Purpose: To evaluate MR-derived histogram parameters in predicting aggressiveness and surgical outcomes in patients with PDAC, by correlating them to pathological features, recurrence-free survival (RFS), and overall survival (OS).
Methods: Pre-operative MR examinations of 103 patients with PDAC between July 2014 and September 2018 were retrospectively analyzed. Morphologic features and whole-tumor histogram-derived parameters were correlated to pathological features using Fisher's exact or Mann-Whitney U tests and receiver operating characteristic (ROC) curves were constructed for significant parameters. Cox regression analysis and Kaplan-Meier curves were used to determine the association of clinical-pathological variables, morphological features, and histogram-derived parameters with RFS and OS.
Results: T1entropy, ADCentropy, T2kurtosis, and ADCuniformity had the highest area under the curve (AUC) for prediction of vascular infiltration, nodal metastases, microscopic vascular invasion, and peripancreatic fat invasion (.657, .742, .760, and .818, respectively). Poor tumor differentiation (P = 0.002, hazard ratio-HR = 4.08), nodal ratio (P = 0.034, HR 6.95), and ADCmaximum (P = 0.021, HR 1.01) were significant predictors of RFS. Poor tumor differentiation (P = 0.05, HR 2.82), ADCuniformity (P = 0.02, HR 3.32), and arterialentropy (P = 0.02, HR 6.84) were the only significant predictors of death; patients with higher arterialentropy had significantly shorter OS than patients who did not meet this criterion (P = 0.02; median OS 24 vs 31 months).
Conclusion: Histogram-derived parameters may predict adverse pathological features in PDACs. High arterialentropy seems to be associated with short OS after surgery in patients with PDAC.
Keywords: Magnetic resonance imaging; Pancreas; Pancreatic carcinoma; Pancreatic neoplasms.