Uterine papillary serous and clear cell carcinomas predict for poorer survival compared to grade 3 endometrioid corpus cancers.

To compare the survival of women with uterine papillary serous carcinoma (UPSC) and clear cell carcinoma (CC) to those with grade 3 endometrioid uterine carcinoma (G3EC). Demographic, pathologic, treatment, and survival information were obtained from the Surveillance, Epidemiology, and End Results Program from 1988 to 2001. Data were analysed using Kaplan-Meier and Cox proportional hazards regression methods. Of 4180 women, 1473 had UPSC, 391 had CC, and 2316 had G3EC cancers. Uterine papillary serous carcinoma and CC patients were older (median age: 70 years and 68 vs 66 years, respectively; P<0.0001) and more likely to be black compared to G3EC (15 and 12% vs 7%; P<0.0001). A higher proportion of UPSC and CC patients had stage III-IV disease compared to G3EC patients (52 and 36% vs 29%; P<0.0001). Uterine papillary serous carcinoma, CC and G3EC patients represent 10, 3, and 15% of endometrial cancers but account for 39, 8, and 27% of cancer deaths, respectively. The 5-year disease-specific survivals for women with UPSC, CC and G3EC were 55, 68, and 77%, respectively (P<0.0001). The survival differences between UPSC, CC and G3EC persist after controlling for stage I-II (74, 82, and 86%; P<0.0001) and stage III-IV disease (33, 40, and 54; P<0.0001). On multivariate analysis, more favourable histology (G3EC), younger age, and earlier stage were independent predictors of improved survival. Women with UPSC and CC of the uterus have a significantly poorer prognosis compared to those with G3EC. These findings should be considered in the counselling, treating and designing of future trials for these high-risk patients.

Histological classification of gastric adenocarcinoma for epidemiological research: concordance between pathologists.

Epidemiology of gastric adenocarcinoma suggests that intestinal-type and diffuse-type cancers develop through distinct causal pathways. To examine the differences in risk factors and molecular changes between the histological types, reliable data on histological typing are essential. We evaluated the concordance between two pathologists in assessment of 95 gastric adenocarcinomas for Laurén classification and tumor grade. Two pathologists, each blinded to the other's assessment, reviewed H&E-stained slides of gastric tumor. The responses of the two pathologists for histological type were considered as concordant if they fell on one of the three categories (intestinal type, diffuse type, or other). Tumor grade was classified into three categories (well, moderately, or poorly differentiated). The pathologists agreed on the classification of histological type for 71 of 92 (77%) tumors. Kappa coefficient was 0.59 (95% confidence interval, 0.44-0.73). Concordance for tumor grade was 87%, with a kappa coefficient of 0.72 (95% confidence interval, 0.57-0.87). Both observed concordance and kappa coefficient for histological type and tumor grade were similar across three calendar periods of study. Interobserver agreement was virtually identical between tumors with biopsy specimens only and those with surgical specimens. Although the level of disagreement for histological type observed in this study is comparable with that in other studies, the resulting misclassification would lead to the reduction in observed differences in prevalence and odds ratio estimates between two histological types.

Malignant granular cell tumor of the vulva in a 17-year-old: Case report and literature review.

Granular cell tumors are uncommon soft tissue tumors. Although the majority of these tumors are benign, a rare malignant variant exists which is aggressive, with local recurrence rates up to 70% and 3-year survival rates of less than 50%. We present a case of malignant granular cell tumor of the vulva in a 17-year-old, the sixth such case to be reported at this site. She was treated with a left hemivulvectomy and ipsilateral groin node dissection followed by postoperative radiation therapy. She remains free of disease at 16 months. Patients with malignant granular cell tumor or granular cell tumor of malignant potential are best managed with wide local excision and regional lymph node dissection.

Inhibin and CD99 (MIC2) expression in uterine stromal neoplasms with sex-cord-like elements.

Uterine mesenchymal neoplasms with sex-cord-like elements are designated as endometrial stromal tumor with sex-cord-like elements (ESTSCLE) or uterine tumor resembling ovarian sex-cord tumor (UTROSCT), depending on the extent of sex-cord-like differentiation. Occasionally, sex-cord elements similar to those in ESTSCLE and UTROSCT occur in uterine adenosarcomas. To determine whether the sex-cord-like elements in these tumors show immunohistological evidence of sex-cord differentiation, we studied a series of uterine neoplasms for expression of inhibin, a peptide hormone expressed by normal ovarian granulosa cells and ovarian sex-cord neoplasms, and CD99, a protein also expressed by granulosa cells, Sertoli cells, and some ovarian sex-cord tumors. Thirty uterine mesenchymal neoplasms (five epithelioid or plexiform smooth muscle tumors, three endometrial stromal tumors, two mixed endometrial stromal and smooth muscle tumors, 10 ESTSCLE, five UTROSCT, and five miscellaneous stromal processes) and five epithelial neoplasms were evaluated for expression of CD99 (clone 12E7) and inhibin (clone R1) in formalin-fixed, paraffin-embedded tissue. Three of 10 (30%) ESTSCLE and five of five (100%) UTROSCT were inhibin and CD99 immunoreactive. Inhibin staining was confined to the areas with sex-cord-like differentiation, and staining was generally much stronger and more extensive in areas featuring prominent foam cells. There were no differences in the degree or intensity of staining for inhibin in premenopausal and postmenopausal women. CD99 expression tended to correlate with inhibin and was typically confined to similar cell types in the individual neoplasms. Weak CD99 immunoreactivity was seen in one additional epithelioid smooth muscle tumor, whereas all other mesenchymal and epithelial neoplasms studied for inhibin and CD99 were negative. These results provide further immunohistological support for true sex-cord differentiation within uterine mesenchymal proliferations and suggest that the degree of sex-cord differentiation may correlate with the expression of these markers.

Diffusely infiltrative endometrial adenocarcinoma: an adenoma malignum pattern of myoinvasion.

The prognostic significance of a diffusely infiltrative intramyometrial growth pattern was evaluated in 110 cases of low-stage (stages I and II) endometrial adenocarcinoma. Fifty cases were associated with diffuse infiltration (DI group), and 50 cases had more conventional granulation tissue type intramyometrial infiltration (GTT group). Ten cases with carcinomatous involvement of deeply situated adenomyosis (ADMY group) were also studied. The diffusely infiltrative "adenoma malignum" growth pattern featured typically round, regular individual glands, clearly within myometrium but with minimal or absent stromal or inflammatory cell response. Myoinvasion of the conventional sort was characterized by irregular, sharply angulated abnormal glands within myometrium without interposed normal glands or endometrial stroma. The abnormal glands were surrounded, at least focally, by edematous stroma with granulation tissue type reaction and/or an inflammatory cell infiltrate. Mean follow-up was 77.8 months (range 3-219 months) for the patients with diffusely infiltrative myoinvasion and deep adenomyosis and 86.9 months (range 1-206 months) for the patients with conventional myoinvasion. Recurrence-free survival for patients with stage I disease and conventional myoinvasion (94%) was similar to that of patients with diffuse adenoma malignum infiltration (98%; p = 0.13). Survival rates for both groups were also similar. Two (4%) of the 50 patients with diffusely infiltrative adenoma malignum pattern of myoinvasion died of endometrial carcinoma 36 and 72 months after hysterectomy, and 2 (4%) of the 50 patients with conventional myoinvasion died 34 and 67 months after hysterectomy (p = 0.41). Survival in these patients correlated with depth of myometrial invasion and stage. There were no recurrences in the patients with deep adenomyosis. These results suggest that although endometrial carcinomas with diffuse myometrial infiltration are fully capable of aggressive clinical behavior, they do not appear to behave any more aggressively than those with conventional myometrial invasion. Prognostic indicators of clinically aggressive disease are similar to those that have been previously identified for endometrial carcinomas with the more conventional pattern of myometrial infiltration. They include cervical involvement, deep myometrial invasion, higher histologic grade, and lymph-vascular space invasion. Endometrial carcinomas with extensive involvement of adenomyosis and adjacent foci of minimal myometrial infiltration appear to have very low malignant potential, but the number of cases with this finding and adequate clinical follow-up is limited. This finding needs to be confirmed in a much larger series of cases.

Chemo-radiotherapy for localized pancreatic cancer: increased dose intensity and reduced acute toxicity with concomitant radiotherapy and protracted venous infusion 5-fluorouracil.

PURPOSE: Although concomitant radiation therapy (RT) and bolus 5-Fluorouracil (5-FU) have been shown to improve survival in locally confined pancreatic cancer, most patients will eventually succumb to their disease. Since 1994, we have attempted to improve efficacy by administering 5-FU as a protracted venous infusion (PVI). This study compares treatment intensity and acute toxicity of consecutive protocols of concurrent RT and 5-FU by bolus injection or PVI. METHODS AND MATERIALS: Since 1986, 74 patients with resected or locally advanced pancreatic cancer were treated with continuous course RT and concurrent 5-FU by bolus injection (n = 44) or PVI throughout the course of RT (n = 30). Dose intensity was assessed for both 5-FU and radiotherapy. Toxicity endpoints which could be reliably and objectively quantified (e.g., neutropenia, weight loss, treatment interruption) were evaluated. RESULTS: Cumulative 5-FU dose (mean = 7.2 vs. 2.5 gm/m2, p < 0.001) and weekly 5-FU dose (mean = 1.3 vs. 0.5 gm/m2/wk, p < 0.001) were significantly higher for patients receiving PVI 5-FU. Following pancreaticoduodenectomy, 95% of PVI patients maintained a RT dose intensity of > or = 900 cGy/wk, compared with 63% of those receiving bolus 5-FU (p = 0.02). No difference was seen for patients with locally advanced disease (72% vs. 76%, p = n.s.). Grade II-III neutropenia was less common for patients treated with PVI (13% vs. 34%, p = 0.05). Grade II-III thrombocytopenia was uncommon (< or = 3%) in both treatment groups. Mean percent weight loss (3.8% vs. 4.1%, p = n.s.) and weight loss > or = 5% of pre-treatment weight (21% vs. 31%, p = n.s.) were similar for PVI and bolus treatment groups, respectively. Treatment interruptions for hematologic, gastrointestinal or other acute toxicities were less common for patients receiving PVI 5-FU (10% vs. 25%, p = 0.11). CONCLUSION: Concurrent RT and 5-FU by PVI was well tolerated and permitted greater chemotherapy and radiotherapy dose intensity with reduced hematologic toxicity and fewer treatment interruptions compared with RT and bolus 5-FU. Longer follow-up will be needed to assess late effects and the impact on overall survival.

Value of inhibin in the identification of granulosa cell tumors of the ovary.

Inhibins are peptide hormones that participate in the regulation of the pituitary-gonadal feedback system and are selectively expressed by cells of sex cord-stromal derivation. To determine the efficacy of this marker for distinguishing granulosa cell tumors, 134 primary and metastatic lesions of the ovary were evaluated for expression of the alpha-subunit of inhibin in routinely processed formalin-fixed, paraffin-embedded tissue. A variety of sex cord-stromal tumors (SCST), including 35 adult and juvenile granulosa cell tumors, 14 fibroma-thecomas, and 18 other sex cord-stromal proliferations, were studied. In addition, 33 surface epithelial neoplasms, 12 germ cell tumors, 11 metastases, and 11 miscellaneous ovarian neoplastic proliferations were evaluated. Among the non-granulosa cell neoplasms, special emphasis was placed on primary neoplasms and metastases that histologically simulated granulosa cell tumors. Thirty-three of 35 (94%) granulosa cell tumors were immunoreactive compared with 2 of 12 (17%) primary ovarian endometrioid tumors, one of nine (11%) primary ovarian transitional cell (Brenner) proliferations, and 3 of 17 (18%) other primary and metastatic poorly differentiated (undifferentiated) carcinomas. In 31 of the 35 granulosa cell tumors, inhibin staining was of moderate to strong intensity or was present in at least half of the constituent cells, whereas only 2 of 33 primary surface epithelial neoplasms fulfilled the same criteria, showing weak staining of 70% to 80% of the cells. In contrast, 10 of 14 (71%) ovarian fibroma-thecomas and 17 of 18 (94%) other sex cord-stromal proliferations were positive for inhibin. Nonneoplastic luteinized stromal cells stained for inhibin in 29 of 85 cases in which they could be evaluated. The results of this study show that although it is not completely specific and cannot reliably distinguish granulosa cell tumors from fibroma-thecomas or other ovarian sex cord-stromal proliferations, inhibin can be used to help distinguish sex cord-stromal neoplasms from most primary and metastatic non-SCST. Caution should be exercised in the interpretation of inhibin-positive cells, because a wide variety of primary and metastatic ovarian tumors may contain significant numbers of positively staining luteinized cells.

Lymphangioleiomyomatosis of the uterus simulating high-stage endometrial stromal sarcoma.

Symptomatic uterine lymphangioleiomyomatosis (LAM) simulating high-stage uterine sarcoma in a patient with tuberous sclerosis complex is reported. A 49-year-old female presented with abdominal pain and anemia. Preoperative workup revealed a uterine mass and a large amount of peritoneal free fluid and possible metastatic implant along the lateral edge of the liver. The patient also had a large right pleural effusion. A fungating anterior uterine fundal mass with apparent perforation and intraabdominal hemorrhage was found on laparotomy. A portion of the mass was excised and initially interpreted as an endometrial stromal sarcoma. Microscopic examination revealed multiple vascular epithelioid smooth muscle proliferations in the uterus and serosal surface of the fallopian tube and periaortic lymph node lymphangioleiomyomas. The uterine, fallopian tube, and nodal lesions were positive for smooth muscle actin, desmin, and HMB-45, findings characteristic of LAM. Additional examination of the patient revealed stigmata of tuberous sclerosis complex. Although uterine LAM is uncommon, it may be associated with pelvic and/or abdominal symptoms and may simulate a primary uterine mesenchymal neoplasm.

Desmoplastic and spindle-cell malignant melanoma. An immunohistochemical study.

The clinical, histologic, and immunohistologic features of 22 desmoplastic melanomas (DMM), 10 mixed desmoplastic and spindle-cell melanomas (DMM/SMM), and two cellular spindle-cell melanomas (SMM) were studied. Patients ranged in age from 35 to 91 years (mean, 67) and included 23 men and 11 women. Seventeen cases occurred in sun-damaged skin of the head and neck. 11 were on the extremities, and six on the trunk. Except for two cases, all were Clark's level IV or V. Twenty-two (65%) cases were associated with a recognizable overlying pigmented lesion. Thirty of 32 (94%) DMM and DMM/SMM were clearly positive for S100. S100 staining was limited to < 5% of the spindle cells in two DMM/SMM. All DMM were negative when stained with HMB45. Three DMM/ SMM were immunoreactive with HMB45, as were both SMM. CD68 staining was limited to < 5% of the spindle cells in two of 32 DMM and DMM/SMM and 20% of the cells in one of two SMM. Nine (32%) DMM and DMM/SMM contained significant numbers of spindle cells immunoreactive for SMA but not desmin. In five cases, the number of actin-positive spindle cells. Two color stains for SMA and S100 demonstrated that these smooth-muscle actin positive cells constituted a separate spindle-cell population, consistent with reactive myofibroblasts. This study indicates that the immunohistologic features of desmoplastic melanoma differ from those of conventional melanoma. If a problematic spindle-cell skin lesion is a suspected melanocytic process, HMB45 is unlikely to provide confirmatory (or exclusionary) evidence for the diagnosis of DMM. Similarly, because of the variability in S100 expression in this neoplasm, the absence of S100 staining should not be relied on too heavily to exclude DMM if the clinical and histologic features favor that diagnosis. Caution should be exercised in the interpretation of numerous actin-positive spindle cells in isolation of additional confirmatory or exclusionary data as desmoplastic melanomas may contain significant numbers of these cells.

Adenoid cystic carcinoma of the submandibular gland with symptomatic ovarian metastases.

We report the clinical and pathologic features of an adenoid cystic carcinoma of the submandibular gland that metastasized to the ovaries 10 years after initial presentation. A 30-year-old woman underwent excision of a right submandibular adenoid cystic carcinoma followed by regional external beam radiation therapy. Three years later, she underwent extended hepatic resection and localized radiotherapy to the hepatic region for metastatic disease. The patient was without evidence of disease for 7 years when she developed pelvic pain and a pelvic mass was found. A solid and cystic 10-cm left ovarian mass and a single metastatic tumor nodule involving the right ovary were excised via the laparoscope. Histologically, the tumor was identical to the patient's initial salivary gland neoplasm. The neoplastic cells were CAM 5.2 positive, S100 positive, muscle-specific actin positive, and smooth muscle actin positive. Ultrastructurally, characteristic pseudocysts (pseudolumina) with abundant basal lamina and true glandular lumina lined by short microvilli were present. Other than a single anecdotal account of a parotid gland adenoid cystic carcinoma, this case represents the first documented report of an adenoid cystic carcinoma of salivary gland origin that was associated with symptomatic ovarian metastases. This case demonstrates that the ovary is a potential site for metastatic disease many years following the diagnosis and treatment for a primary neoplasm however uncommon or remote the site of origin. Since metastatic adenoid cystic carcinoma can rarely present as an ovarian mass, a clinical history of this neoplasm should be heavily weighed in the differential diagnosis of any unusual ovarian tumor with a predominant cribriform, trabecular, or tubular pattern.

Atypical polypoid adenomyofibroma (APA) versus well-differentiated endometrial carcinoma with prominent stromal matrix: an immunohistochemical study.

It can be extremely difficult in some cases to distinguish atypical polypoid adenomyomas (APAs) from invasive adenocarcinoma in an endometrial curettage or biopsy specimen. In order to determine if immunophenotypic features can be exploited to differentiate between these two entities in problematic cases, a series of APAs and myoinvasive well-differentiated endometrial carcinomas (WDCAs) were studied with a panel of standard immunohistochemical markers. All 23 APAs had stromal smooth muscle actin (SMA) reactivity, 12 of 23 had variable degrees of stromal desmin reactivity, and nine of 22 had CD34-positive stromal cells. All epithelial components of the APAs were cytokeratin (AE1 and CAM5.2) positive, whereas 22 of 23 were positive for estrogen receptor (ER) and progesterone receptor (PR). Among the 10 myoinvasive WDCAs, all contained at least some SMA-positive stromal cells, seven of 10 desmin-positive stromal cells, and four of eight CD34-positive stromal cells. All carcinomas studied demonstrated CAM5.2 and PR-positive epithelia; nine of 10 were ER positive. We conclude that the immunophenotype of APAs does not differ significantly from well-differentiated endometrial adenocarcinoma and that immunophenotyping is of little value in distinguishing APA from carcinoma. Because the stroma in APAs histologically and immunophenotypically more closely resembles a hybrid myofibromatous stroma, we prefer to refer to these lesions with the modified designation "atypical polypoid adenomyofibroma," although "APA" may be retained for clinical use.

Transitional cell neoplasms of the ovary and urinary bladder: a comparative immunohistochemical analysis.

Walthard cell nests, the Brenner tumor (benign, proliferating, low malignant potential, and malignant), and primary ovarian transitional cell carcinoma are considered to be primary female genital tract proliferations of transitional-type (urothelial) epithelium on conventional light microscopic grounds. In order to further investigate the similarities (or dissimilarities) of proliferations of female genital tract transitional epithelium and urothelium, we compared transitional cell proliferations (TCPs) of the female genital tract (n = 25) and urinary bladder (n = 15) using antibodies to carcinoembryonic antigen (CEA; clone 0062), carbohydrate determinant 19-9 (CA19-9; clone 1116-NS-19-9), cytokeratin 7 (CK-7; clone OV-TL 12/30), and cytokeratin 20 (CK-20; clone Ks 20.8), four monoclonal antibodies that have been shown to stain transitional cell urothelial proliferations. Both groups of tumors exhibited significant staining for CEA, CA19-9, and CK-7, and the difference in numbers of cases staining was not significant. CA19-9 was present in 15 of 25 female genital tract TCPs as compared with 12 of 15 bladder TCPs; CEA was present in 17 of 25 female genital tract TCPs and nine of 15 comparable bladder TCPs. CK-7 was present in all cases studied with the exception of one Walthard cell nest and a malignant Brenner tumor that was not immunoreactive with the other antibodies tested. In contrast, 13 of 15 bladder TCPs were CK-20 positive, whereas only one of 25 female genital tract TCPs was positive (< 5% of cells). Walthard cell nests and benign Brenner tumors were more likely to be CA19-9 positive than were Brenner tumors of low malignant potential, malignant Brenner tumors, and primary transitional cell carcinoma of the ovary. We conclude that despite their apparent morphologic and immunologic similarity to TCPs of the urinary bladder (particularly at the histologically low-grade end of the transitional cells spectrum), Walthard cell nests and ovarian Brenner tumors constitute an immunophenotypically distinct form of TCP.

Atypical polypoid adenomyofibromas (atypical polypoid adenomyomas) of the uterus. A clinicopathologic study of 55 cases.

We present the clinicopathological and immunohistochemical features of 55 atypical polypoid adenomyofibromas, a definitional expansion of an entity previously reported as "atypical polypoid adenomyoma" (APA) of the uterus. Patients ranged in age from 25 to 73 (mean, 39.9) years. All but two of the patients were premenopausal, and 14 were undergoing evaluation for infertility. Histologically, the lesions featured a biphasic proliferation of architecturally complex and cytologically atypical endometrial glands within a myofibromatous stroma. The histologic pattern ranged from widely separated and loosely clustered irregular but branched glands embedded in broad zones of cellular myofibromatous stroma to those possessing crowded, markedly complex, branching glands separated by sparse intersecting fascicles of fibromuscular tissue. The stroma in all cases was actin or desmin positive or both. Morular/squamous metaplasia was present in all but two cases and florid in most. All cases exhibited architecturally complex glands, and in 25 cases the architectural complexity was indistinguishable from that of well-differentiated endometrial adenocarcinoma, as we have defined it; that is, they had a high architectural index. Twenty-nine patients were initially treated with polypectomy or curettage followed by hormonal therapy; persistent or recurrent APA developed in 45% of the patients in this group (33% with low architectural index vs. 60% with high architectural index). Five patients had successful pregnancies despite persistent disease. Superficial myoinvasion was identified in the hysterectomy specimen in two of 12 APAs with a high architectural index but not in 21 APAs with a low architectural index. All patients are alive and well 1 to 112 months after diagnosis (mean, 25.2 months). On the basis of this study, we propose that APAs with markedly complex glands (high architectural index) be designated "atypical polypoid adenomyofibromas of low malignant potential" (APA-LMP) to emphasize the potential risk for myometrial invasion. A treatment program featuring local excision accompanied by close follow-up is warranted for APA despite the presence of recurrent or persistent disease. Patients with APA-LMP may also, in selected cases, be managed with less than hysterectomy, although (as with the usual well-differentiated carcinoma) there is a small but definite risk associated with this approach.

Estrogen and progesterone receptors and anti-gross cystic disease fluid protein 15 (BRST-2) fail to distinguish metastatic breast carcinoma from eccrine neoplasms.

Cutaneous metastases of breast carcinoma can be histologically similar to primary skin tumors with eccrine differentiation. We compared the immunohistochemical staining characteristics of 15 metastatic breast carcinoma skin lesions in 12 patients to those of a series of primary eccrine tumors using estrogen receptor, progesterone receptor, and anti-gross cystic disease fluid protein-15 markers. Anti-gross cystic disease fluid protein-15 positivity was noted in 7 of 15 breast carcinoma skin metastases, 0 of 5 benign eccrine tumors, 1 of 6 microcystic adnexal carcinomas, and 1 of 1 metastatic sweat gland adenocarcinoma. Estrogen receptor positivity was found in 1 of 15 metastatic breast carcinoma skin lesions, 0 of 5 benign eccrine tumors, 2 of 8 microcystic adnexal carcinomas, and 1 of 1 metastatic sweat gland adenocarcinoma. Progesterone receptor positivity was identified in 15 of 15 metastatic breast carcinoma skin lesions, 2 of 5 benign eccrine tumors, 5 of 8 microcystic adnexal carcinomas, and 1 of 1 metastatic sweat gland adenocarcinomas. These results indicate that standard immunohistochemical staining for estrogen receptors, progesterone receptors, and gross cystic fluid protein-15 markers will not reliably distinguish primary (or metastatic) eccrine tumors from cutaneous metastases of breast carcinoma.

Esophageal submucosal gland duct adenoma.

An 81-year-old man with a 3-year history of dysphagia underwent endoscopic resection of a 1-cm-diameter distal esophageal mass. Examination revealed a submucosal neoplasm with a circumscribed growth pattern composed of tubules, cysts, and papillae in association with a marked interstitial lymphoid infiltrate. The cyst lumens and papillae were lined by two to six layers of cytologically bland cuboidal to columnar cells with rare mitotic figures. The basal layer of cells was uniformly positive for smooth-muscle actin. Mucin-positive intracytoplasmic lumens were focally present, but cytoplasmic mucin was not seen. There was no evidence of Barrett's metaplastic epithelium. These features are similar to those in two, possibly three, previously reported cases of esophageal adenomas and bear a resemblance to sialadenoma papilliferum, a rare neoplasm of the minor salivary glands. Their clinicopathologic and immunohistologic features suggest that these neoplasms derive from the submucosal gland ducts. Comparison with the previously reported cases indicates that although the proportions of the various components (tubules, cysts, and papillae) may vary, all cases appear to pursue a slowly growing, clinically indolent course with no evidence of recurrence after complete resection.

Proposed criteria for the diagnosis of well-differentiated endometrial carcinoma. A diagnostic test for myoinvasion.

Existing criteria for separating clinically benign but architecturally complex or cytologically atypical endometrial proliferations (hyperplasia or metaplasia) from well-differentiated endometrial carcinoma are underspecified and poorly reproducible, in part due to the absence of a uniformly agreed on methodologically independent outcome against which to judge the efficacy of competing sets of criteria. Because myoinvasion is the first unambiguous indicator of clinically aggressive behavior for proliferations in this spectrum, we have employed the presence or absence of myoinvasion as a tool to develop clinically meaningful diagnostic criteria for the separation of complex atypical hyperplasia/metaplasia from well-differentiated carcinoma (CAHM/WDCA). We obtained the paired endometrial samplings and hysterectomy specimens of 520 patients; these were split into a training set of 306 cases and a test set of 214. The presence or absence of myoinvasion was assessed from an examination of the hysterectomy specimen. For the purposes of this study, myoinvasion was defined as the presence of irregular intramyometrial glands surrounded by a granulation tissue-like response. To determine the morphologic features that were most predictive of myoinvasion, a series of endometrial architectural, cytological, and stromal features was initially evaluated on the training set (149 myoinvasive and 157 nonmyoinvasive). Using a variety of exploratory data techniques including the classification algorithm CART, we developed a diagnostic rule for predicting myoinvasion that employed one architectural feature (glandular complexity captured by a pictorial architectural index) and two cytological features (nuclear pleomorphism and prominence of nucleoli). Extensive squamous differentiation, fibroblastic stroma, necrosis, stromal foam cells, and other cytologic features did not provide additional predictive value when cross-validated. The true misclassification rate of the CART-generated prediction rule was further assessed by applying the rule to the test set drawn largely from community hospitals. The sensitivity and specificity of this rule for detecting myoinvasion was 99.5 and 57%. The likelihood ratio was 2:1, (i.e., using prior odds of myoinvasion in the CAHM/WDCA spectrum of 1:10, the posterior odds on myoinvasion using the CART-generated rule would be 1:5). Comparison of the CART-generated myoinvasion prediction rule with the Kurman and Norris endometrial stromal invasion criteria for well differentiated endometrial carcinoma (25), using receiver operator characteristic curve (ROC) techniques, demonstrated a significant improvement in the ability to separate myoinvasive from non-myoinvasive endometrial proliferations with the CART-generated rule; the average area under the curve for the CART-generated rule was 0.78 (SE = 0.02) versus 0.67 (SE = 0.03) for the endometrial stromal invasion criteria.(ABSTRACT TRUNCATED AT 400 WORDS)