration peritumoral mean diffusivity surrounding metastasis than in peritumoral non-enhancing lesions; lower imaging around glioblastoma. However, no significant differences resolution by the relatively low signal-to-noise ratio of 1.5T in peritumoral fractional anisotropy have been observed MRI; the aforementioned T2-shine-through effect in DWI; (11, 12). Lee et al. (2) found that the peritumoral minimum reactive gliosis and infiltration into the surrounding tissue ADC value in glioblastoma was significantly lower than that in the later-staged metastatic lesions. of metastasis. Recently, Bauer et al. (31) have shown that During assessment of the extent of peritumoral non- a multiparametric approach with advanced MR imaging enhancing lesion, we observed that metastases had might be useful. They showed that the combination of significantly more extensive edema than glioblastomas diffusion-weighted imaging, DSC perfusion, and dynamic had. Metastases are surrounded by massive amounts contrast-enhanced perfusion MR metrics in peritumoral of edema, often extending to regions far from the site T2 hyperintensity area can help the differentiation of of a small metastatic focus. Pure vasogenic edema of glioblastoma from solitary brain metastasis with an metastases represents an underlying disturbance in accuracy of 98%. vascular permeability, in which plasma proteins and other Although advanced MRI methods help to distinguish macromolecules pass freely into the perivascular space and glioblastoma from metastasis, we suggest the use of a more consequently into the interstitial extracellular space. The intuitive method with easily implemented MRI sequences, extent of a peritumoral non-enhancing lesion is unrelated i.e., conventional T2-weighted and diffusion-based images to the size of the enhancing tumor or, necessarily, to such as DWI, ADC maps, and exponential DWI. However, the clinical status of the patient (28). Maurer et al. (17) this present study represents a simple visual morphologic suggested that measuring the extent of peritumoral non- assessment and does not describe a quantitative enhancing lesion has diagnostic potential for differentiating methodology. metastasis from glioblastoma. This study had several limitations. Biopsy of peritumoral Numerous studies that have used advanced MRI non-enhancing lesions was not done as part of the techniques, such as MRS, PWI, DTI, and ADC maps, have histologic examination at the time of surgery. Furthermore, provided valuable information about the differentiation we used a retrospective design and a limited number of of glioblastoma from metastasis in peritumoral regions patients; because of the small sample, factors that were (3, 9-12). On MRS, an elevated choline to creatinine (Cho/ less useful for differentiating glioblastoma from metastasis Cr) ratio is found in peritumoral regions of glioblastomas may not have been included, and important combinations because of neoplastic cell infiltration, but there is no of factors may have also been missed. Depending on the increase in the Cho/Cr ratio in the peritumoral regions isocitrate dehydrogenase (IDH1) mutation status, 1p19q of metastasis (3, 6, 9, 11). There is also no appreciable codeletion and methylation status of O-6-methylguanine- difference in the peritumoral N-acetylaspartate to creatine DNA methyltransferase (MGMT), glioblastoma has had (NAA/Cr) ratios of the two groups, because of an absence different treatment successes, different patient populations, of neuronal replacement or destruction (3). On PWI, and different imaging findings in both conventional and several studies have revealed that the peritumoral regions advanced MR imaging (32-34). In our study, patients’ of glioblastomas exhibit increased relative cerebral blood diagnoses with glioblastoma were based only on histological volume (rCBV), calculated in accordance with the ratio phenotypes, because they were diagnosed before WHO between CBV in the pathologic area and in contralateral 2016 revision. Therefore, we failed to evaluate the different white matter, because of neoangiogenesis and tumor imaging findings of glioblastoma based on genetic infiltration (2, 3, 11, 29). Sunwoo et al. (30) showed that mutations and molecular markers. both intratumoral and peritumoral perfusion using arterial In conclusion, we found that visual assessment of tumoral spin labeling (ASL) perfusion MR imaging can aid in the and peritumoral regions, using conventional MRI and differentiation of glioblastoma from brain metastasis. Their diffusion-based techniques, allows us to distinguish these study demonstrated that glioblastomas had significantly two types of tumor. higher intratumoral and peritumoral perfusion than brain metastases had, and that peritumoral perfusion in Acknowledgments particular provided stronger differentiation power. On DTI, We thank In Seong Kim, MR scientist of Siemens, for several investigators have reported significantly greater assistance in obtaining exponential DWI.

32 www.i-mri.org https://doi.org/10.13104/imri.2021.25.1.23

REFERENCES 1987;66:865-874 14. Strugar J, Rothbart D, Harrington W, Criscuolo GR. Vascular 1. Louis DN, Ohgaki H, Wiestler OD, et al. The 2007 WHO permeability factor in brain metastases: correlation classification of tumours of the central nervous system. with vasogenic brain edema and tumor angiogenesis. J Acta Neuropathol 2007;114:97-109 Neurosurg 1994;81:560-566 2. Lee EJ, terBrugge K, Mikulis D, et al. Diagnostic value of 15. Kelly PJ, Daumas-Duport C, Scheithauer BW, Kall BA, peritumoral minimum apparent diffusion coefficient for Kispert DB. Stereotactic histologic correlations of computed differentiation of glioblastoma multiforme from solitary tomography- and magnetic resonance imaging-defined metastatic lesions. AJR Am J Roentgenol 2011;196:71-76 abnormalities in patients with glial neoplasms. Mayo Clin 3. Law M, Cha S, Knopp EA, Johnson G, Arnett J, Litt AW. Proc 1987;62:450-459 High-grade gliomas and solitary metastases: differentiation 16. Al-Okaili RN, Krejza J, Wang S, Woo JH, Melhem ER. by using perfusion and proton spectroscopic MR imaging. Advanced MR imaging techniques in the diagnosis of Radiology 2002;222:715-721 intraaxial brain tumors in adults. Radiographics 2006;26 4. Furnari FB, Fenton T, Bachoo RM, et al. Malignant astrocytic Suppl 1:S173-189 : genetics, biology, and paths to treatment. Genes 17. Maurer MH, Synowitz M, Badakshi H, et al. Glioblastoma Dev 2007;21:2683-2710 multiforme versus solitary supratentorial brain metastasis: 5. Cha S, Lupo JM, Chen MH, et al. Differentiation of differentiation based on morphology and magnetic glioblastoma multiforme and single brain metastasis by resonance signal characteristics. Rofo 2013;185:235-240 peak height and percentage of signal intensity recovery 18. Park YW, Ahn SJ. Comparison of contrast-enhanced T2 derived from dynamic susceptibility-weighted contrast- FLAIR and 3D T1 black-blood fast spin-echo for detection enhanced perfusion MR imaging. AJNR Am J Neuroradiol of leptomeningeal metastases. Investig Magn Reson 2007;28:1078-1084 Imaging 2018;22:86-93 6. Chiang IC, Kuo YT, Lu CY, et al. Distinction between high- 19. Burger PC, Vogel FS, Green SB, Strike TA. Glioblastoma grade gliomas and solitary metastases using peritumoral multiforme and anaplastic . Pathologic criteria 3-T magnetic resonance spectroscopy, diffusion, and and prognostic implications. Cancer 1985;56:1106-1111 perfusion imagings. Neuroradiology 2004;46:619-627 20. Oh J, Cha S, Aiken AH, et al. Quantitative apparent 7. Opstad KS, Murphy MM, Wilkins PR, Bell BA, Griffiths JR, diffusion coefficients and T2 relaxation times in Howe FA. Differentiation of metastases from high-grade characterizing contrast enhancing brain tumors and gliomas using short echo time 1H spectroscopy. J Magn regions of peritumoral edema. J Magn Reson Imaging Reson Imaging 2004;20:187-192 2005;21:701-708 8. Tang YM, Ngai S, Stuckey S. The solitary enhancing cerebral 21. Smirniotopoulos JG, Murphy FM, Rushing EJ, Rees JH, lesion: can FLAIR aid the differentiation between glioma Schroeder JW. Patterns of contrast enhancement in the and metastasis? AJNR Am J Neuroradiol 2006;27:609-611 brain and meninges. Radiographics 2007;27:525-551 9. Ishimaru H, Morikawa M, Iwanaga S, Kaminogo M, Ochi 22. Kunimatsu A, Kunimatsu N, Kamiya K, Watadani T, Mori M, Hayashi K. Differentiation between high-grade glioma H, Abe O. Comparison between glioblastoma and primary and metastatic brain tumor using single-voxel proton MR central nervous system using MR image-based spectroscopy. Eur Radiol 2001;11:1784-1791 texture analysis. Magn Reson Med Sci 2018;17:50-57 10. Cho SK, Na DG, Ryoo JW, et al. Perfusion MR imaging: 23. VandenBerg SR. Current diagnostic concepts of astrocytic clinical utility for the differential diagnosis of various brain tumors. J Neuropathol Exp Neurol 1992;51:644-657 tumors. Korean J Radiol 2002;3:171-179 24. Fink KR, Fink JR. Imaging of brain metastases. Surg Neurol 11. Lee EJ, Ahn KJ, Lee EK, Lee YS, Kim DB. Potential role of Int 2013;4:S209-219 advanced MRI techniques for the peritumoural region 25. Takei H, Rouah E, Ishida Y. Brain metastasis: clinical in differentiating glioblastoma multiforme and solitary characteristics, pathological findings and molecular metastatic lesions. Clin Radiol 2013;68:e689-697 subtyping for therapeutic implications. Brain Tumor Pathol 12. Lu S, Ahn D, Johnson G, Cha S. Peritumoral diffusion tensor 2016;33:1-12 imaging of high-grade gliomas and metastatic brain 26. Provenzale JM, Engelter ST, Petrella JR, Smith JS, MacFall tumors. AJNR Am J Neuroradiol 2003;24:937-941 JR. Use of MR exponential diffusion-weighted images to 13. Kelly PJ, Daumas-Duport C, Kispert DB, Kall BA, Scheithauer eradicate T2 "shine-through" effect. AJR Am J Roentgenol BW, Illig JJ. Imaging-based stereotaxic serial biopsies 1999;172:537-539 in untreated intracranial glial neoplasms. J Neurosurg 27. Engelter ST, Provenzale JM, Petrella JR, Alberts MJ, DeLong www.i-mri.org 33 Morphologic Assessment for Glioblastoma and Solitary Metastasis | Bo Young Jung, et al.

DM, MacFall JR. Use of exponential diffusion imaging to of solitary brain metastasis from glioblastoma multiforme: determine the age of ischemic infarcts. J a predictive multiparametric approach using combined MR 2001;11:141-147 diffusion and perfusion. Neuroradiology 2015;57:697-703 28. Penn RD. and neurological function: CT, 32. Kwon YW, Moon WJ, Park M, et al. Dynamic susceptibility evoked responses, and clinical examination. Adv Neurol contrast (DSC) perfusion MR in the prediction of long-term 1980;28:383-394 survival of glioblastomas (GBM): correlation with MGMT 29. Calli C, Kitis O, Yunten N, Yurtseven T, Islekel S, Akalin promoter methylation and 1p/19q deletions. Investig Magn T. Perfusion and diffusion MR imaging in enhancing Reson Imaging 2018;22:158-167 malignant cerebral tumors. Eur J Radiol 2006;58:394-403 33. Smits M, van den Bent MJ. Imaging correlates of adult 30. Sunwoo L, Yun TJ, You SH, et al. Differentiation of glioma genotypes. Radiology 2017;284:316-331 glioblastoma from brain metastasis: qualitative and 34. Yamashita K, Hiwatashi A, Togao O, et al. MR imaging- quantitative analysis using arterial spin labeling MR based analysis of glioblastoma multiforme: estimation of imaging. PLoS One 2016;11:e0166662 IDH1 mutation status. AJNR Am J Neuroradiol 2016;37:58- 31. Bauer AH, Erly W, Moser FG, Maya M, Nael K. Differentiation 65

34 www.i-mri.org