Validation and Simplification of the Radiation Therapy Oncology Group Recursive Partitioning Analysis Classification for Glioblastoma

摘要: Purpose Previous recursive partitioning analysis (RPA) of patients with malignant glioma (glioblastoma multiforme [GBM] and anaplastic astrocytoma [AA]) produced six prognostic groups (I–VI) classified by factors. We sought here to determine whether the classification for GBM could be improved using an updated Radiation Therapy Oncology Group (RTOG) database excluding AA considering additional baseline variables. Methods Materials The new considered 42 variables 1,672 from expanded RTOG database. Patients receiving radiation only were excluded such that all received radiation+carmustine. “Radiation dose received” was replaced “radiation assigned.” RPA models compared original model applying them a test dataset comprising 488 other trials. Fitness evaluated explained variation. Results more variations in survival than did (20% vs. 15%) therefore chosen further analysis. It reduced combining Classes V VI produce three classes (Classes III, IV, V+VI), as had indistinguishable dataset. simplified not improve performance (explained variation 18% 20%) but is easier apply because it involves four variables: age, status, extent resection, neurologic function. Applying this resulted distinct median times 17.1, 11.2, 7.5 months V+VI, respectively. Conclusions final model, VI, defined This will used future