A Comparison of Local Failure and Necrosis Following Different Radiosurgery Strategies for Large Brain Metastases

Abstract

Large brain metastases (LBMs) are associated with poor local control with single-fraction stereotactic radiosurgery (SRS) alone. Various alternative strategies have been developed, including fractionated SRS (FSRS) and staged SRS (SSRS) for intact LBMs, and resection with postoperative-SRS (postop-SRS) or preoperative-SRS (preop-SRS) for operable LBMs. The objective of this study is to compare local failure (LF) and radiation necrosis (RN) outcomes among these four management strategies to determine the optimal treatment paradigm. Consecutive patients diagnosed with LBM (≥2 cm in maximum dimension) between July 2017 and January 2022 and treated with one of the aforementioned strategies at a single tertiary institution were evaluated. All immobilization, target contouring, margins, dose- and prescription selection followed pre-defined institutional guidelines. Primary endpoints included LF, symptomatic RN, or a composite endpoint of these two variables. Gray's test was used to compare the cumulative incidence of the LF and the composite endpoint, with death as a competing risk. A total of 234 LBMs in 188 consecutive patients met the inclusion criteria. The median age was 65 years (range: 31-98), the median KPS was 80 (range: 50-100), and 58% were female. The most common primary tumors were lung (48%) and breast cancer (17%). The median maximum tumor diameter was 3.0 cm (range: 2.0-5.6). 47 (20%) lesions were treated with FSRS, 66 (28%) with SSRS, 74 (32%) with postop-SRS, and 47 (20%) with preop-SRS. With a median follow-up of 12 months, 22 (9%) LF and 11 (5%) RN events occurred. The 6-month and 1-year cumulative incidences of LF for the entire cohort were 5% (95% CI: 3%-9%) and 8% (95% CI: 5%-12%), respectively. The 6-month and 1-year LF rates were 4% (95% CI: 1%-13%) and 8% (95% CI: 3%-20%) for FSRS; 8% (95% CI: 3%-20%) and 8% (95% CI: 3%-20%) for SSRS; 7% (95% CI: 3%-15%) and 8% (95% CI: 3%-16%) for postop-SRS; 0 and 7% (95% CI: 2%-20%) for preop-SRS (p>0.05). The 1-year OS rates were favorable in resected patients (61% for postop-SRS and 82% for preop-SRS) compared to SRS alone strategies (45% for FSRS and 56% for SSRS) (p = 0.004). Similarly, RN events were significantly lower in resected patients treated with either bimodality approach (0 for postop-SRS and 4% for preop-SRS) than SRS standalone strategies (9% for FSRS and 8% SSRS) (p = 0.024). At 12 months, the cumulative probabilities of the composite endpoint were 13% (95% CI: 5%-25%) for FSRS, 15% (95% CI: 7%-25%) for SSRS, 9% (95% CI: 3%-17%) for postop-SRS, and 12% (95% CI: 4%-24%) for preop-SRS and not significantly different between the groups. For medically operable patients with surgically resectable LBMs, a strategy of surgery and SRS, regardless of timing, is associated with favorable local control and reduced risk for RN. For unresected patients, either SSRS or FSRS is associated with similar local control, but slightly higher RN risk. Prospective comparative evaluation is warranted.