Autoimmune pancreatitis (AIP) was first described as a clinical entity in 1961 by Sarles et al. who described a type of sclerosing pancreatitis associated with hypergammaglobulinemia [1]. In 1995, Yoshida et al. summarized the clinical features of these patients and named the disease autoimmune pancreatitis [2]. Since then, its further clinical characterization has taken place, resulting in the development of the International Consensus Diagnostic Criteria (ICDC) for AIP [3]. According to these criteria, a diagnosis of AIP is based on five cardinal features; namely, pancreatic parenchymal and ductal imaging, serology, pancreatic histopathology, other organ involvement, and response to steroid therapy. Furthermore, two different histopathological patterns have been recognized with distinct clinical profiles [4]. The predominant type, histologically described as lymphoplasmacytic sclerosing pancreatitis, is a pancreatic manifestation of an IgG4-related systemic disease characterized by elevated serum IgG4 levels and the involvement of other organs due to abundant tissue infiltration of IgG4-positive plasma cells [5]. It shows a male predilection (3–4:1), peaking in prevalence in the 6th decade of life [6]. A final diagnosis can be made based on a combination of clinical features and frequently does not require histopathological confirmation [3]. In contrast, a different type of AIP not related to IgG4 and referred to as idiopathic duct-centric pancreatitis has been reported mainly in North America and Europe ^[7][8][7,8]. The primary histological feature of this type is the presence of a granulocytic epithelial lesion, characterized by focal disruption and destruction of the duct epithelium due to the invasion of neutrophilic granulocytes [7]. This type of the disease usually shows no or very few (<10 cells/HPF) IgG4-positive plasma cells and is not associated with serum IgG4 elevation or the involvement of other organs, as typically seen in lymphoplasmacytic sclerosing pancreatitis ^[6][9][6,9]. However, inflammatory bowel disease is commonly described in association with AIP type 2, with variable prevalence being reported among different studies and ranging from 10.4% to 100% ^{[6][9][10][11][12]}[6,9,10,11,12]. The patients are on average a decade younger than patients with lymphoplasmacytic sclerosing pancreatitis and there is no sex predilection [6]. Because there is no serological biomarker or other organ involvement, a histological analysis of the pancreatic tissue is necessary for a definitive diagnosis ^[3][9][3,9]. Because histopathology is not always available, the terms type 1 and type 2 have been introduced to describe the clinical profiles associated with lymphoplasmacytic sclerosing pancreatitis and idiopathic duct-centric pancreatitis, respectively [3]. If the distinction between the two subtypes is not possible based on ICDC, these cases apply to the category of AIP-not otherwise specified (AIP-NOS). The diagnosis of AIP-NOS is therefore established as the standard diagnosis in the absence of conclusive criteria for the diagnosis of probable or definitive AIP type 1 as well as the absence of histological criteria for AIP type 2 or concurrent IBD [3]. Although the understanding of AIP has advanced significantly over the last decade, diagnosis of AIP remains challenging because the clinical and radiological presentation is often indistinguishable from pancreatic cancer [13]. As a result, a significant number of patients receive a major operative procedure for a disease process that generally responds well to corticosteroid therapy ^[14][15][16][14,15,16]. On the other hand, delay in diagnosis and treatment of pancreatic cancer has disastrous consequences, and prompt surgical treatment offers the only chance of survival [14]. These two diseases are most reliably distinguished by histopathological examination of resected specimens. Therefore, previously reported series of patients that underwent pancreatic resection for a presumed pancreatic malignancy but were diagnosed with AIP based on histology findings are of great research value for describing the histopathological, demographic, clinical, and radiological characteristics of AIP, as well as any diagnostic pitfalls.

2. Autoimmune Pancreatitis as a Diagnostic Challenge

Currently, the ICDC provides the most accurate guidelines for diagnosing AIP [3]. Despite well-defined diagnostic criteria, clinical and radiographic features that distinguish AIP from pancreatic cancer can be subtle and difficult to acknowledge. Patients with AIP typically present with obstructive jaundice, abdominal pain, and weight loss, which overlap considerably with the symptoms usually seen in pancreatic malignancy [17]. Furthermore, typical imaging findings of a diffusely enlarged pancreas, especially with a capsule-like rim, and irregular narrowing of the main pancreatic duct without upstream dilatation are frequently lacking. Instead, a focal inflammatory mass and gland atrophy, with or without ductal dilatation ^[13][17][13,17], is commonly revealed. Serological biomarkers can potentially aid in definitive diagnosis and should be measured if IgG4-related gastrointestinal disease is suspected [18]. However, no specific biomarkers for AIP have been identified to date [19]. Serum IgG4 levels lack the sensitivity and specificity to establish a diagnosis of AIP or to distinguish it from other diseases such as primary sclerosing cholangitis, pancreatic cancer, and other hepatopancreatobiliary diseases ^[20][21][20,21]. Fold elevation above normal rather than absolute values is recommended, with an elevation >2 times the upper limit of normal being strongly suggestive of AIP in the setting of pancreatic mass ^[18][21][18,21]. Nevertheless, type 2 AIP is generally not associated with IgG4, and a proportion of type 1 AIP patients are seronegative ^[6][9][6,9]. Similarly, serum levels of CA 19-9, which is used as a tumor marker of pancreatobiliary malignancy, are influenced by other factors and can be elevated in benign biliary diseases presenting with jaundice and nonspecific causes of pancreatic inflammation. Therefore, the diagnostic accuracy of CA 19-9 in differentiating AIP from pancreatic cancer is limited when used alone ^[22][23][22,23]. The involvement of other organs as part of IgG4-related systemic disease is common in patients with type 1 AIP and has been demonstrated in the extrapancreatic bile ducts; salivary, lacrimal, and thyroid glands; kidneys; lungs; mediastinal and abdominal lymph nodes; the aorta; and the retroperitoneum [24]. Patients may therefore present with obstructive jaundice due to IgG4-related sclerosing cholangitis causing bile duct strictures, bilateral lacrimal and salivary gland swelling due to dacryo- and sialadenitis, or lymphadenopathy, either generalized or localized. Alternatively, they may develop tubulointerstitial nephritis, pneumonitis, aortic aneurysms or dissections due to aortitis, mediastinal or retroperitoneal fibrosis [24]. Extrapancreatic manifestations can be diagnosed by imaging, clinical examination, or histological evaluation of the affected tissue. The well-described pattern of other organ involvement is an important clue and suggests AIP rather than pancreatic cancer [3]. When present, it strengthens the diagnosis of AIP and possibly allows for histological confirmation, especially since some other tissues are readily amenable to biopsy [3]. However, extrapancreatic manifestations do not necessarily occur simultaneously but may precede or be subsequent to AIP, therefore not pointing toward a diagnosis of AIP at the time [24]. Again, type 2 AIP has no relation to IgG4 and usually lacks the involvement of other organs, but it should be strongly considered in the presence of concurrent inflammatory bowel disease ^[9][10][9,10]. Finally, the two types of AIP display a typical pathohistological pattern; however, histology is not usually available because adequate pancreatic tissue is difficult to obtain [3]. Tissue obtained by endoscopic ultrasound (EUS) fine-needle aspiration (FNA) may not be representative, given the limited caliber of the needle and architectural distortion ^[25][26][25,26]. Moreover, the results are commonly suggestive of pancreatic cancer demonstrating ductal atypia ^[27][28][27,28]. Therefore, EUS FNA is not recommended for diagnosing AIP according to ICDC guidelines. Instead, tissue acquisition from either surgical resection or a pancreatic core biopsy specimen may be used ^[3][29][3,29]. Core biopsies, conducted through EUS-trucut biopsy (EUS-TCB) or newer-generation EUS fine-needle biopsy (EUS-FNB), enable greater tissue specimen size and better histopathologic review [29]. According to European guidelines on IgG4-related digestive disease, EUS may demonstrate hypoechoic pancreatic enlargement and other features suggestive of AIP. Furthermore, it is used for obtaining tissue samples for histological diagnosis [30]. EUS-guided tissue acquisition with a core biopsy using a 19-gauge needle is recommended; however, even a 22-gauge needle can be used to obtain a sample allowing for histological evaluation ^[30][31][30,31]. Additionally, IgG4 immunostaining of biopsy specimens from the major papilla may advance a diagnosis of AIP as it is often involved ^[3][30][32][3,30,32]. Future advances in endoscopic technologies, enabling them to take sufficiently large biopsy samples, will likely improve the diagnostic process. New types of EUS-guided needles for adequate pathological samples have now been developed, but their diagnostic utility is yet to be evaluated in future prospective studies [29]. In patients with a suspicion of AIP and negative work-up for cancer including EUS FNA, a diagnostic steroid trial may be helpful in confirming the diagnosis of AIP [3]. Administration of high-dose prednisone (0.6–1 mg/kg/day) is followed by a reassessment of imaging and CA 19-9 after 2–4 weeks of treatment. Rapid definite improvements in imaging abnormalities and a decrease in CA 19-9 levels, if elevated before treatment, are expected ^[24][30][33][24,30,33]. After treatment response, the prednisone dose is tapered by 5 mg each week until discontinuation ^[24][30][24,30]. In case of poor response to steroid therapy, the diagnosis should be reconsidered. However, a steroid trial needs to be conducted with caution in well selected patients since various conditions, including peritumoral pancreatitis, potentially respond to immunosuppression [3].

3. Surgical Experience in Autoimmune Pancreatitis

Given the diagnostic challenges described above, pancreatic resection may be required to exclude pancreatic malignancy and establish a diagnosis of AIP. Among pancreatic resections performed due to suspected malignancy, benign conditions have been found in 8 to 10% of patients upon final histological examination. Among these, AIP represents about a third of cases, accounting for 2.5% of all pancreatic resections ^{[15][16][34][35]}[15,16,34,35]. These cases should be prevented, since the postoperative course after pancreatic resections can be marked by several life-threatening complications. The most significant one is pancreatic fistula, with its septic and hemorrhagic sequelae. Various modalities, such as several modifications of anastomotic techniques, the placement of pancreatic duct stents and prophylactic use of somatostatin analogues, have been implemented to improve postoperative outcomes. Although perioperative mortality has decreased significantly with gaining surgical experience and critical care management, morbidity rates remain high, even in high-volume centers ^[36][37][36,37]. Therefore, adequate preoperative workup is crucial in order to improve patient selection to avoid unnecessary surgery. A review of surgical case series published in the past 10 years clearly demonstrates the difficulties in AIP diagnosis (Table 1). Patients with AIP that were candidates for surgery frequently lacked the typical findings of AIP. Therefore, AIP was not necessarily suspected before surgery, which resulted in an insufficient preoperative workup. For example, only a few studies reported preoperative serum IgG4 levels, which were measured in a range from 14 to 57% of the study cohort ^{[15][34][35][38][39][40]}[15,34,35,38,39,40]. Ikeura et al. determined serum IgG4 levels to be present in 77% of the study cohort; however, these are cumulative measurements, taken before or after surgery [41]. Furthermore, serum levels of IgG4 were generally not helpful in differential diagnosis between AIP and malignant lesions because they were elevated in only about a third of patients in whom AIP type 1 was pathologically confirmed after surgery ^{[15][34][38][40]}[15,34,38,40]. In contrast, CA 19-9 was more commonly assessed, and elevated CA 19-9 levels were found in 25 to 52% of patients ^{[15][34][40][42]}[15,34,40,42]. Javed et al. even reported a greater proportion of patients with AIP that presented with elevated CA 19-9 than with elevated IgG4 (47.1 vs. 26.1% for CA 19-9 and IgG4, respectively) [34]. Preoperative tissue sampling was performed in 0 to 73% of patients, among whom suspicion of malignancy was reported in 5 to 25%, although they eventually proven to be false positives ^{[15][34][35][39]}[15,34,35,39]. However, histopathological confirmation of a malignant disease is not necessary before proceeding with surgical resection of a pancreatic solid mass according to the International Study Group of Pancreatic Surgery [14]. Again, limited data on preoperative tissue sampling may indicate the lack of preoperative suspicion of AIP. The presence of extrapancreatic manifestations of IgG4-related systemic disease may to help raise the suspicion of AIP; however, these were reported in 13 to 71.4% of cases and sometimes occurred only after surgery had already been performed ^{[15][35][40][41][43]}[15,35,40,41,43].