Tumor control probability modeling of pulmonary metastases in oligometastatic head and neck squamous cell carcinoma treated with stereotactic body radiation therapy

BACKGROUND AND PURPOSE: Local control rates for pulmonary metastases from head and neck squamous cell carcinoma (HNSCC) treated with stereotactic body radiation therapy (SBRT) vary substantially in the literature, likely due to differences in the doses applied. This study, therefore, aims to develop tumor control probability (TCP) models.

METHODS: Dose-time modeling of TCP was performed for three key dose descriptors: 3-fraction equivalent dose (3fxED) prescription dose, 3fxED planning target volume (PTV) D95%, and 3fxED PTV D2%. Each dose descriptor was stratified into 4 dose groups using a k-medians clustering algorithm, and the Kaplan-Meier estimator was computed in each group separately to enable time-dependent dose-response modeling. Logistic regression models were fitted using maximum likelihood estimation with model parameters determined separately for 1-year and 2-year local control.

RESULTS: This retrospective multicenter analysis included 318 pulmonary metastases from 215 patients. The median 3fxED prescription dose was 45.0 Gy (IQR, 39.2-46.5), corresponding to a biologically effective dose (α/β = 20 Gy) (BED20) of 78.8  Gy (IQR, 64.8-82.5). After a median radiographic follow-up of 13.3 months (IQR, 6.6-32.0), 21 local failures were observed. All models indicated a dose-dependent effect on the local control probability. A local control rate of 95 % was achieved at 1 and 2 years when 3fxED prescription doses and PTV D95% exceeded about 51 Gy (94 Gy BED20), and when PTV D2% reached 66 Gy (139 Gy BED20).

CONCLUSION: These TCP models suggest a dose-dependent probability of local control in pulmonary HNSCC metastases treated with SBRT and provide a basis for future clinical assessments in this population.