Magnetic Resonance Imaging Assessment For Clinical Target Volume Delineation Of The Postoperative Cavity In Breast Radiotherapy

In partial breast (PBI) and boost radiotherapy (RT) treatments, the postsurgical lumpectomy cavity (SxC) has been widely accepted as the basis for the contoured volume from which the clinical and planning target volumes are derived. Given the imaging limitations of soft tissue structures with CT, significant inter-physician variations in SxC delineation have been observed. We examined the impact of magnetic resonance (MR) imaging of the breast on SxC delineation by five radiation oncologists (RO). Twenty patients, with clinical stage Tis-2N0 breast tumors, imaged on a prospective study were analyzed. All patients underwent a treatment planning CT directly followed by an MR (axial T2w and contrast-enhanced FSPGR). The relevant clinical, operative, imaging and pathology reports were made available to the ROs. All ROs contoured the SxC, in a blinded manner, on the CT and again on the CT/MR fusion images after a two week interval. All SxCs were assigned a CT cavity visualization score (CVS), as described by Landis et al. The clinical target volume (CTV) was defined as SxC+10 mm and planning target volume (PTV) as CTV+10 mm. Conformity indices (CI), the ratio of the common volume to the total encompassing volume, for the CT and CT/MR based target volumes were calculated for each observer pair. The mean time from surgery to imaging was 49 days (26-76) and CVS was 3.7 (2-5) with no correlation observed between the two (r = -0.077, p = 0.75). Of the 19 evaluable cases, the inclusion of MR to the CT resulted in a significant increase in the delineated SxC volumes (p < 0.001). The median CT and CT/MR SxC volumes were 16.62 cc (4.84-177.93) and 20.39 cc (6.2-198.35), respectively. The mean CI for the CT based SxC, CTV, and PTV were 0.65, 0.80, and 0.84, respectively, compared to 0.59, 0.76, and 0.80 for the CT/MR based SxC, CTV, and PTV. For both the CT and CT/MR SxC volumes a significant correlation between CVS and mean CI was observed (r = 0.91 and 0.84 respectively, p ≤ 0.01). Higher CIs were observed in 21% of the cases (4 patients) with the CT/MR compared to CT only. The MR detection of SxC imaging features, in addition to the CT detected cystic cavity, was noted to result in a reduction in conformity. Higher inter-physician variability was observed in SxC delineation on CT/MR compared to CT alone. Our findings indicate that while the provision of MR imaging provides superior soft tissue contrast, ROs may require specialized training and experience in breast MR image interpretation prior to introducing this modality into the target volume delineation process.