Uterine Endometrial Carcinoma Research Articles

G i protein activation of gonadotropin-releasing hormone–mediated protein dephosphorylation in human endometrial carcinoma

OBJECTIVE: Gonadotropin-releasing hormone receptor is demonstrated in uterine endometrial carcinomas. This study was performed to determine gonadotropin-releasing hormone receptor–mediated membrane events and to identify the guanosine triphosphate binding protein (G protein) subtypes linked to gonadotropin-releasing hormone receptor in the tumors. STUDY DESIGN: Endometrial carcinomas surgically removed had been screened for gonadotropin-releasing hormone receptor expression before plasma membrane isolation. The phosphoprotein level was observed in the phosphorus 32–labeled incorporation from [γ- 32P]adenosine triphosphate into the isolated plasma membranes. The G i (α subunit) protein was detected by immunoblotting and pertussis toxin–catalyzed adenosine diphosphate ribosylation. RESULTS: Incubation of phosphorus 32–labeled membranes with a gonadotropin-releasing hormone analog in the presence of guanosine thiotriphosphate caused a remarkable loss of phosphoprotein from 35 kd protein. This dephosphorylation action was dose dependent of the gonadotropin-releasing hormone analog, and the maximal effect (90% loss) occurred at 100 nmol/L. Pertussis toxin brought about adenosine diphosphate ribosylation of an immunodetected Gα i. Gonadotropin-releasing hormone analog alone or guanosine thiotriphosphate alone had no effect. Pretreatment of the membrane with the pertussis toxin completely inhibited gonadotropin-releasing hormone–mediated dephosphorylation of the 35 kd protein. CONCLUSION: These data demonstrate the coupling of gonadotropin-releasing hormone receptor to protein dephosphorylation through G i, raising the possibility that the antimitogenic action of gonadotropin-releasing hormone may occur by release of the action of protein phosphorylation to promote cell growth.(Am J Obstet Gynecol 1997;176:371-6.)

Coabnormal expression of cyclin D1 and p53 protein in human uterine endometrial carcinomas.

In the normal cell cycle, tumor suppressor gene products (p53) and cyclin (cyclin D1) cooperate. Abnormalities in the cooperation of these factors may result in malignant transformation of the cell. Mutant p53 protein overexpression is defined in many human cancers, including endometrial carcinoma. This study investigated the role of cyclin D1 in the development of human uterine endometrial carcinoma. Seventy-four patients whose pathology slides contained either normal or hyperplastic endometrium adjacent to endometrial carcinoma were studied. Immunohistochemical staining of the serial paraffin sections was performed using antibodies to p53 and cyclin D1. The expression of cyclin D1 was restricted to only a few cells of normal and hyperplastic endometrium, whereas it was preferentially expressed in 40% (30/74) of endometrial carcinomas. The cells that overexpressed cyclin D1 also overexpressed p53. Moreover, all 30 cases with varied distributions of cyclin D1-positive cells corresponded identically with the distribution of p53-positive cells. Diffuse positivity for cyclin D1 was specifically observed in clinically advanced stages of pathologic G2 and G3 tumors. The data suggest that coabnormal expression of cyclin D1 and p53 protein may contribute to the development of endometrial carcinoma and may also be involved in the progression to malignancy.

Coabnormal expression of cyclin D1 and p53 protein in human uterine endometrial carcinomas

BACKGROUND In the normal cell cycle, tumor suppressor gene products (p53) and cyclin (cyclin D1) cooperate. Abnormalities in the cooperation of these factors may result in malignant transformation of the cell. Mutant p53 protein overexpression is defined in many human cancers, including endometrial carcinoma. This study investigated the role of cyclin D1 in the development of human uterine endometrial carcinoma. METHODS Seventy-four patients whose pathology slides contained either normal or hyperplastic endometrium adjacent to endometrial carcinoma were studied. Immunohistochemical staining of the serial paraffin sections was performed using antibodies to p53 and cyclin D1. RESULTS The expression of cyclin D1 was restricted to only a few cells of normal and hyperplastic endometrium, whereas it was preferentially expressed in 40% (30/74) of endometrial carcinomas. The cells that overexpressed cyclin D1 also overexpressed p53. Moreover, all 30 cases with varied distributions of cyclin D1-positive cells corresponded identically with the distribution of p53-positive cells. Diffuse positivity for cyclin D1 was specifically observed in clinically advanced stages of pathologic G2 and G3 tumors. CONCLUSIONS The data suggest that coabnormal expression of cyclin D1 and p53 protein may contribute to the development of endometrial carcinoma and may also be involved in the progression to malignancy. Cancer 1996;78:1248-53.

Expression of component desmosomal proteins in uterine endometrial carcinoma and their relation to cellular differentiation

While the assessment of the malignancy of neoplasms is based on morphologic studies of cells and tissues, use of objective molecular markers is leading to a better understanding and more biologically meaningful classification of neoplasms. In recent years, changes in the expression of cell adhesion molecules, especially E-cadherin, catenin, and adenomatous polyposis coli (APC), in carcinomas have attracted the attention of researchers. However, little is known about desmosomes in the uterine endometrium or in endometrial carcinomas. In this study, we semiquantified the desmosomal components desmoplakin I and II and desmoglein, in tissue sections using confocal laser scanning microscopy (LSM), and examined their relationship to the pathological type, the occurrence of lymph node metastases, and the extent of myometrial invasion. Frozen sections of 31 specimens of normal endometrium, 5 specimens of atypical hyperplasia, and 41 specimens of endometrial carcinoma were stained by the immunofluorescence method using antidesmoplakin I and II and antidesmoglein, and these markers were then semiquantified in tissue sections by LSM. The expression and location of desmoplakin I and II and desmoglein were similar, and their expression decreased with loss of differentiation. The expression was lower in cases of lymph node metastasis than in negative cases and was lower in the cases with > one-half myometrial invasion than in cases with < one-half myometrial invasion. Reduction of desmoplakin I and II and desmoglein expression may play an important role in the invasiveness and metastatic activity of human endometrial carcinoma. They can therefore be used as differentiation markers for endometrial carcinoma.

Expression of GalCer sulfotransferase by human uterine endometrial carcinoma cell lines

We investigated the expression of sulfoglycolipids in several gynecological cancer cell lines by metabolic labeling with S-35-sulfate as well as the activity of GalCer sulfotransferase (ST) and arylsulfatase A (ASA), enzymes which are responsible for the synthesis and degradation of sulfoglycolipids. The endometrial carcinoma cell lines expressed sulfoglycolipids and showed ST activity, indicating that increased synthesis of sulfoglycolipids due to elevated ST activity is a characteristic of endometrial carcinoma that distinguishes it from other carcinomas. These cell lines could provide a useful model for studying the functions of sulfoglycolipids as well as biological properties of ST in cancer cells.

OVEREXPRESSION OF C-ERBB-2 PROTEIN IN ENDOMETRIAL CARCINOMA IS ASSOCIATED WITH ADVANCED-STAGE DISEASE

Expression of the c-erbB-2 product (p185(c-erbB-2)) was examined in plasma membrane isolations from 14 uterine endometrial carcinomas and 3 normal endometrial tissues. Overexpression of p185(c-erbB-2) was found in 8 of 9 endometrial carcinomas of stages III and IV and in none of 5 early stage specimens, when compared with the level in normal endometrial tissues. The expression of p185(c-erbB-2) was independent of histological grade. When the p185(c-erbB-2) immunoprecipitated with anti-p185(c-erbB-2) antibodies was exposed to adenosine triphosphate, enhanced self-phosphorylation occurred more frequently in specimens from the tumors carrying p185(c-erbB-2) overexpression. These findings demonstrated positive correlation between disease spread, p185(c-erbB-2) expression and autophosphorylation of p185(c-erbB-2) in human endometrial carcinoma. The prognostic value of these observations awaits continued study.

DIFFERENT PATTERNS OF P53 GENE-MUTATIONS IN ENDOMETRIAL CARCINOMAS AND ENDOMETRIOID CARCINOMAS OF THE OVARY

Mutation and overexpression of the p53 gene frequently occur in uterine endometrial carcinomas and endometrioid carcinomas of the ovary. To define the pathogenetic relationship between endometrial neoplasms we examined p53 mutations and allelic losses in these diseases. Genomic deoxyribonucleic acid (DNA) was extracted from 12 endometrioid carcinomas of the ovary and 41 adenocarcinomas and 8 adenosquamous carcinomas of the uterus. Exons 4 through 8 of the p53 gene were amplified by means of the polymerase chain reaction. Mutations in each exon were detected with single-strand conformation polymorphism analysis and characterized with nucleotide sequencing analysis. Five (42%) and 6 (75%) of 12 endometrioid carcinomas were found to have mutations in exons 5 and 7, which encode for the functional domains of p53 molecules and allelic loss of exon 4, respectively. All these patients with aberrant p53 presented at an advanced clinical disease stage. Eighteen percent of the endometrial carcinomas demonstrated a random mutation in exons 5 through 7, and no association was observed between the mutation and the clinical stage and histological grade. The finding of different patterns of p53 mutations in endometrial carcinoma and endometrioid carcinoma suggest that each may follow a different molecular developmental pathway.

Neoplastic lesions of questionable significance to humans.

Many compounds giving a positive result in animal carcinogenicity studies through mechanisms involving secondary carcinogenesis pose little or no risk to humans. This article provides an overview of current understanding, with particular reference to renal tumors in male rats with alpha 2mu-globulin nephropathy, urinary bladder neoplasia in rodents, mesovarian leiomyomas induced in rats by beta 2-receptor stimulants, carcinoid tumors in the rodent stomach induced by prolonged suppression of acid secretion, thyroid follicular cell tumors in rodents, canine mammary neoplasia due to administration of progestagens, rodent mammary neoplasia induced by estrogens, uterine endometrial carcinomas of rats induced by dopamine agonists, Leydig cell tumors in the testis of rats, and ovarian tubulostromal adenomas in mice. A positive result on a rodent carcinogenicity study should not automatically preclude further development of a compound; future progress in this field should increase the accuracy of the rodent carcinogenicity study as a tool in human safety assessment.

A putative new proteinous factor negative for stromal growth. Purification and identification from endometrial carcinoma extract

Uterine endometrial carcinoma has been reported to synthesize and secrete some putative mitogens that elicit either a positive or negative proliferation response in endometrial fibroblasts. The purposes of this study were to isolate and to identify the negative growth factor(s) from endometrial carcinoma extract. The factor was isolated by a sequence of molecular size exclusion filtration, anion exchange chromatography, gel filtration and Affi-Gel Blue chromatography, followed by NH2-terminal amino acid sequencing. Mitogenicity was determined by [3H]thymidine incorporation into the endometrial fibroblasts. The purification procedure yielded a single active protein band (68 kDa). The protein, purified approximately 20,000-fold, evoked 90% inhibition of [3H]-thymidine incorporation into endometrial fibroblasts in the nanomolar range. This potent growth inhibitor is a previously unidentified protein molecule as revealed by amino acid sequences. Endometrial carcinoma could produce a new protein that may act as a paracrine factor to suppress the growth of its stroma endometrial fibroblasts.

Possible evidence for estrone-specific binding sites in human uterine endometrial carcinoma.

Estrone (E1), a principal estrogen in postmenopausal women, may have profound consequences in the maintenance and progression of hormone-sensitive endometrial carcinoma. This study was designed to investigate the specific binding sites for E1 in the tumor and in the normal endometrium. The binding of [3H]E1 to the cytosolic fraction and KCl-soluble nuclear fraction from 3 endometrial carcinomas showed saturation kinetics with an apparent equilibrium dissociation constant of approximately 37 and 50 nM, respectively. The specific binding site of E1 was also found in 3 normal endometria. However, the binding capacity in the endometrial carcinoma increased to 1.5-fold of that in the normal endometrium. E2 did not affect [3H]E1 binding to any subcellular fraction from endometrial carcinoma specimens. These findings demonstrate the presence of specific binding sites for E1 in human endometrial carcinoma. The endometrial carcinoma growth may be mediated, at least in part, by E1 and its binding site interaction.

Prolactin binds to human endometrial fibroblasts and inhibits mitogenicity of an endometrial carcinoma extract.

Uterine endometrial carcinoma has been reported to synthesize and secrete a putative peptide mitogen that elicits a potent proliferative response in endometrial fibroblasts. The extract from endometrial carcinoma stimulated [3H]thymidine incorporation into human endometrial fibroblasts in a dose-dependent manner. Concomitant exposure of the fibroblasts to prolactin (PRL) led to a remarkable inhibition of the extract-stimulated mitogenic activity of the fibroblasts. This inhibition was dependent on PRL dose, and maximal effect occurred at 1 microM of PRL. PRL (10 nM) suppressed an apparent maximal activity of the extract by 50%, and the half-maximal stimulated effect of the extract on thymidine incorporation was observed at the same concentration in the absence or presence of PRL. This noncompetitive manner may imply that PRL acts at a stage after the interaction of the mitogen in the extract with the specific receptor for mitogen. When the fibroblasts were first exposed to the extract for 12 hr and then to PRL, PRL suppressed the mitogenic activity with no lag. The rapid growth-inhibitory effect of PRL was mimicked by prostaglandin E1, but the combination of both types of ligand was not additive in the inhibitory action on growth. PRL and prostaglandin E1 may inhibit a similar mitogenic signaling cascade. Specific receptor sites for PRL were detected in the endometrial fibroblasts, showing high binding affinity (Kd = 16.1 nM) and low binding capacity (Bmax = 1.59 pmol/mg protein). Treatment of the fibroblasts with the endometrial carcinoma extract induced no changes in the level of PRL receptor, excluding the possibility that PRL competes for the binding sites with the mitogen. These findings would suggest that PRL may block the mitogenic activity of the fibroblasts stimulated by the endometrial carcinoma-derived mitogen via a PRL receptor-mediated mechanism, perhaps prostaglandin production.

Endometrioid Carcinoma of the Prostate. The Diagnostic Value of Leu7 and Prostatic Specific Antigen

In a case of endometrioid carcinoma of the prostate the expression of Leu7 and prostatic specific antigen (PSA) was studied immunohistochemically on paraffin sections. The same markers were studied in 14 cases of prostatic cancer and 14 cases of uterine endometrial carcinoma. Endometrioid carcinoma of the prostate was negative for Leu7 and PSA, while all prostatic cancer specimens were positive for the above markers. All uterine endometrial carcinomas were negative for PSA and only a small percentage of cells in 4 specimens were positive for Leu7. The absence of Leu7 and PSA in this case of endometrioid carcinoma strongly suggests a müllerian origin.

A study of cis-diamminedichloroplatinum(II) suppositories for the treatment of rabbit uterine endometrial carcinoma.

cis-Diamminedichloroplatinum (II) (cisplatin: CDDP) suppositories containing NaCl at different concentrations were prepared as a local chemotherapeutic agent for the treatment of uterine endometrial carcinoma and were administered to rabbits implanted with uterine VX2 tumor. The intrauterine CDDP histological level, as well as the antitumor effects and side effects of the suppositories to the liver and kidney were studied. The results showed high intrauterine tissue CDDP level in all suppository administrations. In particular, the NaCl-added suppositories enhanced the intrauterine CDDP level. As for antitumor effects, while the tumor growth rate of the NaCl-added suppository group was likely to be suppressed, the suppositories could not suppress tumor growth completely. The plasma platinum (Pt) level was 1.5 micrograms/ml or less and that of the liver and kidney was as low as 0.31 to 0.48 micrograms/g. No difference in levels depending on NaCl concentration was observed, nor was any abnormality found in the biochemical analysis including glutamate oxaloacetate transaminase (GOT) and blood urea nitrogen (BUN). Histopathological study revealed the degeneration of tumor cells in the NaCl-added suppository group. Minimal congestion and hemorrhage were observed in the endometria, possibly resulting from CDDP. By adding NaCl to CDDP suppositories, the uterine CDDP level and antitumor effects increased while no serious renal dysfunction was noted. Therefore, we conclude that NaCl-added CDDP suppositories are a useful local chemotherapy for endometrial carcinoma.

Gonadotropin-releasing hormone stimulates phospholipase C but not protein phosphorylation/dephosphorylation in plasma membrane from human epithelial ovarian cancer.

In view of advances in treatment of certain hormone-dependent cancers with analogues of gonadotropin-releasing hormone (Gn-RH), this study was undertaken to establish the signal transduction events interacting with Gn-RH receptor in a cell-free system prepared from human ovarian mucinous cystadenocarcinoma samples. A high affinity specific binding (Kd=8 x 10-9 M) of [3H] Gn-RH was demonstrated in two from two plasma membrane preparations. Gn-RH showed no effects on the rate of protein phosphorylation from [gamma-32P] adenosine triphosphate in the plasma membrane preparations. On the other hand, incubation of plasma membrane isolated form [3H]inositol-labeled specimens with Gn-RH in the presence of guanosine thiotriphosphate resulted in the rapid production of inositol phosphates. The Gn-RH effects was concentration-dependent, and half-maximal activation occurred with 1-3 nm Gn-RH. The Gn-RH-stimulated membrane event was observed in all plasma membrane isolations tested, but not in those from uterine endometrial carcinoma of a given case. These results provide the first direct evidence that Gn-RH receptor is coupled to phosphoinositide hydrolysis but not to certain membrane protein phosphorylation/dephosphorylation in ovarian carcinoma plasma membrane. Though the functional role of this event in human ovarian cancer is still obscure, it might be part of a possible point of attack for therapeutic approaches using Gn-RH analogues in this malignancy.

Brief Report: Lessons Learned from Studies on Tumor Suppression by Microcell-Mediated Chromosome Transfer

One approach for identifying chromosomes that carry putative tumor-suppressor genes is the introduction of individual, normal human chromosomes to the tumor cells of interest (1-4). In order to identify human chromosomes that can suppress or modulate tumor-associated phentotypes of tumor cells, we performed chromosome transfer experiments via microcell fusion into cell lines derived from a variety of tumors: neuroblastoma (SK-N-MC), fibrosarcoma (HT1080), uterine endometrial carcinoma (HHUA), renal cell carcinoma (YCR), choriocarcinoma (CC1), uterine cervical carcinoma (SiHa), rhabdomyosarcoma (A204), Wilms' tumor (SK-NEP-1), and Kirsten sarcoma virus-transformed NIH 3T3 cells (DT). We first isolated mouse A9 cells containing a single, normal human fibroblast-derived chromosome integrated with pSV2neo plasmid DNA (5). Following fusion of microcells from these A9 cells with tumor cells, we isolated microcell hybrids with the introduction of a specific chromosome and examined their tumorigenicity and in vitro properties (6-11). The following results were obtained (Table 1). The introduction of chromosome 11 suppressed tumorigenicity of HT1080, SiHa, A204, and SK-NEP-1, but not of SK-N-MC, YCR, CC1, and DT, indicating the function of the putative suppressor gene(s) on chromosome 11 is effective only in specific tumors. The tumorigenicity of YCR was modulated by the introduction of chromosome 3p, but not by chromosomes X and 11, supporting a hypothesis that loss and/or mutational inactivation of a gene on 3p may play a crucial role in the development of human renal cell carcinoma.

Cisplatin Suppository: Preparation, Release Characteristics and Clinical Evaluation.

Cisplatin (CDDP) has attracted attention as a chemotherapeutic agent for the treatment of uterine endometrial carcinoma but causes serious side effects, including renal toxicity. CDDP suppositories containing NaCl at different concentrations were prepared to enhance the efficacy and to reduce the side effects of CDDP. The release characteristics, melting point and viscosity of the suppositories were first studied. The rate of CDDP release increased as the NaCl concentration increased: it was 12% 12 h after administration of suppositories containing no NaCl, but 32% with 0.2% NaCl. The melting point was raised by addition of NaCl: 35.5 degrees C without NaCl and 36.5 degrees C with 0.2% NaCl. Addition of 0.2% NaCl doubled the viscosity. Clinically, the suppository containing 0.06% NaCl was given to 3 patients with endometrial carcinoma twice a week for 3 weeks to examine serum CDDP levels and endometrial absorption. Patients with endometrial carcinoma showed different peak plasma platinum (Pt) levels which were as low as 0.12, 0.06 and 0.22 micrograms Pt/ml with similar patterns of change in the level. Radiographic analysis revealed many Pt particles in sections of necrosed endometria after 21d of the treatment. No side effects of CDDP were found in biochemical testing or subjective symptoms.

Transfer of a normal human chromosome 11 suppresses tumorigenicity of some but not all tumor cell lines

The complete suppression of tumorigenicity of a human cervical cancer cell (HeLa) and a Wilms' tumor cell line (G401) following the introduction via microcell fusion of a single chromosome t(X;11) has been demonstrated by Stanbridge and co-workers. To determine whether other tumor cell lines are suppressed by chromosome 11, we performed chromosome transfer experiments via microcell fusion into various human tumor cell lines, including a uterine cervical carcinoma (SiHa), a rhabdomyosarcoma (A204), a uterine endometrial carcinoma (HHUA), a renal cell carcinoma (YCR-1), and a rat ENU-induced nephroblastoma (ENU-T1). We first isolated a mouse A9 cell containing a single human chromosome 11 with integrated pSV2-neo plasmid DNA. Following microcell fusion of the neo-marked chromosome 11 with the various tumors mentioned above, we isolated clones that were resistant to G418 and performed karyotypic analyses and chromosomal in situ hybridization to ensure the transfer of the marked chromosome. Whereas the parental cells of each cell line were highly tumorigenic, SiHa and A204 microcell hybrid clones at early passages were nontumorigenic in nude mice and HHUA was moderately tumorigenic. On the other hand, YCR-1 and ENU-T1 microcell hybrid clones were still highly tumorigenic following the introduction of chromosome 11. Thus, the introduction of a normal chromosome 11 suppresses the tumorigenicity of some but not all tumors, suggesting that the function of the putative suppressor gene(s) on chromosome 11 is effective only in specific tumors.

Monoclonal antibody-defined antigen of human uterine endometrial carcinomas is Leb.

A monoclonal antibody, MSN-1, generated by immunizing a mouse with a human uterine endometrial adenocarcinoma cell line, SNG-II, was strongly and specifically reactive with neutral glycosphingolipids from cancer tissues of patients with uterine endometrial adenocarcinomas. The glycosphingolipid antigen was purified from pooled human meconia, which contained the antigen at the concentration of 1.95 mumol/g dry weight. Its structure was determined by NMR, negative ion FABMS, permethylation analysis, and TLC-immunostaining with monoclonal anti-Lc4Cer antibody, and was concluded to be the III4IV2Fuc2Lc4Cer,Leb antigen of the human Lewis blood system. On ELISA, the monoclonal antibody was found to be strongly reactive with Leb, slightly with Lea and not at all with A, B, H, or IV2FucGg4Cer. The amount of Leb in cancerous regions in the patients with the Lea-b+ blood group was significantly increased compared to that in normal regions in the same patients, and it was a newly appearing antigen in the cancerous tissue of a patient with the Lea+b- blood group.

Adenocarcinoma of the prostate with endometrioid features. A light microscopic and immunohistochemical study of ten cases

The authors reviewed the histologic slides of 2600 prostatic carcinomas seen at Memorial Hospital from 1963 to 1983. In ten cases, resection specimens had a predominantly endometrioid appearance. Six patients had polypoid lesions in and around the verumontanum, and one had a polypoid lesion away from the verumontanum. Two patients had no mucosal lesions and one was not cystoscoped. Histologically, the tumors showed a tall pseudostratified columnar epithelium, usually with amphophilic cytoplasm. The cells were arranged either along papillae or in complexes of large acini or in single glands. In eight of the ten cases, the endometrioid carcinomas were associated with a prior or coexistent typical microacinar prostatic adenocarcinoma. In four cases, the endometrioid pattern existed in a pure form, although in two such cases with urethral tumors, the patients had histories of successfully treated microacinar adenocarcinomas of the posterior prostatic lobe. In one case, a urethral endometrioid tumor coexisted with a small posterior lobe microacinar adenocarcinoma. In five cases, both endometrioid and microacinar carcinomas were seen, including endometrioid and microacinar carcinomas found at the same site at different times (2 cases), tumors with a predominantly endometrioid, yet focally microacinar pattern (1 case), and primary tumors where lymph node metastases had different histologic features (2 cases). Of the three patients with a pure or predominantly endometrioid pattern treated with diethylstilbestrol, two had a marked clinical response. All ten endometrioid prostatic adenocarcinomas showed prostate-specific antigen and prostate-specific acid phosphatase immunoreactivity, in contrast to none of the control uterine endometrial carcinomas. In material spanning a 20-year period, the authors have not seen a single prostatic tumor entirely analogous to the uterine endometrial carcinoma. Until such proof exists, prostatic carcinomas with endometrioid features are best classified and treated as variants of prostatic duct carcinomas.

Prostatic adenocarcinoma with endometrioid features. Clinical, pathologic, and ultrastructural findings.

Thirteen cases of prostatic adenocarcinoma with endometrioid features were reviewed. The patients were older men (49-81 years) presenting with symptoms of hematuria and urinary obstruction. Each of the tumors displayed exophytic growth into the prostatic urethra, with involvement of the verumontanum. The urethral orifices of the large (primary) prostatic ducts were uniformly involved, and coexistent invasive (acinar) adenocarcinoma was identified in 10 cases (77%). The tumors exhibited a complex glandular pattern strikingly similar to uterine endometrial carcinoma, with prominent papillary formation in six cases. All cases demonstrated intense cytoplasmic immunoreactivity for prostatic acid phosphatase and prostate-specific antigen in at least part of the tumor. Focal staining for carcinoembryonic antigen was seen in three cases. Five tumors examined ultrastructurally demonstrated typical features of prostatic adenocarcinoma. Follow-up information was available on all 13 patients (6-83 months). Seven patients died of metastatic tumor (9-70 months after diagnosis), and the other six patients exhibited recurrent local or metastatic tumor. The sites of metastases were identical to those seen with invasive "acinar" prostatic adenocarcinoma, including pelvic lymph nodes, bones, and lungs. Crude 5-year survival was 15%, with a mean survival of 37 months. Adjuvant therapy provided palliative relief for many patients, but did not appear to influence survival. These findings indicate that endometrioid carcinoma is a histologically distinct variant of prostatic adenocarcinoma, with a more aggressive clinical behavior than previously thought.