In modern neuro-oncology, no variable affects therapeutic decisions and prognostic estimation more than tumor classification. Among high grade gliomas, glioblastomas and anaplastic oligodendrogliomas follow markedly different clinical courses; oligodendrogliomas have a more favorable prognosis and remain the only subtype of glioma that commonly responds to chemotherapy. Unfortunately, many malignant gliomas are diagnostically challenging; these non-classic lesions can be difficult to classify according to histological features, creating considerable interobserver variability and limited diagnostic reproducibility. We demonstrated recently that class prediction models, based on gene expression profiles, could classify these diagnostically challenging malignant gliomas in a manner that better correlates with clinical outcome than standard pathology. We have since attempted to use genes identified in our microarray analysis to develop accurate immunohistochemical markers for the classification of high grade gliomas. One potential marker identified in our gene expression study was YKL-40, a protein that has been implicated in pathologic conditions of extracellular matrix degradation and angiogenesis. High serum levels of YKL-40 have been detected in a number of cancers, including glioma. We optimized an immunohistochemical protocol for YKL-40 (Quidel Corporation, San Diego CA) and stained six tumors from our original gene expression microarray analysis. Three anaplastic oligodendrogliomas that were negative for YKL-40 RNA did not demonstrate YKL-40 staining by immunohistochemistry. Positive YKL-40 staining was seen in three glioblastomas that were found to express YKL-40 RNA. Next, we stained tissue microarrays containing an independent set of glioblastomas (n=20) and anaplastic oligodendrogliomas (n=18) sampled in quadruplicate. All but one of the 20 glioblastomas stained positive for YKL-40 whereas the majority of anaplastic oligodendrogliomas were negative; when positive staining was found in oligodendrogliomas, it was most often in microgemistocytes and/or endothelial cells. Furthermore, YKL-40 appeared to provide a better class distinction than GFAP, the current standard immunohistochemical marker used clinically to help distinguish diagnostically challenging gliomas. This study demonstrates that candidate markers identified in gene expression-based microarray analyses can be used successfully to develop accurate immunohistochemical markers for the classification of high grade gliomas. The identification of YKL-40 and additional markers could allow improved stratification of groups of patients with gliomas in clinical trials and molecular marker studies.

[Proc Amer Assoc Cancer Res, Volume 45, 2004]