Diagnostic pitfalls in effusion fluid cytology.

• Many of the false-positives in effusion fluid cytology are caused by the atypical features of reactive mesothelial cells associated with a variety of underlying nonneoplastic processes, including acute pancreatitis , [ 4 ] tuberculosis , [ 1 ] ovarian fibroma , [ 1 ] pulmonary infarction , [ 5 ] chemotherapy , [ 6 ] and cirrhosis (especially post-hepatitic secondary to chronic active hepatitis) . [ 1 ] These clinical conditions may induce remarkable changes in mesothelial cells, resulting in morphologic appearances overlapping those with malignant cells. This may lead to the pitfall of misinterpreting these floridly reactive mesothelial cells with atypical features as cancer cells [see Figure 5c ]. Proliferation spheres (metastatic mammary carcinoma, pleural fluid). The neoplastic cells tend to show radial orientation along the periphery (arrows). [a–d, PAP-stained ThinPrep preparation (a,c, 40X; b,d, 100X).] Acinar pattern (metastatic ovarian adenocarcinoma, peritoneal fluid). Acinar structure with a hollow center (arrow). Others (arrowheads) may resemble proliferation spheres, but these are smaller in size and show central lumina usually discerned by adjusting the fine focus. [PAP-stained SurePath preparation (100X).] Proliferation spheres (metastatic small cell carcinoma of lung, pleural fluid). The cellular details can be evaluated at the periphery of the spheres. The cells have scant cytoplasm and nuclei with ‘salt and pepper’ chromatin without conspicuous nucleoli (arrow inset of b). [PAP-stained SurePath preparation (a,c, 100X; b, 100X; inset of c, 100X zoomed).] Metastatic ovarian endometrioid carcinoma (ascitic fluid). Cohesive clusters of neoplastic cells appear to be conglomerations of ‘proliferation spheres’ (arrows) leading to papillary-like configurations. The patient had endometrioid carcinoma of ovary without papillary component. [a–c, PAP-stained SurePath smear; d,e, DQ-stained Cytospin smear (a, 40X; b,c, 100X zoomed; d, 40X; e, 100X zoomed).] Degenerative vacuoles in reactive mesothelial cells (ascitic fluid). Note relatively fuzzy boundaries of vacuoles (arrowheads in b,c) without any secretions (compare with Figure 7b). The secretory vacuoles containing mucin in neoplastic cells usually show secretion with a targetoid appearance (compare with Figure 7a). Some of these cells may have nuclear features overlapping with cancer cells (c) and may be misinterpreted as cancer cells, especially in patients with clinical history of adenocarcinoma. RM, reactive mesothelial cells. [a–c, PAP-stained SurePath preparation (a, 100μ F1 (Focus 1) and F2 (Focus 2); b,c, 100X zoomed).]

False-positive results in effusion cytology are usually misinterpretations related to the aforementioned pitfalls. [ 1 , 4 - 6 , 29 ] But the so-called false-negative results are usually not because of misinterpretation alone. Malignant cells may not be identified in the clinically proven malignant effusions in about 5% of cases. Such cases with negative results should not be considered as false-negatives or ‘misinterpretations.’ The causes responsible for such negative results are: The effusion may be secondary to blockage of the lymphatics by neoplastic cells that have not exfoliated into the serous cavity. The effusion may simply show reactive mesothelial cells with or without inflammatory cells and lack any malignant cells. This is usually observed with neoplasms that spread by lymphatics. Malignant cells may induce increased capillary permeability due to their chemical mediators, such as VEGF (vascular endothelial growth factor), leading to an accumulation of fluid with the absence of neoplastic cells. Neoplasms such as low-grade sarcomas and spindle cell mesotheliomas usually do not exfoliate the cells into effusions. Neoplastic cells may not exfoliate into the effusion fluid because of an organized thick fibrinous layer covering the serosa. This encapsulation is usually observed in pleural cavities with epithelioid mesothelioma. The neoplastic cells decrease in number over time and eventually may disappear totally. The effusion may be secondary to other associated pathologies such as irradiation, obstructive or aspiration pneumonia, atelectasis, pleuritis, and infarction. Not all effusions from patients with cancer show malignant cells. When the initial effusion smear is negative, a repeat cytologic examination is recommended for recurrent or persistent effusions if there is continued clinical suspicion of malignancy. Because the detection rate of malignancy is increased when multiple specimens are examined, these negative cases may eventually show tumor cells in recurrent effusions. [ 1 , 40 , 41 ] The effusion may be secondary to blockage of the lymphatics by neoplastic cells that have not exfoliated into the serous cavity. The effusion may simply show reactive mesothelial cells with or without inflammatory cells and lack any malignant cells. This is usually observed with neoplasms that spread by lymphatics. Malignant cells may induce increased capillary permeability due to their chemical mediators, such as VEGF (vascular endothelial growth factor), leading to an accumulation of fluid with the absence of neoplastic cells. Neoplasms such as low-grade sarcomas and spindle cell mesotheliomas usually do not exfoliate the cells into effusions. Neoplastic cells may not exfoliate into the effusion fluid because of an organized thick fibrinous layer covering the serosa. This encapsulation is usually observed in pleural cavities with epithelioid mesothelioma. The neoplastic cells decrease in number over time and eventually may disappear totally. The effusion may be secondary to other associated pathologies such as irradiation, obstructive or aspiration pneumonia, atelectasis, pleuritis, and infarction. Not all effusions from patients with cancer show malignant cells. When the initial effusion smear is negative, a repeat cytologic examination is recommended for recurrent or persistent effusions if there is continued clinical suspicion of malignancy. Because the detection rate of malignancy is increased when multiple specimens are examined, these negative cases may eventually show tumor cells in recurrent effusions. [ 1 , 40 , 41 ]

False positivity in up to 0.5% and false negativity in up to 30% of cases have been reported in effusion fluid cytology . [ 1 ] Most of these discrepant diagnoses, especially false-positive results, could be prevented if the interpreter is conversant with the following factors responsible for potential diagnostic pitfalls in this area of cytopathology [ Table 1 ]. Various pitfalls in effusion cytology as potential causes of misinterpretation. FP: False-positive, FN: False-negative, Misintp: Misinterpretation, X: Usually responsible for this type of misinterpretation surface tension-related alterations in cytomorphology improper specimen processing Many faces of reactive mesothelial cells, overlapping with those of cancer cells proliferation -related features degenerative changes, such as nuclear hyperchromasia and cytoplasmic vacuolation unexpected patterns and unusual entities. surface tension-related alterations in cytomorphology improper specimen processing Many faces of reactive mesothelial cells, overlapping with those of cancer cells proliferation -related features degenerative changes, such as nuclear hyperchromasia and cytoplasmic vacuolation unexpected patterns and unusual entities.

If staining, cell-block preparation, immunostaining, and other specimen processing steps are not organized properly to address various objectives associated with interpretation of effusion cytology, it may lead to suboptimal results. This may range from improper specimen collection or storage to failure of making smears for proper staining such as Diff-Quik staining and cell-block preparation . Additional factors such as improper orientation and organization of immunostained cell-block sections for evaluation of ‘ subtractive coordinate immunoreactivity pattern ’ (SCIP) [ 2 , 3 ] may further compromise the final interpretation. [ 1 , 2 ]

The presence of some unexpected patterns and unusual entities in serous cavity fluid specimens, including effusions and washings, may create an interpretation challenge and lead to the diagnostic pitfalls. Some of the patterns and entities are described below. Reactive lymphoid effusions with chronic inflammation showing numerous lymphoid cells [ Figure 9 ] may be misinterpreted as one of the blue cell tumors especially in pediatric population with history of lymphoma, neuroblastoma, Ewing’s/primitive neuroectodermal tumors (PNETs), Wilms’ tumor, or desmoplastic small round cell tumor (DSRCT). [ 29 , 30 ] Chronic inflammatory cells with a few reactive mesothelial cells (pleural fluid). a. With DQ stain, the typical nuclear morphology helps to interpret them as polymorphic lymphocytes, which is consistent with chronic inflammatory cells. b. With PAP stain, these chronic inflammatory cells may resemble cells of lymphoma (compare with Figure 12d) and round blue cell tumors, especially in children. RM, reactive mesothelial cell. [a, DQ-stained Cytospin smear; b, PAP-stained SurePath smear (a, 100X; b, 100X).] Polymorphic lymphocytes of low-grade lymphoproliferative neoplasms, such as some low-grade follicular lymphomas [ Figure 12 ], may be misinterpreted as reactive chronic inflammatory cells. [ 31 ] Similar to reactive lymphocytes in effusion with chronic inflammatory cells, these polymorphic low-grade lymphoma cells resemble cells of small round cell tumors and vice versa, especially in PAP-stained preparations [see Figure 12d,e,f ]. Immunophenotyping with flow cytometry and immunostaining of cell-block sections or Cytospin smears are very useful ancillaries to cytomorphology for objective interpretation in addition to cytogenetics, in-situ hybridization, and other tests as indicated. [ 31 ] Large cell carcinoma of lung (pleural fluid). Numerous isolated carcinoma cells (arrow in c) seen as the predominant population. Occasional mitotic figures (arrowhead in c) are present. [a–c, PAP-stained SurePath preparation (a, 10X; b, 40X; c, 100X).] Metastatic mammary carcinoma (pleural fluid). The immunostaining (i–l) shows that neoplastic cells form the predominant population (immunoreactive for Ber-EP4 in k and non-immunoreactive for vimentin in l) with scant mesothelial cells (arrow RM in i) and rare histiocytes. Proliferation spheres without cores (c,d,g,h) show a radial arrangement of neoplastic cells at their periphery, better seen in a PAPstained preparation (c,d). Although the morphology of cells is better seen with a PAP stain (inset of b), smaller groups and single cells are difficult to distinguish from reactive mesothelial cells in PAP-stained smears (a–c). However, such rare reactive mesothelial cells (arrow RM) are easy to identify in DQ-stained preparations (e,f). IC, inflammatory cells; RM, reactive mesothelial cells. [a–d, PAP-stained SurePath smear; e–h, DQ-stained Cytospin smear; i–l, immunostained cell-block section (a, 40X; b,c, 100X; inset of b, 100X zoomed; d, 100X zoomed; e, 40X; f,g, 100X; h, 100X zoomed; i–l, 40X)]. Follicular lymphoma (peritoneal fluid). The lymphoid population with rare reactive mesothelial cells (arrowheads RM in a,d) resemble chronic inflammatory cells [see also Figure 9 ] and cells of round blue cell tumors, especially in PAP-stained preparations (d). The typical nuclear morphology in a DQ-stained preparation (a–c) helps to interpret the round cells as atypical lymphocytes (arrows b,c,e,f in a,d). The flow cytometry demonstrated a monoclonal lymphoid population. The patient had follicular lymphoma with a colon mass (g). RM, reactive mesothelial cells. [a–c, DQ-stained Cytospin preparation; d–f, PAP-stained SurePath preparation; g, hematoxylin and eosin (HE)-stained paraffin-embedded tissue section of colon mass (a, 100X; b,c, 100X zoomed; d, 100X, e,f, 100X zoomed; g, 100X).] A predominant population of scattered isolated cells of low-grade carcinoma from some primary sites, such as ovary and breast, may resemble floridly reactive mesothelial cells [ Figures 10 , 11e ]. As a DQ stain highlights a second population more distinctly, the failure of including a DQ-stained preparation in the evaluation protocol may compromise proper interpretation in such clinical situations and lead to the pitfall of misinterpreting such specimens as negative for neoplastic cells because of a failure to detect the second population in a PAP-stained preparation alone . Further immunocytochemical evaluation with a properly tailored immunopanel (including two color vimentin (red) with BerEP4 (Brown) immunostaining 32 ), after initial suggestion of a two-cell population in DQ-stained preparation, is helpful to demonstrate their non-mesothelial nature and confirm the second foreign population. Psammoma bodies are concentrically laminated calcific spherules encountered in 3.7% of effusions. [ 33 ] In pleural and pericardial effusions, they are usually associated with various papillary neoplasms such as metastatic papillary carcinoma of thyroid, bronchioloalveolar carcinoma of lung, and serous papillary cystadenocarcinoma of ovary. However, they may be associated with non-neoplastic processes such as papillary mesothelial hyperplasia, endometriosis, and endosalpingiosis in up to 30% of peritoneal effusions, washings, [ 34 ] and culde-sac aspirates, where they are a significant pitfall and may lead to a false-positive interpretation for malignancy. An experience graciously shared by Dr Naylor emphasizes this point. A case of reactive effusion with psammoma bodies in culdocentesis fluid presented at a diagnostic seminar in 1968 was misinterpreted by all four cytopathologists, including him, as adenocarcinoma. [ 35 ] Reactive papillary groups , gland-like epithelial structures, [ 34 , 36 ] and Müllerian inclusions , [ 37 ] especially in peritoneal washings, may be misinterpreted as malignant. Müllerian inclusions are usually seen as tubular or papillary structures. They often form a single layer of cells with some atypia and may be associated with psammoma bodies. This is an important pitfall in peritoneal specimens, especially washings. Its recognition is important to avoid a misdiagnosis of disseminated cancer. [ 37 ] Megakaryocytes are large cells with large, irregular, lobulated, hyperchromatic nuclei. Pleural effusions with fresh blood, as a result of bleeding from the pulmonary microvasculature, may contain megakaryocytes. [ 38 ] They may be present in effusions associated with myeloproliferative disorders or in cases with extensive replacement of bone marrow by metastatic carcinoma. [ 39 ] Their morphology is comparable with the megakaryocytes observed in Romanowsky-stained bone marrow smears. The presence of these cells in effusions may lead to the pitfall of misinterpretation as neoplastic cells, especially in PAP-stained smears.

Degenerative and other changes secondary to improper storage and handling of specimens introduce various atypical morphologic features, including nuclear hyperchromasia and cytoplasmic vacuolations . They are not uncommon in free-floating cells in a fluid medium. Degenerative hyperchromasia seen in Papanicolaou (PAP)-stained smears and other changes in PAP and Diff-Quik (DQ)-stained smears may lead to the pitfall of misinterpreting such cells in effusions as malignant cells [ Figure 5 ]. In chronic effusions, mesothelial cells with degenerative changes show many small vacuoles and may resemble foamy macrophages. The vacuoles may join with each other and form a single large cytoplasmic vacuole displacing the nucleus to the periphery of the cell, leading to a signet-ring cell appearance resembling an adenocarcinoma cell [see Figure 5 ]. Similar changes may also be produced when effusion specimens are left at room temperature for a long time. These artifacts are frequent in effusions collected during the weekend (especially with warmer ambient conditions) and not processed immediately. • Due to this, reactive mesothelial cells, with degenerative intracytoplasmic vacuoles may be misinterpreted as adenocarcinoma cells with mucin vacuoles [see Figure 5 ] . This may be aggravated further by the presence of mesothelial cells with large degenerated nuclei. Similarly, the neoplastic cells with degenerative cytoplasmic vacuoles are much more frequent than adenocarcinoma cells with true mucin vacuoles. Consequently, the nonmucin-producing neoplastic cells with degenerative changes may be misinterpreted as mucin-producing adenocarcinoma [ Figure 6 ]. Metastatic ovarian serous papillary carcinoma (ascitic fluid). Adenocarcinoma cells with degenerative cytoplasmic vacuoles (arrows), which may resemble adenocarcinoma cells with true secretory vacuoles, such as those seen in ovarian mucinous cystadenocarcinoma. [a–c, PAPstained SurePath preparation (a–c, 100X; insets, 100X zoomed).] Degenerative intracytoplasmic vacuoles usually do not occupy the entire cytoplasm of a cell and do not show ballooning. The borders of such degenerative vacuoles are usually ill-defined [see Figure 5b ,c]. In comparison, the true intracytoplasmic vacuoles with secretion usually balloon the entire cell and occupy most of the cytoplasm, and may show secretion in it (targetoid vacuole) [ Figure 7a ]. These vacuoles usually have well-defined borders [ Figure 7b ]. However, it is not always possible to distinguish reactive mesothelial cells with degenerative intracytoplasmic vacuoles from neoplastic cells with secretory vacuoles by cytomorphology alone with certainty. Ancillary tests, including histochemistry, such as a periodic acid–Schiff (PAS) stain with diastase digestion and a mucicarmine stain, may help to discriminate between these entities [ Figure 8 ]. Secretory cytoplasmic vacuoles. a. Metastatic papillary carcinoma of thyroid (pleural fluid). Targetoid secretory vacuole with colloid in neoplastic cell (blue arrow). b. Metastatic colonic adenocarcinoma (ascitic fluid): intracytoplasmic vacuole with welldefined margin (red arrow). [a, DQ-stained SurePath smear; b, PAP-stained Cytospin smear (a,b, 100X zoomed).] Metastatic mucinous adenocarcinoma (pleural fluid). Intracytoplasmic mucicarmine-positive mucin (red arrow). [cell-block section, mucicarmine stain (100X).]

Changes in cell morphology secondary to nutrient-rich fluid medium, which allows continued proliferation of exfoliated cells, lead to various diagnostic pitfalls, including: proliferation spheres increased number of mitotic figures prominent nucleoli. proliferation spheres increased number of mitotic figures prominent nucleoli. The malignant cells may continue to proliferate even after they are exfoliated into a serous cavity fluid to give rise to ‘three dimensional globular structures’ known as ‘ proliferation spheres ’ [ Figures 1, 3 ]. These proliferation spheres are three-dimensional, solid or hollow aggregates without a stromal core. They are unique to metastatic cancer cells in serous cavity fluids. In contrast, urine and cerebrospinal fluid are not conducive to proliferation of neoplastic cells; therefore, urothelial carcinoma in urine and metastatic cancer cells in cerebrospinal fluid do not form proliferation spheres. The periphery of the proliferation spheres often shows a radial arrangement due to the rapid proliferation of their constituent cells, resulting in an increase in the size of these groups [see Figure 1 ]. Acinar and glandular structures may also resemble proliferation spheres at lower magnification. However, these structures are smaller and a central space can usually be seen at higher magnification by adjusting the fine focus [ Figure 2 ]. Proliferation spheres are not observed in recently developed malignant effusions because of lack of time necessary for proliferation, and so they are usually observed at a later stage . They continue to grow and may reach up to 0.5 mm in diameter, which are readily visible to the naked eye either in the fluid or on the slide. These are observed in effusions secondary to many types of malignancies, especially ductal carcinoma of the breast, epithelioid mesothelioma, and poorly differentiated small cell carcinoma of the lung [ Figure 3 ]. Proliferation spheres are not formed in effusions secondary to cancers that lack significant intercellular cohesion. Examples of these include anaplastic gastric carcinoma (linitisplastica type), non-cohesive type of adenocarcinoma of the lung, non-cohesive epithelioid mesothelioma, pleomorphic giant cell carcinoma of the pancreas, giant cell carcinoma of the lung, lobular carcinoma of the breast, adrenocortical carcinoma, and lymphomas. Irrespective of effusion duration, such effusions usually contain a high proportion of isolated cells. Some proliferation spheres may conglomerate together, especially during specimen processing, to form groups that may impart a papillary configuration [ Figure 4 ] . Proliferation spheres simulating papillary structures are relatively common in effusion smears from a variety of neoplasms and do not denote a papillary architecture at the primary lesion. Consequently, papillary-like structures are not uncommon in effusions associated with non-papillary adenocarcinomas of colon and pancreas. Proliferation spheres are not observed in recently developed malignant effusions because of lack of time necessary for proliferation, and so they are usually observed at a later stage . They continue to grow and may reach up to 0.5 mm in diameter, which are readily visible to the naked eye either in the fluid or on the slide. These are observed in effusions secondary to many types of malignancies, especially ductal carcinoma of the breast, epithelioid mesothelioma, and poorly differentiated small cell carcinoma of the lung [ Figure 3 ]. Proliferation spheres are not formed in effusions secondary to cancers that lack significant intercellular cohesion. Examples of these include anaplastic gastric carcinoma (linitisplastica type), non-cohesive type of adenocarcinoma of the lung, non-cohesive epithelioid mesothelioma, pleomorphic giant cell carcinoma of the pancreas, giant cell carcinoma of the lung, lobular carcinoma of the breast, adrenocortical carcinoma, and lymphomas. Irrespective of effusion duration, such effusions usually contain a high proportion of isolated cells. Some proliferation spheres may conglomerate together, especially during specimen processing, to form groups that may impart a papillary configuration [ Figure 4 ] . Proliferation spheres simulating papillary structures are relatively common in effusion smears from a variety of neoplasms and do not denote a papillary architecture at the primary lesion. Consequently, papillary-like structures are not uncommon in effusions associated with non-papillary adenocarcinomas of colon and pancreas. In smears of fresh effusions, mitotic figures may be seen at the periphery of proliferation spheres. Mitotic figures are frequent if the patient is not receiving chemotherapy and the specimen has not been refrigerated. After effective chemotherapy, neoplastic cells often show apoptosis with karyorrhexis in solitary and loose groups of cells. Mitotic figures with multinucleation may also be noted in peritoneal dialysis fluids. [ 28 ]

Although the interpreters may be familiar with the conventional morphology of various neoplastic cells, surface tension of the effusion fluid may lead to alteration in usual cellular morphology. • A classical example in this category would be predominance of polyhedral cells in metastatic sarcoma, in contrast to spindle cells in other cytology specimens such as fine-needle aspiration biopsy (FNAB) smears .