Novel Radiomics Model May Predict Cardiac Outcomes After Radiotherapy for Lung Cancer

By: Joshua D. Madera, MD
Posted: Wednesday, May 22, 2024

The use of 18F-fluorodeoxyglucose (FDG) PET/CT imaging may accurately predict the extent of cardiotoxicity in patients receiving radiotherapy for lung cancer, according to a study published in JCO Clinical Cancer Informatics. This cardiac radiomic model may additionally benefit patients by identifying existing cardiac conditions and predicting which patients may be more likely to experience cardiac-related complications after radiotherapy, suggested Wookjin Choi, PhD, of Sidney Kimmel Medical College at Thomas Jefferson University, Philadelphia, and colleagues.

“This study is the first to present the concept of functional cardiac radiomics and provide seminal data for hypothesis-generating research using an automated cardiac uptake functional biomarker to predict clinical outcomes,” the study authors stated.

From 2020 to 2022, 18F-FDG PET/CT scans were collected from a total of 209 patients with lung cancer before they received treatment. A single clinician analyzed all the scans and classified them according to the clinical cardiac guidelines. Scans were characterized as demonstrating no uptake, diffuse uptake, or focal uptake. A total of 210 functional radiomics characteristics were used when characterizing the uptake observed in the scans. The data were further stratified into training and validation sets before the classification models were constructed.

Based on the clinician’s analysis of the scans, 39.6% demonstrated no evidence of uptake, 25.3% showed diffuse uptake, and 35.1% showed focal uptake. In addition, the study authors identified nine clinically pertinent radiologic features on the scans. Furthermore, of the models they constructed, the most effective one demonstrated a predictive accuracy of 93% in the training data set. Moreover, this model showed predictive accuracies of 80% and 92% in two external validation data sets.

Disclosure: For full disclosures of the study authors, visit ascopubs.org.

JCO Clinical Cancer Informatics