Changing practice patterns of Gamma Knife versus linear accelerator-based stereotactic radiosurgery for brain metastases in the US

OBJECTIVE Single-fraction stereotactic radiosurgery (SRS) is a crucial component in the management of limited brain metastases from non small cell lung cancer (NSCLC). Intracranial SRS has traditionally been delivered using a frame based Gamma Knife (GK) platform, but stereotactic modifications to the linear accelerator (LINAC) have made an alternative approach possible. In the absence of definitive prospective trials comparing the efficacy and toxicities of treatment between the 2 techniques, nonclinical factors (such as technology accessibility, costs, and efficiency) may play a larger role in determining which radiosurgery system a facility may choose to install. To the authors' knowledge, this study is the first to investigate national patterns of GK SRS versus LINAC SRS use and to determine which factors may be associated with the adoption of these radiosurgery systems. METHODS The National Cancer Data Base was used.to identify patients > 18 years old with NSCLC who were treated with single-fraction SRS to the brain between 2003 and 2011. Patients who received "SRS not otherwise specified" or who did not receive a radiotherapy dose within the range of 12-24 Gy were excluded to reduce the potential for misclassification. The chi-square test, t-test, and multivariable logistic regression analysis were used to compare potential demographic, clinicopathologic, and health care system predictors of GK versus LINAC SRS use, when appropriate. RESULTS This study included 1780 patients, among whom 1371 (77.0%) received GK SRS and 409 (23.0%) underwent LINAC SRS. Over time, the proportion of patients undergoing LINAC SRS steadily increased, from 3.2% in 2003 to 30.8% in 2011 (p < 0.001). LINAC SRS was adopted more rapidly by community versus academic facilities (overall 29.2% vs 17.2%, p < 0.001). On multivariable analysis, 4 independent predictors of increased LINAC SRS use emerged, including year of diagnosis in 2008-2011 versus 2003-2007 (adjusted OR [AOR] 2.04, 95% CI 1.52-2.73, p < 0.001), community versus academic facility type (AOR 2.04, 95% CI 1.60-2.60, p < 0.001), non-West versus West geographic location (AOR 4.50, 95% CI 2.87-7.09, p < 0.001), and distance from cancer reporting facility of < 20 versus >= 20 miles (AOR 1.57, 95% CI 1.21-2.04, p = 0.001). CONCLUSIONS GK remains the most commonly used single-fraction SRS modality for NSCLC brain metastases in the US. However, LINAC-based SRS has been rapidly disseminating in the past decade, especially in the community setting. Wide geographic variation persists in the distribution of GK and LINAC SRS cases. Further comparative effectiveness research will be needed to evaluate the impact of these shifts on SRS-related toxicities, local control, and survival, as well as treatment costs and efficiency.