357 - Validation of a Predictive Score for Vertebral Compression Fracture after Stereotactic Body Radiation Therapy for Spinal Metastases in an International Multi-Institutional Cohort

09:35am - 09:45am ET

Room 160

Presenter(s)

Brianna Galvin, MD - Mayo Clinic Rochester, Rochester, MN

B. C. Galvin¹, K. Dibs², L. Zeng³, J. Lucido¹, L. Burgess³, G. M. Spears⁴, J. N. Hobson¹, P. D. Brown¹, D. M. Blakaj², K. W. Merrell¹, J. D. Palmer⁵, A. Sahgal³, and R. O. Kowalchuk¹; ¹Department of Radiation Oncology, Mayo Clinic, Rochester, MN, ²Department of Radiation Oncology, The Ohio State University Wexner Medical Center, Columbus, OH, ³Department of Radiation Oncology, Odette Cancer Centre, Sunnybrook Health Sciences Centre, University of Toronto, Toronto, ON, Canada, ⁴Division of Biomedical Statistics and Informatics, Mayo Clinic, Rochester, MN, ⁵Department of Radiation Oncology, James Cancer Hospital/Wexner Medical Center, The Ohio State University, Columbus, OH

Purpose/Objective(s):

Vertebral compression fracture (VCF) is a common and significant adverse event following stereotactic body radiation therapy (SBRT) for spinal metastases, affecting 15-20% of patients. Our institution previously developed a risk stratification model to predict VCF after SBRT. We performed model validation using a large cohort from two outside institutions with varying fractionation regimens.

Materials/Methods:

We reviewed patients treated with spinal SBRT between 2009 and 2021 at one institution and between 2013 to 2019 at another. Cohorts were analyzed separately. The risk model was applied using one point for lumbar location, SINS >6, and epidural extension. While GTV volume was part of the original model, it was excluded from validation due to heterogeneous practice patterns regarding contouring approach, however, CTV abided by consensus guidelines. Patients were stratified into very low (0 points), low (1 point), intermediate (2 points), and high (3 points) risk groups. The previously published predictive model reported a 2-year cumulative incidence of VCF of 6.7%, 17%, 35.4%, in the low (0-1 point), intermediate (2 points), and high (3 points) risk groups, with treatments primarily involving single fraction. Fine and Gray subdistribution hazard modeling was used to estimate hazard ratios for VCF by risk level while accounting for death as a competing risk, and cumulative incidence functions were calculated. Analysis was conducted using R, version 4.4.1.

Results:

The two cohorts included 1419 spinal segments (median follow-up 30.7 months, 95% CI: 28.8-32.5; 17.2% VCF rate) at one institution primarily treated with 2 fractions and 173 segments (median follow-up 43.4 months, 95% CI: 38.9-49.3; 17.3% VCF rate) at another institution primary treated with 3 fractions. In the first group, there were 549, 523, 282, and 65 spinal segments in the very low, low, intermediate, and high-risk groups respectively, resulting in a cumulative incidence of VCF at 2 years of 9.2% (95% CI 6.5-11.8%), 16.7% (13.2-20.3%), 25.9% (20.0-31.7%), and 34.1% (21.2-47.0%). In the second group there were 53, 69, 42, and 9 spinal segments in the very low, low, intermediate, and high risk groups respectively, resulting in a cumulative incidence of VCF at 2 years of 4.2% ( 95% CI 0.0-9.9%), 10.9% (3.2-18.7%), 31.1% (16.8-45.4%), and 64.4% (0.0-100.0%). The hazard ratio for VCF per 1 point increase in risk score was 1.63 (95% CI 1.42-1.86, p<0.0001) and 2.38 (95% CI 1.67-3.39, p<0.001) in the first and second groups.

Conclusion:

Multi-institutional validation reveals robust model performance across one, two, and three fraction regimens. Patients at high risk for VCF are consistently identified using factors available prior to SBRT delivery. We propose this model could stratify patients for prophylactic measures to reduce VCF risk.