The following is a summary of “Quantitative analysis of interstitial lung abnormalities on computed tomography to predict symptomatic radiation pneumonitis after lung stereotactic body radiotherapy,” published in the June 2024 issue of Oncology by Yoneyama et al.
Symptomatic radiation pneumonitis (SRP) represents a significant complication following thoracic stereotactic body radiotherapy (SBRT), necessitating improved predictive tools beyond conventional visual assessments. AI-based quantitative computed tomography image analysis software (AIQCT) offers a promising avenue to enhance SRP risk prediction by analyzing high-resolution computed tomography (HRCT) images. This study aimed to leverage AIQCT to develop a predictive model for SRP among patients undergoing SBRT for stage I lung cancer. AIQCT automatically segmented HRCT images to quantify features such as reticulation + honeycombing (Ret + HC), consolidation + ground-glass opacities, bronchi (Br), and normal lung tissue (NL), associating these metrics with SRP incidence. Through recursive partitioning analysis (RPA) on a training cohort, distinct risk groups were delineated based on NL-Dmean and Br-volume: high-risk (NL-Dmean ≥ 6.6 Gy), intermediate-risk (NL-Dmean < 6.6 Gy and Br-volume ≥ 2.5 %), and low-risk (NL-Dmean < 6.6 Gy and Br-volume < 2.5 %).
This stratification revealed varying SRP incidences in the training cohort (62.5%, 38.4%, and 7.5%, respectively) and was validated in an independent testing cohort (50.0%, 27.3%, and 5.0%, respectively). These findings underscore AIQCT’s capability to identify critical CT biomarkers associated with SRP, offering a robust predictive framework for personalized risk assessment in SBRT planning. Integration of such predictive models into clinical practice holds the potential to optimize treatment strategies, mitigate SRP risk, and ultimately improve outcomes for patients undergoing SBRT. Prospective studies are warranted to further validate and refine these findings, aiming toward routine implementation of AIQCT-based predictive tools in radiotherapy oncology.
Source: sciencedirect.com/science/article/abs/pii/S0167814024006789
