Challenges in the Interpretation of Peritoneal Cytologic Specimens

Peritoneal washing cytology has been used in the evaluation of gynecologic malignancies since the early 1950s. The International Federation of Gynecology and Obstetrics (FIGO) introduced peritoneal washing as part of the staging of ovarian carcinomas in 1975. For ovarian carcinomas, the presence of malignant cells in peritoneal washings is important for substaging patients with disease that is apparently organ-confined (FIGO stage I) and for the detection of recurrent carcinoma. Tumor confined to an ovary is considered FIGO stage IA, whereas ovarian carcinomas involving the ovarian surface and/or detectable in a peritoneal washing specimen is assigned FIGO stage IC. A 5-year survival rate of 90% in patients with stage IA, grade I ovarian carcinomas contrasts with 68% and 56% in patients with stage I lesions showing ovarian excrescences or ovarian capsular rupture, respectively. Similar results were found in second-look laparotomy patients.1 Survivals in patients with ovarian carcinoma have also been reported to differ significantly, based on peritoneal cytology, regardless of disease stage. The progression-free interval and survival rate are reportedly influenced significantly by tumor grade and the presence of microscopic tumor identified on the basis of biopsy or cytology.2–6 The presence of malignant cells in peritoneal washings from patients with endometrial carcinoma assigns the tumor to FIGO stage IIIA or higher. Most studies suggest that cytology is a predictor of poor outcome,378 but the clinical significance of positive peritoneal washings in patients with non–myometrial invasive, FIGO grade I endometrioid adenocarcinoma is debatable.8 The presence of malignant cells in washings is also an independent, negative prognostic factor in patients with advanced-stage disease. These patients have a reported recurrence rate of 23%, which is similar to that of patients with occult lymph node metastasis identified through surgical staging. Currently, patients with cytologically defined stage IIIA endometrial cancer are included among patients with advanced-stage disease,9 but this may be reassessed in an upcoming revision of the guidelines for FIGO staging of endometrial carcinoma. As stated previously, the presence of malignant cells in serous effusions signifies spread of disease beyond the organ of origin and is associated with significant therapeutic and prognostic implications. Unfortunately, the morphologic examination of cytologic specimens has not proven to be a sensitive or specific diagnostic tool. The malignant cells might be few in number and might be unrecognized among a large population of mesothelial cells and/or macrophages. Mesothelial cells, on the other hand, might show reactive changes due to a wide variety of stimuli. These changes consist of marked nuclear enlargement, hyperchromasia, and cytoplasmic alterations, which can be misinterpreted as malignant cells.10–12 Also, postradiation changes may lead to cellular changes that can mimic a malignancy. Additionally, benign conditions such as endosalpingiosis may be extremely difficult to differentiate from borderline ovarian tumors (tumors of low malignant potential) or low-grade serous carcinomas in cytologic specimens. Consequently, the false-positive rate of cytology in the evaluation of serous effusions has been reported to be as high as 4.5%. The main reason for a false-positive diagnosis is the misinterpretation of reactive mesothelial cells as malignant epithelial cells. Other potential pitfalls include the presence of psammoma bodies, endometriosis, and endosalpingiosis.413–15 The false-negative rate for cytology in serous effusions is believed to be approximately 20%. Factors that contribute to this relatively high false-negative rate include poor distribution of peritoneal washings, infrequent exfoliation of malignant cells, and interpretive errors. As a positive peritoneal washing diagnosis leads to treatment with chemotherapy in many instances, diagnostic accuracy is essential. CYTOLOGIC EVALUATION Serous carcinoma usually presents as cellular specimens containing single cells or poorly cohesive irregular cell clusters with large, pleomorphic nuclei and prominent nucleoli. Peritoneal washings involved by endometrioid carcinomas display loose, 3-dimensional clusters of cells with eccentric, pleomorphic nuclei and abundant delicate cytoplasm, coarse chromatin pattern, and prominent nucleoli. Clear cell carcinoma also demonstrates similar cytologic features. The main differential diagnosis of adenocarcinoma in a peritoneal washing is reactive mesothelial cells, but endosalpingiosis and endometriosis also frequently contribute to diagnostic difficulties. Reactive mesothelial cells usually present as clusters of epithelioid cells with occasional cell ball or papillary cluster formation. The reactive mesothelial cells show cellular enlargement, dense cytoplasm and large nuclei with increased nuclear to cytoplasmic ratio. Occasionally the cells might be vacuolated or contain prominent nucleoli. The presence of cellular “windows” might help to identify the cells as mesothelial. Cells from fallopian tubes in pelvic washings may also lead to false-positive diagnosis. They are usually present as a few small fragments of ciliated columnar epithelium in flat sheets. Endosalpingiosis might be nearly impossible to distinguish from well-differentiated serous neoplasms such as serous borderline tumor and low-grade serous carcinoma. Cases of endosalpingiosis display organized, tight clusters with occasional nonbranching papillary formation. The cells are uniform with scant basophilic cytoplasm. The nuclei have smooth nuclear membranes, a fine chromatin pattern, and small nucleoli. It is important to remember that psammoma bodies might be present in cases of endosalpingiosis. The distinction of endosalpingiosis from serous borderline tumor is based on the presence of large papillary clusters with architectural disorganization in the latter. These findings are better appreciated in cell blocks. The presence of nuclear molding and nucleoli in papillary clusters also suggest a neoplastic process. Endometriosis is another potential pitfall. It is characterized by the presence of round to oval cells arranged in 3-dimensional clusters, tubular structures, and sheets. The nuclei are round or bean-shaped with fine chromatin and rare nucleoli. The cytoplasm is scant and vacuolated. The most sensitive finding in endometriosis is the presence of hemosiderin-laden macrophages. The report format used in peritoneal washings from patients with serous borderline tumors can be problematic. If the patient has a frozen section diagnosis of a serous borderline tumor and the peritoneal washings show the presence of atypical cells, the pathologist might elect to release the report describing the presence of neoplastic cells along with the differential diagnosis, which includes serous borderline tumor and carcinoma. Alternatively, the pathologist could wait for evaluation of the surgical pathology specimen before releasing the final diagnosis. In the setting of a prophylactic salpingo-oophorectomy to reduce the risk of ovarian, tubal, and peritoneal serous carcinoma, it is probably best to finalize the cytology report after reviewing the surgical pathology slides. This reduces the chance of making a false-positive diagnosis. Endosalpingiosis and endometriosis can be reported as benign glandular cells. ANCILLARY STUDIES To improve the accuracy of cytology in peritoneal washings and serous effusions a series of ancillary studies have been reviewed. Most studies are based on immunocytochemical studies, although there is a limited number of studies using flow cytometry and molecular studies. Immunocytochemical studies can be performed in cytospins, ThinPrep preparations, and cell blocks.16 There is some controversy on what kind of preparation works the best. Fetsch and colleagues1718 suggested that cell blocks provided the best morphologic interpretation and immunocytochemical study. However, Mulvany4 reportedly obtained superior results with smear preparations. At Memorial Sloan-Kettering Cancer Center (New York, New York), ThinPrep slides are used for immunocytochemical studies in most fluid specimens requiring it. Several panels of antibodies have been proposed in the literature to distinguish cells of mesothelial origin from adenocarcinoma. The antibodies include markers of mesothelial and epithelial origin. Mesothelial Markers Mesothelial markers previously evaluated include D2-40, calretinin, mesothelin, cytokeratin 5/6, WT-1, and HBME-1. D2-40, a monoclonal antibody directed against oncofetal antigen M2A present on cell membranes, has been described as a useful marker of mesothelial cells. It has been shown to have a good sensitivity and specificity for distinguishing epithelioid malignant mesothelioma and reactive mesothelial cells from adenocarcinoma.19–24 Saad et al25 found that D2-40 has high specificity and good sensitivity for distinguishing epithelioid mesotheliomas from other primary non–small cell lung carcinomas in pleural fluid cytology. Calretinin, a calcium-binding protein widely expressed throughout the central and peripheral nervous systems, is also expressed by mesothelial cells in the cytoplasm and frequently in the nucleus. Calretinin expression has also been demonstrated in a small fraction of adenocarcinomas.1820–2224–40 WT-1 is a DNA-binding protein predominantly located in the nucleus that plays a critical role in the development of the genitourinary tract. In adult tissues, it is expressed by mesangial cells of the kidney, Sertoli cells of the testis, ovarian stromal cells and surface epithelium, mesothelial cells, and some other stromal cells in the female genital tract. WT-1 is also expressed by malignant mesotheliomas and by tumors derived from the ovarian surface epithelium. Although some studies have confirmed the specificity of WT-1 for mesothelioma, several others have demonstrated a low sensitivity. One of the problems concerning WT-1 in the distinction of mesothelial cells from adenocarcinomas is the presence of expression in ovarian serous carcinomas.2540–42 Cytokeratin 5 is expressed by normal mesothelium, squamous and transitional epithelia, and myoepithelial cells in the cytoplasm. Unfortunately, a significant proportion of breast carcinomas also express cytokeratin 5, precluding its use if breast adenocarcinoma involving effusion fluid is in the differential diagnosis.214043 Mesothelin is a surface protein that may be involved in cell-cell adhesion. Its expression is seen in mesothelial cells and surface ovarian epithelial cells and pancreatic ductal carcinomas.4044 HBME-1 is a mouse monoclonal antibody prepared from a human malignant epithelial mesothelioma; however, its specificity is not very high for mesothelial cells.* The markers discussed previously are useful to characterize mesothelial cells and are particularly useful when the cells in question are abundant in the specimen. However, they are not particularly useful in cases in which it is necessary to characterize a small population in an effusion specimen containing abundant mesothelial cells. Negative staining can be due to lack of sensitivity or technical reasons. For these reasons we prefer to use epithelial markers, which can positively identify epithelial cells. Epithelial-Associated Markers Antibodies for the detection of epithelial cells include carcinoembryonic antigen (CA) 125, MOC-31, Ber-EP4, CA 19-9, and B72.3. Carcinoembryonic antigen was one of the first antibodies used to distinguish epithelial cells from mesothelial cells. Immunoreactivity varies considerably, depending on the antibody clone and on the range of tumor types used. Mesothelial cells are generally negative for carcinoembryonic antigen staining, as are most carcinomas derived from gynecologic sites, which limits its utility in the evaluation of peritoneal washings from patients thought to have ovarian or endometrial carcinoma.50–54 Several investigators have also reported that carcinoembryonic antigen stains macrophages and other inflammatory cells owing to nonspecific cross-reactivity and that it has been associated with a 5% to 15% false-positive rate.10313247495556 CA 125 is well established as a tumor marker for ovarian carcinoma, but it can also be expressed in pancreatic and lung carcinomas.57 CA 125 expression has been shown to have low specificity and low sensitivity in effusion samples.4649 The MOC-31 antibody recognizes an epithelial-associated transmembrane glycoprotein of unknown function in the GLS-1 small cell lung carcinoma cell line and in epithelial tumors. The Ber-EP4 antibody is generated from mice immunized with cells from the MCF-7 breast carcinoma cell line and 2 noncovalently bound glycopeptides. Most studies have documented that MOC-31 and Ber-EP4 are highly effective in distinguishing adenocarcinoma from reactive mesothelial cells, although a few authors have reported possible expression of these markers in mesothelial cells.† CA 19-9 is a sialylated lacto-N-fucopentose II molecule that is related to the Lewis a (Lea) blood group substance. Its expression is usually seen in ovarian and gastrointestinal adenocarcinomas. Ueda et al described that 89% of serous adenocarcinoma samples and none of the reactive mesothelial cells showed immunoreactivity for the CA 19-9 antibody.49 The monoclonal antibody B72.3 is generated with a membrane-enriched fraction of human mammary carcinoma and recognizes the glycoprotein TAG-72 that is expressed in a wide variety of adenocarcinomas. In a study by Ko et al, B72.3 was expressed in 69% of adenocarcinomas, and it lacked affinity for reactive mesothelial cells.31 Immunocytochemistry: Conclusions There is no single marker that appears sensitive and specific enough to correctly diagnose peritoneal washings. Therefore, a panel of antibodies appears to be the best approach in the discrimination between gynecologic adenocarcinoma and reactive mesothelial cells. Ueda et al49 recently proposed a combination of MOC-31, Ber-EP4, CA 19-9, and carcinoembryonic antigen antibodies as a suitable panel. MOC-31 and Ber-EP4 appear to be the most effective markers to distinguish adenocarcinoma from mesothelial cells. Mesothelial markers are indicated to distinguish peritoneal involvement by adenocarcinoma from mesotheliomas. Flow Cytometry and Molecular Studies The evaluation of flow cytometry and molecular studies in effusion specimens has been limited.61 Dong et al62 have shown that it is possible to analyze epithelial and mesothelial markers (Ber-EP4 and epithelial membrane antigen), CD138, and integrin subunits with a 4-color flow cytometer. Holloway et al63 applied a real-time polymerase chain reaction–based assay involving 17 genes to 13 independent cytologic samples from biopsy-proven malignant mesothelioma and lung adenocarcinomas. They used microarray data for 190 lung adenocarcinomas and 33 malignant mesotheliomas to identify genes differentially expressed between the 2 diseases. Their test was able to distinguish between lung adenocarcinoma and mesothelioma in cells collected from pleural effusions. SUMMARY In summary, peritoneal effusions may represent major interpretive challenges to the pathologist and their status has important clinical implications. B72.3, MOC-31, and Ber-Ep4 represent the best available markers to identify epithelial cells. Caution is advised to not overdiagnose endometriosis or endosalpingiosis as adenocarcinoma.