Radiosurgery in the treatment of malignant gliomas: The University of Florida experience

OBJECTIVE: To review a single-institution's 12-year experience of treating malignant gliomas with linear accelerator-based radiosurgery. METHODS: One hundred patients were treated for malignant gliomas at the University of Florida between May 1, 1989, and June 12, 2002. Patient variables were entered into a radiosurgery database in a prospective manner. The endpoint of the study was length of survival from the time of diagnosis. The minimum length of follow-up was 18 months or until death. In an attempt to control for selection bias, we used the Radiation Oncology Therapy Group's (RTOG) recursive partitioning categories to compare survival in our group of radiosurgically boosted patients with the RTOG historical database. RESULTS: Recursive partitioning analysis classification was significantly associated with survival. Compared with historical controls, this cohort of patients demonstrated a decreased survival for recursive partitioning analysis Class I and II patients, similar survival for Class III and IV patients, and increased median survival for Class V patients. Other variables demonstrating a statistically significant impact on survival were eloquent location and recurrent disease. CONCLUSION: The results of this study and those of RTOG 93-05 suggest that the reported benefit of upfront radiosurgical boost for malignant gliomas is the result of selection bias. The possibility remains that radiosurgical boost is of benefit in recurrent tumors, especially those in RTOG Class V.