Enhanced MLC modeling did not outperform optimized DLG modeling for single-isocenter SRS

KEY POINTS

• Investigators evaluated 60 clinically delivered HyperArc VMAT-SRS plans: 30 single-isocenter single-target and 30 single-isocenter multi-target plans containing 2–25 targets. Target volumes extended down to 0.007 cm³, with lesions up to 74.1 mm from isocenter.
• Ninety patient-specific QA measurements were performed using SRS MapCHECK and point-dose normalization. Eclipse calculations compared the physics-based enhanced leaf model (ELM) with conventional dosimetric leaf gap (DLG) modeling using Acuros XB and, for ELM, AAA.
• At the clinically emphasized 2%/1 mm criterion, the optimized Acuros XB-DLG model with a 0.5×0.5-mm² spot size achieved the highest overall mean gamma passing rate at 97.4%. Acuros XB-ELM configurations produced approximately 96.3–96.5%.
• Under the much stricter 1%/1 mm criterion, differences became more apparent: optimized DLG achieved a mean passing rate of 92.6%, compared with 90.6% for the vendor-recommended Acuros XB-ELM configuration.
• Absolute dose agreement was essentially equivalent. Across measurements, mean calculation-minus-measurement difference was approximately −0.7% for both ELM and DLG, while the paired ELM-versus-DLG difference averaged 0.0%, ranging from −2.7% to +1.8%.
• ELM was remarkably insensitive to spot-size selection: gamma results for 0×0 versus 0.5×0.7 mm² configurations were strongly correlated (R²=0.982). That makes ELM attractive because it reduces dependence on institution-specific parameter tuning.
• Neither approach eliminated the known challenge of small off-axis targets in multi-target SRS. The optimized DLG model performed slightly better overall, but it had been specifically fine-tuned using SRS measurements, while ELM achieved clinically acceptable agreement without equivalent site-specific tuning.

CLINICAL TAKEAWAY

SOURCE

Physics and Imaging in Radiation Oncology