Predictive value of pretreatment MRI texture analysis in patients with primary nasopharyngeal carcinoma

To determine the predictive value of pretreatment MRI texture analysis for progression-free survival (PFS) in patients with primary nasopharyngeal carcinoma (NPC). Ethical approval by the institutional review board was obtained for this retrospective analysis. In 79 patients with primary NPC, texture analysis of the primary tumour was performed on pretreatment T2 and contrast-enhanced T1-weighted images (T2WIs and CE-T1WIs). The Cox proportional hazards model was used to determine the association of texture features, tumour volume and the tumour-node-metastasis (TNM) stage with PFS. Survival curves were plotted using the Kaplan-Meier method. The prognostic performance was evaluated with the receiver operating characteristic (ROC) analyses and C-index. Tumour volume (hazard ratio, 1.054; 95% confidence interval [CI], 1.016–1.093) and CE-T1WI-based uniformity (hazard ratio, 0; 95% CI, 0–0.001) were identified as independent predictors for PFS (p < 0.05). Kaplan-Meier analysis showed that smaller tumour volume (less than the cut-off value, 11.699 cm${}^3$) and higher CE-T1WI-based uniformity (greater than the cut-off value, 0.856) were associated with improved PFS (p < 0.05). The combination of CE-T1WI-based uniformity with tumour volume and the overall stage predicted PFS better (area under the curve [AUC], 0.825; Cindex, 0.794) than the tumour volume (AUC, 0.659; C-index, 0.616) or the overall stage (AUC, 0.636; C-index, 0.627) did (p < 0.05). A texture parameter of pretreatment CE-T1WI-based uniformity improves the prediction of PFS in NPC patients.