Study on the value of preoperative contrast-enhanced MRI features assessment in evaluating pathological differentiation and postoperative prognosis of pancreatic ductal adenocarcinoma

Objective To evaluate the differential value of preoperative MRI contrast-enhanced features and clinical features for patients with pancreatic ductal adenocarcinoma (PDAC) of different pathological differentiations.

Methods A retrospective analysis was conducted on the clinical and preoperative MRI data of 669 PDAC patients treated with radical surgery. Finally, 385 cases that met the criteria were included, and divided into the poorly differentiated group (190 cases) and moderate-well differentiated group (195 cases) according to the postoperative pathology. All patients received 1.5T MRI examinations before the operation, which included T1WI, T2WI, diffusion-weighted imaging (DWI) and multi-phase contrast-enhanced scans. The differences in clinical-imaging characteristics between two groups were compared. Multivariate logistic regression was used to screen the independent predictors, and a combined evaluation model was constructed. The overall survival (OS) and recurrence-free survival (RFS) in two groups were evaluated by the Kaplan-Meier method.

Results The proportion of patients with CA19-9 ＞ 100 U/mL in poorly differentiation group was significantly higher than that in moderate-high differentiation group (70.53% vs 48.21%, P ＜ 0.01). The results of MRI analysis showed that the poorly differentiated PDAC was more prone to low signal in the arterial phase (93.68% vs 74.87%, P ＜ 0.01), marginal enhancement (57.37% vs 38.46%, P ＜ 0.01) and MRI lymph node enlargement (39.47% vs 17.95%, P ＜ 0.01). The results of multivariate regression results showed that online CA19-9 ＞ 100 U/mL (OR = 2.316), low signal in the arterial phase (OR = 4.441), marginal enhancement (OR = 1.728) and MRI lymph node enlargement (OR = 2.415) were the independent predictors of the poorly differentiated state. The efficacy (AUC) of the image-clinical combined assessment model reached 0.724 (95%CI: 0.676–0.768).

Conclusions Low signal in the arterial phase, marginal enhancement, MRI lymph node enlargement and CA19-9 ＞ 100 U/mL are the important radiological and clinical features for preoperative non-invasive assessment of the differentiation degree of PDAC, which can provide an important basis for preoperative non-invasive assessment of the differentiation degree of PDAC.