Practical Approach to Reporting Based on the International System for Serous Fluid Cytopathology.

All authors conceptualised, designed, wrote and revised the review.

Factors influencing sample adequacy are sample volume, cellular content and cellular preservation (Figure  2 ). Evaluation of serous effusion sample adequacy. Generally, 75 mL is optimal volume [ 2 ] for cytological assessment, but for pericardial fluid, 60 mL is adequate [ 3 ]. Smaller samples should not be rejected but low volume should be commented in a sample report. This is particularly important if there is a clinical suspicion of malignancy, but the sample remains non‐diagnostic (ND) or negative for malignancy (NFM), because low volume and low cellularity samples may not be reliable to exclude malignancy. In contrast, there is no recommended maximum volume for an effusion specimen and clinicians should be encouraged to submit all fluid that is collected to the laboratory. If specimens are to be divided to perform multiple tests such as microbiologic culture or biochemical analysis, the fluid should be well mixed and aliquoting is best performed at the time of sample. Larger volume samples might be the matter of transportation from distant hospitals to the pathology laboratory and a matter of storage space in the laboratory, not a diagnostic issue. The higher the volume, the smaller the number of samples reported as ND or atypical [ 1 ]. Expected cell populations in serous effusions are mesothelial and inflammatory cells (Figure  3 , Table  1 ). Do we need to see mesothelial cells in an adequate sample? According the TIS, it is acceptable to find only lymphocytes (e.g., in the case of tuberculosis or chylous effusions) or neutrophils (e.g., in acute bacterial infections) in benign effusions and the cellular content may be adequate even without mesothelial cells under the aforementioned circumstances. The diagnosis of malignancies may also be made with a one cell population of malignant cells without mesothelial cells. In other words, when the cell population is representative of the disease process behind the pathological effusion, it may be adequate without the presence of mesothelial cells. Reactive effusion with mix of lymphocytes, macrophages and mesothelial cells. May–Grünwald Giemsa stain, 100× magnification. Evaluating the cell populations in serous effusions. Epithelial Mesothelial Inflammatory Mesenchymal Unknown/Other High Low Medium Almost exclusive population of one cell type Heterogeneous population of cells Can a sample be ND despite being cellular? Indeed yes; it may be so because of degenerative changes. Degenerative changes lead to loss of quality of cellular preservation. This may be due to delays in transport to the laboratory with resulting cellular degenerative changes, accompanied by bacterial or fungal overgrowth. Furthermore, the presence of technical artefacts and contaminants may limit interpretation. Other factors that may prevent reliable interpretation are totally obscured cells by blood or inflammatory infiltrate or poor technical preparation. Mechanical factors may have distorted cells during smear preparation or the cells may have dried. The sample may also be totally haemolysed. Poor preservation may result in cellular degeneration that worsens stain penetration. Degenerated cells should not be considered adequate nor interpreted as atypical. In addition to bacterial overgrowth, also other contaminants like talc powder, debris or squamous cells from the skin surface may render the sample difficult to interpret [ 1 ]. A key question regarding sample adequacy is whether the sample represents or provides useful information of the underlying disease process. A typical ND sample is acellular, extremely degenerated or haemorrhagic. If diagnostic criteria of any category other than ND are present, the sample is adequate despite the cellularity or sample volume being low [ 1 ]. In our opinion, the term ‘non‐diagnostic’ is clearly better than ‘insufficient’, regarding the aspect that the sample may be abundant in volume or in cellular content, but it may still not be possible to make a diagnosis.

The referral to cytology laboratory should include sufficient clinical information regarding age, gender and medical history. The following three scenarios are the most common encountered in clinical practice (Figure  4 ). Common presentations of effusions relating to clinical information. In assessment of cell population (Table  1 ), there might be one dominant population indicating a specific cause of effusion (e.g., lymphocytes). In the absence of malignant cells, most samples are likely to be correctly reported as NFM. According to TIS the criteria for a ‘negative for malignancy’ category, there are no features of malignancy either of mesothelial or of non‐mesothelial origin. The sample consists mostly of mesothelial cells that may be arranged as single cells, flat sheets or small loosely cohesive clusters, often with ‘windows’ between each other due to microvilli. Rare binucleation or multinucleation in mesothelial cells is acceptable but cellular atypia should be minimal or absent and the cells should be round or oval in shape. Cytoplasm may be vacuolated in reactive mesothelial cells. The nuclei are either centrally placed or eccentric. The size of mesothelial cell nucleus is usually 1.5–2 times the size of neutrophil. The nuclear membrane is smooth. The nucleolus may be prominent. In the background there may be variable inflammatory cells like histiocytes, giant cells, lymphocytes, neutrophils or more rarely eosinophils. Other benign background components may also be seen. Inflammatory, reactive and metaplastic changes are categorised as NFM [ 1 ]. A ‘normal effusion’ does not exist. Any effusion is pathological, because the serous cavities (pleural, pericardial and abdominal) normally contain only few millilitres of lubricating serous fluid. Effusion may be exudate or transudate, which is an important diagnostic clue in clinical assessment of the aetiology of effusion. In general, transudate (protein content under 30 g/L) is usually a result of systemic disease like cardiac insufficiency, liver or renal disease. Exudate (protein content over 30 g/L) indicates a local disease and the commonest aetiologies are inflammation, infection, trauma and neoplastic processes. Thus, in any effusion specimen, a certain amount of reactive mesothelial cells is present with various degrees of atypia. Those changes do not need to be reported. One exception is a peritoneal washing sample, where the peritoneal cavity lining cells are forced to exfoliate utilising a physiologic balanced saline solution and in that particular situation, a pathological effusion does not necessarily exist at all. In this case the exfoliated mesothelial cells may appear perfectly normal and exfoliate as monolayer sheets rather than individual cells. Unexpected but benign background elements in a sample may provide a clue to the underlying disease process and may include psammoma bodies, collagen balls, detached ciliary tufts, asbestos bodies, Curschmann's spirals, lupus erythematosus cells, necrotic material, fistula material, bile, surgical material, amnionitic fluid, talc, intestinal contents including plant cells, crystals and infectious organisms. Some of the aforementioned materials are extremely rare [ 1 ] (Table  2 ). Evaluating the background in serous effusions. Red blood cells Inflammatory cells Keratin Mucin Asbestos bodies Colloid Amyloid Diathesis/Debris Collagen bodies Psammoma bodies Micro‐organisms (parasites, fungi) If a specific pattern of reactive effusion is present (Table  3 ) such as eosinophilic or lymphocytic, it is recommended to suggest possible clinical causes in the report. For instance, possible causes of an eosinophilic effusion are recent pleural fluid aspiration, trauma leading to pneumothorax or haemothorax, pulmonary infarction, infection, malignancy and allergic conditions such as hypereosinophilic syndrome. An idiopathic eosinophilia is also possible [ 4 , 5 ]. Inflammatory patterns in benign reactive effusions and their possible causes. Viral infection Sarcoidosis Chylous effusions Rheumatoid disease Tuberculosis Chronic/unspecific inflammation Purulent effusion Empyema Bacterial infection Rheumatoid arthritis Other acute inflammation Prior pleural fluid aspiration/paracentesis/thoracocentesis Trauma leading to pneumothorax or haemothorax Idiopathic Infection Pulmonary infarction Hypersensitivity reactions and other allergic conditions Chronic/Unspecific inflammation Granulomatous process Foreign body reaction In a reactive lymphocytic effusion, viral infections, sarcoidosis, rheumatoid disease, chylothorax and tuberculosis might be suggested. Sometimes it is difficult to distinguish on morphology alone, whether a monomorphic appearing lymphocyte‐rich infiltrate is reactive or represents indolent lymphoproliferative disease or low‐grade non‐Hodgkin lymphoma (NHL). Some low‐grade NHL may even show more heterogeneneous cytomorphology and thus mimic a reactive pattern. Without any history or clinical concern of lymphoma, ancillary testing is usually not needed [ 1 ]. A neutrophilic effusion or empyema is usually indicative of bacterial infection, but rarely of other aetiologies like rheumatoid arthritis [ 6 ] and occasionally may also be encountered in malignancy, for example, in a rupture of lung squamous cell carcinoma into the pleural cavity. An effusion specimen that consists predominantly of histiocytes/macrophages is quite common. Macrophages may be difficult to distinguish from mesothelial cells and sometimes they may even remind atypical/malignant epithelial cells (Table  4 ). Differential diagnosis between adenocarcinoma, mesothelial cells and macrophages. Abbreviations: BAP1, breast cancer 1‐associated protein 1; CD68, cluster of differentiation 68; EMA, epithelial membrane antigen; HEG1, protein HEG homologue 1; IC, immunochemistry; BerEP4 and MOC31 are antibodies targeting epithelial cell adhesion molecule; NC ratio, nuclear‐to‐cytoplasmic ratio; WT1, Wilm's tumour 1. If malignant cells are identified, ancillary testing to confirm the known primary should be considered especially if this is the first such sample sent to the cytology laboratory (Table  5 ). The International System for Serous Fluid Cytopathology malignant category (MAL) definition and explanatory note. A clinical dilemma sometimes arises whether or not to perform ancillary tests in the absence of abnormal cells on routine stains in effusions from patients with known malignancy. In general, immunochemistry (IC) is not recommended on such samples. There may be two scenarios when ancillary testing may be considered in samples NFM on routine stains. The first is when there is a strong clinical concern for oligometastatic disease to be identified. Even small numbers of malignant cells may provide comprehensive genomic information by next‐generation sequencing (NGS) [ 7 , 8 ]. The second situation is when the pathologist is concerned that small numbers of malignant cells with a bland appearance may have been overlooked in a very cellular sample. In patients with known lymphoma, it may be necessary to exclude involvement by flow cytometry or IC on lymphocyte‐rich effusions even when the morphology appears benign. For NFM samples in patients with known malignancy, the volume recommendation for adequacy may be stated in the conclusion of the report, for example, ‘Pleural fluid (haemorrhagic, 20mL): negative for malignancy. Recommended volume for serous fluid samples = 75mL’ to avoid potential false‐negative results. In the assessment of serous fluid cytology specimens, the basic components of cytology are evaluated to define whether the findings are normal or abnormal (Tables  1 , 2 , 6 , 7 , 8 , 9 ) and whether the number of atypical cells is sparse or numerous. Evaluating the cells in serous effusions. Small Large Variable Normal Round Oval Other Evaluating the nucleus in serous effusions. Small Large Medium Variable N/C ratio Oval Round Bean‐shaped Variable Spindle Nuclear membrane irregularities Chromatin texture Prominent nucleoli Intranuclear inclusions Abbreviation: N/C ratio, nuclear‐to‐cytoplasmic ratio. If sparse atypical cells are present, request a cell block and IC. Could we practice without indeterminate category like atypia of undetermined significance (AUS) completely? Some experienced cytologists do not use it at all. It is uncommonly used as a diagnostic category in effusions and it should be a diagnosis of exclusion. It is optional to use a two‐step process—a preliminary report as AUS and a final report where the diagnosis may be updated after ancillary tests. But what is atypia in serous effusions and what is not? The scenarios were the category of AUS can be used according TIS are as follows: Cytological features are indefinite for a diagnosis of suspicious for malignancy (SFM) or NFM either quantitatively or qualitatively and the specimen is diagnostic. In AUS, the cytological features more closely resemble reactive, degenerative and benign conditions and the cells are more likely mesothelial cells or macrophages, but malignancy or atypical epithelial cells are not possible to completely exclude even after ancillary testing. The atypical appearing cells may be sparse in number, degenerated and the specimen as a whole has low cellularity [ 1 ]. Cytological criteria for AUS category include mild to moderate nuclear enlargement and slightly increased nuclear to cytoplasmic ratio. The nucleoli are variable or prominent. Nuclear membrane irregularities are mild. Both chromatin and cytoplasm structures are altered. Architecturally there is only mild atypia, for example, pseudoglandular or papillary groups [ 1 ]. Mesothelial proliferations may fall in the AUS category if they lack sufficient features to be designated as SFM or malignant‐primary (MAL‐P), but this should be rare when ancillary tests remain inconclusive. Clinical correlation is crucial. Pitfalls when a serous effusion specimen may contain atypical appearing but ultimately benign reactive mesothelial cells are following clinical conditions: liver cirrhosis, peritoneal dialysis, ectopic pregnancy, tubo‐ovarian abscess, chemotherapy, radiotherapy and pelvic tumours. Pericardial fluid specimens are also known to frequently consist of atypical appearing reactive mesothelial cells [ 1 ]. Lymphocytosis of unknown aetiology may also fall in to AUS category when reactive process is favoured, but low‐grade lymphoproliferative disorder cannot completely be ruled out. Epithelial cells in peritoneal washings fall mostly to the category of AUS if they have benign features and their origin remains unknown. The second scenario are epithelial or other cells from benign or borderline tumours of the ovary or abdominal/pelvic cavities. However, when the epithelial cells are reliably interpreted as endometriosis, endosalpingiosis or detached tubal fimbrial epithelial cells, they should be reported NFM whenever possible [ 1 ]. Proposed AUS algorithm: (Figure  5 ). AUS algorithm in The International System for Serous Fluid Cytopathology. AUS, atypia of undetermined significance; IC, immunochemistry; NFM, negative for malignancy; SFM, suspicious for malignancy. Adapted from Chandra et al. [ 1 ]. Malignant features may be present in architecture, cell populations, cells, cytoplasm, nucleus or in the background (Tables  1 , 2 , 6 , 7 , 8 , 9 ). If only sparse malignant cells are present, request cell block and IC. The SFM category may be used in a few scenarios. If there is only a small number of cells or groups with nuclear pleomorphism, the ancillary tests are needed for confirmation of malignancy. Lymphocytic effusions with a monotonous population may fall into SFM category. Proposed SFM algorithm (Figure  6 ). SFM in The International System for Serous Fluid Cytopathology. IC, immunochemistry; MAL‐S, malignant, secondary; SFM, suspicious for malignancy. Adapted from Chandra et al. [ 1 ]. What are the differences in cytological features in specimens categorised as AUS and SFM? In AUS samples, the cytological abnormalities, such as nuclear enlargement and hyperchromasia are only mild and the number of atypical cells is usually small, the cells are dispersed or in occasional small groups. In SFM categorised samples, the degree of cytological abnormality is greater and there are also architectural features such as occasional three‐dimensional (3D) groups. The total number of atypical cells is usually still small. There is still another pitfall: Cancer cells with only mild cytological atypia. In certain entities, malignant cells may show bland appearances or only mild pleomorphism. This kind of cells may be numerous for example in gastric carcinoma. In breast carcinoma, the number of bland cells may be small and even the cell size may be small like in lobular breast carcinoma, where the cells may be arranged in a linear pattern. Both gastric and breast carcinomas may consist of signet ring cells and have single cell pattern. Various well‐differentiated carcinomas often cause diagnostic challenges. Another challenge is the presence of mucin in the sample background (Table  2 ). Special attention is required regarding background with mucinous material with few or no cells in ascitic fluid, because of the possibility of pseudomyxoma peritonei. The differences in cell lineage in AUS and SFM categories are that in AUS the benign cell type is favoured, but epithelial or other malignant cell of origin is not excluded. In SFM malignant cells of either epithelial or other origin are strongly favoured. In IC, the outcomes may be benign, SFM/malignant or inconclusive in preliminary AUS category as opposed to SFM category, where the outcomes are usually malignant or inconclusive. This diagnostic category requires a recognisable abnormal cell population present and adequate for robust diagnosis on which clinical management may be based. Malignant cell type should be specified on morphology alone or supported by IC. Separation must be made to malignant‐primary category (MAL‐P) and malignant‐secondary category (MAL‐S). MAL‐P corresponds mainly to mesothelioma although other very rare primary serosal malignancies exist like primary effusion lymphoma or primary serosal sarcomas. MAL‐S must be further specified as adenocarcinoma, small cell neuroendocrine carcinoma, squamous cell carcinoma, lymphoma, melanoma, sarcoma, leukaemia, etc. The most common malignant category (MAL) scenario in effusion specimens is metastatic adenocarcinoma. Primary organ site may need to be investigated for adenocarcinoma with organ‐/tissue‐specific antibodies (Table  10 ). Architectural patterns of malignant infiltration may provide a clue in the diagnosis and primary site of tumour (Table  8 ; Figures  7 , 8 , 9 , 10 ). Evaluating the architecture in serous effusions. Papillary Three dimensional Single cells Sheets (peritoneal washing) Mulberry clusters Tightly cohesive Dispersed Loosely cohesive (with gaps/windows) Papillary clusters in serous effusion. Papanicolaou stain, 40× magnification. Morules of mesothelial cells in serous effusion. Papanicolaou stain, 200× magnification. Tightly cohesive groups of epithelial cells in serous effusion. May–Grünwald Giemsa stain, 200× magnification. Dispersed cell pattern in serous effusion. Papanicolaou stain, 200× magnification. Large 3D groups are not unique to site. Large papillary clusters may indicate, for example, serous carcinoma of ovary or uterus or papillary renal cell carcinoma. In breast carcinoma there are variable appearances; for example, singly dispersed, small groups and sharply edged large spherical groups. In ovarian carcinomas there may be 3D groups, psammoma bodies, pleomorphic cells and mucin. In mesothelioma, there may be small equally sized groups. The key cytologic features in the conventional type of pancreatic adenocarcinoma include cohesive clusters of ductal cells in glandular crowding and disorganised ‘drunken honeycomb’ pattern or intercalated duct‐like structures with anisonucleosis, cytoplasmic vacuoles and concomitant ‘Indian‐file’ configuration. Undifferentiated pancreatic carcinoma is comprised of enlarged, undifferentiated, pleomorphic malignant cells. Pancreatic adenosquamous carcinoma could show glandular and/or squamous differentiation. Colloid carcinoma is composed of 3D cancer cell clusters floating in thick mucin. In small cell neuroendocrine carcinoma, the key cytomorphologic features are small moderately cohesive groups of cancer cells with scant delicate cytoplasm, sometimes with degenerative vacuolisation mimicking adenocarcinoma, nuclear moulding and stippled chromatin, nucleoli are not prominent. The cells may be arranged as single cells and short chains. Tumour cells are approximately two to three times larger than small mature lymphocytes. Nuclei are oval to irregular and mitoses, apoptosis and nuclear crushing artefact may be present. In the background there might also be necrosis [ 1 ]. Ancillary tests are recommended to confirm the diagnosis of small cell neuroendocrine carcinoma. Morphological overlap may occur with hematolymphoid proliferations, metastatic Merkel cell carcinoma, metastatic basaloid squamous cell carcinoma and other carcinomas with basaloid morphology, poorly differentiated carcinomas and small round blue cell tumours. In IC it is necessary to note that TTF1 (thyroid transcription factor 1) may be also positive in small cell neuroendocrine carcinomas of non‐pulmonary sites and in cells of thyroid gland origin. As the primary site of small cell neuroendocrine carcinoma is most commonly the lung, the pleural cavity is the most common serous cavity to be involved [ 1 , 9 ]. Single‐cell infiltrate may be a feature in many cancers. If the single cells appear as signet ring cells they may be indicative of gastric, pancreatic or urothelial primary site. If the single cells are plasmacytoid with eccentric, round or oval nuclei it could point towards melanoma, breast, gastric, pancreatic, urothelial and plasma cell neoplasms. In case of a plasma cell neoplasm, ancillary studies are needed to confirm the clonal nature of plasma cells. Lymphoid single‐cell infiltrate would point to lymphoma/leukaemia and the features could further be indicative of either small, intermediate or large lymphoid cells and it should also be estimated whether the lymphoid population is monomorphic or pleomorphic. Small lymphoid cells are approximately two times larger than erythrocytes, which equals or is smaller than a resting lymphocyte. Large lymphoid cells are two to three times larger than resting lymphocytes and the size of medium‐sized lymphoid cells is between small and large lymphocytes [ 10 ]. The estimate of the predominant size of cells in lymphoid proliferations helps to narrow the differential diagnosis roughly towards low‐grade or high‐grade NHL. Together with other ancillary testing like flow cytometry, IC staining and cytogenetic/molecular genetic studies the diagnosis of hematolymphoid malignancy can be made in serous effusion samples, but regarding the rapidly evolving spectrum of classification systems, the diagnostics mainly requires a specialist of hematopathology. Only a few single scattered large atypical cells, either binucleated/multinucleated forms (Reed–Sternberg cells) or mononuclear variants (Hodgkin cells) could be seen in Hodgkin lymphoma (HL), although this type of lymphoma is rarely encountered in effusion specimens. The large atypical cells of HL have abundant cytoplasm and large nuclei with prominent nucleoli and there would be mixed inflammatory infiltrate in the background. When a single cell patterned infiltrate consists of medium‐sized or large atypical lymphoid cells, a MAL diagnosis is more obvious and the first step is to confirm the lymphoid nature of the cells with a preliminary IC panel and to exclude entities as follows: poorly differentiated carcinoma, melanoma, sarcoma, small cell neuroendocrine carcinoma, acute myeloid leukaemia and plasma cell neoplasm [ 1 ]. Monomorphic single‐cell infiltrate can be a feature of lymphoid neoplasms, sarcoma, macrophages or mesothelial cells. Pleomorphic single cell infiltrate could be seen in melanoma, poorly differentiated carcinoma, germ cell tumours and sarcoma. Single‐cell infiltrate consisting of clear cells could point towards renal, ovarian or endometrial carcinoma. Small round and monotonous undifferentiated cells with high nuclear–cytoplasmic ratio give rise to wide differential diagnosis of mesenchymal and non‐mesenchymal small round cell tumours. Diagnosis of tumour subtype can be achieved on effusion samples with a combination of clinical history, imaging, cytomorphological features and ancillary testing [ 11 ]. It is extremely rare to encounter metastatic melanoma in serous effusion samples. Cytological criteria include epithelioid, plasmacytoid or spindled cells arranged as single cells or loosely cohesive groups. The nuclei of melanoma cells are variable and may appear as enlarged, oval, elongated, eccentrically placed or pleomorphic. Macronucleoli, intranuclear inclusions, binucleation and multinucleation may occur. The cytoplasm is granular, eosinophilic and may contain melanin pigment. Clinical history and immunochemical markers will assist making the final diagnosis [ 1 ]. Squamous cell carcinoma is also rarely encountered in serous fluids although it is a common malignancy. Pleural cavity is the most commonly affected serous cavity and the lung is the most common primary site [ 1 , 12 , 13 ]. According to TIS, the cytological criteria and cytomorphological features of metastatic squamous cell carcinoma in serous fluids [ 13 ] are following in a well‐differentiated or keratinising squamous cell carcinoma: single‐cell infiltrate or small clusters and in the background there may be seen keratin debris and keratin pearls. The nuclei are enlarged, irregular and hyperchromatic. The cytoplasm is dense, cyanophilic to eosinophilic or dyskeratotic/orangeophilic on Papanicolaou staining. The amount of cytoplasm is variable. Cells may be polygonal or anucleated and tadpole cells and fibre cells may also be seen. Cell borders are sharply defined. In contrast, in poorly differentiated or non‐keratinising squamous cell carcinoma, the cells may be arranged in clusters or syncytial groups and nuclear‐to‐cytoplasmic ratio is high. Cytoplasm is more ill‐defined, granular to dense. The chromatin is coarse, but otherwise the nuclei are similar as in well‐differentiated forms [ 1 ]. Other carcinomas like urothelial carcinoma, renal cell carcinoma and hepatocellular carcinoma may also involve serous cavities. Typical cytomorphological features in urothelial carcinoma are single cells or small clusters of cells with increased nuclear‐to‐cytoplasmic ratio, hyperchromatic nuclei, irregular nuclear membrane and variably prominent nucleoli. Signet ring cells, plasmacytoid cells and glandular or squamous differentiation may occur in urothelial carcinoma. Hepatocellular and renal cell carcinomas have typically cohesive clusters or single‐cell pattern in serous effusion samples. In renal cell carcinoma, the nuclei are enlarged; nuclear membrane is variable from smooth to irregular and nucleoli may be prominent. Cytoplasm is moderate to abundant in amount and may be clear, vacuolate or granular. In hepatocellular carcinoma nuclei are oval to irregular with macronucleoli and cytoplasm may be granular, eosinophilic, polygonal or may contain bile pigment [ 1 ]. Sarcomas are rarely encountered in malignant effusions. There are only a few studies of cytomorphological features of sarcomas in serous effusions. Diagnostic subtyping of sarcomas in serous effusion specimens is very difficult. Tumour cells tend to round up in fluid specimens and the original cell morphology like spindled, pleomorphic, epithelioid or small round cell morphology is more difficult to evaluate [ 1 , 14 ]. Differential diagnosis includes melanoma, poorly differentiated carcinoma, malignant mesothelioma and other small round blue cell tumours. Melanomas and sarcomas are extremely rare in serous effusions and so is the case also with germ cell and sex cord‐stromal tumours. Diagnosis based only on cytomorphology is challenging. If morphology is classic and immunostains (Table  11 ) are confirmatory—report as malignant‐primary: mesothelioma. Clinical and imaging correlation is essential. If morphology is classic but immunostains are not confirmatory—report as suspicious for mesothelioma (SFM category). If morphology is not classic and immunostains are not confirmatory—report as atypical mesothelial proliferation (AUS category). Further investigation is advised [ 15 ]. Only epithelioid and mixed variants of malignant mesothelioma tend to exfoliate into serous cavities and thus the sarcomatoid and desmoplastic subtypes of mesothelioma are rarely encountered in effusion cytology specimens. There are many features that are helpful in distinguishing these three cell types regarding architecture, cohesion, cytoplasm, nucleus, nuclear/cytoplasmic ratio (NC ratio), nucleoli and IC profile (Table  4 ). Especially the evaluation of the quality of cytoplasm may be helpful (Table  9 ). Evaluating the cytoplasm in serous effusion cellular component. Abundant Scant Moderate Variable Vacuolated Eosinophilic Basophilic Clear Fibrillar Two‐tone Foamy Granular Inclusions Indistinct Blebs Prominent Cilia Submembranous vacuoles Regarding cytomorphology, adenocarcinoma typically shows variable sized groups of tightly cohesive cells with abundant and coarse vacuolated cytoplasm, a pleomorphic nucleus with prominent nucleoli and a high NC ratio. Mesothelial proliferation typically shows similar sized groups of cells with gaps or windows between individual cells. Roughly equal‐sized morules may be seen in mesothelioma although occasional giant clusters are also seen. The cells have basophilic, two‐tone cytoplasm with peripheral blebs and submembranous vacuoles while the nuclei are monomorphic with prominent single nucleoli in mesothelioma and multiple small nucleoli in reactive mesothelial cells. Mesothelial cells have similar cytomorphogical features, whether they are normal quiescent, reactive or malignant, but these features become clearer and marked in malignant mesothelioma. Macrophages are dispersed in loosely aggregated groups of cells with foamy cytoplasm, bean‐shaped and bland nucleus, no prominent nucleoli and a low NC ratio. Immunochemically these three cell types may be distinguished with a panel of immunochemical stains. A good panel to start with is to do two mesothelial markers like WT1 (Wilm's tumour 1) and calretinin, two epithelial markers like MOC31 and BerEP4 (both are antibodies targeting epithelial cell adhesion molecule) and one macrophage marker, for example CD68 (cluster of differentiation 68). Good adenocarcinoma markers are MUC4 (mucin 4) and claudin 4, and if available, they could be useful to be included in the panel. Other stains that may be considered in the second panel are PAS‐D (Periodic acid–Schiff with diastase), CEA (carcinoembryonic antigen) and depending on the sex, age and effusion site a panel of site‐specific stains is recommended (Tables  4 and 10 ). Ascertaining the primary organ site for adenocarcinoma in serous effusions. Abbreviations: 34BE12, cytokeratin 34 beta E12; BAP1, breast cancer 1‐associated protein 1; ca, carcinoma; CA125, cancer antigen 125; CAIX, carbonic anhydrase IX; CCRCC, clear cell renal cell carcinoma; CDX2, caudal type homeobox transcription factor 2; CK20, cytokeratin 20; GATA3, GATA3 binding protein; GCDFP‐15, gross cystic disease fluid protein 15; NKX3.1, homeobox protein NKX‐3.1; p40, p40 protein, also known as ΔNp63; p53, p53 protein; p63, p63 protein; PAX8, paired box gene 8; PRAP, prostatic acid phosphatase; PSA, prostate‐specific antigen; PSMA, prostate‐specific membrane antigen; RCC, renal cell carcinoma; SATB2, special AT‐rich sequence‐binding protein 2; TTF‐1, thyroid transcription factor 1; WT1, Wilm's tumour 1. Ancillary testing in cases of mesothelial proliferations. Abbreviations: BAP1, breast cancer 1‐associated protein 1; EMA, epithelial membrane antigen; FISH, fluorescence in situ hybridisation; IC, immunochemistry; MTAP, methylthioadenosine phosphorylase; NFM, negative for malignancy; P16/CDKN2A, p16INK4a/cyclin‐dependent kinase inhibitor 2A. It is important to remember IC pitfalls: GATA3 (GATA3‐binding protein) stains 50% of mesotheliomas. WT1 is positive in carcinomas of female genital tract and basal type of breast carcinoma and even in some squamous cell carcinomas. CK5/6 (cytokeratin 5/6) is positive in both squamous cell carcinoma (SQCC) and mesothelioma. Calretinin and D2‐40 (a monoclonal antibody that reacts to podoplanin) may also be positive in both SQCC and mesothelioma. Use p63 (p63 protein) or p40 (p40 protein, also known as ΔNp63) (positive in SQCC) and WT1 (positive in mesothelioma) to distinguish between SQCC and mesothelioma in this situation. Recently a new promising mesothelial marker, HEG1 (protein HEG homologue 1), has been introduced. As a pitfall it may be positive in ovarian carcinomas [ 16 ]. Apply TIS criteria. Consider patient related factors, especially sex and age. Take into account also effusion site, whether pleural, ascitic or pericardial. If the cytomorphology is compatible with metastatic adenocarcinoma—primary site needs to be investigated. Then probabilities according to age and sex may be applied when considering site‐specific markers (Table  10 ). In females with pleural or pericardial effusion the most likely primary sites of adenocarcinoma would be breast, lung and ovary and with ascitic fluid ovary, uterus, gastrointestinal and hepato‐biliary tracts. In males, the most likely primary site of adenocarcinoma in pleural/pericardial effusion fluid is lung and in ascitic fluid gastrointestinal and hepato‐biliary tracts [ 17 ]. If cytomorphology is not that of adenocarcinoma, consider small cell neuroendocrine carcinoma, melanoma, lymphoma, leukaemia, squamous cell carcinoma or sarcoma. Ancillary tests are usually needed, because of the overlapping patterns of malignant infiltrations. Applying TIS improves diagnostic accuracy and reproducibility and allows comparison of data and sharing the knowledge. In the future, new molecular markers and related technologies continue to emerge and make the diagnosis of effusion even more accurate, efficient, convenient and economical. We are already looking forward to the second edition of the TIS.

The International System for Serous Fluid Cytopathology (TIS) [ 1 ] standardises classification and nomenclature of effusion cytological specimens. It improves communication among pathologists, clinicians and patients. Each diagnostic category has an estimated risk of malignancy (ROM) and provides evidence‐based recommendations for clinical management (Figure  1 ). Here we provide a practical approach to serous effusion cytology based on the TIS. The International System for Serous Fluid Cytopathology diagnostic categories and clinical management. LBP, liquid‐based preparation.

The authors declare no conflicts of interest.