Hippocampal avoidance whole brain radiotherapy: developing a RapidPlan model

Abstract

Background: Hippocampal-avoidance whole brain radiotherapy (HA-WBRT) has emerged as an approach to retain intracranial tumour control while minimizing cognitive decline. However, the contouring and planning requirements are more complex and time consuming compared to standard WBRT. RapidPlan is an automated treatment planning software that is designed to increase planning efficiency whilst maintaining plan quality. Our group developed an automated HA-WBRT RapidPlan model (Auto) and compared it to a manually optimised standard template (Manual) to assess plan quality and planning efficiency.Methods: A Radiation Oncologist first contoured the hippocampi on 31 patient CT brain data sets fused with MRI with a brain planning target volume (PTV) minus hippocampal avoidance structure, optic chiasm, optic nerve, and lens structures also created. Manual standard template plans were created by an experienced radiation therapist for the first 21 patients with all plans needing to achieve Radiation Therapy Oncology Group (RTOG) 0933 protocol requirements prior to inclusion for creation of the automated RapidPlan model. This Auto model was then tested on a set of 10 separate patients and compared with a Manual plan.The dosimetric parameters and number of optimisations required to achieve protocol requirements were recorded for both.Results: Both the Auto and Manual plans achieved protocol requirements with Auto plans able to achieve these requirements on 1st optimisation for all 10 patients. In contrast, Manual plans were only able to produce acceptable plans in a single optimisation for 5 patients, with 4 patients requiring 2 optimisations and 1 patient requiring 3 optimisations. PTV coverage met RTOG recommendations for all plans but Auto plans were able to achieve lower doses to the organs at risk (OARs) compared to Manual plans, including significantly lower doses to the hippocampi. Independent dose calculation and patient specific dosimetry measurement had a greater than 99% pass rate.Conclusions: A department-created automated HA-WBRT RapidPlan model is feasible and allows for more efficient plan creation with significantly better hippocampal doses compared manually optimised plans and physics check confirming deliverability. Auto plans were able to be created with reduced planning time and resource utilization compared to manual plan creation, allowing for streamlining of workflow and reduced time to treatment for patients.