Preliminary results of this study were presented as a poster at the XXI International Congress of the ISAC; San Diego, California; 4–9 May 2002.
Flow cytometric immunophenotyping including Bcl-2 detection on fine needle aspirates in the diagnosis of reactive lymphadenopathy and non-Hodgkin's lymphoma†
Article first published online: 24 JAN 2005
Copyright © 2005 Wiley-Liss, Inc.
Cytometry Part B: Clinical Cytometry
Volume 64B, Issue 1, pages 34–42, March 2005
How to Cite
Laane, E., Tani, E., Björklund, E., Elmberger, G., Everaus, H., Skoog, L. and Porwit-MacDonald, A. (2005), Flow cytometric immunophenotyping including Bcl-2 detection on fine needle aspirates in the diagnosis of reactive lymphadenopathy and non-Hodgkin's lymphoma. Cytometry, 64B: 34–42. doi: 10.1002/cyto.b.20043
- Issue published online: 17 FEB 2005
- Article first published online: 24 JAN 2005
- Manuscript Accepted: 29 JUN 2004
- Manuscript Received: 26 JAN 2004
- Karolinska Institute
- Cancer Society of Stockholm
- East Europe Committee of the Swedish Health Care Community, Stockholm, Sweden
- Swedish Institute
- World Federation of Scientists, Lausanne, Switzerland
- flow cytometry;
Fine-needle aspiration (FNA) with immunophenotyping by immunocytochemistry (IC) on cytospins has recently received increased consideration in the diagnosis of lymphoma. The aim of our study was to establish the diagnostic value of a four-color flow cytometric (FCM) panel, including cytoplasmic Bcl-2, in cytologic diagnosis of malignant non-Hodgkin's lymphoma (NHL) and reactive lymphoid hyperplasia (RH).
We investigated 424 FNAs from 396 patients. FCM panel included lambda/kappa/CD19/CD5, CD23/CD10/CD20/CD19, CD4/CD7/CD8/CD3 and Bcl-2/CD10/CD19/CD3 in fluorescein isothiocyanate, phycoerythrin, and peridinin chlorophyll protein or a tandem conjugate of R-phycoerythrin and indodicarbocyanine and allophycocyanin. Bcl-2 expression was evaluated separately for gated B and T cells.
In 97% of 172 RH samples, FCM was concordant with the diagnosis. FCM gave correct immunologic diagnosis in 95% of low-grade B-cell NHLs, 78% of high-grade B-cell NHLs, and 53% of T-cell lymphomas. Malignant B cells had higher Bcl-2 expression than did reactive B and T cells. This helped to establish a correct diagnosis especially in cases where no clear-cut monoclonality could be shown by kappa/lambda staining or where there was no expression of surface light chain. The highest Bcl-2 expression was found in follicular lymphomas.
Our FCM panel allowed precise classification of NHL in FNA material in 89.5% of all samples. Bcl-2 staining can be recommended for primary differentiation between reactive hyperplasia and NHL. © 2005 Wiley-Liss, Inc.
The new World Health Organization classification of hematopoietic and lymphoid tumors classifies lymphoid malignancies as distinct biologic entities based on morphology, immunophenotype, genetics, and clinical features (1). The relative importance of each of these methodologies varies among different lymphomas (1). The diagnosis of non-Hodgkin's lymphoma (NHL) on cytologic material has historically been controversial (2). However, our previous publications (3–6) and those of many other investigators (7–12) have shown that when fine-needle aspiration (FNA) is combined with immunophenotyping by immunocytochemistry (IC), the accuracy of NHL detection can exceed 90%. The IC on cytospins is rather time consuming, and routinely only one antigen can be assessed per cell. Immunophenotyping can also be accomplished with multiparameter flow cytometry (FCM), which is a rapid and sensitive method to detect lymphoid marker expression (13). FCM evaluates several antigens on one cell, gives quantitative results, and can detect small abnormal cell populations against a reactive background. Further, current techniques allow detection of intracytoplasmic antigens, thus closing the gap between FCM and IC. These features significantly improve the diagnostic sensitivity and therefore are particularly useful in lymphoma diagnostics (14). Several studies have evaluated FCM immunophenotyping in FNA material, showing a high detection rate (up to 88%) of NHL (15–30). All these studies have emphasized the role of FNA in diagnosis of primary or recurrent NHL and have shown good concordance with subsequent surgical biopsies. However, most of the studies cited are based on retrospective material. Further, large studies evaluating the reliability of FCM by comparison with cytologic diagnosis based on IC in the same FNA samples from patients with lymphoma remain scarce (16, 24, 31, 32).
The aim of this prospective study was to establish the value of FCM panel, consisting of four-color antibody combinations, in diagnosis and classification of lymphomas in nodal and extranodal sites by FNA cytology. We applied six surface B-cell–associated markers (CD19, CD20, CD23, CD10, kappa, lambda) and five T cell-associated markers (CD3, CD4, CD5, CD7, CD8). Moreover, in one tube we included detection of intracytoplasmic levels of Bcl-2 protein in B and T cells present in the sample.
MATERIALS AND METHODS
We prospectively included samples from 3,038 patients with suspect lymphoma referred to the Division of Clinical Cytology, Karolinska Hospital, Stockholm, Sweden, for FNA between May 1999 and October 2001. From a total of 3,038 FNAs performed during that period, IC was performed in 1,088 samples. Samples with cell concentrations of fewer than 5 × 105 cells/ml, with large cells suggesting Hodgkin's lymphoma, with metastasis of other tumors, and from lymph nodes smaller than 1 cm in diameter from children were excluded from the study. In total, 424 FNA samples obtained from 396 patients (187 male, 209 female), ranging in age from 4 to 94 years (mean ± standard deviation 56 ± 19) were analyzed by FCM. Of 396 investigated patients, 369 had a single FNA sample taken. Twenty-six patients had FNA twice and one patient had FNA three times. In these patients, FNAs were performed at different occasions and/or from different lesions. Of the 424 FNAs studied by FCM, 352 samples (83%) were taken from lymph nodes and 72 (17%) from extranodal tissue (17 from the parotis, 14 from subcutaneous tissue, nine from breast tissue, seven from the orbita, seven from the tonsil, five from the spleen, four from skin lesions, three from submandibular glands, two from kidney tumors, two from the thyreoidea, one from the gum, and one from the liver). Reactive hyperplasia of lymphoid tissue (RH) was diagnosed in 172 samples from 164 patients and NHL in 239 samples from 228 patients. In 141 patients a primary lymphoma was diagnosed. Recurrent lymphoma was found in 97 samples from 87 patients. Thirteen samples (seven cases of Hodgkin's lymphoma, two cases of angioimmunoblastic lymphadenopathy, two cases of dermatopathic lymphadenopathy, one case of lateral neck cyst, and one case of seroma) were excluded from further analysis.
Definitions of Diagnosis and Classification Used
The initial diagnosis on FNA was made by two cytopathologists. Cytologic and immunologic criteria for the classification of NHL and RH are described in detail in our previous publication in which categories of the Revised European-American Classification of Lymphoid Neoplasms were defined in terms of cytologic findings (33, 34). Primary lymphoma was defined as a previously undiagnosed lymphoma in which the FNA was the primary diagnostic test. Recurrent lymphoma was defined as one in which the diagnosis of lymphoma had been made by a previous biopsy procedure. For the purpose of this study, the same observers reviewed all samples with discrepant diagnoses. For histopathologic diagnosis, Revised European-American Classification of Lymphoid Neoplasms and World Health Organization classifications were used (1, 33). The final diagnosis was recorded according to histopathologic or cytologic report if a biopsy was not performed.
All FNA biopsies were performed by an experienced cytopathologist who used a needle 0.4 to 0.6 mm in diameter, according to the procedure described by Zajicek (35). Deep-seated lesions were aspirated under ultrasound guidance. One part of the aspirate was used to prepare smears, which were air dried or methanol fixed and stained by May-Grünwald-Giemsa or Papanicolaou technique, respectively. Additional air-dried smears were fixed in freshly prepared 4% buffered formalin and used for proliferation study (see below). The other part of the aspirate was suspended in phosphate buffered saline. Cell viability by trypan blue exclusion and cell concentration in suspensions was assessed immediately after the FNA procedure. Cytospin preparations in a cytocentrifuge (Shandon, Cheshire, UK) for IC analysis (3) were made from one half of the cell suspension and the other half was used for FCM study.
IC on cytospins was performed in 60% (253 of 424) of samples and during the first study year in 70% (73 of 103) of samples. IC was not performed in all cases because after the first year of study we saw an excellent correlation between IC and FCM in low-grade B-cell NHLs (LG-NHL) and RH. Therefore, immunophenotyping was done subsequently only by FCM in most cases of RH and of recurrent LG-NHL.
A three-step alkaline phosphatase anti-alkaline phosphatase method (using alkaline phosphatase anti-alkaline phosphatase reagents from DakoCytomation, DAKO, Glostrup, Denmark) was employed (3). The following monoclonal antibodies were used: anti-kappa (clone A8B5, dilution 1:10), CD10 (clone SS2/36, dilution 1:20), CD20 (clone L26, dilution 1:50), Bcl-2 (clone 124, dilution 1:40; DAKO) and anti-lambda (clone 1-155-2, dilution 1:50), CD3 (clone SK7, dilution 1:40), and CD5 (clone L17F12, dilution 1:40; Becton-Dickinson [BD], San Jose, CA, USA). Proliferation index (PI) analysis was performed with antibody Ki-67 (MIB-1) from Immunotech (Marseille, France) detected by a peroxidase-avidin-biotin complex (DAKO) method. The percentage of proliferating tumor cells was determined by counting at least 200 presumably neoplastic cells at a high-power field magnification in randomly selected areas of the smears (36).
Lymph node excision (n = 91) or biopsies from extranodal tissues (n = 33) were performed in 124 of 396 patients. Tissue specimens were fixed in formalin or B5 fixative and routinely stained with hematoxylin and eosin, Giemsa, periodic acid-Schiff, and Gordon-Sweet methods. Immunophenotyping was performed by FCM on lymph node cell suspensions or on paraffin sections by a standard immunoperoxidase method.
FNA aspirates collected in tubes containing phosphate buffered saline were processed within 2 h. A “stain and then lyse/wash” technique was used. The panel of applied monoclonal antibodies is listed in Table 1. Intracytoplasmic Bcl-2 staining was performed with Intrastain (DAKO), according to the manufacturer's instructions.
|FITC||PE||PerCP or PE-Cy5||APC|
Data acquisition and analysis.
Data acquisition was performed with a FACS-Calibur flow cytometer (BD) equipped with CellQuest software (BD). Light scatter characteristics and autofluorescence levels of normal peripheral blood lymphocytes were used as reference values for instrument settings. Instrument setup and calibration were performed with CaliFlow beads (Spherotech, Inc., Libertyville, IL, USA). On average, 8,874 nongated events (range 544 to 10,000) were acquired per tube. Data analysis was performed with CellQuest (tubes 1–3) and Paint-a-Gate (BD; tube 4; see Fig. 1 for details). Bcl-2 expression was evaluated by mean fluorescence intensity (MFI) for separately gated CD3+ T cells, CD19+ B cells, and CD10+/CD19+ B cells. Bcl-2 MFI for normal T cells present in the samples remained the same throughout the study (MFI 93.6 ± 31.8, median 89), which confirmed stability of the measurements allowing comparison of Bcl-2 expression in various NHL categories.
FCM criteria for phenotypic classification of LG-NHL and RH are presented in Table 2. Only follicular lymphoma (FL), chronic lymphocytic leukemia (CLL), and mantle cell lymphoma (MCL) displayed characteristic immunophenotypes (Table 2). FCM differentiation between immunocytoma/lymphoplasmocytic lymphoma (IC/LPL) and extranodal marginal zone B-cell lymphoma of mucosa-associated lymphoid tissue type/nodal marginal zone lymphoma (MALT/NMZL) was not possible, but localization and clinical data allowed correct distinction in most cases. Scatter characteristics was one of the important features in diagnosis of high-grade B-cell NHL (HG-NHL). Monoclonality was defined as a kappa/lambda ratio (K/L) higher than 6 or lower than 0.3 (28).
Descriptive statistics, analysis of variance, and Dunnett's T3 post hoc tests were used in SPSS 9.0 (SPSS, Inc., Chicago, IL, USA) to analyze differences in frequencies of B and T cells, PI, and Bcl-2 expression across NHL categories. P ≤ 0.05 was considered statistically significant.
Accuracy of Diagnosis Across NHL Subtypes
Our FCM panel was very efficient in detecting LG-NHL. The LG-NHL categories were correctly diagnosed in 95% (177 of 186) of cases (Table 3). In total, histopathologic examination was available in 46% (86 of 186) of LG-NHL cases. Cytologic diagnosis was concordant with histopathology in all these cases (100%). Eight samples were diagnosed by FCM as unspecified LG-NHL. In two samples histopathology confirmed CD10− FL based on morphologic observations and Bcl-6 expression. Another two samples that were reported by FCM as unspecified LG-NHL appeared to be MCL. On review of FCM plots of both samples, we found a low level of CD5 expression that was not correctly evaluated during initial analysis. Cyclin D1+ MCL was confirmed in one of these cases by lymph node histopathology and in another one by bone marrow examination. In four patients LG-NHL subclassification was not possible. Cytologic findings suggested FL because of a predominance of small to medium-size centrocyte-like cells with cleaved nuclei and few centroblast-like cells. However, these lymphomas were CD10− and fluorescent in situ hybridization (FISH) analysis did not show t(14;18), t(11;14) or t(11;18) in three studied cases. FISH analysis showed extra copies of chromosomes 11, 14, and 18 in one case and trisomy 18 in one case. In one case material for genetic studies was not available. Biopsy was not performed because patients were older than 76 years and the biopsy was considered uninformative regarding treatment decisions. In one sample of recurrent MALT/NMZL, FCM did not detect a monoclonal B-cell population. This sample, obtained from soft tissue, had a very low cell count; on average only 1,500 events were registered per tube.
|Cytopathologic diagnosis||Cases (n)||FCM concordancea||Histopathologic confirmation||FCM discordanceb|
|FL||47||45/47 (96%)||28/47||2/47 (4%)|
|CLL||26||26/26 (100%)||10/26||0/26 (0%)|
|IC/LPL||8||8/8 (100%)||2/8||0/8 (0%)|
|MALT/NMZL||15||15/15 (100%)||2/15||0/15 (0%)|
|MCL||6||4/6 (67%)||4/6||2/6 (33%)|
|HG-NHL||29||22/29 (76%)||10/29||7/29 (24%)|
|TCL||7||3/7 (43%)||5/7||4/7 (57%)|
|Total||142||123/142 (87%)||61/142||19/142 (13%)|
|FL||33||33/33 (100%)||0/33 (0%)|
|CLL||32||32/32 (100%)||0/32 (0%)|
|IC/LPL||2||2/2 (100%)||0/2 (0%)|
|MALT/NMZL||8||7/8 (88%)||1/8 (12%)|
|MCL||5||5/5 (100%)||0/5 (0%)|
|HG-NHL||7||6/7 (86%)||1/7 (14%)|
|TCL||10||6/10 (60%)||4/10 (40%)|
|Total||97||91/97 (94%)||6/97 (6%)|
HG-NHL was diagnosed in 36 FNA samples from 35 patients (31 cases of diffuse large B-cell lymphoma [DLBCL], two of Burkitt-like lymphoma, two of T-cell–rich B-cell lymphoma [TCRBCL], and one of anaplastic B-cell lymphoma). Anaplastic B-cell lymphoma and both TCRBCL diagnoses were confirmed by tissue biopsy. Immunoglobulin (Ig) heavy-chain gene rearrangement analysis was done in one case of TCRBCL, which showed clonal rearrangement.
In 78% (28 of 36) of samples FCM suggested a diagnosis of lymphoma. Ig light-chain restriction was detected in 22 samples of DLBCL and 2 of Burkitt-like NHL. In four samples malignant B-cells did not express surface Ig (sIg). FCM did not provide a correct diagnosis in one case of anaplastic B-cell lymphoma, two cases of TCRBCL, and five cases of DLBCL. In these samples analysis showed a dominant reactive T-cell population with very few B cells.
T-cell lymphoma (TCL) was found in 17 patients. TCL diagnosis was confirmed in 13 cases by tissue histopathology, and T-cell receptor gene rearrangement analysis was done in four primary and three recurrent cases. In 53% of TCL cases (three of seven primary lymphomas and six of 10 recurrent lymphomas) FCM showed aberrant findings within a T-cell population. These aberrant findings were a high CD4/CD8 ratio (11–23) in three samples, CD4+CD7− population in three samples, CD4/CD8 double negative in one sample, or double positive population in one sample. T-cell precursor leukemia/lymphoma diagnosis was confirmed by an additional study that showed expression of terminal deoxynucleotidyl transferase. In one sample FNA and cytology showed reactive changes in an enlarged lymph node. However, in this case CD4/CD8 ratio was 17.7 and recurrence of TCL was confirmed by T-cell receptor gene rearrangement analysis.
Accuracy of Diagnosis in Reactive Lymphadenopathy
In 97% of 172 RH samples FCM was concordant with the diagnosis. In four aspirates FCM and IC suggested a possibility of lymphoma. In the first case FCM showed a K/L of 10.6. Cytology showed a mixed-cell population of small lymphocytes and relatively large blast-like cells were present. Clonality analysis by IC was not adequate due to background staining. In the second case a subpopulation of lambda monoclonal CD10+ B cells with high Bcl-2 expression was found. Therefore, a possible partial engagement of FL was reported by FCM. In these two patients a subsequent lymph node biopsy could not confirm a diagnosis of lymphoma and Ig heavy-chain rearrangement analysis showed a polyclonal pattern. In another patient FCM indicated suspicious lymphoma due to a subpopulation of B cells expressing dim kappa, dim CD5, and high levels of Bcl-2. The IC finding was discordant to FCM by showing dominant lambda-positive B cells. In the fourth case of suspected lymphoma FCM and IC showed dominant lambda-positive B cells (K/L 0.6). No lymph node excision biopsy was performed in the latter two patients, but FISH analysis showed normal chromosomes 11, 14, and 18. However, none of these patients developed lymphoma during 2-year clinical follow-up. In the fifth discrepant sample FCM showed dim kappa positivity in the CD19+ B-cell population, but FNA cytology and IC were suggestive for infectious mononucleosis, which was later confirmed by laboratory tests.
B-Cell and T-Cell Counts
Frequencies of B cells, T cells, and CD4/CD8 ratios in different NHL categories are listed in Table 4. The B-cell count was significantly higher in LG-NHL compared with RH (P < 0.001 for FL, CLL, IC/LPL, MALT/NMZL; P = 0.002 for MCL). No difference in B-cell counts was found between RH and HG-NHL (P = 0.998). T-cell counts showed a significant difference between RH and all B-cell NHL categories (P < 0.001). Among B-cell lymphomas B-cell count was lowest in HG-NHL, with a significant difference between HG-NHL and categories of LG-NHL (P < 0.001 for CLL, P = 0.001 for IC/LPL, P = 0.041 for MALT/NMZL, and P = 0.007 for MCL), except for FL (P = 0.085). We also found a significant difference in B-cell count and T-cell count between CLL and FL (P < 0.001 for both cell subsets). Results of statistical analysis performed in 124 histopathologically confirmed cases did not differ from overall results.
|FL||64.8 ± 18.2 (80)||26.1 ± 14.9 (80)||3.9 ± 2.7 (80)||21.0 ± 17.1 (74)|
|CLL||79.5 ± 13.3 (58)||16.6 ± 12.5 (58)||3.6 ± 2.1 (58)||10.6 ± 5.2 (51)|
|IC/LPL||79.1 ± 13.6 (10)||17.8 ± 11.5 (10)||3.7 ± 2.4 (10)||8.0 ± 4.9 (10)|
|MALT/NMZL||69.7 ± 19.8 (23)||19.6 ± 15.1 (23)||3.6 ± 2.2 (23)||12.3 ± 11.8 (16)|
|MCL||82.2 ± 20.5 (11)||12.9 ± 15.0 (11)||1.8 ± 1.6 (11)||23.0 ± 16.3 (8)|
|HG-NHL||50.4 ± 25.3 (35)||25.7 ± 19.5 (35)||2.0 ± 2.1 (35)||55.0 ± 18.7 (27)|
|TCL||27.6 ± 21.7 (17)||61.6 ± 22.3 (17)||7.8 ± 7.2 (15)||36.3 ± 27.5 (13)|
|RH||45.6 ± 17.6 (172)||44.9 ± 17.1 (172)||4.5 ± 2.9 (172)||15.8 ± 12.7 (69)|
K/L was assessed within a gated CD19+ B-cell population. On average, K/L in RH was 1.7 (median 1.6, range 0.4–4.7). Of 222 B-cell NHL samples, 128 showed monoclonal kappa+ and 77 monoclonal lambda+ populations. In eight cases (four HG-NHL, three FL, and one CLL) no expression of sIg was found. Two of these cases were also negative for both light chains by IC. IC showed expression of cytoplasmic lambda in four samples and cytoplasmic kappa in one and was unsuccessful in one due to technical reasons.
FCM showed polyclonal B cells in nine lymphoma samples (one case of anaplastic B-cell lymphoma, two of TCRBCL, five of DLBCL, and one MALT/NMZL). Two samples were monoclonal for kappa and one for lambda by IC. In five samples clonality could not be determined by IC due to background staining. Kappa/lambda staining by IC was not performed in one case of recurrent MALT/NMZL because of characteristic cytologic findings.
The CD4/CD8 ratio was highest in RH (4.5 ± 2.9) and lowest in MCL (1.8 ± 1.6, P = 0.002) and HG-NHL (2.0 ± 2.1, P < 0.001; Table 4). Among different lymphoma categories, CD4/CD8 ratio was significantly lower in MCL and HG-NHL than in FL (P = 0.013 and P = 0.001, respectively) and lower in HG-NHL than in CLL (P = 0.008). When only histopathologically confirmed cases were included in statistical analysis, the results remained the same.
As expected, the PI was highest in HG-NHL and lowest in IC/LPL (Table 4). The PI was significantly higher in HG-NHL than in RH (P < 0.001) and LG-NHL (P < 0.001 for FL, CLL, IC/LPL, MALT/NMZL and P = 0.007 for MCL). The PI in RH was significantly higher than in IC/LPL (P = 0.022), but it was not higher than the PI in other LG-NHL. Among LG-NHL, the PI was significantly higher in FL than in CLL (P = 0.003) and IC/LPL (P < 0.001). Histopathologically confirmed cases of CLL had a higher PI (14.1 ± 11.4) than did all other CLL cases, because open biopsy was performed mostly in cases in which transformation was suspected.
Expression of Bcl-2 was examined in 72 cases of FL, 52 cases of CLL, eight cases of IC/LPL, 18 cases of MALT/NMZL, 10 cases of MCL, 29 cases of HG-NHL, and 137 cases of RH (Fig. 2). As expected, similar levels of Bcl-2 expression were found in T cells in B-cell NHL (MFI 94.7 ± 44.7) and RH (MFI 94.8 ± 28.3). In contrast, malignant B cells had significantly higher mean Bcl-2 MFI values (266.8 ± 207.6) than did reactive CD10− B cells (105.3 ± 31.6). Further, normal germinal center CD10+ B cells in RH expressed considerably lower levels of Bcl-2 (MFI 54.4 ± 30.9) than did T cells and CD10− B cells. The mean MFI value for Bcl-2 in malignant B cells was highest in FL (347.0 ± 248.6). We also detected high levels of Bcl-2 expression in CLL (MFI 212.5 ± 70.5), MCL (MFI 197.6 ± 68.7), IC/LPL (MFI 179.8 ± 30.4), and MALT/NMZL (MFI 150.0 ± 82.3). Statistical analysis of Bcl-2 expression showed that B cells in RH samples expressed significantly lower levels of Bcl-2 than did B cells in lymphomas (Fig. 2). In HG-NHL we noted a heterogeneous Bcl-2 expression. Ten of 29 HG-NHL samples were considered “low” in Bcl-2 because the level of Bcl-2 expression was lower than or equal to that of reactive T cells. In nine samples of CD10+ HG-NHL very high levels of Bcl-2 were found (MFI 440.6 ± 268.1).
We also analyzed corrected MFI values for Bcl-2 defined as the ratio of Bcl-2 MFI in malignant B cells to that in T cells from the same sample. Analysis confirmed the results obtained with MFI alone. The highest corrected MFI value for Bcl-2 was found in FL (3.69 ± 2.62) and the lowest value was found in RH (1.14 ± 0.27). We found no substantial overlap between Bcl-2 MFI in B cells of RH and B-cell NHL. Only in nine cases did results obtained in reactive B cells exceed the first quartile of corrected MFI value for Bcl-2 of malignant B-cells. The first, second, and third quartile values of corrected Bcl-2 MFI in B cells were 0.97, 1.10, and 1.27 in RH and 1.55, 2.34, and 3.40 in B-cell NHL, respectively. Statistical analysis of Bcl-2 expression in histopathologically confirmed cases of B-cell NHL and RH showed similar results.
The addition of a monoclonal antibody combination that included Bcl-2 was especially helpful in lymphoma samples with presence of rests of normal germinal centers. In these samples the results of determination of K/L could be misleading, but a population of cells with high Bcl-2/CD10 expression allowed the correct diagnosis. Moreover, in other cases determination of K/L could be difficult due to weak or negative expression of sIg. We found 49 such cases (30 of FL, three of CLL, one of MCL, five of MALT/NMZL, and 10 of HG-NHL). Of those, Bcl-2 expression was analyzed in 39 cases and, in 35 of those, high expression of Bcl-2 supported a lymphoma diagnosis. In addition, Bcl-2 analysis helped to establish a diagnosis in one case of CLL, one of HG-NHL, and three of sIg-negative FL.
Of particular interest was one case of primary DLBCL in which kappa/lambda staining by FCM did not show clonality even in large cell subpopulation (B cells were sIg negative). These large cells demonstrated a CD10+/CD19+ immunophenotype with very high levels of Bcl-2. By IC the malignant large cells expressed cytoplasmic lambda and had a proliferation rate of 50%.
In this study we demonstrate that a simple four-color FCM panel allows quick and precise detection of NHL subtypes and reactive hyperplasia in FNA material. We found that inclusion of Bcl-2 into FCM panel was useful because malignant B cells in most samples showed a higher level of Bcl-2 expression than did normal B and T cells. We found that FCM and IC are equally effective in distinguishing NHL from RH and also in the diagnosis of primary and recurrent lymphomas.
Both techniques have advantages and limitations. The main advantage of IC over FCM is the requirement of a small number of malignant cells and the preservation of cellular morphology. Staining pattern, intensity of staining, and background can be assessed easily. Moreover, if neoplastic cells are fragile (as in many cases of HG-NHL), they may be destroyed during FCM analysis but successfully assessed in cytospins. However, with IC it is not possible to evaluate double antigen expression (e.g., CD5/CD23 or CD10/Bcl-2) and scoring is semiquantitative. Immunophenotyping by FCM is rapid and sensitive, provides quantitative results, and can detect small abnormal cell populations in a reactive background. The main disadvantage of FCM is unawareness of cytomorphology. For that reason a close cooperation between cytopathologists and the FCM laboratory is required.
To the best of our knowledge, there are only two published studies that directly compared FCM with IC on FNA aspirates in large series of cases (16, 24). These investigators concluded that IC and FCM are excellent methods for immunophenotyping FNA specimens for detection of NHL and can be used interchangeably. The overall concordance of the FCM and IC results in these studies was similar to that in our study.
FCM has been used as an aid to FNA diagnostics in several other studies, with an increasing rate of samples correctly diagnosed and classified by FCM: from 70–80% in early studies to more than 90% in the most recent publications (18, 25, 26, 28).
Our diagnostic FCM panel suggested lymphoma in five samples recognized as RH. In four of these samples IC and FCM were concordant in suggesting lymphoma. Many studies have reported no discordant cases in the differential diagnosis between RH and lymphoma (17–19, 21, 26, 29, 37). Our results are in agreement with those of Meda et al. (25) who reported four cases of RH that were suggestive of lymphoma in FNA. Also, Levy et al. (38) described 12 patients with RH, in whom sIg studies showed monoclonal Ig staining patterns.
Our FCM panel was very useful in detecting low-grade B-cell lymphomas (95% of cases diagnosed and classified accurately). CD10− follicular lymphoma was the only problematic category. Interfollicular neoplastic cells may lack CD10 (39) and in these cases lymphoma subtype cannot be diagnosed correctly by FCM. Approximately 10% of FLs are reported to be CD10− (40). The diagnostic accuracy in the detection of HG-NHL in our material was much lower than that of LG-NHL. Our results are similar to the study of Verstovsek et al. (41) who reported 27% false-negative FCM results in HG-NHL. A preferential loss of large cells may occur due to increased death of malignant cells. In HG-NHL acquired cell counts were approximately 1000 events lower than average (7831 vs. 8874, respectively). Also, B-cell counts were lowest in HG-NHL.
Diagnosis of T-cell NHL by FCM immunophenotyping is not simple because there is no sensitive marker for clonality. In our series FCM detected a pathologic immunophenotype in T-cell NHL in 53% of cases. This finding is similar to results reported by Meda et al. (25), but the frequency of a pathologic phenotype was somewhat lower than obtained by Yao et al. (42) who found aberrant expression of T-cell markers in 67% (14 of 21) of peripheral T-cell lymphoma samples.
One of the aims of our study was to evaluate the usefulness of FCM analysis of Bcl-2 expression in differential diagnosis of malignant lymphoma. This has been addressed in only two studies of lymph node biopsies (43, 44). Cornfield et al. (43) reported that simple dual staining with monoclonal antibodies to Bcl-2 and CD20 is valuable in differentiating neoplastic from benign germinal center cells (FL from RH). Cook et al. (44) evaluated a three-color FC panel using antibodies against CD10, CD20, and Bcl-2. Our four-color panel allowed evaluation of Bcl-2 expression simultaneously in B and T cells and, if present, in CD10+ B cells. Using this approach nonmalignant T cells present in the sample served as an internal control for the comparison of levels of Bcl-2 expression. Similar to previous studies we found that the presence of CD10+ B cells with high Bcl-2 expression was highly predictive for FL (43, 44). In contrast, CD10+ B cells in RH showed much lower levels of Bcl-2 expression than did T cells and CD10− B cells. We also noted high Bcl-2 expression in most cases of LG-NHL, which was not investigated in detail in previous studies (44). Cornfield et al. (43) unexpectedly found that T cells express higher levels of Bcl-2 in FL than in RH. In contrast, in our study Bcl-2 expression in T cells was very similar between RH and B-cell NHL. Bcl-2 expression was most useful in FL cases with partial node involvement and reactive germinal centers that made evaluation of light-chain restriction difficult. In HG-NHL Bcl-2 expression was not as informative because malignant B cells may downregulate Bcl-2 (45). Our results are in agreement with the recent report of Menendez et al. (46) who also found that malignant B cells in bone marrow and peripheral blood samples from patients with mature B-cell neoplasms, except Burkitt's lymphoma, express consistently higher levels of Bcl-2 than their normal counterparts.
The major weakness of our study is that histopathologic confirmation was available in only 29% of cases (124 of 424). Our institution has a very long tradition of diagnostic cytopathology (47). Therefore, in most cases clinicians feel confident with cytologic diagnosis, which is usually coupled with ancillary techniques. Biopsy was performed usually in diagnostically difficult cases or when histopathology was needed for making treatment decisions. Our recent report showed a good concordance between FNA and histopathology and defined some lymphoma entities that were not diagnosed with the FNA technique (6). In the present study cytologic diagnosis was concordant with histopathology in all lymphoma cases. However, there was diagnostic discordance in four cases in which cytology and FCM suggested lymphoma that could not be confirmed by histopathologic examination and/or clinical follow-up.
In summary, our proposed panel of four-color antibody combinations allowed precise classification of NHL in FNA material in 89.5% of all samples. Staining for Bcl-2 can be recommended for differentiation between reactive hyperplasia and NHL and may be very helpful in samples with partial lymph node involvement. Based on the present results we recommend performing quick stain of an FNA smear. If small to medium-size lymphatic cells predominate, indicating low-grade lymphoma, FCM should be the method of choice for immunophenotyping. When large cells predominate, IC is preferable because FCM has a high false-negative rate. Hodgkin's lymphoma and some NHLs (e.g., T-cell–rich B-cell lymphoma and anaplastic large cell lymphoma) cannot be reliably detected by FCM. For the best diagnostic accuracy a strong communication should exist between cytopathologists and FCM laboratories.
We thank Yrsa Bringensparr, Britt Lundh, Shalah Tarahumi, and Margareta Waern for excellent technical assistance and Joanna Mazur for statistical analysis. We thank Miroslav Djokic for fruitful scientific discussions. E. Laane was partly supported by the Swedish Institute and the World Federation of Scientists, Lausanne, Switzerland. FISH analysis was performed at the Department of Clinical Genetics, Karolinska Hospital.
- 1JaffeES, HarrisNL, SteinH, VardimanJW, editors. World Health Organization classification of tumours. Pathology and genetics of tumours of haematopoietic and lymphoid tissues. Lyon: IARC Press; 2001.
- 34Lymphoreticular system. In: GrayW, McKeeGT, editors. Diagnostic cytopathology. 2nd ed. London: Churchill Livingstone; 2003. p 499–548., .
- 35Aspiration biopsy cytology: part 1. Supradiaphragmatic organs. In: WiedGL, editor. Monographs in clinical cytology. Volume 4. Basel: S. Karger; 1974. p 1–29..