Purpose/Objective(s):
Stereotactic ablative radiation (SABR) is an effective and increasingly used treatment for primary and metastatic lung tumors. Post-radiation fibrosis can mimic or obscure recurrent tumor on follow-up imaging, complicating surveillance and response assessment. Little is known about clinical factors that determine fibrosis severity after radiation. At our institution, a phase I trial combining dose-de-escalated SABR (12 Gy in 1 fraction) with pulsed electric field (PEF) ablation provided a low-dose cohort to better describe this relationship. Quantitative imaging may capture differences in intensity distribution and texture not apparent by visual review alone. We performed a radiomics-based analysis to differentiate residual tumor from fibrosis and assist with tumor response assessment.
Materials/Methods:
We included a retrospective cohort of patients with primary lung tumors or lung metastases treated with SABR at our institution from 2020 to present, with a minimum of 6 months of follow-up. Regions of interest (ROIs) were contoured using MIM (v7.2.3). Using a lower range-lock threshold of -700 HU, fibrosis was contoured on follow-up scans within the irradiated lung volume by a radiation oncologist. Fibrosis volume was modeled on the log scale; BED was calculated using an a/ß of 10 (BED10). Radiomic analysis was performed using PyRadiomics (v3.1.0) and Python (v3.8). Statistical analysis was performed in RStudio (v4.5.2). Paired comparisons used Wilcoxon signed-rank tests with false discovery rate correction.
Results:
A total of 68 tumors were included in the tumor-level analysis. BED10 ranged from 26.4 to 149.6, and fibrosis volume ranged from 4.0 to 175.9 cc. In univariate linear regression, BED10 (ß=0.01; p=0.021) and age (ß=0.02; p=0.004) were significantly associated with increased fibrosis volume. In multivariable linear regression adjusting for sex, tumor type, tumor centrality, and breath-hold technique, both BED10 and age remained significantly associated with increased fibrosis volume. This corresponds to a 10.5% increase in fibrosis volume per +10 Gy BED10 and a 22% increase per +10 years of age. Radiomic analysis demonstrated significant differences across multiple first-order and texture features. Residual GTV was consistently higher density (70th percentile HU, mean HU, median HU, and 10th percentile HU; p<0.001), while fibrosis demonstrated greater heterogeneity and more complex texture (higher IQR and entropy, and higher GLCM/GLSZM texture metrics; p<0.001).
Conclusion:
Higher BED10 and older age were independently associated with larger post-radiation fibrosis volume, indicating a clinically meaningful dose–fibrosis relationship. Fibrosis demonstrated a distinct imaging phenotype compared with residual tumor in paired radiomics comparisons. These findings support incorporating quantitative imaging to improve post-treatment assessment and estimate recurrence risk in patients treated with lung SABR.
