348 - Characterizing Stereotactic Radiosurgery Plan Quality across the Community: Imaging and Radiation Oncology Core (IROC) Phantom Audit Program

Purpose/Objective(s): SRS demands highly localized and precise dose delivery. The IROC QA Center performs standardized SRS audits to verify delivery accuracy for participants in NCI clinical trials, testing whether the dose delivered to an anthropomorphic head phantom matches the institution's calculated plan. These audits assess delivery independent of plan quality; institutions utilize their own clinical planning protocols and objectives. We analyzed a decade of SRS audit data to characterize plan quality variability across the community and determine if the plan itself serves as a driver of physical delivery failure.

Materials/Methods: We analyzed 958 IROC Houston SRS head phantom irradiations from 680 institutions across various platform categories: Varian Modern (n=522), Varian Old (n=146), Elekta (n=102), CyberKnife (n=97), Gamma Knife (n=58), TomoTherapy (n=16), and Halcyon/Ethos (n=9). Delivery accuracy was assessed using thermoluminescent dosimeter (TLD) point-dose ratios (target ±5%) and orthogonal planar gamma analysis (=85% at 5%/3 mm). Plan quality was retrospectively evaluated using Paddick Conformity Index (CI), RTOG CI, Gradient Index (GI), R50, Coverage, Selectivity, Homogeneity, V12Gy, and Tumor Control/Normal Tissue Complication Probability (TCP/NTCP) using the HyTEC logistic model. Associations were evaluated using Kruskal–Wallis tests, stratified Mann–Whitney U tests, and Spearman correlation.

Results: The overall audit pass rate was 88%. Gamma Knife showed the highest pass rate (94.8%), followed by Varian Modern (90.4%), while other platforms ranged from 80% to 85%. While the audit does not mandate specific planning goals, machine platform imparted a significant dosimetric signature across all quality metrics (Kruskal–Wallis p<0.001; HI e²=0.302, GI e²=0.159). Notably, plan quality was a strong predictor of physical delivery success: failing audits had significantly lower CI than passing ones (median 0.788 vs. 0.846; p<0.001). Within Varian Modern, VMAT (n=384) outperformed IMRT/static fields (n=118) in both quality metrics and delivery accuracy, showing higher CI (0.862 vs. 0.833; p=0.012) and higher gamma passing rates (98.5% vs. 97.5%; p=0.013). GammaKnife achieved the steepest dose gradients and lowest radionecrosis NTCP, while TomoTherapy exhibited the shallowest gradients and highest radionecrosis risk. The Dosimetric Leaf Gap (DLG) was the strongest TPS-level predictor of systematic delivery error (?=-0.284 vs. TLD ratio, p<0.001), followed by MLC transmission factor (?=-0.116, p=0.02).

Conclusion: SRS plan quality varies substantially across the community based on platform and technique. While IROC audits focus strictly on delivery accuracy, our findings demonstrate that poor plan quality is intrinsically linked to increased delivery failure. This suggests that optimizing plan parameters is not only a clinical necessity for tissue sparing but also a critical factor in ensuring the deliverability of the intended dose.