Predictors of Local Failure and Leptomeningeal Disease After Adjuvant Fractionated Stereotactic Radiosurgery for Resected Brain Metastases

Abstract

Adjuvant stereotactic radiosurgery (SRS) is widely used for resected brain metastases, with improved local control compared to observation and decreased risk of neurocognitive side effects compared to whole brain radiation. However, local failure and leptomeningeal disease (LMD) remain significant sources of morbidity for these patients. We sought to identify clinical and dosimetric factors predictive of local failure and LMD following adjuvant SRS.A total of 65 consecutive patients with one resected brain metastasis who received Linac-based SRS in 3 to 5 fractions to the tumor bed from Oct 2016 to Feb 2020 at a single institution were included in a retrospective analysis. Local failure and LMD were calculated by cumulative incidence with death as a competing risk. Univariate and multivariate analysis were done using the Fine and Gray method; variables with p-value ≤0.10 on univariate analysis were included in the multivariate model.Median follow up was 11.5 months and median overall survival was 13.1 months from completion of SRS. Of the 65 resected brain metastases, 35 were from non-small cell lung cancer, 9 were from breast, 4 from colon, 4 from renal cell carcinoma, 4 from melanoma, 4 from esophagus, and 1 each from sarcoma, bladder, oropharynx, ovarian, and prostate. The median SRS prescription dose was 30 Gy in 5 fractions, with median biologically effective dose (BED10) of 48 (range 35.7-51.3). A total of 13 patients (20%) had local failure, and 15 patients (23%) had LMD. Four patients had both local failure and LMD. For local failure risk, there was a trend towards a significant association with increasing planning target volume (PTV) measured in cm3 (HR 1.024, P = 0.056) on univariate analysis. Increasing number of synchronous intact brain lesions (P = 0.0013), and near total or subtotal resection (P = 0.035) were significantly associated with increased risk of LMD on univariate analysis. Breast primary histology (P = 0.089), BED10 maximum dose (P = 0.075), and BED10 mean dose (P = 0.071) trended towards significance on univariate analysis for LMD risk and were also included in the multivariate model. On multivariate analysis, only increasing number of synchronous intact brain lesions (HR 1.502 [95% CI 1.078-2.074], P = 0.014), and near total or subtotal resection (HR 4.605 [95% CI 0.913-6.713], P = 0.035) remained independent predictors of LMD. BED10 prescription dose, maximum dose, mean dose, minimum dose in the PTV, and PTV coverage with prescription dose were not significant predictors of local failure or LMD.Increasing number of synchronous intact brain lesions and less than gross total resection were independent risk factors for development of LMD following adjuvant fractionated SRS for resected brain metastases. Multiple SRS dosimetric factors, including BED10 prescription dose, maximum dose, mean dose, minimum dose in the PTV, and PTV coverage with prescription dose, were not significant predictors of local failure or LMD.K.V. Chaung: None. M.Z. Kharouta: None. A.J. Gross: None. G.C. Pereira: None. A.M. Kumar: Stock; AbbVie. D.B. Mansur: None. A.E. Sloan: None. T.R. Hodges: None. S. Choi: None.