The use of cytokeratin 7 and EMA in differentiating ovarian yolk sac tumors from endometrioid and clear cell carcinomas

Yolk sac tumors (YSTs) have a variety of morphologic patterns, some of which can resemble either endometrioid adenocarcinoma (EAC) or clear cell carcinoma (CCC). Immunohistochemical staining for alpha-fetoprotein (AFP) is usually only focal and thus is not always helpful in the diagnosis of YST, and pancytokeratin (CK) is expressed by all three tumors. We studied a battery of immunohistochemical markers with specific attention to the utility of cytokeratin 7 (CK7) in differentiating YST from EAC and CCC. A total of 46 ovarian tumors were retrieved for this study: 16 YST, 19 EAC, and 11 CCC. The three groups were analyzed for the expression of CK7, AFP, Leu-M1 (CD15), EMA, and WT1 by immunohistochemistry. In addition, CK and c-kit (CD117) were studied in the YSTs. All of the YSTs tested (100%) were positive for CK. CK7 was considered negative in all 16 YST cases (100%), although a few tumor cells (1%-2%) stained in 4 cases. In contrast, 17 of 19 EACs and all 11 CCCs had diffuse 3+ to 4+ positivity for CK7; the two other EACs showed 2+ positivity for CK7 (40% and 30% of the tumors). AFP was positive in 12 of 15 YSTs (80%), but was generally focal with 1+ staining in 10 cases (67%); only 2 cases were 3+. All of the EACs and CCCs were negative for AFP. Leu-M1 was 1+ in 9 of 15 YSTs (60%), while the remaining 6 were considered negative. Leu-M1 was positive in 10 of 15 EACs tested (67%), but the staining was variable with 1 case 3+, 3 cases 2+, and 6 cases 1+. In the CCCs, 10 cases (91%) were 3+ to 4+, and 1 case was 1+. EMA was essentially negative in 15 of 15 YSTs (100%), with 3 completely negative and 12 showing very focal (<5%) staining. Eight of 12 EACs showed 4+ staining, 3 showed 3+ staining, and 1 showed 2+ staining. All of the 11 CCCs (100%) showed 4+ staining. WT1 was negative in all cases of YST and CCC; 16 of 18 EAC tested (89%) were negative for WT1, but 2 (11%) were 4+ positive. C-kit was negative in all YSTs. In conclusion, it is important for pathologists to be aware that YSTs may mimic EACs and CCCs and that this distinction is important for the clinical management of patients with these tumors. AFP staining is focal in most YST, so an absence of staining does not exclude this diagnosis. CK7 and EMA are essentially negative in YST but are diffusely positive in CCC and EAC, making them useful markers for differentiating YSTs from both CCCs and EACs. Leu-M1 may also be helpful for distinguishing YSTs from CCCs.     

A limited panel of immunomarkers can reliably distinguish between clear cell and high-grade serous carcinoma of the ovary

The distinction of ovarian clear cell carcinomas (CCCs) from high-grade serous carcinomas (HG-SCs) is sometimes a diagnostic challenge. With the recognition that CCCs respond poorly to conventional chemotherapy there are efforts to initiate clinical trials for CCC, making accurate diagnosis critical. The purpose of this study was to test and validate a set of antibodies that could aid in the diagnosis of CCC, using a series of cases from different centers in North America. Using a test set of 133 CCCs, we identified the following markers: Cyclin E, estrogen receptor, hepatocyte nuclear factor (HNF)-1beta, Ki-67, p21, p53, and Wilms tumor (WT)1 that show significant discrimination from 200 HG-SCs. For validation, these markers were characterized on an independent set of 104 CCCs from 3 other centers. There were no significant differences in expression of these 7 markers between the independent test and validation sets of CCC. Combining all CCC cases (N=237), HNF-1beta showed the highest sensitivity (82.5%) and specificity (95.2%) for CCC, and WT1 for HG-SC (sensitivity: 79.9%, specificity: 97.4%). A diagnostic panel consisting of WT1, ER, and HNF-1beta demonstrated nearly identical performance as a panel using all 7 markers in distinguishing CCCs from HG-SCs, correctly classifying 84% of cases. Three percent of cases were misclassified and 13% carried an uninformative triple negative immunophenotype. CCCs show a distinct, reproducible immunophenotype, compared with HG-SCs, and a panel of 3 immunomarkers can serve as a diagnostic aid in problematic cases.     

Calretinin,a more sensitive but less specific marker than alpha-inhibin for ovarian sex cord-stromal neoplasms: an immunohistochemical study of 215 cases

Although inhibin has been shown to be a sensitive marker for ovarian sex cord-stromal and fibrous neoplasms, it may be negative in some cases. Calretinin, a mesothelial marker, has shown promise as a marker for sex cord-stromal neoplasms. The aim of this study was to evaluate and compare calretinin and inhibin as immunohistochemical markers for sex cord-stromal and fibrous neoplasms. A total of 215 ovarian neoplasms were immunostained with commercially available antibodies to calretinin and inhibin. These tumors included 87 sex cord-stromal (39 granulosa cell, 13 Sertoli-Leydig, 4 Sertoli, 9 thecomas, 14 fibrothecomas, and 8 other stromal tumors), 37 fibrous (20 fibromas, 9 adenofibromas, and 8 fibrosarcomas), 65 epithelial, 22 germ cell, and 4 miscellaneous neoplasms. The staining was evaluated on a 0-4 scale based on percentage of neoplastic cells labeling: 0 = none; 1+ = 1-25%; 2+ = 26-50%; 3+ = 51-75%; 4+ = 76-100%. Calretinin reactivity was detected in 100% of sex cord-stromal and 90% of fibrous neoplasms, including 32 that were inhibin negative (2 granulosa cell tumors, 1 Sertoli-Leydig cell tumor, 1 thecoma, 3 fibrothecomas, 16 fibromas, 6 adenofibromas, and 3 fibrosarcomas). All four calretinin-negative fibrous neoplasms were inhibin negative. Calretinin staining was also detected in 22% of epithelial neoplasms but none of the germ cell and miscellaneous neoplasms tested. Inhibin staining was detected in 92% of sex cord-stromal neoplasms, 22% of fibrous neoplasms, 2% of epithelial neoplasms, and none of the germ cell and miscellaneous neoplasms tested. Calretinin has a 97% sensitivity and 85% specificity for sex cord-stromal and fibrous neoplasms, whereas inhibin has a 71% sensitivity and 99% specificity. This study shows that both calretinin and inhibin are useful in the diagnosis of ovarian sex cord-stromal and fibrous neoplasms. Calretinin is a more sensitive but less specific immunohistochemical marker than inhibin. Calretinin is particularly useful in the diagnosis of sex cord-stromal and fibrous neoplasms that are inhibin negative. The high frequency of calretinin in fibrous neoplasms suggests that a subgroup of these neoplasms may be derived from specialized gonadal stromal cells, perhaps thecal cells.     

Glypican 3: a novel marker in testicular germ cell tumors

Glypican 3 (GPC3), a membrane-bound heparin sulfate proteoglycan, may play a role in promoting embryonic cell growth and differentiation. GPC3 is mutated in Simpson-Golabi-Behmel syndrome, characterized by tissue overgrowth and an increased risk of embryonal malignancies. Recently, GPC3 was reported to be one of the over-expressed genes in testicular yolk sac tumors by gene expression microarray analysis. However, the presence of the GPC3 protein in germ cell tumors has never been investigated. The purpose of the study was to investigate the GPC3 expression in various histologic components of testicular germ cell tumors using immunohistochemistry and to assess its possible utility as a diagnostic marker. Tumors from 71 patients were examined, including components of 42 seminomas, 37 embryonal carcinomas, 24 yolk sac tumors, 20 teratomas with mature elements, 16 teratomas with immature elements, and 7 choriocarcinomas. Cytoplasmic and membranous immunoreactivity was semiquantitatively evaluated. All yolk sac tumor (24/24) and choriocarcinoma (7/7) components were immunoreactive for GPC3, whereas only 38% of teratomas with immature elements and 8% of embryonal carcinomas expressed GPC3. There was no immunoreactivity in benign testicular tissue, intratubular germ cell neoplasia, seminomas (0/42), or teratomas with mature elements (0/20). We conclude that the oncofetal protein GPC3 is a novel immunohistochemical marker in testicular germ cell tumors with differential expression in defined histologic subtypes. Our findings suggest a possible role of GPC3 in tumor cell differentiation. Furthermore, GPC immunostaining may be useful in the pathologic diagnosis of nonseminomatous germ cell tumors, particularly yolk sac tumor, and choriocarcinoma.     

Congenital teratoma: a clinicopathologic study of 22 fetal and neonatal tumors

Extragonadal teratoma is the most common congenital tumor. The prognostic significance of the grade of immaturity and the presence of small foci of conventional yolk sac tumor (YST) in fetal and neonatal teratomas have not been determined. We report detailed histologic studies of 22 congenital teratomas, including eight tumors resected in utero for developing hydrops, and correlate the histologic features with initial serum alpha-fetoprotein (AFP) levels and clinical outcome. All fetal tumors that required in utero intervention were grade 3 immature teratomas, with admixed conventional YST in 44%. Among tumors resected postnatally, those presenting in utero were more commonly immature (71% vs. 50%). All initial post-surgical serum AFP levels were high, as expected in a neonate. No correlation was found between AFP elevation above the mean for gestational age and the presence of YST, hepatic differentiation, or immature endodermal glands in the tumor. Among 15 survivors with follow-up, 5 patients had malignant mixed germ cell tumors (immature teratoma with foci of conventional YST) and 5 had immature teratomas with foci of hepatic differentiation or immature endodermal glands with subnuclear vacuoles (so-called "well-differentiated YST"). No patient has developed recurrent or metastatic disease after treatment by complete surgical excision alone (mean follow-up, 37.6 months). The clinical behavior of congenital teratomas is determined predominantly by whether or not the tumor can be completely resected and in our study did not correlate with the grade of the teratoma or with the presence or absence of foci of hepatic tissue, immature intestinal glands, or foci of conventional YST.     

Detection of AFP and HCG in metastatic testicular cancer after treatment with chemotherapy or radiation therapy

We used an indirect immunoperoxidase technique to detect alpha-fetoprotein (AFP) and human chorionic gonadotropin (HCG) in tissue sections of nine metastatic germ cell tumors excised after treatment with chemotherapy or radiation therapy, and correlated the results with the serum levels of AFP and HCG. In all but 1 case yolk sac tumor (YST) was the only histologic type that reacted for AFP (AFP+) and syncytiotrophoblasts (STB) were the only histologic type that reacted for HCG (HCG+). Among 5 cases with normalization of the serum AFP before surgery, 3 were associated with YST-/AFP-, 1 with YST+/AFP+, and 1 with YST+/AFP- metastases; and among 4 cases with normalization of the serum HCG all were associated with STB-/HCG- metastases. Among 3 cases with persistent elevation of the serum AFP, 1 was associated with YST+/AFP+, 1 with YST+/AFP-, and 1 with YST-/AFP- metastases; and of 2 cases with persistent elevation of the serum HCG, 1 was associated with STB-/HCG- and 1 with STB+/HCG+ metastases. These data suggest that marker normalization in the face of persistent tumor results primarily from eradication of YST and STB, but also from treatment-induced inhibition of AFP and HCG synthesis or secretion.     

Distinction of hepatocellular carcinoma from benign hepatic mimickers using Glypican-3 and CD34 immunohistochemistry.

Distinguishing a well-differentiated hepatocellular carcinoma (HCC) from normal and cirrhotic liver tissue or benign liver nodules, such as hepatic adenoma (HA) and focal nodular hyperplasia (FNH), may be very difficult in some cases, particularly in small needle core biopsies. We studied the expression of Glypican-3 (GPC3) and CD34 in 107 cases of HCC, 19 cases of HA, and 16 cases of focal nodular hyperplasia (FNH). In addition, we studied GPC3 expression in 225 cases of nonhepatic human tumors with epithelial differentiation. Ninety-four of 107 cases (88%) of HCC showed focal or diffuse cytoplasmic GPC3 staining, whereas all HA and FNH cases were GPC3-negative, and only 7 of 225 cases (3%) of nonhepatic tumors with epithelial differentiation expressed GPC3. The sensitivity and specificity of GPC3 for HCC was 88% and 97%, respectively. There were three CD34 staining patterns observed in hepatic tissue: negative, incomplete positive, and complete positive. In negative staining pattern, only blood vessels in portal triads or rare sinusoidal spaces immediately adjacent to portal tracts were positive. The negative staining pattern was seen in normal or cirrhotic liver tissue only. The complete CD34 staining pattern showed virtually all sinusoidal spaces with CD34-positive staining throughout the lesion. The complete CD34 staining pattern was seen in virtually all cases of HCC and in only some cases of HA and FNH. The incomplete CD34 staining pattern was characterized by either CD34 positivity in virtually all sinusoidal spaces in some but not all nodules or CD34 positivity in the peripheral sinusoidal spaces adjacent to portal triads. The incomplete CD34 staining pattern was seen in rare cases of HCC and in most cases of HA and FNH. We conclude that GPC3 is a very specific marker not only for differentiating HCC from nonhepatic tumors with epithelial differentiation, but also for differentiating HCC from HA and FNH. GPC3 immunoreactivity, in combination with a complete CD34 immunostaining pattern, greatly facilitates the accuracy of distinguishing between malignant hepatic lesions and benign mimickers.     

Primary and metastatic mucinous adenocarcinomas in the ovaries: incidence in routine practice with a new approach to improve intraoperative diagnosis

Mucinous carcinomas are reported to comprise 6-25% of ovarian carcinomas (mean 12%), but recent refinements in the interpretation of histologic features of noninvasive and metastatic mucinous carcinomas suggest that this may be an overestimate. Mucinous carcinomas in the ovaries are commonly metastatic, but the proportion of primary versus metastatic mucinous carcinomas in unselected patients is unknown. To evaluate the histologic-type distribution of ovarian surface epithelial carcinomas, a consecutive series of 124 ovarian carcinomas was reviewed using uniform current criteria. Microinvasive and intraepithelial carcinomas and carcinomas arising in germ cell and stromal tumors were excluded. To evaluate the proportions of primary and metastatic tumors among the mucinous carcinomas, 52 consecutive mucinous carcinomas from nonreferral patients were reviewed. Three of 124 primary ovarian carcinomas were pure mucinous carcinomas (2.4%). Among 52 cases of mucinous carcinomas in the ovaries, 40 (77%) were metastatic and 12 (23%) were primary. Among the 12 primary mucinous tumors, three were atypical proliferative mucinous tumors with microinvasion and nine were invasive mucinous carcinomas. Among the 40 metastatic mucinous tumors, 18 (45%) were from the gastrointestinal tract, 8 (20%) were pancreatic, 7 (18%) were gynecologic malignancies (5 cervical, 2 endometrial), 3 (8%) were from the breast, and 4 (10%) were of unknown primary site. A simple rule that classifies all bilateral mucinous carcinomas as metastatic, unilateral mucinous carcinomas <10 cm as metastatic, and unilateral mucinous carcinomas >or=10 cm as primary correctly classified 90% of the neoplasms. This algorithm for distinguishing primary and metastatic mucinous carcinomas in the ovary can be used at the time of intraoperative consultation to guide surgical management. After careful exclusion of noninvasive, microinvasive, and metastatic tumors, pure mucinous adenocarcinoma primary in the ovary appears to be substantially less common than previously reported.     

Clear Cell Carcinoma and Clear Cell Odontogenic Carcinoma: a Comparative Clinicopathologic and Immunohistochemical Study

Clear cell carcinoma or hyalinizing clear cell carcinoma (CCC) and clear cell odontogenic carcinoma (CCOC) are rare, low-grade and typically indolent malignancies that can be diagnostically challenging. In this study the clinicopathologic, histologic, and immunohistochemical features of 17 CCCs and 12 CCOCs are examined. The differential diagnosis of clear cell malignancies in the head and neck is discussed. The relationship of CCCs and CCOCs to other clear cell tumors on the basis of their immunohistochemical staining patterns is postulated.     

The frequency and histology of hepatic tissue in germ cell tumors

The frequency of hepatic tissue and its histological characteristics were examined in 516 germ cell tumors. Hepatic tissue was observed in 48 cases (9.3%). The incidence of hepatic tissue was low in tumors of the ovary (5%), high in both retroperitoneal (27%) and sacro-coccygeal (24%) tumors, and low in both mature (0.3%) and immature teratomas (11%). It was usually encountered in infancy, and the frequency was high in both yolk sac tumors (48%) and mixed germ cell tumors (52%). The hepatic tissue found mainly in mature or immature grade 1 teratomas was similar to adult normal human liver tissue (Ha-type). Tissue in areas consisting of some immature somatic elements of a mixed germ cell tumor was similar to embryonic or fetal liver tissue (Hf-type). Many hepatic nests found in a polyembryoma were of both Ha- and Hf-types. The hepatic tissue found in close relation to yolk sac elements showed predominantly hepatocellular carcinoma-like features (HCLS). Immunohistochemically, the cytoplasm of adult liver-type cells was positive for alpha-1-antitrypsin (AAT), human albumin (ALB), and the third (C3) and fourth (C4) components of the complement system. The cytoplasm of fetal liver-type cells showed the same positivity; in addition, these cells were positive for alpha-fetoprotein (AFP) in 25% of the cases. The cytoplasm of hepatic cells of HCLS was positive for AFP, AAT, ALB, C3, and C4. A weakly positive reaction for CEA and CA19-9 was observed in bile duct-like structure in some Hf-type cases.     

Oxyphilic clear cell carcinoma of the ovary. A report of nine cases

Nine clear cell carcinomas (CCC) of the ovary with a prominent component of cells with abundant eosinophilic cytoplasm are reported. The majority of these tumors--which we have designated "oxyphilic clear cell carcinomas"--were misinterpreted by the referring pathologists as tumors of other types. Each specimen, however, had one or more features establishing it as a clear cell carcinoma, including tubules and cysts lined by cuboidal, hobnail, or flattened cells; nests and sheets of cells with abundant clear cytoplasm containing glycogen; and an adjacent adenofibromatous component. Two tumors were bilateral. The ages of the patients (average, 55 years), their clinical presentation, and the gross appearance of the neoplasms were similar to those of clear cell carcinomas in general. The diagnosis of clear cell carcinoma should always be considered in the differential diagnosis of an ovarian tumor with oxyphil cells, particularly if the patient is postmenopausal. Thorough sampling should be undertaken in such cases to identify other, more typical foci of clear cell carcinoma in order to avoid misdiagnosis.     

Paraganglioma of the ovary: report of three cases of a rare ovarian neoplasm, including two exhibiting inhibin positivity

Paraganglioma is one of the rarest neoplasms to involve the ovary, whether primary or metastatic, with only two previous reports. We describe three examples that occurred in patients 22, 58, and 68 years of age. Two patients had hypertension. Two tumors involved the left ovary and one the right ovary; they ranged from 8 to 22 cm, were solid, and were tan, brown, or yellow. One tumor was confined to the ovary; in the second case, there were tumor deposits on the posterior surface of the uterus and the contralateral ovary; in the other case, there was peri-aortic lymph node involvement and peritoneal deposits. In all cases, however, radiologic investigations did not reveal an alternative primary site. On microscopic examination, all three tumors showed a predominantly nested "zellballen" pattern with groups of cells surrounded by a vascular stroma. Tumor cells largely had abundant granular eosinophilic cytoplasm with, in 2 cases, focal clear cell areas. In 1 case, bizarre tumor giant cells were present. Immunohistochemically, all neoplasms were cytokeratin negative and diffusely positive with neuroendocrine markers. In 1 case, there was an S-100-positive population of sustentacular cells. Two cases were positive for inhibin, one focal and one diffuse, and the other was focally positive for calretinin. Electron microscopy performed in 2 cases revealed dense core neuroendocrine granules. One patient has been followed up for 15 years and is alive and well. Although metastatic spread from an undetected primary outside the ovary cannot be totally excluded for the 2 cases with extraovarian disease, we think that the neoplasms most likely represent primary ovarian paragangliomas. Because various neoplasms in the sex cord-stromal and steroid categories are likely to enter into the differential diagnosis, inhibin and calretinin positivity represents a significant potential diagnostic pitfall. The differential is broad and may include many other ovarian tumors, particularly those with an oxyphilic cell type. Possible theories of histogenesis of primary ovarian paraganglioma include an origin from extra-adrenal paraganglia in the region of the ovary or unidirectional differentiation within a teratoma.     

雌激素受体α、β在卵巢癌中的表达及意义

目的　研究雌激素受体 ( ER) α、β在卵巢癌中的表达及意义。　方法　应用免疫组织化学 ABC法检测了卵巢癌 2 6例 ,其中浆液性癌 3例、内膜样腺癌 1 0例、颗粒细胞癌 1 1例、粘液性癌 2例 ,以及 1 2例正常卵巢中 ERα和 ERβ的表达。　结果　 ERβ在正常卵巢组织均呈阳性表达 ,而在卵巢癌组织中则表达不一 ,1 1例颗粒细胞癌全部为强阳性表达 ,1例浆液性癌、6例内膜样癌和 1例粘液性癌呈阳性表达。 ERβ表达阳性率在正常卵巢和卵巢癌分别为 1 0 0 %和 73 .1 % ,两者比较有显著差异 ( P<0 .0 5) ;ERβ阳性细胞密度颗粒细胞癌 ( 574.0 4± 45.3 6)较正常卵巢 ( 2 1 9.55± 46.84)明显增加 ( P<0 .0 1 )。在正常卵巢和卵巢癌中均未发现 ERα的阳性表达。　结论　 ERβ的表达与变异可能与卵巢癌的类型有关 ;ERβ在卵巢颗粒细胞癌的形成中可能发挥着重要的作用 ;而 ERα与卵巢癌的发生可能无相关性     

Yolk sac differentiation in germ cell tumors. A morphologic study of 50 cases with emphasis on hepatic, enteric, and parietal yolk sac features

We assessed 50 germ cell tumors with areas of yolk sac tumor (YST) for a variety of features including histologic patterns; hyaline droplets; syncytiotrophoblastic elements; hepatic, enteric, and parietal yolk sac differentiation; and granulomatous reaction. Of prime interest was the fact that many YSTs formed hepatic-like foci (22%), enteric-like glands (34%), and parietal yolk sac structures (92%). Hepatoid areas were characterized by nests and cords of polygonal, acidophilic cells with prominent nucleoli and intense cytoplasmic staining for alpha-fetoprotein. Enteric differentiation occurred as well-defined glands with a sharp, striated border and relatively bland nuclear features. Ultrastructurally these glands had apical microvilli with associated glycocalyx and long anchoring rootlets. The apical cytoplasm and luminal contents stained for carcinoembryonic antigen. Parietal yolk sac differentiation was characterized by the intercellular accumulation of basement membrane substance as generally thick and longitudinally arranged bands of eosinophilic material. Such material, by electron microscopy, was both intra- and extracellular, and had irregular outlines and inhomogeneous electron density. It contrasted with the strictly intracellular, round, homogeneous, hyaline globules that, we believe, represent visceral yolk sac differentiation. This intercellular material stained positively for laminin, a basement membrane component. Assessment of 22 embryonal carcinomas and 24 germinomas failed to show hepatic, enteric, and parietal yolk sac features, with one possible exception. We believe these features, especially parietal yolk sac differentiation, are helpful in differentiating YSTs from embryonal carcinomas and germinomas.     

Hepatocyte antigen as a marker of hepatocellular carcinoma: an immunohistochemical comparison to carcinoembryonic antigen,CD10, and alpha-fetoprotein

Hepatocyte monoclonal antibody (Hep) (alternatively Hep Par 1 for Hep paraffin 1) has been reported to stain normal hepatic tissue and hepatocellular carcinoma (HCC) with high specificity. We have studied the Hepatocyte expression in 96 cases of HCC and 311 cases of nonhepatic epithelial tumors. All cases of HCC were also stained with CEA-Gold 5, CD10, and alpha-fetoprotein. Hep, CEA-Gold 5, CD10, and alpha-fetoprotein immunostains were performed on formalin-fixed, paraffin-embedded tissue sections. Hep immunoreactivity was detected in 88 of 96 cases of HCC (92%), with a cytoplasmic and granular pattern of staining. The level of Hep expression in HCC corresponded to the nuclear grade and growth pattern. All 50 cases of nuclear grade 1 and nuclear grade 2 HCC were positive (100%), whereas 37 of 44 nuclear grade 3 (84%) and 1 of 2 nuclear grade 4 (50%) were positive. Sixty-seven of 68 cases of HCC with a trabecular, pseudoglandular, or scirrhous growth pattern were positive (98%), whereas 22 of 27 cases of HCC with a compact growth pattern were positive (81%). CEA-Gold 5, CD10, and alpha-fetoprotein immunoreactivity was detected in 76% (73 of 96), 52% (50 of 96), and 31% (30 of 96) cases of HCC, respectively. The positive predictive value of the combination of all four antibodies was 97%. Three cases of HCC were negative for all four antibodies; these cases had a high nuclear grade or a sarcomatoid or compact growth pattern. Twenty of 311 cases of nonhepatic tumors were positive for Hep (6%): 15 were adenocarcinomas and five were neuroendocrine tumors. The negative predictive value of Hep in HCC was 94%. The Hep-positive nonhepatic epithelial tumors were easily distinguished from HCC by the expression of keratin 7 or keratin 20 for adenocarcinoma and chromogranin and synaptophysin for neuroendocrine tumors because HCC does usually not express these markers. With the exception of two cases of hepatoid gastric carcinoma, all Hep-positive nonhepatic epithelial tumors were negative for alpha-fetoprotein, CEA-Gold 5, and CD10. Our study demonstrates that Hep is a relatively specific marker for HCC. It is useful in differentiating HCC from primary hepatic cholangiocarcinoma and metastatic tumors when combined with other immunomarkers.