Use of Virtual CT and On-Treatment MRI to Reduce Radiation Dose and Anesthesia Exposure Associated With the Adaptive Workflow in Pediatric Patients Treated With Intensity Modulated Proton Therapy

Purpose: The purpose of this study was to determine whether virtual computed tomography (vCT) derived from daily cone beam computed tomography (CBCT), or on-treatment magnetic resonance imaging (MRI_tx) can replace quality assurance computed tomography (qCT) in our clinical workflow to minimize imaging dose and potentially anesthesia exposure in patients requiring plan adaptation. Methods and Materials: Pediatric patients (age <24 years) treated from 2020 to 2023 with intensity modulated proton therapy with at least 1 qCT during proton therapy were eligible. For cases that required plan adaptation, the dose was recalculated on vCT and compared with same-day qCT as well as the original planning computed tomography (pCT). Anatomic changes triggering plan adaptation were grouped into categories. Two pediatric radiation oncologists verified whether these changes could be detected using CBCT, qCT, and/or MRI_tx. A new adaptive imaging workflow was proposed to limit imaging dose and anesthesia exposure. Results: One hundred sixty-eight pediatric patients were treated from 2020 to 2023. Across all patients, there were 517 qCT scans and 61 MRI_tx acquired. The median number of qCT scans per patient was 3 (range, 1-5). The treatment plans for 20 patients (12%) were adapted. In all patients requiring plan adaptation, there was a correlation between dose differences in target coverage and maximum body dose when comparing vCT with pCT and qCT with pCT (n = 20, r² = 0.79, P < .01, and r² = 0.32 P = .01, respectively). The most common reason for adaptation was tissue change (eg, inflammation, changes in abdominal gas, or diaphragmatic variability) in the beam path (10/20) and changes in tumor volume (6/20). All cases of weight change, tissue change in beam path, and unreproducible setup could be detected on CBCT. All cases of change in tumor volume within the brain were detected on MRI_tx. Replacing the qCT with the vCT was associated with an estimated median reduction of imaging dose by 50% and anesthesia exposure by 1.5 hours. Conclusions: vCT derived from daily CBCT only or MRI_tx can safely replace qCT for monitoring dosimetric changes to trigger a new pCT in our clinical workflow. This change would potentially reduce imaging dose and anesthesia exposure.