Response prediction to PRRT in progressing and metastatic GEP-NET undergoing restaging 68Ga-DOTA PET/CT: A preliminary multicenter radiomics study

#3112

Introduction: Radiomics aims to identify quantitative features from biomedical images, such as PET, also to improve the prediction of patient outcome.

Aim(s): To evaluate the potential predictive applications of a radiomics model using 68Ga-DOTA-PET/CT images in GEP-NETs.

Materials and methods: We retrospectively analysed 14GEP-NETs (1G1-2G2-1G3; mean age 56.2y) who underwent staging 68Ga-DOTA-PET/CT before complete PRRT with 177Lu-DOTATOC (range activity 27.9-29.9 GBq; 5 cycles for each). FU data about clinical, laboratory and radiological exams were recorded for at least 6 months after the last cycle. 127 lesions with high PET expression of somatostatin receptors were examined. LifeX tool was used to extract 65 imaging features from each lesion by defining two subgroups (parenchymal and LN+BL). CgA values pre-PRRT and histological grading features were considered. A novel statistical system based on point-biserial correlation coefficient and logistic regression analysis was implemented for feature reduction and selection, while DA was used to obtain the prediction model.

Conference: 18th Annual ENETS Concerence (2021)

Presenting Author:

Authors: Laudicella R, Comelli A, Spataro A, Stefano A, Vento A,

Keywords: radiomics, PRRT, GEP-NET, 68Ga-DOTA, PET/CT,