Interfraction positional variation in pancreatic tumors using daily breath-hold cone-beam computed tomography with visual feedback.

We assessed interfraction positional variation in pancreatic tumors using daily breath-hold cone-beam computed tomography at end-exhalation (EE) with visual feedback (BH-CBCT). Eleven consecutive patients with pancreatic cancer who underwent BH intensity-modulated radiation therapy with visual feedback were enrolled. All participating patients stopped oral intake, with the exception of drugs and water, for > 3 hr before treatment planning and daily treatment. Each patient was fixed in the supine position on an individualized vacuum pillow. An isotropic margin of 5 mm was added to the clinical target volume to create the planning target volume (PTV). The prescription dose was 42 to 51 Gy in 15 fractions. After correcting initial setup errors based on bony anatomy, the first BH-CBCT scans were performed before beam delivery in every fraction. BH-CBCT acquisition was obtained in three or four times breath holds by interrupting the acquisition two or three times, depending on the patient's BH ability. The image acquisition time for a 360° gantry rotation was approximately 90 s, including the interruption time due to BH. The initial setup errors were corrected based on bony structure, and the residual errors in the target position were then recorded. The magnitude of the interfraction variation in target position was assessed for 165 fractions. The systematic and random errors were 1.2 and 1.8 mm, 1.1 and 1.8 mm, and 1.7 and 2.9 mm in the left-right (LR), anterior-posterior (AP), and superior-inferior (SI) directions, respectively. Absolute interfraction variations of > 5 mm were observed in 18 fractions (11.0%) from seven patients because of EE-BH failure. In conclusion, target matching is required to correct interfraction variation even with visual feedback, especially to ensure safe delivery of escalated doses to patients with pancreatic cancer.