This house believes that the use of Functional Imaging for treatment planning of head and neck tumors needs to be carefully considered. IMRT for Head and Neck Tumors Vincent GREGOIRE, M.D., Ph.D., Hon. FRCR Head and Neck Oncology Program, Radiation Oncology Dept. & Center for Molecular Imaging and Experimental Radiotherapy, Université Catholique de Louvain, St-Luc University Hospital, Brussels, Belgium Larynx PRV Spinal cord PTV 55.5 Gy Right parotid Left parotid PTV 69 Gy Oropharyngeal SCC T2-N0-M0 SIB-IMRT: 30x2.3 Gy 30x1.85 Gy This house believes that Heterogeneity in H&N TV delineation RO will be (even more) multidisciplinary RO is conformal (e.g. IMRT, proton, hadrons) RO will be tailored (individualized) (based on imaging and molecular profiling) and adaptive RO will be associated with targeted agents Harari et al., 2004
5 cm 5 cm 5 cm Target selection and delineation Betrayal of images This is not an apple R. Magritte Segmentation methods: one by lab! Manual delineation Visual interpretation Subjective approach! Histogram-based Fixed threshold (SUV, %) Adaptive threshold (SBR) Soft thresholding (probabilistic) Feature-based (Fuzzy) clustering Clustering of TAC Image-based Probabilistic modeling Gradient based Experimental conditions Data sets Validation studies Visual rendering Complex model unrecoverable target NO CONSENSUS Variability in PET image segmentation for H&N tumors Macroscopy CAT Scan 18F-FDG PET Schinagl et al., 2007 Daisne et al, 2003
Raw image Patient 1 2 3 4 5 6 7 8 9 mean From PET image to tumor: a fixed threshold? Surgical Specimen 8.3 5.2 30.9 4.1 5.6 8.6 17.3 15.4 24.3 13.3 GTV (ml) 40% 23.3 10.9 29.7 8 37.9 9.7 35.9 41.8 39.9 26.5* 50% 11.8 6.7 21.7 5.3 19.2 6.6 25.5 28.3 30 17.2 55.1 55.7 38.6 57 70.9 43.2 59 67.5 56.5 55.9 0 Gy 50 Gy Image-Guided Radiation Therapy in HNSCC The 4th dimension FDG-PET Geets et al, 2006 Geets et al, 2003 PET image segmentation during RxTh UG 4mm PET in radiotherapy planning for NSCLC Pathological correlation? Clinical validation? SBR Image processing Image segmentation BG 6mm + deconvolution W&C NSCLC with atelectasis J. Lee & X. Geets, 2005 Deruysscher, 2007
Images Methods Figures of merit NCAT Zubal Clinical image used as an example and associated segmentation NURBS surface NSCLC digital MC PET phantom Different segmentation methods PET in radiotherapy planning for NSCLC Werner-Wasik et al., 2011 Wanet et al., Radioth. Oncol. 20 PET in radiotherapy planning for NSCLC Objective evaluation of the available segmentation methods Statistical methods Simulated PET images (MC) Gradient detection Pathology validated patient images + + = Scanner model 3D phantom 3D tumor model Simulated 3D PET Precision Shape/volume Accuracy Reproducibility/robustness Wanet et al., Radioth. Oncol. 20 TG211, 2011
PET imaging: a wide range of molecular probes Glucose metabolism 18F-FDG Protein synthesis 11C-MET 18F-FET Proliferation 18F-FLT Hypoxia 18F-FMISO 18F-FAZA Receptors 18F-FES 68GA-Traztuzumab (HER2) 64Cu-DOTA-panitumumab (EGFR) proliferation A word of caution hypoxia perfusion 5 µm Courtesy of Geets, 201 From Kaanders., 2001 Biological heterogeneity [ 18 F]-FDG TEP Registered autoradiography Effect of resolution 2,5 mm 2,7 mm 2,0 mm 3,0 mm AR 100 µm 1,5 mm 3,5 mm Résolution 2.3 mm Résolution 0.1 mm
Dice Similarity Index (%) In Vivo experimental set-up PET-tracer, Hoechst Pimonidazole Tumor sectioning FAZA autoradiogram & pimonidazol SiHa SCCVII Micro PET Autoradiography (2-D distribution of FAZA) SiHa UT-SCC-33 Immuno-fluorescence microscopy (hypoxia, vessels, perfusion) Busk et al Acta Oncol, 2008;47:1201-121 Busk et al IJRBOP 2008;70:1202-12 Busk et al Acta Busk Oncol;, et al Acta 2008;47:1201-1210 Oncol, 2008;47:1201-12 Busk et al IJRBOP 2008;70:1202-12 Scaling issue Scaling issue Tumor: FSA II Mouse n 2_1 Tum Volume: 2.30 ml 18 F-FDG 300 µci 100 90 80 70 Volume corresponding to the highest 10 percent of activity AR Volume 60 50 40 30 20 FSA II (n=5) SCC VII (n=5) FSA II + RT (n=5) PET Volume 10 0 10 20 30 40 50 60 70 80 90 100 Mismatch ( / ) % of Overall Tumor Volume N. Christian, 2007
Effect of resolution PET imaging: a wide range of molecular probes r ² 0.88 0.84 Mosaic PET 0.86 0.87 0.86 0.84 % vol Mouse T ø 0.0 7.0 8.7 10.0 11.0 12.1 12.7 13.1 13.9 14.5 15.0 Human T ø 0.0 13.9 17.5 20.1 22.1 23.8 25.3 26.6 27.8 28.9 30.0 Glucose metabolism 18F-FDG Protein synthesis 11C-MET 18F-FET Proliferation 18F-FLT Hypoxia 18F-FMISO 18F-FAZA Receptors 18F-FES 68GA-Traztuzumab (HER2) 64Cu-DOTA-panitumumab (EGFR) Courtesy of Geets, 201 Comparison 18 F-FDG / 14 C-EF3 Comparison 18 F-FDG / 14 C-EF3 Anova: p = 0.19
Dose painting by number Dose painting : the physics issue Flat dose Non-flat dose Mean Tumor Dose = 2 Gy Survival is non-flat (higher in resistant areas) More similar survival across entire tumor Courtesy of D. De Ruysscher Molecular imaging dose painting by number Tomotherapy Hi-Art H&N SCC: T4N2bM0 60 Gy + SIB of 30 Gy Hypoxia (Cu-ATSM) Molecular imaging dose painting by number Dose escalation protocol DPBN based on FDG- PET Median dose of 80.9 Gy (n=7) et 85.9 Gy (n=14) No grade 4 acute toxicity Deveau et al., 2010 Duprez et al., 2010
Functional imaging for H&N treatment planning Great potential but Don t trust your physician Rely on user s independent methods Be aware of the various conceptual limitations Never forget the clinical knowledge At the end, the proof is in the pudding The Bridge at Argenteuil C. Monet, 1874