ESTRO 37 Abstract book

S635

ESTRO 37

Results After a minimal follow-up of 72 months, 23 of 210 (11.0%) patients developed local recurrences. Tumors with WHO type I pathology had significantly poor local control than those with WHO type II/III (5-year rate = 60.0% vs. 90.6%, P=0.0088). The 10-year local control rate in patients with T4 disease was lower than those with T1-3 disease (81.6% vs. 91.2%, P=0.0719). The 10-year local control rates in patients with post-IndCT biopsy showing absence and presence of viable tumor cells were 91.1% and 80.6%, respectively (P=0.0830). The interval between IndCT and RT (> vs. ≦ 7 days), RT dose (= vs. > 70 Gy), RT technique (IMRT vs. non-IMRT), and use of concurrent chemotherapy (yes vs. no) did not affect local control (P = 0.5487, 0.1109, 0.6673, and 0.8330). Conclusion Under our IndCT + RT protocol, long-term local control for advanced NPC patients is encouraging. Patients with WHO type II/III, T1-3 disease and post-IndCT biopsy showing no viable tumor cells had relatively higher local control rates. EP-1128 Proton Therapy in adenoid cystic carcinoma at the West German Proton Therapy Center Essen (WPE) B. Winckler-Saleske 1,2,3 , T. Steinmeier 2,3 , D. Ahmad Khalil 1,2,3 , P.H. Kramer 2,3 , D. Geismar 1,2,3 , B. Timmermann 1,2,3 1 University Hospital, Clinic for Particle Therapy, Essen, Germany 2 University Hospital, West German Proton Therapy Center Essen WPE, Essen, Germany 3 University Hospital, West German Cancer Center WTZ, Essen, Germany Purpose or Objective Proton Therapy (PT) is used for the treatment of adenoid cystic carcinoma as it has the potential to restrict radiation doses to the target volumes while sparing organs at risk. We report early data on feasibility and outcome after PT. Material and Methods Between June 2015 and August 2017, 21 patients (20 adults, 1 child; 15 male, 6 female), median 56.8 years (range, 11.3-81.5 years) with adenoid cystic carcinomas (ICD-O-3 morphological code 8200/3) were included in the prospective register studies. Neoplasms originated from paranasal sinus (42.9%), glandula (gl.) parotis (14.3%), oral cavity (14.3%), nasopharynx (9.5%), gl. lacrimalis (9.5%), nasal cavitiy (4.8 %) and gl. submandibularis (4.8%). Two patients were treated for a recurrence in the brain, or in the paranasal sinus, respectively. One patient presented with disseminated disease (lung). In 57.1% of the cases, patients were presenting with skull base infiltration. Two patients received concomitant chemotherapy, one due to his young age, one due to squamous cell carcinoma differentiation. PT was delivered as definitive (52.4%), adjuvant (42.9%) or palliative (4.8%) treatment. In 2 patients neck nodes were irradiated. Either pencil beam (57.1%) or uniform scanning (38.1%) or both (4.8%) was used. Median applied dose was 70 Gy (range, 60-71 Gy), administered in daily fractions of median 2 Gy (range, 1.65-2.2 Gy), 5x/week. Side effects were documented according to CTCAE v4.0 before, during and after PT. Results Median follow-up (FU) time after last fraction was 0.9 years (range, 0.0-1.7 years). During PT, dermatitis radiation, oral mucositis and fatigue of grade 2 were documented in 9, 6 and 4 patients, respectively. In 2 patients, oral mucositis grade 3 toxicities occurred. In 2 patients, preexisting grade 4 optic nerve disorders were reported already before PT. 90.5% of patients finished PT without interruption of more than three days. Three months after PT, FU data was available on 14 patients.

tumour volume (GTV PET ). Contours were derived using sixteen automated segmentation methods (PET-AS), including the machine learned method ATLAAS. The MTV, as well as the Dice Similarity Coefficient (DSC) of the PET-AS contours, was calculated in comparison to the clinician-derived GTV PET . Results Figure 1 shows the mean DSC for the sixteen PET-AS methods used to contour the MTV compared to the GTV PET . The mean DSC (+/- SD) for ATLAAS, 60% Peak Thresholding (PT60), Adaptive Thresholding (AT), General Clustering Means (GCM3) and Watershed Thresholding (WT) was 0.72 (+/- 0.10), 0.61 (+/- 0.20), 0.63 (+/- 0.15), 0.55 (+/- 0.20), 0.60 (+/- 0.21) respectively. The mean MTV [range] derived by ATLAAS was 6.01 [1.3 – 24] mL, PT60 8.66 [3.28 – 44.01] mL, AT 3.85 [1.30 – 8.75] mL, GCM3 8.36 [1.78 – 32.20] mL, WT 7.20 [0.54 – 27.55] mL. The mean MTV of the GTV PET was 6.22 [1.12 – 21.25] mL. A Kruskall Wallace test showed significant difference (P = 0.0003) between the GTV PET MTV and the MTVs delineated by ATLAAS, PT60, AT, GCM3 and WT PET-AS methods.

. Conclusion

Compared to other PET-AS methods, ATLAAS had the highest DSC, smallest SD and the volume was closest to the GTV PET , demonstrating machine learning can improve the robustness and accuracy of automated segmentation in H&N cancer. ATLAAS segmentation is suitable for GTV PET delineation in post-treatment 18 F-FDG PET imaging in H&N cancer. EP-1127 Prognostic factors analysis in patients with advanced NPC received induction CT followed by RT C. Wu 1 , W.Y. Wang 2 , Y.C. Liu 3 , J.C. Lin 3 1 Changhua Show-Chwan Memorial Hospital, Department of Radiation Oncology, Changhua, Taiwan 2 Hung Kuang University, Department of Nursing, Taichung, Taiwan 3 Taichung Veterans General Hospital, Department of Radiation Oncology, Taichung, Taiwan Purpose or Objective To investigate the prognostic factors affecting local control in patients with advanced nasopharyngeal carcinoma (NPC). Material and Methods We retrospectively review 210 previously untreated NPC patients who received a uniform protocol of induction chemotherapy (IndCT) + radiotherapy (RT). Baseline characteristics of patients were median age 45 (range 24- 74), male/female=156/54, WHO I/II/II pathology= 5/170/35, stage II/III/IV=10/68/132. The IndCT consists of cisplatin 60 mg/m2, alternating with 5-fluorouracil 2500 mg/m2 plus leucovorin 250 mg/m2 (P-FL) by an outpatient weekly schedule for a total of 10 weeks. RT ≥ 70 Gy by conventional fractionation was delivered one week later. Eleven of 210 patients also received cisplatin-based concurrent chemotherapy during RT.