MRI-based volumetric assessment of cardiac anatomy and dose reduction via active breathing control during irradiation for left-sided breast cancer.

PURPOSE: Heart dose-volume analysis using computed tomography (CT) is limited because of motion artifact and poor delineation between myocardium and ventricular space. We used dedicated cardiac magnetic resonance imaging (MRI) to quantify exclusion of left ventricular (LV) myocardium via active breathing control (ABC) during left breast irradiation and to determine the correlation between irradiated whole heart and LV volumes.

METHODS AND MATERIALS: Fifteen patients who completed adjuvant irradiation for early-stage left breast cancer participated. Treatment consisted of 45 Gy to the entire breast using ABC followed by a 16-Gy electron boost to the lumpectomy cavity. Patients underwent planning CT scans in free breathing (FB) and moderate deep inspiration breath hold (mDIBH). Electrocardiogram-gated cardiac MRI was performed in the treatment position using alpha-cradle immobilization. MRI scans were acquired in late diastole (LD), mid-diastole (MD), and systole (S) for both FB and mDIBH. After image fusion with the patients' radiation therapy planning CT scan, MRI LV volumes were defined for the three examined phases of the cardiac cycle, and comparative dose-volume analysis was performed.

RESULTS: Cardiac volume definition was found to differ significantly because of combinations of respiratory and intrinsic heart motion. The fraction of LV myocardium receiving 50% (22.5 Gy) of the prescribed whole breast dose (V(22.5)) was reduced by 85.3%, 91.8%, and 94.6% via ABC for LD, MD, and S, respectively. Linear regression revealed strong correlation between MRI-defined whole heart and LV V(22.5) reduction via ABC, suggesting that LV myocardium accounts for up to approximately 50% of the excluded heart volume through this technique. Significant but weaker correlations were noted between CT-defined whole heart and LV V(22.5) reductions with marked variability in the measurements of patients with larger amounts of heart in the treatment field.

CONCLUSIONS: Cardiac MRI demonstrated a significant reduction in LV myocardium irradiated with the use of ABC. The correlation between reduction in V(22.5) values for LV wall and CT-defined whole heart suggests that CT is adequate for determining which patients are likely to benefit from ABC treatment, but inaccuracies inherent to standard CT dictate that more detailed imaging studies such as MRI are required for accurate cardiotoxicity assessment.