Diagnosis, treatment of Type (cid:0) autoimmune pancreatitis and management of relapse and side effects after steroid treatment in China

Background: Type (cid:0) autoimmune pancreatitis (AIP) have a dramatic response to steroid therapy, but the relapse rates (RRs) is very high, and the side effects of steroid therapy are inevitable. However, the management of relapse and side effects of steroid therapy have been poorly investigated and no series have been reported previously, so this article focuses on the management of relapse and side effects. Methods: A single-centre, retrospective, cohort study of the type (cid:0) AIP patients admitted to Xiangya Hospital Central South University from September 2008 to September 2019. Collection and retrospectively analyzed the clinicopathologic data and outcomes of type (cid:0) AIP patients. Results: 82 patients with type (cid:0) AIP were included. The ratio of male with female was 2.73: 1 with median age of onset of 52 years old. 73.2% were histologically conrmed, 62.2% were showed other organ involvement (OOI). 78.0% treated by medications: 40 cases (62.5%) prednisolone, 24 cases (37.5%) prednisolone plus cyclophosphamide. The effective rate of prednisolone and prednisolone plus cyclophosphamide was no signicant difference (37/40 vs 22/24, P = 0.904). The RRs of prednisolone and prednisolone plus cyclophosphamide after remission have no signicant difference (14/40 vs 7/24, P = 0.630). Increasing the dosage of prednisolone, the effective rate of the prednisolone treatment and prednisolone plus cyclophosphamide treatment relapsed patients were 78.57% and 71.43%, respectively. The side effects were diabetes mellitus (DM) in 12.5%; central obesity in 15.6%; hyperlipidemia in 18.8%; gastric ulcer in 6.3%; osteoporosis in 9.4%; bone fracture in 1.6%. For those side effects patients performed low dose prednisolone and expectant treatment have a ideal results. Conclusions: There was no signicant difference between prednisolone and prednisolone plus cyclophosphamide in the treatment of type (cid:0) AIP. Increasing the dose of prednisolone can effectively treat relapsed patients. Low dose steroid and expectant treatment should be performed when side effects arised. mycophenolate mofetil, cyclophosphamide or rituximab. Treatment with prednisolone plus cyclophosphamide has not been explored in AIP. In our study we use initial prednisolone xed-dose (30mg/d) or prednisolone (30mg/d) plus cyclophosphamide (100 mg/d) regimen. We found that the effective rate of prednisolone and prednisolone plus cyclophosphamide was no signicant difference [37/40(92.5%) vs 22/24(91.7%), P = 0.904].

prednisolone plus cyclophosphamide. The effective rate of prednisolone and prednisolone plus cyclophosphamide was no signi cant difference (37/40 vs 22/24, P = 0.904). The RRs of prednisolone and prednisolone plus cyclophosphamide after remission have no signi cant difference (14/40 vs 7/24, P = 0.630). Increasing the dosage of prednisolone, the effective rate of the prednisolone treatment and prednisolone plus cyclophosphamide treatment relapsed patients were 78.57% and 71.43%, respectively.
The side effects were diabetes mellitus (DM) in 12.5%; central obesity in 15.6%; hyperlipidemia in 18.8%; gastric ulcer in 6.3%; osteoporosis in 9.4%; bone fracture in 1.6%. For those side effects patients performed low dose prednisolone and expectant treatment have a ideal results.
Conclusions: There was no signi cant difference between prednisolone and prednisolone plus cyclophosphamide in the treatment of type AIP. Increasing the dose of prednisolone can effectively treat relapsed patients. Low dose steroid and expectant treatment should be performed when side effects arised.

Background
Autoimmune pancreatitis (AIP) is mediated by autoimmunity and is characterized by obstructive jaundice, pancreatic enlargement, irregular stenosis of the pancreatic duct, lymphoplasmacytic in ltration and brosis, and other organ involvement (OOI). AIP is a particular type of chronic pancreatitis which was rst described by Sarles in 1961 [1] and Yoshida et al proposed the concept of AIP in 1995 [2]. Japanese scholars have reported that AIP accounts for 5%-6% of chronic pancreatitis [3]. Atsushi Masamune et al reported that the overall prevalence rate of 10.1 per 100,000 persons and annual incidence rate of 3.1 per 100,000 persons in Japan [4].
In recent years, AIP is increasingly recognized as a new clinical entity with its protean pancreatic obiliary and systemic presentations [5]. Its unique pathology and overlap of clinical and radiological features foster the disease's unique position. Two histological subtypes in AIP have been recognized, type I and type II [6,7]. Over the past decade, many national case studies of AIP have been published [8,9], but rare reported from China which suffering from a high prevalence of type AIP.
Because of the similarities between clinical manifestations and imaging ndings of AIP and pancreatic cancer, clinicians and radiologists do not have enough knowledge about it, so that some patients with AIP experienced unnecessary surgical treatment due to misdiagnosis as pancreatic cancer. AIP is a particular type of chronic pancreatitis which are dramatic response to steroids [10][11][12][13][14][15][16][17][18]. But various RRs, ranging from 24% to 63%, have been reported for type AIP [19][20][21]. Furthermore, several studies have reported steroid side effects related to AIP and other autoimmune diseases [23][24][25][26][27][28][29]. The aims of this study were to analyze the data of AIP patients and summarizes the diagnosis and treatment experience of type AIP in China. Due to the RRs of AIP is very high, and the side effects of steroid therapy are inevitable. Therefore this study also focus on the management of relapse and side effects.

Patients
The study was conducted in Xiangya Hospital Central South University. Collection and retrospectively analyzed the clinicopathologic data and outcomes of these patients who was hospitalized in our center during September 2008 to September 2019. According to International Consensus of Diagnostic Criteria (ICDC) [10], all patients were diagnosed as AIP. Excluded the patients with incomplete information collection and the patients with type II AIP. A total of 82 cases were included.

Follow-up
Follow-up period was the time from AIP diagnosis until death or the last visit. During follow-up, patients underwent CT or MRI/MRCP at 3-month intervals during the rst year, 6-month intervals during the second year and then once a year to monitor relapse and OOI. Serum IgG4 levels and biochemical blood tests were performed at every visit to the outpatient clinic. For the steroid therapy patients we also asked if there were any side effects.

Statistical analysis
Continuous variables were presented as means ± standard deviation (SD), and were compared using Student's t test. Categorical variables were compared using the Chi square test, and P-value < 0.05 was considered to indicate statistically signi cant differences. All the statistical analyses were performed using the Statistical Product and Service Solutions (SPSS) 22.0 statistical software package (IBM Analytics, Armonk, NY) in this study.

Results
Eighty-two patients with type AIP were included in this study, the ratio of male with female was 2.73: 1 with mean ages at diagnosis were 53.4 ± 11.2 years. Patients' characteristics are summarized in Table 1. Serological results and radiographic results are summarized in Table 2.
CT examination 80 patients underwent CT examination, 58 patients (72.5%) considered AIP, 22 patients (27.5%) considered pancreatic biliary tract tumors, 52 cases (65.0%) showed focal pancreatic enlargement, 28 cases (35.0%) showed diffuse enlargement, pancreatic head enlargement in 23 cases (28.8%). The diffuse AIP CT scan showed typical pancreas diffuse enlargement was "sausage-like", pancreatic parenchymal density decreased, dynamic enhancement was homogeneous and delayed intensi cation; some patients had clearly de ned low density envelope-like margins around the pancreas, which was characteristic of AIP ( Fig. 1a, b). After the treatment with steroid, the pancreatic parenchyma was reduced and evenly strengthened. The clearance around the pancreas was clear and the feather-like structure was restored ( Fig. 1c, d). The focal manifestation of focal AIP is a low-density mass. After dynamic enhancement, delayed and homogenous enhancement of the mass can be seen (Fig. 2a). After treatment, the mass disappears, and the parenchymal mass of the pancreas is evenly enhanced and the structure is clear (Fig. 2b).

Treatment and outcomes
The initial dose of prednisolone is at least 30 mg/d, the dose of cyclophosphamide was 100 mg/d, 2-4 weeks later to review if the effect is well, start to reduce the prednisolone dose 5 mg/d every 1-2 weeks, according to the results of clinical manifestations, abdominal CT and IgG4 to adjust the dose. Reexamine abdominal CT and IgG4 every 4-6 weeks; maintain the dose of 5mg/d for at least 6 months after the clinical manifestations, abdominal CT and IgG4 results are normal. For those relapsed patients, we increasing the dose of prednisolone to initial dose. For those patients, who have side effects due to high dose of prednisolone treatment, we reduce the dose of prednisolone to 5mg/d, and reviewed very two week, until the clinical manifestations, abdominal CT and IgG4 results are normal, also continued maintain treatment for 6 months.
All of the 82 patients were followed up for 12-142 months, with a median follow-up time of 92 months. 64 patients (78.0%) underwent treatment with different medications: 40 cases (62.5%) prednisolone, 24 cases (37.5%) prednisolone plus cyclophosphamide. The effective rate of prednisolone and prednisolone plus cyclophosphamide was no signi cant difference [37/40(92.5%) vs 22/24(91.7%), P=0.904]. And the RRs of prednisolone and prednisolone plus cyclophosphamide also have no signi cant difference ( 35.0% vs 29.2%, P = 0.630). 14 cases (35.0%) of the prednisolone treatment patients relapsed after remission, 8 (20.0%) of them had more than two relapse events. 5 (12.5%) of them had twice relapse events, 2 patients (5.0%) no longer treated and relapsed three times; One patient (2.5%) had four relapse events, and died due to intolerance to prednisolone therapy. 11 of the 14 prednisolone treatment relapsed patients have remission by increasing the dose of prednisolone to initial dose and gradually decreased the dose, nally withdrawal the prednisolone after 6 months maintenance treatment. 7 (29.2%) of prednisolone plus cyclophosphamide treatment patients relapsed after remission, 4 (16.7%) of them had more than two relapse events; one of them had three relapse events, and died due to repeated of cholangitis, one case cannot rule out pancreatic cancer after recurrence, and then treated surgically, but died due to postoperative complications. Increase the prednisolone to initial dose and gradually decreased the dose, nally withdrawal the prednisolone after 6 months maintenance treatment, 5 of 7 prednisolone plus cyclophosphamide treatment relapsed patients have remission. Therapeutic effect between prednisolone and prednisolone plus cyclophosphamide in Table 3.
The side effects of the 64 drug treatment patients were DM in 8 patients (12.5%); central obesity or moon face in 10 cases (15.6%); hyperlipidemia in 12 cases (18.8%); gastric ulcer in 4 cases (6.3%); osteoporosis in 6 case (9.4%); bone fracture in 1 case (1.6%). For these patients we reduce the dose of prednisolone, and low dose maintenance treatment was performed, furthermore, for the DM patients oral hypoglycemic drugs, for the central obesity or moon face and hyperlipidemia patients oral statins, for the gastric ulcer patients oral proton pump inhibitor(PPIs) and gastric mucosal protective drugs, osteoporosis patients oral calcium supplements, for the fracture patient we withdraw prednisolone and surgical treatment of fracture, but three months later relapse of AIP, we performed low dose prednisolone maintenance treatment, the effect is also very good.
18 cases (22.0%) were not treated with prednisolone, of which 12 cases (14.6%) were misdiagnosed as cholangiocarcinoma or pancreatic cancer and received surgical treatment, and 2 cases died due to relapse after cholangiojejunostomy; 4 cases (4.9%) prednisolone experimental therapy patients could not exclude pancreatic cancer, and then underwent surgical treatment, and 2 cases died due to postoperative complications. 4 cases (4.9%) unwillingness prednisolone treatment, and 3 (75.0%) of them died due to recurrent cholangitis. 2 case (3.7%) was found due to physical examination, and had no symptoms all the time, 1 case died for natural causes.

Discussion
To the best of our knowledge, this is the only one retrospective study ever conducted in a tertiary hospital in China that summarize the experience of diagnosis and treatment of AIP, which speci cally focus on the management of relapse and side effects. We found the misdiagnosis rates of AIP is very high, so it is very important to improve the accurate diagnosis rate of AIP. There was no signi cant difference between prednisolone and prednisolone plus cyclophosphamide in the treatment of type AIP. Increasing the dose of steroid can be effective on therapy relapse patients. Low dose steroid and symptomatic treatment should be performed when side effects arise from a high dose of steroid.
AIP is an enigmatic disease and sometimes di cult to diagnose [30]. The clinical manifestations of AIP are complex, in particular, focal AIP is very similar to pancreatic cancer in imaging manifestation, so it is di cult to distinguish focal AIP from pancreatic cancer. But AIP is sensitive to hormone therapy and does not need surgical treatment, whereas pancreatic cancer requires surgical treatment. Therefore, the differential diagnosis between AIP and pancreatic cancer is great importance. AIP has a variety of diagnostic criteria, and the applicable conditions and methods of each criterion are different, so the AIP diagnosed in different countries may sometimes differ [16]. In 2011, Shimosegawa et al rst proposed the International Consensus of Diagnostic Criteria (ICDC) of AIP, and divided the AIP into Type I and Type II, which is the most widely accepted diagnostic criteria [10]. The AIP in China is mainly type I , that is, IgG4-related diseases affect the pancreas, most of them onset in the elderly, and about 70% of cases can be diagnosed in experienced treatment centers without the pathological specimens [17]. Elevated gamma globulinemia and immunoglobulin IgG, especially elevated IgG4, have been considered as characteristic indicator of AIP [18]. As many as 94% of AIP patients have elevated IgG4 levels [31], and IgG4 levels are closely related to disease activity [32]. Van Heerde et al. [33] studied the serum IgG4 and CA19-9 in patients with AIP and pancreatic cancer found that the sensitivity and speci city for diagnosis of AIP was 73% and 74% respectively with CA19-9 < 74 kU/L as the cut-off value. with IgG4 > 2.6 g/L as the threshold, the sensitivity of IgG4 for diagnosis of AIP was 70%, and the speci city was 100%. The sensitivity and speci city of diagnosis of AIP with combined CA19-9 < 74 kU/L and IgG4 > 1.0 g/L was 94% and 100% respectively. Therefore, simultaneous detect IgG4 levels and CA19-9 can improve the differential diagnosis rate between AIP and pancreatic cancer. The results of this study were basically consistent with the report. 72 patients were tested for IgG and 62 patients (86.1%) were elevated; IgG4 was detected in 62 cases, and 56 cases (90.3%) were positive.
Chari et al. [12] reported that 30% of AIP patients needed steroid experimental therapy, needle biopsy and surgical resection to make a de nite diagnosis. In this study, 16 cases (19.5%) were misdiagnosed as pancreatic cholangiocarcinoma and were diagnosed as AIP after surgery. Focal mass AIP is di cult to be diagnosed by imaging examination and endoscopic ultrasound biopsy [34], and endoscopic ultrasound biopsy is not available in some hospitals, so biopsy of the involved organs is helpful for diagnosis. In this study, 44 patients (53.7%) underwent tissue biopsy. The results showed that biopsy of involved organs is bene cial to the diagnosis of AIP. If cholangitis is combined, do ERCP and duodenal papillary biopsy can help to exclude pancreatic cancer and cholangiocarcinoma. When the differential diagnosis is not clear, steroid experimental therapy is helpful to diagnose AIP, but in order to avoid delay in the treatment of malignant tumors, imaging examination have performed 2 to 4 weeks after hormone therapy, the mass must be nearly completely relieved, any no relief of the mass must do exploratory surgery [35].
The rst goal of therapy in type I AIP is to induce remission. Response to steroid therapy in patients with AIP is dramatic and consistently leads to clinical improvement regardless of the subtypes [19,36]. As a result, steroids have become the standard therapy for inducing remission in AIP [37,38]. In the ICDC the starting dose of steroid for remission induction is de ned as 0.6-1 mg/kg per day [10]. Hart et al. [39] suggested the usefulness of a step-up approach to treating AIP with azathioprine, 6-mercaptopurine, mycophenolate mofetil, cyclophosphamide or rituximab. Treatment with prednisolone plus cyclophosphamide has not been explored in AIP. In our study we use initial prednisolone xed-dose (30mg/d) or prednisolone (30mg/d) plus cyclophosphamide (100 mg/d) regimen. We found that the effective rate of prednisolone and prednisolone plus cyclophosphamide was no signi cant difference [37/40(92.5%) vs 22/24(91.7%), P = 0.904].
Although steroid therapy is effective, but the type I AIP has a high RRs, ranging from 24% to 63% [19][20][21]. The Japanese consensus guidelines for AIP recommend re-administration or dose-up of steroid in patients who relapse after successful remission induced by initial steroid therapy [20]. Kazuichi Okazaki et al. [40] reported that most relapsed AIP cases, remission can be achieved with the same prednisolone dose as the initial dose, although it may be necessary to taper more gradually. In our study the relapsed patients achieved the same prednisolone or prednisolone plus cyclophosphamide dose as the initial dose, and most of the patients who relapsed had good results.
Current guidelines recommend low-dose (5 mg/day) maintenance steroid treatment (MST) for 2-3 years to reduce the relapse rate ( < 30%) [41]. However, steroid side effects are inevitable, the risk of corticosteroid-associated side effects, as well as the lifetime cumulative steroid dose, must be considered, the short-and long-term corticosteroid treatment may induce side effects, including chronic glycometabolism, obesity, an immunocompromised status against infection, cataracts, glaucoma, osteoporosis, and myopathy [42]. The majority of side effects, including secondary diabetes mellitus and osteoporosis, are treatable and/or preventable. Severe side effects, such as infection, bone fracture, femoral head necrosis and cardiovascular diseases, are life-threatening and should be avoided [41]. In this research the side effects of the 64 medications treatment patients were DM in 8 cases (12.5%); central obesity or moon face in 10 cases (15.6%); hyperlipidemia in 12 cases (18.8%); gastric ulcer in 4 cases (6.3%); osteoporosis in 6 case (9.4%); bone fracture in 1 case (1.6%). For these patients we reduce the dose of prednisolone, and low dose maintenance treatment was performed, furthermore, oral hypoglycemic drugs for DM patients, oral statins for central obesity or moonface and hyperlipidemia patients, oral proton pump inhibitors (PPIs) and gastric mucosal protection drugs for gastric ulcer patients, and oral calcium supplements for osteoporotic patients. For the fracture patient we withdrew prednisolone and operated on fracture, but AIP recurred three months later, and we performed low dose prednisolone maintenance treatment, the effect is also very good.
This study has several limitations. This study is a retrospective study, retrospective analysis may lead to some bias. Additionally, some patients with incomplete information are excluded, which may have some in uence on the results. Lastly, China as a high incidence of type AIP countries, the number of cases is not enough, so there may be partial bias. Further multicenter, prospective original studies would provide more precise data to reduce potential confounding results.

Conclusion
In conclusion, the misdiagnosis rate of AIP in china is very high. There was no signi cant difference between prednisolone and prednisolone plus cyclophosphamide in the treatment of type AIP. RRs is very high, follow-up should be continued after drug withdrawal, and the follow-up time should last more than 3 years. Increasing the dose of prednisolone can be effective on therapy relapse patients. Low dose steroid therapy and expectant treatment are useful when side effects arise from a high dose of steroid.  Pancreatic imaging ndings of diffuse enlargement AIP. a. Diffuse pancreas enlargement, reduced in parenchymal density, becoming "Sausage-like". b. Uniform pancreas, delayed enhancement, pancreatic pericardium-like margin (as indicated by the arrow). c. After the treatment with steroid, the pancreatic parenchyma was signi cantly reduced, the pancreatic density was slightly higher than before. d. After the treatment, the pancreatic parenchyma was evenly intensi ed, and the surrounding clearance was clear, restoring the feather-like structure.

Figure 2
Pancreatic imaging ndings of focal pancreatic enlargement AIP. a. Low-density lump in the tail of the pancreas, uniform and delayed strengthening. b. The tail of the pancreas is signi cantly reduced after treatment with steroid, uniformly strengthened, and the surrounding structure is clear. Figure 3 MRCP ndings of AIP. a.b. Intrahepatic and extrahepatic bile ducts dilate, the main pancreatic duct is irregularly narrowed (as indicated by the red arrows)