T. Kollipara1, D. Klebaner1, J. A. Jaoude1, A. Pratapneni1, I. C. Gibbs1, E. Pollom1, E. Rahimy1, K. A. Kumar Jr2, M. Mendoza1, A. Persad3, Y. Hori3, D. Park3, S. D. Chang3, K. N. Chou3, and S. G. Soltys1; 1Department of Radiation Oncology, Stanford University School of Medicine, Stanford, CA, 2Department of Radiation Oncology, University of Texas Southwestern Medical Center, Dallas, TX, 3Department of Neurosurgery, Stanford University School of Medicine, Stanford, CA
Purpose/Objective(s): Spine metastases with epidural spinal cord compression (ESCC) present unique challenges when treated with spinal stereotactic radiosurgery (SRS), given the need to balance spinal cord toxicity with local control. We examined our institutional experience with SRS for patients with ESCC. We hypothesized that higher Bilsky grade of ESCC is associated with local failure (LF) following SRS for spine metastases. Materials/
Methods: We performed an IRB-approved retrospective analysis of patients with spinal metastases treated with SRS from 2002 to 2022 with at least one post-treatment imaging. From the pre-treatment MRI scan, the extent of ESCC was classified per Bilsky: 0 (bone only), 1a (epidural impingement without thecal sac deformation), 1b (thecal sac deformation without spinal cord abutment), 1c (thecal sac deformation with spinal cord abutment, no compression), 2 (cord compression with visible CSF), or 3 (cord compression with no visible CSF).For LF analyses, these lesions were grouped into Bilsky Grade 0, Grade 1a-c and Grade 2-3, with Bilsky Grade 0 as the reference. Univariate and multivariate Fine-Gray competing risk regression was conducted to estimate subdistribution hazard ratios (SHR) with 1-year LF as the outcome, death as a competing risk, and Bilsky grade group as the exposure. Multivariate analyses adjusted for gastrointestinal (GI) histology, minimum dose (Dmin) to the GTV, single fraction equivalent dose (SFED, calculated with an a/b of 10), and fractionation (single vs. multi-fraction). Results: We identified 224 patients with 651 spinal metastases. Of these 651 lesions, 436 (67%) had Grade 0 ESCC, 179 (28%) had Grade 1a-c ESCC, and 36 (5%) had Grade 2-3 ESCC. Median SFED was 18.0 (95% CI 16.4-20.0) for Grade 0, 18.0 (95% CI 16.4-20.0) for Grade 1a-c, and 16.4 (95% CI 15.2-17.2) for Grade 2-3 (p<0.001). Higher ESCC grade was associated with increased fractionation of SRS, with 270 (62%) Grade 0 segments receiving single-fraction SRS, compared to 74 (41%) for Grade 1a-c and 13 (36%) for Grade 2-3 (p<0.001). The 1-year cumulative incidence of LF increased with increasing Grade of ESCC: 8% (95% CI 6%-11%) for Grade 0, 13% (95% CI 8%-18%) for Grade 1a-c, and 26% (95% CI 13%-41%) for Grade 2-3 (p<0.001). In the multivariate model, increasing ESCC Grade was associated with 1-year LF, with an estimated two-fold increase in the risk of LF for Grade 1a-c lesions (SHR 2.2, 95% CI 1.2-3.8, p=0.009) and a four-fold increase in the risk of LF for Grade 2-3 lesions (SHR 4.3, 95% CI 1.7-10.8, p=0.002) compared to Grade 0 ESCC. SFED and Dmin had no significant association with LF. Conclusion: Increasing Bilsky Grade of ESCC was associated with increasing rates of LF after spinal SRS. With 1-year LF of 8% for Grade 0 and 13% for Grade 1a-c, SRS provides high rates of tumor control. Further analyses will attempt to identify dosimetric correlates of local progression, particularly in those with Grade 2-3 ESCC at the highest risk of progression.
