ESTRO 2024 - Abstract Book

S2228

Clinical - Upper GI

ESTRO 2024

2089

Digital Poster

Development and validation of a risk model for local failure in pancreatic cancer after SBRT.

Eliana La Rocca 1 , Luisa Altabella 2 , Tiziana Rancati 3 , Giulia Volpi 1 , Gabriella Rossi 1 , Roberto Rossi 1 , Pier Giorgio Esposito 2 , Fabio Casciani 4 , Giuseppe Malleo 4 , Carlo Cavedon 2 , Renzo Mazzarotto 1 1 Azienda Ospedaliera Universitaria Integrata di Verona, Radiation Oncology Department, Verona, Italy. 2 Azienda Ospedaliera Universitaria Integrata di Verona, Medical Physics Unit, Verona, Italy. 3 Fondazione IRCCS Istituto Nazionale dei Tumori, Data Science Unit, Milan, Italy. 4 Pancreas Institute,University of Verona, General and Pancreatic Surgery Unit, Verona, Italy

Purpose/Objective:

Coverage of the Gross Tumor Volume (GTV) with high doses in SBRT of pancreatic cancer (PC) patients is often limited due to the proximity of highly radiosensitive organs at risk (OARs). This analysis aims to develop and validate a model predicting local failure (LF) based on the geometric extension of the GTV uncovered by the high dose prescription (50Gy).

Material/Methods:

We identified two cohorts of advanced or borderline PC patients in our institutional database. All patients received chemotherapy and SBRT in 5 fractions:-30 Gy to the Planning Target Volume(PTV);-50 Gy with a simultaneous integrated boost (SIB) to the region of vessel abutment/encasement;-25 Gy of simultaneous integrated protection (SIP) to the overlap between PTV and Planning OAR volumes.SBRT was delivered with a flattening filter free linac, in free breath after a daily ConeBeam CT. We trained the model on the first 50 treated patients. We used univariate/multivariable survival analysis (UVA/MVA, including Cox regression and Kaplan-Meier Curves evaluated through theLog-Rank test) to assess the association of LF with the GTV size and the amount of the GTV not covered in the SIB. Specifically, we defined a set of incomplete GTV-covering levels: GTV75 (=yes if less than 75% of the GTV was included in the 50 Gy SIB isodose) and GTV70, GTV50, and GTV30 in a similar way (Figure1). We included in MVA variables with p≤0.2 at UVA. We considered MVA models relevant if their Log Likelihood was significantly higher when compared to UVA (Likelihood Ratio Test). The validation cohort included 91 SBRT patients treated subsequently in our institution.