Treatment planning for total marrow irradiation (TMI) is challenging due to large number of structures involved and lack of experience in most centers. In this study, we aimed to evaluate the feasibility of applying the knowledge-based treatment planning technique to simplify and automate treatment planning for TMI on conventional Linacs with volumetric modulated arc therapy (VMAT) fields, by using historical helical tomotherapy (HT) TMI plans as input data for dosimetric model construction. We retrospectively retrieved 30 clinical TMI treatment plans from an institutional clinical trial with uniform prescription dose of 12 Gy to planning target volumes (PTVs) including skeletal bones, lymph nodes, spleen, and spinal canal, for 14 male and 16 female patients. All the plans were generated on a HT planning station with a jaw size of 5.0 cm. The dosimetric data for all the structures were extracted to a commercial knowledge-based planning system for construction of dosimetric models for 29 structures, including 4 PTVs and 25 organs at risk (OARs). To evaluate the efficacy of model-based optimization, five new TMI treatment plans using VMAT fields were generated. Each VMAT plan included 8 full arc fields with four isocenters arranged along the longitudinal direction to cover the complete target volumes from the top of skull to mid-thigh, using 6-MV photon beams. Structure-specific objectives generated from the dosimetric models were used to guide VMAT plan optimization. All the treatment plans were normalized so that 85% of the skeletal bone PTV received at least the prescription dose. DVH estimation models for 3 PTVs and 19 OARs were constructed and trained, while four gender-specific OARs failed due to insufficient statistics and three other structures failed due to insufficient in-field matches. In the VMAT-based TMI plans, all the PTVs had comparable dose coverage as in the HT plans. In the 19 OARs with constructed dosimetric models, 11 showed lower average median dose (D50) compared to the HT plans, and 8 shower higher average D50 compared to the HT plans; the difference was not statistically significantly (p > 0.05 in t-tests) except for 2 OARs. The table compares the dose statistics for some representative OARs between the VMAT and the HT plans. This study demonstrated that with the knowledge-based treatment planning system, prior treatment planning experience and dosimetric data from historical HT plans could be used to simplify TMI treatment planning with other delivery techniques by automating the optimization process and help achieve high plan quality. The dosimetric models can be made available to aid TMI planning in centers with limited prior experience.Abstract 1166; TableOrganAverage D50 /GyHT PlansVMAT PlansBladder6.9±1.55.9±0.8Bowel5.5±1.16.1±1.1Brain6.3±1.06.0±0.7Esophagus4.6±0.65.3±0.8Kidney, Lt5.2±0.95.0±0.8Kidney, Rt4.9±0.94.7±1.0Liver6.3±0.85.6±0.7Lung, Lt5.5±0.55.7±0.3Lung, Rt5.2±0.45.5±0.1Thyroid6.1±1.54.0±0.2 Open table in a new tab