R Finnegan^1,2*, E Lorenzen^3,4, J Dowling^1,5,6, D Thwaites¹, I Jensen⁷, M Berg⁸, M Thomsen⁹, B Offersen^10,11, C Brink^3,4, L Holloway^1,2,6,12,13, (1) Institute of Medical Physics, University of Sydney, Camperdown, NSW, AU, (2) Ingham Institute for Applied Medical Research, Liverpool, NSW, AU, (3) Institute of Clinical Research, University of Southern Denmark, Odense, Denmark, (4) Laboratory of Radiation Physics, Odense University Hospital, Odense, Denmark, (5) The Australian e-Health and Research Centre, CSIRO Health and Biosecurity, Herston, QLD, AU, (6) School of Mathematical and Physical Sciences, University of Newcastle, Newcastle, NSW, AU, (7) Department of Medical Physics, Aalborg University Hospital, Aalborg, DK, (8) Department of Medical Physics, Vejle Hospital, Vejle, DK, (9) Department of Medical Physics, Aarhus University Hospital, Aarhus, DK, (10) Institute of Clinical Medicine, Aarhus University, Aarhus, DK, (11) Department of Oncology, Aarhus University Hospital, Aarhus, DK (12) South Western Sydney Clinical School, University of New South Wales, Sydney, NSW, AU, (13) Centre for Medical Radiation Physics, University of Wollongong, Wollongong, NSW, AU

• 
  R Finnegan

(Tuesday, 7/14/2020) 2:00 PM - 3:00 PM [Eastern Time (GMT-4)]

Room: Track 3

Purpose:
To investigate cardiac substructure dosimetry in a large, retrospective radiotherapy trial and to evaluate the extent to which dose to the whole heart can be used to infer and predict this dosimetry.

Methods:
An open-source, multi-atlas based segmentation framework was used to automatically delineate the whole heart and 16 cardiac substructures on planning CT scans of 1518 Danish breast cancer patients participating in a hypofractionation (versus standard fractionation) trial during 2009-2014. These patients were treated using the same tangential field-in-field technique, and dosimetry for cardiac volumes was extracted from the radiotherapy plans. The predictive power of whole heart dosimetry to estimate substructure dosimetry was assessed using two models: a linear relationship with only mean whole heart dose, and a second-order relationship including other whole heart-based dosimetry and patient parameters.

Results:
The mean and maximum doses to particular cardiac structures vary by approximately an order of magnitude, even within a single treatment laterality. Patients in the hypofractionation trial arm receive lower physical doses, which are evident with correspondingly lower mean whole heart dose. However, the average dose to the LADCA is similar between arms. Evaluation of model performance indicates that, averaged over the cardiac substructures, the mean whole heart dose accounts for 44% of the inter-patient variation in mean dose to cardiac substructures, and 21% of the variation in D5%. The more complex model demonstrates improved performance, accounting for 80% and 71% of the variation in mean and D5% dose to cardiac substructures, respectively.


Conclusion:
Improved sparing of cardiac structures is possible, with trends towards overall lower doses. Mean whole heart dose alone is not sufficient to characterise inter-patient variation in cardiac dosimetry, even within a homogeneous treatment group. Additional parameters, based on the dose to the whole heart and patient anatomy, can be used to accurately predict substructure dosimetry.

Keywords

Not Applicable / None Entered.

Taxonomy

Not Applicable / None Entered.

Contact Email