Fractionated Stereotactic Intensity-Modulated Radiotherapy for Large Brain Metastases: Comprehensive Analyses of...

The objective was to explore dosimetric predictors of brain necrosis (BN) in fractionated stereotactic radiotherapy (SRT).

After excluding collinearities carefully, multivariate logistic models were developed for comprehensive analyses of dosimetric predictors in patients who received first-line fractionated SRT for brain metastases (BMs). The normal brain volume receiving an xx Gy biological dose in 2 Gy fractions (VxxEQD2) was calculated from the retrieved dose-volume parameters.

Thirty Gy/3 fractions (fr) SRT was delivered to 34 patients with 75 BMs (median target volume, 3.2 cc), 35 Gy/5 fr to 30 patients with 57 BMs (6.4 cc), 37.5 Gy/5 fr to 28 patients with 47 BMs (20.2 cc), and 40 Gy/10 fr to 20 patients with 37 BMs (24.3 cc), according to protocols, depending on the total target volume (p < 0.001). After excluding the three-fraction groups, the incidence of symptomatic BN was significantly higher in patients with a larger V50EQD2 (adjusted odds ratio: 1.07, p < 0.02), V55EQD2 (1.08, p < 0.01), or V60EQD2 (1.09, p < 0.01) in the remaining five- and ten-fraction groups. The incidence of BN was also significantly higher in cases with V55EQD2 > 30 cc or V60EQD2 > 20 cc (p < 0.05). These doses correspond to 28 or 30 Gy/5 fr and 37 or 40 Gy/10 fr, respectively.

In five- or ten-fraction SRT, larger V55EQD2 or V60EQD2 are BN risk predictors. These biologically high doses may affect BN incidence. Thus, the planning target volume margin should be minimized as much as possible.