Subclassification of lymphoproliferative disorders in serous effusions

A 10-year experience


  • Leung Chu Tong MD, FRCPC,

    1. Department of Laboratory Medicine and Pathobiology, University of Toronto, Ontario, Canada
    2. Laboratory Medicine Program, University Health Network, Toronto, Ontario, Canada
    Current affiliation:
    1. Department of Pathology, Memorial Sloan-Kettering Cancer Center, New York, New York
    Search for more papers by this author
  • Hyang-Mi Ko MD, PhD,

    1. Department of Laboratory Medicine and Pathobiology, University of Toronto, Ontario, Canada
    2. Laboratory Medicine Program, University Health Network, Toronto, Ontario, Canada
    Search for more papers by this author
  • Mauro Ajaj Saieg MD, PhD,

    1. Department of Laboratory Medicine and Pathobiology, University of Toronto, Ontario, Canada
    2. Department of Pathology, Santa Casa Medical School, Sao Paulo, Brazil
    Search for more papers by this author
  • Scott Boerner MD, FRCPC,

    1. Department of Laboratory Medicine and Pathobiology, University of Toronto, Ontario, Canada
    2. Laboratory Medicine Program, University Health Network, Toronto, Ontario, Canada
    Search for more papers by this author
  • William R. Geddie MD, FRCPC,

    1. Department of Laboratory Medicine and Pathobiology, University of Toronto, Ontario, Canada
    2. Laboratory Medicine Program, University Health Network, Toronto, Ontario, Canada
    Search for more papers by this author
  • Gilda da Cunha Santos MD, PhD, FRCPC, FIAC

    Corresponding author
    1. Department of Laboratory Medicine and Pathobiology, University of Toronto, Ontario, Canada
    2. Laboratory Medicine Program, University Health Network, Toronto, Ontario, Canada
    • Corresponding author: Gilda da Cunha Santos, MD, PhD, FRCPC, FIAC, Department of Laboratory Medicine and Pathobiology, University of Toronto, University Health Network, 200 Elizabeth Street, 11th Floor, Eaton Wing, Toronto, Ontario M5G 2C4; Fax: (416) 340-5517;

    Search for more papers by this author



Rare studies have reported the application of multiple ancillary tests to the diagnosis of lymphoproliferative disorder in serous effusions. In the current study, the authors evaluated the effectiveness of using an algorithm for the triage of serous effusions and the contribution of ancillary studies to achieve a specific subtype of lymphoproliferative disorder.


Serous effusion samples that had a final diagnosis of lymphoproliferative disorder or suspicious for lymphoma were selected from cases that were diagnosed between 2001 and 2010. Data were collected on patient and sample characteristics as well as results from immunophenotype and molecular studies.


In total, 168 serous effusions were identified from 110 patients. The most common site of involvement was the pleural cavity (n = 133) followed by the peritoneal cavity (n = 30) and pericardial cavity (n = 5). The volume of serous effusions ranged from 2 mL to 1000 mL (mean, 238 mL). In 42 patients (38.2%), serous effusions were the primary source of diagnosis. In 129 patients who had a diagnosis of LPD, either generic (n = 82) or specific (n = 47) ancillary tests were performed as a single test in 58 samples (67.4%) or as a combination of multiple studies in 19 samples (23.2%). Immunophenotyping was successful in almost all samples that had a specific subtype with 16 B-cell and 4 T-cell lymphomas being diagnosed. More samples with a specific subtype of lymphoma underwent molecular tests compared with those who had a generic diagnosis (19.1% vs 13.4%).


Successful, specific subtyping of lymphoproliferative disorders was achieved in approximately 33% of cases that were tested for ancillary studies following an approach for the triage and aliquoting of serous effusion samples. Cancer (Cancer Cytopathol) 2013;121:261–70. © 2012 American Cancer Society.


The involvement of serous effusions (SEs) by non-Hodgkin lymphoma (NHL) is not rare. The reported frequency of lymphoma in malignant effusions has ranged between 10% and 15% and constitutes one of the most common causes of malignancy in SEs after carcinoma of lung, breast, and ovary.[1]

Lymphoid cells in SEs pose a diagnostic challenge in cytopathology, because a reactive lymphocytosis consisting predominantly of small cells may mimic indolent lymphoma. In contrast, large cell lymphomas in effusions, although they are easier to diagnose morphologically, may present problems with respect to ancillary studies because of hypocellular or degenerated samples.

Various strategies for subclassifying lymphoproliferative disorders in SE specimens have been described. Some authors have reported the subclassification of NHL in SEs with the aid of immunophenotyping (IP),[4] immunohistochemistry (IHC),[5] and molecular assays.[8] However, there are rare reports about the routine application of multiple ancillary tests to the diagnosis of lymphoma in SEs.[6, 9] Techniques and lymphoma classification systems have changed significantly in the last decade with the introduction of molecular studies and the description of new entities. The current World Health Organization (WHO) classification[10] highlights the use of multiple ancillary techniques for subclassifying NHL, and this multiparametric approach has been recommended as the standard of practice because of its importance for patient treatment.

To apply a multiparametric approach to daily practice, a specimen triage algorithm may be necessary to obtain the maximum amount of diagnostic, prognostic, and predictive information. An efficient method should provide clinically useful data and should avoid the over investigation of fluids with reactive lymphocytosis. An early documented protocol for specimen triage in SEs included details like cell count, cytology processing, IHC, flow cytometry, and electron microscopy.[9] More recently, another algorithm for both fine-needle aspirates (FNAs) of lymph nodes and SEs has been described.[11] However, to date, there has been no publication on the contribution of multiple ancillary techniques for lymphoma subclassification in SEs based on the latest WHO classification as part of an established algorithm for sample triage.

The objective of the current study was to evaluate the effectiveness of using a triage and testing algorithm for the diagnosis of lymphoproliferative disorder in SE and to assess the contribution of ancillary studies to achieve a specific diagnosis using the current WHO classification.


Study Cases

SE cytology reports from October 2001 to November 2010 with a diagnosis of lymphoproliferative disorder or suspicious for lymphoma were collected retrospectively from the electronic pathology database of our institution. The following clinicopathologic characteristics were extracted from the reports: age at the time of diagnosis, sex, site, gross volume, and final diagnosis. Previous history of lymphoma was extracted from the electronic medical record. Results from ancillary tests also were retrieved from the final reports, including cell-surface IP, IHC, polymerase chain reaction (PCR) for B-cell or T-cell clonality, human herpesvirus 8 (HHV-8) status, fluorescence in situ hybridization (FISH), and in situ hybridization for the detection of Epstein-Barr virus-encoded RNA (EBER) assays.

Algorithm for Ancillary Studies

For SEs that were submitted with either a stated history or a clinical suspicion of lymphoma, an immediate morphologic assessment was performed to triage lymphoid-rich effusions for appropriate ancillary studies, including molecular assays, similar to the algorithm used to assess lymph node FNAs[12] (Fig. 1). Romanowsky-stained cytospin preparations were produced and submitted to the cytopathologist for rapid assessment. In addition, samples that were submitted without any history of lymphoma or request for IP but were deemed to have intense lymphocytosis with some morphologic features suggestive of lymphoma after examination of the ThinPrep slide also were triaged if any fresh material remained. The implementation of this algorithm permitted the application of at least one ancillary technique to the samples. Therefore, the final cytology diagnosis incorporated information from cytomorphology, IP, IHC, and molecular assays.

Figure 1.

This flowchart illustrates the algorithm used for triaging and aliquoting lymphoid-rich serous effusions for ancillary tests. LPD indicates lymphoproliferative disorders; TP, ThinPrep; FFPE, formalin-fixed, paraffin-embedded; CP, cytospin preparation; CB, cell blocks.

After rapid morphologic assessment, the sample was judged in terms of both cellularity and the presence of necrosis/degeneration to determine whether it was fit to undergo IP. If the preliminary assessment based on cytomorphology suggested a lymphoproliferative disorder, then a cytotechnologist would perform a cell count. The number of lymphoid cells present in the specimen would dictate which ancillary test was to be performed. If there were sufficient numbers of viable lymphoid cells, then cell-surface IP was performed using either flow cytometry or laser-scanning cytometry. IP results were interpreted by one of the authors (G.d.C.S., S.B. and W.R.G.).

The differential diagnosis that was formulated during rapid assessment also initiated the aliquoting of available material for concurrent IHC and/or molecular assays, depending on the amount of sample remaining after cell-surface marker IP. The decision to submit the specimen for further testing also was based on a judicious evaluation of the volume, previous clinical history of lymphoma, the subtype, and any specific clinical request or question. The most common clinical questions were to confirm recurrence and to detect whether transformation to an aggressive subtype occurred in a patient who had a previously diagnosed indolent lymphoma. Then, based on the results from either flow cytometry or laser-scanning cytometry and the differential diagnosis formulated during the rapid assessment, additional ancillary tests were ordered.

For IHC, a cytospin preparation or a formalin-fixed, paraffin-embedded (FFPE) cell block was used. FFPE cell blocks were produced according to the HistoGel technique (Thermo Fisher Scientific Inc [Richard Allan Scientific], Kalamazoo, Mich). For PCR-based molecular tests, either an aliquot of fresh fluid or an unstained 20-micron cell block section was submitted to the molecular genetics laboratory; and, for FISH assays, 2 or 3 additional cytospin unstained slides were prepared and submitted to the cytogenetics laboratory.


Flow cytometry

Flow cytometry was performed on SE samples using a previously described whole blood lysis technique.[13] Selected antibodies conjugated to fluorescein isothiocyanate (FITC), phycoerythrin (PE), phycoerythrin-Texas Red-x, PE-cyanine 5.1 (PECy5), and PE-cyanine 7 (PECy7) were used at concentrations that were titrated for optimal staining. From October 2001 to November 2010, there was a transition from 3-color flow cytometry (FITC, PE, and PECy5) to 4-color flow cytometry (additional PE-Texas Red),and 5-color flow cytometry (additional PECy7). IP was performed using multiparameter flow cytometry (either Epics XL or Cytomics FC-500; Beckman Coulter Canada Inc., Mississauga, Ontario, Canada). The events were gated using forward scatter versus side scatter. The following panels were used for flow-cytometric IP analysis using 3-color fluorochromes: polyclonal kappa FITC/polyclonal lambda PE/CD19 PECy5, CD5 FITC/CD10 PE/CD19 PECy5, FMC7 FITC/CD23 PE/CD20 PECy5, CD4 FITC/CD8 PE/CD3 PECy5, CD7 FITC/CD11c PE/CD2 PECy5, and an automated, fluorescent negative control sample. The panels were slightly modified with 4-color fluorochromes and included an additional antibody (CD45). The panels for 5-color flow cytometric IP included 3 additional antibodies (CD7, CD38, and CD79a).

Laser-scanning cytometry

From November 2003 to November 2010, laser-scanning cytometry using the Clatch protocol was the primary IP method.[14, 15] This technique has demonstrated the ability to provide results equivalent to those from flow cytometry.[13] Imaging cytometers (either laser-scanning cytometry or iCys; CompuCyte, Cambridge, Mass) and WinCyte software or iCys 3.4 (CompuCyte) with events contoured on the basis of forward light scatter/laser light absorbance were used for analysis. A full panel consisted of the following monoclonal antibodies: CD19, CD20, CD3, CD5, CD10, kappa, lambda, CD22, CD23, FMC7, CD4, CD8, CD56, CD7, CD2, CD25, CD14, CD11c, and CD16. The antibody clones have been described previously along with concentrations, fluorochromes, and suppliers.[13]


IHC was performed on either FFPE cell blocks or cytospin preparations. IHC on cytospin preparations followed the same protocol that was used for FFPE cell blocks.

For the workup of differential diagnoses, the following monoclonal antibodies were used: AE1/AE3, Cam 5.2, 34 beta E12, and CD25 as well as polyclonal antibodies myeloperoxidase (MPO) and lysozyme. For subclassification of lymphomas, the following monoclonal antibodies were selected: CD10, CD15, CD20, CD22, CD30, CD43, CD45, CD79a, CD138, B-cell chronic lymphocytic leukemia/lymphoma 2 (Bcl-2), Bcl-6, HHV-8, MIB-1, cyclin D1 (CCND1), CD2, CD3, CD4, CD5, CD7, CD8, CD1a, T-cell intracellular antigen 1 (TIA-1), anaplastic lymphoma kinase (ALK), and CD56. Polyclonal antibodies kappa and lambda light chains and terminal deoxynucleotidyl transferase (TdT) also were included. Either the iVIEW DAB avidin-biotin detection kit or the Ventana ultraview DAB chromogen kit (Ventana Medical Systems, Tucson, AZ) was used as the detection system.

Polymerase Chain Reaction Assays

DNA was extracted from samples and assessed for the presence of T-cell receptor gamma (TCRγ) gene rearrangements by PCR using 2 pairs of Vγ and Jγ primers.[16] PCR analysis of B-cell clonality was performed with primers to the framework region III and joining region of the immunoglobulin heavy-chain (IgH) gene essentially by following the previous reported protocol.[17] The presence of HHV-8 was assessed by PCR using primers for specific region of the virus as previously described.[18] Appropriate positive, negative, and internal control samples were run with each specimen. The results of PCR for B-cell or T-cell monoclonality and HHV-8 were recorded.

Fluorescence In Situ Hybridization Assays

FISH for the detection of translocations was performed on cytospin preparations that were produced from the samples. Assays were completed using one or more of the following probes: IGH/BCL2 to detect the t(14;18) (q32;q21.3), IGH/CCND1 to detect t(11;14)(q13;q32) (dual-color, dual-fusion probes), and MYC (break-apart probe) to detect rearrangements. Details have been published previously describing the probes, the hybridization protocol, and the scoring criteria used for the assays.[19]

Epstein-Barr Virus In Situ Hybridization

In situ hybridization for the detection of EBER was completed on either cytospin preparations or cell block sections with an XT INFORM probe supplied by Ventana Medical Systems on an automated stainer (Ventana BenchMark XT). Visualization was achieved with the ISH iVIEW system with alkaline phosphatase and nitro blue tetrazolium/5-bromo-4-chloro-3′-indolyphosphate substrates using nuclear Fast Red as contrast.[20] An external positive and negative control accompanied each run.


Clinicopathologic Characteristics

From October 2001 to November 2010, there were 168 SEs from 110 patients with a final cytologic diagnosis of suspicious for lymphoma or lymphoproliferative disorder. There were 72 men and 38 women, and the mean patient age was 58.7 years (range 19-92 years). The most common site of involvement was the pleural cavity (n = 133), followed by the peritoneal cavity (n = 30) and the pericardial cavity (n = 5) (Table 1). The mean volume of SE samples was 238 mL (range 2-1000 mL). Forty-two patients (38.2%) did not have a previous history of lymphoproliferative disorder, and the SE sample represented the primary source of diagnosis. In 46 patients, the fluid was collected to rule out recurrence. For the 22 remaining patients, no information about lymphoproliferative disorders was available (Table 1).

Table 1. Patient and Sample Characteristics
 No. of Patients (%)
Patient characteristics 
Men72 (65.5)
Women38 (34.5)
Total110 (100)
Age, y 
Median [range]59 [19-92]
Previous history of lymphoma 
Yes46 (41.8)
No42 (38.2)
Unknown22 (20)
Sample characteristics 
Volume: Mean [range], mL238 [2-1000]
Type of effusion 
Pleural133 (79.2)
Peritoneal30 (17.9)
Pericardial5 (2.9)

Lymphoproliferative Disorder Subclassification

One hundred twenty-nine SE samples (77%) were diagnosed as lymphoproliferative disorder; whereas, in 39 SE samples (23%), the final diagnosis was suspicious for lymphoma (Fig. 2). The distribution of patients according to the final cytologic diagnosis is presented in Table 2. These diagnoses were divided into 2 categories: a generic diagnosis and a specific subtype. Generic diagnoses (82 patients) included 26 large B-cell lymphoma; 22 large cell lymphoma; 18 NHL, not otherwise specified (NOS); 14 small B-cell NHL; and 2 T-cell lymphoma, large cell type. The specific subtype diagnoses (47 patients) included 13 chronic lymphocytic leukemia/small lymphocytic lymphoma; 11 follicular lymphoma; 5 mantle cell lymphoma; 4 Burkitt lymphoma; 4 primary effusion lymphoma, 4 T-cell acute lymphoblastic leukemia/lymphoma (T-ALL), 3 post-transplant lymphoproliferative disorder; 2 peripheral T-cell lymphoma, NOS; and 1 myeloma.

Figure 2.

This flowchart illustrates generic and specific diagnoses of lymphoproliferative disorders and the multiple ancillary tests performed. IP indicates immunophenotype; IHC, immunohistochemistry; PCR, polymerase chain reaction; FISH, fluorescence in situ hybridization; EBER, Epstein-Barr virus-encoded RNA.

Table 2. Type of Ancillary Tests Performed in 129 Samples According to the Diagnosis
DiagnosisNo. of CasesIP with resultsIP no resultsIHCFISHPCREBER
  1. Abbreviations: BL, Burkitt lymphoma; CLL/SLL, chronic lymphocytic leukemia/small lymphocytic lymphoma; EBER, Epstein-Barr virus-encoded RNA; FISH, fluorescence in situ hybridization; FL, follicular lymphoma; IHC, immunohistochemistry; IP, immunophenotype; LBCL, large B-cell lymphoma; LCL, large cell lymphoma; ML, mantle cell lymphoma; NHL, NOS, non-Hodgkin lymphoma, not otherwise specified; PCR, polymerase chain reaction; PEL, primary effusion lymphoma; PTCL, NOS, peripheral T-cell lymphoma, not otherwise specified; PTLD, post-transplant lymphoproliferative disorder; Small B NHL, small B-cell non-Hodgkin lymphoma; T-ALL, T-lymphoblastic leukemia/lymphoma; TCL, T-cell lymphoma.
Generic diagnosis82632115560
NHL, NOS181012200
Small B NHL141301110
Specific subtype472319333
PTCL, NOS2100000

Ancillary Studies

One hundred twenty-nine samples with a final diagnosis of lymphoproliferative disorder were available for our analysis of the contribution of ancillary studies. The 39 samples that were diagnosed as suspicious for lymphoma were excluded from the analysis because in 25 samples (68%), no ancillary tests were performed, and only 14 samples were sent for ancillary studies either as a single test in 11 samples (10 IP and 1 PCR) or as a combination of 2 tests in 3 samples (2 IP plus PCR, 1 IP plus FISH). No samples were sent for IHC. Thirteen of 39 samples underwent IP by either laser scanning cytometry or flow cytometry, and the results were non-contributory in 7 samples because of a T-cell predominant population with reactive phenotype (5 samples) or because samples were rendered non-interpretable by non-specific binding (2 samples). Four samples revealed light-chain restriction; however, further subclassification could not be achieved. Two samples had a B-cell population; however, light-chain restriction was not demonstrated. The PCR results either were negative for B-cell monoclonality (2 samples) or were not conclusive because of poor DNA quality (1 sample). One sample that was sent for FISH analysis revealed an IGH/BCL2 translocation.

In the group of 88 patients in which a diagnosis of lymphoproliferative disorder was rendered, IP, IHC, PCR, FISH, and EBER analyses were performed as a single test or as a combination of multiple studies. In 58 samples, only one type of assay was used. Concurrent ancillary studies using either IP and IHC, or IP and molecular assays, or IHC and molecular assays were conducted in 10 samples. All 3 tests, including IP, IHC, and molecular assays, were performed in 9 samples (Fig. 2).


In 22 of 86 samples that were submitted specifically for IP, no conclusive results were obtained, and the reasons included a predominance of non-viable cells (12 samples), reactive T-cell lymphocytosis (7 samples), and non-specific antibody binding (3 samples). More of the samples that had a generic diagnosis of lymphoma (21 samples) had a null result from IP compared with samples that had a specific subtype (1 sample). Among the samples that had non-contributory IP results, the vast majority were large cell lymphoma (20 samples) (Table 2).

The success rate for IP was 74.4% (64 of 86 samples), which included the total number of samples that yielded either positive or negative results. Sixteen samples that were diagnosed as B-cell lymphoma were subclassified based on successful IP results: 5 were subclassified as chronic lymphocytic leukemia/small lymphocytic lymphoma, 4 as follicular lymphoma, 3 as mantle cell lymphoma, 3 as Burkitt lymphoma, and 1 as a primary effusion lymphoma (Fig. 3). The primary effusion lymphoma that had no IP results was diagnosed based on molecular studies. IP revealed a population of abnormal T cells with an aberrant T-cell antigen expression profile in 3 samples that were diagnosed as T-ALL and in 1 sample that was diagnosed as peripheral T-cell lymphoma, NOS. The mean number of cell surface markers performed for the subclassification of B-cell lymphomas was 5.8 markers versus 9.8 markers for T-cell lymphomas. Furthermore, 2 samples that were diagnosed as post-transplant lymphoproliferative disorder were subclassified based on a B-cell population that was identified on IP.

Figure 3.

Serous effusions samples diagnosed as lymphoproliferative disorders. (A) This cytospin preparation of a large B-cell lymphoma reveals small lymphocytes and large cells with scant blue cytoplasm and multiple nucleoli (centroblasts) (Romanowsky stain; ×100). (B) Liquid-based preparation (ThinPrep) showing the predominance of small lymphoid cells in a case of follicular lymphoma (Papanicolaou stain; ×100). (C) This cytospin preparation reveals follicular lymphoma with small lymphoid cells that have angulated nuclei and small or inconspicuous nucleoli (centrocytes) mixed with macrophages and occasional small lymphocytes (Romanowsky stain; ×100). (D) Laser-scanning cytometry histograms from the same case reveal a monoclonal B-cell population that expresses CD19, CD20, CD10, and surface immunoglobulin (sIg) with lambda light-chain restriction.


IHC was performed in 24 samples (17 cell block samples and 7 cytospin preparations). This included 2 samples that had noncontributory results for lambda and kappa light chains because of high background cytoplasmic staining. The panel for IHC was selected depending on the cytomorphology of each sample. For some samples, the panel was used to work up a broad differential diagnosis; and, for other samples, the panel was used for lymphoma subclassification. For B-cell NHL subtyping, IHC was performed in 2 follicular lymphomas, 2 Burkitt lymphomas, and 2 primary effusion lymphomas. For the other lymphoproliferative disorders, 1 sample each of T-ALL, post-transplantation lymphoproliferative disorder, and myeloma was submitted for IHC (Table 2).

Molecular Assays

Among the samples that had a final diagnosis of lymphoproliferative disorder, 9 samples were submitted for PCR, 8 for FISH, and 3 for EBER analysis. Relatively more samples that had a specific subtype of lymphoma (9 of 47 samples; 19.1%) were sent for molecular tests compared with samples that had a generic diagnosis (11 of 82 samples; 13.4%) (Table 2). Among the samples that had a specific subtype identified, PCR analysis for B-cell or T-cell monoclonality was performed in 3 samples: Two of those samples were positive for B-cell monoclonality, and 1 sample was negative for both B-cell and T-cell monoclonality. One of the monoclonal B-cell samples also was positive for HHV-8 by PCR (primary effusion lymphoma).

FISH analyses for MYC rearrangement and/or IGH/BCL2 translocation were performed on 3 samples. EBER analysis was performed on 3 samples, in which 2 were positive (post-transplant lymphoproliferative disorder) and 1 was negative (primary effusion lymphoma) for EBER.


The current study demonstrates that the routine use of a triage and testing algorithm for SE samples allowed a generic diagnosis of lymphoproliferative disorder in the majority of cases and a specific subtype in cases that underwent multiple ancillary testing. IP (laser scanning cytometry or flow cytometry) was successful in almost all samples that had a specific subtype. Ancillary studies, either as a single test or in combination, were completed in a considerable number of samples. To the best of our knowledge, this is the largest study to date examining the use of a multiparametric approach for subtyping lymphoproliferative disorders in SE samples and, more importantly, using the latest WHO classification.

In our study, the specific subtyping of samples was feasible in the context of both the initial diagnosis of lymphoma or as a way to confirm recurrence. The diagnoses of lymphoproliferative disorder were based only on cytomorphologic and ancillary test findings. For patients who had a previous history of lymphoma, the results helped to confirm recurrence or transformation to an aggressive subtype. For cases in which the initial diagnosis was provided by the effusion sample, we did not confirm the accuracy by doing a cytohistologic correlation. To perform such an evaluation would have required a different study design, which was beyond the scope of the current report. However, when a complete combination of morphologic evaluation, IP, and all additional ancillary tests is consistent with a specific subtype, this should suffice for a definitive diagnosis. Furthermore, in our large series of SE samples, both B-cell and T-cell lymphomas were successfully subclassified as well as cases of post-transplant lymphoproliferative disorder and myeloma. In the past 40 years, few studies have attempted to examine the best way to characterize lymphoproliferative disorders in fluids. Moreover, different classification systems and techniques have been used, which has made comparison of the results very difficult.

Multiple ancillary tests were performed with a high success rate as part of our multiparametric approach for lymphoma subtyping. Our approach in handling SE differs from that reported in prior studies essentially in the total number of cases and ancillary tests performed for each patient. Previous reports evaluated few parameters, such as cytomorphology only[21, 22]; cytohistologic correlation[23]; cytoautopsy confirmation[24]; a dual combination of cytomorphology and IP[4, 25]; or a trimodality approach of cytomorphology, IHC, and flow cytometry for nucleic acid content.[6] By incorporating the results from IHC (24 of 84 samples), PCR (27 of 95 samples), and Southern analysis (13 of 63 samples), 20 lymphomas in 95 lymphoid-rich effusions were successfully diagnosed.[8] Previous series consisted of a small number of cases and included diagnoses of lymphoma in 16,[26] 33,[25] 7,[27] and 48 cases.[4] When IHC was used in conjunction with cytomorphology, one study demonstrated a diagnosis of 22 lymphomas by performing IHC analyses for CD45, CD3, CD20, and CD30[5]; and another study examined 54 cases of lymphoma that were submitted for CD20 and CD45RO analyses.[7] Several other techniques, such as flow cytometry for DNA, RNA, and proliferative activity,[6] computer-interactive morphometry[28, 29] and reverse transcriptase in situ PCR to detect HHV-8 virus,[30] were studied, but they were not used commonly in routine cytology practice.

The use of a triage algorithm applied to SE samples enabled us to select cases that would yield results from multiple ancillary tests and maximize the yield from even small-volume samples. Indeed, cell-surface IP (flow cytometry or laser scanning cytometry) was performed routinely as the first ancillary technique, because it requires the specimen to remain in fresh state without any fixatives and allows testing the sample for multiple markers using a small volume as well as the possibility of receiving results on the same day. According to our triage algorithm, if flow cytometry provides conclusive results or if there is no residual material after flow cytometry, then the samples may not be submitted for IHC, which explains the small number of cases that were tested by IHC. In addition, some tests, such as flow cytometry, have evolved over the years, and new markers are available to allow the possibility of performing a wide range of these tests in small-volume samples. Even when performing an immediate assessment, the failure rate for cell-surface IP was 28%. In our study, the results from IP were non-contributory in the majority of our samples that were diagnosed as large-cell lymphoma, similar to what has been reported for FNAs. The cells from large-cell lymphoma are particularly fragile and, thus, frequently are under represented in cell-surface markers studies like flow cytometry.[31, 32] In such cases, testing for B-cell or T-cell clonality by PCR can be helpful. Another alternative to be fully explored for those samples is the use of a split-signal IGH FISH-chromogenic in situ hybridization DNA probe. A recent study demonstrated that this probe was effective in detecting IGH translocations and diagnosing B-cell lymphoproliferative disorders in several FNAs and in a few effusion cases.[33]

The main reasons for non-conclusive ancillary tests were similar to those reported in other studies and included the presence of non-viable cells, reactive T-cell lymphocytosis, and non-specific antibody binding. Only one previous study clearly stated the percentage of failed tests (26.3%) for immunocytochemistry because of high background or insufficient material.[8] In our series, 2 of 24 samples that were submitted for IHC had failed results because of high cytoplasmic background staining for kappa and lambda light chain.

A limitation in the current study is that we examined the application of our algorithm in a retrospective manner; therefore, there was often more than one SE sample submitted from the same patient on the same day, and only one of the samples was tested, causing the percentage of cases submitted to ancillary testing to be lower than expected. In addition, our study encompassed a long period during which there were changes in methodology and markers, making it difficult to have all cases assessed for the same tests and all panels performed for each marker. However, with the ever-increasing number of molecular markers/assays and the advances in flow cytometry, a prospective study conducted with the same tests and panels being performed for each marker may be unrealistic.

In summary, specific subtyping of lymphoproliferative disorders was successfully achieved in approximately 33% of cases that were tested for ancillary studies using an approach for the triage and aliquoting of SE samples. This approach enabled us to obtain results from multiple tests even in small-volume specimens. Additional marker studies also may be possible using the same algorithm and may increase the amount of diagnostic, prognostic, and predictive information that is becoming increasingly important for patient management.


Mauro A. Saieg was a research fellow supported by the Terry Fox Foundation Strategic Health Research Training Program in Cancer Research at the Canadian Institutes of Health Research (grant number TGT-53 912).


The authors made no disclosures.