Immune Checkpoint Inhibition and Single Fraction Stereotactic Radiosurgery in Non-Small Cell Lung Cancer Brain Metastases: An International Multicenter Study of 395 Patients

Abstract

Brain metastases most commonly arise from non-small cell lung cancer (NSCLC). In recent years, immune checkpoint inhibitors (ICI) have demonstrated improvements in overall survival (OS) in NSCLC. However, concerns remain about the risk of radiation necrosis (RN) when ICI are administered with stereotactic radiosurgery (SRS). Logistic regression was used to evaluate prognostic factors associated with the development of any grade RN and symptomatic RN. Cumulative incidence of RN was evaluated using competing risks analysis and the Fine and Gray model, where the null hypothesis was rejected for p < 0.05. There were 395 patients with 2,513 brain metastases treated across 11 international institutions included in the analysis. The median follow-up was 14.2 months. Median patient age was 67 years (Interquartile Range [IQR]: 61-73), 53.4% were male, the median Karnofsky Performance Status was 80 (IQR: 80-90), and 88.6% has active extracranial disease at the time of SRS. The median margin dose was 19 Gy (IQR: 18-20), 97.5% of patients were treated on the Gamma Knife ®, 3.8% underwent prior whole brain radiation therapy (WBRT). The median V12 Gy was 5.2 cm3 and 36.5% of patients had a V12 Gy ≥ 10 cm3, anti-PD-1 agents were administered in 91.6% of patients. A V12 Gy ³ 10 cm3 was associated with an increased risk of developing any grade RN; odds ratio (OR): 2.12, p = 0.04 and OR: 2.18; p = 0.03 on univariable and multivariable analysis, respectively. Similarly, a V12 Gy ≥ 10 cm3 was associated with an increased risk of developing symptomatic RN; OR: 3.80, p = 0.003 and OR: 3.95; p = 0.003 on univariable and multivariable analysis, respectively. Receipt of concurrent ICI and prior WBRT were not statistically significant. At 1-year, the cumulative incidence of any grade and symptomatic RN was 4.8% and 3.8%, respectively. The cumulative incidence of any grade RN was 3.8% vs. 5.3% for the concurrent and non-concurrent groups at 1-year, respectively (p = 0.35). The cumulative incidence of symptomatic RN was 3.8% vs. 3.6% for the concurrent and non-concurrent groups at 1-year, respectively (p = 0.95). The risk of any grade and symptomatic RN following SRS and ICI administration for NSCLC brain metastases increases as the V12 Gy exceeds 10 cm3. Concurrent ICI and SRS does not appear to increase this risk. Radiosurgical planning techniques should aim to minimize the V12 Gy.