Diagnosis and Treatment of Autoimmune Pancreatitis in China: A Systematic Review

Aims To provide comprehensive data on the diagnosis and treatment of autoimmune pancreatitis (AIP) patients in China. Design A systematic review. Methods All clinical studies concerning AIP from China published between January 2006 and June 2014 were retrospectively reviewed and analyzed. Results A total of 26 original articles involving 706 AIP patients were included with an estimated proportion of type 2 AIP as 4.7%. In the 706 AIP patients, the range of mean/median age was 48.6–67.0 years old and the male to female ratio was 4.47:1. The common presentations included obstructive jaundice (pooled rate: 63.4%, 95%CI: 55.4%–71.0%) and abdominal symptoms (pooled rate: 62.3%, 95%CI: 52.4%–71.7%). Biliary involvement was the most common extrapancreatic manifestations, especially the lower part of the common bile duct (pooled rate: 62.3%, 95%CI: 49.9%–73.9%). According to the imaging examinations, 53.8% and 41.6% patients were classified into focal-type and diffuse-type, respectively. Notably, upstream pancreatic duct dilatation was found in parts of patients (pooled rate: 13.8%, 95%CI: 6.6%–23.1%). The levels of serum IgG4 were elevated in most patients (pooled rate: 86.0%, 95%CI: 74.2%–94.6%). Nearly three tenths AIP patients received surgery (pooled rate: 29.7%, 95%CI: 18.1%–42.8%) due to mimicked malignancy. Steroid treatment was given to 78.4% patients (95%CI: 65.3%–89.1%) with a pooled remission rate of 96.2% (95%CI: 94.0%–97.9%). The pooled relapse rate was 13.8% (95%CI: 7.2%–22.0%) with the mean follow-up time ranging from 12 to 45 months. Conclusion Type 1 is the predominant type of Chinese AIP patients and the clinical features, diagnostic modalities and therapeutic regimen were similar with those in other countries. Knowledge of AIP should be more widespread to avoid unnecessary surgery.


Introduction
Since it was first reported in 1961 by Sarles et al [1] and was termed as autoimmune pancreatitis (AIP) in 1995 by Yoshida et al [2], AIP has gradually attracted attention in recent years. Nowadays, two subtypes of AIP are known-type 1 is a multi-organ disease associated with IgG4 while type 2 appears to be a pancreas-specific disorder [3]. In recent years, diagnostic criteria of AIP have been proposed in Japan, United States, Korea and Italy, and an International Consensus Diagnostic Criteria (ICDC) for AIP has been reached [4].
China has a high prevalence of pancreatic disease [5], and if the prevalence rate of AIP in Japan (2.2 per 100,000 populations) was used, the estimated amount of AIP patients would be 29 thousand [6]. However, there have been only a few reports about AIP from China [7][8][9][10]. In this study, we collected all well-designed clinical studies of AIP from China, and provided comprehensive data on the clinical presentation, diagnosis and treatment of Chinese AIP patients.

Strategy, criteria, and procedures for the literature search
We conducted a systematic literature search of PubMed, Web of Science and Chinese literature database including Wanfang Data and CNKI on June 20, 2014, and all publications related to AIP between January 2006 and June 2014 were retrieved. Indexing terms used for searching were autoimmune pancreatitis, IgG4-related disease, lymphoplasmacytic sclerosing pancreatitis, idiopathic duct centric pancreatitis, chronic pancreatitis with autoimmune diseases. Inclusion and exclusion criteria were delineated before commencement of the literature search. Inclusion criteria included: (1) cases were Chinese patients; (2) the diagnostic criteria of AIP were clearly defined and consistent with the current accepted ones, which included ICDC [4] and criteria proposed by Japan [11,12], Mayo Clinic [13], Korea [14], and Asian (Korea-Japan symposium) [15]; (3) The number of patients in single article is 10; (4) if multiple articles were from the same institution, the one with most patients were selected. All articles other than original contributions, such as case reports, reviews, meta-analyses, editorials and letters were excluded. This systematic review was performed in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement checklist (S1 PRISMA Checklist).

Definitions
In order to include all relevant articles without loss of the fundamental meanings of the terms, clinical presentation and extrapancreatic manifestation were defined in a relatively broad way. Abdominal symptoms included abdominal pain or discomfort. Because the diagnostic criteria for acute pancreatitis were not clearly given in most of the literature, the presentation of acute pancreatitis mentioned in some articles was classified as abdominal symptoms. Bile duct involvement means bile duct stricture, divided into lower part of the common bile duct stricture and hilar/intrahepatic bile ducts stricture. Because ICDC clearly defines the subtypes of AIP, the proportion of type 1 and type 2 AIP was estimated according to studies adopting ICDC as criteria.

Data extraction from eligible articles
Data on the clinical presentations, laboratory tests, imaging examinations, pathological examinations, treatment and outcomes of the patients were retrieved. Missing data or indeterminate definitions were resolved by direct contact with authors if possible and were otherwise considered not available. Two authors (Q.Q.M., L.X.) determined the relevant original articles and extracted the data independently, whereas the third author (W.Y.L) checked the results. If a disagreement existed, the relevant programs were repeated until a consensus was achieved among the authors.

Statistical analysis
Descriptive analysis or meta-analysis was conducted to provide the overall data of clinical presentation, laboratory test, imaging examination, pathological examination, treatment and outcome of the patients. In the meta-analysis, Cochran's Q test was used to test the heterogeneity of results across studies, and I 2 was calculated to quantify the heterogeneity among studies. If the p value was less than 0.10 (Cochran's Q test), the results were considered heterogeneous and random-effects model was used to obtain the pooled estimates. Otherwise, fixed-effects model was applied. The potential presence of publication bias was evaluated statistically by the Begg-Mazumdar's and Egger's test. Meta-analysis was performed with StatsDirect Statistical software, version 2.7.8 (StatsDirect Ltd, Altrincham, UK).

Results
A total of 26 articles (5 English-language and 21 Chinese-language articles) were enrolled in the study ( [8,9,16] and S1 File). The detailed search procedures are outlined in Fig 1. The detailed data related to diagnostic criteria for AIP are outlined in Table 1. In the 5 studies which adopted ICDC as criteria, the proportion of type 2 AIP was 4.7% (8/171). Detailed data and forest plots for all of the variables are listed in S2 File.

Demographics and Clinical Presentation
Among the 706 AIP patients, 577 were men (male/female: 4.47:1) and the range of mean/ median age of the patients was 48.6-67.0 years old.

Imaging examinations
The general manifestation of pre-treatment computed tomography (CT), magnetic resonance imaging (MRI) or ultrasound examination was reported in 627 patients from 25 articles. Among these 627 patients, 337 (53.8%) were classified into diffuse-type, 261 (41.6%) were focal-type and 29 (4.6%) with atypical imaging manifestation could not be clearly classified into either type. Most lesions of focal-type were located in the head (74.4%, 174/234), and the other were located in the body (11.5%, 27/234) and the tail ( (Table 3).

Laboratory tests
Various serum markers were measured in different amount of patients (

Pathology
In the 706 patients, pancreatic cytology specimens were obtained in 79 (11.2%) by endoscopic ultrasonography (EUS)-guided or CT-guided fine-needle aspiration (FNA) and no malignant or atypical cells were found. Pancreatic histology specimens were obtained from 220 (

Discussion
It has been nearly two decades since Yoshida et al. proposed the term autoimmune pancreatitis in 1995 [2] and AIP has now become a focus in research. Our study provided the comprehensive data of Chinese AIP patients. This study showed that the clinical features of AIP patients in China were consistent with the reports from other countries [6,13,17] except for the considerable proportion of pancreatic duct dilation (13.8%). Nearly one-third patients (29.7%) underwent surgeries due to suspected malignancy. The steroid regimen with an initial dose of 30 mg, 40 mg or 0.6mg/kg prednisone per day was effective in most patients with low rates of relapse and complications. AIP is a rare disease and still a diagnostic challenge despite of increasing study in recent years. As a region with high prevalence of pancreatic disease, China would have a large amount of AIP patients. However, the concept of AIP has just been introduced in this decade and the first clinical report about Chinese AIP patients was published in 2008 [7]. In this study, we could only collect 706 patients from 26 articles, which suggested a large gap in the study of AIP between China and developed countries. Nowadays, several AIP diagnostic criteria including Japanese criteria, Mayo Clinic's HISORt criteria, Korean criteria, Asian diagnostic criteria and ICDC were proposed. Our study showed that the criteria above were adopted by different institutions in China, and 58.9% of AIP patients were diagnosed by Asian diagnostic criteria. Asian diagnostic criteria are simple and have good practicability, and may be more suitable for Chinese AIP patients due to similar ethnic background. In ICDC, AIP is classified into two subtypes. Type 1 is recognized to be part of IgG4-related disease and type 2 is characterized by intraductal neutrophilic infiltration and no IgG4 elevation [4]. Type 2 only covered a proportion of 1% to 6% AIP patients in East Asia, while this proportion increased to 15% to 38% in Europe and North America [3,[18][19][20]. The present study revealed that type 2 AIP covered a proportion of 4.7% in China. This low proportion was consistent with the previous studies from Japan and Korea. Notably, type 2 AIP is much more difficult to be diagnosed because there is no specific serum marker and histological examination is always required. However, in clinical practice, an adequate specimen of the pancreas is not frequently available. In this study, 11.2% AIP patients received EUS-guided or CT-guided FNA but most of them only obtained cytology diagnosis. Several recent studies also showed the difficulty of obtaining adequate specimens for histopathological evaluation through FNA [21,22]. From this perspective, the true prevalence of type 2 AIP in China may be underestimated. Due to the low proportion of type 2, our study mainly showed the features of type 1 AIP.
CT and MRI are main pancreatic parenchymal imaging modalities and play an essential role in the diagnosis of AIP. Our study showed some differences compared with ICDC. In ICDC, diffuse enlargement and segmental/focal enlargement are proposed to be typical and indeterminate findings, respectively [4]. However, in our study, 41.6% patients were classified into focal-type. The relatively high proportion of focal-type may be attributed to the fact that about one third patients obtained diagnosis of AIP after surgery and focal-type patients tends to receive surgery with suspect of malignancy, which led to potential selection bias. Moreover, pancreatic duct imaging was adopted by ICDC as an important diagnostic evidence for AIP and long and multiple strictures were recognized as the main manifestation, while dilation of the pancreatic duct was recognized as a rare manifestation to differentiate AIP from pancreatic cancer. In our study, about two-thirds patients showed stricture of pancreatic duct (pooled rate: 68.1%, 95%CI: 49.3%-84.3%) but dilatation was also found in 13.8% patients. In the previous study from our center [16], marked upstream dilatation (duct size > 5 mm) upon stricture was seen in 16.0% (4/25) patients and in another report from China [9], dilation of pancreatic duct were found in 45.5% (10/22) type 1 AIP patients. Theoretically, the pathogenesis of AIP is different with ordinary chronic pancreatitis and dilated and irregularly shaped medium-sized and large ducts are rarely found in histology examination. Therefore, further study is needed to clarify the pancreatic duct imaging manifestation of Chinese AIP patients.
It was reported that the sensitivity of IgG4 elevation varies between 44%-95% in different settings [17,26]. In our study, the level of serum IgG4 exceeds the upper limit of normal were observed in 86.0% patients in 9 (34.6%) articles, which indicated the similar diagnostic value for Chinese AIP patients. In the other 17 (65.4%) studies, IgG4 test could not be performed and the diagnosis of AIP depended on other evidences. However, IgG4 was the only recommended serological marker for type 1 AIP in the ICDC, performing IgG4 test would improve the accuracy and simplicity of diagnosing AIP.
Steroid treatment was given to nearly 80% AIP patients in China, including many patients who received surgery for suspected malignancy. AIP is highly responsive to steroid treatment. To date, although the ICDC defines the initial dose of steroid for induction of remission as 0.6 to 1 mg/kg per day, a steroid regimen has not been standardized, and there is no consensus on the duration of induction, the tapering schedule, and the optimal dose and duration of maintenance therapy [4,27]. In China, an initial of 30mg or 40mg prednisone per day is adopted as the most commonly used regimen and our study showed the pooled remission rate as 96.2%, which proved the excellent response of Chinese AIP patients to steroid. Relapse after steroid treatment is an important issue. According to a multicenter, international survey, the relapse rate was 35.8% (245/684) after a long-term follow-up [28]. A study from Mayo Clinic showed that 44.8% (52/116) AIP patients experienced relapse during a median follow-up of 47 months [29]. In our study, the pooled rate of relapse was only 13.8% (37/269) with the mean follow-up time ranging from 12 to 45 months. A shorter follow-up time may explain the relatively low relapse rate in China. However, a report from Japan also showed the relapse was 12.5% after a mean follow-up of 43.5 months [30]. We think different steroid regimens may be another explanation. Maintenance steroid therapy, which could decrease relapse [31], was not adopted in most reports from Western countries but was always adopted in Eastern countries [32]. From this perspective, maintenance therapy may be a better choice for patients at high risk for relapse.

Limitations
There are several limitations of this study. First, due to the nature of this systematic review, publication bias could not be avoided and significant heterogeneity existed in many aspects. We believed this is a suboptimal and feasible option to provide an overview of AIP in China, when a national study is not available. Second, type 1 and type 2 were clearly defined in ICDC in 2011; however, not all enrolled studies adopted these criteria, so we could not exactly distinguish type 1 and type 2 AIP. However, with low proportion of type 2, our study mainly showed the features of Chinese type 1 AIP patients.

Conclusions
In conclusion, type 1 is the predominant type of Chinese AIP patients and the clinical features, diagnostic modalities and therapeutic regimen were fundamentally similar with those in other countries. Knowledge of AIP should be more widespread to avoid unnecessary surgery.
Supporting Information S1 PRISMA Checklist. Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement checklist. (DOC) S1 File. Enrolled studies except for these cited in the text.