Radiologic imaging of the pancreas is an essential component of the Japanese criteria where it is mentioned as a “must” for diagnosing AIP. The criterion is stated as diffuse narrowing of the main pancreatic duct with irregular wall (more than 1/3 length of the entire pancreas) and diffuse enlargement of the pancreas which can be identified with ERCP (endoscopic retrograde cholangiopancreatography) and CT, respectively[9]. Although abdominal US is most frequently performed imaging method for the screening of pancreatic disease, overlying bowel gas or obesity often hinders the sonographic visualization of the pancreatic gland, rendering the examination limited in scope and quality.

On CT, most cases of AIP reveal a diffusely enlarged pancreas with no or mild peripancreatic fat infiltration. Phlegmonous changes of the pancreas, peripancreatic fluid collection or pseudocysts formation are rare. Diffuse enlargement of the pancreas correlates with the pathology of marked stromal edema, which manifests as diffuse swelling on gross examination[16,17]. Surprisingly, CT does not reveal any typical findings of chronic pancreatitis, such as multiple parenchymal calcifications or pancreatic ductal stones. Rather, AIP resembles mild form of acute pancreatitis according to Balthazar classification[18]. Some cases of AIP reveal peculiar CT findings, such as a capsule-like low-density rim surrounding the pancreas (Figure 1)[19]. This rim is thought to correspond to the histologic findings of an inflammatory process that contains fibrous changes involving the peripancreatic adipose tissue[20]. Delayed enhancement of the pancreatic parenchyma is another distinguishing feature of AIP. On arterial enhanced phase, the pancreas appears hypodense when compared to the spleen. On the delayed phase, attenuation increases when compared to early images. This reflects fibrosis with an associating inflammatory process. These characteristic patterns observed on CT are important in differentiating AIP from pancreatic cancer[15].

The term “enlargement” of the pancreas can be subjective and vague. The size of the pancreas may be affected by many factors, including body mass, ethnic group, gender and age. There are individual variations in the size of the gland, with smaller atrophic glands seen in older individuals[21,22]. While a pancreas may seem normal for a large young man, the same size can be described as “enlarged” for an elderly patient with small body mass. Nishino et al[23], therefore, describe that the pancreas is considered to be enlarged when the width of the pancreatic body or tail was more than two thirds of the transverse diameter of the vertebral body or the width of the pancreatic head is more than the full transverse diameter of the vertebral body.

The Japanese imaging criterion has described the ductal pathology as diffuse narrowing of the main pancreatic duct with irregular wall (more than one third of the length of the entire pancreas), which can be observed on direct pancreatogram[9]. However, we were confronted with confusion when applying this criterion. This may be because the terms used were imprecise and vague in the meaning. First, in the previous international symposium on chronic pancreatitis[24,25], pancreatic lesion was classified as focal, segmental, or diffuse according to the extent of involvement. In this international classification, “diffuse” is used when a process involves the entire pancreatic duct or at least more than two-thirds of the entire length of the main pancreatic duct, whereas lesions that are not continuous and involve the head, body or tail are defined as “segmental” (Figures 2 and 4)[24,25]. In Japanese criteria, the terms “diffuse” and “at least one third of the entire length” in the same sentence are contradictory to each other. Thus, it may be more appropriate that the term “diffuse” irregular narrowing of main pancreatic duct in Japanese imaging criterion (Table 1) is changed to “diffuse or segmental” irregular narrowing. In AIP cases of segmental involvement, the intervening normal-appearing duct upstream to the stricture shows no or minimal dilatation in spite of the long stricture (Figure 4). These findings differentiate AIP from pancreatic cancers, which reveal stricture associated with marked upstream ductal dilatation[26,27]. Second, “narrowing” of the main pancreatic duct is a term based on subjective assessment. At the present time, there are no references for how thin “narrowing” is. Like the size of pancreatic gland, duct diameter also varies by age, gender, and size of patient, etc. In the cases of AIP, most patients are elderly, which complicates matters because the diameter of the main pancreatic duct physiologically increases with age[28,29]. In patients with AIP, the ductal diameter widens to normal caliber after steroid therapy (Figures 2 and 4).

Open in New Tab   Full Size Figure   Download Figure

Figure 4 ERCP findings. A: Segmental irregular narrowing (arrow heads) of the main pancreatic duct that involves the pancreatic head and tail portion is noted in a 53-year-old man; B: After steroid therapy, the narrowing sites on preceding ERCP were resolved. Biliary stenting for narrowing of intrapancreatic common bile duct is noted.

Some authors suggested that as a Japanese length criterion of direct pancreatogram, “more than 1/3” of ductal narrowing (Table 1) should be changed to “more than 1/4” because the extent of ductal narrowing was between 1/4 and 1/3 of the entire length of main pancreatic duct in some cases of AIP[30]. Actually, in our series, the extent of irregular narrowing was less than 1/3 of the entire length of main pancreatic duct in 7 of 28 (25%) patients (Figure 3, Table 3). Moreover, a variant form of AIP that is characterized by focal parenchymal swelling with a localized short stenosis of the main pancreatic duct and evident upstream dilatation has been reported[31]. Because this focal type is a rare form of AIP and mainly diagnosed after laparotomy, further discussion is beyond the scope of this review. However, studies reported that segmental narrowing progressed to diffuse narrowing on serial ERCPs without steroid treatment[32,33]. In other words, “focal” and “segmental” types can evolve into “diffuse” with time, which implies that the total length of the stricture can be less than one third at an earlier stage of the disease[32,33]. Last, “irregular wall” is another confusing term. While Japanese criteria use the term of “irregular” to portray the marginal irregularity of the narrowed ductal wall[30], we have experienced AIP patients with direct pancreatograms that reveal smooth margins not uncommonly. This may be related to the fact that pathologic specimens often reveal inflammation confined to the subepithelial space with an intact ductal epithelial lining[34,35]. Therefore, the term “irregular wall” may not always be appropriate for AIP. On the other hand, the Cambridge classification for chronic pancreatitis uses “irregular” (i.e., ‘irregular dilatation’ of main pancreatic duct) to describe the overall contour of the main pancreatic duct[24]. These pancreatographic findings may be explained by the heterogenous pattern of inflammatory infiltrates and fibrosis noted in pathology[34].

The intriguing point of imaging is that CT findings resembling mild acute pancreatitis are quite unusual when associating ERCP findings, which reveal diffuse or segmental irregular narrowing of the main pancreatic duct, are considered. Narrowing of the main pancreatic duct with or without intrapancreatic common bile duct stenosis on ERCP is a finding that is rarely seen in acute pancreatitis, but often in chronic pancreatitis[36]. While diffuse enlargement without calcification or stone is common in acute pancreatitis, this finding is rare in chronic pancreatitis and parenchymal atrophy is more frequently observed in ordinary chronic pancreatitis. CT findings have no evident chronic parenchymal changes in contrast to the ductal pathology detected on ERCP[37]. Coexistence of such contradicting radiologic findings (CT and ERCP) in the same patient was a rare yet peculiar association that was consistently observed in patients with AIP[38].

We believe that clinicians should vigorously obtain direct pancreatograms when patients show unusual clinical features or atypical clinical manifestation in patients suspected of pancreaticobiliary malignancies. Although direct pancreatogram holds a critical position in diagnosing AIP, hardships follow in obtaining it. Due to the possible risk related to the procedure, the visualization of pancreatic duct is not always carried out when the main pancreatic duct is not dilated on US or CT in patients with diffuse pancreatic swelling alone[39,40]. There are also pitfalls in interpreting pancreatograms. Because the main pancreatic duct is extrinsically compressed, it is technically difficult to attain whole ductal images and, sometimes, sufficient amount of contrast is not introduced into the main pancreatic duct. An “abrupt cutoff” of the main pancreatic duct may conceal a true “irregular narrowing” when ERCP is performed inadequately. Deep cannulation of catheter up to the tail portion guided by a thin guide wire can help circumvent this problem. While magnetic resonance cholangiopancreatography (MRCP) is replacing diagnostic ERCP in many pancreaticobiliary diseases, it seems to be limited in delineating the detailed pathology of main pancreatic duct associated with AIP. AIP is basically a “narrow-duct” disease and the resolution of MRCP is inferior to that of ERCP in this aspect[41,42].

Abnormally elevated levels of serum gamma globulin and/or IgG or the presence of autoantibodies are described as the laboratory criterion of AIP by the Japan Pancreas Society[9]. In general, these features are known facets of autoimmune diseases, which provide laboratory evidence of an autoimmune pathogenesis[43]. By measuring various autoantibodies, a number of candidates have emerged for clinical usage in AIP. Detection rates for autoantibodies, however, have varied among reports[44,45]. This difference may be accounted by the number of measured autoantibodies, which can affect detection rates of autoantibodies; rates tend to increase as more autoantibodies are checked[46]. Among the conventional autoantibodies that are commonly investigated in autoimmune diseases, anti-nuclear antibody and rheumatoid factor are more frequently detected in AIP. Other markers, including anti-smooth muscle antibody and anti-mitochondrial antibody, have failed to show detection rates above 10%. On the other hand, anti-lactoferrin antibody (ALF) and anti-carbonic anhydrase II antibody (ACAII) are relatively organ-specific autoantibodies, which reveal the highest detection rates (over 50%) among autoantibodies in AIP[47]. Lactoferrin is normally present in the pancreatic acinus and carbonic anhydrase II in the ductal cells of the pancreas. ALF and ACA II, however, require a special laboratory for measurement that is unavailable to many clinicians. And, even if ALF and ACA II are measured, these markers are not 100% sensitive[48,49]. This implies that although carbonic anhydrase II and lactoferrin are considered to be the most likely target antigen for AIP, cases without ALF and ACAII, despite a good response to steroid therapy, also exist. ALF and ACAII are not specific to AIP as they may be elevated in other diseases[48,49]. Elevated levels of ACAII were also reported in patients with pancreatic cancer or ordinary chronic pancreatitis[10,50,51].

Elevated IgG and hypergammaglobulinemia are not always seen in patients with AIP. IgG4, a subtype of IgG, levels have been reported to be able to distinguish AIP from other pancreatic disorders with a high sensitivity (95%) and specificity (97%)[52]. Moreover, some cases of AIP show elevated serum levels of IgG4 in spite of normal IgG levels[53]. We, therefore, believe that both IgG and IgG4 levels should be measured in all patients suspected with AIP. In Japanese criteria, however, the increased IgG4 level was not included in the laboratory criteria and only mentioned in the appendix that there are cases with elevated levels of IgG4[9]. The Japanese criteria for laboratory data states that IgG levels should be higher than 1 800 mg/dL (normal range, 614-1295 mg/dL[54]). Although cutoff value of IgG was set at 1 800 mg/dL in Japanese criteria, another group uses different cutoff value (1 700 mg/dL) of IgG for the diagnosis of AIP[55]. If we set cutoff value of IgG at higher level, the specificity of IgG increases but the sensitivity decreases. More evidence on sensitivity and specificity according to each cutoff value should be provided for IgG before cutoff level is established.

While serologic markers have provided a base in understanding AIP, there are still controversies. As autoantibodies and IgG can rise non-specifically during the course of various injuries and diseases, the mere increment can not indicate a cause-and-effect relationship. Moreover, elevation of IgG4 level has been reported in patients suffering from pancreatic carcinoma and other types of chronic pancreatitis[50]. A number of groups have tried to find other laboratory indicators of AIP. HLA may identify patients susceptible to AIP. One report mentioned that frequencies of DRB1*0405 and DQB1*0401 were significantly higher in patients with AIP when compared with chronic calcifying pancreatitis[56]. At the present time, however, further studies are required to evaluate the value of each laboratory indicator and find a more reliable one.

The Japanese criteria described the histopathologic findings of AIP as marked inflammatory infiltrates mostly of lymphocytes and plasma cells and dense fibrosis of the pancreatic tissue[9]. In one report, however, lymphoplasma cell infiltration was minimal in one third of AIP cases. In general, marked infiltration of inflammatory cells is noted in early stage of an autoimmune disease, but fibrosis becomes predominant as the disease progresses. This suggests that degree of inflammatory cell infiltrates and fibrosis and predominance of either one may be dependent on the stage of AIP[57]. The characteristic lymphoplasmacytic infiltration and fibrosis may not be pathognomonic to AIP. Alcoholic chronic pancreatitis (ACP) also showed an abundant amount of lymphoplasma cell infiltration and fibrosis[58]. Although periductal inflammation and fibrosis were observed in both ACP and AIP, the detailed histologic patterns differed. While ACP mainly revealed dense interlobular fibrosis with a cirrhosis-like appearance, AIP showed loose fibrosis in both interlobular and intralobular region[58-60]. AIP showed more severe and diffuse acinar atrophy than ACP. As for extracellular matrix proteins, collagen type IV, a component of the intact basement membrane that ensures accurate regeneration of tissue, was preserved in cases associated with AIP, whereas ACP showed a significant loss of this collagen subtype[58]. Not only may this help differentiate AIP from ACP, but also aid our understanding of the regeneration of acinar structures and regression of fibrosis following steroid treatment. More specific histologic features, i.e., the pattern of fibrosis and inflammation, should be supplemented for histopathologic criterion of AIP.

Periductal fibrosis and inflammatory infiltrates surrounding the duct like a cuff compress the ductal lumen into a star-like structure[61]. Intrapancreatic portion of common bile duct is often involved and the biliary involvement in AIP develops by the same mechanism as the pancreatitis[23]. Some reports of AIP show predominant neutrophilic or eosinophilic granulocytes infiltrating into the pancreatic parenchyma rather than lymphoplasma cell infiltration[62]. This acute inflammatory component of AIP is characterized by focal detachment, disruption and destruction of the duct epithelium due to invading granulocyte, which has been named “granulocytic-epithelial lesions” of the ducts. The extension and severity of chronic and acute changes in AIP vary from case to case, and even from area to area within the same pancreas.

Many clinicians come across difficulties in obtaining adequate pancreatic specimens for histologic evaluation[48]. Fine-needle aspiration biopsy often fails to gain sufficient amounts of pancreatic tissue. Large-bore needle biopsies may be used to yield adequate amounts for pathologic examinations. However, this is at the cost of increasing risks of procedure-related complications. Because of the patchy nature of inflammation seen in AIP, percutaneous biopsy may not be diagnostic due to sampling error problems[13]. Another problem is the potential risk of tumor seeding during biopsy in patients in whom cancer can not be omitted. Due to this reason, endoscopic ultrasonogram (EUS)-guided biopsy is recommended for patients in whom pancreatic cancer can not be excluded[63]. This approach is useful because the pancreatic head is the most frequently involved portion in AIP. One recent paper has reported the usage of EUS-guided trucut biopsy, which may help surmount the above problems and allow optimal histologic examination in AIP[64].

Although histologic evaluation offers a gold standard in diagnosing many disease entities, its role maybe a little different in autoimmune conditions, as seen in primary sclerosing cholangitis and autoimmune hepatitis[65,66]. In these conditions, the histopathologic findings are not disease-specific. Biopsy is often performed to exclude other entities that coexist or show resemblance rather than to make a diagnosis. Taken together, the role of histopathologic examination in patients suspected of AIP may be the exclusion of other diseases, such as malignancy, rather than the confirmation of diagnosis.

Although the dramatic response to the steroid is a well-known phenomenon in AIP, a detailed steroid schedule has not been fully established at the present time. Prednisolone is usually initiated at 30-40 mg per day and tapered after confirmation of the response to the steroid. The response to the steroid is defined as improvement in clinical symptoms, negative conversion of detected autoantibodies, normalization of elevated levels of IgG, and reversion of abnormal pancreatic imaging, including CT and endoscopic retrograde pancreatography[58]. In cases of obstructive jaundice associated with intrapancreatic common bile duct narrowing, however, biliary stenting is often needed additionally. The dose of oral corticosteroid may be tapered by 5 mg each 2-4 wk. Eventually, the steroid is completely discontinued or maintained at a dose of 2.5-10 mg/d according to the preference of the doctor[50,52,67].

The response to the oral steroid provides a circumstantial evidence of underlying autoimmune pathogenesis[68]. Among the responses to the oral steroid therapy, recovery of pancreatic ductal narrowing is top priority in the differential diagnosis of AIP. For the recovery of pancreatic ductal narrowing, histologic recovery including periductal fibrosis should be accompanied. We already reported histologic recovery, especially regression of pancreatic fibrosis in patients with AIP after steroid therapy[69]. Relief of pancreatic ductal narrowing by steroid administration is a unique and specific finding that can not be seen in any other type of chronic pancreatitis or pancreatic cancer (unpublished observation, Myung-Hwan Kim, MD). And, because marked improvement of pancreatic ductal narrowing can be observed at as early as 2 wk after steroid therapy[32,50], steroid trial may be a practical diagnostic tool that has clinical impact, especially when differentiation from cancer is an issue. This is analogous to autoimmune hepatitis where steroid treatment is justified when the entity is highly suspected, and the response to treatment is incorporated into the diagnostic scoring system[66]. If steroid therapy fails to show clinical improvement, imaging studies should be performed again to differentiate AIP from pancreatic cancer. There are opposes against including the response to the steroid into the diagnostic criteria of AIP. They argue that steroid is not usually prescribed in any other type of chronic pancreatitis and one has to strongly suspect AIP in the first place to ever consider steroid therapy. To observe the response to the oral steroid in patients suspected of AIP, however, may be a diagnostic trial as well as therapeutic trial. Actually, in one Japanese university hospital and Italian group, they use the response to the steroid as one diagnostic criterion of AIP[50,70].

There are also concerns of the possibility of cancer progression during a trial of steroid therapy in an operable patient[6]. Despite this risk, we believe that a trial with steroids can be used to guide diagnosis when used in a proper fashion. If a patient shows typical pancreatic images of AIP, a short course (about 2 wk) of steroid may differentiate AIP from pancreatic cancer due to the fact that pancreatic ductal images of malignancy do not change[32,50]. And if the results are equivocal or do not favor AIP, the diagnosis of AIP should undergo reevaluation and the possibility of laparotomy should be considered. By including the response to the steroid into the diagnostic criteria, we can overcome the fact that there are patients who do not satisfy laboratory data and histologic findings, yet reveal typical images and an excellent response to the steroid.

Autoimmune diseases tend to cluster, and one patient may have more than one autoimmune conditions[43]. This is also the case in AIP which is frequently associated with

Sjögren’s syndrome, retroperitoneal fibrosis, primary sclerosing cholangitis, primary biliary cirrhosis, and inflammatory bowel disease[71]. This association may be explained by the fact that carbonic anhydrase II and lactoferrin are present in the salivary gland and biliary duct as well as the pancreas. An autoimmune response against these common antigens may result in “autoimmune exocrinopathy” which describes an autoimmune disease that involves multiple exocrine organs[72]. On the basis of the absence or presence of a systemic autoimmune diseases, Okazaki et al[14] divided AIP into 2 major groups: Primary or secondary forms of the disease. In contrast to the Japanese criteria, Italian group has included the association of other postulated autoimmune disease into their diagnostic criteria of AIP (Table 4)[50].

The prevalence of other postulated autoimmune diseases, however, varies among papers with some reporting rates that exceed 50%[50]. Comorbidities are manifested at various time points on the natural course of AIP; some manifest before AIP and some simultaneously and others after remission[32]. Thus, primary form of AIP may need to be changed to secondary form in cases where other autoimmune diseases are not recognized at diagnosis of AIP but develop later. For patients with AIP, the presence of an associated autoimmune disease may help the diagnosis, but if not present at the onset of the disease, we must carefully search for it.

In addition, pancreatitis in patients suffering from other autoimmune diseases does not always indicate that the cause of pancreatitis is autoimmunity. For example, when pancreatitis occurs in patients with systemic lupus erythematosus, pancreatitis can be related to various causes, such as vasculitis or medications[73]. This also adds confusion in reporting AIP in patients with underlying autoimmune conditions. In these cases, the remarkable response to steroids can help identify autoimmunity as the cause of pancreatitis[74].