3005 - Evaluation of a Prototype VMATp Optimizer with Automatic Static Angle Modulated Port Placement for Lung Radiotherapy

Purpose/Objective(s):

A recently introduced VMATp technique (commonly referred as RAD) integrates rotational arcs with dynamic collimator rotation and Static Angle Modulated Ports (STAMPs) and has demonstrated improved OAR protection over conventional VMAT in lung radiotherapy. Manual STAMP selection is user dependent. This study compares a prototype optimizer for automatic STAMP parameter selection to manual for clinical benchmarks.

Materials/Methods:

Eleven patients with locally advanced lung cancer (mean target volume: 560.4±195.3 cc) prescribed to 66 Gy in 33 fractions were retrospectively selected. For each patient, three plans were compared: clinical VMAT, RAD with manual STAMP selection, and RAD with automatic STAMP selection. Manual STAMP selection began with 7 evenly distributed ports. Based on prior experience, static–arc weighting was set to be static-dominant. Then one at a time, ports traversing large amounts of normal tissue were removed, with full re-optimization after each removal. This generated six manually created RAD plans per patient. The plan with the best overall dosimetry was selected for analysis. In the automatic approach, port angles, number of ports, and static–arc weighting were determined directly by the prototype optimizer. All RAD plans used dynamic collimator rotation during delivery to conform to the target. Plans were normalized to PTV D95%=98%, and evaluated using OAR metrics and MUs. Wilcoxon signed-rank tests were performed for pairwise comparisons.

Results:

All RAD plans met clinical goals and achieved substantially improved dosimetry compared with clinical VMAT, particularly in lung V_20Gy and V_5Gy, cord D_max, heart V_5Gy, and esophagus D_mean. Auto-STAMP selection generated 2 STAMPs, whereas manual selection used 2-6. For smaller (<400 cc) or peripheral tumors (4 out of 11 cases), auto-STAMP achieved dosimetric results comparable to manual selection. For larger, centrally located tumors with mediastinal nodal involvement, manual selection of 3–6 STAMPs provided small but significant improvements over automatic selection (2 STAMPs) in lung V_5Gy, (53.9±13.9% vs 54.8±13.9%), V_20Gy, (22.0±6.0% vs 22.3±6.5%), D_mean (13.6±3.6 vs 14.1±3.4 Gy), and cord D_max (38.1±6.5 vs 41.0±5.2 Gy) (p<0.05). MUs were similar between automatic and manual plans, indicating comparable plan complexity. Both RAD approaches required more MUs than VMAT due to additional modulation from STAMPs. However, simulated delivery times were not significantly different among VMAT (66.0±13.4 s), automatic (66.6±7.2 s), and manual (68.1±10.8 s) plans.

Conclusion:

RAD with automatic STAMP angle and weight selections produced clinically acceptable plans in all cases, with improved lung and heart sparing compared to clinical VMAT baseline. Plan quality with auto-STAMP was comparable to manual settings for smaller, peripheral thoracic tumors. Individualized STAMP configuration is still necessary to optimize lung and cord sparing particularly in larger, centrally located tumors.