The results of our cohort study suggest that the majority (73.2%) of patients with PC had CP and 26.8% had other pancreatic diseases. This observation is consistent with that of the previous study by Campisi et al.. [7] But the disease spectrum was more diverse in our study than that reported by Campisi et al. In the CP spectrum, there were three kind of CP, including benign CP, maligant CP and CP combined with IPMN in our study. The non-CP diseases include not only the P-NETs, the IPMN, the SCN, but also the SPT and the MCN. In addition to that, the rate of malignant CP was high (14.9% in our series vs. 3.9% in Campisi’s study) (Fig. 1, Table 1).
One of the key features of CP is the presence of pancreatic intraductal stones and diffuse distributed parenchymal calcifications. Together, these features have high sensitivity (72.7%) and specificity (100%) for diagnosing CP. (Table S2, Table 2). All our findings are similar to the revised Japanese clinical diagnostic criteria for CP. [4] But our series also showed that there is a small possibility of presence of P-NETs or malignant IPMN which can mimic CP (Table S2).
The other key points for the differentiation of CP and non-CP were the presence of typical features of the non-CP diseases. (Table S2, Table 2). The curvilinear or peripheral eggshell-like calcifications were the main feature of SPT. Point or block calcification(s) at the edge of the lesion was found in malignant IPMN (80%), P-NETs (60%), MCN (66.7%) and SPT (20.0%). Stellate central calcification was the main finding of SCN (Table 2, Table S2). These findings of SPT were slightly different from that of the previous studies, where the calcifications of SPT were characteristically peripheral and punctate.[21] The calcifications of P-NETs were usually focal, coarse, irregular, and centrally located, [22, 23] calcifications of MCN were peripheral eggshell or focal punctate calcifications, [24] calcifications of IPMN were punctate calcifications (87%) and coarse calcification (33%). [25, 26] In our series, one of the important identification points was that calcifications of pseudocyst are located in the cystic wall, whereas in the non-CP diseases, the cystic pancreatic lesions usually have only segmental calcifications.
The second important purpose of this study was to determine whether PC can be used to suggest a diagnosis of malignancy. On a whole, the presence of diffuse distribution of parenchymal calcifications alone and intraductal stones alone ruled out maliganacy with high specificity (95.1% and 94.3% respectively) (Table S3). But in our study group, the majority of the patients had CP and IPMN with only one case of mixed ductal-endocrine carcinoma of the pancreas and mucinous cystadenocarcinoma each. Also, the difference in the incidence of PC was significantly lower in benign IPMN than malignant IPMN (p = 0.002) in our study. Previous studies have also reported that the presence of coarse calcifications is associated with high risk for malignancy. [27] Not much is known about PC and malignancy in other non-CP diseases, and some of our results were somewhat contradictory. In MCN, multilocular macrocystic lesions and lesions with calcifications in the wall had a higher risk of malignancy. [28] But in our study we could not find any difference between the benign and malignant group (2.5% vs. 3.6%, p = 0.560), which similar to that reported by Wu et al. [29]. In a previous study, 7 of 10 P-NETs with calcifications had malignant features, unlike the observations of this study.[6] However, in SPT, PC was not useful to differentiate the benign and malignant forms of SPT as found in this study and previous studies (Table 3). [6, 10, 11]
There are few studies showing that PC is associated with cancer. [2, 3] In a study (n = 48) reported by Mohamed et al., [8] the presence of a pancreatic mass in CP is suggestive of malignancy, especially when PC are observed. Our results in aggreement with these that PC was found to a risk factor with the highest odds ratio. We also found that presence of PC, in combination with other variables, such as advanced age (> 55 years), high BMI (> 24 kg/m2), atrophy of the parenchyma and pancreatic mass, were shown to be independent predictors of PDAC but with the lowest odds ratio (Table 4). This may be due to the patients were all received surgical treatment or the number of patients were relatively large in our study. And, on the other hand, the PC is an important warning sign for malignancy who had surgical indication, even the patients had no pancreatic mass.
Common bile duct dilation or bile duct stenting was a protective factor (OR = 0.2, p = 0.032). This was contrary to the observations made by previous study. [8] But a similar study by Ruan et al. [30] showed that dilated bile ducts passed through the lesions in 79.17% of massforming CP and in 16.67% cases of pancreatic carcinoma (p < 0.001). The double duct sign, which indicates the expansion of both the pancreatic duct and the bile duct, was similar in both groups. However, the number of cases were small (massforming CP, n = 24; pancreatic carcinoma, n = 30). The rate of calcification in massforming CP was significantly more than in pancreatic carcinoma cases (58.3% vs. 10%, p < 0.001). In our study, the incidence of calcification in CP with PDAC was more than CP without PDAC cases (75.0% vs. 22.6%, p = 0.000 (Table S4).
About 5–10% of "mass-forming" pancreatitis often masquerades as PDAC. due to overlap of the imaging findings. [31, 32] In this study, we found that suspicious lymph nodes, abutment or encasement of the arteries around the pancreas, venous involvement, and the tumor size have no abilty to differentiate between CP and CP with PDAC. Many other clinical features such as the duration of symptoms before admission, jaundice, high-colored urine, diabetes mellitus, total bilirubin levels, and the carbohydrate antigen 19 − 9 (CA19-9), the carbohydrate antigen 125 (CA125) were insignificant using logistic regression, which are in agreement with these in preivout studies. [20, 33–36] However, the appearance of these clinical features should raise the suspicion of PDAC and may complement other clinical findings to improve diagnostic accuracy. [34]
Our study has some limitations. First, the sample size was relatively small. Second, only the patients undergoing surgery were included. This method guaranteed the accuracy of the diagnosis, but many patients with PC were not included in our study who received medical or endoscopic intervention therapy. Third, because the focus of this study was PC, findings of imaging modalities other than CT were not studied. Also, the morphological features of bile ducts and pancreatic ducts, including the double duct sign and whether they passed through the lesion or interrupted the lesion areas, were not further evaluated. Fourth, this study was retrospective in nature. The technical parameters adopted in different periods were slightly different. The scan thickness of 5mm in the early-phase of the study could have increase the risk of overlooking PC. Future larger prospective studies are required to validate the findings of this study. Finally, we were not sure whether the CT scan parameters of other hospitals were consistent with those of our hospital. However, during the pre-operative discussion at our pancreatic center, the imaging data of other hospitals was reviewed and no inconsistency with regards to the presence of PC was found.
In summary, the incidences of PC in the pancreatic diseases were lower than these in previous studies, but the disease spectrum, including three kind of CP and five kind of non-CP diseases, was more diverse than these in privious studies. On the other hand, PC may be related to malignant IPMN. And, PC, in combination with other variables, were shown to be related to PDAC in the setting of CP, not only in cases with pancreatic mass, but also in cases without panceatic mass.