T. Murai1,2, Y. Kasai3, Y. Eguchi3, S. Takano2, N. Kita2, A. Torii2, T. Takaoka2, N. Tomita2, Y. Shibamoto4, and A. Hiwatashi2; 1Department of Radiation Oncology, Shonan Kamakura General Hospital, Kamakura, Japan, 2Department of Radiology, Nagoya City University Graduate School of Medical Sciences, Nagoya, Japan, 3Department of Radiology, Nagoya City University Hospital, Nagoya, Japan, 4Narita Memorial Proton Center, Toyohashi, Japan
Purpose/Objective(s): Since stereotactic radiotherapy (SRT) for large brain metastases (BMs) is known to be associated with a higher risk of brain necrosis (BN), different fractionation protocols depending on tumor size are often employed. However, dosimetric predictors of the normal brain have not been clarified yet in fractionated SRT. Materials/
Methods: Patients (pts) who were treated by fractionated SRT protocols were analyzed. Following pts were excluded in this study: 1) previous surgery or radiotherapy history; 2) meningitis carcinomatosa; 3) melanoma, hematologic malignancy, and small cell lung cancer; 4) WHO performance status of 3–4; 6) BM number > 10. In the first protocol employed from 2012 to 2016, the basic prescribed dose was 35 Gy/5 fr. For small lesions (maximum diameter < 1.5 cm), 30 Gy/3 fr was permitted. For large lesions (= 3 cm), 37.5 Gy/5 fr was used. From 2017 to 2021, the protocol for the large BMs was revised to decrease the incidence of BN. For large lesions (= 15 cc), treatment with 35 Gy or 37.5 Gy/5 fr was superseded by 40 Gy/10 fr. These pts were followed once every < 3 months with physical examinations and contrast-enhanced MRI. BN was pathologically verified or diagnosed by functional MRI findings. To explore dosimetric predictors of BN, the dose-volume data of the normal organs and targets were extracted from the radiation planning systems. The biological equivalent doses to the brain with an a/ß ratio of 2 Gy were calculated using the linear quadratic model to assess different fractionation schedules. Multivariate logistic regression models were developed to explore BN predictors. Results: In total, 112 pts (median age 68 years) with 215 BMs were treated according to these protocols. Among them, 76 pts had lung cancers and 106 had extracranial lesions. The 3-fraction (-fr) treatment was delivered to 34 pts with 74 BMs, the 5-fr treatment to 58 with 106 BMs, and the 10-fr treatment to 20 with 35 BMs. Mean target volumes in the 3-, 5- and 10-fr groups were 4.3, 15.4, and 25.9 cc, respectively (p < 0.001). The overall survival rate at 1 year was 54% (median, 13 months). One-year local tumor control rates were 92% in the 3-fr group, 94% in the 5-fr group, and 82% in the 10-fr group (p = 0.11). BN of grade (G ) 2 or higher occurred in 4 pts in the 3-fr, 7 in the 5-fr, and 0 in the 10-fr groups. BNs (= G1) were observed in 7 in the 3-fr, 9 in the 5-fr, and 1 in the 10-fr groups. In the 5- or 10-fr groups, BNs were more frequently observed in pts with a larger brain volume receiving 55 Gy biologically equivalent dose in 2 Gy fractions (V55GyE). The odds ratios of = G1 and = G2 BNs were 1.06 ((95% codidence interval: 1.01-1.12, p < 0.03) and 1.06 (1.01-1.12, p < 0.02), respectively. G2 BNs occurred in 3% (V55GyE = 15 cc), 8% (> 15 cc, = 30 cc), and 22 % (> 30 cc). In the 3-fr group, no risk predictor was identified. Conclusion: BrainV55GyE emerges as a significant dosimetric predictor for BN in cases treated with 5- or 10-fr SRT.
