2377 - In Silico Comparison of SPArc and IMPT throughput

D. Mumaw¹, J. Iwrey^2,3, P. Wang⁴, X. Cong¹, S. Both⁵, B. De Jong⁵, E. W. Korevaar⁶, Z. Wang⁷, P. M. Liu¹, X. Li¹, X. Cao¹, G. Liu⁸, R. L. Deraniyagala Jr¹, C. W. Stevens¹, X. Gao^9,10, and X. Ding¹; ¹Department of Radiation Oncology, Corewell Health William Beaumont University Hospital, Royal Oak, MI, ²University of Michigan, Ann Arbor, MI, ³William Beaumont University Hospital - Corewell Health, Royal Oak, MI, ⁴Inova Schar Cancer Institute, Fairfax, VA, ⁵Rijksuniversiteit Groningen, Groningen, Netherlands, ⁶Department of Radiation Oncology, University of Groningen, University Medical Center Groningen, Groningen, Netherlands, ⁷Department of Oncology, Yizhou Pronton Therapy Center, Zhuozhou, Hebei, China, ⁸Cancer Center, Union Hospital, Tongji Medical College, Huazhong University of Science and Technology, Wuhan, China, ⁹Peking University First Hospital, Beijing, China, ¹⁰Hebei Yizhou Tumor Hospital, Zhuozhou, China

Purpose/Objective(s): Spot-scanning Proton Arc Therapy (SPArc) is a new delivery technique promising reduced treatment times compared to existing step-and-shoot Intensity Modulated Proton Therapy (IMPT) techniques. There has been an increase in demand for proton therapy resulting in many new centers being built and existing centers being retrofitted with additional vaults. To explore the added value of additional vaults in light of the increased throughput offered by SPArc, we constructed a model. Materials/

Methods: A model was developed incorporating SPArc and IMPT machine delays obtained from de Jong et al. (PMID: 36373893); patient couch setup and egress time (13.3 min.); irradiation time (IMPT: 2.4 min./field, SPArc: 4 min./arc); and beamline switching time (30 sec.). Additionally, an optional anesthesia delay (20 min.) could be included. The models throughput and duty-cycle predictions were validated with clinical logfiles from two institutions. The model was then applied to simulate a 15-hour clinical operation day to quantify the effects of vault number and anesthesia requirements on throughput and per-patient on-couch waiting time.
Results: Validation was successful. With a single vault for 11 hours, the model predicted 22.9 treatments compared to 21 reported by the logfiles (20% anesthesia). With a 2-vault setup for 8.3 hours, the model predicted 33.4 treatments compared to 35 logged treatments (5% anesthesia). In the latter scenario, duty cycle logs were available: the model prediction (23%) and logfiles (26%) agreed. Upon simulating a cohort with 20% anesthesia requirements, median daily patient throughput increased logarithmically with increasing vault number (#) for both SPArc (throughput = 33 + 96 x log(#): r²=0.96) and IMPT (throughput = 32 + 63 x log(#): r² =0.99). SPArc was able to treat 24, 25, 29, 37, and 49% more patients than IMPT in 1–5 vaults, respectively (p<0.001 for all). Comparing 20% to 0% anesthesia requirements, with SPArc, anesthesia incurred a relatively constant 18±1% loss in throughput irrespective of vault number, while IMPTs loss improved linearly from 17–6% from 1–5 vaults. Median on-couch waiting times (minutes) increased quadratically with increasing vault number (#) for both SPArc (wait = -0.54# + 0.29#^2 + 0.86: r² =0.99) and IMPT (wait = 1.25# + 0.63#^2 + 0.01: r² =1.0). Compared to IMPT, SPArc reduced on-couch wait times by 77±4%—consistent across vault counts (p<0.001 for all). Anesthesia incurred increases in on-couch wait time which worsened linearly with increasing vault count from 1–5 for both SPArc (0–30%) and IMPT (1–15%).
Conclusion: This model suggests that the reduced treatment times that result from use of SPArc technology can improve the treatment workflow efficiency across all proton center configurations, as measured by improved patient throughput and reduced on-couch waiting time relative to IMPT. The optimal number of vaults appears to be limited by anesthesia requirements and on-couch waiting time, but this warrants further analysis.