Gallbladder papillary neoplasms share pathological features with intraductal papillary neoplasm of the bile duct

Intraductal papillary neoplasm of the bile duct (IPNB) has been widely recognized. However, the knowledge of intracystic papillary neoplasm of the gallbladder (IPNG) including papillary adenoma and adenocarcinoma is not well defined. In this study, we compared the clinicopathological and immunohistochemical features between 32 IPNG cases and 32 IPNB cases. IPNG-1 (low-high grade dysplasia) exhibited an earlier onset age, smaller tumor size and lower level of CK20 expression compared to IPNG-2 (invasive carcinoma). Histologically, pancreaticobiliary and intestinal subtype accounted for nearly half of IPNG or IPNB (44.4% and 48.1% vs. 44.0% and 44.0%), respectively. Immunohistochemically, 88.9% of IPNG and 92.0% of IPNB cases were positive for MUC1, and 96.3% and 92.0% for CK7, respectively. CDX2 and MUC2 were more highly expressed in the intestinal subtype than in other subtypes. CK20 expression increased in parallel with tumor progression. In addition, 53.1% of IPNG cases and 68.6% of IPNB cases exhibited invasive carcinoma, and showed significant survival advantages to conventional gallbladder adenocarcinoma and cholangiocarcinoma, respectively. In conclusion, papillary adenoma and adenocarcinoma of the gallbladder can be recognized as different pathological stages of IPNG, and they share pathological features with IPNB.


INTRODUCTION
Intraductal papillary neoplasm of the bile duct (IPNB) is a class of tumor that is characterized by predominant intraductal papillary growth that may occur anywhere along the biliary tree [1][2][3][4][5][6][7][8]. Its' definition does not include tumors originating from the gallbladder. However, some gallbladder neoplasms exhibit a papillary growth pattern that projects into the lumen of the gallbladder, including papillary adenocarcinoma and adenoma. Adsay et al. examined mass-forming (≥ 1 cm) pre-invasive neoplasms of the gallbladder and found that these tumors exhibited a spectrum of dysplastic changes, variable configuration, and different levels of MUCs, CKs, and CDX2 expression in different cell lineages, which resembled the presentation of IPNB [9]. However, the authors included tubular neoplasms of the gallbladder in their research, and they did not conduct comparative analyses with IPNB. Therefore, we selected gallbladder papillary adenocarcinoma, adenoma and IPNB cases from the same population and compared their clinicopathological features, histological subtypes and expression profile of MUCs, CKs, and CDX2 to investigate whether papillary adenoma and papillary adenocarcinoma of the gallbladder can be recognized as different pathological stages of the same disease, which we named intracystic papillary neoplasm of the gallbladder (IPNG), and whether papillary Research Paper www.impactjournals.com/oncotarget tumors from the gallbladder and bile duct exhibited similar pathological features.

RESULTS
Thirteen of the 27 cases of IPNG were identified as intestinal, 12 cases as pancreaticobiliary, and 2 cases as oncocytic subtype. The numbers of intestinal, pancreaticobiliary, and oncocytic subtypes in the 25 cases of IPNB were 11, 11, and 3, respectively ( Table 1). None of the IPNG or IPNB cases in the present cohort were classified as gastric subtype. However, three cases of IPNG exhibited a minor component of gastric cells (Figure 1). A total of 59.3% of IPNG (16 cases) and 48% (12 cases) of IPNB contained two or more histological subtypes of tumor cells. MUC1 was expressed primarily in the apical membrane and occasionally in the cytoplasm of tumor cells. MUC2, CK7, and CK20 were present primarily in the cytoplasm, and CDX2 was located in the nucleus ( Figure 2).

Clinicopathological features
IPNG and IPNB showed significantly different clinical manifestations. However, they shared similarities in pathological features. A total of 28.1% of IPNG and 18.8% of IPNB lesions were multiple (P = 0.376). The proportion of cases with invasive carcinoma was similar in IPNG and IPNB (53.1% vs. 68.8%, P = 0.200). Only 5 cases of IPNG and 4 cases of IPNB exhibited lymph node involvement (P = 1.000) ( Table 1).

Histological subtype and immunohistochemical staining
Most cases of pancreaticobiliary subtype in IPNG were negative for MUC2 (11/12), and 9 of 13 cases of intestinal subtype were positive (P = 0.005, Figure 4A). CDX2 was not expressed in half of the pancreaticobiliary subtype (6/12), and it was diffusely expressed in 8 of 13 cases of the intestinal subtype (P = 0.007, Figure  4B). Among the immunohistochemical markers, the expression of CDX2 was positively correlated with MUC2 and CK20, with coefficients of 0.391 (P = 0.044, Figure 4C) and 0.514 (P = 0.006, Figure 3E), respectively.
Multivariate analysis suggested that positive surgical margin and lymph node involvement were independent risk factors of IPNG. The median overall survival of patients with negative surgical margin was 117.0 months, which was much longer than that of positive cases (9.0 months, P = 0.023, Figure 5D), with a hazard ratio of 8.7 (95% CI, 1.3-56.7). The median overall survival of cases without lymph node involvement was also much longer than that of cases with lymph node involvement (117.0 vs. 26.0 months, P = 0.021, Figure 5E), with a hazard ratio of 7.6 (95% CI, 1.4-42.1). However, the independent risk factor of IPNB was invasive carcinoma (P = 0.040, Figure 5F). The median overall survival of IPNB-1 was 103.0 months (95% CI, 60.9-145.1 months), which was much longer than that of IPNB-2 (44.0 months), with a hazard ratio of 2.0 (95% CI, 1.0-3.8).

DISCUSSION
IPNB is widely recognized as a pathological entity that includes papillary cholangiocarcinoma and precursor lesions. Our findings demonstrated that IPNB with or without invasive carcinoma exhibited nearly identical clinical features, and gradual changes in histological subtypes and immunohistochemical expression files during tumor progression. For example, the main histological subtype of IPNB-1 was pancreaticobiliary (77.8%), and most IPNB-2 cases were classified as intestinal subtype (62.5%). Furthermore, CDX2, CK20 and MUC2, which are known as markers of intestinal metaplasia [4,[10][11][12], were expressed more frequently in IPNB-2 than in IPNB-1.
It is believed that adenomas of the gallbladder play a minor role in the pathway of gallbladder carcinogenesis, but some studies suggest that papillary adenoma is involved in the "adenoma-adenocarcinoma" process [13][14][15]. Gallbladder adenocarcinoma originating from adenoma is relatively uncommon may partly due to papillary adenoma accounts only a small portion of gallbladder adenoma [16,17]. Several previous studies demonstrated that gallbladder papillary neoplasms appeared macroscopically as cauliflower-like masses projecting into the lumen of the gallbladder and microscopically as papillary proliferation of epithelial cells with delicate fibrovascular stalks, and the prognosis of gallbladder papillary adenocarcinoma was better than that of gallbladder adenocarcinoma (NOS), which was extremely similar to the presentation of IPNB [18][19][20].
Therefore, we performed comparative analyses between IPNG-1 and IPNG-2 to investigate the relationship between gallbladder papillary adenoma and adenocarcinoma. Clinicopathologically, the mean tumor size of IPNG-2 was 4.0 ± 2.2 cm, which was much larger than that of IPNG-1 (1.5 ± 1.2 cm, P < 0.001). The average onset age of IPNG-1 was 56.5 ± 16.9 years, which was younger than that of IPNG-2. These findings suggested that IPNG-1 was an early stage of IPNG. The IPNG-1 lesions grew gradually over time, gained invasiveness, and progressed to IPNG-2. Moreover, the distribution of histological subtype was similar in IPNG-1 and IPNG-2 and in all of the immunohistochemical markers except CK20, which was expressed more frequently in IPNG-2 than in IPNG-1. These findings were consistent with previous studies focused on IPNB, which demonstrated that CK20 was expressed more frequently in intraductal papillary neoplasm of the liver than nonneoplastic bile ducts and non-papillary intrahepatic cholangiocellular carcinoma, and its incidence increased significantly in parallel with lesion progression [21,22]. The younger onset age and smaller tumor size of IPNG-1 and the increased expression of CK20 during tumor progression suggest that gallbladder papillary adenoma and adenocarcinoma, which were separated in clinical practice, can be different stages of a single pathological entity, which we propose to name intracystic papillary neoplasm of the gallbladder.
Thereafter, we compared IPNG to IPNB and found them with great similarities. Firstly, pancreaticobiliary and intestinal subtype accounted for nearly half of IPNG or IPNB (44.4% and 48.1% vs. 44.0% and 44.0%), respectively. Secondly, the distribution of immunohistochemical expression of MUCs, CKs, and CDX2 were similar, and MUC1 and CK7 expression were prevalent. Thirdly, CDX2 and MUC2 were more highly expressed in the intestinal subtype than in other subtypes. Fourthly, CDX2 expression was positively correlated with MUC2 and CK20 expression. Fifthly, 53.1% of IPNG cases and 68.6% of IPNB cases exhibited invasive carcinoma, and showed similar significant survival advantages to conventional gallbladder adenocarcinoma and cholangiocarcinoma, respectively. Sixthly, CK20 expression increased in parallel with tumor progression.
On one hand, 88.9% of IPNG and 92.0% of IPNB cases were positive for MUC1, and 96.3% and 92.0% for CK7, respectively. MUC1, a marker of pancreaticobiliary differentiation and reported to be associated with the pancreaticobiliary subtype of IPNB in previous studies [1,3], was prevalent in almost all cases in our study. CK7, as a marker of biliary epithelium [23], was also expressed in almost every case of IPNG and IPNB. Therefore, we speculated that IPNG and IPNB likely retained pancreaticobiliary phenotype (MUC1/CK7) during tumorigenesis. On the other hand, CDX2 and MUC2 expression levels in IPNG or IPNB were associated with the intestinal subtype, and CDX2 was positively correlated with MUC2 and CK20 expression, which is consistent with previous studies [1,2,4,11,12,22] and suggests that intestinal metaplasia plays a role in the tumorigenesis of IPNG and IPNB. The immunohistochemical markers CDX2 and CK20 in IPNB and CK20 in IPNG were associated with invasive carcinoma, which suggests that intestinal metaplasia participated in tumorigenesis, and some internal molecular changes of this pathological process also play a role in tumor progression. The specific intrinsic mechanisms may be heterogeneous in tumors originating from different locations.  that of cholangiocarcinoma and gallbladder adenocarcinoma, respectively. There was no significant difference in overall survival between IPNB and IPNG (C). Multivariate analysis suggested that positive surgical margin (D) and lymph node involvement (E) were independent risk factors of IPNG. However, the independent risk factor of IPNB was invasive carcinoma (F).

Case selection and clinicopathological data
Hospital were retrieved. A total of 32 cases of gallbladder tumors that were papillary or contained papillary ingredients (10 adenoma and 22 adenocarcinoma in accordance with previous diagnostic criteria) and 32 cases of IPNB (7 intrahepatic, 8 hilar, and 17 extrahepatic cases) were included. We also included 97 cases of gallbladder adenocarcinoma (NOS) and 303 cases of cholangiocarcinoma (NOS, 92 intrahepatic, 79 hilar, and 132 common bile duct cases) from the same period for comparative survival analyses. Diagnoses were reconfirmed, and 15 of 32 IPNG cases were identified as low-high-grade dysplasia (IPNG-1). The other 17 cases were diagnosed with invasive carcinoma (IPNG-2). Ten of the 32 IPNB cases were low-high-grade dysplasia (IPNB-1), and the other 22 cases were invasive carcinoma (IPNB-2). Clinicopathological data of all patients were obtained from medical records. All participants provided written informed consent, and the Peking Union Medical College Hospital Ethics Committee approved all study procedures.

Statistical analysis
Comparisons of continuous variables were performed using Student's t test. Comparisons of categorical variables were performed using the Pearson chi-square test and Fisher's exact test. Spearman's rank correlation test was used to evaluate the correlation between categorical variables. Kaplan-Meier curves and the log-rank tests were used to compare overall survival. Cox regression analysis was performed to identify prognostic factors in each group. Statistical significance was set at P < 0.05.

CONCLUSIONS
Papillary adenoma and adenocarcinoma of the gallbladder can be recognized as tumors at different stages of a single pathological entity, namely, IPNG. The distribution of histological subtypes and immunohistochemical phenotype of IPNG resembled that of IPNB. During tumorigenesis, IPNG and IPNB likely retained pancreaticobiliary phenotype (MUC1/CK7), and nearly half of the cases obtained intestinal phenotype (CDX2/MUC2/CK20), and intrinsic molecular changes in the process of intestinal metaplasia such as CK20 expression may contribute to the progression of these tumors.

Authors᾿ contributions
Xueshuai Wan, Jie Shi, and Anqiang Wang contributed equally to performing experiments and writing the article; Yuan Xie, Xiaobo Yang, Chengpei Zhu, Haohai Zhang, Liangcai Wu, Shanshan Wang, Hanchun Huang, Jianzhen Lin, and Yongchang Zheng collected data and revised the manuscript; Zhiyong Liang, Xinting Sang, and Haitao Zhao contributed equally to the concept and design of this study.