3259 - Implementing CT-Guided Online Adaptive Radiation Therapy (CT-ART) without Destroying the Clinic: A First Year Workflow

12:45pm - 02:00pm PT

Hall F

Screen: 30

POSTER

Presenter(s)

Zachary Kiss, DO - Fox Chase Cancer Center, Philadelphia, PA

Z. A. Kiss¹, J. Fredette², A. Lukez¹, R. M. Shulman¹, H. N. Yankey¹, A. Eldib Jr¹, J. Panetta¹, C. M. C. Ma¹, L. Chen¹, D. M. Yang¹, T. Lin¹, I. Veltchev¹, X. Chen¹, J. K. Wong¹, S. S. Kumar¹, J. G. Price¹, M. A. Hallman¹, E. M. Horwitz¹, J. E. Meyer¹, and T. J. Galloway¹; ¹Department of Radiation Oncology, Fox Chase Cancer Center, Philadelphia, PA, ²Department of Biostatistics and Bioinformatics, Fox Chase Cancer Center, Philadelphia, PA

Purpose/Objective(s):

CT-ART is a treatment technique involving the generation of two unique plans (adaptive vs scheduled) for selection prior to each fraction (fx), based upon daily imaging with contour adjustment. A scheduled plan applies the original plan from simulation to the current anatomy while an adaptive plan generates a new optimization for the current anatomy. We evaluated workflow characteristics by fx throughout implementation of CT-ART.

Materials/Methods:

All patients treated with CT-ART from program initiation until present were evaluated (2023-2024; n=233; 1293 fxs). Parameters pertinent to treatment delivery included covering vs prescribing physician, adaptive vs scheduled treatment (adaptive fxs entail de-novo contours and QA prior to delivery), fx duration (initial CBCT to treatment completion in minutes), number of fractions delivered in a day, and disease site. Clinical variables were compared using Chi-square and Fisher’s Exacts tests. Continuous variables were compared using Kruskal-Wallis tests. Pearson correlation tests were used to determine trends from the date of program initiation. An ordinary least squares regression model was created to determine association between fx duration with variables of interest. Regression models controlled for physician and clustered by anonymized patient ID. Analysis was conducted in R v4.4.2.

Results:

The most common treatment course was 5 fxs (range 1-34). The adaptive plan was selected for 88% of fxs. Prostate was the most frequent treatment target (n=476 fxs; 37%). Overall mean fx duration was 32 minutes (IQR 25-37): 33 min for adaptive fxs (IQR 25-38) and 26 min for scheduled fxs (IQR 20-30). The longest mean fx duration by site was pancreas (44 min; IQR 36-50) and shortest was prostate (28 min; IQR 23-31). Since program initiation, median number of daily fractions rose from 3 to 4 in the first vs final 90 days of the study period (r=0.4; CI 0.3, 0.5). Amongst sites, mean fx duration decreased for pancreas from 51 to 45 minutes in the first vs final 90 days of the study period (r= -0.3; CI -0.5, -0.1). Controlling for covariates, receiving an adaptive fraction predicted a 0.22 increase in log fx duration (p<0.01). Prescribing physicians delivered 42% of their own fxs, otherwise these were treated by a covering physician. Whether the prescribing physician covered their own fractions was not significantly associated with fraction duration (34 vs 31 min; p=0.8).

Conclusion:

Over the study period, the number of daily fractions increased without significant change in overall mean fraction duration. Trends in fraction duration were found in some abdominal targets, suggesting a learning curve specific to those disease sites. An adaptive plan takes significantly longer than a scheduled plan. Whether the prescribing physician was also the covering physician did not impact mean fraction duration.