Objective To explore the predictive value of computed tomography (CT) texture analysis for the recurrence in children with acute pancreatitis (AP).

Methods The clinical diagnostic test was conducted. The clinical data of 56 children with primary AP who were admitted to Wuhan Fourth Hospital from January 2016 to January 2018 were collected. There were 13 males and 43 females, aged from 3.5 to 13.0 years, with a median age of 5.5 years. Based on follow-up in other hospitals, 20 children with recurrence of AP were allocated into recurrence group, and 36 children without recurrence were allocated into non-recurrence group. All the 56 children underwent abdomen plain and enhanced CT scan within 24 hours after first admission. Observation indicators: (1) comparison of clinicopathological features between two groups of children with AP. (2) comparison of CT texture parameters between two groups of children with AP. (3) diagnostic efficacy of clinical features and CT texture parameters. Follow-up using outpatient reexamination and telephone interview was conducted to detect recurrence of AP up to February 2020. The duration of follow-up required more than or equal to 24 months. The Shapiro Wilk test was used to analyze normality of measurement data. Measurement data with normal distribution were repre-sented as Mean±SD, and comparison between groups was conducted using the t test. Measurement data with skewed distribution were represented as M (range) or M (P25,P75), and comparison between groups was conducted using the Mann-Whitney U test. Count data were described as absolute numbers, and comparison between groups was analyzed using the chi-square test. Comparison of ordinal data was conducted using the Mann-Whitney U test. Clinical parameters and CT texture parameters with statistical differences were multivariate analyzed using the Logistic regression model. Receiver operating characteristic curve (ROC) analysis was used to evaluate the predictive efficacy of parameters for recurrence of AP.

Results (1) Comparison of clinicopatholo-gical features between two groups of children with AP: cases with or without complications were 14 and 6 for the recurrence group, versus 7 and 29 for the non-recurrence group, showing a significant difference between the two groups (χ²=14.021, P<0.05). Cases with minimal, moderately severe or severe disease (severity of disease) were 2, 5, 13 for the recurrence group, versus 19, 11, 6 for the non-recurrence group, showing a significant difference between the two groups (Z=5.414, P<0.05). (2) Comparison of CT texture parameters between two groups of children with AP: the energy value in the arterial phase on CT examination was 0.186(0.174,0.206)for the recurrence group and 0.413(0.405,0.425) for the non-recurrence group, showing a significant difference between the two groups (Z=9.413, P<0.05). The energy value and entropy value in the venous phase on CT examination were 0.084(0.078,0.092) and 0.961(0.210,1.720) for the recurrence group, versus 0.135(0.124,0.156) and 0.372(0.210,0.535) for the non-recurrence group, showing significant differences between the two groups (Z=4.763, 7.243, P<0.05). (3) Diagnostic efficacy of clinical parameters and CT texture parameters: results of multivariate analysis showed the complications, severity of disease, energy value in the arterial phase on CT examination were related factors for recurrence in children with AP, energy value and entropy value in the venous phase on CT examination were related factor for recurrence in children with AP (odds ratio=0.874, 0.765, 0.837, 0.902, 0.813, 95% confidence interval as 0.802‒0.985, 0.581‒0.914, 0.753‒0.897, 0.862‒0.948, 0.765‒0.873, P<0.05). Results of ROC analysis showed that that areas under curve (AUC) of complications, severity of disease in the clinical parameters were 0.734 and 0.832, the AUC of single CT texture parameter was 0.811‒0.867, the AUC of clinico-pathological parameters combined with CT texture parameters was 0.882.

Conclusion CT texture analysis can early and non-invasively predict the recurrence of AP in children, and the combination of clinicopathological parameters with CT texture parameter has a better predictive efficacy.