VMAT for Head and Neck Cancer: What Is the Impact of the Optimization Parameters on the Treatment Plan Quality?

To evaluate the effectiveness of different optimization parameters on radiotherapy plan quality for seventeen head and neck cancer patients.

Volumetric Modulated Arc Therapy with Simultaneous Integrated Boost (VMAT-SIB) plans, involving up to three tumors, were generated. For each participant, a reference plan (Plan_Ref) was created using dual-arc with 180 control points, 20° gantry-angle increment and 1 cm minimum segment width. Modified plans were developed with dose constraints and optimization settings constant by changing to single-arc, 150 and 200 control points, 0.5 cm minimum segment width, and 30° and 40° gantry-angle increments. These plans were referred to as Plan_Arc1, Plan_CP150, Plan_CP200, Plan_SW0.5, Plan_Inc30, and Plan_Inc40, respectively. D_95% of planning target volumes (PTVs), homogeneity index (HI), monitor units (MUs), maximum dose (D_max) of spinal cord, mandible, and brainstem were recorded. Statistical and Bland-Altman analysis was performed comparing the modified plans to Plan_Ref.

Average D_95% values for PTV1, PTV2, and PTV3 ranged from 93.13 to 98.82%. Plan_SW0.5 provided superior target coverage and homogeneity with higher MUs than Plan_Ref. Plan_Arc1 significantly reduced PTV coverage and dose homogeneity, while increasing MUs compared with Plan_Ref (p < 0.05). The average D_max as derived from all planning approaches was up to 43.86 Gy, 65.86 Gy, and 48.85 Gy for spinal cord, mandible and brainstem, respectively. For spinal cord, Plan_Ref led to significantly lower doses compared to Plan_Arc1 and Plan_Inc30, while the brainstem recorded statistically higher D_max doses than Plan_Arc1. Significantly higher D_max was observed for the mandible using Plan_SW0.5 (p < 0.05). However, for D_max, the comparison plans showed good agreement with Plan_Ref based on Bland-Altman analysis.

The VMAT plan quality is strongly affected by the minimum segment width whereas no differences were observed with the modification of the number of control points.