Distribution of nuclear size and internuclear distance are important criteria for grading astrocytomas

Clin Neuropathol. Jan-Feb 2006;25(1):48-56.


Aim: The differentiation between low-grade astrocytomas and anaplastic astrocytomas is susceptible to considerable inter-observer variability. In order to contribute to a better standardization of astrocytoma-grading based on quantitative data, the present study focuses on two important aspects not being considered in previous morphometric studies: elaboration of a decision flow chart for tumor grading based on morphometric parameters and appropriate cut-off-values, easily performed using low-cost equipment such as measuring oculars; investigation of the distribution (histograms) of parameters describing nuclear size and internuclear distance, which had been represented in previous studies by their mean and standard deviation only.

Material and methods: At least 300 tumor cell nuclei per case were investigated in paraffin sections from surgical specimen of 75 patients with astrocytomas WHO grade II (n = 23) and anaplastic astrocytomas WHO grade III (n = 52) by means of a digital image analysis system.

Results: The morphometric data showed significant differences between both groups of tumors. According to multivariate analysis, the best contribution to tumor grading was achieved by means of parameters concerning the distribution of values for nuclear diameters and internuclear distances. A decision tree was constructed using a knowledge based algorithm, which provided astrocytoma grading based on the distribution of values for nuclear diameter, as well as the numerical nuclear density and proliferation index. Measurements using a measuring ocular took an acceptable amount of time (1.5 hour per case) and showed good reproducibility when compared with measurement by means of digital image analysis.

Conclusion: The study demonstrates that a morphometric examination of tumor cell nuclei in paraffin sections supports the clinically important differential diagnosis between low-grade and high-grade astrocytomas. The method for classification and the data published in the present study constitute a good basis for a standardized and reproducible grading procedure for astrocytomas, which can be performed in any histologic laboratory even without a digital image analysis system.

MeSH terms

  • Algorithms
  • Astrocytoma / classification*
  • Astrocytoma / pathology*
  • Astrocytoma / ultrastructure
  • Brain Neoplasms / classification*
  • Brain Neoplasms / pathology*
  • Brain Neoplasms / ultrastructure
  • Cell Nucleus / ultrastructure*
  • Decision Trees
  • Humans
  • Image Processing, Computer-Assisted
  • Reproducibility of Results