ESTRO 36 Abstract Book

S921 ESTRO 36 _______________________________________________________________________________________________

parameters of 18F-FDG PET/CT image in patients with early stage non-small cell lung cancer (NSCLC) treated by stereotactic body radiation therapy (SBRT). Material and Methods Thirty patients with early stage NSCLC (T1-2N0M0) were retrospectively investigated. SBRT was delivered with total dose of 40-48Gy in 4 fractions for peripheral regions or 50-60Gy in 7-15 fractions for central regions or regions nearby other organ at risk. All patients underwent 18F- FDG PET/CT scan before treatment. Each tumor was delineated using PET Edge (MIM Software Inc., Cleveland) and texture parameters were calculated using open- source code CGITA (Fang, et.al., 2014). From 18F-FDG PET/CT image, three conventional parameters including metabolic tumor volume (MTV), maximum standardized uptake value (SUV) and total lesion glycolysis (TLG) and four textural parameters including entropy and dissimilarity derived from co-occurrence matrix and high- intensity large-area emphasis and zone percentage derived from size-zone matrix were analyzed. Reproducibility was evaluated using two independent delineation conducted by two observers using intraclass correlation coefficients (ICC). The ability to predict local control (LC) was tested for each parameter using Cox proportional hazards model. Results Median follow-up period was 30.1 month and 8 (23%) patients occurred local relapse. Between two observers, six parameters besides zone percentage (ICC value 0.59) showed ICC value ranged between 0.81 and 1.00. In univariate analysis, there were significant correlations between LC and tumor diameter>30mm (hazard ratio 7.21, p=0.02), MTV≥5.14cm3 (HR 9.38, p=0.01), TLG≥59.7 (HR 5.86, p=0.04), entropy≥-34.3 (HR 0.13, p=0.02), dissimilarity≥2235 (HR 6.87, p=0.03) and treatment biological equivalent dose≥100Gy (HR 0.02, p=0.04), respectively. Maximum SUV≥10.4 was not a significant predictor for LC (p=0.09).

needed.

EP-1688 Voxelbased analysis of FMISO-PET and diffusion-weighted MRI of two different HNSCC models in mice R. Winter 1 , S. Boeke 2 , M. Krueger 3 , A. Menegakis 2 , E. Sezgin 2 , L. Wack 1 , G. Reischl 3 , B. Pichler 3 , D. Zips 2 , D. Thorwarth 1 1 University Hospital Tübingen, Section for Biomedical Physics, Tübingen, Germany 2 University Hospital Tübingen, Radiation Oncology, Tübingen, Germany 3 Werner Siemens Imaging Center, Preclinical Imaging and Radiopharmacy, Tübingen, Germany Purpose or Objective Hypoxia is an important prognostic marker for radiotherapy (RT) response, particularly for head and neck squamous cell carcinoma (HNSCC) and may be measured using PET-tracers such as 18 F-FMISO. Moreover, parameters derived from functional MRI have been correlated with response to RT, such as ADC. Our hypothesis is that multiparametric PET/MRI, i.e. a combination of different parameters derived from PET and functional MRI, allows a better prediction in terms of RT response than single parameters do. The aim of this study was to distinguish two different HNSCC cell-lines grown as xenografts in mice, based on voxel-wise image analysis of simultaneously acquired FMISO-PET and ADC data. Material and Methods 11 immunodeficient nude mice were injected into the hind leg with tumor-cells of human HNSCC cell-lines FaDu (n=7) or CAL-33 (n=4). Once a tumor reached its target size (~300 mm³), simultaneous PET and MR imaging was performed on a 7T-PET/MR scanner (Bruker) at two time points: before (d0) and after two weeks (d14) of fractionated irradiation (10x 2Gy). The protocol included dynamic FMISO-PET (90min), anatomical T2- and diffusion- weighted MRI. An image of the FMISO uptake was reconstructed from the last 5 min of the acquired PET data. An ADC map was calculated from a set of 9 diffusion-weighted MR images (b=0-800 s⁄mm²). On the anatomical MR image, tumor and muscle were defined as regions of interest (ROIs). ROIs and ADC map were then resampled to the PET image grid for consistent image analysis on the voxel level. FMISO tumor- to-muscle-ratios (TMRs) were determined at both time points for ROI-based and voxel-by-voxel comparison with ADC values. Results The median (d0/d14) TMRmean was 1.43/1.06 and 1.25/1.00, median ADCmean was 780/929 and 1095/1286 x10 ⁻ ⁶ mm²/s, median FMISO TMRmax was 2.55/1.57 and 1.80/1.52, median slope m of a regression line through voxelbased FMISO TMR and ADC scatter data was -2.29/- 1.25 and 0.02/-0.26 x10 ⁻ ⁴ , median ADCmean of a thresholded subregion of the tumor where FMISO TMR≥1.4 was 730 (d0) and 1145 (d0) x10 ⁻ ⁶ mm²/s for FaDu and CAL- 33 tumor ROIs, respectively. Parameter values for all tumors are presented in Fig1; a scatter plot of voxelbased FMISO TMR and ADC values for one FaDu and one CAL-33 tumor at d0, in Fig2. Out of five parameters, three had strong potential for differentiation of the HNSCC cell-line, when measured at d0: TMRmax, slope m of the regression line and ADCmean of the FMISO positive region (TMR≥1.4). Conclusion Voxelbased analysis of FMISO-PET and ADC data proved to have high potential for discrimination of tumor cell-lines presenting different radiobiological properties. Three parameters were found to be suitable to distinguish the two cell-lines with well-known difference in

Conclusion Texture analysis based on gradient-based delineation method has high reproducibility in most parameters. Entropy and dissimilarity calculated from co-occurrence matrix is potentially beneficial to predict LC with reproducibility in patients with NSCLC treated by SBRT. To establish utility of texture analysis in 18F-FDG PET/CT image, further study including prospective trial will be