Chronic Pancreatitis Finding by Endoscopic Ultrasonography in the Pancreatic Parenchyma of Intraductal Papillary Mucinous Neoplasms Is Associated with Invasive Intraductal Papillary Mucinous Carcinoma

Intraductal papillary mucinous neoplasms (IPMNs) are a well-characterized group of intraductal mucin-producing cystic neoplasms of the pancreas with malignant potential [1,2]. In the 2012 international consensus guidelines, high-risk IPMN included two clinical categories with “high-risk stigmata” and “worrisome features,” and different therapeutic strategies were recommended for both groups. It is recommended that patients with “high-risk stigmata” be treated by surgical resection [3,4,5,6,7]. Patients who have “worrisome features” and additional high-risk signs should also undergo resection, while patients with “worrisome features” alone should receive careful follow-up. The “high-risk stigmata” of IPMNs that can be detected by endoscopic ultrasonography (EUS) are an enhancing solid component inside the cyst and dilation of the main pancreatic duct (>10 mm) [3,8,9,10,11]. Additional high-risk signs among the “worrisome features” are a definite mural nodule and suspected main duct involvement [12,13]. Thus, the current guidelines focus on the morphological features of IPMN as revealed by EUS.

On the other hand, although not described in previous reports, clinical experience suggests that evidence of chronic pancreatitis (CP) is frequently identified in the background pancreatic parenchyma of IPMN patients by EUS (EUS-CP findings). CP is a well-known risk factor for pancreatic malignancy, including pancreatic ductal adenocarcinoma [14,15,16]. However, the relationship between malignant transformation of IPMNs and pathological changes of the pancreatic parenchyma, such as atrophy, inflammation, and fibrosis, remains unclear. Moreover, changes of the background parenchyma are not considered to be a high-risk sign for malignant IPMN in the current guideline. We hypothesized that EUS-CP findings might be associated with malignant IPMN.

To test this hypothesis, we utilized a database of 69 consecutive patients with IPMNs who underwent preoperative EUS and surgical resection at Kobe University Hospital. We examined the association of EUS-CP findings with the incidence of malignant IPMN and with pathological changes of the background parenchyma in the resected specimens (atrophy, inflammation, and fibrosis).

The clinical data of 69 consecutive patients with IPMNs in whom EUS was performed before surgical resection at Kobe University Hospital from April 2010 to October 2014 were collected prospectively. Patients with concomitant pancreatic carcinoma (defined by the presence of two discrete tumors) were excluded. We investigated various clinical and tumor characteristics, including age, sex, abdominal symptoms (presence vs. absence), serum CEA (<5 vs. ≥5 ng/mL), serum CA19-9 (<37 vs. ≥37 U/mL), diabetes mellitus (presence vs. absence), habitual drinking (none/sometimes/daily), alcohol consumption (<50 vs. ≥50 g/day), macroscopic type of IPMN (main duct, mixed, or branch duct type), cyst location (head vs. body/tail), main pancreatic duct diameter (<5, 5-10, >10 mm), mural nodule in the cyst (presence vs. absence), cyst diameter (<30, 30-40, >40 mm), and tumor mucin phenotype (intestinal, gastric, pancreatobiliary, or oncocytic). According to the 2012 international consensus guidelines [3], the macroscopic type of IPMN was classified as main duct type (MD-IPMN), mixed type (Mix-IPMN), or branch duct type (BD-IPMN) based on the results of imaging studies, mainly magnetic resonance cholangiopancreatography (MRCP). Tumors were histologically classified into four types (gastric, intestinal, pancreatobiliary, or oncocytic) according to the appearance on hematoxylin and eosin-stained sections and immunohistochemistry with antimucin antibodies (MUC1/MUC2/MUC5AC/MUC6). In patients with multiple morphologically distinct IPMNs, each neoplasm was classified depending on the type of the dominant component. Tumors were also classified into low-grade dysplasia, high-grade dysplasia, or invasive intraductal papillary mucinous carcinoma (IPMC). The diameter of the main pancreatic duct and the cyst size were measured by MRCP in most cases. In patients who did not undergo MRCP, measurement was done by an alternative method, such as endoscopic retrograde cholangiopancreatography or computed tomography. The presence of mural nodules was evaluated by EUS and magnetic resonance imaging. Abdominal symptoms were defined as including abdominal or back pain/discomfort, weight loss, appetite loss, and jaundice.

This study was conducted in accordance with the Declaration of Helsinki and its amendments (UMIN-CTR ID: 000021345). The study protocol was approved by the ethics committee of Kobe University School of Medicine (No. 1864). All authors had access to the study data and reviewed and approved the final manuscript.

Three echo-endoscopes (GF-UCT240, GF-UCT260, or GF-UE260; Olympus, Tokyo, Japan) were used in this study. EUS-CP findings refer to the parenchymal features of CP in the Rosemont classification recently proposed as EUS diagnostic criteria for this condition [17]. These findings included “hyperechoic foci with shadowing,” “lobularity with honeycombing,” “lobularity without honeycombing,” “hyperechoic foci without shadowing,” and “stranding” (Fig. 1). “Cysts” is also a parenchymal feature of CP in the Rosemont classification, but it was excluded because IPMN itself affects cyst formation. For the same reason, ductal features of CP in the Rosemont classification were excluded, including “main pancreatic duct calculi,” “irregular main pancreatic duct contour,” “dilated side branches,” “main pancreatic duct dilation,” and “hyperechoic main pancreatic duct margin.”

EUS-CP findings were reviewed prospectively by an endosonographer (M. Takenaka) who was unaware of the other data. All cases were re-examined by a second endosonographer (H. Shiomi), who was also unaware of the other data, and there was a good correlation of EUS-CP findings between the two endosonographers (κ = 0.80 for score 0 vs. ≥1, p < 0.001). Both endosonographers had experience with more than 1,000 EUS examinations.

Indications for surgical resection of IPMN at our hospital were decided according to the 2012 International Consensus Guidelines. According to the guidelines, resection is recommended for MD-IPMN. In the case of Mix-IPMN and BD-IPMN, resection is recommended if the patient has “high-risk stigmata.” In patients with “worrisome features” and additional high-risk signs, resection is also recommended.

Hematoxylin and eosin-stained tissue sections from all resected specimens of IPMN were reviewed by a pathologist (Y. Zen), who was blinded to all clinical features. Pancreatic parenchymal atrophy, infiltration of inflammatory cells, and fibrosis were evaluated in the background pancreatic parenchyma. The severity of parenchymal atrophy, inflammation, or fibrosis was graded as 0 (none), 1+ (mild), 2+ (moderate), or 3+ (severe). Representative examples are shown in Figure 2.

All statistical analyses were conducted using JMP version 11 (SAS Institute, Cary, NC, USA), and all p values were two-sided. The χ² test (or Fisher exact test, if appropriate) was used to assess associations between categorical variables. To assess the association of the number of EUS-CP findings (0 vs. ≥1) with the incidence of invasive IPMN, we conducted multivariate logistic regression analysis. The binary categorical variable (presence and absence) of the incidence of invasive IPMN was used as an outcome variable. Multivariate binary logistic regression analysis was performed to adjust for potential confounders. The OR was adjusted for age, alcohol consumption, and macroscopic type of IPMN, the p value of which was <0.2 in univariate analysis. Kappa coefficients were calculated to assess agreement of EUS-CP findings between the two observers. In all analyses, p < 0.05 was considered statistically significant.

We categorized 69 IPMN patients by clinical features and factors associated with EUS-CP findings. There were 29 patients without EUS-CP findings and 40 patients with at least one EUS-CP finding. The clinical and morphological features of IPMN are summarized with respect to EUS-CP findings in Table 1. Patients with EUS-CP findings were older than those without EUS-CP findings (70.2 ± 1.4 vs. 65.0 ± 1.7 years, p = 0.02). Among the patients with EUS-CP findings, invasive IPMC was significantly more frequent than among those without EUS-CP findings (42.5% [17/40] vs. 3.4% [1/29], p = 0.0002) (Table 2). Multivariate analysis showed that detection of EUS-CP findings in the background pancreatic parenchyma was associated with a higher prevalence of invasive IPMC (multivariate OR = 21.9, 95% CI = 3.8-423.1, p = 0.0001). Assessment of each EUS-CP finding showed that invasive IPMC was significantly more frequent in the patients with “stranding” than in those without (55% [11/20] vs. 14% [7/49], p = 0.002). In addition, invasive IPMC was significantly more frequent in the patients who had “lobularity without honeycombing” than in those without (83% [5/6] vs. 21% [13/63], p = 0.004) (Table 3).

Table 4 shows the association between the total number of EUS-CP findings and the pathological findings (atrophy/inflammation/fibrosis) in the background pancreatic parenchyma of the IPMN patients. Among patients with EUS-CP findings, higher grades of pancreatic atrophy and inflammation were observed than in those without EUS-CP findings (atrophy: 72.5% [29/40] vs. 34.5% [10/29], p = 0.003; inflammation: 45.0% [18/40] vs. 20.7% [6/29], p = 0.04). Assessment of the pathological correlates with each EUS-CP finding revealed higher grades of pancreatic atrophy, inflammation, and fibrosis in patients with “stranding” than in those without (atrophy: 90%[18/20] vs. 43% [21/49], p = 0.0004; inflammation: 56% [11/20] vs. 27% [13/49], p = 0.03; fibrosis: 75% [15/20] vs. 35% [17/49], p = 0.003) (Table 5).

Figure 3 displays the relationship between pathological changes (atrophy, inflammation, and fibrosis) in the background pancreatic parenchyma and invasive IPMC. Higher grades of atrophy and inflammation were significantly associated with the presence of invasive IPMC (atrophy: p = 0.002; inflammation: p = 0.002). Higher grades of fibrosis were also associated with invasive IPMC, but the relationship was not significant (p = 0.057).

In this study, we found that the frequency of invasive IPMC was significantly higher in patients with EUS-CP findings than in those without. In other words, IPMN patients with “normal background pancreatic parenchyma” on EUS had a significantly lower rate of malignancy. To the best of our knowledge, this is the first study to examine the association between EUS-CP findings and malignant transformation of IPMNs.

The Rosemont classification has been proposed as EUS criteria for diagnosing CP [17]. The EUS-CP findings used in this study corresponded to the parenchymal features of the Rosemont classification. However, the relationship between EUS findings and pathological findings in patients with CP has not been well characterized, even in the Rosemont classification. One of the reasons for this is that tissue specimens for evaluating the pancreatic parenchyma often cannot be obtained in patients with CP because they rarely undergo surgery. In this study, we were able to examine the relationship between EUS-CP findings and pathological changes of the background pancreatic parenchyma because we used the resected specimens from IPMN patients.

In the Rosemont classification, each EUS finding has a predictive value, and the findings are categorized into major or minor criteria. The major criteria of CP are “hyperechoic foci with shadowing” and “lobularity with honeycombing.” On the other hand, “stranding” and “lobularity without honeycombing” are classified as minor criteria because of poor specificity. In the present study, “stranding” and “lobularity without honeycombing” were frequently observed in the background parenchyma of IPMN patients, which was a discrepancy between our results and the Rosemont classification. This difference might have arisen because the causes of CP were different. The Rosemont classification was mainly based on alcoholic CP. On the other hand, our study did not target alcoholic CP, but instead assessed CP caused by mucin stagnation in IPMN patients. It has been reported that hypersecretion of mucin by IPMNs causes obstruction of the main pancreatic duct, which may induce low-grade pancreatitis [2]. This difference in the pathogenesis of the parenchymal changes might be associated with the difference of major EUS findings.

Alcohol consumption is strongly associated with CP [18,19,20]. In our study, there were no significant differences in alcohol consumption between the groups with and without EUS-CP findings, but the presence of these findings was associated with pathological changes of the background parenchyma, such as atrophy and inflammation. This indicates that the EUS-CP findings we investigated were not associated with alcohol consumption and that an increase in mucin production with malignant transformation of IPMN might be associated with changes in the background parenchyma.

The present study has some limitations. First, it was difficult to demonstrate the association of each EUS-CP finding with specific pathological features. Even in the Rosemont classification, only “hyperechoic foci with shadowing” and “cysts” (excluded from this study) have histological correlates such as calcification and pseudocyst, while other factors (including “stranding” and “lobularity with honeycombing”) are not associated with specific histological features. Further studies will be required to clarify the pathological correlates of these findings. Next, this was a retrospective single-center study, and the number of subjects was limited. Thus, larger studies will be necessary to confirm our results.

In conclusion, detection of EUS-CP findings in the pancreatic parenchyma of IPMN patients was associated with invasive IPMC. When EUS examination of IPMN is performed, the background parenchyma should be evaluated in addition to the morphological features of the tumor itself.

We would like to thank the staff of the Center of Clinical Research (Dr. Omori) at Kobe University Hospital for their valuable contributions (advice about statistical analysis). The authors assume full responsibility for the analysis and interpretation of the data in this report.

This work was supported by grants from JSPS KAKENHI (Grants-in-Aid for Scientific Research; grant No. 15624848 to A. Masuda and grant No. 15612795 to H. Kutsumi). This work was also supported by the Research Committee of Intractable Pancreatic Diseases from the Ministry of Health, Labor, and Welfare of Japan (A. Masuda, H. Shiomi, and H. Kutsumi).

The authors have no conflicts of interest to declare.

M. Takenaka, A. Masuda, and H. Shiomi contributed equally to this work.