Comparison of Proliferation Indices in Glioblastoma Multiforme by Whole Tissue Section vs Tissue Microarray

Anatomic Pathology / WHOLE TISSUE SECTION VS MICROARRAY Comparison of Proliferation Indices in Glioblastoma Multiforme by Whole Tissue Section vs Tissue Microarray Andres G. Chiesa-Vottero, MD, 1 Lisa A. Rybicki, MS, 2 and Richard A. Prayson, MD 1 Key Words: Tissue microarray; Proliferation indices; Ki-67; Glioblastoma multiforme; Cell proliferation Abstract Tissue microarrays (TMAs) reduce the amount of tissue analyzed with the assumption that protein and gene expression patterns are homogeneous throughout tumors. Many tumor types, including glioblastoma multiforme (GBM), are heterogeneous in many regards, including cell proliferation. We retrospectively compared Ki-67 labeling indices (LIs) determined by whole tissue section (WTS) vs TMA in a series of 50 GBMs from 45 patients. A paired t test indicated that the difference between average LIs obtained from a TMA vs a WTS was not significant (P =.51). There was no correlation between TMA and WTS (r = 0.042; P =.77), indicating that the methods yielded very different results in individual tumors. The Ki-67 LI did not always correlate with the tissue section in an individual tumor; however, when evaluating a large number of tumors on a TMA, the LI range and mean LI were roughly comparable with the LI range and mean LI determined from the WTS. In an attempt to increase throughput expression profiling of tumor samples, tissue microarray (TMA) can be used. It is a method of harvesting small cores of tissue and placing them in an array on a recipient block, such that hundreds of cases can be easily analyzed simultaneously. 1 The reflex criticism of this technique is that the small amount of tissue analyzed may not be representative of the tumor, particularly in heterogeneous tumors. Constituting approximately 25% of all intracranial tumors, glioblastoma multiforme (GBM) is the most common primary tumor of the central nervous system. This highly malignant neoplasm grows quickly, and, without therapy, 95% of patients die within 3 months of diagnosis. 2 Among neuroepithelial tumors, it generally is difficult to identify prognostic factors among high-grade tumors, ie, factors that predict why a small subset of patients with GBM survive longer. In GBM, besides age, Karnofsky performance status, and location, no other prognostic factors have been reliably identified that influence survival. Sex, tumor size, preoperative conditions, specific neurologic findings, extent of tumor resection, diverse histologic features of the resected tumor, response to radiotherapy, reoperation, and postoperative adjuvant treatment each have been regarded as significant by some authors and relatively unimportant by others. 3-13 The ability to prognosticate for patients with gliomas has been a primary goal in neuro-oncology for decades. 14 With this purpose, assessment of proliferative activity in astrocytic tumors has been the subject of considerable research interest during the last decade. While some studies have noted an inverse relationship between cell proliferation and patient survival, others have failed to observe such a relationship. 3,15-18 Recently, it has been suggested that the individual growth 902 Am J Clin Pathol 2003;120:902-908 Downloaded 902 from https://academic.oup.com/ajcp/article-abstract/120/6/902/1759043

Anatomic Pathology / ORIGINAL ARTICLE potential of a GBM results in regrowth after a case-specific time, which is shorter in tumors with a high cell proliferation index and longer in tumors with a low index. 19 This was demonstrated through a strong inverse correlation between time to relapse and the labeling index (LI). The purpose of the present study was to evaluate how representative TMAs are in comparison with whole tissue section (WTS) for the study of proliferation indices in 45 cases of GBM, a tumor well known for its heterogeneity. Materials and Methods We retrospectively compared Ki-67 LIs determined by WTS vs TMA in a series of 50 GBMs from 45 patients. Tumors were graded according to the most recent criteria of the World Health Organization classification of tumors of the nervous system 3 Image 1. The TMAs were built using the Tissue Arrayer (Beecher Instruments, Silver Spring, MD). Briefly, a fresh H&Estained slide was obtained from each block and used as a guide to select the regions for sampling. Two 1.5-mm-diameter cores were punched from each donor block and arrayed in a recipient block. Immunohistochemical analysis was performed on tissue sections generated from the original WTS paraffin block and the TMA block using the Benchmark Automated Slide Stainer (Ventana Medical Systems, Tucson, AZ). Tissues were incubated for 32 minutes with the monoclonal antibody Ki-67 (clone MM1, Novocastra, Newcastle upon Tyne, England) at a 1:5 dilution after a high-temperature antigen unmasking technique. The signal was subjected to amplification using the Ventana Amplification Kit (Ventana Medical Systems). Antibody binding was visualized via the avidinbiotin complex technique, using diaminobenzidine as a chromogen. The developed slides were counterstained with hematoxylin. The LI was determined by manually evaluating 1,000 cells, representing anywhere from a half TMA core to 2 cores, depending on the cellularity of the individual tumor. The 1,000 cells were counted only once. Only nuclear staining was considered positive. The Ki-67 LI was expressed as a percentage of positive cells. The area with the most positivity was chosen for evaluation, based on the assumption that the areas with the highest proliferation index will best predict the biologic behavior of the tumor. Fifty pairs of LIs from TMA vs WTS were generated Image 2 and Image 3. Positive and negative controls were performed. The paired t test was used to compare results from TMA with those from WTS. The Pearson correlation was calculated between the 2 measurements. Recursive partitioning analysis for survival data was used to see whether cut-points could be identified for TMA or WTS that optimize survival. This method did not identify any viable cut-points for TMA or for WTS. Cox proportional hazards analysis was applied to test whether there was an association of TMA or WTS with survival. Results The patients included 17 women and 28 men whose age at diagnosis ranged from 34 to 81 years (median, 59 years; Image 1 Microvascular proliferation and pseudopalisading necrosis in a World Health Organization grade IV astrocytoma (glioblastoma multiforme) (H&E, original magnification 200). Image 2 (Case 19) Ki-67 immunostaining, tissue microarray (labeling index 2.3%) (original magnification, 400). Downloaded from https://academic.oup.com/ajcp/article-abstract/120/6/902/1759043 Am J Clin Pathol 2003;120:902-908 903 903 903

Chiesa-Vottero et al / WHOLE TISSUE SECTION VS MICROARRAY mean, 59.1 years). The tumors were located in the cerebral hemispheres, with extension to the basal ganglia in 2 cases. Tumor size ranged from 1 to 7.4 cm in 33 cases (mean, 4.2 cm; median, 4 cm); the tumor was described as large in 7 additional cases. Information about size was not known in the remaining 5 tumors. Thirty-nine patients underwent surgical debulking, and 3 underwent biopsy only. In 3 cases, the extent of surgical resection was not known. Eight patients underwent a second debulking procedure, and 1 patient underwent a third debulking procedure. Thirty patients were known to have completed postoperative radiotherapy. Gamma knife treatment was administered to 7 patients. The postoperative Karnofsky performance status was available in 21 cases; scores ranged from 40 to 100 (mean, 76; median, 80). Four patients completed systemic chemotherapy treatment, and 9 additional patients started chemotherapy but were unable to complete it. Eight patients received local chemotherapy (insertion of polifeprosan 20 with carmustine implant [Gliadel Wafer], Guilford Pharmaceuticals, Baltimore, MD) at the time of their second debulking. Two patents completed immunotherapy. Five patients were involved in a tamoxifenhypothyroid experimental therapeutic trial. At last follow-up, 7 patients were alive at 1 (n = 3), 2.5 (n = 1), 3.5 (n = 1), 23.5 (n = 1), and 45 (n = 1) months. Thirty-eight patients had died between 1.5 and 23 months after surgery (mean, 10.3 months; median, 10.25 months). In comparison with the theoretically more representative WTS, TMA underestimated the LI in 28 specimens and overestimated the LI in 22 specimens Table 1. For WTS, the LI ranged between 3.0% and 76.4% (mean, 20%; Image 3 (Case 19) Ki-67 immunostaining, whole tissue section (labeling index 9.6%) (original magnification, 400). SD, 15.3%; median, 14.8%). For TMA, the LI ranged between 0.0% and 76.4% (mean, 18%; SD, 13.8%; median, 14.6%). The difference ranged between 56.6 (negative difference indicates that TMA yielded a smaller LI than WTS) and 64.2% (positive difference indicates that TMA yielded a larger LI than WTS), with a mean of 2.0, an SD of 21.0, and a median of 4.2. The paired t test indicated, however, that the difference between the average LI obtained from TMA vs that from WTS was not significant (P =.51) Table 2. The Pearson correlation (r) was 0.042 (P =.77), indicating that the 2 methods were not correlated significantly Figure 1. Survival study using Cox proportional hazards analysis resulted in a hazard ratio of 1.00 for TMA (95% confidence interval, 0.98-1.02; P =.87) and 0.99 for WTS (95% confidence interval, 0.97-1.02; P =.79). This indicates that Ki-67 LI did not correlate with survival by TMA or by WTS. Discussion Although the paired t test in the present study indicated that the difference between the Ki-67 LI obtained from TMA vs WTS was not significant, the lack of correlation indicates that the 2 methods result in very different values. This finding may be a consequence of tumor heterogeneity, lack of standardization of immunohistochemical analysis and Ki- 67 immunostaining in particular, and/or the TMA not being representative of the WTS. Many central nervous system tumors, most notably gliomas, are heterogeneous. 20 Burger and Kleihues 21 mapped out a number of untreated gliomas from autopsied brains and demonstrated that there was considerable variability with regard to histologic appearance and grade from area to area. Regional heterogeneity also has been demonstrated in cell proliferation. 22-25 Studies of glioma cell explants have shown that the regions of highest mitotic activity extend along the advancing edge of solid tumor tissue, ie, at the interface of solid explant and culture medium. 26,27 Selection of a block of representative tumor for immunostaining does not necessarily guarantee that the most proliferative area of the neoplasm is present in that block. 20 Maruno and colleagues 28 found that the distribution of Ki-67 positive nuclei was variable in different sections of solid GBM tissue. Dalrymple et al 25 confirmed these findings in both in vitro and in vivo studies by demonstrating spatial variability in proliferative activity. They observed a higher proportion of proliferating cell nuclear antigen positive glioma cells at the periphery of the solid tissue component and relatively little reactivity within tumor cells located in the surrounding infiltrated parenchyma. 25 904 Am J Clin Pathol 2003;120:902-908 Downloaded 904 from https://academic.oup.com/ajcp/article-abstract/120/6/902/1759043

Anatomic Pathology / ORIGINAL ARTICLE Table 1 Survival and Comparison of Tissue Microarray and Whole Tissue Section Ki-67 Labeling Indices in 50 Data Pairs Labeling Index Data Pair No. Tissue Microarray Whole Tissue Section Difference Survival (mo) 1 19.8 76.4 56.6 11.5 2 19.4 69.5 50.1 5.5 3 14.6 44.8 30.2 6.5 4 15.4 45.0 29.6 19 5 12.9 42.1 29.2 4 6 15.8 44.6 28.8 12.5 7 0.0 19.9 19.9 19.5 8 5.9 25.4 19.5 17 9 12.2 28.2 16.0 20 10 2.7 17.9 15.2 7 11 (same patient as pair 10) 3.0 17.9 14.9 7 12 14.2 26.2 12.0 23 13 (same patient as pair 12) 14.5 26.2 11.7 23 14 3.8 14.2 10.4 11.5 15 9.4 19.1 9.7 4 16 28.2 37.1 8.9 Alive, 1 17 6.6 15.0 8.4 10.5 18 7.8 15.4 7.6 8 19 2.3 9.6 7.3 Alive, 2.5 20 26.2 33.4 7.2 Alive, 3.5 21 6.8 12.9 6.1 11.5 22 7.3 13.0 5.7 15.5 23 12.9 18.2 5.3 Alive, 23.5 24 15.5 20.6 5.1 7 25 8.9 13.8 4.9 15 26 8.7 12.2 3.5 5 27 1.1 4.0 2.9 15 28 (same patient as pair 14) 12.8 14.2 1.4 11.5 29 19.9 19.4 0.5 7.5 30 13.7 12.7 1.0 5.5 31 13.9 12.8 1.1 6 32 10.6 9.4 1.2 14.5 33 7.6 5.9 1.7 9.5 34 (same patient as pair 7) 22.6 19.9 2.7 19.5 35 13.4 10.6 2.8 10 36 23.1 19.8 3.3 11 37 26.2 22.6 3.6 7 38 18.7 14.6 4.1 Alive, 1 39 24.4 13.9 10.5 2.5 40 17.9 5.0 12.9 10.5 41 22.6 8.9 13.7 1.5 42 23.7 6.8 16.9 Alive, 1 43 25.4 7.8 17.6 Alive, 45 44 45.0 23.7 21.3 20 45 33.4 7.6 25.8 13.5 46 34.2 6.7 27.5 4.5 47 30.6 3.0 27.6 5 48 (same patient as pair 35) 44.2 10.6 33.6 10 49 44.0 8.7 35.3 1.5 50 76.4 12.2 64.2 13.5 Table 2 Summary of Mean Ki-67 Labeling Index (%) Generated by Evaluation of the TMA vs WTS Variable N Mean SD Median Minimum Maximum P TMA 50 18.0 13.8 14.6 0.0 76.4 WTS 50 20.0 15.3 14.8 3.0 76.4 Difference * 50 2.0 21.0 4.2 56.6 64.2.51 TMA, tissue microarray; WTS, whole tissue section. * Difference = TMA WTS. Downloaded from https://academic.oup.com/ajcp/article-abstract/120/6/902/1759043 Am J Clin Pathol 2003;120:902-908 905 905 905

Chiesa-Vottero et al / WHOLE TISSUE SECTION VS MICROARRAY LI From WTS 100 80 60 40 20 0 0 20 40 60 80 100 LI From TMA Figure 1 Scatter plot of Ki-67 labeling indices (LIs) from tissue microarray (TMA) vs whole tissue section (WTS). The identity line is graphed and indicates where results from TMA and WTS are equal. The LI is smaller from TMA than from WTS for all points to the left of the line and larger for all points to the right of the line. The Pearson correlation (r ) was 0.042 (P =.77), indicating that the 2 methods result in very different LI values. Variations in methods that can affect the determination of the proliferation LI include the following: how many cells should be evaluated (from hundreds to thousands), how one definitively excludes nontumor cells from evaluation, what area of the tumor should be evaluated, where counting should be performed, and whether counts should be manually determined or evaluated by using an imaging analysis system. 20 A comparative study of MIB-1 staining indices of gliomas measured by the National Institutes of Health Image Analysis Program and the conventional manual cell counting method showed that the paired t test demonstrated no significant difference between methods. 29 The authors used the same areas or fields of each slide for the National Institutes of Health and manual count methods. 29 On the other hand, MIB-1 LIs obtained by the CAS 200 cell analysis system method (Becton Dickinson, San Jose, CA), according to Giannini et al, 30 are significantly lower than the MIB-1 LIs obtained by manual counts (approximately 30%). By convention, counts usually are performed in the area with the most immunostaining, 20 but different workers have used different selection methods. Alternatively, some have recommended counting regions with high and low immunostaining or assessing random fields. 31 An attempt is made to exclude nontumor cells, which may be variably easy to do. 20 Some have advocated counting positive tumor nuclei in the total number of all cells in the fields examined, whereas others describe counting positive nuclei in the total number of cells examined, excluding nontumor cells from the total. 32 Counts should be performed at high magnification and only nuclear staining interpreted as positive. 20 From a practical standpoint, the relative rate of cell proliferation, which can be assessed readily by visual inspection, is more important than the precise number that is generated from a count. 20 According to Grzybicki et al, 32 discrete cutoff values are not useful clinically because pathologists do not agree that, for any given tumor type or grade, indices fall on one side or the other of the cutoff point. Furthermore, absolute or cutoff values for proliferation indices are not necessarily extrapolatable to another laboratory. Standardization of methods is advocated. 30 A final important factor to consider in evaluating LI results includes intraobserver and interobserver variability. 20 In a recent study by Prayson et al, 33 a slide from each of 30 oligodendrogliomas, stained with MIB-1 antibody, was distributed to 6 pathologists to independently determine the LI. Overall agreement was good ( 0.75), with a concordance coefficient of 0.832 (confidence interval, 0.700-0.909). According to the authors, although there is clearly unavoidable interobserver variability in the interpretation of MIB- 1 immunostained sections, this study suggests that correlation between observers is very good and the differences are, in most cases, probably not clinically important. Other studies have demonstrated that findings obtained on TMAs are highly representative of their donor tissues, despite the small size of the individual specimen. 34 Fernebro et al 35 concluded that the TMA technique for immunohistochemical staining in rectal cancer yielded staining of good quality and expression data for Ki-67 and p53 comparable to those obtained with whole tissue staining. Nocito et al 36 studied 2,317 bladder tumors that had been analyzed previously for histologic grade and Ki-67 LI and constructed 4 replica TMAs from different areas of each tumor to determine the representativeness of the results obtained by using TMAs. Despite discrepancies in individual cases, the grade and Ki-67 information obtained on minute arrayed samples were highly similar to the data obtained on large sections (P <.0001). Most important, every individual association between grade or Ki-67 LI and tumor stage or prognosis (recurrence, progression, tumor-specific survival) that was observed in large section analysis could be reproduced fully on all 4 replica TMAs. The authors concluded that intratumor heterogeneity did not significantly affect the ability to detect clinicopathologic correlations on TMAs, probably because of the large number of tumors that can be included in TMA studies. 36 Sampling strategies for both WTS and TMA may not accurately represent cell proliferation in a given neoplasm owing to great regional variation in GBMs in individual cases. Tumor heterogeneity seems to reflect the genetic instability that is a fundamental property of gliomas and is central to the process of tumor progression. The observed heterogeneity in proliferative activity also is undoubtedly a function of the method used to assess it, 24 906 Am J Clin Pathol 2003;120:902-908 Downloaded 906 from https://academic.oup.com/ajcp/article-abstract/120/6/902/1759043

Anatomic Pathology / ORIGINAL ARTICLE although TMA experiments can achieve an unprecedented level of standardization. 34 The use of TMA for determination of the Ki-67 LI did not always correlate with the theoretically more representative tissue section in an individual tumor. The data obtained from the TMA cannot be used reliably for prognostication in individual cases. However, when evaluating a large number of tumors on a TMA, the LI range and mean LI were roughly comparable with the LIs determined with WTSs, therefore suggesting that TMA may be useful for assessing a large population. From the 1 Division of Pathology and Laboratory Medicine and the 2 Department of Biostatistics and Epidemiology, the Cleveland Clinic Foundation, Cleveland, OH. Address correspondence to Dr Prayson: the Cleveland Clinic Foundation, 9500 Euclid Ave, L25, Cleveland, OH 44195. References 1. Rimm DL, Camp RL, Charette LA, et al. Tissue microarray: a new technology for amplification of tissue resources. Cancer J. 2001;7:24-31. 2. Bredel M, Piribauer M, Marosi C, et al. High expression of DNA topoisomerase IIα and Ki-67 antigen is associated with prolonged survival in glioblastoma patients. Eur J Cancer. 2002;38:1343-1347. 3. Kleihues P, Cavenee WK. Pathology and Genetics of Tumors of the Nervous System. Lyon, France: IARC Press; 2000. World Health Organization Classification of Tumors. 4. Burger PC, Green SB. Patient age, histologic features, and length of survival in patients with glioblastoma multiforme. Cancer. 1987;59:1617-1625. 5. Salmon I, Dewitte O, Pasteels JL, et al. Prognostic scoring in adult astrocytic tumors using patient age, histopathological grade, and DNA histogram type. J Neurosurg. 1994;80:877-883. 6. Sneed PK, Prados MD, McDermott MW, et al. Large effect of age on the survival of patients with glioblastoma treated with radiotherapy and brachytherapy boost. Neurosurgery. 1995;36:898-904. 7. Nazzaro JM, Neuwelt EA. The role of surgery in the management of supratentorial intermediate and high-grade astrocytomas in adults. J Neurosurg. 1990;73:331-344. 8. Leibel SA, Scott CB, Loeffler JS. Contemporary approaches to the treatment of malignant gliomas with radiation therapy. Semin Oncol. 1994;21:198-219. 9. Lesser GJ, Grossman S. The chemotherapy of high-grade astrocytomas. Semin Oncol. 1994;21:220-235. 10. Albert FK, Forsting M, Sartor K, et al. Early postoperative magnetic resonance imaging after resection of malignant glioma: objective evaluation of residual tumor and its influence on regrowth and prognosis. Neurosurgery. 1994;34:45-61. 11. Devaux BC, O Fallon JR, Kelly PJ. Resection, biopsy, and survival in malignant glial neoplasms: a retrospective study of clinical parameters, therapy, and outcome. J Neurosurg. 1993;78:767-775. 12. Wood JR, Green SB, Shapiro WR. The prognostic importance of tumor size in malignant gliomas: a computed tomographic scan study by the Brain Tumor Cooperative Group. J Clin Oncol. 1988;6:338-343. 13. Kreth FW, Warnke PC, Scheremet R, et al. Surgical resection and radiation therapy versus biopsy and radiation therapy in the treatment of glioblastoma multiforme. J Neurosurg. 1993;78:762-766. 14. Vaquero J, Zurita M, Morales C, et al. Prognostic significance of endothelial surface score and MIB-1 labeling index in glioblastoma. J Neurooncol. 2000;46:11-16. 15. Haapasalo HK, Sallinen PK, Helen PT, et al. Comparison of three quantitation methods for PCNA immunostaining: applicability and relation to survival in 83 astrocytic neoplasms. J Pathol. 1993;171:207-214. 16. Torp SH, Helseth E, Dalen A, et al. Relationships between Ki-67 labelling index, amplification of the epidermal growth factor receptor gene, and prognosis in human glioblastomas. Acta Neurochir (Wien). 1992;117:182-186. 17. Wakimoto H, Aoyagi M, Nakayama T, et al. Prognostic significance of Ki-67 labeling indices obtained using MIB-1 monoclonal antibody in patients with supratentorial astrocytomas. Cancer. 1996;77:373-380. 18. Pigott TJ, Lowe JS, Palmer J. Statistical modelling in analysis of prognosis in glioblastoma multiforme: a study of clinical variables and Ki-67 index. Br J Neurosurg. 1991;5:61-66. 19. Schroeder R, Feisel KD, Ernestus RI. Ki-67 labelling is correlated with the time to recurrence in primary glioblastomas. J Neurooncol. 2002;56:127-132. 20. Prayson RA. Cell proliferation and tumors of the central nervous system, part II: radiolabeling, cytometric, and immunohistochemical techniques. J Neuropathol Exp Neurol. 2002;61:663-672. 21. Burger PC, Kleihues P. Cytologic composition of the untreated glioblastoma with implications for evaluation of needle biopsies. Cancer. 1989;63:2014-2023. 22. Deckert M, Reifenberger G, Wechsler W. Determination of the proliferative potential of human brain tumors using the monoclonal antibody Ki-67. J Cancer Res Clin Oncol. 1989;115:179-188. 23. Allegranza A, Girlando S, Arrigoni GL, et al. Proliferating cell nuclear antigen expression in central nervous system neoplasms. Virchows Arch A Pathol Anat Histopathol. 1991;419:417-423. 24. Coons SW, Johnson PC. Regional heterogeneity in the proliferative activity of human gliomas as measured by the Ki- 67 labeling index. J Neuropathol Exp Neurol. 1993;52:609-618. 25. Dalrymple SJ, Parisi JE, Roche PC, et al. Changes in proliferating cell nuclear antigen expression in glioblastoma multiforme cells along a stereotactic biopsy trajectory. Neurosurgery. 1994;35:1036-1045. 26. Carlsson J. A proliferation gradient in three dimensional colonies of cultured human glioma cells. Int J Cancer. 1977;20:129-136. 27. Haynes LW, Davis BE, Mitchell J, et al. Cell proliferation in explant cultures of malignant gliomas. Acta Neuropathol (Berl). 1978;42:87-90. 28. Maruno M, Kovach JS, Kelly P, et al. Transforming growth factor-α, epidermal growth factor receptor, and proliferating potential of benign and malignant gliomas. J Neurosurg. 1991;75:97-102. Downloaded from https://academic.oup.com/ajcp/article-abstract/120/6/902/1759043 Am J Clin Pathol 2003;120:902-908 907 907 907

Chiesa-Vottero et al / WHOLE TISSUE SECTION VS MICROARRAY 29. Amin MR, Kurosaki M, Watanabe T, et al. A comparative study of MIB-1 staining indices of gliomas measured by NIH Image Analysis Program and conventional manual cell counting method. Neurol Res. 2000;22:495-500. 30. Giannini C, Scheithauer BW, Burger PC, et al. Cellular proliferation in pilocytic and diffuse astrocytomas. J Neuropathol Exp Neurol. 1999;58:40-53. 31. McKeever PE, Ross DA, Strawderman MS, et al. A comparison of the predictive power for survival in gliomas provided by MIB-1, bromodeoxyuridine and proliferating cell nuclear antigen with histopathologic and clinical parameters. J Neuropathol Exp Neurol. 1997;56:798-805. 32. Grzybicki DM, Liu Y, Moore SA, et al. Interobserver variability associated with the MIB-1 labeling index: high levels suggest limited prognostic usefulness for patients with primary brain tumors. Cancer. 2001;92:2720-2726. 33. Prayson RA, Castilla EA, Hembury TA, et al. Interobserver variability in determining MIB-1 labeling indices in oligodendrogliomas. Ann Diagn Pathol. 2003;7:9-13. 34. Bubendorf L, Nocito A, Moch H, et al. Tissue microarray (TMA) technology: miniaturized pathology archives for highthroughput in situ studies. J Pathol. 2001;195:72-79. 35. Fernebro E, Dictor M, Bendahl PO, et al. Evaluation of the tissue microarray technique for immunohistochemical analysis in rectal cancer. Arch Pathol Lab Med. 2002;126:702-705. 36. Nocito A, Bubendorf L, Maria Tinner E, et al. Microarrays of bladder cancer tissue are highly representative of proliferation index and histological grade. J Pathol. 2001;194:349-357. 908 Am J Clin Pathol 2003;120:902-908 Downloaded 908 from https://academic.oup.com/ajcp/article-abstract/120/6/902/1759043