An Efficient and Objective Method for Identifying Radiation-Related Cardiac Toxicity after Definitive Conventional Radiation for Locally Advanced Non-Small Cell Lung Cancer

Document Type


Publication Date


Publication Title

International Journal of Radiation Oncology, Biology, Physics


Purpose/Objective(s): Reported series associating heart dose and mortality or toxicity are limited by small patient numbers or absence of detailed cardiac event data. Traditional cardiac toxicity monitoring involves patient or physician reported outcomes at the time of follow up or manual chart review to identify events. This methodology is limited, inherently subjective, and potentially introduces bias. Here, we report the largest series using an objective method for determining RT-related cardiac toxicity by searching the electronic medical record system EPIC for relevant ICD-10 codes. Materials/Methods: With IRB approval, we identified locally advanced lung cancer patients treated with definitive RT +/- chemotherapy at our institution between 2009-2019 and searched their EPIC records for 100 cardiac-related ICD-10 codes and classified them as pre- and post-RT events. Through the patient data portal, we searched all patients for the ICD10 codes of interest at once. Results were then manually verified. Correlation of dosimetric parameters and cardiac toxicity was examined using univariate (UVA), and multivariate subdistribution hazard analysis with death as a competing risk. Results: 396 patients with available dosimetric data were analyzed with median follow-up of 20 months. 98 patients (25%) had at least 1 post-RT cardiac event, including, but not limited to, ischemic, dysrhythmic, and heart-failure related events. Manual verification of the electronically queried events showed 91% accuracy of the employed methodology. Inaccuracy resulted from erroneous ICD-10 code recording for a patient encounter. Median time to first event was 13 months (range 0.2-111). Dosimetric analysis of mean and max heart dose, and volume receiving 5- 70 Gy (in 5 Gy increments) identified mean heart dose, and V45-V60 as the most significant predictors of cardiac toxicity and mortality. MVA accounting for use of chemotherapy, prior cardiac events, age, smoking status, and Charlson Comorbidity Index showed an increase of 3-5% in cumulative incidence of cardiac toxicity and overall mortality per 1 Gy of mean heart dose or additional %volume receiving 45-60 Gy (all statistically significant). ROC analysis suggested a target mean heart dose 15 Gy to minimize toxicity. Conclusion: Direct EMR query allowed for efficient, objective and accurate characterization of cardiac toxicity after thoracic RT and may provide a more comprehensive method for identifying RT-related toxicity compared to traditional means of data collection. Mean heart dose and V45-V60 were the




3 Supplement

First Page


Last Page