CpG islands frequently contain gene promoters or exons and are usually unmethylated in normal cells. Methylation of CpG islands is associated with delayed replication, condensed chromatin and inhibition of transcription initiation. The investigation of aberrant CpG-island methylation in human cancer has primarily taken a candidate gene approach, and has focused on less than 15 of the estimated 45,000 CpG islands in the genome. Here we report a global analysis of the methylation status of 1,184 unselected CpG islands in each of 98 primary human tumours using restriction landmark genomic scanning (RLGS). We estimate that an average of 600 CpG islands (range of 0 to 4,500) of the 45,000 in the genome were aberrantly methylated in the tumours, including early stage tumours. We identified patterns of CpG-island methylation that were shared within each tumour type, together with patterns and targets that displayed distinct tumour-type specificity. The expression of many of these genes was reactivated by experimental demethylation in cultured tumour cells. Thus, the methylation of particular subsets of CpG islands may have consequences for specific tumour types.

Combined loss of 1p and 19q is a statistically significant predictor of prolonged survival in patients with pure oligodendroglioma, independent of tumor grade. Given the lack of this association in patients with astrocytic neoplasms and the previously demonstrated chemosensitivity of oligodendrogliomas, a combined approach of histologic and genotypic assessment could potentially improve existing strategies for patient stratification and management.

reviewed independently by 4 neuropathologists to determine interobserver diagnostic concordance rates. Cases wherein diagnostic disagreements arose were re-1 Division of Neuropathology, Barrow Neurologviewed jointly to identify and refine the histologic criteria that were adversely ical Institute, St. Joseph's Hospital and Medical affecting diagnostic reproducibility. Using the criteria developed in the study, a Center, Phoenix, Arizona.set of 315 gliomas with known survival data was evaluated in order to validate the 2 Department of Pathology and Laboratory Medusefulness of the criteria. icine, Mayo Clinic and Mayo Foundation, Roch- RESULTS.There was significant improvement in diagnostic concordance with each ester, Minnesota.session (P Å 0.02). For the first session, the concordance rates were as follows: all 3 Department of Neuropathology, The Ohio 4 reviewers, 52%; any 3 reviewers, 60%; 2 reviewers, 70%. For the fourth session, State University, Columbus, Ohio.the respective rates were 69%, 75%, and 80%. Although features important in grading, particularly microvascular proliferation, were sometimes problematic, 4 Department of Statistics, The Ohio State University, Columbus, Ohio.most disagreements related to the classification of tumors. Much of the improvement related to the refinement of criteria distinguishing diffuse astrocytomas from oligodendrogliomas/oligoastrocytomas and pilocytic astrocytomas. It was concluded that the presence of any typical oligodendroglioma was sufficient to remove a tumor from the astrocytoma category. CONCLUSIONS.The authors' data indicate that oligodendroglial tumors comprise up to 25% of gliomas, a significantly higher proportion than was previously recognized. The data also suggest that the wide range of survival times reported for patients with anaplastic astrocytoma may reflect ''contamination'' resulting from misdiagnosis, particularly of oligodendroglial tumors and pilocytic astrocytomas.

There is no benefit to the treatment of high-grade astrocytomas in children with eight-drugs-in-1-day chemotherapy compared with CCNU, vincristine, and prednisone. Extent of tumor resection and histopathologic diagnosis are significant prognostic variables. The overall outcome for children with high-grade astrocytomas remains poor.