ESTRO 36 Abstract Book

S481 ESTRO 36 2017 _______________________________________________________________________________________________

Conclusion The availability of MRI images during target delineation of pancreatic cancer on 3DCT and 4DCT improved the interobserver variation. Furthermore, delineated volumes are smaller on CT+MRI than on CT only. An approach of 3DCT GTV delineation with margins may be preferable to 4DCT iGTV delineation since the latter increases uncertainty.

Poster: Physics track: (Quantitative) functional and biological imaging

PO-0885 Assess Tumor Voxel Dose Response (SF2) Using Multiple FDG PET Images D. Yan 1 , S. Chen 2 , D. Krauss 2 , P. Chen 2 , G. Wilson 2 1 Beaumont Health System, Radiation Oncology, Royal Oak MI, USA 2 Beaumont Health system, Radiation Oncology, Royal Oak- MI, USA Purpose or Objective To determine human tumor voxel dose response with using bio-marker (SF2) derived from multiple FDG PET images and evaluate the pattern of tumor local radioresistance and failure. Material and Methods Multiple FDG PET/CT images obtained at the pre- and weekly during the treatment (35x2Gy) for 15 HN cancer patients were used for the study. from 0 to 70Gy, for each tumor voxel, v, such that TMR(v, d) = SUV(v, d)/SUV(v, 0). TMR of each patient was constructed following voxel-by- voxel deformable PET/CT image registration. Assuming ln(TMR(v, d)) = k . ln(SF(v, d)), the optimal linear regression with unknown break-points was used to determine the tumor voxel survival fraction, SF at different treatment dose levels. Therefore, SF2(v, d), the tumor voxel survival fraction after 2Gy, was obtained and used as the tumor voxel dose response marker. Tumor voxel SF2 distribution at different treatment dose level was analyzed to evaluate the pattern of tumor voxel dose response, specifically the tumor local radioresistant pattern; meanwhile the effects of tumor voxel reoxygennation and proliferation with respect to tumor local failure were also evaluated. The tumor volumes with different SF2 cutoffs were also correlated to treatment outcome. Results Human tumor has significant inter- and intra-tumor variations in their voxel dose response (Table). Tumor voxels with the mean SF2 > 0.7 shows significant prediction power (p < 0.01) after the treatment dose of 20~40Gy on tumor local failure. Tumor local recurrence showed strong correlation to the tumor local radioresistance determined using SF2 (Figure). Two of 3 failures showed a clear reoxygennation effect after 20Gy, therefore could be potentially controlled by escalating dose to destroy the local radioresistant niches. One failure had a small (< 1cc) local radioresistant sub-volume and showed the minimal reoxygennation, therefore it may need a local ablation.

Conclusion Tumor voxel metabolic ratio determined us ing multiple FDG PET images can be used to assess tumor voxel dose response, SF2, which shows an excellent predictive value for tumor local radioresistance and failure. Our results demonstrate that a heterogeneous dose distribution in tumor should be prescribed to optimize cancer radiotherapy. Tumor voxel TMR determined at the early treatment will be good bio-parametric matrix to determine a new tumor dose prescription function at the voxel level for the treatment of adaptive dose-painting- by-number. PO-0886 Early changes of FDG-PET markers predict the outcome after chemo-radiotherapy for pancreatic cancer S. Broggi 1 , P. Passoni 2 , E.G. Vanoli 3 , C. Fiorino 1 , G.M. Cattaneo 1 , C. Gumina 2 , P. Mapelli 3 , E. Incerti 3 , L. Gianolli 3 , N. Slim 2 , M. Picchio 3 , R. Calandrino 1 , N.G. Di Muzio 2