ESTRO 36 Abstract Book

S324 ESTRO 36 _______________________________________________________________________________________________

26 German Cancer Research Center - Heidelberg, DKTK partner site Tübingen, Tübingen, Germany 27 Faculty of Medicine and University Hospital Tübingen- Eberhard Karls Universität Tübingen, Department of Radiation Oncology, Tübingen, Germany 28 TU Dresden - Med. Faculty Carl Gustav Carus, Institute of Pathology, Dresden, Germany 29 TU Dresden - Med. Faculty Carl Gustav Carus, Tumour- and Normal Tissue Bank- University Cancer Centre UCC, Dresden, Germany 30 TU Dresden - Med. Faculty Carl Gustav Carus, Medical Systems Biology - University Cancer Centre UCC, Dresden, Germany Purpose or Objective A gene signature predicting loco-regional control (LRC) of locally advanced head and neck squamous cell carcinoma (HNSCC) after postoperative radiochemotherapy (PORT- C) will be evaluated using nanoString and RNA microarray data. The prognostic power of the signature as well as the correlation between both methods is evaluated to underline the robustness of the proposed signature. Material and Methods Gene expression analyses were performed using nanoString technology and the GeneChip® Human Transcriptome Array 2.0 (Affymetrix) on a multicentre retrospective patient cohort of 191 patients with HNSCC who received postoperative radiochemotherapy. The nanoString gene expression panel of 209 genes was composed hypothesis-driven, including genes which are involved in proliferation, invasion and metastasis as well as in radio(chemo)resistance associated with tumour hypoxia, cancer stem cell markers, cisplatin-resistance and DNA repair. A gene signature which optimally predicts LRC was extracted from the nanoString gene expression data. Different statistical methods for signature selection and outcome prediction were compared. In parallel, this gene signature was evaluated using gene expression data of the GeneChip® Human Transcriptome Array analyses. The prognostic performance of both methods, measured by the concordance index (CI), was compared. Results The extracted nanoString gene signature contained genes related to cellular proliferation, migration, invasion, and tumour hypoxia. From the different statistical methods, Cox regression performed best and was chosen for outcome prediction. Internal 3-fold cross validation during model building showed a CI≈0.7, indicating a good performance of the model. Evaluating the signature using the gene expression data generated with the GeneChip® Human Transcriptome Array led to similar results. The expression values of each gene within the signature were significantly correlated between nanoString and RNA microarray data with R>0.4. Conclusion We determined a gene signature for the prediction of LRC in a cohort of 191 patients with locally advanced HNSCC after postoperative radiochemotherapy based on nanoString gene expression data. The signature showed a good prognostic value and was validated by internal and external validation. Using gene expression data from the GeneChip® Human Transcriptome Array a similar prognostic value was obtained, underlining the robustness of the proposed signature. PO-0620 Partial Laryngeal IMRT for T2N0 Glottic Cancer: Impact of Image Guidance and Radiotherapy Regimen K. Rock 1 , S. Huang 1 , A. Tiong 1 , L. Lu 2 , W. Xu 2 , A. Bayley 1 , S. Bratman 1 , J. Cho 1 , M. Giuliani 1 , A. Hope 1 , J. Kim 1 , J. Ringash 1 , B. O'Sullivan 1 , J. Waldron 1 1 Princess Margaret Cancer Centre, Department of Radiation Oncology, Toronto, Canada

2 Princess Margaret Cancer Centre, Department of Biostatistics, Toronto, Canada

Purpose or Objective Since 2006 we have used partial laryngeal IMRT due to the ease of sparing the contralateral carotid and arytenoid cartilage. Since then the protocol has undergone two changes. Initially, the matching surrogate for image guidance changed from cervical vertebrae bone (IGRT- bone) to laryngeal soft tissue (IGRT-larynx). Secondly, the IMRT dose/fractionation for T2N0 glottic cancer changed from hypofractionation (RT-hypo) 60 Gy in 25 fractions in 5 wks (60 Gy/25f/5w) to moderate accelerated (RT-acc) 66-70 Gy/33-35f/5.5-6w. This study investigates the impact of these two changes on local control (LC) following partial laryngeal IMRT for T2N0 glottic cancer patients. Material and Methods All clinical T2N0 glottic patients receiving partial laryngeal IMRT from Jan 2006 to Dec 2013 were reviewed. GTV was delineated based on endoscopic/radiological findings. CTV1 was (GTV +0.3-0.5cm) to high dose and CTV2 (GTV +1cm) to elective dose. PTV was CTV + 0.5cm expansion axially but 1cm superior and inferiorly. LC, overall survival (OS), and grade 3-4 late toxicity (LT) were compared between IGRT-bone vs IGRT-larynx, and RT-hypo vs RT- acc. Univariable analysis (UVA) was used to identify potential factors for local failure (LF). Variables studied included IGRT matching technique (IGRT-bone vs IGRT- larynx), tumor volume and extension (supra- vs sub- glottic), RT regimen, anterior commissure involvement and smoking status. Since IGRT surrogate and RT regimen were highly correlated, two separate multivariable analysis (MVA) models were constructed. Results A total of 139 patients were identified. T2 category was assigned as follows: supra- or sub-glottic extension (54%); impaired vocal cord mobility (47%); or both (49%). Anterior commissure involvement was present in 72% of patients. IGRT-larynx and IGRT-bone were used in 92 (66%) and 47 (34%) patients, respectively. RT-hypo and RT-acc were given to 71 (51%) and 68 (49%) patients, respectively. Median follow-up was 5.03 yrs. A total of 28 local (IGRT- bone:15/47, IGRT-larynx:13/92), 6 regional, 2 distant failures were identified. Compared to IGRT-bone, IGRT- larynx had higher 5 year LC (85% vs 68%, p=0.02) and OS (87% vs 65%, p=0.007), but identical LT (7% vs 7%, p=0.73). Higher LC was also observed for RT-acc vs RT-hypo (89% vs 70%, p=0.008). UVA revealed that IGRT-larynx (HR 0.42, p=0.02), RT-acc (HR 0.33, p=0.01) were associated with reduced risk for LF while current smoker did not impact LC (p=0.49). MVA adjusted for GTV and smoking status confirmed that IGRT-larynx reduced risk of LF vs IGRT- bone (HR=0.40, 95% CI 1.2-5.3, p=0.02); RT-acc also reduced risk of LF (HR 0.34, 0.15-0.79, p=0.012) Conclusion Moderate accelerated IMRT with laryngeal soft tissue guidance for T2N0 glottic cancers results in high LC (>80%) with minimal toxicity. The transition from bone vertebrae to laryngeal soft tissue matching, together with accelerated schedules and higher biological equivalent dose is safe and potentially shows superior outcomes. PO-0621 Validation of tumor delineation on HE stained sections with cytokeratin staining as gold standard H. Ligtenberg 1 , S. Willems 2 , E. Jager 1 , C. Terhaard 1 , C. Raaijmakers 1 , M. Philippens 1 1 UMC Utrecht, Radiotherapy, Utrecht, The Netherlands 2 UMC Utrecht, Pathology, Utrecht, The Netherlands Purpose or Objective The extent of microscopic tumor growth determines the clinical target volume (CTV) in head-and-neck tumors. In imaging validation with histopathology, the CTV-margin needed to include all microscopic tumor, can be