Cerebrospinal fluid cytologic findings in multiple sclerosis. A comparison between patient subgroups

OBJECTIVE: To compare cytologic findings in cerebrospinal fluid (CSF) in various subgroups of multiple sclerosis (MS) patients. STUDY DESIGN: CSF from 77 patients with clinically definitive or probable MS was examined by means of qualitative cytology. After the cell count was determined in a Fuchs-Rosenthal chamber, slides were prepared by the cytosedimentation method and stained with May-Grünwald-Giemsa stain and oil red O and, whenever possible, with Papanicolaou stain and toluidine blue. In addition to the differential cell count, the lymphocyte/monocyte ratio, percentage of activated forms in the lymphocytic and monocytic series, presence and percentage of lymphoplasmacytes and mature plasma cells, presence of lipophages, lymphophages and presence of mitotic figures were evaluated. RESULTS: The following statistically significant differences were found between the various MS subgroups: (1) higher prevalence of mitotic figures in the primary progressive MS subgroup; (2) higher prevalence of foam cells and lymphophages and lower prevalence of CSF pleocytosis in more severely disabled patients; (3) lower cell count, lower prevalence of CSF pleocytosis, lower lymphocyte/monocyte ratio and lower prevalence of lymphoplasmacytes in treated patients; and (4) higher prevalence of mature plasma cells and lipophages in MS patients with disease of longer duration. CONCLUSION: The differences observed in the various MS subgroups may reflect certain aspects of MS pathogenesis. Qualitative CSF cytology may therefore be useful for both clinicians and neuroimmunologists. Qualitative cytology of CSF is an important diagnostic method that should never be omitted from an examination of CSF from patients with MS.     

Cytologic features of cerebrospinal fluid in Lyme disease

The cytomorphologic findings in the cerebrospinal fluid (CSF) of four patients with lymphocytic meningoradiculitis (Lyme disease) related to a tick bite (due in at least two cases to Borrelia burgdorferi) are reported. In all cases, the May-Grünwald-Giemsa-stained centrifuge preparations of the CSF showed a cellular pattern consisting of a lymphocytic pleocytosis composed mainly of immunoblasts and plasma cells associated with numerous foamy macrophages. Direct immunofluorescence studies in one case showed the polyclonal pattern of the immunoblasts and the plasma cells. These CSF findings can be considered as suggestive of the spirochetal origin of a lymphocytic meningitis following a tick bite.     

Increased prothrombin, apolipoprotein A-IV, and haptoglobin in the cerebrospinal fluid of patients with Huntington's disease

Huntington's disease (HD) is a progressive neurodegenerative disease caused by an unstable CAG trinucleotide repeat expansion. The need for biomarkers of onset and progression in HD is imperative, since currently reliable outcome measures are lacking. We used two-dimensional electrophoresis and mass spectrometry to analyze the proteome profiles in cerebrospinal fluid (CSF) of 6 pairs of HD patients and controls. Prothrombin, apolipoprotein A-IV (Apo A-IV) and haptoglobin were elevated in CSF of the HD patients in comparison with the controls. We used western blot as a semi-quantified measurement for prothrombin and Apo A-IV, as well as enzyme linked immunosorbent assay (ELISA) for measurement of haptoglobin, in 9 HD patients and 9 controls. The albumin quotient (Qalb), a marker of blood-brain barrier (BBB) function, was not different between the HD patients and the controls. The ratios of CSF prothrombin/albumin (prothrombin/Alb) and Apo A-IV/albumin (Apo A-IV/Alb), and haptoglobin level were significantly elevated in HD. The ratio of CSF prothrombin/Alb significantly correlated with the disease severity assessed by Unified Huntington's Disease Rating Scale (UHDRS). The results implicate that increased CSF prothrombin, Apo A-IV, and haptoglobin may be involved in pathogenesis of HD and may serve as potential biomarkers for HD.     

Characterization of four lipoprotein classes in human cerebrospinal fluid.

Lipoprotein metabolism in brain has not yet been fully elucidated, although there are a few reports concerning lipids in the brain and lipoproteins and apolipoproteins in the cerebrospinal fluid (CSF). To establish normal levels of lipoproteins in human CSF, total cholesterol, phospholipids, and fatty acids as well as apolipoprotein E (apoE) and apoA-I levels were determined in CSF samples from 216 individuals. For particle characterization, lipoproteins from human CSF were isolated by affinity chromatography and analyzed for size, lipid and apolipoprotein composition. Two consecutive immunoaffinity columns with antibodies, first against apoE and subsequently against apoA-I, were used to define four distinct lipoprotein classes. The major lipoprotein fraction consisted of particles of 13;-20 nm containing apoE and apoA-I as well as apoA-IV, apoD, apoH, and apoJ. In the second particle class (13;-18 nm) mainly apoA-I and apoA-II but no apoE was detected. Third, there was a small number of large particles (18;-22 nm) containing no apoA-I but apoE associated with apoA-IV, apoD, and apoJ. In the unbound fraction we detected small particles (10;-12 nm) with low lipid content containing apoA-IV, apoD, apoH, and apoJ. In summary, we established lipid and apolipoprotein levels in CSF in a large group of individuals and described four distinct lipoprotein classes in human CSF, differing in their apolipoprotein pattern, lipid composition, and size. On the basis of our own data and previous findings from other groups, we propose a classification of CSF lipoproteins.     

Cerebrospinal fluid infiltration in Hodgkin lymphoma: a case report

BACKGROUND: Central nervous system (CNS) involvement by Hodgkin lympboma is a rare event. Involvement of the cerebrospinal fluid (CSF) in such cases is even more uncommon. We report a case of Hodgkin lymphoma in which the patient developed infiltration of the CSF while on chemotherapy. CASE: A 45-year-old woman was diagnosed with Hodgkin lymphoma by fine needle aspiration and subsequent biopsy of the cervical lymph node. She complained of headache during the course of chemotherapy, for which CSF examination was undertaken. Cytocentrifuge sediment of the CSF revealed marked eosinophilic pleocytosis, accompanied by scattered monocytes, polymorpbs, lymphocytes, plasma cells and histiocytes. An occasional large mononudlear cell with a large, round nucleus and prominent irregular nucleolus with a moderate amount of basophilic cytoplasm conformning to the morphology of Hodgkin's cells was noted. Binucleated Reed-Sternberg cells were not seen. Following intratbecal methotrexate, a reduction in the cellular infiltrate was observed. CONCLUSION: CSF cytology is important for the diagnosis of CNS involvement by Hodgkin lymphoma and may be positive before lesions can be visualized by magnetic resonance imaging or computed tomograpby scans.     

Is CSF cytology a useful diagnostic procedure in staging paediatric CNS tumours?

Objectives:  To evaluate whether there are any factors that predict malignant cells being found in paediatric cerebrospinal fluid (CSF) samples. To determine whether CSF provides useful staging information not provided by magnetic resonance imaging (MRI) in paediatric patients with primary central nervous system (CNS) malignancy.
Methods:  We compared the CSF cytology and spinal MRI staging results in paediatric patients with primary CNS malignancy at a UK tertiary referral centre, over a decade.
Results:  Of 159 CSF samples, 72 samples were from 72 patients with primary CNS malignancy with spinal MRI available for comparison. Eight of these 72 had positive cytology (seven malignant and one suspicious). All had a high clinical suspicion of tumour at the time of sampling. Of the 72 patients, only two had evidence of CSF spread on MRI spinal staging and CSF cytology; ten had MRI without cytological evidence and six had cytological without MRI evidence.
Conclusions:  In paediatric patients with primary CNS tumours, CSF cytology provides useful staging information. Spinal MRI alone may miss some patients with CSF spread who would be identified with CSF cytology.     

Diagnosis of metastatic tumors in cerebrospinal fluid samples using thin-layer cytology

OBJECTIVE: To evaluate the application of ThinPrep liquid-based cytology (LBC) and present our experience using LBC in the diagnosis of metastatic tumors in cerebrospinal fluid (CSF) samples. STUDY DESIGN: We examined 38 cytologic specimens of CSF, processed by ThinPrep technique. Of these, 18 presented with a previously diagnosed primary malignancy. Various immunocytochemical markers were performed. RESULTS: ThinPrep technology provided preservation of cytomorphologic features, high cellularity per slide and clear background. Analysis revealed 8 breast carcinomas, 5 lung carcinomas, 4 lymphomas, 3 adenocarcinomas of the gastrointestinal tract, 1 squamous cell carcinoma of uterine cervix and 1 urinary bladder carcinoma. Fifteen samples were negative for malignancy. CONCLUSION: CSF cytology is the only examination that verifies the presence of malignancy. Thin monolayer technology is suggested as an appropriate diagnostic method for metastatic tumors in CSF in everyday routine and seems to be a valuable tool for further management and planning of treatment.     

Castleman's disease, plasma-cell type. Diagnosis of central nervous system involvement by cerebrospinal fluid cytology

A case of Castleman's disease of the plasma-cell type is reported in which central nervous system (CNS) involvement was diagnosed by cerebrospinal fluid (CSF) cytology. The patient had multicentric disease with constitutional symptoms, immunologic abnormalities and peripheral blood cytopenias requiring cytotoxic agents and steroids for treatment. CNS symptoms and diagnostic cytologic findings in CSF occurred in the absence of morphologic lesions demonstrable by computed tomography or magnetic resonance imaging of the brain.     

Chronic meningitis and Lyme disease in Sweden.

We studied 35 patients with chronic meningitis. The neurological abnormalities included aseptic meningitis, cranial neuropathy (mostly facial palsy), motor and sensory peripheral radiculoneuropathy, and myelitis. Neurological symptoms were sometimes preceded by erythema chronicum migrans or an insect bite and were often accompanied by fever, malaise, profound fatigue, and weight loss. The cerebrospinal fluid (CSF) abnormalities consisted of a predominantly mononuclear pleocytosis, an elevated CSF protein (mean 2.3 g/l), intrathecal synthesis of oligoclonal immunoglobulin G, and, in half of the patients, a fall in the CSF/blood glucose ratio. High antibody titers to the Lyme spirochete and the Swedish Ixodes ricinus spirochete were demonstrated by immunofluorescence in 26 of the 35 patients. By imprint immunofixation of electrofocused samples of serum and CSF, intrathecal production of oligoclonal Lyme-spirochete-specific IgG was demonstrated in one patient with chronic meningitis. Four sequential paired samples of serum and CSF from this patient showed local synthesis of spirochete-specific antibodies in CSF. The 35 patients improved or recovered, sometimes dramatically, during a two-week course of intravenous penicillin G.     

Neurotensin-like immunoreactivity (NTLI) concentration in the cerebrospinal fluid of children and its alteration in a febrile aseptic meningitis

Neurotensin-like immunoreactivity (NTLI) concentrations in the cerebrospinal fluid (CSF) of normal children and patients with febrile aseptic meningitis, aged 7 months to 15 years, were studied. The NTLI concentrations in CSF of 27 children with normal CSF findings were 160.1 +/- 54.6 pg/ml (mean +/- S.D.). The NTLI concentration in CSF of 26 patients in an acute phase of aseptic meningitis was 110.6 +/- 51.1 pg/ml which was significantly (P less than 0.01) lower than the controls. These patients had a mean temperature of 101.4 +/- 1.5 degrees F which remained elevated for an average of 3.5 days. The NTLI concentrations in CSF of 23 patients in a recovery phase (after blood and CSF findings became normal with no fever) were 166.5 +/- 57.8 pg/ml, which did not differ significantly from the normal. There were no statistical correlations between the NTLI concentration in CSF and the protein concentration or total cell count in CSF. These results suggest that NTLI concentration changes during a febrile aseptic meningitis and that it may be associated with thermoregulation.     

Apolipoprotein A-II: beyond genetic associations with lipid disorders and insulin resistance

PURPOSE OF REVIEW: Apolipoprotein A-II, the second major HDL apolipoprotein, was often considered of minor importance relatively to apolipoprotein A-I and its role was controversial. This picture is now rapidly changing, due to novel polymorphisms and mutations, to the outcome of clinical trials, and to studies with transgenic mice. RECENT FINDINGS: The -265 T/C polymorphism supports a role for apolipoprotein A-II in postprandial very-low-density lipoprotein metabolism. Fibrates, which increase apolipoprotein A-II synthesis, significantly decrease the incidence of major coronary artery disease events, particularly in subjects with low HDL cholesterol, high plasma triglyceride, and high body weight. The comparison of transgenic mice overexpressing human or murine apolipoprotein A-II has highlighted major structural differences between the two proteins; they have opposite effects on HDL size, apolipoprotein A-I content, plasma concentration, and protection from oxidation. Human apolipoprotein A-II is more hydrophobic, displaces apolipoprotein A-I from HDL, accelerates apolipoprotein A-I catabolism, and its plasma concentration is decreased by fasting. Apolipoprotein A-II stimulates ATP binding cassette transporter 1-mediated cholesterol efflux. Human and murine apolipoprotein A-II differently affect glucose metabolism and insulin resistance. A novel beneficial role for apolipoprotein A-II in the pathogenesis of hepatitis C virus has been shown. SUMMARY: The hydrophobicity of human apolipoprotein A-II is a key regulatory factor of HDL metabolism. Due to the lower plasma apolipoprotein A-II concentration during fasting, measurements of apolipoprotein A-II in fed subjects are more relevant. More clinical studies are necessary to clarify the role of apolipoprotein A-II in well-characterized subsets of patients and in the insulin resistance syndrome.     

Comparative Proteomics Analysis of Cerebrospinal Fluid of Patients with Guillain–Barré Syndrome

To better understand the pathophysiologic mechanisms underlying Guillain-Barré syndrome (GBS), Comparative proteomic analysis of cerebrospinal fluid (CSF) between patients with GBS (the experiment group) and control subjects suffering from other neurological disorders (the control group) was carried out using two-dimensional gel electrophoresis (2-DE) technique, in combination with matrix-assisted laser desorption ionization time of flight mass spectrometry (MALDI-TOF MS) and database searching to determine abnormal CSF proteins in GBS patients. Image analysis of 2-DE gels silver stained revealed that 10 protein spots showed significant differential expression between the two groups of CSF samples. The expression of cystatin C, transthyretin, apolipoprotein E and heat shock protein 70 were decreased. However, haptoglobin, alpha-1-antitrypsin, apolipoprotein A-IV and neurofilaments were elevated. The subsequent ELISA measured the concentration of cystatin C and confirmed the result of the proteomic analysis. These identified proteins may be involved in the pathophysiological process of GBS and call for further studying the role of these proteins in the pathogenesis of the disease.     

Caspase-1 levels in biological fluids from patients with multiple sclerosis and from patients with other neurological and non-neurological diseases

In multiple sclerosis (MS), pathological white matter damage in the central nervous system is sustained by immune-inflammatory response. Caspase-1 plays a pivotal role in immune-mediated inflammation, as it regulates the cellular export of IL-1beta and IL-18. We carried out a preliminary in vitro study of the kinetics of extracellular caspase-1 release. We then measured caspase-1 levels in paired serum and cerebrospinal fluid (CSF) samples of 75 MS patients, 15 healthy subjects, and patients with other neurological diseases. Paired synovial fluid and serum samples of patients with juvenile idiopathic arthritis, and paired sputum and serum samples of asthma patients were also studied. Mean serum caspase-1 concentrations did not differ between groups. Caspase-1 was detected in the CSF of patients with acute, but not stable, MS [7.5 +/- (SEM) 0.9 pg/ml; test's sensitivity, 56% and specificity, 100%]. Its levels correlated with pleocytosis. The highest mean caspase-1 levels were found in the arthritic synovial fluids (945.5 +/- 126.6 pg/ml, which correlated with erythrocyte sedimentation rate), and in the sputum samples (370.1 +/- 71.0 pg/ml, which correlated with the number of macrophages in the sputum). On condition that caspase-1 is determined in the fluids pertaining to the disease-specific inflammatory sites, its level is a reliable marker of ongoing immune-inflammatory response. The enzyme measurement in CSF can also help define state-trait in MS.     

Cells with natural killer activity in the cerebrospinal fluid of normal mice and athymic nude mice with acute Sindbis virus encephalitis

Sindbis virus causes an acute, nonfatal inflammatory encephalitis in weanling BALB/c mice. Mononuclear inflammatory cells are present in the cerebrospinal fluid (CSF) as well as in the parenchyma of the brain. Both aspects of this inflammatory response were eliminated by treatment with cyclophosphamide. Athymic nude (nu/nu) mice developed no inflammation in the brain, but did develop a CSF pleocytosis that peaked on day 2 after infection. The time course of the appearance of cells in the CSF was earlier in nu/nu mice than their heterozygote (nu/+) littermates. The pleocytosis in nu/nu mice reached a peak on day 2, whereas in nu/+ mice the peak was on day 4, as it is in normal BALB/c mice. To determine whether some of the CSF cells in nu/nu mice may be natural killer (NK) cells, NK activity was measured in a 4-hr assay by using a YAC-1 target cell. NK cell activity in the spleen and peripheral blood was induced by infection with Sindbis virus in nu/nu mice with a similar time course to that of nu/+ mice (peak 1 day after infection). CSF from nu/nu mice had NK activity present 2 days after infection that was greater than that present in either the peripheral blood or spleen. BALB/c and nu/+ mice had insufficient cells present for assay at day 2, but BALB/c mice had NK activity present in the CSF 3 and 5 days after infection that exceeded that in the peripheral blood or spleen. Brain interferon was detectable on day 1 in nu/nu mice, but not until day 2 in nu/+ mice even though the amounts of brain virus were the same in the two groups at all time points. It is concluded that cells with NK activity contribute to the CSF pleocytosis induced by acute Sindbis virus encephalitis.     

Dysregulation of retinoid transporters expression in body fluids of schizophrenia patients

This study aims to find the biomarkers or associated proteins in body fluids of schizophrenia patients so that we can further understand the etiology of schizophrenia. We applied proteomic technologies combining two-dimensional electrophoresis with Coomassie blue staining and mass spectrometry and identified a procedure for the clinical screening of disease-influenced body fluid proteins in two sets of samples, plasma from 19 schizophrenia patients and cerebrospinal fluid (CSF) from 35 drug-treated schizophrenic patients and 36 healthy controls. The expression of transthyretin (TTR) tetramer increased significantly in plasma of schizophrenic patients after a valid 2 months in-hospital antipsychotic treatment. Conversely, the expression of the TTR tetramer and apolipoprotein E (ApoE) was down-regulated by up to 1.68 and 3.62 times, respectively, in the CSF of schizophrenia patients compared to that of normal controls, which has not been reported previously. Considering that the TTR tetramer and ApoE are both retinoid transporters, retinoid dysfunction might be involved in the pathology of schizophrenia.     

Cellular Composition of Cerebrospinal Fluid in HIV-1 Infected and Uninfected Subjects

In order to characterize the cellular composition of cerebrospinal fluid (CSF) in a healthy state and in the setting of chronic pleocytosis associated with HIV-1 (HIV) infection, multi-parameter flow cytometry was used to identify and quantitate cellular phenotypes in CSF derived from HIV-uninfected healthy controls and HIV-infected subjects across a spectrum of disease and treatment. CD4+ T cells were the most frequent CSF population and the CD4:CD8 ratio was significantly increased in the CSF compared to blood (p = 0.0232), suggesting preferential trafficking of CD4+ over CD8+ T cells to this compartment. In contrast, in HIV-infection, CD8+ T cells were the major cellular component of the CSF and were markedly increased compared to HIV-uninfected subjects (p<0.001). As with peripheral blood, the CSF CD4:CD8 ratio was reversed in HIV-infected subjects compared to HIV-uninfected subjects. Monocytes, B cells and NK cells were rare in the CSF in both groups, although absolute counts of CSF NK cells and B cells were significantly increased in HIV-infected subjects (p<0.05). Our studies show that T cells are the major cellular component of the CSF in HIV-infected and uninfected subjects. The CSF pleocytosis characteristic of HIV infection involves all lymphocyte subsets we measured, except for CD4+ T cells, but is comprised primarily of CD8+ T cells. The reduced proportion of CD4+ T cells in the CSF may reflect both HIV-related peripheral loss and changes in trafficking patterns in response to HIV infection in the central nervous system.     

The latex agglutination test versus counterimmunoelectrophoresis for rapid diagnosis of bacterial meningitis

A modified latex agglutination (LA) test was compared with Gram-staining and counterimmunoelectrophoresis (CIE) for the rapid detection in the cerebrospinal fluid (CSF) of antigen to Haemophilus influenzae type b, Neisseria meningitidis groups A, B and C, Escherichia coli K1, Streptococcus pneumoniae and group B streptococci, seven frequent causes of bacterial meningitis in children. Of 50 CSF samples from patients with culture-proven bacterial meningitis 90% were correctly shown by the LA test to contain antigen of the responsible organism. Gram-staining revealed organisms in 80% of 45 of these samples. In 75% of the 40 samples that were of sufficient volume for CIE, positive results for the appropriate antigen were obtained. The concentration of antigen detected in the CSF by the LA test varied from undetectable to 800 000 ng/ml. Patients with a high concentration (more than 2000 ng/ml or a positive result at dilutions of CSF over 1/8) were significantly more likely to have a poor response to therapy (two died and two had persistent pleocytosis or bacteria in the CSF) than patients with a lower concentration (4/16 v. 0/18, P < 0.05). After appropriate therapy was begun the concentration of antigen fell dramatically, but measurable amounts of antigen persisted in the CSF for up to 6 days. The LA test detected bacterial antigen at concentrations 2 to 70 times below the lower limit detected by CIE. In seven additional patients who had received antibiotics before lumbar puncture was performed the LA test detected antigen from meningitis-causing bacteria even though cultures of the CSF were sterile. In another 145 patients who did not have meningitis the results of the LA test were negative. The LA test, done as described in this article, is easier to perform than CIE and should be a useful addition to the diagnostic tests carried out on the CSF of any patient suspected of having meningitis.     

Characterization and functional studies of lipoproteins, lipid transfer proteins, and lecithin:cholesterol acyltransferase in CSF of normal individuals and patients with Alzheimer's disease

We investigated the lipoprotein distribution and composition in cerebrospinal fluid (CSF) in a group of patients with Alzheimer's disease (AD) or affected by other types of dementia in comparison to non-demented controls. We found slightly decreased apolipoprotein (apo)E and cholesterol concentrations in CSF of AD patients and moderately increased apoA-I concentrations, while in patients suffering from other types of dementia the apoA-I CSF concentration was increased. ApoA-IV concentrations varied widely in human CSF, but were not associated with any clinical condition. HDL(2)-like apoE-containing lipoproteins represent the major lipoprotein fraction. In CSF of normal controls, only a minor HDL(3)-like apoA-I-containing lipoprotein fraction was observed; this fraction was more prevalent in AD patients. ApoA-II was recovered mostly in the HDL(3) density range, while apoA-IV was not associated with lipoproteins but appeared in a lipid-free form, co-localizing with LCAT immunoreactivity. Bi-dimensional analysis demonstrated pre-beta and alpha apoA-I-containing particles; apoE and apoA-II were detected only in alpha-migrating particles. ApoA-IV distributed both to pre-beta and gamma-migrating particles; the LCAT signal was co-localized in this gamma-migrating fraction. Enzymatically active LCAT was present in human CSF as well as PLTP activity and mass; no CETP mass was detected. In CSF from AD patients, LCAT activity was 50% lower than in CSF from normal controls. CSF lipoproteins induced a significant cholesterol efflux from cultured rat astrocytes, suggesting that they play an active role in maintaining the cholesterol homeostasis in brain cells.     

Leukocyte Attraction by CCL20 and Its Receptor CCR6 in Humans and Mice with Pneumococcal Meningitis

We previously identified CCL20 as an early chemokine in the cerebrospinal fluid (CSF) of patients with pneumococcal meningitis but its functional relevance was unknown. Here we studied the role of CCL20 and its receptor CCR6 in pneumococcal meningitis. In a prospective nationwide study, CCL20 levels were significantly elevated in the CSF of patients with pneumococcal meningitis and correlated with CSF leukocyte counts. CCR6-deficient mice with pneumococcal meningitis and WT mice with pneumococcal meningitis treated with anti-CCL20 antibodies both had reduced CSF white blood cell counts. The reduction in CSF pleocytosis was also accompanied by an increase in brain bacterial titers. Additional in vitro experiments showed direct chemoattractant activity of CCL20 for granulocytes. In summary, our results identify the CCL20-CCR6 axis as an essential component of the innate immune defense against pneumococcal meningitis, controlling granulocyte recruitment.     

Cytologic features of pilocytic astrocytoma in cerebrospinal fluid specimens

OBJECTIVE: To illustrate the cytomorphologic features of pilocytic astrocytoma (PA) in cerebrospinal fluid (CSF) samples. STUDY DESIGN: A search of records from 1965 to 2001 was performed to identify all patients with a diagnosis of PA in whom CSF samples were examined. Slides from CSF samples originally reported as atypical, suspicious or positive were reviewed and the cytomorphologic features assessed. RESULTS: Two hundred ninety-three patients with a diagnosis of PA were identified. Of these, 44 had a total of 65 cytologic preparations of CSF. In 34 patients (77.2%) the CSF cytology was negative, in 5 (11.4%) either atypical or suspicious, and in 5 (11.4%) positive for neoplastic cells. The tumors in the 5 positive cases arose in the cerebellar hemispheres (2), cerebellar vermis (1), thalamus (1) and tectum with extension into the fourth ventricle (1). All positive samples were hypercellular, with an average of 5 cell clusters per case (range, 3-11). The clusters were composed of cohesive epithelioid cells with a mean of 8 cells per cluster. In addition, some cases had scattered, isolated, single cells. These single neoplastic cells had prominent, hairlike cytoplasmic processes. The cells in clusters appeared epithelioid, with oval nuclei, mild nuclear pleomorphism, finely or slightly coarsely granular chromatin and cobweblike cytoplasm. CONCLUSION: The cytomorphologic features of PAs recapitulate their histologic characteristics. The tumor cells are recognizable in CSF samples and readily distinguishable from histiocytes and ependymal cells.