Local control after stereotactic radiation for thyroid cancer: a histology risk-stratified analysis.

Abstract

Purpose/objectives: Recurrent and metastatic non-anaplastic thyroid cancer has traditionally been considered relatively radioresistant. We evaluated whether anatomic site or histology risk group better predicts local control (LC) following stereotactic radiation therapy (SBRT/SRS).

Materials/methods: We retrospectively reviewed 32 patients with 66 treated lesions (42 skeletal, 24 non-skeletal) treated between 2012 and 2025. Histology was dichotomized into low-risk (papillary, follicular, medullary, Hürthle cell; n=61) and high-risk (poorly differentiated and squamous cell carcinoma; n=5). BED₁₀ was calculated using α/β = 10 Gy. LC was defined per RECIST 1.1. Kaplan-Meier and log-rank tests were used. Intra-patient correlation was addressed via generalized estimating equations (GEE) and Cox shared frailty models. A multivariable GEE model included histology risk, site, BED₁₀, PTV volume, prior surgery, and age.

Results: Median follow-up was 15.0 months. Six local failures occurred (crude LC 90.9%); 1- and 2-year LC were 96.5% and 87.1%. Histology risk was the dominant determinant: low-risk histology achieved 95.1% crude LC versus 40.0% for high-risk (log-rank p=0.002). In multivariable GEE analysis, low-risk histology was independently associated with LC (OR 158.0, 95% CI 1.72-14,479.7; p=0.028), whereas skeletal site was not (OR 1.07, 95% CI 0.02-51.8; p=0.974). BED₁₀ was not associated with LC after adjustment (p=0.752). Skeletal and non-skeletal LC were similar after histology stratification (95.0% vs 95.2% among low-risk). Treatment was well tolerated; no Grade ≥3 radiation toxicity, myelopathy, or fractures occurred.

Conclusion: In stereotactic radiation for metastatic non-anaplastic thyroid cancer, histology risk grouping-not anatomic site or dose escalation-is the dominant predictor of local control. These findings support histology-based risk stratification for SRT patient selection.