Proliferative Potential of Astrocytomas and Glioblastomas

Abstract

Bromodeoxyuridine (BUdR) labeling studies were performed to estimate the growth potential of 462 neuroectodermal tumors in situ. The BUdR labeling index (LI), or S-phase fraction, reflected the proliferative activity of individual tumors and revealed differences in the proliferative potential of histologically similar tumors. A higher LI correlated with a shorter duration of survival. Statistical analysis with the Cox proportional hazards stepwise model showed that age was the best predictor of survival among glioblastoma patients; adding the BUdR LI did not increase the predictive power. However, the LI correlated with age, as indicated by the lower LIs of younger patients. In patients with highly (HAA) and moderately anaplastic astrocytoma (MoAA), the BUdR LI was the best predictor of survival. Multivariate analysis revealed that age and BUdR LI were the best predictors of survival for glioblastoma and HAA patients; adding a term to distinguish between HAA and glioblastoma did not improve the prediction. From the fit of the proportional hazards model, equations were derived to estimate survival probability based on age, LI, and diagnosis. The prognosis for MoAAs was best defined by the BUdR LI, while the prognoses for glioblastomas and HAAs were a function of age and LI. Entering the same LI into the equations resulted in a different survival probability for MoAA patients than for glioblastoma/HAA patients. Without effective treatments for specific tumor phenotypes, the survival of patients with intracranial gliomas appears to depend on the proliferative potential of individual tumors.