Nomogram based on MRI and clinical features to predict progression-free survival in patients with stage IIIC1r cervical squamous cell carcinoma: A two-center study.

To develop a nomogram based on MRI and clinical features to predict progression-free survival (PFS) of 2018 FIGO stage ⅢC1r cervical squamous cell carcinoma (CSCC). 144 consecutive patients with stage ⅢC1r CSCC from two independent institutions were stratified into training cohort (from Institution 1, n=100) and independent validation cohort (from Institution 2, n=44). Univariate and multivariate Cox regression analyses of MRI and clinical features before treatment were performed to determine independent risk factors for PFS in training cohort. Nomogram was developed based on them. Concordance index (C-index), calibration curves, and receiver operating characteristic (ROC) analyses were performed to assess and validate the nomogram. In training cohort, 2009 FIGO stage, maximum length of the primary tumor, short diameter and roundness index of the maximum metastatic lymph node were independent risk factors of PFS in patients with stage IIIC1r CSCC (all P- values < 0.05). Nomogram based on them to predict 1- and 3-year PFS achieved C-indexes of 0.835 (95% confidence interval (CI): 0.809–0.862) and 0.789 (95%CI: 0.683–0.895) in the training and validation cohorts, respectively. Areas under ROC curves for the nomogram to predict 1- and 3-year PFS were 0.891 (95%CI: 0.829–0.954), 0.921 (95%CI: 0.861–0.981) in training cohort, and 0.902 (95%CI: 0.803–0.999), 0.885 (95%CI: 0.778–0.992) in validation cohort, respectively. Calibration curves indicated the nomogram predictions were in good agreement with actual observations. The nomogram based on MRI and clinical features has high accuracy and stability in predicting PFS of patients with stage IIIC1r CSCC. • Nomogram based on MRI and clinical features was developed to predict PFS of stage IIIC1r CSCC. • The prediction nomogram achieved a C-index of 0.789 in validation cohort. • Nomogram had good predictive value with AUCs larger than 0.85 in validation cohort. [ABSTRACT FROM AUTHOR]