Incidence and Risk Factors for Vertebral Compression Fractures From Spine Stereotactic Body Radiation Therapy: Results of a Large Institutional Series

Abstract

There is a paucity of data regarding the incidence of and risk factors associated with the development of vertebral compression fractures (VCF) post-spine stereotactic body radiation therapy (sSBRT). Patients who underwent sSBRT, typically for spine metastases, were included in an IRB-approved database. Gender, consult KPS, epidural disease, thecal sac compression, histology, spine tumor location, VCF, pre-existing VCF, and dosimetric factors (prescription dose, target volume, maximum target dose) were retrospectively collected. Pain relief post-sSBRT was prospectively collected using the Brief Pain Inventory. These factors were correlated to the development of new VCF or the progression of a pre-existing VCF. Survival was calculated using Kaplan-Meier analysis and Cox proportional hazards modeling was utilized to evaluate the association between the clinical factors and the occurrence of VCF. A total of 507 treatments (348 patients) were included in this study. Median KPS, follow-up, and prescription dose were 80 (range, 30-100), 8 months (range, 1-72), and 15 Gy in one fraction (range, 7-18), respectively. Fifty-seven percent of patients were prior smokers whereas only 9% were actively smoking. The reasons for treatment were pain (73%), pain and neurological deficit (13%), asymptomatic (10%), and only neurological deficit (3%). Renal cell carcinoma (24%), lung cancer (14%), and breast cancer (11%) were the most common histologies. Multiple myeloma accounted for 7.3% of treatments. Fifty-nine percent, 32%, and 9% of tumors were located in the cervical/thoracic, lumbar, and sacral spine, respectively. Some patients received prior external beam radiation therapy (26%), sSBRT (4%), surgery (21%) and kyphoplasty (5%). Pre-existing VCF was present in 31% of treatments. Actuarial survival, radiographic control and pain control at 6/12 months were 71%/55%, 74%/64% and 82%/ 75%, respectively. Progression of a pre-existing VCF or development of a new VCF occurred in 15% of treatments. Univariate analysis demonstrated that prior smoking (HR 1.7, p = 0.04), current smoking (HR 1.9, p = 0.03), pain as a reason for treatment (HR 15.4, p < 0.01), lumbar vs sacral spine (HR 4.8, p = 0.03) and pre-existing VCF (HR 5.7, p < 0.01) were predictors of VCF. Multivariate analysis showed that only pre-existing VCF (HR 5.6, p < 0.01) and pain as a reason for treatment (HR 14.9, p < 0.01) were significant predictors of VCF. In our series of over 500 treatments, the incidence of VCF was low (15%). Important predictors of VCF were pre-existing VCF, pain as the reason for treatment, smoking, as well as lumbar spine location. This data shows that the rate of VCF is acceptable at sSBRT doses up to 16 Gy in one fraction.