Objective:To investigate the risk factors for the relapse of autoimmune pancreatitis (AIP) after steroid therapy.Methods:Clinical data of 72 AIP patients treated with steroids in Nanjing Drum Tower Hospital from January 2012 to December 2023 were collected retrospectively. AIP patients were divided into relapse group ( n=25) and non-relapse group ( n=47) based on the presence or absence of their relapse after steroid therapy. Patients' age of onset, gender, history of diabetes mellitus, first clinical manifestations, serum IgG4 and CA19-9 level, imaging features and other organ involvements were recorded. Oral prednisone was used at an initial dose of 0.6 mg·kg -1·d -1, gradually reduced to 5-10 mg/d and then maintained at a low dose. The follow-up period started from steroid initiation to the last follow-up or relapse. The presence of maintenance steroid treatment, time interval between onset and steroid initiation, the presence of significant IgG4 decrease and the presence of persistently enlarged pancreas after therapy were recorded. The cumulative relapse rate curve after steroid therapy was drawn by Kaplan-Meier method. Univariate and multivariate analyses were performed by Cox proportional hazard regression model. The receiver operator characteristic curves (ROC) were plotted and the area under the curve (AUC) was calculated. The Log-Rank test was used to analyze the differences on the relapse between different groups. The subgroup forest plot was drawn to assess the effect of risk factors on the relapse of AIP in different subgroups. Results:The 72 patients with AIP had a median follow-up of 42 (12-127) months. 34.7% (25/72) of patients relapsed after steroid therapy during the follow-up period. The percentages of patients whose first clinical manifestation was abdominal distension or acute pancreatitis, whose interval between onset and steroid initiation was more than 1 year and whose pancreases were persistently enlarged after steroid therapy in the relapse group were higher than those in the non-relapse group, and the differences were all statistically significant (all P value <0.05). The 1-, 3- and 5-year cumulative relapse rate after steroid therapy was 20.8%, 34.1% and 37.8%, respectively. Univariate analysis found that the first clinical manifestations of abdominal distension or acute pancreatitis, interval between onset and steroid initiation more than 1 year, and persistently enlarged pancreas after steroid therapy were all significantly associated with relapse (all P value <0.05). Multivariate analysis found that interval between onset and steroid initiation more than 1 year and persistently enlarged pancreas after steroid therapy were independent risk factors for relapse of AIP [hazard ratio ( HR)=3.606 and 6.515, 95% confidence interval (95% CI) 1.362-9.854 and 2.088-20.326]. Kaplan-Meier survival curves showed that the relapse rate after steroid therapy was higher in AIP patients whose interval between onset and steroid initiation was more than 1 year than in those whose interval was less than 1 year (55.6% versus 27.8%), and the relapse rate in AIP patients with persistently enlarged pancreas after steroid therapy was higher than that in those without it (77.8% versus 28.6%), and the differences were both statistically significant (both P<0.05). Subgroup forest plot showed that persistently enlarged pancreas after steroid therapy was an independent risk factor for relapse of AIP regardless of the presence of a diabetes mellitus history, the first manifestation of abdominal pain, the diffuse or focal type in pancreatic imaging, and the presence of dilated pancreatic duct or not (all P value <0.05). Conclusions:Time interval between onset and steroid initiation more than 1 year and persistently enlarged pancreas after steroid therapy were independent risk factors for the relapse of AIP after steroid therapy.