A Volumetric Dosimetry Analysis of Vertebral Body Fracture Risk after Spine Stereotactic Radiosurgery

Abstract

Vertebral compression fractures (VCF) are a complication of spine stereotactic radiosurgery (sSRS). It is currently unclear how the heterogeneity of radiation dose delivered to a vertebral body affects the occurrence of VCF. In this study, we sought to analyze how volumetric dosimetry and clinical factors were associated with the development of VCF. We evaluated 173 spinal segments undergoing single fraction sSRS in 85 patients from a retrospective database at a single tertiary care institution. Vertebral bodies were contoured and detailed dosimetric variables were collected. Competing risk models were used to evaluate the effect of clinical and dosimetry variables on the risk of VCF. Sensitivity analysis was performed to obtain cutoffs of 90% sensitivity. Our primary endpoint was development of post-sSRS VCF. Fractures were noted in 21/173 vertebrae (12.1%); of these, 12 were de novo and 9 were progressive. The median time to fracture was 11 months. Median follow up time was 14.2 months. In total, 69 vertebral bodies were lumbar, 82 were thoracic, and 22 were cervical. The median prescribed dose for both groups was 18 Gy (IQR 16-18). On multivariable analysis, the percentage of vertebral body volume receiving >20 Gy (HR 1.04, p=0.03) and >24 Gy (HR 1.10, p=0.04) were significantly associated with increased risk of VCF. No other patient or dosimetric variables were found to be significant on multivariable analysis. Sensitivity analysis revealed that the percentages of vertebral body volume receiving >20 Gy and >24 Gy required to obtain 90% sensitivity for predicting vertebral body fracture are 24% and 0%, respectively. sSRS treatment plans that permit increased doses of >20 Gy and >24 Gy to larger vertebral body volumes are associated with increased risk of VCF. In order to minimize the risk of post-sSRS VCF, the percentage of vertebral body volume receiving >20 Gy and >24 Gy should be less than 24% and 0%, respectively. These results may help guide clinicians in treatment planning for patients at high risk for VCF following sSRS.