Representative sampling of gliomas at biopsy is essential for correct assignation of histological grade and subsequent patient management. If sampling is unrepresentative, tissue devoid of mitoses or necrosis may be obtained and result in a falsely low glioma grade. The objective of the present study was to assess whether the MIB‐1 labelling index of glioma tissue cores specifically constructed so as to replicate ‘unrepresentative' astrocytoma biopsies would predict the real or actual glioma grade. Tissue microarrays were prepared from 134 samples of low‐grade astrocytoma (LGA; WHO grade II), anaplastic astrocytoma (AA) and glioblastoma multiforme (GBM). Donor blocks were sampled in such a way as to avoid mitoses, necrosis and endothelial hyperplasia. Immunohistochemistry was performed using the Ki‐67 (MIB‐1) proliferation marker, and the percentage of MIB‐1‐positive cells per core was calculated. Mean MIB‐1% values for LGA (n = 47), AA (n = 38) and GBM (n = 46) were mean (SD) 0.54 (0.82), 5.68 (6.69) and 7.21 (7.98); ranges 0–3.07, 0–30.08 and 0–29.08, respectively. An MIB‐1% count of >3.07 excludes LGA, but an MIB‐1% count of <3.07 does not exclude GBM or AA. It is concluded that in unrepresentative glioma biopsy material, MIB‐1 labelling may be used to exclude LGA, but cannot be used to distinguish between GBM and AA.
Astrocytomas in adults demonstrate considerable cellular heterogeneity, and pathological grading is the most significant predictor of clinical outcome.1 While grading may be suggested radiologically, it is recommended that all astrocytomas be biopsied to establish a correct diagnosis and grade. The presence of mitoses and necrosis is a key adjunct in the grading of astrocytomas. Mitoses are a feature of high‐grade gliomas, including anaplastic astrocytomas (AAs), and would not generally be seen on a biopsy of a low‐grade astrocytoma (LGA). Moreover, the presence of palisaded necrosis in an astrocytoma distinguishes GBM from AA.
Inadequate or unrepresentative tissue sampling at biopsy may lead to erroneous diagnosis of an astrocytoma having a lower grade than is actually the case, with adverse consequences in terms of failure to administer supplementary treatments including radiotherapy. If sampling is inaccurate, glioblastoma multiforme (GBM) may be devoid of mitoses and necrosis, and AAs may be devoid of mitoses.
Evaluation of proliferative activity using the Ki‐67 (MIB‐1) antibody has been used as an adjunct in the assessment of tumour grade in human astrocytomas,2,3 and its prognostic value in distinguishing subtypes is recognised.4,5 We sought to evaluate the ability of this marker to predict the underlying histological grade of gliomas using tissue microarray (TMA) technology to simulate inadequately sampled biopsy specimens. TMA methodology is now widely used in cancer research6 and is a validated method for tumour immunoprofiling.7 More recently, it has been used in the study of non‐neoplastic disease of the central nervous system.8 In this study, we hypothesised that the MIB‐1 labelling index in tissue cores designed to replicate unrepresentative astrocytoma biopsies would predict the histological grade in three astrocytoma subtypes.
Patients and methods
Subject material
Adult astrocytomas were selected from the neuropathology tissue archives in Beaumont Hospital, Dublin, Ireland. The material that was chosen consisted of surgically resected formalin‐fixed paraffin‐embedded tumour tissue. Astrocytomas were graded according to the WHO International Classification of brain tumours.1 Forty‐eight were classified as LGA (WHO grade II); 39 as AA (WHO grade III); and 47 as GBM (WHO grade IV), yielding 134 cores. WHO grade I or pilocytic astrocytomas, glioneuronal tumours and mixed oligoastrocytomas were specifically excluded.
Tissue microarray construction
TMAs were constructed by extracting ‘core biopsies' from a precise morphologically representative area of the original paraffin “donor” block, with subsequent re‐embedding of carefully selected cores into a “recipient” microarray paraffin block. This was done using a precision instrument (Beecher Instruments, Silver Spring, MD, USA), employing separate core needles for punching donor and recipient blocks. Glioma TMAs were prepared using cores designed to replicate intraoperative inadequate astrocytoma sampling. Cores (0.6 mm diameter) were prepared from resected LGA, AA and GBM. During core sampling, tissue areas were targeted to avoid mitoses, necrosis and endothelial hyperplasia. Specifically, two experienced neuropathologists examined haematoxylin and eosin‐stained sections of the original tumour resection, and astrocytoma grade was agreed. Core biopsies were not used as they did not provide sufficient tissue for sampling nor could we have been certain that such true in vivo biopsies accurately reflected correct tumour grade. Areas of highest cell density but without mitoses were marked with a Nikon optical circular marker having a diameter of 1 mm. With grade II astrocytomas, which by definition are devoid of mitoses, we sought to sample the most cellular areas. The donor block was overlaid with the marked slide; cores were obtained from within the marked circle and subsequently arranged in the recipient tissue block together with location and immunocytochemical controls. Later, all cores in the recipient block were again examined to ensure that mitoses were not inadvertently included.
Immunohistochemistry
TMAs were immunostained with the Ki‐67 (MIB‐1) antibody (DakoCytomation, Glostrup, Denmark) using a standard avidin–biotin peroxidase technique with the Vectastain® Elite ABC kit (Vector Labs, Burlingame, CA, USA) with diaminobenzidine as the chromogenic substrate. The sections were counterstained with haematoxylin.
Scoring of MIB‐1 immunostaining
The tissue arrays were labelled so that the investigator (K.M.S.) was blinded as to the origin of the individual cores. Using image analysis (Olympus DP‐soft version 3) with a grid overlay, the individual cores were evaluated for total cell count and total number of MIB‐1‐positive cells. The results were then expressed as the percentage of MIB‐1‐positive tumour cells per core. MIB‐1 positivity was defined as definitive brown nuclear staining. The mean number of cells per core analysed was 516, 504 and 942 for LGA, AA and GBM, respectively. Statistical analysis was performed using a t test. A p value of <0.05 was considered statistically significant.
Results
LGA tumours had significantly lower MIB‐1% positivity in comparison with GBM (T = −5.7; p<0.0001) and AA (T = −4.6; p<0.0001) (table 1, fig 1). AA was marginally lower than GBM, but this did not reach statistical significance (T = 0.6; p = 0.54).
Table 1 Mean MIB‐1% labelling indices in each of the glioma subtypes.
| GBM (n = 46) | AA (n = 38) | LGA (n = 47) | |
|---|---|---|---|
| Mean | 7.21 | 5.68 | 0.54 |
| SD | 7.9 | 6.69 | 0.82 |
| Range | 0–29.08 | 0–27.16 | 0–3.07 |
GBM, glioblastoma multiforme; AA, anaplastic astrocytoma; LGA, low‐grade astrocytoma
Figure 1 Representative cores from each of the three glioma subtypes. (A) Low‐grade astrocytoma: total cell count 728; total MIB‐1‐positive cells 2; MIB‐1% index 0.27%. (B) Anaplastic astrocytoma: total cell count 985; total MIB‐1‐positive cells 67; MIB‐1% index 6.8%. (C) Glioblastoma multiforme: total cell count 1350; total MIB‐1‐positive cells 256; MIB‐1% index 18.9%.
Mean (SD) MIB‐1% values for LGA (n = 48), AA (n = 39) and GBM (n = 47) were 0.54 (0.82), 5.67 (6.69) and 7.21 (7.98); ranges 0–3.07, 0–30.08 and 0–29.08, respectively.
An MIB‐1 count >3.07% excludes an LGA but, conversely, an MIB‐1 count of <3.07% does not exclude either AA or GBM (fig 2). Thus, GBM and AA have similar proliferation rates in unrepresentative biopsies and are not distinguishable by this method.
Figure 2 Distribution of MIB‐1% positivity in each of the three glioma types. GBM, glioblastoma multiforme; AA, anaplastic astrocytoma; LCA, low‐grade astrocytoma.
A summary of the MIB‐1 labelling indices obtained in each of the tumour types is given in table 1.
Discussion
In this study, which was designed to simulate unrepresentative astrocytoma sampling, we demonstrated that an MIB‐1 index >3.07% excludes LGA. An MIB‐1 index of <3.07% does not exclude a higher grade astrocytoma. Furthermore, the significant overlap in MIB indices between AA and GBM renders distinction between these two grades of astrocytoma unreliable in unrepresentative biopsies.
Take‐home messages
Biopsy sampling of gliomas is essential for correct classification of histological grade.
Unrepresentative sampling that lacks mitoses or necrosis may result in a falsely low grade.
Ki‐67 (MIB‐1) labelling is a useful adjunct in the assessment of glioma grade and is of prognostic value.
In this study, we assessed whether the MIB‐1 labelling index in tissue cores designed to replicate unrepresentative biopsies would predict the histological grade in three astrocytoma subtypes.
Our findings indicate that in unrepresentative glioma biopsy material, MIB‐1 labelling may be used to exclude low‐grade astrocytoma, but cannot be used to distinguish between glioblastoma multiforme and anaplastic astrocytomas.
In the post‐surgery management of astrocytomas in adults, it is vital to distinguish between LGA and AA, whereas distinction between AA and GBM is less critical in that there is no debate about the need for maximum tumour reduction followed by whole brain irradiation (RT). The benefits of RT for LGA are less evident, as is the benefit/risk ratio of surgery in LGA. Many centres adopt an expectant approach to LGA—that is, biopsy with interval imaging, reserving RT for tumours which show clinical or radiological evidence of progression. A false diagnosis of LGA could result in a patient being denied the possible benefits of early RT. In practice, the presence of enhancement with contrast in an astrocytoma over‐rides the predictive value of a diagnosis of LGA on biopsy. In cases of adult supratentorial astrocytoma without contrast enhancement, an MIB‐1 index of >3.07% is, in our practice, used to sway a treatment decision in favour of early RT. We did not attempt to relate the MIB‐1 index to outcome as we would have required a larger patient sample to control for other variables such as patient age, residual post‐surgical tumour volume and degree of neurological disability, factors which of themselves are independent determinants of outcome in astrocytoma.
Footnotes
Competing interests: None declared.
References
- 1.Kleihues P, Sobin L H.WHO classification of brain tumours. Pathology and genetics of tumours of the nervous system Lyon: IARC Press, 2000
- 2.Burger P C, Shibata T, Kleihues P. The use of the monoclonal antibody Ki‐67 in the identification of proliferating cells: application to surgical neuropathology. Am J Surg Pathol 198610611–617. [DOI] [PubMed] [Google Scholar]
- 3.Parkins C S, Darling J L, Gill S G.et al Cell proliferation in serial biopsies through human malignant brain tumours: measurement using Ki‐67 antibody labeling. Br J Neurosurg 19915289–298. [DOI] [PubMed] [Google Scholar]
- 4.Heesters M A, Koudstaal J, Go K G.et al Analysis of proliferation and apoptosis in brain gliomas: prognostic and clinical value. J Neurooncol 199944255–266. [DOI] [PubMed] [Google Scholar]
- 5.Sallinen P K, Haapasalo H K, Visakorpi T.et al Prognostication of astrocytoma patient survival by Ki‐67 (MIB‐1), PCNA, and S‐phase fraction using archival paraffin‐embedded samples. J Pathol 1994174275–282. [DOI] [PubMed] [Google Scholar]
- 6.Kononen J, Bubendorf L, Kallioniemi A.et al Tissue microarrays for high‐throughput molecular profiling of tumor specimens. Nat Med 19984844–847. [DOI] [PubMed] [Google Scholar]
- 7.Gulmann C, Butler D, Kay E.et al Biopsy of a biopsy: validation of immunoprofiling in gastric cancer biopsy tissue microarrays. Histopathology 20034270–76. [DOI] [PubMed] [Google Scholar]
- 8.Goldstine J, Seligson D B, Beizai P.et al Tissue microarrays in the study of non‐neoplastic disease of the nervous system. J Neuropathol Exp Neurol 200261653–662. [DOI] [PubMed] [Google Scholar]