Purpose: Conformal avoidance of the hippocampus during whole brain radiation therapy (WBRT) is associated with quality of life and memory preservation. The purpose of this study is to implement a knowledge-based treatment planning model for hippocampus avoidance whole brain (HAWB) volumetric modulated arc therapy (VMAT) that leverages optimization features in Varian Eclipse V15.6.
Methods: Fifty previously treated HAWB VMAT plans were collected for this study. Thirty out of the 50 patients were re-planned by a skilled dosimetrist with more than 10 years of experience. The VMAT plans use two co-planar full arcs and two non-coplanar half arcs to spare the hippocampus and skin while achieving high levels of target coverage. All plans were optimized using convergence mode. Thirty plans were used to train a knowledge-based planning model using the RapidPlan (RP) modeling tool. The final RapidPlan model was tested on the remaining 20 patients to automatically generate 20 HAWB VMAT plans.
Results: On average, the RP plans outperformed the original plans in almost all metrics analyzed. The PTV D2% values were 32.3±0.1 Gy vs. 33.2±0.9 Gy for the RapidPlan and original plan respectively while the PTV D98% values were 28.4±0.6 Gy vs. 26.9±2.0 Gy. These results indicate that the RapidPlan has higher target coverage with lower hot spots than the original plans. The hippocampus V9Gy decreased from 98.2%±3.7% to 62.8%±18.4% using the RapidPlan model. The mean dose to the eyes and the max dose to the lens were reduced by half using RapidPlan. The optic nerve max doses for RP plans were similar to those of the original plans (31.5 ±1.46 Gy vs. 31.2±1.55 Gy).
Conclusion: We have implemented a knowledge-based treatment planning model to automatically perform HAWB VMAT planning. Our results showed substantial improvement in target coverage and better sparing of the eyes, lens and hippocampus.