Investigating the Impact of Coronary Artery Dosimetry on Major Adverse Cardiac Events after Thoracic Radiation Therapy

Abstract

<h3>Purpose/Objective(s)</h3> Minimizing cardiac toxicity after thoracic radiation therapy (RT) remains a clinical challenge. Mean heart dose is associated with major adverse cardiac events (MACE) and recent studies suggest RT dose to coronary arteries may better quantify risk after conformal RT. Prior studies have not evaluated RT dose to individual coronary arteries in risk modeling. We therefore sought to develop an integrated dosimetric model accounting for RT doses received by each coronary artery. <h3>Materials/Methods</h3> We performed a retrospective study of a randomly selected representative cohort of patients with locally advanced non-small cell lung cancer. We scored cardiac events as (1) ischemic, (2) constrictive, (3) valvular, or (4) conductive. We contoured and measured doses received by the left circumflex (LCX), left anterior descending (LAD), left main coronary (LMCA), and right coronary arteries (RCA). RT doses were measured as equivalent dose in 2-Gray (Gy) fractions. Dosimetric parameters associated with cardiac toxicity were examined using cumulative incidence, competing risk regression models, receiver operator characteristic (ROC), and log-rank statistics with significance defined as two-tailed <i>P</i><0.05. <h3>Results</h3> We identified 100 patients who received definitive RT (median=66 Gy, range=60-74 Gy) from 2006-2020. Median follow-up was 4.1 years (range, 0-11.0). Median age was 70 years (range, 38-84). Half the cohort was male (50%), a majority had smoking history (81%), and all received chemotherapy (100%). Thirty (30%) experienced MACE at a median of 13 months post-RT. Most events were ischemic (n=17, 57%) or conductive (n=13, 43%), with only 3 cases (10%) of valvular disease. Mean heart, LCX, LAD, and LMCA doses were predictive of cardiac toxicity (AUC=0.59, 0.58, 0.59, and 0.63, respectively) and were each significantly associated with MACE on univariable analysis (<i>P</i><0.05). For patients with full labs available, 10-year atherosclerotic cardiovascular disease (ASCVD) risk was also associated with MACE (HR=1.31, <i>P</i>=0.04). Chemotherapy, age, diabetes, hypertension, prior cardiac event, and mean RCA dose were not associated with MACE. Using the log-rank statistic, we identified threshold mean doses of 1.82, 12.71, 15.45, and 31.29 Gy for the heart, LCX, LAD, and LMCA, respectively. Patients exceeding all three threshold doses to the LCX, LAD, and LMCA had a two-year rate of MACE of 57.1% versus 15.9% (<i>P</i>=0.005). On multivariable analysis after adjusting for ASCVD, patients exceeding all identified doses to the LCX, LAD, and LMCA were at nearly 5 times the risk of MACE (HR=4.76, <i>P</i>=0.0005). <h3>Conclusion</h3> Our results suggest that patients receiving doses above our identified thresholds to all of the left coronary arteries (LCX, LAD, and LMCA) are at highest risk of MACE. If validated, our findings may inform RT treatment planning, where sparing of at least one vessel may lead to lower risk of MACE.