ESTRO 2022 - Abstract Book

S1563

Abstract book

ESTRO 2022

(G) features individually as well as with a separate combined (R+G). Algorithms were trained on two thirds of the dataset (124 patients) and tested on the remaining third (60 patients). The model dimensions were reduced by removing the highly correlated variables (with Pearson correlation r > 0.8) and by performing feature selection using Least Absolute Shrinkage and Selection Operator (LASSO) and ElasticNet regularisations. Models were optimised using 10 folds stratified cross- validation, replicated 100 times within the training set. Area Under the ROC Curves (AUC) were used for model optimisation and accuracy, Youden Index (YI) and F1-scores were calculated. Groups of GS6, GS7(3+4), GS7(4+3) and GS8 to 10 were considered. Thirty-two patients had tumours with GS6, 41 patients had GS7(3+4), 32 patients had GS7(4+3) and 79 patients had GS8 or higher. Prostate tumours were also classified as low, intermediate or high risks based on GS, initial PSA and TNM stage. Four patients had low risk cancer, 36 had intermediate risk and 144 had high risk cancer. Results Table 1 summarises the AUCs, accuracy, YI and F1-scores for the training and test results, and the number of selected features in each model. Classification between GS6 vs >GS6 (N= 32 vs 152) showed AUC R = 0.55, AUC G = 0.63 and AUC R+G = 0.63 within the test set. Classification between GS7(3+4) vs GS7(4+3) (N=41 vs 32) showed AUC R = 0.54, AUC G = 0.69 and AUC R+G = 0.68 on the test set. Finally, AUC R = 0.57, AUC G = 0.62 and AUC R+G = 0.65 were observed on the test set for the classification of GS7 vs >GS7 (N=73 vs 79). Regarding the risk group classification, only classification between intermediate and high risk (N=36 vs 144) was performed due to the reduced number of low risk patients in this cohort. For these subgroups, the test results showed AUC R = 0.63, AUC G = 0.79 and AUC R+G = 0.79.

Conclusion Genomics models outperformed the radiomics models for risk stratification in prostate cancer. Our combined radiogenomics model improved performance for the classification between GS7 and GS higher than 7 (>GS7). External validations are warranted to verify these findings. [1] Osman et al. Int. J. Radiat. Oncol. Biol. Phys. 2019; 105(2):448-456

PO-1759 Delta Radiomics can predict complete pathological response in rectal cancer patients

A. Angrisani 1 , T. Di Pietro 1 , E. D’Ippolito 1 , V. Nardone 1 , A. Sangiovanni 1 , A. Reginelli 2 , C. Guida 3 , S. Cappabianca 1

1 "L. Vanvitelli" University of Campania, Precision Medicine - Radiotherapy Unit, Naples, Italy; 2 "L. Vanvitelli" University of Campania, Precision Medicine - Radiology Unit, Naples, Italy; 3 Ospedale del Mare, Radiotherapy, Naples, Italy Purpose or Objective The present study was designed to evaluate MRI delta texture analysis (D-TA) in predicting the outcome in terms of the complete pathological response of patients with locally advanced rectal cancer undergoing neoadjuvant chemoradiotherapy (C-RT) followed by surgery. Materials and Methods We performed a retrospective analysis on 100 patients with locally advanced rectal adenocarcinoma undergoing C-RT and radical surgery in three different centers between January 2013 and December 2019. The gross tumor volume (GTV) was evaluated at both baselines and after C-RT MRI and contoured on T2, DWI, and ADC sequences. Multiple texture parameters were extracted with LifeX Software, and D-TA was calculated as the percentage variations in the two-time points. By performing univariate analysis and multivariate analysis (logistic regression), these TA parameters were then correlated with patients' pathological outcomes. Complete pathological response (pCR, with no viable cancer cells: TRG 0) was chosen as the statistical end-point. ROC Curves were calculated on the three different datasets. Results In the whole cohort, 21 patients (21%) showed a pCR. At univariate analysis and binary logistic analysis, the only parameter that resulted significantly correlated with pCR in the Training dataset was ADC GLCM-Entropy. The binary logistic regression was repeated in the two Validation Dataset. AUC for pCR was 0.87 in the Training Dataset and respectively 0.92 and 0.88 in the two Validation Datasets.