Rapid Detection Of Neoplastic Cells In Serous Cavity Effusions In Children With Flow Cytometry Immunophenotyping

The aim of the study was to evaluate the role of flow cytometric immunophenotyping (FCI) as a quick diagnostic tool for pediatric malignancy in serous cavity effusions (SCE). FCI results on103SCE in a pediatric population were compared with retrospective clinical outcome (RCO). As to the CM examination,75.7%(73cases) was negative,5.8%(6cases) was suspicious, and20.6%(19samples) was positive. The sensitivity and specificity of CM in effusions examination were61.0%and96.9%, respectively. Of41patients assessed as having malignancies by RCO,36patients were diagnosed as lymphoma (25), AML (2), neuroblastoma (8) and retinoblastoma (1) by FCI, respectively. The sensitivity and specificity of FCI for detecting neoplastic cells were87.8%and98.4%, respectively. The concordance of FCI data with final diagnoses of the patients was94.2%. FCI data for lymphoma was concordant with the final diagnosis in89.3%of cases. When Hodgkin lymphoma was excluded, the overall correlation increased to96.1%.13cases of T cell lymphoma (including1ALCL and1LGL) were diagnosed by FCI in pleural effusions and only1case in ascites. Six cases of B cell lymphoma were diagnosed in ascites and4cases in pleural effusions. The distribution of T and B cell lymphoma in ascites or pleural effusions was significantly different (Fisher’s Exact Test, P=0.009). FCI with CD81+/CD56bright+/CD45-combination showed a100%of concordance with the final diagnosis in neuroblastoma patients. FCI is a useful tool for rapid and reliable diagnosis of pediatric non-Hodgkin lymphoma in SCE samples and also to suggest the presence of non-lymphoid malignancies especially neuroblastoma.Materials and MethodsOne hundred and three consecutive SCE samples from103children were collected including68pleural effusions,32peritoneal effusions and3cerebrospinal fluid specimens (CFS). The assessment criteria of the RCO were defined on comprehensive parameters including both laboratory and clinical evidences by at least two experienced hemato-oncologist(s) regardless of the results obtained by FCI and CM. The detailed assessment criteria of RCO for having a malignancy (positive RCO) or not (negative RCO) are described as follows:RCO positive for malignancy was defined as positive histology plus subsequent response to chemotherapy or other treatments specific for malignancy, as documented by clinical, radiological and imaging data. RCO negative for malignancy was defined as no any clinical (including clinical presentation, diagnosis and treatment response during the course of the disease) and laboratory (including blood test, tumor marker, tissue biopsy, bone marrow morphology and immunology, cytogenetic and molecular analyses etc.) evidences to support the diagnosis of a malignancy in patient.Cytological Preparation and Classification of Effusions:For routine cytological preparations, air-dried and alcohol-fixed cytocentrifuge or smear preparations were made depending on the cellularity. These were routinely stained with Wright-Giemsa stain, and classified into negative, atypical/suspicious, and positive malignancy. The cytomorphologic diagnosis was rendered independent of the FCI results.Flow Cytometric Immunophenotyping:The effusions were analyzed for various antigens on a flow cytometer (FACSCaliburTM; Becton Dickinson, San Jose, CA). Data acquisition and analysis were performed using CellQuest software (Becton Dickinson). Immunophenotyping was performed by four-color immunofluorescent staining using the following commercially available fluorescence-labeled monoclonal antibodies (MoAbs):Mouse IgG1, Mouse IgG2a, CDla, CD2, CD3, CD4, CD5, CD7, CD8, CD10, CD11c, CD13, CD19, CD20, CD22, CD23, CD25, CD30, CD33, CD34, CD38, CD45, CD56, CD69, CD81, CD103, CD117, CD138, HLA-DR, CD45RA, CD45RO, FMC7, TCRα/β, TCRγ/δ, Cμ, SmlgM, Kappa(κ), and Lambda(λ)(Becton-Dickinson, San Jose, CA).Statistical Analysis:Data were described as median when continuous, as absolute and relative frequency when categorical. True positive and true negative results were assessed as those matching with positive and negative RCO, respectively. The agreement with RCO, sensitivity, specificity, positive predictive value and negative predictive value were calculated for each method. The distribution comparisons were performed by means of a test (Fisher’s Exact Test). All statistical analysis was performed using the Statistical Package for the Social Sciences (spss) software, version12.0(SPSS Inc., Chicago, IL, USA). In our study, P<0.05was considered to be statistically significant.ResultsCytomorphology and Flow Cytometric ImmunophenotypingWith a median follow-up time of15.2months,41cases (39.8%) were finally diagnosed as malignancies, and62(60.2%) as non-malignanices based on RCO. The malignancy group consisted of26cases of non-Hodgkin lymphoma,2Hodgkin lymphoma (HL),2acute myeloid leukemia (AML),8neuroblastoma,1retinoblastoma and2soft tissue sarcoma. The non-malignancy group consisted of30cases of pneumonia,9septicaemia,7tuberculous pleuritis,3tuberculous peritonitis,3infectious mononucleosis,2lung abscess,2hematophagocytic syndrome,1Kawasaki disease,1lupus erythematosus,1Crohn’s disease,1aplastic anemia,1thrombocytopenia and1congenital choledochal cyst.As to the CM examination,75.7%(78cases) was negative,5.8%(6cases) was suspicious, and20.6%(19samples) was positive. The sensitivity and specificity of CM in effusions examination were61.0%and96.9%, respectively. Of41patients with positive RCO,36cases (87.8%) were diagnosed as malignancies by FCI in effusions, including lymphoma (25pts), AML (2pts), neuroblastoma (8pts) and retinoblastoma (1pt). Neoplastic cells in total cells ranged from0.7%-97.2%(median66.4%) on FCI. Of5patients with negative FCI results,2had Hodgkin lymphoma,1had anaplastic large cell lymphoma (ALCL) and2had soft tissue sarcomas. The sensitivity and specificity of FCI for detecting neoplastic cells in all SCE were87.8%and98.4%, respectively. The concordance of FCI data with final diagnoses was94.2%for all groups. FCI results were concordant with the final diagnoses in89.3%(25/28) of cases with lymphoma. When HL was excluded, the overall concordance rate was increased to96.1%. FCI result from a case of B-NHL is shown as an example (Figure1). In a normal T cell rich serous fluid, FCI detected a minor proportion (9.3%) of aberrant T cell population which showed high light forward scatter and expressing CD4, CD7, CD2, CD45RO, CD30and negative for CD5and CD69. This case was diagnosed as ALCL by FCI referring to the literature. Another ALCL case did not show or diminution of pan-T antigens and FCI could not confirm the diagnosis. With reference to the2008WHO classification of tumors of hematopoietic and lymphoid tissues, the effusions from25lymphoma patients analyzed by FCI were categorized into B cell lymphoma (12pts), T cell lymphoma (11pts), anaplastic large cell lymphoma (1pt), and large granular lymphocyte leukemia (LGL)(1pt).In ascites and pleural effusions, FCI with CD81+/CD56bright+/CD45- combination showed a100%of concordance with the final diagnosis in neuroblastoma patients. Figure3illustrates the FCI result from a case of neuroblastoma. A CSF sample presenting a phenotype of CD81+/CD56+/CD45-was finally diagnosed as retinoblastoma based on the histological examination of the mass in the eye.13cases of T cell lymphoma (including1ALCL and1LGL) were diagnosed by FCI in pleural effusions and only1case in ascites. Six cases of B cell lymphoma were diagnosed in ascites and4cases in pleural effusions. The distribution of T and B cell lymphoma in ascites or pleural effusions was significantly different (Fisher’s Exact Test, P=0.009).Sixty-two patients were RCO negative for malignancy. In all these patients, the effusions disappeared after conventional therapy. Only one ascites was initially suspected to be caused by a putative T cell lymphoma based on FCI data, however, the patient was eventually diagnosed as Epstein-Barr virus (EBV) infection (infectious mononucleosis) based on the evidences of EBV serology, EBV-DNA analysis and clinical follow up.ConclusionsFCI can be used as an adjunct to cytomorphology for rapid and reliable diagnosis and classification of pediatric NHL and also to suggest the presence of non-lymphoid tumors particularly neuroblastoma.