Click here to


Are you sure ?

Yes, do it No, cancel

Impact of Parotid Gland Migration On Xerostomia Prediction

R Astaburuaga1,2,3,4*, H Gabrys1,2,3 , B Sanchez-Nieto4,5, R O Floca3,6 , S Klueter3,7, K Schubert3,7, H Hauswald3,7,8, M Bangert1,3, (1) Department of Medical Physics in Radiation Oncology, German Cancer Research Center (DKFZ), Heidelberg, Germany, (2) Medical Faculty of Heidelberg, Universitat Heidelberg, Heidelberg, Germany, (3) Heidelberg Institute for Radiation Oncology , Heidelberg, Germany, (4) Institute of Physics, Pontificia Universidad Catolica de Chile, Santiago, Chile, (5) Center UC Investigation in Oncology, Pontificia Universidad Catolica de Chile, Santiago, Chile, (6) Medical Image Computing, German Cancer Research Center (DKFZ), Heidelberg, Germany, (7) Department of Radiation Oncology, University Hospital Heidelberg, Heidelberg, Germany, (8) Clinical Cooperation Unit Radiation Oncology, German Cancer Research Center (DKFZ), Germany


(Thursday, 8/2/2018) 7:30 AM - 9:30 AM

Room: Davidson Ballroom A

Purpose: To predict late grade two xerostomia (G2x) following radiotherapy under the impact of parotid gland (PG) migration.

Methods: A cohort of 139 head and neck cancer patients was used to study the association and interdependence of the planned mean dose (PMD) to PGs, the average dose gradient within PGs in the lateral direction (gradâ‚“), and PG migration with regard to G2x at 6-36 months after radiotherapy. PG migration one month after the treatment start was estimated for a subset of 88 patients. It was quantified based on daily MVCTs, using the volumetric change of the external contour at the level of the C2 vertebral body (Barker et. al. 2004).

Results: For clarity, results only cover the contralateral PG.PMD failed to predict G2x (AUC=0.57). Instead, G2x patients were well recognized by gradâ‚“ (AUC=0.72). We hypothesize that gradâ‚“ predicts G2x as a proxy of the increase in the delivered mean dose during treatment. For random setup errors, a sampling study confirmed a strong linear relation between uncertainty in the delivered mean dose and gradâ‚“ (correlation coefficient). In order to differentiate between the patients where PG migration may have a higher and lower impact on the delivered mean dose, the subset was split into high- and low-gradient groups. Supporting our hypothesis, we found that (1) the G2x patients with low gradx were successfully recognized by the PMD (AUC=0.91) and (2) the G2x patients with high gradx were well recognized by PG migration (AUC=0.77) and gradâ‚“ (AUC=0.75).

Conclusion: This work provides evidence that xerostomia prediction models may benefit from information about inter-fractional PG migration during radiotherapy. It motivates further work on image-guided NTCP modelling by relating treatment outcomes with the delivered dose instead of the planned dose. An ongoing patient-specific reconstruction of the delivered mean dose may further substantiate our findings.

Funding Support, Disclosures, and Conflict of Interest: Funding Support: DAAD Scholarship


NTCP, Organ Motion, Setup Errors


TH- response assessment : CT imaging-based

Contact Email