Morphology of proliferating and non-proliferating tumor cell nuclei in glioblastomas correlates with preoperative data from proton-MR-spectroscopy.

In contrast to the growing interest in proton-MR-spectroscopy (1HMRS) for preoperative examination of patients with brain tumors, there is nearly no knowledge about a correlation between data from 1HMRS and histomorphology as confirmed by quantitative morphological methods. Whether a correlation can be confirmed between data from 1HMRS and quantitative histomorphology of glioblastomas representing the most frequent type of brain tumors was investigated in the present study. Furthermore, it was of interest, whether correlations between spectroscopic data and histomorphology can be confirmed for proliferating and non-proliferating tumor cell nuclei independently. Using stringent inclusion criteria for this study, 24 patients were investigated by means of preoperative 1HMRS and by means of digital image analysis of paraffin sections from the surgical specimen. Proliferating and non-proliferating tumor cell nuclei were investigated separately in the region with the highest proliferative activity in each tumor using immunohistological staining for the proliferation marker Ki67. Main results showed highly significant correlations between total creatine and variables of nuclear size, as well as correlations between choline and variables of nuclear shape. These results were confirmed for both proliferating and non-proliferating tumor cell nuclei. A significant correlation between N-acetyl-aspartate level and topometric variables (number of neighbors per nucleus, variables describing distances between tumor cell nuclei) was confirmed for proliferating tumor cell nuclei. Discriminant analysis provided a good separation of cases with high and with low values for these spectroscopic variables based on histomorphometric data. In conclusion, the results confirm a direct correlation between data from preoperative 1HMRS and histomorphological characteristics of glioblastomas supporting the biological relevance of spectroscopic data for the examination of brain tumor patients.