Abstract
INTRODUCTION: We analyzed the ability of various patient- and treatment-related factors to predict radiation-induced esophagitis (RE) in patients with non-small cell lung cancer (NSCLC) treated with three-dimensional conformal radiation therapy (3D-CRT), intensity-modulated radiation therapy (IMRT), or proton beam therapy (PBT). METHODS AND MATERIALS: Patients were treated for NSCLC with 3D-CRT, IMRT, or PBT at MD Anderson from 2000 to 2008 and had full dose-volume histogram (DVH) data available. The endpoint was severe (grade≥3) RE. The Lyman-Kutcher-Burman (LKB) model was used to analyze RE as a function of the fractional esophageal DVH, with clinical variables included as dose-modifying factors. RESULTS: Overall, 652 patients were included: 405 patients were treated with 3D-CRT, 139 with IMRT, and 108 with PBT; corresponding rates of grade≥3 RE were 8%, 28%, and 6%, respectively, with a median time to onset of 42 days (range, 11-93 days). A fit of the fractional DVH LKB model demonstrated that the fractional effective dose was significantly different (P=.046) than 1 (fractional mean dose) indicating that high doses to small volumes are more predictive than mean esophageal dose. The model fit was better for 3D-CRT and PBT than for IMRT. Including receipt of concurrent chemotherapy as a dose-modifying factor significantly improved the LKB model (P=.005), and the model was further improved by including a variable representing treatment with >30 fractions. Examining individual types of chemotherapy agents revealed a trend toward receipt of concurrent taxanes and increased risk of RE (P=.105). CONCLUSIONS: Fractional dose (dose rate) and number of fractions (total dose) distinctly affect the risk of severe RE, estimated using the LKB model, and concurrent chemotherapy improves the model fit. This risk of severe RE is underestimated by this model in patients receiving IMRT.