Exploring plan quality: using plan complexity to quantitatively analyse the tradeoff between clinical suitability and dosimetric accuracy

PURPOSE: This study investigates how controlling the complexity of Volumetric Modulated Arc Therapy (VMAT) plans can quantitatively balance clinical suitability and dosimetric accuracy. MATERIALS AND METHODS: Thirty clinical cases treated with VMAT plans were re-optimized to limit plan complexity to varying degrees using Monitor Units (MU) objective and the Aperture Shape Controller (ASC) in the Eclipse TPS. The resulting plans were evaluated based on surrogate measures of plan complexity (single and aggregate complexity metrics), dosimetric accuracy (γ-index passing rate), and clinical suitability (Adjusted Plan Quality Metric). RESULTS: Stricter MU limits and higher ASC penalties resulted in beam's-eye-view regularization and reduced leaf travel while increasing modulation of gantry speed and dose rate. Limiting plan complexity enhanced dosimetric accuracy, as shown by a higher γ-index passing rate. However, this was accompanied by a moderate reduction in clinical suitability, reflected by a lower Adjusted Plan Quality Metric (APQM%), with variations across treatment sites. Complexity metrics related to MLC apertures were the most significant indicators for characterizing dosimetric accuracy and clinical suitability. An aggregate complexity score was established, demonstrating a monotonic relationship with both dosimetric accuracy and clinical suitability, enabling the assessment of their trade-off. CONCLUSION: By defining appropriate proxies, clinical suitability and deliverability can be quantitatively balanced through managing plan complexity in VMAT plans. While tailored to this study's clinical routine, the method holds potential for standardizing plan quality assessment and improving VMAT treatment outcomes across diverse clinical settings. Insights gained could contribute to greater consistency and effectiveness in evaluating the overall plan quality in radiation therapy.