ESTRO 2021 Abstract Book

S1246

ESTRO 2021

Figure 2. AUC to indicate the accuracy of the nomogram (original and externally validated).

Conclusion For the first time, a prediction model focusing on the timely implementation of innovations is successfully built and externally validated. This model can now be widely used to innovate more successfully in radiotherapy. References 1. Collins GS, Reitsma JB, Altman DG, Moons KGM. Transparent Reporting of a Multivariable Prediction Model for Individual Prognosis or Diagnosis (TRIPOD). Circulation. 2015 Jan 13;131(2):211–9. PO-1522 Pseudoprogression versus true progression in glioblastoma patients : A multiapproach analysis A. Laprie 1 1 Institut Claudius Regaud/Institut Universitaire du Cancer de Toulouse – Oncopôle, , radiation Oncology, Toulouse, France Purpose or Objective Differentiating early tumor progression (ETP) from pseudoprogression (PSP) in patients with glioblastoma (GBM) is crucial to improve prognosis. Objective of this study was to investigate the ability of multiapproach analysis including clinical data, diffusion weighted imaging (DWI), perfusion magnetic resonance imaging (MRI) and 3D magnetic resonance spectroscopic imaging (MRSI) to distinguish ETP from PSP in patients with GBM. Materials and Methods Among 180 patients included in prospective phase III Randomized SPECTROGLIO trial (NCT01507506)(1), 46 patients were suspected with progression within 6 months after end of radiotherapy. Ratios of choline/creatine (Cho/Cr), choline/N-acetyl aspartate (Cho/NAA) and lactate/N-acetyl aspartate (Lac/NAA) were extracted after co-registration of 3D-MRSI with the FLAIR and T1-weighted images before and after gadolinium administration. Voxels were analysed according to the region to which they belonged: contrast- enhanced (CE) lesion, necrosis and infiltrative oedema. Apparent diffusion coefficient (ADC) and cerebral blood volume (CBV) maps were calculated and registered to T1 CE MRI. Differences between ETP and PSP were evaluated using Mann-Whitney tests and p-values were adjusted with Bonferroni correction. Logistic regressions with backward elimination procedure and area under receiver operator characteristic curve (AUC) analyses were used to assess the predictive ability of imaging parameters. Overall survival (OS) were calculated from inclusion until death or last-follow-up. Results Based on the trial recommendations and RANO criteria on new MRI one month after progression suspicion, 28