Multiparametric MRI-based Radiomics Signature for Preoperative Estimation of Tumor-Stroma Ratio in Rectal Cancer.

Fudan University,Huang Dan,Liu Fangqi,Fu Caixia

DOI: https://doi.org/10.1007/s00330-020-07403-6

IF: 7.034

2020-01-01

European Radiology

Abstract:To determine whether a radiomics signature (rad-score) outperforms ADC in TSR estimation by developing a radiomics biomarker for preoperative TSR diagnosis in rectal cancer. This study included 149 patients (119 and 30 in the training and validation cohorts, respectively). All patients underwent T2-weighted, diffusion-weighted, and contrast-enhanced T1-weighted imaging. A rad-score was generated using the least absolute shrinkage and selection operator (LASSO) algorithm and stepwise multivariate logistic regression. Meanwhile, the mean ADCs were calculated from ADC maps. For both the mean ADC and rad-score, binary logistic regression and Spearman correlation coefficients were used to determine associations with the TSR, and the area under the receiver operating characteristic (ROC) curve was used to assess the diagnostic performance. The reliability of the rad-score was quantified by comparing the imaging-estimated TSR with the actual TSR of each patient. Both the mean ADC and rad-score were positively correlated with the TSR in the training cohort (mean ADC: p < 0.001, r = 0.566; rad-score: p < 0.001, r = 0.559) and validation cohort (mean ADC: p < 0.001, r = 0.671; rad-score: p = 0.002, r = 0.536). The rad-score, with AUCs of 0.917 (95% CI 0.869–0.965) and 0.787 (95% CI 0.602–0.972) in the training and validation cohorts, respectively, outperformed the mean ADC (training cohort: AUC = 0.776, 95% CI 0.693–0.859; validation cohort: AUC = 0.764, 95% CI 0.592–0.936) in TSR estimation. The ADC possesses potential diagnostic value for TSR estimation in rectal cancer, and the rad-score shows increased diagnostic value over the ADC and may be a promising supplemental tool for patient stratification and informing decision-making. • Tumor-stroma ratio has been verified as an independent prognostic factor for various solid tumors including rectal cancer. • The ADC and multiparametric MRI-based radiomics features were significantly and positively correlated with the tumor-stroma ratio in rectal cancer. • The radiomics signature outperformed the ADC in discriminating TSR in rectal cancer.