Abstract
Tumor cell nuclei in 161 primary resected astrocytic tumors have been analyzed by means of digital image analysis together with the determination of the proliferation index Ki-67. All cases had been reevaluated retrospectively by two histopathologists and were typed and graded according to the WHO-classification of central nervous system tumors: Pilocytic astrocytomas (n=5), diffuse astrocytomas grade II (n=15), anaplastic astrocytomas grade III (n=44) and glioblastomas grade IV (n=97). The main aim was to contribute to the question, whether planimetric variables are of primary importance in grading of astrocytic tumors and to test the degree of statistical separation of the different tumor grades. There was a nearly 100% correct reclassification of all astrocytomas by means of discriminant analysis with the exception of one single case. This result was achieved exclusively by shape variables determined by Fourier-analysis. The variability of nuclear shape within the single tumor case showed a less pronounced contribution to grading than mean and median values of shape variables. In contrast, nuclear size, variation of nuclear size and proliferation index showed a considerable overlap between the tumor grades. Different from astrocytomas, glioblastomas revealed a pronounced heterogeneity of nuclear morpholgy. The detailed analysis of nuclear shape by means of Fourier-analysis provided a better separation of the tumor grades than a simple shape factor. Together with other reports on quantitative morphological analyses of astrocytic tumors, the results provide a realistic perspective for a quantitatively assessed grading of astrocytic tumors, especially for the differentiation between astrocytomas grade II and III according to the WHO-classification.
Get full access to this article
View all access options for this article.
