VISTA expression is associated with a favorable prognosis in patients with high-grade serous ovarian cancer.

Blockading programmed death ligand 1 (PD-L1) shows promising results in patients with some cancers, but not in those with ovarian cancer. V-domain Ig suppressor of T cell activation (VISTA) is a recently discovered immune checkpoint protein that suppresses T cell activation. This study aimed to investigate the expression and clinical significance of VISTA in ovarian cancer as well as its relationship with PD-L1. VISTA and PD-L1 levels in 146 ovarian cancer samples were assessed using immunohistochemistry. We investigated the association between VISTA and other clinicopathological variables, including survival. The associations between the VISTA-encoding C10orf54 gene, other immune checkpoints, and survival were analyzed. VISTA was detected in 51.4% of all samples and 46.6% of PD-L1-negative samples; it was expressed in 28.8%, 35.6%, and 4.1% of tumor cells (TCs), immune cells (ICs), and endothelial cells, respectively. Furthermore, VISTA expression was associated with pathologic type and PD-L1 expression. Moreover, VISTA expression in TCs, but not in ICs, was associated with prolonged progression-free and overall survival in patients with high-grade serous ovarian cancer. The expression of C10orf54 mRNA was associated with prolonged overall survival and immune escape-modulating genes. These results showed that VISTA expression in ovarian tumor cells was associated with a favorable prognosis in patients with high-grade serous ovarian cancer; however, additional studies are required to better understand the expression and role of VISTA in ovarian cancer.

Quantitation of secretory component levels in cyst fluids, ascitic fluids, and sera from ovarian adenocarcinoma patients.

A secretory component (SC) was detected by radioimmunoassay in the cyst fluids, ascitic fluids, and sera from patients with ovarian adenocarcinomas. Serous cyst fluids and ascitic fluids showed lower levels (expressed as means +/- SE) of SC (1.37 +/- 0.37 and 1.24 +/- 0.24 microgram/ml, respectively) than mucinous cyst fluids (181.50 +/- 50.40 microgram/ml). SC levels in the sera of all patients with ovarian adenocarcinoma were high (12.67 +/- 1.43 microgram/ml) when compared to SC levels in the sera of normal individuals (2.34 +/- 0.41 microgram/ml). Sera from patients with ovarian cancers diagnosed as serous, mucinous, papillary, and poorly differentiated adenocarcinomas showed SC levels of 9.93 +/- 1.68, 22.44 +/- 3.24, 7.35 +/- 1.13, and 10.10 +/- 1.58 microgram/ml, respectively.

Leukocyte migration in ovarian carcinoma: comparison of inhibitory activity of tumor extracts.

Inhibition of migration of leukocytes from patients with serous cystadenocarcinoma of the ovary was studied by the use of several different types of ovarian carcinoma extract as antigen. KCl extract of an ovarian carconoma was found to be the most effective antigen preparation in comparison with saline, deoxycholate, and perchloric acid extracts. Low concentrations of KCl ovarian carcinoma extract significantly inhibited migration of leukocytes from 11 of 17 patients with ovarian carcinoma (migration index, less than 0.74). Leukocytes from patients with breast, colon, or endometrial carcinoma showed minimal reactivity with ovarian carcinoma KCl extract, and leukocytes from patients with ovarian carcinoma showed minimal reactivity with KCl extracts of breast, colon, and endometrial carcinoma. These results suggested that the 3 M KCl procedure is superior for the isolation of antigens active in the leukocyte migration inhibition test and that this test may be of use for the isolation of tumor-associated antigen and the immunodiagnosis of ovarian carcinoma.

Carcinoembryonic antigen in human ovarian neoplasms.

The presence and location of carcinoembryonic antigen (CEA) was examined in tissue sections of 14 mucinous and 13 serous cystadenomas and cystadenocarcinomas of the ovary. CEA was demonstrated in the mucinous tumors but was not present in the serous tumors. In the mucinous tumors, CEA was located in the glycocalyx and apical portions of the absorptive-type epithelium, with only trace quantities in goblet cells, a pattern identical to that seen in colonic neoplasms. Endocervical-type epithelium in the mucinous tumors contained little or no CEA.

Expression of cell-mediated immunity and blocking factor using a new line of ovarian cancer cells in vitro.

A cell-mediated cytotoxicity test, quantitated by postlabeling with tritiated thymidine, was used to asses immune reactivity of cancer patients to the HeW cell line derived from serous cystadenocarcinoma of the ovary. Lymphocytes from 71.4% of serous and mucinous cystadenocarcinoma patients demonstrated a cytotoxic response towards the HeW cells, whereas no reactivity was observed towards target cells derived from nonovarian cancer. These observations indicated that the HeW cells express tumor-associated antigen. In some patients bearing similar tumors, cytotoxicity was blocked by ascitic fluid from other patients with cystadenocarcinoma. In addition, antigen obtained from the spent culture fluid of HeW cells exhibited blocking activity in a typical dose-response fashion, suggesting that blocking factor may be free tumor-associated antigen or an antigen-specific suppressor molecule. Thus, blocking of the lymphocytotoxic response of cystadenocarcinoma patients towards HeW cells may be utilized to monitor the isolation of ovarian carcinoma-associated antigen.

Monoclonal antibodies recognizing tumor-associated antigen of human ovarian mucinous cystadenocarcinomas.

Spleen cells from BALB/c mice immunized with human ovarian cystadenocarcinoma extract were fused with the mouse myeloma cell line P3/NSI/1-Ag 4 in the presence of polyethylene glycol (Mr 4000). Of the 46 hybrids obtained, four secreted antibodies preferentially reactive to the immunizing ovarian tumor extract. Two of these hybrids, which showed no reaction with normal controls, were selected for cloning by the limiting dilution method. The numerous clones obtained from each hybrid were screened against a panel of five ovarian tumor extracts, pooled normal ovary extracts, and pooled normal human sera. One clone from each hybrid that showed specificity for ovarian mucinous cystadenocarcinomas was recloned to assure monoclonality and to establish a permanent hybridoma cell line. The antibodies secreted by these cell lines were of IgG1 subclass with kappa light chains. These antibody-producing hybridomas were selected for further analysis of the antibody specificity by a solid-phase radioimmunoassay and quantitative absorption tests. The monoclonal antibodies recognized an antigenic determinant present only in mucinous cystadenocarcinomas of the ovary. These did not react with any other gynecological or nongynecological tumor thus far tested. The antigen was not demonstrable in any normal adult tissues tested. Among fetal tissues examined, only fetal intestine extract showed a positive reaction. The antigenic determinant recognized by these monoclonal antibodies was unrelated to carcinoembryonic antigen, normal glycoprotein, normal human serum components, or human ABO blood group materials. These antibodies, which have relatively high affinity and can be produced in large amounts, will be useful for the isolation and immunochemical characterization of this antigen. The purified antigen and the specific antibodies could be then combined in a sensitive radioimmunoassay for the early detection of the antigen in the sera and body fluids of patients with ovarian mucinous cystadenocarcinomas.

CA54/61 as a marker for epithelial ovarian cancer.

Using a new one-step, double-determinant enzyme immunoassay, we performed quantitative measurements of a mucin-type glycoprotein antigen (CA54/61) that we recently detected in sera of ovarian carcinoma patients. When the cutoff value was set at 12 units/ml, at which a high diagnostic efficiency was demonstrated [or at 20 units/ml (mean + 3 SD of healthy females)], the positive rates of ovarian serous, mucinous, clear cell, and endometrioid carcinomas were 76% (or 63%), 63% (or 55%), 57% (or 52%), and 50% (or 38%), respectively. Even in mucinous cystadenocarcinoma, more than one-half of the cases were positive, indicating the potential utility of the assay in the diagnosis of mucinous tumors. In sera from patients with benign ovarian tumors, only 9% (or 4%) of the cases were positive, indicating the quite high specificity of this test for ovarian carcinomas. To make a comparison between CA54/61 and CA125, we set the cutoff level of CA125 at 110 units/ml, at which value a high diagnostic efficiency was demonstrated [or at 35 units/ml (mean + 3 SD of healthy females)]. When both CA54/61 and CA125 were assessed in sera from 36 patients with mucinous cystadenocarcinoma, the positive rates of CA54/61 and CA125 were 64% (or 56%) and 36% (or 56%), respectively, suggesting that CA54/61 is of clinical value as a new tumor marker for ovarian cancers, including mucinous tumors.

A study of cyst fluid and plasma carcinoembryonic antigen in patients with cystic ovarian neoplasms.

Cyst fluid and plasma carcinoembryonic antigen (CEA) levels were measured in 11 patients with ovarian cyst-adenocarcinoma and in 16 patients with benign ovarian neoplasms. In patients with ovarian cancer, plasma CEA levels were not elevated above 2.5 ng/ml unless cyst fluid CEA levels were 4 to 16 mu-g/ml. In this series, cystic and plasma CEA levels were elevated most consistently in patients with mucinous ovarian tumors. Furthermore, on the basis of molecular size and immunoreactivity by immunodiffusion, ovarian cancer cyst fluid CEA and colonic cancer CEA had similar immunochemical properties. Consistent with the findings in other neoplasms, follow-up studies showed that plasma CEA levels returned to the normal range between 2 and 12 weeks after surgical excision of the ovarian tumor. It is concluded that plasma CEA is of value in the management of patients with ovarian mucinous cystadenocarcinoma.

Generation of monoclonal antibodies reacting with human epithelial ovarian cancer.

Fusion of the murine myeloma line P3-X63-Ag8-U1 with spleen cells from a mouse immunized with a membrane preparations (CM) of a mucinous ovarian cystoadenocarcinoma yielded two monoclonal antibodies, MOv1 and MOv2, which reacted by solid-phase radioimmunoassay with immunizing tumor CM but were unreactive with normal kidney CM as well as with plasma proteins and peripheral blood cells from the immunizing carcinoma patient. MOv1 and MOv2 were further tested by solid-phase radioimunoassay on a panel of different CM from fresh surgical specimens of ovarian and nonovarian carcinomas, benign ovarian tumors, normal ovary and kidney tissues, and on various tumor cell lines. In addition, the antibodies were characterized by immunofluorescence on live cells from cell lines and surgical specimens, and on frozen sections of benign and malignant ovarian tumors, of nonovarian tumors, and of normal tissues. The results obtained indicate that MOv1 and MOv2 recognize two different epitopes on molecules present on malignant and benign ovarian mucinous tumors and colonic glands. In addition, the antigen recognized by MOv2 was also detected in carcinmas of lung, colon, stomach, and breast; in gastrointestinal glands; and in the glandular lumina of normal lactating breast.

Immunoperoxidase localization of a high-molecular-weight mucin recognized by monoclonal antibody 1D3.

The distribution of an antigen recognized by murine monoclonal antibody 1D3 (Bhattacharya, M., Chatterjee, S.K., Barlow, J. J., and Fuji, H. Cancer Res., 42: 1650-1654, 1982) was investigated in various types of human malignant and normal adult tissues by indirect immunoperoxidase assay in fixed paraffin-embedded sections. One hundred percent of ovarian mucinous cystadenocarcinomas expressed high levels of the antigen with intense staining of 80 to 100% of the tumoral area, thus confirming our previous finding with radioimmunoassay and absorption analyses. About 51% of colonic carcinomas, 33% of gastric carcinomas, and 22% of pancreatic carcinomas were also positive for this high-molecular-weight mucoprotein antigen. All other ovarian and nonovarian carcinomas tested including carcinoma of lung, breast, endometrium, cervix, and prostate were not stained by 1D3. In addition, sarcomas, melanomas, and lymphomas also did not express any detectable level of the antigen. When surveyed against various normal adult tissues, 1D3 had reactivity limited to the colon. Normal colon, however, exhibited reduced staining intensities compared to tumors or to the apparently normal colon adjacent to tumors. The antigen thus appears to be a colorectal tissue-specific antigen showing increased levels in ovarian mucinous cystadenocarcinomas and in some gastrointestinal tumors. 1D3 antigen is a potential tumor marker for mucinous ovarian and colonic tumors.

Immunomodulation in patients with epithelial ovarian cancer after adoptive transfer of tumor-infiltrating lymphocytes.

The immunomodulation determined by natural killer cell activity, delayed-type hypersensitivity to purified protein derivative and phytohemagglutin, and phenotypic changes of peripheral blood lymphocytes was characterized in 12 patients with epithelial ovarian cancer who received adoptive transfer of tumor-infiltrating lymphocytes (TILs) after cisplatin-containing chemotherapy (TIL group). As a control, 10 patients with epithelial ovarian cancer who did not receive infusions of TIL were also examined in the same fashion. In the TIL group, peripheral blood lymphocytes showed increased percentages of cells bearing the CD8 antigen, in contrast to stable percentages of CD4 antigen-bearing cells, resulting in a decreased ratio of CD4+ to CD8+ cells. The percentages of CD16 and CD56 antigen-bearing cells also increased in proportion to augmentation of natural killer cell activity against K562 cells. Additionally, with regard to cell-mediated immunity determined by delayed-type hypersensitivity to phytohemagglutin and purified protein derivative, significantly and slightly enlarged erythema was observed 2 and 8 weeks, respectively, after the injection of TILs (phytohemagglutin, P < 0.05; purified protein derivative, not statistically significant). The control group showed no major changes in any of the immunological markers. These results suggest the possibility that the adoptive transfer of TILs induces immunoactivation of cellular immunity and enhances natural killer activity in patients with epithelial ovarian cancer.

Adjunct to the diagnostic distinction between adenocarcinomas of the ovary and the colon utilizing a monoclonal antibody (COL-4) with restricted carcinoembryonic antigen reactivity.

Malignant ovarian tumors may represent either primary ovarian cancers or metastatic lesions (from patients with demonstrated primary cancers at other body sites) whose distinction may be difficult using clinical, surgical, and pathological criteria. Monoclonal antibody (MAb) COL-4, reactive with carcinoembryonic antigen, has previously been shown to react preferentially with adenocarcinomas of the colon versus a variety of normal tissues. We report here that MAb COL-4 is strongly reactive with primary colonic carcinomas (N = 50), as well as regional (N = 42), and distant (N = 20) metastases of colonic adenocarcinoma. In contrast, MAb COL-4 demonstrated little to no reactivity with primary (N = 53) and metastatic carcinomas of the ovary (N = 23) including serous, mucinous, and poorly differentiated adenocarcinomas using immunohistochemical techniques. This differential reactivity was statistically significant (P less than 0.001), suggesting the potential clinical utility of MAb COL-4 in the differentiation of ovarian from colonic adenocarcinoma. Solid-phase quantitative radioimmunoassays and Western blotting techniques confirmed these results. Data are also presented that the carcinoembryonic antigen molecules or epitopes recognized by a more classical broadly reactive anti-carcinoembryonic antigen MAb are distinct from those recognized by MAb COL-4. Other carcinomas which also metastasize to the ovary and may be confused clinically with a primary ovarian tumor such as adenocarcinomas of the stomach and breast were also evaluated for reactivity with MAb COL-4. COL-4 was also reactive with all gastric carcinomas evaluated, but failed to react with breast carcinomas. Hence, COL-4 can now be utilized as an immunohistochemical adjunct for the differentiation of ovarian from gastrointestinal adenocarcinoma which can be difficult to distinguish by clinical, surgical, and histological parameters.

IL-27 suppresses SKOV3 cells proliferation by enhancing STAT3 and inhibiting the Akt signal pathway.

Ovarian cancer continues to be the most lethal gynecologic malignancy worldwide. IL-27 is a novel member of the IL-12 cytokine family. The aim of this study was to investigate the effects of IL-27 on the ovarian cystadenocarcinoma cell line SKOV3 and determine possible mechanisms underlying its effect. We stably transfected an IL-27 plasmid, empty vector, IL-27 shRNA or negative control into SKOV3 cells. Cell proliferative activity was evaluated using a WST-1 cell proliferation assay kit. Cell viability was quantified by measurements of lactate dehydrogenase release. The mRNA levels of nine genes were tested by q-PCR. Western blotting was used to verify apoptosis and signal pathways. We found that the IL-27 plasmid significantly enhanced cytotoxicity and inhibited the proliferation of SKOV3 cells. Caspase-3 protein was augmented by IL-27 plasmid and abated by IL-27 shRNA. The incremental expression of IL-27 activated the STAT3 pathway and attenuated the Akt pathway. The over-expression of IL-27 could significantly upregulate a series of antitumor cytokines including IL-6, IL-12 and interferon-γ and down-regulate protumor factors such as TLR4 and NF-κB1. Our data show that IL-27 has direct antitumor capacity in ovarian cancer cells via enhancing apoptosis by inducing the STAT3 pathway and restraining the Akt pathway. Précis: IL-27 enhanced the cytotoxicity and suppressed the proliferation of ovarian cancer cells by activating STAT3 and inhibiting the Akt signal pathway. IL-27 plays an important role in antitumor activity against epithelial ovarian cancer.

Clinical evaluation of circulating serum sialyl Tn antigen levels in patients with epithelial ovarian cancer.

Sialyl Tn antigen (NeuAc alpha 2----6GalNac alpha 1----0-Ser/Thr [STN]) with antigenic specificity in the core structure of mucin-type carbohydrate chains has been determined. In the present study, we evaluated the clinical significance of this new carbohydrate antigen, STN, in patients with epithelial ovarian cancer. With the use of a radioimmunoassay developed to detect STN antigen in serum, elevated (greater than or equal to 32.6 U/mL) antigen levels were observed in 50.0% of patients with ovarian cancer. In contrast, 3.8% of healthy individuals had STN antigen levels greater than or equal to 32.6 U/mL. In 9.6% of patients with benign gynecologic diseases and 0% of pregnant women, there were elevated levels of STN antigen. There was a significant difference (P less than .001) in STN antigen levels between patients with ovarian cancer and patients with benign gynecologic diseases, pregnant women, or the controls. The mean +/- SD for all evaluated samples of ovarian cancer was 109.2 +/- 146.8 U/mL. Both the mean values and the positive rate increased as the stage advanced. Classified according to the histologic type, the highest positive rate (61.0%) was observed in mucinous adenocarcinoma. The usefulness of STN antigen as a circulating tumor marker in ovarian cancer was estimated as follows: sensitivity 50.0%, specificity 93.5%, positive predictive value 72.2%, negative predictive value 84.7%, and diagnostic value 46.8%. Serum STN antigen levels were elevated in 12 of 33 patients with ovarian cancer who had serum CA 125 antigen levels less than 35 U/mL. While CA 125 antigen levels were elevated in 74.6% and STN antigen levels were elevated in 50.0% of the same population, the use of both assays indicated the sensitivity of detection of 83.8% in the population studied.

The cellular composition in the peritoneal cavity and the cytotoxic function of the peritoneal cells from patients with ovarian cancer; effect of tumor necrosis factor-alpha treatment.

In patients with epithelial ovarian cancer (EOC), the cellular composition in the peritoneal cavity and the functional capacities of the peritoneal cells (PC) are unknown. Especially the peritoneal macrophages (m phi) could play an important role in defense against tumor cells. To study the cellular composition in the peritoneal cavity and the functional capacities of PC, these cells were obtained from three patients with EOC. The PC were immunophenotyped and tested functionally in vitro in a cytotoxicity assay. One of the patients was treated intraperitoneally (i.p.) with a single dose of 0.06 mg/m2 tumor necrosis factor-alpha (TNF-alpha). PC were obtained before the treatment, after 24 h and after 1 week. PC from healthy women undergoing laparoscopic sterilization served as controls. It appeared that patients with EOC have a lower percentage of macrophages (m phi) in the peritoneal cavity than healthy persons. These m phi of patients were also less capable of killing U 937 tumor cells as compared to the peritoneal m phi of control persons. However, in the patient treated i.p. with TNF-alpha the cytotoxic capacities of the peritoneal m phi were strongly improved. The percentage cytotoxicity at an effector to target ratio of 10, increased from 17% to 80%. Thus, the peritoneal m phi in this patient were activated in vivo to a tumoricidal state. These findings indicate that PC in patients with EOC differ from controls, but further investigation is necessary to define the contribution of the disease and/or prior chemotherapy to this defect.

Expression of "myelomonocytic" antigens in mesotheliomas and adenocarcinomas involving the serosal surfaces.

Because of the demonstrated efficacy of Leu-M1 as a discriminant between malignant epithelioid mesothelioma (MEM) and adenocarcinomas involving the serosal surfaces (ACSs), the authors assessed the reactivities of related "myelomonocytic" antigens in this context. Paraffin sections from 41 MEMs and 43 ACSs (pulmonary, "serous surface papillary," and metastatic mammary adenocarcinomas) were evaluated for their expression of Leu-M1, LN1, LN2, and the Mac 387 antigen. Diagnoses were based in each case on the results of conventional histologic and electron microscopic examinations. Leu-M1 was detected only in minute foci of three peritoneal MEMs and was absent entirely in pleural mesotheliomas. Conversely, 38 of 43 ACSs expressed this marker. Three cases of peritoneal MEM and one pleural mesothelioma were multifocally LN2 positive, as were 39 of 43 ACSs. LN1 was the most frequently expressed antigen in MEM, being observed in 18 such tumors (10 pleural; 8 peritoneal); it was also detected in 37 of 43 ACSs. Mac 387 failed to label any of the neoplasms assessed in this series. These results demonstrate similar patterns of "myelomonocytic" antigen expression by diverse ACS and a general absence of Leu-M1 and LN2 in MEM. LN1 and the Mac 387 antigen appear to have no additional value when compared with Leu-M1 and LN2 in the immunohistochemical evaluation of serosal neoplasms.

Immunohistochemical and ultrastructural localization of T antigen in ovarian tumors.

Thomsen-Friedenreich (T) antigen, the immediate precursor antigen of the human blood group MN system, has been found in many carcinomas, but it is suppressed in normal tissues and nonmalignant diseases. Using a monoclonal antibody specific for the T epitope and an indirect immunoperoxidase technique at light and electron microscopic levels, the authors studied the expression of T antigen and its potential diagnostic value in ovarian tumors. Among 30 serous and mucinous ovarian cystadenocarcinomas, 20 (67%) were positive and 10 (33%) were negative for T antigen. In carcinomas, positive rates increased in parallel with the tumor grade and were 37%, 75% and 80% for grade 1, 2, and 3 tumors, respectively. Of the nine patients with metastasis, seven (78%) had positive and two had negative reactions in their primary and metastatic tumors. T antigen staining was observed at the intraluminal cell surfaces and peripheral cell membranes. The ultrastructural localization of T antigen revealed electron-dense reaction products at the cell surface and microvillous surfaces. Of the ten benign ovarian tumors, three (30%) were weakly positive and seven (70%) were negative for T antigen. These findings indicate a positive correlation between the presence of immunoreactive T antigen and conventional unfavorable prognostic indicators in ovarian carcinoma. The surface location of T antigen suggests that it may have a functional role at the cell membrane and the membrane may be involved in secretion (shedding) of T antigen. Detection of T antigen may be a useful marker of prognosis in ovarian carcinoma.

An immunohistochemical study of distribution of carcinoembryonic antigen in epithelial tumours of the ovary.

An immunohistochemical study of the tissue CEA content of 82 epithelial neoplasms of the ovary has shown that mucinous tumours contain more of this substance than do their serous counterparts; otherwise a knowledge of tissue CEA content appears to be of little value in the differential histological diagnosis of this group of neoplasms. Among mucinous tumours there is only a partial correspondence between their degree of malignancy, as assessed histologically, and their content of CEA. It is postulated that immunohistological study of tissue CEA may add a degree of finesse to morphological analysis of these neoplasms and thus allow for a more precise grading of their degree of malignancy.

Expression of CEA, CA125, CA19-9 and human milk fat globule membrane antigen in ovarian tumours.

The expression of five different antigens in ovarian tumours was studied by means of an immunohistochemical test with anti-CEA, HMFG1 and HMFG2, NS19-9 and OC125 antibodies. Considerable variation was noted not only between different histological types and between tumours of one type but also between areas in a single tumour. HMFG1 and HMFG2 were the most reactive of all the antibodies; NS19-9 and OC125 were expressed by different populations of cells. It is concluded that specific combinations of antibodies are more effective both for the monitoring of ovarian cancer as well as for immunodiagnosis and treatment, than any single one used.

Endocrine cells in hepatobiliary cystadenomas and cystadenocarcinomas.

We investigated the distribution of endocrine cells in hepatobiliary cystadenoma (n = 5, two associated with mesenchymal stroma) and cystadenocarcinoma (n = 3) immunohistochemically. In normal livers (n = 20) and livers affected by hepatolithiasis (n = 15) used as controls, endocrine cells revealed by chromogranin immunostaining were located exclusively in normal or proliferating intrahepatic peribiliary glands. In the eight cases of hepatobiliary cystadenoma and cystadenocarcinoma, endocrine cells were present in four cases (50%) (1 cystadenoma, 1 cystadenoma with mesenchymal stroma, and 2 cystadenocarcinomas). Endocrine cells tended to be located beneath and among the columnar epithelial cells. Intrahepatic peribiliary glands were located in the vicinity of cystadenoma or cystadenocarcinoma in six (75%) of the eight cases, and they frequently showed cystic dilatation and contained endocrine cells. Intrahepatic peribiliary glands were located in the vicinity of the endocrine cells in all cystadenomas and cystadenocarcinomas that were positive for endocrine cells. These data show that about 50% of hepatobiliary cystadenomas and cystadenocarcinomas contain endocrine cells and suggest that hepatobiliary cystadenoma and cystadenocarcinoma may originate from intrahepatic peribiliary glands.