ESTRO 2020 Abstract Book

S152 ESTRO 2020

PD-0297 Estimating the probability of active tumor in cervical cancer derived from PET/MR-imaging L. Kravic 1 , H.E.S. Pettersen 1 , L.B. Hysing 1,2 , M. Daniel 3 , D. Georg 3 , E.M. Høye 1 , C. Ekanger 1 , D. Thorwarth 4 , S.M.C. Pilskog 1,2 1 Haukeland University Hospital, Department of Oncology and Medical Physics, Bergen, Norway ; 2 University of Bergen, Faculty of Mathematics and Natural Sciences, Bergen, Norway ; 3 Medical University of Vienna/AKH Vienna, Department of Radiation Oncology, Vienna, Austria ; 4 Eberhard Karls University Tübingen, Section for Biomedical Physics- Department of Radiation Oncology, Tübingen, Germany Purpose or Objective The aim of this study was to implement a multivariate logit model to derive the probability distribution of active cancer (PAC) in cervical cancer and to evaluate to which extent the PAC is affected by the different imaging parameters. Material and Methods Nine patients with locally advanced cervical cancer who underwent simultaneous multiparametric imaging (diffusion weighted MR, contrast enhanced MR and [ 18 F]FMISO PET) performed on a hybrid MR/PET (mp- MR/PET) scanner were analysed. Parameters (ADC, K trans and normalized PET-uptake values) derived from the mp- MR/PET images were analysed with a logit model to assess the PAC within the delineated GTV. To calibrate the logit model we reviewed literature for studies using mp-MR/PET parameters to discriminate between tumor and normal cervical tissue. From these studies we extracted classification cutoff values verified by pathology with their corresponding sensitivity and specificity as well as general details on the imaging- and patient data. For each patient, the voxelwise probability of active cancer (PAC) was calculated for the mp images utilizing the MICE Toolkit software with the multivariate logit model. Reviewed parameter thresholds (Table) and the reported sensitivity were used as weights to convert image intensity to PAC. PAC was derived for the GTV, on both the individual parameters and from a combination of all imaging parameters. Peaks in the histogram of the PAC distribution and the spatial distribution of PAC were compared for the nine patients. Results More than 145 papers with relevant input data for mp- PET/MR were reviewed (Table).

>500 cm 3 for acute toxicity, while none was correlated with late toxicity.

Conclusion Toxicity was mild and acceptable with high pts’ satisfaction. Local control was acceptable and overall survival was 97.8% at 3 years. In this series, the use of 3DCRT or IMRT for the treatment of elderly BC pts treated with adjuvant once-weekly hypofractionated RT did not differ in terms of toxicity and efficacy. Such a scheme was well accepted by pts, especially by those with difficulties in commuting and with disabling associated disease.

The combined model with ADC+FMISO+K trans resulted in a peak of PAC-values around 78-99%, which was both more distinct and shifted towards higher probabilities. An average PAC peak value of 98% for the combined model was obtained at 6 patients, while 3 patients had not a distinct peak in the combined model as well as for single parameters. Single parameter models from either FMISO, ADC, K trans had lower probabilities in the range 18-97%; 79- 96%; 17-96%, respectively, and the PAC histogram distributions varied substantially. In the six patients with a distinct peak of PAC for the combined parameter model, all single parameters contributed to that result, especially

Poster discussion: PH: Quantitative functional and biological imaging

PD-0296 FAPI-PET/CT as a novel diagnostic imaging probe – first clinical experience in radiation oncology

Abstract withdrawn