Click here to


Are you sure ?

Yes, do it No, cancel

Impact of Enhanced CT-Based Heart Model On Estimating Radiation Therapy Related Late-Onset Cardiac Disease in the Childhood Cancer Survivor Study

S Shrestha1, 2*, Q Liu3, J Bates4, Y Yasui5, A Gupta1, 2, C Owens1, 2, S Smith1, R Weathers1, C Lee6, B Hoppe7, W Leisenring8, K Oeffinger9, L Constine10, D Mulrooney5, G Armstrong5, R Howell1, 2, (1) MD Anderson Cancer Center, TX (2) UTHealth Graduate School of Biomedical Sciences, TX (3) University Of Alberta, Canada (4) University Of Florida, FL (5) St. Jude Children's Research Hospital, TN (6) National Cancer Institute, MD (7) Mayo Clinic, FL (8) Fred Hutchinson Cancer Research Center, WA (9) Duke University, NC (10) University of Rochester Medical Center, NY


(Sunday, 7/12/2020)   [Eastern Time (GMT-4)]

Room: AAPM ePoster Library

Purpose: We previously evaluated late-onset cardiac disease in the Childhood Cancer Survivor Study (CCSS). Since individuals in CCSS were mostly treated without computed tomography (CT)-based planning, heart doses were estimated by reconstructing each individual’s radiation therapy (RT)-treatment on an age-scaled phantom with a simple atlas-based heart. We recently enhanced our phantom by adding six CT-based heart models from international reference phantoms; one model was identified (“new base-heart”) as anatomically most representative across the CCSS age-range (infant-adolescent). The purpose of this study was to examine the impact of using this enhanced heart model on estimating cardiac risk.

Methods: The CCSS includes 24,214 individuals diagnosed 1970-1999, median age at diagnosis 7.0 (range 0–20.9) years and at last follow-up 27.5 (range 5.6–58.9) years. For those treated with RT (n=11,667), mean heart dose D(m), V5, and V20 were calculated for six heart models. We evaluated dose-response relationships using piecewise-exponential models, adjusting for attained age at evaluation, sex, diagnosis age, race, smoking history, diagnosis year, and chemotherapy exposure. Each individual’s D(m), V5, and V20 were assigned using: [1]the new base-heart and [2]a heart model matched to closest age/sex.

Results: For new base-heart, relative rates (RR) of any cardiac disease increased linearly with D(m)=10Gy (P<0.001), and both V5,?20?0=50% and V20<30% were associated with elevated risks (RR=1.4; 95% CI=1.0-2.1) and (RR=1.6; 95% CI=0.9-2.8), respectively; risks captured by RRs from new base-heart, when compared to atlas-based heart, were larger for D(m), similar for V5, and smaller for V20. The RRs calculated using the new base-heart and age/sex-matched hearts agreed on average within 10%, suggesting robustness.

Conclusion: With an anatomically realistic heart, cardiac disease risk increased linearly with D(m), V5, and V20. However, further investigation of substructure dose response is warranted because of the complex nature of the changes with D(m), V5, and V20 compared to previous atlas-based heart.

Funding Support, Disclosures, and Conflict of Interest: This work was supported by National Cancer Institute (CA55727, G.T. Armstrong, PI), Cancer Center Support (CORE) grant (CA21765, C. Roberts, PI), American Lebanese-Syrian Associated Charities (ALSAC), Marilyn and Frederick R. Lummmis, Jr. MD Fellowship in Biomedical Sciences, Sylvan Rodriguez Foundation Scholarship.


Dose Response, Phantoms, Radiation Risk


TH- Response Assessment: Modeling: other than machine learning

Contact Email