3338 - Left Ventricular Sphericity as a Radiomic Predictor of Cardiac Outcomes after Thoracic Radiotherapy for Lung Cancer

M. Bakhtiar¹, C. E. Kehayias², C. V. Guthier², J. He², R. H. Mak², and K. M. Atkins³; ¹Harvard Radiation Oncology Program, Boston, MA, ²Department of Radiation Oncology, Dana-Farber Cancer Institute/Brigham and Women’s Hospital, Harvard Medical School, Boston, MA, ³Department of Radiation Oncology, Cedars-Sinai Medical Center, Los Angeles, CA

Purpose/Objective(s): Pathophysiologic cardiac remodeling manifests as a rounded left ventricle (LV), quantified by LV sphericity. LV sphericity has been established as an independent predictor for risk of cardiac events, including heart failure/cardiomyopathy (HF/CM) in the general population. Given the baseline cardiovascular (CV) risk among patients with lung cancer, we sought to measure LV sphericity and its association with baseline CV disease and post-radiotherapy (RT) cardiac events in patients who received RT for lung cancer. Materials/

Methods: We retrospectively analyzed 698 patients with non-small cell lung cancer (NSCLC) treated with RT between 2001-2014. Sphericity was calculated using open source software on LV structures generated using a deep learning auto-segmentation algorithm. Post-RT major adverse cardiac events (MACE; unstable angina, myocardial infarction [MI], heart failure [HF], coronary re-vascularization, or cardiac death) and grade =3 cardiac common terminology criteria for adverse events (CTCAE), including HF/CM, were collected. Descriptive statistics were performed to compare sphericity when patients were grouped by CV disease history. We performed area under the receiver operating characteristic curve (AUC) analysis and Fine & Gray regressions (non-cardiac death as a competing risk) to test the association between baseline sphericity and time to cardiac events.
Results: The median LV sphericity was 0.75 (interquartile range [IQR] 0.71-0.79) on RT planning CT scans. There was a trend toward difference in baseline sphericity values between patients when grouped by baseline coronary heart disease (CHD, p=0.16) or MI (p=0.12), but not congestive heart failure or arrhythmia (p>0.05). Sphericity modestly predicted development of grade =3 HF/CM (AUC 0.53) in all patients, and predicted HF/CM to a greater degree in those with a history of MI (AUC 0.61). The median follow-up was 19 months (IQR 7-45). Adjusting for age, history of hypertension, diabetes, arrhythmia, CHD, and RT dose to the LV myocardium, there was an association between baseline sphericity (=0.80, upper-quartile) and time to grade =3 HF/CM event, with a trend towards significance (adjusted hazard ratio [aHR] 1.50, 95% CI 0.85-2.67; p=0.16), but no association with risk of overall MACE, coronary events, or arrhythmias (p>0.05). Among patients with a history of MI (n=86) prior to RT, adjusting for age, hypertension, and RT dose to the LV myocardium, baseline sphericity =0.80 was associated with a significantly increased risk of grade =3 HF/CM (aHR 3.72, 95% CI 1.32-10.48; p=0.013).
Conclusion: Among patients with NSCLC who received thoracic RT, high LV sphericity was associated with baseline CHD or MI, and significantly associated with time to HF/CM in those with a history of MI. LV sphericity warrants further evaluation in larger studies as an automated radiomic data point to assess baseline CV risk and cardiac outcomes in patients receiving thoracic RT.