1197 - Expanding TMLI Access with Automated Planning on a Ring-Gantry Linac: A Single-Institution Clinical Deployment
Presenter(s)
C. Han1, A. Magliari2, L. Rosa2, S. V. Dandapani1, J. Y. C. Wong1, T. M. Williams1, and A. Liu1; 1Department of Radiation Oncology, City of Hope National Medical Center, Duarte, CA, 2Varian Medical Systems, Palo Alto, CA
Purpose/Objective(s): A modern ring-gantry linear accelerator (linac) equipped with a dual-layer multi-leaf collimator (MLC) provides a maximum field size of 28 cm × 28 cm2. This study summarizes our institutional experience implementing total marrow and lymphoid irradiation (TMLI) on this linac model, focusing on workflow streamlining through development of automated planning tools and reporting clinical treatment of the first patient. Our goal is to support adoption of TMLI on this linac model by other centers.
Materials/Methods: We developed a suite of open-source software tools to automate key components of the TMLI treatment planning workflow on this linac platform. The software concatenates upper- and lower-body CT images for a whole-body image. It then automatically configures beam arrangements and isocenter placements on the whole-body CT image. Isocenters are placed either at the patient midline or at laterally shifted positions depending on patient width and anatomical location along the superior-inferior axis, with midline placement consistently used for the head and neck region and lower extremities. The application generates volumetric modulated arc therapy (VMAT) beams covering the entire target volume from the skull to the feet in a head-first CT simulation setup, and it converts lower-extremity VMAT beams for delivery with a feet-first, supine position. Plan optimization was supported by a previously validated knowledge-based planning (KBP) model. The software then creates one- or two-isocenter delivery plans, depending on isocenter spacing, which will be exported to the treatment delivery system. The automated workflow was tested and validated using simulation cases. One patient was subsequently treated using this workflow.
Results: The first patient treated with TMLI on this linac had a maximum width of 46.5 cm and a planning target volume (PTV) length of 165 cm. Fourteen VMAT beams across seven isocenters were used for the upper body through mid-thigh, and ten VMAT beams across five isocenters were used for the lower extremities. For a prescription dose of 12 Gy in 8 fractions, the treatment plan optimized with the KBP technique achieved a mean dose of 6.6 Gy, 5.7 Gy, 5.6 Gy, and 5.3 Gy for the lungs, kidneys, liver, and bowel, respectively. Treatment delivery required eight delivery plans, six of which contained two isocenters with a 10.5-cm spacing. All eight fractions were delivered successfully. The average time from acquisition of the first cone-beam CT to delivery of the final VMAT beam was 82±13 minutes (range: 56 – 95 minutes).
Conclusion: TMLI was clinically implemented on this ring-gantry linac with acceptable treatment times and satisfactory plan quality using the KBP model. The automated planning software significantly improved workflow efficiency by reducing manual, time-intensive steps. The software is freely available as both open-source code and a compiled executable to facilitate adoption by other institutions.