HyperArc Automated Stereotactic Radiosurgery Planning Enables Accurate a Priori Fractionation Scheme Selection via Adherence to HyTEC Toxicity Thresholds.
Joel A Pogue, John Fiveash, Rex Cardan, Christopher Willey, Natalie Viscariello, Rodney Sullivan, Samuel Marcrom, Luke Moradi, Philip Schmalz, James Markert, Richard Popple
Abstract
Open AccessPurpose: Radiosurgery plan safety is commonly estimated by volumes receiving specific doses (ie, 12 Gy/1 fraction [fx]), which are evaluated postplan generation. However, automated treatment planning can produce highly consistent and thus predictable plans. Thus, we hypothesized that HyperArc (HA) automated stereotactic radiosurgery (SRS) planning enables clinical decision-making prior to plan generation, such as selecting the appropriate SRS fractionation scheme. Methods and Materials: All previously treated single-isocenter HA plans at our institution were queried, totaling 3361 marginless targets without bridging at the 50% isodose level (1495 plans), making this the largest single-institutional SRS dosimetry study to the authors' knowledge. Eight isodose volumes (IDVs; 50.00%-97.60%) were calculated for all HA targets, each corresponding to the ratio of a High Dose per Fraction, Hypofractionated Treatment Effects in the Clinic (HyTEC) brain toxicity dose level and a common prescription dose (eg, 50.00% = 12 Gy/24 Gy). Power law relationships of IDV and target volume ( I D V = a V t a r g e t b ) were generated from a training data set of 361 targets (10.7%) and validated on the remaining 3000 targets (89.3%), allowing grade 1 to 3 brain toxicity rates to be predicted from target volume. Results: Models resulted in high R² values when applied to the validation cohort (≥0.982), allowing targets to be classified as either above or below the HyTEC thresholds (IDV = 5 cm3, 10 cm3, and 20 cm3) with high accuracy (≥97.6%) and precision (≥99.3%). As an example, the 50.0% IDV model predicted that target volumes/diameters of 1.00 cm3/1.24 cm, 2.34 cm3/1.65 cm, and 5.51 cm3/2.19 cm correlate with 3.6%, 4.8%, and 8.6% grade 1 to 3 brain toxicity rates, respectively, when prescribing 24 Gy/1 fx. Conclusion: The resulting models enabled accurate and precise prediction of target volumes/diameters, resulting in 3.6%, 4.8%, and 8.6% brain grade 1 to 3 toxicity rates, according to HyTEC toxicity estimates. Leveraging relative IDVs rather than prescription doses enabled all 3361 targets to be used for modeling 9 common SRS prescriptions (1 fx: 24 Gy, 20 Gy, 18 Gy, 16 Gy, and 15 Gy; 3 fx: 27 Gy and 24 Gy; 5 fx: 30 Gy and 25 Gy), enabling clinicians to estimate brain toxicity a priori via an open-source calculator.