Impact of ATM Mutations on Brain Metastasis Control and Radionecrosis in Non-Small Cell Lung Cancer Patients Undergoing SRS

Abstract

<h3>Purpose/Objective(s)</h3> While Ataxia Telangectasia Mutated gene (ATM) variants have been associated with improved local control after radiation, adverse radiation events have been reported following conventional radiotherapy for extracranial tumors. It remains unknown whether ATM variants affect rates of radiation necrosis (RN) and local intracranial progression (LIP) following stereotactic radiosurgery (SRS) for brain metastases. We questioned whether pathologic ATM variants increase rates of RN and/or decrease rates of LIP following SRS for brain metastasis in non-small cell lung cancer (NSCLC) patients. <h3>Materials/Methods</h3> Patients undergoing an initial course of SRS for NSCLC brain metastases between 1/2015 and 12/2020 were retrospectively identified at our institution. RN and LIP were determined by either biopsy or multi-disciplinary serial MRI review. Cumulative incidence of RN and LIP were estimated via the Kaplan Meier method. Cox proportional hazards testing was performed for variant classification (not detected [ND] vs variant of unknown significance [VUS] or pathogenic variant [PV]). A power analysis estimated that 34 events would need to be observed to detect a >4-fold hazard ratio for event. <h3>Results</h3> 541 patients completed SRS for brain metastasis secondary to NSCLC, of which 260 completed molecular profiling. Comprehensive Genomic Profiling (CGP) was performed on cell free circulating tumor DNA (61%), primary tumor (17%), extracranial metastasis (17%), and brain tumor (18%), including ≥2 sites for 33 (13%) patients. Availability of CGP data was significantly associated with more recent year of SRS completion (2018-20, 64% vs 2015-17, 25%; p<0.001), younger age at brain metastasis diagnosis (64.8 years [IQR 57.1-72.8] vs 67.8 years [IQR 59.6-72.8]; p=0.036), no prior tobacco use (22% vs 10%; p<0.001), adenocarcinoma histology (84% vs 73%; p=0.002), and metastatic disease at initial diagnosis (73% vs 63%; p=0.017). ATM variants were identified in 36 cases (13.8%; 13 PV, 23 VUS). For all patients with CPG, RN incidence was 4.9% (95% CI 1.6 – 8.2%) at 6 months and 9.9% (95% CI 4.8 - 15.0%) at 12 months and LIP was 5.4% (95% CI 2.4 – 8.4%) at 6 months and 9.8% (5.5 – 14.1%) at 12 months. For patients with ATM variants, RN incidence was 5.3% (0.0 - 15.3%) at both 6 and 12 months, with 2 total RN events at 5.7 months and 13.9 months following SRS, respectively; LIP was 3.1% (0.0 - 9.1%) at both 6 and 12 months, with one total LIP event at 2.0 months following SRS (P=0.46 and P=0.26 for RN and LIP, respectively). <h3>Conclusion</h3> We did not detect significant differences in RN or LIP risk following SRS for NSCLC brain metastases associated with ATM variants. These results suggest that at this time, ATM status should not inform SRS use when otherwise indicated for NSCLC brain metastases. This dataset additionally allowed us to estimate the number of patients needed to detect a given magnitude of relative risk for RN or LIP associated with a given variant. These data can inform future studies as CGP data accumulate.