Supratentorial glioblastoma treated with radiotherapy: use of the Radiation Therapy Oncology Group recursive partitioning analysis grouping for predicting survival.

Objective: This study aimed to evaluate the usefulness of recursive partitioning analysis model established by the Radiation Therapy Oncology Group for predicting the survival of patients with supratentorial glioblastoma treated with radiotherapy and to determine prognostic factors for the subgroups of this prognostic model.

Methods: A total of 108 glioblastoma patients treated with radiotherapy between January 1987 and December 2005 were retrospectively reviewed. Recursive partitioning analysis classes III, IV, V and VI included 8, 29, 32 and 39 patients, respectively. These classes were divided into two subgroups: a good prognostic group containing classes III-IV and a poor prognostic group containing classes V-VI. The median radiation dose was 60 Gy. Seventy-five patients received chemotherapy and/or immunotherapy.

Results: The overall survival differed significantly among classes III, IV, V and VI, with median survival times of 34, 15, 11 and 7 months, respectively. Among the good prognostic group, patients with basal ganglia invasion showed poorer survival outcomes than patients without basal ganglia invasion. Among the poor prognostic group, patients with tumor sizes of <5 cm and patients treated with nimustine hydrochloride showed better survival outcomes than those with tumor sizes of > or =5 cm and those without treatment with nimustine hydrochloride, respectively.

Conclusions: This study confirms the prognostic value of the recursive partitioning analysis grouping. Basal ganglia invasion could be a useful predictive factor for survival in the good prognostic group, whereas tumor size and treatment with nimustine hydrochloride could be useful predictive factors in the poor prognostic group.