Pancreatic tumor in type 1 autoimmune pancreatitis: a diagnostic challenge

Autoimmune pancreatitis (AIP) was first proposed as a clinical entity by Yoshida et al. in 1995 [1]. Current consensus suggests AIP encompasses two different types with distinct histological and clinical profiling. Type 1 AIP is recognized as a pancreatic manifestation of IgG4-related systemic disease. Histologically, it is termed lymphoplasmacytic sclerosing pancreatitis (LPSP) or AIP without granulocyte epithelial lesions (GELs). It is characterized by periductal infiltration of lymphocytes, storiform fibrosis, abundant IgG4-positive plasma cells, and obliterative phlebitis. Clinically, predominance has been observed in elderly males, with elevated serum IgG4 levels, accompanying extrapancreatic involvement (like bile ducts, salivary glands, retroperitoneum, kidneys etc.), and an exclusive response to steroids [2‐5]. The prognosis is considered favorable. The long-term survival of type 1 AIP patients has been proved to be comparable to the age- and gender-matched subjects from the general population [4]. The treatment primarily involves inducing remission and reducing relapse with steroids or sometimes immunosuppressants [6, 7].

The synchronous and metachronous occurrences of type 1 AIP and pancreatic tumors (PaT) have been previously reported. In AIP patients, the occurrence rate of PaT is about 0.1–4.8% [5‐12]. This co-occurrence suggests that the definitive diagnosis of AIP cannot rule out the concurrent presence of PaT. When PaT is identified in AIP patients, management changes to diagnostic and therapeutic workup for cancer, where the principal therapy is surgical resection accompanied by chemotherapy, radiation therapy or targeted medical treatment. The prognosis would be variably poorer than that of pure AIP cases, depending on the subtype, stage and therapy choice of the concurrent PaT [13, 14]. However, AIP shares similar clinical and radiological manifestations with PaT where diagnosis can be challenging. This includes: abdominal pain, obstructive jaundice, weight loss, elevation of serum IgG4 and CA-199 levels, mass formation in pancreas, pancreatic atrophy, narrowing of main pancreatic duct (MPD) and common bile duct (CBD), decreased enhancement compared to normal pancreas, and peripancreatic vessel stricture [5, 15]. These features may obscure the diagnosis of concurrent PaT in patients with AIP, which subsequently lead to the delay of appropriate therapy. As far as we know, several studies have attempted to increase the detection of concurrent tumor in AIP, but many non-PaT tumors were also studied [8, 12, 16]. Only one study has investigated the difference between AIP patients with and without PaT, where no significant difference was observed [8]. The purpose of this study was to further evaluate the clinical and radiological markers with the aim to increase the detection of concurrent PaT in type 1 AIP.

In the present study, we found that the gender, pattern of pancreatic enlargement, pancreatic heterogeneity, low-density on enhanced CT and rim enhancement on MRI were significantly different between type 1 AIP patients with and without PaT. 3 of 5 AIP + PaT patients were females. All of them demonstrated focal pancreatic enlargement, heterogeneity and low-density on CT. The details of low-density on CT didn’t work significantly. Only one AIP + PcT patient had the MRI data, and showed the characteristic of rim enhancement. None of the pure AIP patients presented with rim enhancement on MR imaging. According to our knowledge, there was only one comparable study that has attempted to identify the difference between type 1 AIP patients with and without PaT, but no statistically significant differences were concluded [8].

The prevalence of PaT in AIP patients was about 5.4% (3/56) in the first medical institution and 11.1% (2/18) in the second medical institution. The general prevalence of concurrent PaT in type 1 AIP in the two institutions was about 6.7% (5/74), which was higher than other studies [5‐9, 11, 12]. The reason for the higher prevalence of PaT in AIP patients, especially in the second medical institution, may be a result of potential selection bias. The second institution is a cancer specific center, where non-cancer patients such as pure AIP cases may not be referred for surveillance. The definite association between AIP and PaT remained inconclusive. Some researchers have suggested that AIP may be a manifestation of paraneoplastic syndrome [10, 16], some have indicated that chronic pancreatic inflammation in AIP may contribute to carcinogenesis [8, 12, 19], and some have reported that AIP was not associated with an increased incidence of total malignancies [20, 21]. Our results tended to support the hypothesis of paraneoplastic syndrome, which would be briefly explained below.

The concurrent PaTs were identified 0, 3 and 6 months after the diagnosis of AIP, respectively. But during retrospective review, all of the PaTs were concurrently present with AIP (this synchronous occurrence may indicate the association between AIP and paraneoplastic syndrome). Several studies have reported that the occurrence of cancer in AIP patients was significantly higher in the first year than in the subsequent years [12, 16]. Accordingly, some PaTs were identified at the same time of AIP diagnosis and some were found during the follow-up period that could last as long as 186 months [7‐9, 12, 22, 23]. Therefore optimal surveillance duration in AIP patients cannot be concluded for early PaT detection. However, close observation should be warranted during the first year post diagnosis of AIP given the diagnostic challenge and relatively high incidence during the first year. In this cohort, the delay in diagnosing the concurrent PaT was mainly attributed to the negative EUS-FNA results of pancreas at the initial visit. It indicated that the negative EUS-FNA results of pancreas couldn’t rule out the presence of PaT, which was consistent with the results of some previous studies [19, 24]. Several studies showed that serum IgG4 levels at the diagnosis of AIP were significantly higher in AIP patients with cancer than those without cancer [12, 16]. In our study, the serum IgG4 level in 4 of 5 AIP + PaT patients went beyond the 2 times upper limit of normal value, though it didn’t differ significantly between AIP patients with and without PaT. The types of included cancers in the previous studies were mainly non-PaT tumors [12, 16], it remains questionable whether different types of concurrent cancer would result in different levels of serum IgG4 elevation. Age, serum CA-199 levels or extrapancreatic involvements did not vary significantly between the two groups, which were consistent with previous studies [8, 12]. In the current study, the only significant clinical characteristic was gender and 3 of 5 AIP patients with PaT were females. Comparably, Ikeura et al. found that 2 out of 3 AIP patients with PaT were females, though gender difference between the two groups didn’t reach statistical significance [8]. However, there were previous studies where the reported AIP + PaT patients were all males [9, 19]. The reason for the discrepancy between our and previous study is unknown. This may be due to the limited sample size in both current and previous studies. The female AIP + PaT predominance in our cohort may suggest that at least some AIP cases are associated with paraneoplastic syndrome. When type 1 AIP appears in an uncommon population, intense attention should be paid to exclude the concurrent presence of a tumor.

The CT imaging characteristics of focal pancreatic enlargement, pancreatic heterogeneity and low-density on enhanced phase were significantly different between AIP patients with and without PaT in this cohort. All of the three CT imaging features were detected in all of the 5 cases with accompanied PaT. Comparably, Ikeura et al. also found that all of the 3 AIP patients who developed PaT showed the feature of focal pancreatic enlargement, but this characteristic didn’t differ significantly between the comparison groups [8]. When the inclusion criteria was set as focal pancreatic enlargement that underwent resection, the incidence of concurrent PaT in type 1 AIP could be as high as 33.3% (2/6) [19], which was much higher than previously mentioned. Besides, Ikeura et al. found 2 out of 3 AIP + PaT patients presented with pancreatic low-density, but statistical significance was not provided [8]. The international consensus considers pancreatic low-density as an atypical imaging finding of AIP and suggests such patients should be managed as PaT unless an alternative diagnosis is strongly suspected [2]. Therefore concurrent pancreatic malignancy needs to be excluded in the presence of focal pancreatic enlargement and low-density in AIP. Notably, focal pancreatic enlargement, pancreatic heterogeneity and low-density did appear in pure AIP cases as we have shown, so the presence of these features do not validate concurrent PaT. Furthermore, the diagnosis cannot be further made easier by the interpretation of other CT imaging characteristics. 4 of 5 AIP patients with accompanied PaT presented with mass-like low-density despite no statistically significant difference was observed between AIP patients with and without PaT. Notably, the shape of mass-like low-density was also noted in small portions of pure AIP cases (6/21). Our results showed that the sizes of concurrent PaT measured in specimen were comparable with initial CT measurement, which indicated the presence of mass-like low-density was highly suggestive of concurrent tumor. However, when the tumor was too small to exceed the contour of pancreas on imaging, it was hard to conclude the presence of concurrent PaT. In this situation, close surveillance was of vital importance, which was supported by the fact that 3 cases of AIP + PaT became highly suspicious for malignancy during the follow-up period in our cohort. On the other hand, if the PaT was very large and appeared obvious in the background of pancreas on imaging, the diagnosis of AIP + PaT may be easy to make. Consequently, the morphologic criteria should be modified with the stage or the size of the tumor. The effect of size difference between pathology and radiology on the interpretation of mass-like low-density needs future investigations to explore, where larger sample size are warranted. In this cohort, the low-density areas in both AIP and PaT portions presented as hypoattenuating region during arterial phase and either hypoattenuating or isoattenuating during venous phase. There was not statistically significant absolute density value identified to indicate the underlying pathological differences between the two. So the presence of pancreatic low-density regions on CT in patients with AIP was only suggestive for the presence of concurrent PaT. Other imaging characteristics suggestive of PaT include MPD dilatation, pancreatic parenchymal atrophy and peripancreatic vessel stricture. However, these characteristics were also noted in pure AIP cases, which was in accordance with previous studies [3, 14, 25]. These imaging features in AIP may be attributed to the longstanding intra- and peri-pancreatic inflammation and fibrosis in AIP.

In the aspect of MR imaging characteristics, our results showed that AIP + PaT patient could present with rim enhancement on MR imaging, and none of the included pure AIP cases showed this imaging characteristic. So it may be a useful MR imaging characteristic to differentiate pure AIP cases from AIP + PaT cases. However, neither the role of MR imaging characteristics in differentiating AIP patients with and without PaT nor the presence of rim enchantment on MR imaging in pure AIP cases has ever been studied previously [17, 18]. Moreover, there was only one AIP + PaT case included in the present cohort. So it is still too early to draw a final conclusion on the role of MR imaging in differentiating the two groups, and future studies with larger sample size are recommended.

Besides CT and MR, the case of AIP with SEP in our study implies that ¹⁸F-FDG PET/CT may add some value where the difference in FDG uptake may help identify the concurrent PaT. However, Shiokawa et al. [16] reported that none of AIP patients with cancer were identified with ¹⁸F-FDG PET/CT, though there was no case of PaT in their cohort. Further studies are required before definitive conclusion can be made.

There are limitations to our study. Firstly, though we have included the AIP + PaT data of two institutions, only a small size of cases was identified due to its rare entity. Secondly, the AIP patients in the two medical institutions follwed different diagnostic and follow-up pathways due to various reasons where different imaging modalities were used for surveillance which further limited the sample size as some did not meet the selection criteria.

When type 1 AIP patients are females and present with focal pancreatic enlargement, pancreatic heterogeneity, low-density on CT and rim enhancement on MR imaging, it should be prudent to exclude the synchronous presence of PaT. The CT imaging characteristics of pancreatic low-density or other MR imaging characteristics of intra−/peri-pancreatic structures are not reliable to conclude the presence of concurrent PaT.