1134 - Histologic Classifier of Radiosensitivity to Spine SBRT

C. B. Jackson¹, L. A. Boe², L. Zhang³, A. Apte³, A. Jackson³, L. M. Ruppert¹, J. Haseltine¹, A. Schmitt¹, M. Vaynrub¹, W. C. Newman¹, E. Lis¹, O. Barzilai¹, M. Bilsky¹, Y. Yamada⁴, and D. S. Higginson¹; ¹Memorial Sloan Kettering Cancer Center, New York, NY, ²Department of Epidemiology and Biostatistics, Memorial Sloan Kettering Cancer Center, New York, NY, ³Department of Medical Physics, Memorial Sloan Kettering Cancer Center, New York, NY, ⁴Department of Radiation Oncology, Memorial Sloan Kettering Cancer Center, New York, NY

Purpose/Objective(s): Spine stereotactic body radiation therapy (SBRT) outperforms conventional radiotherapy (RT) in local control. Little is known about the incidence of local failure (LF) and vertebral compression fracture (VCF) with spine SBRT in 3 fractions. Materials/

Methods: This is a retrospective cohort study of 1,502 patients treated to 2,134 lesions between 2014-2023. All patients had no prior overlapping RT and had at least 2 months of MRI follow up. LF was defined as progressive disease on MRI, while VCF was defined as new or progressive fracture on MRI in the absence of LF. LF and VCF rates were compared at the lesion level using Gray’s test with death as a competing risk. To generate a dose-response model, we used competing risks regression to predict tumor control probability (TCP) at 2 years as a function of biological effective dose (BED), assuming an a/ß of 10. Overall survival (OS) was calculated at the patient level using the Kaplan-Meier method.
Results: The median follow up after SBRT was 14 months. Median OS was 20 months (95% confidence interval (CI) 18.3-21.6 months). There were 1,201 lesions (56%) that received 27 Gy and 933 (44%) that received 30 Gy. For lesions treated with 30 Gy vs 27 Gy, actuarial incidences of LF at 1, 2, and 3 years were 8% vs 13%, 11% vs 17%, and 13% vs 19%, respectively (p < 0.001). Cholangiocarcinoma (CCa; n=32), colorectal adenocarcinoma (CRC; n=97), and hepatocellular carcinoma (HCC; n=21) lesions had 1-year LF rates of 27%, 22%, and 30%, respectively, while prostate cancer (PCa; n=255) lesions had a 1-year LF rate of 6%. We therefore defined radiosensitive (to SBRT) lesions as PCa, radioresistant lesions as CCa, HCC, or CRCa, and all other histologies as intermediate. On multivariate regression, radioresistant histology was associated with increased risk of LF (HR 2.63, 95% CI 1.89-3.66, p < 0.001); there was a trend for decreased risk of LF with 30 Gy compared to 27 Gy prescription dose, but this trend did not reach statistical significance (HR 0.60, 95% CI 0.35-1.05, p = 0.076). For radiosensitive, intermediate, and radioresistant histologies, our dose-response model predicts that BED₁₀ values of 60, 72, and 77 Gy, respectively, are necessary to achieve 95% TCP 2 years after SBRT. The incidence of any VCF was 4.6% (95% CI 3.7%-5.5%) 2 years after SBRT; the risk was higher for lesions treated to 30 Gy (5.6%) compared with 27 Gy (3.8%); p = 0.022. The cumulative incidence of VCF requiring surgery or kyphoplasty was 2.5% (95% CI 1.8%-3.2%) 2 years after SBRT, and there was no statistically significant difference between 27 Gy (2.0%) compared to 30 Gy (3.1%); p = 0.2.
Conclusion: With 3-fraction spine SBRT, a prescription dose of 30 Gy provides superior local control compared to 27 Gy, with comparable risk of VCF requiring intervention. Histological radiosensitivity was the most important predictor of LF. For particularly radioresistant histologies such as CRC, CCa, and HCC, our data argue in favor of escalation to a more ablative regimen to maximize local control.