2201 - Real-World Patterns of Cranial Radiation Therapy Among Patients with Brain Metastases from Non-Small Cell Lung Cancer in the Era of CNS-Penetrant Tyrosine Kinase Inhibitors

L. L. Zhu¹, S. Waliany², F. K. Keane^3,4, H. Willers^3,4, D. E. Soto⁵, A. E. Marciscano³, J. Gainor², Z. Piotrowska², H. A. Shih^4,6, J. Lin², and M. J. Khandekar^3,4; ¹Harvard Radiation Oncology Program, Boston, MA, ²Massachusetts General Hospital Cancer Center, Boston, MA, ³Department of Radiation Oncology, Massachusetts General Hospital, Boston, MA, ⁴Harvard Medical School, Boston, MA, ⁵Massachusetts General Hospital North Shore Cancer Center, Danvers, MA, ⁶Department of Radiation Oncology, Massachusetts General Hospital and Harvard Medical School, Boston, MA

Purpose/Objective(s): Treatment of brain metastases (BMs) in non-small cell lung cancer (NSCLC) has been transformed by the use of CNS-penetrant tyrosine kinase inhibitors (TKIs) targeting common oncogenic driver alterations. While stereotactic radiosurgery (SRS) and whole brain radiotherapy (WBRT) are often deferred to preserve quality of life, limited real-world data exist regarding upfront and salvage radiation in this setting. Materials/

Methods: We conducted an IRB-approved retrospective study of consecutive patients with NSCLC BMs who received their first course of single fraction SRS at our institution between 1/1/2017 and 12/31/2021. Patients were classified into two cohorts: 1) those with targetable alterations treated with CNS-active TKIs and 2) those without targetable alterations. Date of BM diagnosis, lesions treated, and patterns of failure were extracted from medical records, imaging, and treatment plans. Regression analyses were used to assess the relationship between oncogene status, use of SRS and WBRT, and presence of leptomeningeal disease (LMD).

Results: We identified 158 patients with 353 treated BMs, with 58 patients [118 lesions] in Group 1, and 100 patients [235 lesions] in Group 2. Median age was 64 years; 82% of patients had KPS>70. At time of BM diagnosis, 38% of patients had a single BM, 41% had 2-5, and 20% had >6. The patients in Group 1 had the following oncogene drivers: EGFR (n=32), ALK (n=16), ROS1 (n=8), RET (n=2.) Median follow-up was 18 months for Group 1 and 8 months for Group 2 (p=0.03). There was no difference in the use of upfront WBRT (Group 1: 3/58 (5%), Group 2: 12/100 (12%), p=0.17). Of patients not initially treated with WBRT, Group 1 patients received the first course of SRS at a median of 224±74 days from BM diagnosis compared to a median of 34±17 days in Group 2 pts (p=1.9x10^-7). Lesions treated in Group 1 patients were significantly more likely to have been present at BM diagnosis (Group 1: 74/118 (65%), Group 2: 118/235 (50%), p=0.029). There was no difference in rates of salvage WBRT following SRS or time to WBRT between the 2 groups (Group 1: 8/58 (13.8%) 5.2 months, Group 2: 11/100 (11%), 8.0 months, p=0.60 and p=0.71). Even after controlling for median follow up, there was a trend toward higher rates of LMD among Group 1 patients (10/58 (17%) vs. 6/100 (6%), p=0.054).

Conclusion: To the best of our knowledge, this study represents one of the largest to date evaluating patterns of brain RT in patients with NSCLC BMs in the era of CNS-penetrant TKIs. While these agents significantly delay time to initial SRS compared to patients without targetable alterations, 65% of lesions needing SRS were present at initial BM diagnosis, suggesting that local progression remains a concern. There was no difference in the use of salvage WBRT, despite higher rates of LMD in Group 1 patients. Identification of factors predictive of CNS progression on TKI may enable optimization of the timing and integration of local therapy in conjunction with CNS-active agents.