Clinical, histopathologic, and molecular markers of prognosis: Toward a new disease risk stratification system for medulloblastoma

Purpose To assess the feasibility of performing central molecular analyses of fresh medulloblastomas obtained from multiple institutions and using these data to identify prognostic markers for contemporaneously treated patients. Materials and Methods Ninety-seven samples of medulloblastoma were collected. Tumor content in samples was judged by frozen section review. Tumor ERBB2 protein and MYCC, MYCN, and TRKC mRNA levels were measured blind to clinical details using Western blotting and real-time polymerase chain reaction, respectively. Histopathologic and clinical review of each case was also performed. All data were subjected to independent statistical analysis. Results Sample acquisition and analysis times ranged from 3 to 6 days. Eighty-six samples contained sufficient tumor for analysis, including 38 classic, 30 nodular desmoplastic, and 18 large-cell anaplastic (LCA) medulloblastomas. Protein and mRNA were extracted from 81 and 49 tumors, respectively. ERBB2 was detected in 40% (n = 32 of 81) of tumors, most frequently in LCA disease (P = .005), and was independently associated with a poor prognosis (P = .031). A combination of clinical characteristics and ERBB2 expression provided a highly accurate means of discriminating disease risk. One hundred percent (n = 26) of children with clinical average-risk, ERBB2-negative disease were alive at 5 years, with a median follow-up of 5.6 years, compared with only 54% for children with average-risk, ERBB2-positive tumors (n = 13; P = .0001). TRKC, MYCC, and MYCN expression and histopathologic subtype were not associated with prognosis in this study. Conclusion Central and rapid molecular analysis of frozen medulloblastomas collected from multiple institutions is feasible. ERBB2 expression and clinical risk factors together constitute a highly accurate disease risk stratification tool. (C) 2004 by American Society of Clinical Oncology.