Assessing Accumulated Dose and Geometrical Changes of Esophagus from Weekly Mris Acquired during Radiotherapy of Locally Advanced Lung Cancer

Abstract

Moderate to severe acute esophagitis (AE) is a key toxicity limiting radiotherapy (RT) in locally advanced non-small cell lung cancer (LA-NSCLC). The purpose of this pilot study is to develop a clinical platform for robust evaluation of delivered esophageal dose via weekly MRIs acquired during the RT course, and correlate the associated dose and local geometric changes with AE. Twenty LA-NSCLC patients will be enrolled in an IRB approved study to undergo weekly MR during RT. Nine patients were treated via intensity-modulated RT to 60-70Gy in 2 Gy-fractions with concurrent chemotherapy and had weekly MRIs. Respiratory triggered 2D axial T2-weighted MRIs were acquired on a 3T scanner during the RT courses to monitor tumor/esophagus response. AE grades were defined according to CTCAE and 6 of the 9 patients developed AE2, 3-6 weeks after RT started. The MRIs were first rigidly aligned to the planning CT referenced to the tumor. The esophagus was assumed as a tubular organ and the distances between the medial axis of esophagus on the planning CT and the weekly MRIs were calculated to estimate inter-fractional positional uncertainties. Then, the esophagus contours on the weekly MRIs were propagated onto the planning CT via deformable registration using a regularized large deformation diffeomorphic metric mapping (LDDMM) algorithm. The total dose delivered to esophagus was accumulated voxel by voxel based on the correspondence built via LDDMM. Local geometric changes of esophagus during the RT course were calculated using the Jacobian map obtained from LDDMM. Ultimately, correlation between the accumulated mean esophageal doses and local geometric changes of esophagus with AE status (dichotomization at ≥AE2) were calculated. The esophagus was better visualized and contoured on MRI than planning CT due to improved soft tissue contrast. As a mobile structure, the inter-fractional positional uncertainty of esophagus was 6.4±2.7mm across patients, with a maximum of 14.6mm. Consequently, the population median accumulated esophagus mean dose at the end of RT was 24Gy, which was higher than the planned dose of 23Gy (p=0.01); individual increases of up to 2.9Gy were seen. The population median of the mean accumulated esophagus dose was 20Gy at the time of AE. Both weekly accumulated dose and local geometric changes (expansions) of esophagus measured via Jacobian correlated with AE; the associated correlation coefficients (R) were 0.53, and 0.73, respectively. MRI is well suited for tracking the esophagus changes and accumulating doses along the course of RT. Weekly expansion of esophagus has the potential to be a prognostic indicator of AE in addition to esophagus mean dose. A study recruiting more patients is warranted to build a predictive model of AE, and guide an adaptive RT approach to reduce AE.