ESTRO 38 Abstract book

S539 ESTRO 38

Material and Methods All HNC patients recruited to the VoxTox study (UK CRN ID 13716) were eligible. All were treated with standard radical protocols on TomoTherapy with daily IG. Swallowing OARs (ipsi & contralateral parotids – IPG & CPG, submandibular glands – ISMG & CSMG, superior & middle pharyngeal constrictors – SPC/MPC, oral cavity – OC and supraglottic larynx – SGL) were segmented on planning scans, according to published atlases. An open-source deformable image registration toolbox (Elastix), in-house dose calculation software (CheckTomo), planning kV and IG-MVCTs were used to compute D P and D A OAR DVHs. Mean dose was used for analysis. Toxicity was assessed with CTCAEv4.03 at baseline, and 1 year, using 3 endpoints: xerostomia (Xer), salivary duct inflammation (SDI), dysphagia (Dys). Baseline data on comorbidities, primary site (PDS), TNM staging, surgery, smoking, alcohol and systemic therapy (SACT) were collected. Univariate analysis was undertaken to examine relationships between baseline parameters, D P , D A , and Gr2+ toxicity. Odds ratios (ORs) were calculated for all variables. Dose differences between cases with and without Gr2+ toxicity were assessed with t-tests. AUC of ROC curves were calculated for dosimetric parameters. Results 141 patients had full datasets available for analysis. PDS were: 90 oropharynx, 14 hypopharynx/larynx, 16 salivary gland/sinus/skin, 10 oral cavity, 5 nasopharynx, 6 CUP. Concomitant SACT: cisplatin 78, cetuximab 11, none 52. Mean dose differences (D A – D P , 95% CI) for each OAR were: IPG 1.58Gy (1.36-1.81), CPG 0.93Gy (0.71-1.15), ISMG 1.00Gy (0.84-1.15), CSMG 1.23Gy (1.07-1.39), SPC 0.92Gy (0.80-1.03), MPC 0.73Gy (0.55-0.92), OC 0.62Gy (0.46- 0.79), SGL 0.95Gy (0.71-1.20) (Fig 1).

Conclusion A NTCP model based on delivered dose to SRR 0.01 and HTN was generated and validated for RB. External validation would be desirable. FEA provides a more anatomically representative tool for dose accumulation than planar homogeneous expansion. Improved toxicity prediction based on delivered dose could be useful for decision making in adaptive RT. PO-0984 Univariate toxicity associations are stronger with delivered than planned dose in HNC patients. D. Noble 1 , K. Harrison 2 , A. Hoole 3 , M. Wilson 4 , S. Thomas 3 , A. Bates 5 , L. Shelley 6 , N. Burnet 7 , R. Jena 1 1 University of Cambridge, Oncology, Cambridge, United Kingdom ; 2 University of Cambridge, Cavendish Laboratory, Cambridge, United Kingdom ; 3 Cambridge University Hospitals NHS Foundation Trust, Department of Medical Physics and Clinical Engineering, Cambridge, United Kingdom ; 4 University College London, Department of Medical Physics and Biomedical Engineering, London, United Kingdom ; 5 Cambridge University Hospitals NHS Foundation Trust, Cambridge Clinical Trials Unit- Box 401, Cambridge, United Kingdom ; 6 University of Cambridge, Engineering, Cambridge, United Kingdom ; 7 University of Manchester, Manchester Academic Health Science Centre, Manchester, United Kingdom ) dose to OARs in patients with HNC are well documented; adaptive strategies may reduce such differences, but lack an evidence base. To date, no data exists proving that patient toxicity associates more strongly with D A , compared to D P . This study aims to test that hypothesis. Purpose or Objective Differences between planned (D P ) and delivered (D A

Concomitant SACT increased the risk of all toxicity endpoints (ORs, 95% CIs); Xer 2.22 (1.10–4.52), SDI 3.16 (1.41–7.55), Dys 3.33 (1.04–14.1), as did baseline symptoms for Xer (2.10, 1.01-4.27) and SDI (5.06, 1.68- 15.3), and higher N-stage (2+) for Xer (2.10, 1.04-4.27). Univariate relationships between dose to relevant OARs and toxicity are shown in Table 1.