Abstract Book

S165

ESTRO 37

the tract normal to a computed medial axis (MA), thereby enabling such analysis of regional effects along the tract length. The MA is calculated via a MA transform by thinning followed by a 3D parametric cubic spline fit. MD, FA, and RT dose data were mapped to the normal slices after restricting these data to the censored tracts for each patient. The sensitivity measure was calculated as the percent change in mean FA or MD per patient and slice divided by the mean RT dose per patient and slice, scaled by the mean over all the slices, and averaged over all 49 patients. Results Distinct patterns of FA/MD sensitivity to dose, relative to the tract’s mean, were seen for specific white matter structures, in particular at their terminal ends. For example, for the corticospinal tract, FA sensitivity at terminal ends, right hemisphere, was -39.1±28.5 and - 7.6±11.4 percent change per Gy. For the corresponding left hemisphere, this FA sensitivity was -29.5±32.0 and - 7.1±12.4 percent change per Gy. The sensitivity pattern persisted for corresponding tracts in both the left and right brain hemispheres (Figure 1). Structures which exhibited marked variation seen in both hemispheres included the corticospinal tract, medial lemniscus, and inferior cerebellar peduncle. Conclusion Our analysis suggests that some tracts exhibit significant local variations in sensitivity of diffusivity changes to radiation dose. These results could not have been obtained via axial slicing of the tracts, as very curved or hooked tracts would be sampled in two locations at once during axial slicing, regardless of rotation. Tract-Crawler is a novel tool to visualize and analyze white matter structures.

rates were measured 6 weeks, and 1 year after therapy, and converted into the percentage of baseline. SMG data were fitted to log regression and to the Lyman-Kutcher- Burman (LKB) model, with a complication defined as <25% of the flow before therapy 1 . Thirty patients were treated after unilateral neck dissection. Patients with N2c or N3 nodes were excluded, 38% were N 0 , 32% T 3-4 . Tumors were localized in the oropharynx in 69%, larynx in 14%, nasopharynx in 9%, oral cavity in 4%, and hypopharynx in 3%. Treatment consisted of conventional RT, non SMG sparing IMRT, SMG sparing IMRT, in 34, 84, 82 patients, respectively. The mean dose to the ipsilateral submandibular gland was 62Gy (94%>40 Gy), for the contralateral submandibular gland 47 Gy, range 0-72 Gy. Flow rates were converted to the baseline unilateral SMG flow rate, and compared with the contralateral SMG mean dose. Results There was a highly significant correlation between increasing mean SMG dose and decreasing ab solute and relative SMG flow after 6 weeks and 1 year. In 5 mean contralateral SMG dose groups:≤30 Gy (n=30), 30-40 Gy (n=28) 40-50 Gy (n=32), 50-60 Gy (n=57), > 60 Gy (n=53), the complication rate at 6 weeks was 22%, 50%, 73%, 76% and 86%, respectively; at 1 year 30%, 31%, 75%, 86% and 94%, respectively (p<0.0001). The D50 after 6 weeks was 31 Gy (95% CI 24-37) and 38 Gy after one year (95% CI 35-42) after analysis with the LKB model; 33 Gy (27-40) and 36 Gy (31-42), using logistic regression. Submandibular dose response curve 6 weeks and one year after therapy: Conclusion This NTCP-curve for submandibular gland function is based on the largest database of objective measurements in the literature. External validation of the NTCP-curve will be performed. For planning purposes usually a threshold of 40 Gy is chosen. Based on our results the planning constraint for mean submandibular gland dose should be set to less than 30 Gy, preferably 20 Gy. 1: Dijkema et al. Int J Radiat Oncol Biol Phys 2010:78;449-453. PV-0315 Comparison of NTCP models between using planned and actual delivered dose for head and neck cancer T. Kanehira 1 , S.R. Van Kranen 1 , O. Hamming-Vrieze 1 , T. Janssen 1 , J.J. Sonke 1 1 Netherlands Cancer Institute, Department of Radiation Oncology, Amsterdam, The Netherlands Purpose or Objective Normal tissue complication probability (NTCP) models are typically derived from the planned dose distribution. The planned dose, however, often deviates from the delivered dose due to both daily setup and anatomical variations. The purpose of this study was to investigate the impact of the difference between the planned and the actual delivered dose on NTCP models for head and neck cancer patients.

PV-0314 Model Based Radiotherapy: Submandibular Dose-Response NTCP-curve Based on Objective Measurements. C. Terhaard 1 , J. Vermaire 1 , T. Dijkema 2 , M. Philippens 1 , P. Braam 2 , J. Roesink 1 , C. Raaijmakers 1 1 UMC Utrecht, Radiation Oncology Department, Utrecht, The Netherlands 2 UMCN Radboud, radiation oncology, Nijmegen, The Netherlands Purpose or Objective Submandibular gland (SMG) normal tissue complication probability (NTCP) curves will be part of model based indication for proton therapy in the Netherlands. Subjective measurements of xerostomia and sticky saliva reflect whole saliva, and are not suitable for the determination of the SMG NTCP curve. We performed direct measurements of salivary flow, including the SMG flow. Based on a large data base with a broad mean SMG dose distribution we obtained NTCP curves 6 weeks and 1 year after therapy. Material and Methods We utilized dose-response data of 200 head-and-neck patients, treated between 1999 and January 2017 in prospective salivary gland function studies. SMG flow