White matter changes are correlated significantly with radiation dose. Observations from a randomized dose‐escalation trial for malignant glioma (radiation therapy oncology group 83‐02)

Background. A Phase I/II randomized dose‐seeking trial was performed to document the severity, time course, and significance of white matter changes seen on serial imaging scans (magnetic resonance imaging, computed tomography) associated with bis‐chlorethyl nitrosourea (BCNU) and hyperfractionated cranial irradiation. Methods. Long term survivors (⩾ 18 months) were identified from a prospective randomized dose‐escalation Phase I/II trial designed to evaluate twice‐daily radiotherapy for supratentorial high grade malignant gliomas. All scans were reviewed by a neuroradiologist who had no information about the prescribed dose and fractionaction. In the trial, patients were assigned to receive 64.8 Gy, 72.0 Gy, 76.8 Gy, or 81.4 Gy (all fractionated as 1.2 Gy twice a day [bid]), or 48.0 Gy or 54.4 Gy (both in 1.6‐Gy bid fractions). Bis‐chlorethyl nitrosourea was administered every 8 weeks for 1 year. Of 747 randomized patients, 177 had analyzable scans. The scans reviewed were those acquired preoperatively, immediately postoperatively, 3, 6, 12, and 18 months after radiotherapy. Radiographic endpoints included no white matter change (Grade 0), minimal patchy white matter foci (Grade 1), start of confluence of white matter disease (Grade 2), large confluent areas (Grade 3), confluence with cortical/subcortical involvement (Grade 4), leukoencephalopathy (Grade 5), and possible necrosis (Grade 6) according to the classification of F. Fazekas et al. ⁹The effects were scored relative to the baseline preoperative scans. The dose pairs of 48 Gy and 54.4 Gy, 64.8 Gy and 72 Gy, and 76.8 Gy and 81.4 Gy were grouped together for analysis (low, intermediate, and high dose, respectively). Toxicity was analyzed in three ways: Grade 2 or worse, Grade 3 or worse, and Grade 6. Results. Grade 2 or worse changes were observed in 26.6, 27.6, and 40.4% of patients in the low, intermediate, and high dose groups, respectively. Grade 3 or worse changes were observed in 8.3, 20.0, and 36.5% of patients in the low, intermediate, and high dose groups, respectively. Grade 6 changes were observed in 1.6, 4.6, and 19.2% of patients in the low, intermediate, and high dose groups, respectively. No statistically significant differences were observed among treatment groups when toxicity was evaluated as Grade 2 or worse. For toxicity of Grade 3 or worse, an chi‐square test revealed P values of 0.04 (low vs. intermediate dose), 0.09 (intermediate vs. high dose), and 0.0005 (low vs. high dose). With the endpoint of possible necrosis (Grade 6), P values were 0.21 (low vs. intermediate dose), 0.05 (intermediate vs. high dose), and 0.003 (low vs. high dose). The median time to radiographic appearance of an effect (15 months) was not influenced by total dose or fraction size. Conclusions. A well described toxicity scale for white matter injury was applied successfully to patients with malignant glioma treated with definitive irradiation. Severe white matter changes continued to increase significantly as the total dose of hyperfractionated cranial irradiation was escalated. The time to onset of the white matter abnormalities appeared to be independent of dose. An ongoing Radiation Therapy Oncology Group study will allow correlation of white matter injury with prospective neuropsychometric testing.