Objective To explore the value of apparent diffusion coefficient (ADC) in diffusion weighted imaging (DWI) for pathological grade of pancreatic neuroendocrine neoplasms (pNEN) before surgery.

Methods The MRI and pathological data of 72 patients with pathologically confirmed pNEN of Zhongshan Hospital, Fudan University from March 2015 to June 2019 were analyzed retrospectively, and ADC values of the tumors were measured. pNEN was classified into pNET G1, pNET G2, pNET G3, and pNEC G3 according to WHO 2017 latest classification standards for digestive system tumors. Spearman method was used to analyze the correlation between ADC of pNEN lesions and Ki-67. LSD method was used to pair-wise compare the difference of ADC. And receiver operating characteristic curve (ROC) method was used to analyze the diagnostic efficacy of ADC for pathological grade of pNEN.

Results All the 72 cases of pNEN were single lesion, and the pathological grade was 18 cases of pNET G1, 36 cases of pNET G2, 13 cases of pNET G3, and 5 cases of pNEC G3. ADC of pNEN was negatively correlated with Ki-67 (r=-0.845, P < 0.001). There was no significant difference in the ADC values of tumor foci between every two groups (except for pNET G3 and pNEC G3 groups, P < 0.001). As the cut-off value of ADC was 1.596×10^-3 mm²/s, the sensitivity and specificity for distinguishing grade G1 and G2 pNEN were 97.22% and 83.33%, and the area under the ROC was 0.941 (Z=13.340, P < 0.001). As the cut-off value of ADC was 1.103×10^-3 mm²/s, the sensitivity and specificity for distinguishing grade G2 and G3 pNEN were 83.33% and 100%, and the area under the ROC was 0.968 (Z=18.830, P < 0.001).

Conclusions ADC values of pNEN at different levels are different to some extent. ADC is helpful for the pathological diagnosis of pNEN before surgery.