Adenocarcinoma of gallbladder: an immunohistochemical profile and comparison with cholangiocarcinoma
- Correspondence to Professor Prithi S Bhathal, Department of Pathology, The University of Melbourne, Melbourne, VIC 3010, Australia;
- Received 14 August 2012
- Revised 5 November 2012
- Accepted 9 November 2012
- Published Online First 25 December 2012
Aim Cholangiocarcinomas display intestinal and pyloric gland metaplasia-cell phenotypes. Those that arise in chronically inflamed (fluke infested) bile ducts more frequently express the intestinal metaplasia-cell phenotype and p53 than sporadic cholangiocarcinomas. We wished to determine if adenocarcinomas of the gallbladder display a similar profile.
Methods Adenocarcinoma, adenoma, and dysplastic and metaplastic epithelia were studied in 55 gallbladders. Serial paraffin sections were stained for five foregut antigens characteristically present in pyloric gland metaplasia, three intestinal-specific antigens and p53. Antigen expression was compared with that shown by 65 fluke-associated and 47 sporadic cholangiocarcinomas.
Results Pyloric gland metaplasia in gallbladders with chronic cholecystitis invariably displayed the five foregut antigens. The frequency of expression of these five antigens by the gallbladder cancers and cholangiocarcinomas did not differ significantly. An intestinal goblet-cell marker and p53 were more frequently expressed by gallbladder carcinoma (59% and 45%, respectively) and fluke-associated cholangiocarcinoma (45% and 46%) than by sporadic cholangiocarcinoma (17% and 23%). K20 was more frequently expressed by gallbladder carcinoma (52%) than either fluke-associated (21%) or sporadic (17%) cholangiocarcinoma. Dysplastic epithelium and adenomas also displayed the pyloric gland and intestinal metaplasia-cell phenotypes. Cells staining for pyloric gland metaplasia-cell phenotypes were distinct from the intestinal metaplasia-cell phenotypes when present together in a gallbladder carcinoma, cholangiocarcinoma, dysplastic epithelium or adenoma.
Conclusions Adenocarcinomas of gallbladder generally arise from a foregut cell lineage via a metaplasia-dysplasia-carcinoma sequence. A background of chronic inflammation increases the frequency of expression of an intestinal goblet-cell phenotype and p53 in the cancers.
Chronic cholecystitis due to gallstones is the major risk factor for adenocarcinoma of the gallbladder in Western countries. An adenoma-carcinoma sequence has been demonstrated, but the incidence of adenoma is too low to account for all gallbladder cancers.1 ,2 Dysplasia developing in metaplastic epithelium is regarded as a source of adenocarcinoma by some workers,3–5 while focal epithelial hyperplasia progressing to atypical hyperplasia and carcinoma in situ has been proposed as a frequent pathway in Mexican6 and American Indian7 populations. Although both pyloric gland metaplasia and intestinal metaplasia are found in chronically inflamed gallbladders, intestinal metaplasia is regarded as the more likely indicator of progression to cancer.4 ,8 ,9
We have reported that while both fluke-associated and sporadic cholangiocarcinoma similarly display pyloric gland metaplasia-cell phenotypes, fluke-associated cholangiocarcinoma more frequently express the intestinal metaplasia-cell phenotype and p53 than sporadic cholangiocarcinoma.10 Here, we compare the antigen profile of gallbladder adenocarcinoma with that of cholangiocarcinoma.
Materials and methods
Patients and tissue samples
In all, 46 gallbladders from patients admitted to The Royal Melbourne Hospital, including 28 with 29 adenocarcinomas, six with adenoma, six dysplasia and 16 with chronic cholecystitis were studied. Gallbladders from four livers unsuitable for transplantation and five resected with hepatic metastases or haemangiomas of liver were used as controls. Serial sections taken from paraffin blocks of tissue used for surgical pathology reports and additional strips of mucosa taken after completion of the pathologists’ reports were studied. Tissue was collected as approved by the Ethics Committee of the Royal Melbourne Hospital.
Two of the 29 gallbladder carcinomas were papillary, three mucinous and three carcinoma in situ, with three described as well, 10 moderately and 12 poorly differentiated adenocarcinomas. Patients included eight men aged 39–89 years (median 66 years) and 20 women aged 40–86 years (median 70 years). The two synchronous cancers were 1 and 2 mm in diameter, 2 cm apart in the body of a gallbladder with metastatic carcinoma in the cystic duct lymph node. Gallbladders with chronic cholecystitis were from four men and 12 women aged 44–81 years (median 65 years). The six gallbladders with adenoma, and the six with dysplasia were from two men and 10 women aged 39–88 years (median 76 years). Adenomas consisted of three tubular, two tubulovillous and one papillary adenoma. Gallbladders from organ donors and patients were from six men and three women aged 7–76 years (median 48 years).
The details of antigen expression by 47 sporadic cholangiocarcinoma from Australia and 65 fluke-associated cholangiocarcinoma from Thailand given in table 2 in this article are from table 2 of our previous publication.10
The five foregut antigens studied included D10, 1F6, MUC6, MUC5AC and TFF2. D10 is a high molecular weight mucin that, along with MUC6 and TFF2, is invariably expressed in gastric mucous neck cells, antral glands and, together with 1F6, in Brunner glands, peribiliary and periductal glands of the pancreas and pyloric gland metaplasia.11 ,12 The epitope recognised by the antibody to antigen D10 is highly conserved and present in the homologous tissues of all classes of terrestrial vertebrates tested except birds.13 MUC5AC is found in gastric foveolar epithelium and the surface epithelium overlying the acini of pyloric gland metaplasia.11
Antigen 17NM is restricted to the mucous vacuole of human intestinal goblet cells, goblet cells of intestinal metaplasia and some gastrointestinal tumours.14 Expression of cytokeratin K20, CDX2 and p53 were also studied.
Single or double immunohistochemical staining of sections was carried out as previously described.11 Antibodies to the apoproteins of MUC6 (clone CLH5) and MUC5AC (clone CLH2), and p53 (clone DO-7) were from Novocastra Laboratories (Newcastle-upon-Tyne, UK), K20 (clone Ks20.8) from DakoCytomation Pty. (Botany, Australia), and CDX2 (clone CDX2-88) from Biogenex Laboratories Inc. (San Ramon, California, USA). A polyclonal rabbit antibody to TFF215 was a gift from Professor Andrew Giraud. Sections were counterstained with haematoxylin and 1% Alcian blue for acid mucins.16
Sections of normal tissues were used as positive controls, and 1% bovine serum albumin/buffer solution as a negative antibody control.
To compare the extent of expression of different antigens in tissues, their expression was graded as: −, no staining; 1+, less than 10% of cells or glands stained; 2+, 10%–50% of cells or glands stained; 3+, majority of cells or glands stained.
The significance of differences in frequency of antigen expression (positive or negative) in different cancer groups was determined using Fisher's exact test; a cancer was regarded as positive for antigen provided at least occasional cells (1+) stained.
Normal gallbladder epithelium
There was no epithelial staining for D10, 1F6, 17NM or K20, but focal or more widespread expression of MUC6, MUC5AC and TFF2 was found in the surface epithelium of the majority of the nine normal gallbladders; acid mucin secretion was invariably present (table 1). Metaplasia and staining for CDX2 were absent.
Gallbladders with chronic cholecystitis
Foci of pyloric gland metaplasia ranged from occasional to numerous in the different gallbladders; the acinar cells of the pyloric gland metaplasia invariably stained for D10, MUC6 and TFF2, and more variably for 1F6 (table 1, figure 1A–C). MUC5AC was focally, or generally, present throughout the surface epithelium of all the gallbladders.
Intestinal metaplasia was present in the surface epithelium as occasional, solitary goblet cells staining for 17NM and acid mucins in seven gallbladders, and as focal collections of numerous goblet cells in another seven. Intestinal metaplasia was also occasionally found in the surface epithelium overlying a focus of pyloric gland metaplasia and in the ducts leading to the glands (eg, figure 1D,F).
K20 was present in one or more small, focal areas of surface epithelium in 10 of the gallbladders, with another four showing widespread staining.
Adenocarcinoma of gallbladder
The numbers of the 29 cancers staining for the antigens tested are given in table 1. Two cancers displayed all five foregut antigens, six displayed four, eight displayed three, six displayed two, two displayed one and five cancers did not express any. Of the last, three cancers stained for 17NM (1+), K20 (2+ or 3+) and CDX2; the remaining two were synchronous cancers, and stained only for CDX2.
The proportion of cancer cells expressing D10 and MUC6 differed widely between different cancers, but both were expressed in different cancer cells to those displaying 17NM or MUC5AC (figure 1G,H), resulting in an apparent inverse relationship between extent of expression of the two metaplastic-cell-phenotypes when greater than 50% of one or other was present in a cancer (figure 2). There was no statistical association between expression of any two of the metaplastic antigens tested or with expression of CDX2 or p53.
In all, 18 of the 28 gallbladders had cholelithiasis and the other 10 chronic cholecystitis; nine of the cancers were accompanied by areas of dysplasia, two associated with a carcinoma in situ, three were carcinoma in situ and four arose in an adenoma.
Comparison of gallbladder adenocarcinoma and cholangiocarcinoma
The frequency of expression of the five foregut antigens did not differ significantly between the three cancers (table 2). 17NM was expressed significantly more frequently by gallbladder and fluke-associated cancers than by sporadic cholangiocarcinomas. p53 Was more frequently expressed by both the gallbladder and fluke-associated cancers than sporadic cholangiocarcinomas, but only the fluke-associated cancers differed significantly (table 2).
Antigen expression in adenoma
One tubular adenoma had the majority (3+) of cells staining for 17NM and few cells (1+) expressing D10 and MUC5AC, and none 1F6 or MUC6. The other five adenomas showed variable staining for D10 (0–3+), MUC6 (0–3+) or MUC5AC (0–3+) with few (1+) or no cells staining for 17NM (table 1). Cells staining for D10 or MUC6 were distinct from those staining for 17NM (figure 1E).
Chronic cholecystitis with gallstones is the major risk factor for gallbladder cancer in Western countries, and the presence and significance of metaplasia in this condition have been well documented.3 ,4 ,8 ,9 We found the same metaplastic-cell phenotypes were also displayed by dysplastic epithelium and adenoma present in these gallbladders. Liver fluke infestation gives rise to chronic inflammation of the larger bile ducts with widespread epithelial hyperplasia and dysplasia, both of which display pyloric gland and intestinal metaplasia-cell phenotypes.10 Florid peribiliary gland hyperplasia and de novo gland formation around larger bile ducts accompanies the inflammation, with the glands displaying the pyloric gland metaplasia profile.10 Both gallbladder adenocarcinoma and fluke-associated cholangiocarcinoma therefore arise against a background of chronic inflammation, with the same two metaplastic-cell phenotypes presaging the cancers. Livers containing sporadic cholangiocarcinoma did not show such florid peribiliary gland hyperplasia or frequent examples of epithelial cell hyperplasia and dysplasia as those with fluke-associated cholangiocarcinoma.10
All but two gallbladder cancers displayed one or both of the metaplastic-cell phenotypes, the two negative tumours being unique also in that they were synchronous cancers. Pyloric gland metaplasia-cell phenotypes were expressed with similar frequencies by the gallbladder cancers and both groups of cholangiocarcinoma, suggesting that their expression of this phenotype was not due solely to a background of prior chronic inflammation. The gallbladder cancers and fluke-associated cancers differed from sporadic cholangiocarcinoma in showing higher frequencies of expression of the intestinal metaplasia-cell phenotype (intestinal goblet cell antigen 17NM) and the p53 protein. Although this suggests that a background of chronic inflammation may influence both the metaplastic-cell phenotype displayed and the pathway of tumour development, there was no statistical correlation between 17NM and p53 expression in either cancer.
Cholangiocarcinoma in Asia shows a higher frequency of p53 expression and different types of mutation than in Europe or the USA, and it has been suggested that environmental factors account for such geographic differences.17–19 The higher frequency of p53 expression in the gallbladder carcinomas compared with sporadic cholangiocarcinoma from the same population was not statistically significant but, given the relatively small number of the gallbladder cancers examined, it seems likely that a background of chronic inflammation promoted a higher frequency of p53 protein expression in the ensuing cancers. Other investigators have shown that the mucosa of gallbladders with increasing degrees of chronic inflammation (ie, extent of lymphocyte infiltration) show corresponding increases in expression of p21 (C1P1/WAF1), Ki67 and p53 proteins.20 As suggested above for 17NM expression, a prolonged inflammatory process itself would then lead to the higher frequency of expression of p53 seen in both gallbladder adenocarcinoma and fluke-associated cholangiocarcinoma.
The four tissue antigens (D10, 1F6, MUC6, TFF2) used to identify the profile of pyloric gland metaplasia are all normally present in mucous neck cells, pyloric glands, Brunner glands, and peribiliary and periductal glands of the pancreas: cells and tissues which are cytologically and functionally similar.12 ,21–25 It has been proposed that chronic ulceration in the gastrointestinal tract induces the emergence of a novel cell lineage from gastrointestinal stem cells—the so-called ulcer-associated cell lineage or UACL—which develops into pyloric gland metaplasia.21 ,22 We have previously suggested that the one mucous gland–cell lineage, inherent in foregut mucosa, is responsible for the development and differentiation of these various normal and metaplastic tissues, and certain biliary lesions.11 Our observations indicate that gallbladder adenocarcinoma and cholangiocarcinoma both arise from this same cell lineage through a pyloric gland metaplasia-intestinal metaplasia-dysplasia-carcinoma sequence. Other investigators have presented evidence for the age-related progression of an antral-type metaplasia-intestinal metaplasia-dysplasia sequence in the development of gallbladder cancer.26 The high proportion of gallbladder cancers expressing CDX2 is also in keeping with their origin from such a gastrointestinal cell lineage.
The higher frequency of expression of K20 in gallbladder cancer than in cholangiocarcinoma appears to be an independent phenomenon. We, and others, have reported higher frequencies of expression of K20 in cholangiocarcinoma arising in extra-hepatic bile ducts compared with those from more peripheral bile ducts, and the higher frequency found in gallbladder cancer would be in keeping with such a gradient.10 ,27 ,28
We conclude that dysplastic and malignant cell clones in gallbladder and bile ducts display pyloric gland or intestinal metaplasia-cell phenotypes in keeping with a foregut cell lineage-metaplasia-dysplasia-carcinoma sequence in gallbladder and biliary cancer development.
Take home message
Their consistent display of pyloric gland and intestinal metaplastic-cell phenotypes indicates that both adenocarcinoma of gallbladder and cholangiocarcinoma generally arise from a foregut cell lineage via a metaplasia-dysplasia-carcinoma sequence. A background of chronic inflammation increases the frequency of the intestinal metaplasia-cell phenotype and p53 expression in both cancers.
The authors thank Professor A Giraud for the antiserum to TFF2, and the Editor of the Journal of Clinical Pathology for permission to include some previously published data (from10) in table 2. Consent to use the archived or surgical material for the study was granted by the HREC, Melbourne Research, The University of Melbourne.
Contributors NRH and PSB conception and design and data analysis and interpretation. NRH drafted and PSB revised the manuscript. Both authors take responsibility for contents of the manuscript.
Competing interests None.
Ethics approval HREC, Office for Research, Melbourne Health, Royal Melbourne Hospital, Melbourne, Victoria. The project was later transferred to the University of Melbourne, and ethics approval granted by HREC, Melbourne Research, the University of Melbourne.
Provenance and peer review Not commissioned; externally peer reviewed.