Histopathology and MIB-1 Labeling Index Predicted Recurrence of Meningiomas. A Proposal of Diagnostic Criteria for Patients with Atypical Meningioma

1538 Histopathology and MIB-1 Labeling Index Predicted Recurrence of Meningiomas A Proposal of Diagnostic Criteria for Patients with Atypical Meningioma Donald Ming-Tak Ho, M.D. 1,2 Chih-Yi Hsu, M.D. 1,2 Ling-Tan Ting, M.Sc. 1,2 Hung Chiang, M.D. 1,2 1 Department of Pathology and Laboratory Medicine, Veterans General Hospital-Taipei, Taipei, Taiwan, Republic of China. 2 Department of Pathology, National Yang-Ming University School of Medicine, Taiwan, Republic of China. Supported in part by Veterans General Hospital- Taipei grant 90-365. The authors thank Ms. Rinka H. Chen for her technical assistance. Address for reprints: Donald Ming-Tak Ho, M.D., Department of Pathology and Laboratory Medicine, Veterans General Hospital-Taipei, 201 Shih-Pai Road, Sec. 2, Taipei, 11217, Taiwan; Fax: 886-2-2875-7056; E-mail: mtho@vghtpe.gov.tw. Received August 13, 2001; revision received October 17, 2001; accepted October 24, 2001. BACKGROUND. Although various histopathologic features have been associated with aggressive behavior or recurrence of meningiomas, there is little agreement about which features are the most important and in what combination. The objective of this study was to formulate diagnostic criteria for atypical meningioma. METHODS. Eighty-three patients with meningiomas who underwent macroscopic total resection and had been followed until they developed recurrent disease or for at least 10 years were studied. Thirteen histopathologic parameters that related to recurrence of the tumor were evaluated in each patient. All possible combinations of histologic parameters that were related significantly to recurrence were used to formulate scoring models. The model that included the fewest parameters and that could distinguish tumor recurrence best within 10 years was chosen as the final model. RESULTS. The final model included three parameters: loss of architecture, mitoses 1.5/mm 2, and necrosis. Of the 52 tumors with a score 2 (0 or 1 of the 3 parameters), all except 1 tumor did not recur within 10 years, and they were all considered benign meningiomas. Of the 31 tumors with a score 2(2or3ofthe 3 parameters), 94% recurred within 10 years (76% recurred within 5 years), and they were considered atypical meningiomas. The estimated 5-year and 10-year recurrence rates for the benign meningiomas were 0.0% and 1.9%, respectively, for benign meningiomas and 71.0% and 93.5%, respectively, for atypical meningiomas (P 0.001). The estimated 5-year and 10-year mortality rates also were significantly different (0.0% and 0.0% vs. 22.1% and 26.7%, respectively; P 0.001). The MIB-1 labeling index (LI) for the entire group studied ranged from 0.4 to 33.5 (mean LI, 8.4). Fifty-two tumors with an LI of 10 did not recur within 10 years. Of the 31 tumors with an LI 10, 97% recurred (71% within 5 years). CONCLUSIONS. Histopathology and MIB-1 LI were able to predict clinical outcomes of patients with meningioma. The authors propose that atypical meningioma may be diagnosed when two of the following three criteria are present: loss of architecture, mitoses 1.5/mm 2, and necrosis. Cancer 2002;94:1538 47. 2002 American Cancer Society. DOI 10.1002/cncr.10351 KEYWORDS: meningioma, recurrence, atypical, MIB-1, Ki-67. Meningiomas comprise 15 20% of all primary intracranial tumors. 1,2 They are generally benign tumors, and most patients are cured after surgery and remain free of recurrence. However, some tumors behave in a more aggressive fashion, and patients develop local recurrence or metastasis. Histopathologically, meningiomas currently are separated into three grades, i.e., benign (ordinary) meningioma, atypical meningioma, and anaplastic (malignant) menin- 2002 American Cancer Society

Meningioma/Ho et al. 1539 gioma, that intend to reflect the clinical behavior of the tumors. 3 However, there is no consensus on the diagnostic criteria for atypical and anaplastic meningiomas. 3 13 The histologic criteria in some of the diagnostic schemes also are vague and are subject to considerable interobserver interpretation. For example, in the widely used World Health Organization grading system, there are descriptive terms such as having several of the following features, frequent mitoses, high mitotic index, and far in excess. 3 Although studies have shown increasing recurrence rates in patients with benign, atypical, and anaplastic meningiomas, the same data also depicted the inaccuracy of histopathologic diagnosis. 7,14 16 For example, of the studies that involved patients with completely excised tumors, the 5-year recurrence rate for patients with benign meningiomas was 21%, and the rates for patients with atypical and anaplastic meningiomas were 38% and 78%, respectively. 7,16 A 21% recurrence rate in patients with completely excised benign tumors raised the possibility of misclassification of aggressive tumors into the benign category, whereas a 22% nonrecurrence rate for patients with anaplastic tumors illustrated the possibility of overdiagnosis. The objective of this study was to determine histopathologic criteria that could distinguish different prognostic groups of meningiomas. We also studied the MIB-1 labeling index (LI), a marker of proliferative potential, and correlated it with clinical outcome. MATERIALS AND METHODS Eighty-three patients with meningiomas who were seen from 1976 through 1985 at the Veterans General Hospital-Taipei were included in this study. The criteria used for the selection of these patients included macroscopic total resection of the tumor and longterm follow-up. Macroscopic total resection was defined using surgical records and/or postoperative computed tomography (CT) or magnetic resonance imaging (MRI) studies. Patients who did not undergo macroscopic total resection were excluded. Radiologic images other than CT and MRI were not used. Followup was considered adequate when it encompassed the time of recurrence if a tumor recurred or if it was 10 years. Patients with tumors that did not recur within the follow-up period and with follow-up 10 years were excluded from this study. The tumor sites included cerebral convexity (n 45 patients), parasagittal (n 10 patients), sphenoid ridge (n 6 patients), olfactory groove (n 6 patients), cerebellopontine angle (n 5 patients), supersella (n 2 patients), tentorium cerebelli (n 2 patients), lateral ventricle (n 2 patients), and spinal cord (n 5 patients). The original diagnosis of these patients included benign meningiomas of varies types, i.e., transitional (n 31 patients), fibrous (n 11 patients), meningotheliomatous (n 8 patients), microcystic (n 4 patients), angiomatous (n 2 patients), psammomatous (n 2 patients), and secretory (n 2 patients) tumor types as well as atypical meningiomas (n 9 patients) and anaplastic meningiomas (n 14 patients). Only original biopsy specimens were used in this study. All specimens were recut from the paraffin embedded tissue blocks for hematoxylin and eosin staining for morphologic analysis and immunohistochemical staining for MIB-1 (monoclonal, 1:75; Immunotech, Marseille, France; microwaved 3 times for 5 minutes each). Positive and negative controls were included with each batch of the sections to confirm the consistency of the analysis. Sections from one glioblastoma that had a known average MIB-1 LI of 41.3 were used as positive controls (for counting and reporting of MIB-1 LI, see below). The batch of MIB-1 staining was considered acceptable when the LI of the control section was within the range of 37.2 45.4 (standard deviation [SD] was 2.0 from the mean). Negative controls were accomplished by omitting the primary antibody. Thirteen histologic parameters, including loss of architecture (sheeting), hypercellularity, nuclear pleomorphism, vesicular nuclei, distinct nucleoli (small or prominent), mitoses, small cells, papillary formation, endothelial proliferation, necrosis, brain invasion, invasion of structures other than brain, and MIB-1 LI, were evaluated in each tumor. Except for mitoses and MIB-1 LI, all parameters were recorded as either absent or present. The mitotic count was the highest count of 10 consecutive high-power fields (hpf; 1 hpf 0.196 mm 2 ) and was expressed per 10 hpf, per mm 2 and per 10,000 cells. For MIB-1 staining, distinct nuclear staining was recorded as positive. The MIB-1 LI was defined as the percentage of immunostained cells divided by the total number of cells in the evaluated area. All counts were performed at a magnification of 400 using an ocular graticule consisting of 10 10 100 fields covering an area of 0.0625 mm 2. Ten viable fields from the area of maximal labeling were chosen for counting. From each specimen, 1000 5000 cells were counted, depending on the tumor cellularity. Vascular components and hematogenous cells were excluded from the analysis. All of the histopathologic features described above were reviewed by two of the authors (D.M.-T.H. and C.-Y.H.) who were unaware of the patients outcomes. When there were discrepancies for parameters that were recorded as absent or present, the sections were examined simultaneously by both observers, and the results were recorded after agreement was reached.

1540 CANCER March 1, 2002 / Volume 94 / Number 5 For mitotic counting, a discrepancy of 1/10 hpf was accepted, and the higher count was recorded. For counts that had greater discrepancies, more than 1 set of 10 hpf was recounted simultaneously by both observers, and the highest count was recorded. For the MIB-1 LI, a discrepancy of 10% was accepted, and the higher LI was recorded. For LI values that had a greater discrepancy, 10 fields were counted simultaneously by both observers, and the LI was recorded. The patients medical records were reviewed in detail. Clinical data were recorded, including dates of birth and death, gender, tumor sites, date of surgery for primary and recurrent tumor resections, extent of resection, and use of adjuvant therapy. Evidence of dural, venous sinus, bone, soft tissue, or brain invasion was obtained from the surgical notes and from CT and/or MRI reports. The date of recurrence was defined radiologically or was based on the date of recurrent tumor resection. Recurrence free survival (RFS) was defined as the time from surgery to evidence of recurrence. Overall survival (OS) was calculated as the time from surgery to the date of death or last follow-up for surviving patients. The mean and median of mitoses rate and MIB-1 LI were used as reference values to set cut-off points. The value separating the two subgroups of patients with the most significant difference in RFS was the cut-off point. Univariate analysis of the histologic parameters relating to tumor recurrence was performed. The RFS and OS curves were plotted using the Kaplan Meier method. Their differences were tested using the logrank test. All possible combinations of the histopathologic parameters that were significantly related to recurrence, except for MIB-1 LI, were used to formulate scoring models. We assigned one score for each histopathologic parameter. The sum of scores was the histopathologic score of the tumor. The score that could best differentiate tumors that recurred within 10 years and tumors that recurred after 10 years or not at all was the cut-off score in this model. The model that was comprised of the fewest parameters and that could predict tumor recurrence best within 10 years was the model selected. Because the diagnostic scheme set by Perry et al. primarily used a single criterion, 12 which is in contrast to other schemes that generally used multiple criteria or at least two parameters, 3 11,13 we also applied the model by Perry et al. to our study materials to evaluate the effectiveness of a single criterion. According to Perry et al., atypical meningiomas were diagnosed when any one of the following three criteria was identified: 1) high mitotic index (e.g., 4 mitoses per 10 hpf or 2.5/mm 2 ); 2) the presence of at least 3 of the following 4 features: sheeting architecture, hypercellularity, macronucleoli, and small cell formation; and 3) brain invasion. Anaplastic meningiomas were diagnosed when there was evidence of either of the following criteria: 1) excessive mitotic activity (e.g., 20 mitoses per 10 hpf or 12.5/mm 2 ) or 2) focal or diffuse loss of meningothelial differentiation at the light microscopic level resulting in sarcoma, carcinoma, or melanoma-like appearance. 12 RESULTS Among the 83 patients studied, the mean age was 50.6 years (range, 14 73 years), and the male:female gender ratio was 1.0:0.8. The median follow-up was 158.5 months (range, 17.5 297.4 months). Thirty-nine patients (47%) had recurrent tumors, and the latent time to recurrence ranged from 8.3 months to 269.6 months (median, 43.6 months). Eight patients (9.6%) died, and their survival ranged from 17.5 months to 227.1 months (median, 42.1 months). The initial mean rate of mitoses counted by the 2 observers was 3.4/10 hpf (range, 0 19/10 hpf) and 3.3/10 hpf (0 17/10 hpf), respectively. The mean ( SD) difference in the mitotic count between the two observers was 0.1/10 hpf ( 0.9/10 hpf). The initial mean MIB-1 LI for the two observers was 8.5 (range, 0.3 36) and 8.3 (range, 0.4 31.2), respectively. The mean ( SD) difference in the MIB-1 LI between the two observers was 0.2 ( 1.3). Neither mitotic counting nor MIB-1 LI counting was significantly different between the two observers. The mitotic count for 75 patients (90.4%) and the MIB-1 LI for 70 patients (84.3%) between the two observers were within the acceptable ranges. After recounting for patients with levels that were not within the acceptable ranges, the mean and median mitotic counts were 3.7/10 hpf and 2.0/10 hpf, respectively (range, 0.0 19.0/10 hpf), and the cut-off point for mitoses was 3.0/10 hpf (1.5/mm 2 ). The mean and median MIB-1 LI values were 8.4 and 5.9, respectively (range, 0.4 33.5), and the cut-off point for the MIB-1 LI was 10.0. Table 1 shows the univariate analysis of the relation between histopathologic parameters and tumor recurrence. Of the 13 histopathologic parameters evaluated, 10 were related significantly with tumor recurrence. The parameters included loss of architecture, hypercellularity, distinct nucleoli, mitotic count 3/10 hpf (1.5/mm 2 or 4/10,000 cells), small cells, papillary formation, endothelial proliferation, necrosis, brain invasion, and MIB-1 LI 10. The scoring models were created using all of the possible combinations of the 10 parameters listed above. Of all possible scoring models, the model with loss of architecture, mitoses 3/10 hpf (1.5/mm 2 ), and necrosis distinguished most effectively tumors

Meningioma/Ho et al. 1541 TABLE 1 The Relation between Histopathologic Parameters and Recurrence Recurrence Parameter No. < 5 yrs 5 10 yrs > 10 yrs None a 10-yr RFS (%) Log-rank statistic P value b Loss of architecture Absent 60 6 3 8 43 85.0 68.1 0.001 c Present 23 16 5 1 1 8.7 Hypercellularity Absent 54 3 2 8 41 90.7 61.0 0.001 c Present 29 19 6 1 3 13.8 Nuclear pleomorphism Absent 55 14 3 7 31 69.1 0.4 0.533 Present 28 8 5 2 13 53.6 Vesicular nuclei Absent 14 4 1 3 6 64.3 0.7 0.394 Present 69 18 7 6 38 63.8 Nucleoli Inconspicuous 24 2 0 5 17 91.7 4.5 0.034 c Distinct 59 20 8 4 27 52.5 Mitoses per 10 hpf (1.96 mm 2 ) 3 d 46 1 0 7 38 97.8 59.2 0.001 c 3 37 21 8 2 6 21.6 5 e 62 7 3 9 43 83.9 58.3 0.001 c 5 21 15 5 0 1 4.8 Mitoses per 10,000 cells 4 49 2 0 8 39 95.9 54.6 0.001 c 4 34 20 8 1 5 17.7 Small cells Absent 76 16 7 9 44 69.7 35.3 0.001 c Present 7 6 1 0 0 0.0 Papillary formation Absent 82 21 8 9 44 64.6 7.0 0.008 c Present 1 1 0 0 0 0.0 Endothelial proliferation Absent 82 21 8 9 44 64.6 39.8 0.001 c Present 1 1 0 0 0 0.0 Necrosis Absent 51 2 2 7 40 92.2 51.6 0.001 c Present 32 20 6 2 4 18.8 Brain invasion Absent 72 15 5 9 43 72.2 18.3 0.001 c Present 11 7 3 0 1 9.1 Invasion of structures other than brain Absent 65 15 6 7 37 67.7 2.4 0.125 Present 18 7 2 2 7 50.0 MIB-1 LI 10 52 0 0 9 43 100.0 104.8 0.001 c 10 31 22 8 0 1 3.2 RFS: recurrence free survival; hpf: high-power fields; LI: labeling index a Follow-up 10 years. b Log-rank test. c Statistically significant. d Three mitoses per 10 hpf 1.5 mitoses/mm 2. e Five mitoses per 10 hpf 2.5 mitoses/mm 2. that recurred within 10 years from tumors that did not recur within 10 years (Table 2, Model 1). Using this scoring model, with 1 exception, all patients who had tumors with scores of 0 or 1 (0 or 1 of the 3 parameters) did not have recurrent tumors or had recurrent tumors only after 10 years. Of the patients who had tumors with scores of 2 or 3 (2 or 3 of the 3 parameters), all but 2 patients had recurrent tumors within 10 years (76% had recurrent tumors within 5 years). The RFS and OS curves of these two groups of patients are

1542 CANCER March 1, 2002 / Volume 94 / Number 5 TABLE 2 Sample of Several Scoring Models and their Association with Recurrence Score No. Recurrence (%) < 5 yrs 5 10 yrs > 10 yrs None a Model 1: Loss of architecture, mitoses 3 per 10 hpf, b and necrosis c 0 of 3 39 0 0 5 (12.8) 34 (87.2) 1 of 3 13 0 1 (7.7) d 3 (23.1) 9 (69.2) 2 of 3 14 9 (64.3) 3 (21.4) 1 (7.1) d 1 (7.1) d 3 of 3 17 13 (76.5) 4 (23.5) 0 0 Model 2. Loss of architecture, mitoses 3 per 10 hpf, b and brain invasion c 0 of 3 42 0 0 6 (14.3) 36 (85.7) 1 of 3 18 5 (27.8) d 2 (11.1) d 3 (16.7) 8 (44.4) 2 of 3 16 12 (75) 4 (25) 0 0 3 of 3 7 5 (71.4) 2 (28.6) 0 0 Model 3: Loss of architecture, mitoses 3 per 10 hpf, b necrosis, and small cells c 0 of 4 39 0 0 5 (12.8) 34 (87.2) 1 of 4 13 0 1 (7.7) d 3 (23.1) 9 (69.2) 2 of 4 11 7 (63.6) 2 (18.2) 1 (9.1) d 1 (9.1) d 3 of 4 16 11 (68.8) 5 (31.3) 0 0 4 of 4 4 4 (100) 0 0 0 Model 4. Loss of architecture, mitoses 3 per 10 hpf, b necrosis, and brain invasion c 0 of 4 39 0 0 5 (12.8) 34 (87.2) 1 of 4 12 0 1 (8.3) d 3 (25) 8 (66.7) 2 of 4 11 6 (54.5) 2 (18.2) 1 (9.1) d 2 (18.2) d 3 of 4 15 12 (80) 3 (20) 0 0 4 of 4 6 4 (66.7) 2 (33.3) 0 0 Model 5. Loss of architecture, mitoses 3 per 10 hpf, b necrosis, and hypercellularity c 0 of 4 39 0 0 5 (12.8) 34 (87.2) 1 of 4 10 0 1 (10) d 2 (20) 7 (70) 2 of 4 7 3 (42.9) 0 2 (28.6) d 2 (28.6) d 3 of 4 11 6 (54.5) 4 (36.4) 0 1 (9.1) d 4 of 4 16 13 (81.3) 3 (18.8) 0 0 a Follow-up 10 years. b Three mitoses per 10 hpf 1.5/mm 2. c One parameter was counted as one score. d Tumors that could not be discriminated by the model (see Results). shown in Figures 1 and 2. Patients with tumor scores of 0 or 1 could be diagnosed with benign meningioma, whereas patients with tumors scores of 2 or 3 could diagnosed with atypical meningioma. Although three (3.6%) tumors (Table 2, Model 1, footnote d) could not be discriminated using this model, none of the other scoring models tested surpassed this model in terms of the number of parameters or the ability to discriminate tumor recurrence. Several scoring models are shown in Table 2 to illustrate the latter statement: Model 2, which included loss of architecture, mitoses 3/10 hpf, and brain invasion, showed that 7 patients with tumor scores of 1 had recurrent tumors within 10 years. Model 3, which included all three parameters from Model 1 plus small cells (for a total possible score of 4), showed the same discrimination effect as Model 1. Model 4 included the four parameters from Model 3 with the substitution of small cells for brain invasion, and four tumors could not be discriminated. Model 5 consisted of the parameters from Model 1 plus hypercellularity, and this model missed in six tumors that could not be discriminated. When we applied the criteria set by Perry et al., 12 58 tumors in this study were classified as benign, and 25 tumors were classified as atypical. None of the patients had meningiomas with features that fulfilled the diagnostic criteria for anaplastic meningioma. Table 3 shows the comparison of the model by Perry et al. and our model in terms of correlation between diagnosis (grade) and tumor recurrence. The recurrence rate was significantly different between patients with benign meningiomas and patients with atypical meningiomas who were diagnosed using either model (P 0.001). However, by using the model by Perry et al., 7 of 58 patients (12%) who were diagnosed with benign meningioma had a recurrence within 10 years, whereas 2 of 25 patients (8%) who were diagnosed with atypical meningioma did not have a recurrence. When using our model, only 1 of 52 patients (2%) with benign meningiomas had a recurrence within 10

Meningioma/Ho et al. 1543 FIGURE 1. Recurrence free survival curves for patients with a histopathologic score 2 and patients with a histopathologic score 2. years, and 2 of 31 patients (6.5%) with atypical meningiomas did not have a recurrence or developed a recurrence after 10 years. Table 4 shows the patterns of diagnostic criteria in patients with atypical meningiomas who were diagnosed using the model by Perry et al. 12 and the correlation with tumor recurrence. Of the 6 patterns identified, 3 patterns had 1 criterion (100, 001, and 010; 16 tumors), 2 patterns had 2 criteria (101 and 110; 5 tumors), and 1 pattern had 3 criteria (111; 4 tumors). For patients with tumors that had 2 or 3 criteria, all had recurrent tumors within 10 years, whereas 2 patients with tumors that met only 1 criterion did not have recurrent tumors. The findings described above showed that using multiple criteria was better than using only one criterion for the diagnosis of atypical meningioma. We propose using the criteria we identified (Table 2, Model 1) for the diagnosis. Table 5 summarizes our proposal for the diagnostic criteria for atypical meningioma. When using this diagnostic model for our study material, 52 patients had benign meningiomas (score 2), and 31 patients had atypical meningiomas (score 2). The mean ages of these two groups of patients, 50.3 years (range, 14 73 years) and 51.1 years (range, 23 73 years), respectively, did not differ significantly. The gender ratio (male:female) was 1.0:1.4 for patients with benign meningiomas and 3.4:1.0 for patients with atypical meningiomas. The estimated 5-year and 10- year recurrence rates were 0.0% and 1.9%, respectively, for patients with benign meningiomas and 71.0% and 93.5%, respectively, for patients with atypical meningiomas (P 0.001). The estimated 5-year and 10-year mortality rates were both 0.0% for patients with benign tumors and 22.1% and 26.7%, respectively, for patients with atypical tumors (P 0.001). The MIB-1 LI for the entire group studied range from 0.4 to 33.5 (mean SD, 8.4 6.8). The 52 patients who had tumors with an LI 10 did not have recurrent tumors or developed recurrent tumors after 10 years. Of 31 patients who had tumors with an LI 10, all except 1 patient developed a recurrent tumor within 10 years (71% developed a recurrent tumor within 5 years) (Table 1). Figures 3 and 4 show the RFS and OS curves for patients with LI 10 and LI 10. FIGURE 2. Overall survival curves for patients with a histopathologic score 2 and patients with a histopathologic score 2. DISCUSSION Although various histopathologic features have been associated with aggressive behavior or recurrence of meningiomas, there is little agreement about which features are the most important and in what combination or degree. These features have generally included loss of architecture (sheeting), hypercellularity, nuclear pleomorphism, prominent nucleoli, high nuclear-cytoplasmic ratios (small cell changes), increased mitotic activity, hypervascularity, necrosis, invasion of adjacent structures, and brain invasion. 6,8,9,17 19 The lack of consensus regarding the features associated with aggressive tumor behavior has been responsible

1544 CANCER March 1, 2002 / Volume 94 / Number 5 TABLE 3 Grading of the Current Study Materials Using the Model by Perry et al. and Model 1 from the Current Study Recurrence (%) Study No. < 5 yrs 5 10 yrs > 10 yrs None a P value Perry et al. 12 Benign 58 4 (6.9) 3 (5.2) 9 (15.5) 42 (72.4) 0.001 Atypical 25 18 (72) 5 (20) 0 2 (8) Current study (Model 1) Benign b 52 0 1 (1.9) 8 (15.4) 43 (82.7) 0.001 Atypical c 31 22 (71) 7 (22.6) 1 (3.2) 1 (3.2) a Follow-up time 10 years. b Scores of 0 or 1. c Scores of 2 or 3. TABLE 4 The Patterns of Diagnostic Criteria in Atypical Meningiomas Diagnosed Using the Model by Perry et al. and the Correlation with Recurrence Recurrence (%) Diagnostic criteria a No. < 5 yrs 5 10 yrs > 10 yrs None b 100 12 9 (75) 2 (16.7) 0 1 (8.3) 001 3 2 (66.7) 0 0 1 (33.3) 010 1 1 (100) 0 0 0 101 4 1 (25) 3 (75) 0 0 110 1 1 (100) 0 0 0 111 4 4 (100) 0 0 0 a High mitotic index; the presence of at least three of the following four features: sheeting architecture, hypercellularity, macronucleoli, and small cell formation; and brain invasion (1, present; 0, absent). b Follow-up 10 years. TABLE 5 Proposed Diagnostic Criteria for Atypical Meningiomas Presence of any two of the following three criteria Loss of architecture Mitoses 3/10 hpf (1.5/mm 2 ) Necrosis hpf: high-power fields. for the development of several grading systems for meningiomas. 3 13 In our study of the identification of histopathologic criteria to distinguish between benign and aggressive meningiomas, we only studied patients with meningiomas who underwent macroscopic total resection and were followed until recurrence or for at least 10 years if they did not have a recurrent tumor. This was meant to eliminate the problems inherent in the study of recurrent meningiomas, because tumor recurrence is dependent not only on the proliferative potential of the tumors but also on the completeness FIGURE 3. Recurrence free survival curves for patients with an MIB-1 labeling index (LI) 10 and an MIB-1 LI 10.

Meningioma/Ho et al. 1545 FIGURE 4. Overall survival curves for patients with an MIB-1 labeling index (LI) 10 and an MIB-1 LI 10. of original resection. A minimum of 10-year follow-up of the patients with nonrecurrent tumors assured the true benign nature of these tumors. Our study materials included tumors that were nonrecurrent within 10 years and tumors that recurred within 10 years, categories that were correlated with the histopathologic diagnosis of benign and atypical meningiomas, respectively. There were no anaplastic (malignant) tumors, which we defined as tumors either with metastasis or that fulfilled the diagnostic criteria set by Perry et al., 12 in this study. The rare occurrence of anaplastic meningiomas was compatible with that reported in the series by Perry et al., consisting of only approximately 1% of meningiomas. The histopathologic model that distinguished nonrecurrent (benign) and recurrent (atypical) meningiomas most effectively in our study material was comprised of three parameters, including loss of architecture, mitosis 1.5/mm 2, and necrosis. Benign meningiomas generally had none of the parameters or, at the most, one parameter; whereas atypical meningiomas usually had two or three parameters. Tumors with loss of normally recognized architecture (patterns) of benign meningioma or sheeting were recognized easily. When we studied mitotic activity, we expressed the mitotic count per 10 hpf, per mm 2, and per 10,000 cells. The exact cell count was used in the denominator to express mitotic activity in an attempt to avoid factors that may have affected mitotic counts per 10 hpf or mm 2 due to variation of cellularity in different tumors and the presence of acellular or poorly cellular elements (e.g., fibrosis and necrosis) in the areas of evaluation. In this study, there was good correlation between mitoses per 10,000 cells and mitoses per 10 hpf or mm 2 (data not shown). For practical purposes and convenience, we chose mitoses per 10 hpf or mm 2 to express mitotic activity. Of the necroses seen in meningiomas, there were micronecrosis and large, infarct-like necrosis. Although the latter presumably resulted from ischemia or a degenerative process, both forms of necrosis were related to aggressiveness of the meningiomas. When determining whether a single criterion or multiple criteria were superior for the diagnoses of atypical meningiomas, our observations showed that multiple criteria were superior. To illustrate this point, we used both the comparative study of grading our materials using the model by Perry et al. 12 and our model (Table 3) as well as the study analyzing the diagnostic criteria of the model by Perry et al. for the diagnosis of atypical meningiomas (Table 4). Furthermore, as discussed below, mitoses and brain invasion, which were used as single diagnostic criterion in the model by Perry et al., could not be used as absolute criteria. With regard to the cut-off point chosen for mitotic counts, there had been different values selected in previous studies, e.g., 2/10 hpf, 1 3/10 hpf, 5,12 4/10 hpf, 9 or 6/10 hpf. 20 However, no value by itself fully distinguished benign meningiomas from aggressive meningiomas. Using a lower cut-off point led to having more clinically benign tumors in the high-grade category, whereas using a higher cut-off point led to having more clinically aggressive tumors in the low-grade category. In the current study, when we chose 3/10 hpf (1.5/mm 2 ) as the cut-off point, 1 of 46 tumors (2.2%) with mitoses 3/10 hpf recurred within 10 years, whereas 8 of 37 tumors (21.6%) with mitoses 3/10 hpf did not recur or recurred after 10 years. When we used 5/10 hpf (2.5/mm 2 ) as the cut-off point, which was used in the model by Perry et al., 12 10 of 62 tumors (16.1%) with mitoses 5/10 hpf recurred within 10 years, whereas 1 of 21 tumors (4.8%) with mitoses 5/10 hpf did not recur (Table 1, Mitotses per 10 hpf). In our diagnostic model, we selected the lower cut-off point (1.5/mm 2 ) to assure that the histologically diagnosed benign tumors were truly clinically benign. For the diagnosis of atypical tumor, we needed at least one additional parameter for the diagnosis. Conversely, when using a higher cut-off point (e.g., 2.5/mm 2 ) and using it as the sole criterion, a fair number of aggressive tumors were categorized into the benign category that may have been missed using adjuvant therapy. Although brain invasion had been considered a

1546 CANCER March 1, 2002 / Volume 94 / Number 5 sign of malignancy and has been used as a diagnostic criterion for malignant meningioma, 10,21 23 its significance has been questioned, because it has little prognostic significance by itself. 2,12,17 It was reported that up to 4% of benign meningiomas showed brain invasion. 1 In the current study, although brain invasion was significant prognostically in univariate analysis, it was not chosen as a diagnostic criterion in the final model (Model 1). Our study showed that it was neither a sensitive marker nor an absolutely specific marker, because it was present in only 10 of 30 patients (33%) with tumors that recurred within 10 years and in 1 patient with a nonrecurrent tumor. The MIB-1 monoclonal antibody recognizes a nonhistone, nuclear protein (Ki-67 protein) that is expressed during the G1, S, G2, and M phases of the cell cycles but it is absent in the G0 phase and in the early part of the G1 phase. It can be applied to routinely embedded paraffin sections and has been employed as an operational marker of cell proliferation in various types of tumors. 24,25 Many researchers have studied the MIB-1 LI in patients with meningiomas. 13,15,18,20,26 37 Most of those studies showed that the MIB-1 LI was related significantly to the clinical outcome of the patients. However, the mean, range, 15,20,26 29,31 34,36,37 and cut-off values identified in different studies varied. The reported mean LI for benign meningiomas ranged from 0.7 to 2.2; for atypical meningiomas, it ranged from 2.1 to 9.3; and, for anaplastic meningiomas, it ranged from 11.0 to 16.3. 15,33 35 For the studies of patients with nonrecurrent and recurrent meningiomas, the mean LI of nonrecurrent meningiomas ranged from 0.98 to 3.8, and that of recurrent tumors ranged from 4.0 to 8.8. 15,26,27,33 The cut-off points identified in previous studies were 3.0, 20,27 3.2, 35 and 4.2. 32 There were also some studies in which the results did not show a correlation between the MIB-1 LI and tumor behavior. 13,18,30 The current study showed that the MIB-1 LI was a powerful prognostic marker and the cut-off point of 10 discriminated tumors in patients with different prognoses. However, because of the variation in the MIB-1 LI between different laboratories, presumably due to differences in staining and counting methodologies, the cut-off point identified in one laboratory may not be applicable to other institutions. Nevertheless, it is a powerful tool for those researchers who have experience with the technique and who can verify the validity of the cut-off point. In addition to histopathologic factors, other clinical factors have been associated with prognosis, including patient age, gender, preoperative condition of patients, location and size of tumors, type of surgical resection, hormonal receptors, cytogenetic patterns, and postoperative radiotherapy. 1,19,38 In conclusion, the histopathologic study of a group of patients with meningiomas who underwent macroscopic total resection and had long-term follow-up showed that the presence of two of three criteria (i.e., loss of architecture, mitoses 1.5/mm 2, and necrosis) was correlated with patients with atypical meningiomas who had shorter RFS and OS compared with patients with benign meningiomas. It also was shown that the MIB-1 LI was a significant prognostic marker for meningiomas. REFERENCES 1. Ayerbe J, Lobato RD, de la Cruz J, Alday R, Rivas JJ, Gómez PA, et al. Risk factors predicting recurrence in patients operated on for intracranial meningioma. A multivariate analysis. Acta Neurochir (Wien) 1999;141:921 32. 2. Coke CC, Corn BW, Werner-Wasik M, Xie Y, Curran WJ Jr. Atypical and malignant meningiomas: an outcome report of seventeen cases. J Neurooncol 1998;39:65 70. 3. Kleihues P, Burger PC, Scheithauer BW. International histological classification of tumours. Histological typing of tumours of the central nervous system, 2nd ed. Geneva: World Health Organization, 1993. 4. Zülch KJ. International histological classification of tumours. Histological typing of tumours of the central nervous system. No. 21. Geneva: World Health Organization, 1979. 5. Miller DC, Ojemann RG, Proppe KH, McGinnis BD, Grillo HC. Benign metastasizing meningioma. J Neurosurg 1985; 62:763 6. 6. Jääskeläinen J, Haltia M, Laasonen E, Washlström T, Valtonen S. The growth rate of intracranial meningiomas and its relation to histology. An analysis of 43 patients. Surg Neurol 1985;24:165 72. 7. Jääskeläinen J, Haltia M, Servo A. Atypical and anaplastic meningiomas: radiology, surgery, radiotherapy, and outcome. Surg Neurol 1986;25:233 42. 8. de la Monte SM, Flickinger J, Linggood RM. Histopathologic features predicting recurrence of meningiomas following subtotal resection. Am J Surg Pathol 1986;10:836 43. 9. Maier H, Öfner D, Hittmair A, Kitz K, Budka H. Classic, atypical, and anaplastic meningioma: three histopathological subtypes of clinical relevance. J Neurosurg 1992;77:616 23. 10. Burger PC, Scheithauer BW. Tumors of the central nervous system. In: Rosai J, editor. Atlas of tumor pathology, 3rd series, fascicle 10. Washington, DC: Armed Forces Institute of Pathology, 1994:259 86. 11. Miller DC. Predicting recurrence of intracranial meningiomas. A multivariate clinicopathologic model: interim report of the New York University Medical Center meningioma project. Neurosurg Clin North Am 1994;5:193 200. 12. Perry A, Scheithauer BW, Stafford SL, Lohse CM, Wollan PC. Malignancy in meningiomas. A clinicopathologic study of 116 patients, with grading implications. Cancer 1999;85: 2046 56. 13. Abramovich CM, Prayson RA. Histopathologic features and MIB-1 labeling indices in recurrent and nonrecurrent meningiomas. Arch Pathol Lab Med 1999;123:793 800.

Meningioma/Ho et al. 1547 14. Mirimanoff RO, Dosoretz DE, Linggood RM, Ojemann RG, Martuza RL. Meningioma: analysis of recurrence and progression following neurosurgical resection. J Neurosurg 1985;62:18 24. 15. Kolles H, Niedermayer I, Schmitt C, Henn W, Feld R, Steudel WI, et al. Triple approach for diagnosis and grading of meningiomas: histology, morphometry of Ki-67/Feulgen stainings, and cytogenetics. Acta Neurochir (Wien) 1995;137: 174 81. 16. McCarthy BJ, Davis FG, Freels S, Surawicz TS, Damek DM, Grutsch J, et al. Factors associated with survival in patients with meningioma. J Neurosurg 1998;88:831 9. 17. McLean CA, Jolley D, Cukier E, Giles G, Gonzales MF. Atypical and malignant meningiomas: importance of micronecrosis as a prognostic indicator. Histopathology 1993;23: 349 53. 18. Prayson RA. Malignant meningioma: a clinicopathologic study of 23 patients including MIB1 and p53 immunohistochemistry. Am J Clin Pathol 1996;105:719 26. 19. Perry A, Stafford SL, Scheithauer BW, Suman VJ, Lohse CM. Meningioma grading. An analysis of histologic parameters. Am J Surg Pathol 1997;21:1455 65. 20. Hsu DW, Efird JT, Hedley-Whyte ET. MIB-1 (Ki-67) index and transforming growth factor-alpha (TGF alpha) immunoreactivity are significant prognostic predictors for meningiomas. Neuropathol Appl Neurobiol 1998;24:441 52. 21. Cushing H, Eisenhardt L. Meningiomas. Their classification, regional behaviour, life history, and surgical end results. Springfield: Charles C. Thomas Publishers, 1938. 22. Russell DS, Rubinstein LJ. Pathology of tumours of the nervous system. 5th ed. Baltimore: Williams & Wilkins, 1989: 452 506. 23. Burger PC, Scheithauer BW, Vogel FS. Surgical pathology of the nervous system and its coverings. 3rd ed. New York: Churchill Livingstone, 1991:67 91. 24. Gerdes J, Schwab U, Lemke H, Stein H. Production of a mouse monoclonal antibody reactive with a human nuclear antigen associated with cell proliferation. Int J Cancer 1983; 31:13 20. 25. Cattoretti G, Becker MH, Key G, Duchrow M, Schluter C, Galle J, et al. Monoclonal antibodies against recombinant parts of the Ki-67 antigen (MIB 1 and MIB 3) detect proliferating cells in microwave-processed formalin-fixed paraffin sections. J Pathol 1992;168:357 63. 26. Ohta M, Iwaki T, Kitamoto T, Takeshita I, Tateishi J, Fukui M. MIB1 staining index and scoring of histologic features in meningioma. Indicators for the prediction of biologic potential and postoperative management. Cancer 1994;74: 3176 89. 27. Matsuno A, Fujimaki T, Sasaki T, Nagashima T, Ide T, Asai A, et al. Clinical and histopathological analysis of proliferative potentials of recurrent and non-recurrent meningiomas. Acta Neuropathol 1996;91:504 10. 28. Matsuno A, Nagashima T, Matsuura R, Tanaka H, Hirakawa M, Murakami M, et al. Correlation between MIB-1 staining index and the immunoreactivity of p53 protein in recurrent and nonrecurrent meningiomas. Am J Clin Pathol 1996;106: 776 81. 29. Langford LA, Cooksley CS, DeMonte F. Comparison of MIB-1 (Ki-67) antigen and bromodeoxyuridine proliferation indices in meningiomas. Hum Pathol 1996;27:350 4. 30. Møller M, Brændstrup O. No prediction of recurrence of meningiomas by PCNA and Ki-67 immunohistochemistry. J Neurooncol 1997;34:241 6. 31. Madsen C, Schrøder HD. Ki-67 immunoreactivity in meningiomas determination of the proliferative potential of meningiomas using the monoclonal antibody Ki-67. Clin Neuropathol 1997;16:137 42. 32. Perry A, Stafford SL, Scheithauer BW, Suman VJ, Lohse CM. The prognostic significance of MIB-1, p53, and DNA flow cytometry in completely resected primary meningiomas. Cancer 1998;82:2262 9. 33. Abramovich CM, Prayson RA. MIB-1 labeling indices in benign, aggressive, and malignant meningiomas: a study of 90 tumors. Hum Pathol 1998;29:1420 7. 34. Karamitopoulou E, Perentes E, Tolnay M, Probst A. Prognostic significance of MIB-1, p53, and bcl-2 immunoreactivity in meningiomas. Hum Pathol 1998;29:140 5. 35. Nakasu S, Li DH, Okabe H, Nakajima M, Matsuda M. Significance of MIB-1 staining indices in meningiomas. Comparison of two counting methods. Am J Surg Pathol 2001; 25:472 8. 36. Nakaguchi H, Fujimaki T, Matsuno A, Matsuura R, Asai A, Suzuki I, et al. Postoperative residual tumor growth of meningioma can be predicted by MIB-1 immunohistochemistry. Cancer 1999;85:2249 54. 37. Nagashima G, Aoyagi M, Yamamoto M, Yamamoto S, Wakimoto H, Ohno K, et al. P53 overexpression and proliferative potential in malignant meningiomas. Acta Neurochir (Wien) 1999;141:53 61. 38. Chan RC, Thompson GB. Morbidity, mortality, and quality of life following surgery for intracranial meningiomas. A retrospective study in 257 cases. J Neurosurg 1984;60:52 60.