ESTRO 2021 Abstract Book

S290

ESTRO 2021

TP FP TN FN Sensitivity Specificity

Linear

43 2 148 8 0.84

0.99

Gaussian 27 0 150 24 0.53

1.0

Quadratic 46 13 137 5 0.9

0.91

Cubic

45 87 63 6 0.88

0.42

Conclusion We have shown that it is feasible to predict the shape and volume of NSCLC tumours from routine CBCTs and effectively identify patients with large tumour shrinkage early, giving departments time to allocate resources for ART. OC-0395 Ex vivo validation of PET imaging for response based dose-painting of NSCLC T. Meijer 1 , J. de Swart 2 , C. Groenendijk 3 , L. Maat 4 , R. Kanaar 5 , D. van Gent 5 , J. von der Thüsen 1 , S. Petit 6 1 Erasmus MC, Department of Pathology and Clinical Bioinformatics, Rotterdam, The Netherlands; 2 Erasmus MC, Department of Radiology and Nuclear Medicine, Rotterdam, The Netherlands; 3 TU Delft, Department of Radiation Science and Technology, Delft, The Netherlands; 4 Erasmus MC, Department of Cardio-Thoracic Surgery, Rotterdam, The Netherlands; 5 Erasmus MC, Department of Molecular Genetics and Oncode Institute, Rotterdam, The Netherlands; 6 Erasmus MC Cancer Institute, Department of Radiotherapy, Rotterdam, The Netherlands Purpose or Objective Response based dose-painting could be a pragmatic approach for dose-painting, but would require measuring tumor response non-invasively during treatment e.g. using PET. However, validation of PET imaging for intra- tumor response measurements in 3D is extremely challenging due to lack of a ground truth. Therefore the aim of this study was to set up an experimental procedure that can be used to evaluate the potential of (novel) PET tracers for response assessment of NSCLC tumor tissue ex vivo , without requiring animal experiments or patients undergoing extra procedures. The method was tested using slices of resected human lung tumor that were kept viable ex vivo , and that were administered 18F-fluorodeoxyglucose (FDG), as proof of principle. Materials and Methods Residual fresh NSCLC tissue was prospectively collected from lobectomy specimens and cut in 300 µm thick slices that were cultured, at 5% CO2 at 37 °C (atmospheric oxygen). Next 3.4 – 8.1 MBq FDG was added to the culture medium, incubated on an orbital shaker and washed. Five slices were treated with 30 µg/mL cisplatin to kill all cells before FDG administration, to rule out signal from sticky FDG. The slices were imaged using a micro PET scanner (Siemens Inveon using OSEM3D and MAP). The PET signal of the slices was compared to a Wallac Wizard 1480 gamma counter. Next, tumor slices were fixed and embedded in paraffin. Double immunohistochemistry staining was performed using primary antibodies against pan-keratin and Ki-67 (as surrogate for proliferation). An automated procedure was developed, trained and validated to detect Ki-67 positive cells. Linear multilevel mixed-effects regression analysis was performed to correlate the FDG uptake of the slices with the number of vital tumor cells within the slices (Figure 1).