Pericardial fluid cytology: An analysis of 128 specimens over a 6-year period
Presented in part at the 59th American Society of Cytopathology Annual Meeting; November 4-8, 2011; Baltimore, MD.
Abstract
BACKGROUND
Pericardial fluid (PF) accumulates through various mechanisms and cytology is part of the workup to determine the specific etiology, primarily to rule in or rule out malignancy. To the best of the authors' knowledge, the current study is the largest systematic evaluation of PF cytology performed to date.
METHODS
PF specimens collected over 6 years were retrieved. Clinical history, laboratory, cytologic, and pericardial biopsy results were recorded.
RESULTS
A total of 128 PF specimens were obtained from 113 patients (56 males and 57 females), representing 4.5% of all fluids. Of these, 95 cases (74.2%) were benign, 2 (1.6%) had “severely atypical cells, ” and 31 cases (24.2%) were malignant. The most common etiologies for benign PF specimens were neoplasm (23.1%), idiopathic (19%), infection (14.7%), and connective tissue disease (12.6%). The most common neoplasm producing malignant PF was lung carcinoma, both in males (75%) and females (52.2%), with adenocarcinoma being the most common type (72.2%). In females, breast carcinoma was the second most common neoplasm (39.1%). Approximately 87.1% of patients with malignant PF specimens had a prior history of malignancy and approximately 32.7% underwent a concomitant pericardial biopsy. The false-negative rate for cytology was 14.7% (hematologic malignancies [2 cases], metastatic sarcoma [1 case], and sarcoidosis [1 case] not detected) and that for pericardial biopsy was 40% (metastatic carcinoma [4 cases] not detected).
CONCLUSIONS
PF specimens are uncommon. A specific interpretation is rendered in approximately 98.4% of cases. Lung carcinoma is the most common tumor to produce malignant PF in both males and females. Approximately 87.1% of patients with malignant PF have a known history of malignancy. Although cytology is superior to pericardial biopsy in diagnosing metastatic carcinoma, other tumors may go undetected in the PF. Cancer (Cancer Cytopathol) 2013;121:242–51. © 2013 American Cancer Society.
INTRODUCTION
Cytologic evaluation is just one aspect of the overall workup of pericardial fluid (PF), which, together with general chemical analysis and microbiology cultures, has as its main purpose the determination of the etiology of the PF. To rule in or rule out malignancy is the main contribution of the cytologic evaluation in the laboratory workup of PF. Several well-known conditions can produce a pericardial effusion such as infection, malignancy, connective tissue disease, pericardial injury, metabolic causes, heart disease, or idiopathic causes.1 Each effusion is treated based on the specific etiology and hemodynamic stability of the patient.7 With knowledge of the specific cause that triggered the accumulation of the pericardial effusion, clinicians can tailor the treatment to target that specific cause or simply provide supportive measures.
Although systematic evaluation of pleural and peritoneal fluid cytology is abundant in the literature,8 large series focusing on PF cytology are surprisingly sparse.11 There are several reasons for this discrepancy. First, pericardial effusions are uncommon in the cytology laboratory compared with the other 2 types of effusions. Unless the PF is accumulated in a large amount or produces hemodynamic compromise (ie, cardiac tamponade), the effusion is usually not tapped.1 Patients with small amounts of pericardial effusion can be completely asymptomatic.1 In addition, if a patient has a concomitant pleural and pericardial effusion, the pleural effusion is preferentially tapped unless there is hemodynamic compromise. Therefore, the collection of a large number of PF cytology cases for a systematic analysis is difficult to accomplish. However, with an appropriate sample, conclusions specific to PF cytology can be drawn. A systematic review of the literature revealed only 2 series focused on PF cytology, both of which were published > 15 years ago.11 To the best of our knowledge, the current study of 128 cases collected over 6 years is the largest series published in the literature to date, allowing us to gain insight into the specifics of PF cytology.
The objective of the current study was to analyze a large cohort of PF specimens from a single institution to determine the etiology of the effusions, the diagnostic usefulness of cytology evaluation, how the cytologic results correlate with pericardial biopsy, and the usefulness of immunohistochemical stains in the workup of cases. To provide a useful interpretation of the PF, it is necessary for practicing cytopathologists to know the most common conditions that can lead to PF accumulation and how cytologic evaluation can contribute in distinguishing between them.
MATERIALS AND METHODS
PF specimens collected between January 2005 and December 2010 were retrieved through a computerized search. The specimens had been collected through ultrasound or fluoroscopic-guided pericardiocentesis or at the time of pericardial window. Standard analysis of PF at the time of collection included general chemistry (glucose, protein, and lactate dehydrogenase levels), microbiology (aerobic and anaerobic cultures, acid-fast bacilli stain, mycobacterial culture, and fungal culture), and cytologic evaluation. Each specimen was received fresh in the cytology laboratory. If the specimens were cloudy, turbid, or bloody, a total of 10 mL from each specimen was poured off into a centrifuge tube and saline was added to bring the specimen up to 45 mL. Specimens with a clear appearance on visual inspection were used in their entirety. Specimens were centrifuged for 15 minutes at 1500 revolutions per minute. A drop from the sediment was placed in the cytospin chamber. One Diff-Quik–stained and 3 Papanicolaou-stained cytospin slides were prepared for each specimen and a cell block also was prepared when excess PF was available. Immunohistochemical stains were performed on the cell block in selected cases. PF cytology cases were originally reviewed, interpreted, and reported by different cytopathologists at the time of collection and initial diagnosis. Cases with a malignant cytologic diagnosis were reviewed retrospectively by both authors. Clinical history, laboratory and radiologic results, the original cytologic diagnosis, and all pertinent histologic diagnoses, if available, were recorded. In cases with concomitant pericardial biopsy, false-positive and false-negative rates were calculated using standard statistical methods.
RESULTS
A total of 128 PF specimens were obtained from 113 patients (56 males and 57 females) who ranged in age from 6 years to 85 years (mean, 52.6 years). PF samples comprised 4.5% of fluid specimens processed at our institution during the study period (1292 pleural and 1396 peritoneal fluid specimens). The amount of PF fluid received ranged from < 1 mL to 1150 mL (mean, 61.0 mL).
The cytologic microscopic interpretation was recorded into 3 general categories: “benign” (ie, no malignant cells identified), “severely atypical cells present, ” and “malignant cells present.” For each category, a more detailed description was included in the diagnosis that specifically stated the types of cells present. For example, the description for the benign category (“no malignant cells identified”) usually stated the presence of reactive mesothelial cells, acute and chronic inflammatory cells, and/or blood. For the malignant cases, tumor type and/or favored primary tumor site were stated based on the clinical history, prior pathology information, or immunohistochemical stains performed on the cell block. Cases recorded as “severely atypical cells present” were paucicellular, comprised of rare cells with atypical cytological features, and a definitive diagnosis (benign vs malignant) could not be rendered.
The vast majority of PF specimens (95 cases; 74.2%) were classified as benign, negative for malignancy. In only 2 cases (1.6%) was a cytologic interpretation of “severely atypical cells present” rendered, and malignant cells were present in the PF specimen in 31 cases (24.2%).
PF Specimens With Benign Cytologic Interpretation
There were 95 PF specimens with a benign cytologic interpretation collected from 83 patients. Duplicate samples resulted from 7 patients who underwent repeated cytologic evaluation of PF draining from the pigtail catheter placed at the time of the initial pericardiocentesis (5 patients with 2 PF specimens and 2 patients with 3 PF specimens) and 3 patients had chronic recurrent effusion accumulated at an interval of 3 months to 4 months. The interpretation did not change with repeated cytologic evaluations of the PF. There was a wide range of etiologies for the benign effusions (Table 1), as detailed below.
| Etiology | Total No. of Cases |
|---|---|
| Neoplasms (no. of cases; % of total) | 22 (23.1%) |
| Neoplasm in patients undergoing chemotherapy and/or RT (17 cases; 77.2%) | |
| Primary lung carcinoma (9; 41%) | |
| Small cell carcinoma (5) | |
| Adenocarcinoma (3) | |
| Squamous cell carcinoma (1) | |
| Metastases to lung/mediastinal lymph nodes (5; 22.7%) | |
| Soft tissue sarcoma (1) | |
| Rectal adenocarcinoma (1) | |
| Appendiceal adenocarcinoma (1) | |
| Renal cell carcinoma (1) | |
| Oropharyngeal squamous cell carcinoma (1) | |
| Mediastinal yolk sac tumor (1; 4.5%) | |
| Esophageal carcinoma (1; 4.5%) | |
| Breast carcinoma (1; 4.5%) | |
| Other mechanisms (4 cases; 18.3%) | |
| Multiple myeloma (2) | |
| Chronic lymphocytic leukemia (1) | |
| Acute myeloid leukemia (1) | |
| Primary cardiac lymphoma (1; 4.5%) | |
| Idiopathic | 18 (19%) |
| Infections | 14 (14.7%) |
| Connective tissue diseases | 12 (12.6%) |
| Systemic lupus erythematosus (8) | |
| Scleroderma (3) | |
| Sjogren syndrome (1) | |
| Pericardial injury syndromes | 8 (8.4%) |
| Trauma to chest (4) | |
| Stab wound (2) | |
| Gunshot wound (1) | |
| Motor vehicle accident (1) | |
| Recent acute myocardial infarction (2) | |
| Pericardiotomy for closure of patent foramen ovale (1) | |
| Perforated ventricle after pacer placement (1) | |
| Metabolic causes | 6 (6.3%) |
| Uremia (5) | |
| Hypothyroidism (1) | |
| Medication induced | 6 (6.3%) |
| Minoxidil (3) | |
| Tacrolimus (3) | |
| Sarcoidosis | 2 (2.1%) |
| Hypereosinophilic syndrome | 1 (1.2%) |
| Multifactorial | 6 (6.3%) |
- Abbreviations: PF, pericardial fluid; RT, radiotherapy.
Neoplasms
It is interesting to note that neoplasms represented the most common cause of benign pericardial effusions at the study institution (23.1%). The vast majority of these cases were from patients with a prior or concurrent history of malignancy for which they received systemic chemotherapy and radiotherapy to the chest (77.2%). For example, 9 patients (41%) had advanced stage primary lung carcinoma, with small cell carcinoma being the most common primary lung carcinoma associated with a benign pericardial effusion while receiving treatment. Another 5 patients (22.7%) had a prior history of malignancy diagnosed elsewhere in the body that resulted in metastases to the lung and/or mediastinal lymph nodes that were treated with palliative radiotherapy to the chest. Three patients (13.5%) had a prior history of a treated malignancy in the chest area (esophageal and breast carcinoma and mediastinal yolk sac tumor) in which the disease was locally controlled with no direct involvement of the pericardium, and the PF was considered to be due to treatment effect.
In a minority of patients with a history of neoplasia (18.3%) the PF accumulated through other mechanisms that were not directly related to chemotherapy and/or radiotherapy. For example, in 1 patient with multiple myeloma, the disease involved the mediastinum and subsequent pericardial biopsy confirmed direct pericardial involvement, whereas in the other patient with multiple myeloma the effusion was considered to be the result of uremia secondary to renal involvement by multiple myeloma. In the patient with acute myeloid leukemia, the hemorrhagic pericardial effusion was considered to be due to thrombocytopenia.
Primary malignant tumors of the pericardium were very uncommon, with only 1 case of primary cardiac lymphoma involving the pericardium reported in the current study (4.5%). The patient in this case had a benign pericardial effusion and no ancillary studies were performed. However, the concomitant pericardial biopsy demonstrated the presence of a follicular lymphoma, which was confirmed by immunohistochemical studies and fluorescence in situ hybridization for t(14;18).
Of the 22 benign PF specimens associated with a neoplasm, 9 of the patients underwent a concomitant pericardial biopsy (40.9%). Six cases had a negative result on the pericardial biopsy (2 primary lung adenocarcinomas, 1 metastatic rectal adenocarcinoma, 1 metastatic renal cell carcinoma [RCC], 1 metastatic oropharyngeal squamous cell carcinoma, and 1 esophageal carcinoma). Three cases had a positive pericardial biopsy (1 case each of metastatic soft tissue sarcoma, primary cardiac lymphoma, and multiple myeloma).
Idiopathic
The second largest group of patients with benign PF specimens is represented by those with idiopathic cases (19%), in whom no cause for the effusion could be identified based on the clinical history or comorbid conditions, or from routine evaluation of the fluid.
Infections
Infections represented the third most common cause of benign PF specimens (14.7% of all benign PFs), with the vast majority (71.4%) being bacterial in nature. Table 2 summarizes the various bacteria identified from PF cultures and/or blood cultures. The least common etiologies noted were viral and atypical mycobacteria. Viral serologies were not performed routinely, except in immunocompromised patients. Two patients (14.3%) were identified as having elevated serum titers for Coxsackie virus type A. Both of these patients were immunosuppressed; 1 patient had acute myelogenous leukemia and was status post-bone marrow transplant, and the other patient was an 8-year-old child with hemophagocytic lymphohistiocytosis. Mycobacterial infections were uncommon in the current series as a cause for PF (14.3% of all infectious PFs), with none being produced by Mycobacterium tuberculosis. One patient was a 57-year-old female with chronic neutropenia of unknown etiology who was diagnosed with Mycobacterium kansasii by culture from a mediastinal lymph node biopsy performed at an outside hospital, and by respiratory culture performed at the study institution. The second patient was a 37-year-old man with an untreated human immunodeficiency virus infection who had disseminated Mycobacterium avium complex diagnosed from blood culture and bone marrow biopsy examination.
| Etiologic Organism (No. of Cases and % of All Infectious PFs) | PF Cultures | Blood Cultures | Comments |
|---|---|---|---|
| Bacterial (71.4%) | |||
| Vancomycin-resistant Enterococcus (2) | + | + | |
| Methicillin-resistant Staphylococcus aureus (1) | - | + | Intravenous drug use with endocarditis |
| Methicillin-sensitive Staphylococcus aureus (1) | + | + | Septic shock |
| Staphylococcus spp. coagulase negative (2) | + | NA | |
| Escherichia coli (1) | + | NA | |
| Klebsiella oxytoca (1) | + | NA | |
| Gram-positive cocci (Gemella spp.) (1) | + | NA | |
| Presumed bacterial, specific organism not identified (1) | - | - | Mediastinal abscess in a patient with uncontrolled diabetes mellitusa |
| Viral (14.3%) | - | NA | Elevated serum titers |
| Coxsackie virus (2) | |||
| Atypical mycobacteria (14.3%) | |||
| Mycobacterium avium complex (1) | NA | + | Disseminated infection in untreated HIV-positive patient |
| Mycobacterium kansasii (1) | - | NA | Large mediastinal lymphadenopathy |
- Abbreviations: +, positive; −, negative; HIV, human immunodeficiency virus; NA, not applicable; PF, pericardial fluids; spp, species.
- a The patient had a white blood cell count of 20,800/µL and pneumomediastinum. Antibiotic therapy was initiated and thoracoscopy was performed to drain the empyema a few days prior to the development of the pericardial effusion.
Other Conditions
The other etiologies of benign PFs at the study institution were connective tissue diseases (12.6%), pericardial injury syndromes (8.4%), metabolic causes (6.3%), medication (6.3%), sarcoidosis (2.1%), and hypereosinophilic syndrome (1.2%), which are well known in the literature to be associated with a pericardial effusion. The unusual case of a patient who was status post-liver transplant (with 3 PFs collected days apart) in whom the fluid accumulation was attributed to treatment with tacrolimus, because the dose of tacrolimus was increased just 3 weeks before the development of symptoms of pericardial effusion, should be noted.
Multifactorial
Six patients (6.3%) presented with several comorbid conditions simultaneously, each of which could have been responsible for the PF; however, not one stood out as being the main cause of fluid accumulation. The association of comorbid conditions included uremia, treatment with minoxidil, receipt of radiotherapy for lung carcinoma 4 years prior, infection, hypothyroidism, or undetermined rheumatologic condition.
Despite the long list of conditions that led to the pericardial effusion in these cases, all benign PFs had similar, fairly nonspecific findings on cytologic examination. Specifically, the PFs contained mesothelial cells, which were usually reactive in nature, with a variable mixture of acute and chronic inflammatory cells and macrophages. In addition, hemorrhagic PFs contained hemosiderin-laden macrophages and numerous red blood cells. The mesothelial cells were easily recognized as such, with no need to perform confirmatory immunohistochemical stains on the cell block. Nine of the 14 infectious PFs (64.3%) demonstrated abundant acute fibrinopurulent exudate.
Cytologic Interpretation of PF Specimens With “Severely Atypical Cells”
In 2 cases (1.6%), a cytologic interpretation of “severely atypical cells present” was rendered. One case was that of a 67-year-old man with a lung mass who presented with cardiac tamponade. The PF demonstrated rare clusters of atypical cells and no further workup was performed on the cell block material. A fine-needle aspiration of a right posterior cervical lymph node performed 1 day later indicated metastatic small cell carcinoma. The second case was that of a 45-year-old female with no known history of malignancy who also presented with cardiac tamponade. Rare groups of atypical cells were noted in the PF; however, no further ancillary studies were performed on the cell block. Two months later, the patient presented with enlarging pleural and peritoneal effusions which demonstrated a metastatic adenocarcinoma. However, no primary tumor site was identified.
PF Specimens With a Malignant Cytologic Interpretation
There were 31 PF specimens with a malignant cytologic interpretation that were collected from 28 patients (Table 3). Four patients (1 male and 3 females) had repeated PF cytology at a 1-day interval because of continuous drainage from the pigtail catheter. The interpretation did not change with repeated cytologic evaluations of the PF.
| Primary Site | Males (8 Cases; 25.8%) | Females (23 Cases; 74.2%) |
|---|---|---|
| Lung | 6 (75.0%) | 12 (52.3%) |
| Adenocarcinoma | 5 | 8 |
| Non-small cell carcinoma, NOS | — | 3 |
| Large cell neuroendocrine carcinoma | — | 1 |
| Small cell carcinoma | 1 | — |
| Breast | — | 9 (39.1%) |
| Pancreas | 1 (12.5%) | — |
| Uterine serous carcinoma | — | 1 (4.3%) |
| Diffuse large B-cell lymphoma | — | 1 (4.3%) |
| Adenocarcinoma of unknown primary tumor | 1 (12.5%) | — |
- Abbreviations: NOS, not otherwise specified; PF, pericardial fluid.
All malignant PF cases, with the exception of 1 case, were due to a very short list of metastatic carcinomas (those of the lung, breast, pancreas, and gynecologic tract), as indicated in Table 3. There were no cases of malignant mesothelioma involving the pericardium.
It is interesting to note that malignant PF specimens were more common in females, who represented 74.2% of patients with malignant effusions. This was due primarily to cases of lung carcinoma, which represented > 50% of the malignant PF specimens detected in females. In fact, lung carcinoma was the most common primary carcinoma that produced pericardial metastases in both males and females (6 cases in males [75.0%] and 12 cases in females [52.3%]). The most common type of lung carcinoma metastatic to the pericardium in both males and females was adenocarcinoma (13 cases, representing 72.2% of all lung carcinomas). In males, metastatic pancreatic adenocarcinoma was a distant second (1 case; 12.5%), whereas in females, the breast was the second most common site of metastatic carcinoma (9 cases; 39.1%). The only nonepithelial malignant neoplasm involving the PF was a case of diffuse large B-cell malignant lymphoma (4.3%).
In the vast majority of cases (87.1%), patients had a known, pathologically proven, prior history of malignancy, specifically lung, breast, uterine, and pancreatic carcinoma. Only 4 patients (12.9%) had no documented, pathologically proven prior history of malignancy at the time of PF cytology; however, 2 of these patients were known to have a lung mass based on the imaging studies performed at the time of PF collection. One patient diagnosed with diffuse large B-cell lymphoma had newly diagnosed human immunodeficiency virus infection with diffuse lymphadenopathy, fever, and weight loss. The fourth patient had multiple lung nodules as well as liver and bone metastases with no clear primary tumor site identified.
Cytological evaluation of the PF specimens revealed 3 distinct morphologic patterns in the case of metastatic carcinomas.
One pattern consisted of cellular specimens with malignant cells arranged either in 3-dimensional groups with a depth of focus or dispersed as single cells, or a combination of these 2 patterns. The cytoplasm of malignant cells ranged from scant to moderate, and was usually vacuolated. Characteristically, tumor cell nuclei were markedly enlarged when compared with the benign mesothelial cells and displayed significant pleomorphism, with coarse chromatin, irregular nuclear membranes, and visible nucleoli (Fig. 1). This pattern was the most common, being present in all cases of metastatic lung carcinoma (with the exception of small cell carcinoma), pancreatic adenocarcinoma, uterine serous carcinoma, adenocarcinoma of unknown primary tumor, and 44.4% of metastatic breast carcinoma cases.
Metastatic lung adenocarcinoma is shown. Malignant cells are arranged in groups or dispersed as single cells. Characteristically, tumor cells have markedly enlarged and pleomorphic nuclei with coarse, dark chromatin when compared with the benign mesothelial cells (Papanicolaou stain, × 600).
The second pattern was recognized in only slightly more than one-half of metastatic breast carcinoma cases (55.6%) and was comprised of cellular specimens with malignant cells arranged in 3-dimensional groups with a sharply demarcated “community border” or tubular/duct-like arrangements. Single, discohesive cells were not the predominant feature (Fig. 2). Another characteristic of this pattern was the marked uniformity of the tumor cells with minimal nuclear pleomorphism. In addition, the tumor cell nuclei were not enlarged when compared with the benign mesothelial cells, but they had dark, coarse nuclear chromatin (Fig. 3).
Metastatic breast carcinoma is shown, demonstrating a cellular specimen with numerous 3-dimensional cohesive clusters of malignant cells. Note the tubular/duct-like arrangement of the malignant cells within the clusters (Papanicolaou stain, × 100).
Metastatic breast carcinoma is shown. At a higher magnification, tumor cells are noted to be uniform with minimal pleomorphism, although the nuclear chromatin is distinctly dark and coarse (Papanicolaou stain, × 400).
The only case of metastatic small cell carcinoma in the current study had a pattern that at low power mimicked a benign lymphocytic effusion because the tumor cells were predominantly arranged singly or in very small aggregates (2-10 cells) (Fig. 4). However, at higher magnification, nuclear molding was appreciated in the small cellular aggregates. The tumor cells appeared nearly as “naked nuclei” because they barely had any visible cytoplasm. Another distinctive feature noted at higher magnification was variability in nuclear size and chromatin consistency ranging from small, apoptotic bodies with dark, smudged chromatin to larger nuclei with crisper chromatin details. Nevertheless, the size of the tumor cell nuclei was smaller than that of the nuclei of mesothelial cells (Fig. 5).
Metastatic small cell carcinoma is shown. At a lower magnification, this pattern mimics a benign lymphocytic effusion because the tumor cells scattered in the background as single cells may be confused with inflammatory cells (Papanicolaou stain, × 40).
Metastatic small cell carcinoma is shown. However, at a higher magnification, one can appreciate a morphologic spectrum ranging from small apoptotic bodies to larger cells that are nearly devoid of cytoplasm and with nuclear molding (Papanicolaou stain, × 1000).
Although most commonly encountered, the first morphologic pattern is nonspecific in suggesting a primary tumor site on morphologic grounds alone. This is in contrast with the other 2 patterns, which, when recognized, are very helpful in recognizing metastatic breast carcinoma or small cell carcinoma, respectively.
In addition to routine cytology, immunohistochemical stains were performed in 12 cases (38.7%). It appears that the immunohistochemical stains were performed primarily to confirm the presence of malignant cells and to distinguish them from the reactive mesothelial cells rather than to identify the primary tumor site. The antibodies used included classic adenocarcinoma (MOC-31, BerEP4, and B72.3), mesothelial (calretinin, cytokeratin 5/6 [CK5/6], and Wilms Tumor-1 [WT-1]) markers with the addition of more specific markers depending on the clinical history (eg, thyroid transcription factor 1 [TTF-1], cancer antigen 19-9 [CA 19-9], estrogen and progesterone receptors, BRST-2, human epidermal growth factor receptor 2 [HER2] neu, and p63).
Correlation of PF Cytology With Pericardial Biopsy
Twenty-seven patients with benign PF cytology (corresponding to 33 PF specimens) had a concomitant pericardial biopsy performed at the time of pericardial window (23.9%). Four patients (14.8%) had a false-negative cytologic interpretation because the pericardial biopsy demonstrated the presence of a specific condition such as metastatic soft tissue sarcoma to the mediastinum with direct involvement of the pericardium, multiple myeloma involving the mediastinum with secondary involvement of the pericardium, primary cardiac follicular lymphoma, and nonnecrotizing granulomas consistent with sarcoidosis.
The 2 patients with severely atypical cells in their PF specimens did not undergo a pericardial biopsy.
Ten patients with malignant PF cytology (corresponding to 12 PF specimens) underwent a concomitant pericardial biopsy (8.8%). Four patients (40%) had a false-negative pericardial biopsy result demonstrating nonspecific chronic pericarditis, although the PF cytology identified metastatic carcinoma. In 1 patient (10%), a few severely atypical cells were present in the pericardial biopsy but could not be characterized further because they were not present on deeper levels. In the 5 patients with positive pericardial biopsy (50%), the cytologic diagnosis was confirmed.
Overall, the performance of PF cytology in detecting malignancy was found to be better than that of pericardial biopsy, with a sensitivity of 71% and a specificity of 100% (compared with 64% sensitivity and 85% specificity for the pericardial biopsy).
DISCUSSION
To the best of our knowledge, the current study of 128 PFs specimens is the largest series focusing on PF cytology published to date in the literature.11 Our experience is that the vast majority of cases (98.4%) can be easily classified as either benign or malignant. Cases with a noncommittal diagnosis of “atypical cells, suspicious” are very uncommon, in general due to the limited number of cells present on the cytospin slides and cell block.
It was intriguing to observe in the current series that the most common cause of pericardial effusion at the study institution was malignancy, which can produce both benign (23.1%) and malignant (24.2%) PF specimens. The mechanisms that lead to PF accumulation in the face of malignancy are multiple, including treatment effect (in particular chest radiotherapy), direct extension of the tumor to the pericardium, hematogenous and lymphangitic tumor spread to the pericardium, or other consequences of the malignant process (eg, uremia, thrombocytopenia).13
With regard to malignant PF specimens, we were able to make several interesting observations that to the best of our knowledge have not been mentioned in previous studies.11
In the current study, malignant PF specimens were found to be more common in females (74.2% of all malignant effusions). It is interesting to note that this finding is actually due to metastatic lung carcinoma, not breast carcinoma as one may presume.
Lung carcinoma is by far the most common malignancy that is metastatic to the pericardium in both males (75.0%) and females (52.3%), with adenocarcinoma being the most common histologic type encountered (72.2%).
Metastatic breast carcinoma to the pericardium appears to never be the first manifestation of the disease, because all of the 9 patients in the current study with malignant PF due to breast carcinoma were known to have this disease.
Malignant PF is rarely the first manifestation of the disease. In the vast majority of malignant PF specimens (87.1%), patients had a known, pathologically proven history of malignancy.
In the current study, various other etiologic conditions were identified in the benign PF specimens in addition to neoplasia, which is in keeping with prior literature.3 However, although the etiologic spectrum of benign PF is broad, the cytologic findings are limited and nonspecific. Reactive mesothelial cells in benign PF specimens are usually recognized as such from the routine preparations without the need for confirmatory immunohistochemical stains. Typically, reactive mesothelial cells in benign effusions are present as single cells or small 3-dimensional groups with a scalloped, knobby contour and demonstrate a spectrum of morphological changes. They have a round-to-oval shape with a moderate amount of cytoplasm with prominent borders with a “lacy skirt” periphery; a slit-like space called a “window” forms between 2 adjacent cells due to the presence of surface slender microvilli. The cytoplasm has a biphasic staining pattern with a denser endoplasm and a more lucent, clear ectoplasm. Nuclei are typically round-to-oval and centrally located, with finely distributed nuclear chromatin; binucleated and multinucleated cells can be noted in benign effusions.15
Conversely, malignant mesothelial cells appear as 3-dimensional cell balls (or morules), which are crowded cellular groups of various sizes and shapes in which the nuclei are overlapped and individual cellular details are difficult to appreciate. The lobulated contour of these cell balls is a clue toward their mesothelial origin. Another characteristic feature is the presence of giant atypical mesothelial cells. These are isolated, multinucleated giant cells with significant nuclear atypia, in which the nuclear-to-cytoplasmic ratio is low. This is due to the presence of abundant cytoplasm with the characteristics of mesothelial cells cytoplasm, that is an optically dense central portion and clear periphery.15
The presence of other cytologic elements in addition to the reactive mesothelial cells, such as macrophages or mixed acute inflammatory infiltrate, does not provide any clues to the specific etiology of the benign effusion. One exception to this observation is represented by the presence of abundant acute fibrinopurulent exudate in 64.3% of bacterial infections.
The distinction between reactive mesothelial cells and malignant cells is usually straightforward. Three distinct cytomorphologic patterns are recognized in PF specimens involved with metastatic carcinoma. It is worth highlighting the pattern of small, single malignant cells observed in cases of metastatic lobular carcinoma of the breast and small cell carcinoma.16 This pattern is difficult to recognize and the malignant cells can be overlooked, especially if the prior history of malignancy is not known.10 In both situations, a single file or linear arrangement of molded tumor cells can be noted. The malignant cells of lobular carcinoma of the breast also have an eccentrically placed nucleus and small cytoplasmic vacuoles.16
In the case of malignant PF specimens, immunohistochemical stains are typically needed to confirm the presence of malignant cells through classic adenocarcinoma and mesothelial markers, rather than to determine the primary tumor site.19 This approach is justified by the finding that approximately 93.5% of malignant PF specimens had either a pathologically proven, known history of malignancy or a potential primary tumor site identified through imaging studies (ie, a lung mass in 2 patients) at the time of PF cytologic examination.
Repeated cytologic evaluations occurred in 14 patients (12.4%), involving both benign and malignant PF specimens, and yielded similar interpretations. Analogous results were obtained by Wiener et al in their study.12 Repeated cytologic examinations to increase the probability of detecting malignancy are not necessary in the workup of PF specimens.22
It appears that cytology alone was considered sufficient for pathologic evaluation in the majority of cases, with only one-third of patients (32.7%) undergoing a concomitant pericardial biopsy. Cytologic evaluation was found to correlate well with pericardial biopsy. The rate of false-negative results for cytology was 14.7% in the current study. Cases not identified through cytology included hematologic malignancies, metastatic sarcoma, and sarcoidosis involving the pericardium. This discrepancy could be due to the fact that the effusion is produced through an exudative mechanism but the malignant cells and/or granulomas do not actually shed as easily into the accumulated fluid as carcinoma cells. Conversely, pericardial biopsy had a 40% false-negative rate, indicating nonspecific chronic pericarditis changes when the PF cytology demonstrated metastatic carcinoma. This could be due to sampling error, because the biopsy represents a smaller portion of the pericardium. It appears that a combination of PF cytology and pericardial biopsy, if clinically feasible to obtain, would yield a better sensitivity in diagnosing malignancy.
PFs appear to be unusual in a cytology laboratory, because they represented a mere 4.5% of all body cavity fluids at the study institution. The cytologic evaluation of these specimens is not very different from that of other fluids (pleural or peritoneal) in terms of cytomorphology and the need for immunohistochemical stains. However, there are very specific conditions that produce PF accumulation. Our main role as cytopathologists examining PF specimens is to identify the malignant cases, keeping in mind that some tumors (sarcomas, hematologic malignancies) may go undetected in the PF. Conversely, cytology is superior to pericardial biopsy in diagnosing metastatic carcinomas.
Acknowledgments
We thank Mrs. Patricia R. Strong, Director of the Writing Center and Assistant Professor at the University College at Virginia Commonwealth University in Richmond for her critical review of the article and useful suggestions.
FUNDING SUPPORT
No specific funding was disclosed.
CONFLICT OF INTEREST DISCLOSURES
The authors made no disclosures.
