Program Information
A Novel CT Imaging Biomarker to Quantify Radiation Injury in the Esophagus with Application to Outcome Assessment
J Niedzielski1,2*, J Yang2 , F Stingo3 , Z Liao2 , D Gomez2 , R Mohan2 , M Martel2 , T Briere2 , L Court2 , (1) University of Colorado School of Medicine, Aurora, CO, (2) UT MD Anderson Cancer Center, Houston, TX, (3) University of Florence, Florence, Italy
Presentations
WE-G-FS1-2 (Wednesday, August 2, 2017) 4:30 PM - 6:00 PM Room: Four Seasons 1
Purpose: Uncertainty in quantifying esophageal toxicity with subjective endpoints reduces the understanding of dose-response. We developed an imaging biomarker of esophageal radiation-response and investigated this for outcome assessment with multiple applications.
Methods: An in-vivo imaging biomarker of esophageal radiation-response, esophageal-expansion, was investigated in 134 patients with NSCLC. A comprehensive computational framework was created and validated to quantify esophageal-expansion by utilizing DIR of CT images between the RT plan and weekly 4DCTs during radiotherapy. After validation, esophageal-expansion was used for RT outcome assessment in three distinct applications: (1) An objective endpoint in toxicity prediction modelling;(2) Pre-emptive toxicity detection during the initial weeks of radiotherapy;(3) Objective quantification of radiosensitivity for use in clinical trials.Application (1) compared predictive performance of lasso esophagitis prediction models using esophageal-expansion versus CTCAE grade as the complication endpoint; (2) utilized patient’s esophageal-expansion in the first 3 weeks of treatment to predict esophageal toxicity severity by completion of radiotherapy; and (3) utilized k-means clustering on expansion dose-response to group patients in differing subpopulations of radiosensitivity and used this information in toxicity prediction models.
Results: Biomarker was validated with a mathematical phantom and patient data (AUC=0.87, for grade3). Application results: (1) Esophageal-expansion endpoint-based models outperformed grade endpoint-based models using lasso (AUC=0.72 vs 0.64, respectively). (2) Biomarker value in initial weeks of RT showed strong pre-emptive detection of toxicity by treatment completion (AUC=0.82, week3). (3) Radiation-response was highly variable between patients. K-means optimized patient radiosensitivity subgroups to three clusters, and cluster membership improved the predictive performance of toxicity models (Scaled Brier=12.1% vs 18.2%, with radiosensitivity information vs without).
Conclusion: Esophageal-expansion was validated as a quantitative imaging biomarker of radiation-response and esophageal toxicity. Esophageal-expansion is an objective endpoint that improves toxicity prediction modelling, pre-emptively detects esophageal radiation injury, and objectively quantifies patient's inherent radiation sensitivity.
Contact Email: