AUTHOR=Bouhamama Amine , Leporq Benjamin , Faraz Khuram , Foy Jean-Philippe , Boussageon Maxime , Pérol Maurice , Ortiz-Cuaran Sandra , Ghiringhelli François , Saintigny Pierre , Beuf Olivier , Pilleul Frank 

TITLE=Radiomics combined with transcriptomics to predict response to immunotherapy from patients treated with PD-1/PD-L1 inhibitors for advanced NSCLC

JOURNAL=Frontiers in Radiology

VOLUME=Volume 3 - 2023

YEAR=2023

URL=https://www.frontiersin.org/journals/radiology/articles/10.3389/fradi.2023.1168448

DOI=10.3389/fradi.2023.1168448

ISSN=2673-8740

ABSTRACT=INTRODUCTION	
In this study, we aim to build radiomics and multiomics models based on transcriptomics and radiomics to predict the response by patients treated with PD-L1 inhibitor.

MATERIAL and METHODS
195 patients treated with PD-1/PD-L1 inhibitors were included. 342 Radiomic Features were extracted from pretreatment CT scans for all patients. The training set was built with 110 patients treated in the Centre Léon Bérard comprehensive cancer center. An independent validation cohort was built with the 85 patients treated in Dijon. The two sets were dichotomized into two different classes: patients with  a disease control (DC), and considered as Non-Responders, in order to predict the disease control at 3 months. Various models were trained with different feature selection methods and different classifiers were evaluated to build the models. 
In a second exploratory step, we used transcriptomics to enrich the database and develop a multiomics signature of response to immunotherapy, in a 54-patient subgroup. Finally, we considered the HOT/COLD status. We trained a radiomics model to predict the HOT/COLD status on the one hand and on the other hand we prototyped a hybrid model integrating radiomics and the HOT/COLD status to predict the response to immunotherapy.

RESULTS
Radiomics signature for 3 months PFS classification: The most predictive model had an area under the receiver operating characteristic curve (AUROC) of 0.94 in the training set  and 0.65 in the external validation set. This model was obtained with the t-test selection method and with a SVM classifier. 
Multiomics signature for PFS classification: The most predictive model had an AUROC of 0.95, in the training set, and 0.99 in the validation set. 
Radiomics model to predict HOT/COLD status: the most predictive model had an AUROC of 0.93 in the training set; and 0.86 in the validation set. 
HOT/COLD radiomics hybrid model for PFS classification: the most predictive model had an AUROC of 0.93 in the training set and 0.90 in the validation set. 

CONCLUSION:
In conclusion, radiomics could be used to predict the response to immunotherapy in NSCLC patients. The use of transcriptomics or the HOT/COLD status together with radiomics may improve the prediction models.