Background: Astrocytoma is a primary brain tumor that affects 20,000 Americans each year. To date, only age and histologic grade stand out as independent predictors of survival. There is now increased interest in the use of molecular markers as objective standards against which to establish diagnosis and grade.
Methods: The study evaluated human glioma tumor suppressor genes and associated loci in fresh snap-frozen gliomas from 63 males and 37 females, with a median age of 42 years, including 19 low-grade astrocytomas. The tumor samples were selected so that about equal numbers of glioblastomas from younger and older patients were represented in the series. Methods for suppressor gene and genetic loci evaluation included loss of heterozygosity (LOH) analysis, multiplex polymerase chain reaction analysis, and gene sequencing.
Results: Low-grade astrocytomas had the least number of molecular abnormalities. LOH on 9p and/or CDKN2A deletion occurred more often in glioblastomas (P < 0.001), LOH on 17p/TP53 mutations occurred more frequently in anaplastic astrocytomas (AAs; P = 0.112), and LOH on 10q/PTEN mutation frequency was similar in glioblastomas and AAs (P < 0.001). Poorer survival was associated significantly with the occurrence of either deletion of p16 (P = 0.031), LOH on 9p (P = 0.016), or LOH on 10q (P = 0.0007). The absence of LOH on 17p and the presence of PTEN mutation were associated marginally with survival. Even though TP53 mutations were more frequent among younger patients with glioblastoma, they had no statistically significant effect on survival after adjustment for age (P = 0.62). In all multivariate models, age and grade were the only significant predictors of survival or were nearly significant predictors of survival.
Conclusions: The results suggest that LOH on 9p and p16 deletions may prove to be objective standards for the diagnosis of patients with high-grade gliomas, although the absence of these abnormalities is nonprognostic.