Clinical utility of CD23 and FMC7 antigen coexistent expression in B-cell lymphoproliferative disorder subclassification



Background: CD23 and FMC7 are normal B-cell antigens utilized during diagnostic immunophenotyping of suspected lymphoproliferative disorders. However, the diagnostic utility of coexistent antigenic expression patterns with simultaneous two-color staining and flow cytometric analysis has not been studied extensively. Methods: Using multiparameter flow cytometry, we evaluated the expression pattern of FMC7 and CD23 in 218 cases of B-cell lymphoma from blood and bone marrow specimens. Results: The CD23(+)/FMC7(-) pattern was the most common pattern in patients with chronic lymphocytic leukemia and related variants. The widest variation of patterns was found in patients with follicular cell lymphoma, large cell lymphoma, and Waldenström's macroglobulinemia, a lymphoplasmacytoid disorder, although most cases expressed the CD23-/FMC7(+) pattern. The CD23 and FMC7 antigen, along with the CD5 coexpression pattern, provides critical adjunctive data. These data allow accurate classification of the majority of cases, thereby providing a key aspect of a reliable diagnostic algorithm. The CD23 and FMC7 antigen expression pattern, along with selected other antigens, was predictive of subtypes in >95% of lymphoproliferative cases and narrowed the differential diagnosis in the remaining cases. Conclusion: The flow cytometric CD23/FMC7 expression pattern achieved by multicolor immunophenotyping facilitates accurate and reproducible classification of B-cell lymphomas and has diagnostic utility. Cytometry (Clin. Cytometry) 50:1–7, 2002. © 2002 Wiley-Liss, Inc.

The accurate diagnosis of lymphoproliferative disorder (LPD) is one of the most challenging in surgical pathology. Several ancillary techniques, such as flow cytometry, cytogenetics, and molecular DNA-based studies, play an important role in diagnostic accuracy. None of these is designed to replace histopathology, but enhances the accuracy and precision of lymphoma diagnosis and subclassification. The immunophenotypic approach to lymphoma classification utilizes a variety of monoclonal antibodies, which can be used in different combinations and employed in routine clinical diagnostic practice as either tissue immunohistochemistry or multiparameter flow cytometry. There is little consensus as to the best approach to flow cytometric combinations of monoclonal antibodies. Some attempts to reach consensus agreement on the optimal and minimal acceptable diagnostic approaches in leukemia and lymphoma diagnosis and classification have been published (1–3). However, clinical practice still is largely individualistic and based more on regional apprentice style teaching rather than on published data.

The use of flow cytometry is being recognized increasingly as a necessary diagnostic evaluation for the diagnosis of B-cell LPDs. Many studies have been performed to determine the expression patterns of different antigens (e.g., CD5, CD10, CD20, and CD23) on different types of B-cell LPDs and their utilization for subclassification of the disorders. The antigens CD23 and FMC7 are expressed on normal B cells and are utilized commonly in the immunophenotypic subclassification of suspected LPD (4–9). However, the diagnostic importance of studying the coexistent expression pattern of these two antigens (CD23 and FMC7) has not been studied extensively. Our recent study (10) of CD23 and FMC7 antigen coexistent expression patterns in tissue biopsies from patients with LPD indicated diagnostic usefulness in the classification of lymphoma. For this reason, we designed this study to evaluate coexistent expression patterns of CD23 and FMC7 in cases of B-cell LPD with firmly established diagnosis of chronic lymphocytic leukemia (CLL), prolymphocytoid transformation of chronic lymphocytic leukemia (CLL/PL), prolymphocytic leukemia (PLL), hairy cell leukemia (HCL), mantle cell lymphoma (MCL), marginal zone lymphoma (MZL), Waldenström's macroglobulinemia (WM), large cell lymphoma (LCL), follicular cell lymphoma (FCL), and Burkitt's lymphoma (BL) in blood and bone marrow specimens. We also evaluated other antigens (CD5, CD10, CD11c, CD19, CD20, CD25, and CD103) in all above subclassifications of B-cell LPD in an effort to define a robust diagnostic algorithm integrating our observations on CD23 and FMC7 coexistent expression.


Patient Samples

From January 1997 to August 1999, samples of peripheral blood and bone marrow were analyzed retrospectively in the Analytical Cytometry Lab at William Beaumont Hospital (Royal Oak, MI). The samples were taken from 218 patients with a firmly established diagnosis of B-cell LPD. The study was carried out in accordance with the institutional guidelines for human investigation. All phenotypic studies were correlated with morphologic findings and other laboratory studies (molecular diagnostic gene rearrangements, cytogenetics, and immunohistochemistry) to establish a definitive diagnosis of LPD by trained hematopathologists at William Beaumont Hospital.. The cases were re-reviewed to confirm diagnostic uniformity for the purposes of this study by the authors using the diagnostic criteria of the revised European American lymphoma (REAL) classification.

Flow Cytometric Analysis

Peripheral blood (5 ml of EDTA anticoagulated specimen) and bone marrow (1–2 ml of EDTA or heparinized bone marrow aspirate) were diluted to 15 ml with 0.1% of bovine serum albumin/phosphate-buffered saline (BSA/PBS). This preparation was washed with centrifugation at 1,500 rpm twice more with 0.1% BSA/PBS in an automated cell washer (DAC II, Baxter Scientific, Chicago, IL). Preconjugated monoclonal antibodies used as a three-color panel for lymphoma on 10 aliquots of the sample included CD2, CD3, CD4, CD5, CD7, CD8, CD10, CD16, CD19, CD20, CD22, CD23, CD38, FMC7, CD45, and kappa and lambda light chains. The combination of CD23-PE (Becton-Dickinson Immunocytometry System [BDIS], San Jose, CA) and FMC7-FITC (Immunotech, Marseilles, France, a Beckman Coulter company) was employed on all samples. Immunophenotyping staining procedures were standard “whole blood lysis” techniques using FACSLyse (BDIS). Specimens were analyzed on a FACScan flow cytometer (Becton Dickinson, San Jose, CA) equipped with Cellquest software or on a Cytoron Absolute flow cytometer (Ortho Diagnostic Systems, Johnson & Johnson Co., Raritan, NJ) equipped with Ortho Research software. The flow cytometers acquired a minimum of 10,000 cellular events as list mode files containing forward angle scatter, log side scatter, and three log fluorescence signals. List mode files were analyzed using WinList software (Verity Software House, Topsham, ME) and CD45/side scatter gating methods (11). Expression of the antigen was interpreted and scored for this study as follows: positive (+), negative (-), and partial (<30%) or minor subset expression (±). This arbitrary distinction at the 30% level for partial expression was selected using isotype controls to establish the cursor placement for antigen positivity. Additionally, it refers to 30% of the abnormal B-cell population and not to the entire lymphocyte population, which reflects some degree of subjectivity or judgement.


Examples of the various CD23 and FMC7 antigen expression patterns identified among the 218 cases are shown in Figure 1. As T cells included in any lymphoid population visualized using CD45 side scatter gating analysis of list mode files represent a subpopulation of cells lacking both CD23 and FMC7 expression, interpretation of the clonal B-cell population CD23 and FMC7 expression must take this fact into account. The complete results of CD23 and FMC7 coexpression patterns and LPD diagnostic subtypes are summarized in Table 1.

Figure 1.

Coexistent expression patterns of CD23 and FMC7 commonly seen in various B-cell LPDs. The abnormal lymphoid populations are present in various proportions, but represent a significant population among lymphoid cells gated in blood and bone marrow samples using CD45 and side scatter analysis methods. Patterns of CD23 and FMC7 coexpression as categorized in the study include CD23(−)/FMC7(−) (histogram A), CD23(+)/FMC7(−) (histograms B,H), CD23(±)/FMC7(±) (histogram C), CD23(±)/FMC7(+) (histogram D), CD23(−)/FMC7(+) (histograms E,F), CD23(±)/FMC7(+) (histogram G), and CD23(+)/FMC7(+) (histogram I). Benign T cells in the CD45/side scatter gated population also constitute a population of CD23(−) and FMC7(−) lymphoid cells.

Table 1. Observed Immunophenotypic Patterns of CD23 and FMC-7 in LPD Subtypes*
CD23/FMC-7 patternLPD diagnosis
  • *

    Scoring of antigen expression: +, positive; −, negative; ±, dim or partial expression.



All of the diagnosed cases of CLL (n = 121) expressed CD23 and CD5. Of these 121 cases, 105 cases (88%) were CD23(+), but did not coexpress FMC7. In the remaining 15 cases (12%), only a minor subset of the clonal B lymphocytes coexpressed FMC7.


In this diagnostic category, defined as 10–55% of the lymphocytes showing prolymphocytoid morphology, six of eight cases (75%) were CD23(+). Of these six cases, only one was negative for FMC7, four showed a minor subpopulation of FMC7 positive cells, and one case fully coexpressed CD23 and FMC7. The remaining two cases were FMC7 (+) but did not coexpress CD23.


The two cases with a PLL diagnosis were CD23(-) and FMC7(+).


All 15 cases were CD23(-) and expressed FMC7.


Six cases were reviewed under this category, which is considered a form of lymphoplasmacytoid leukemia/lymphoma (LPL). Four cases (67%) were CD23(-) and FMC7(+). One case coexpressed CD23 and FMC7 and the other was CD23(+) and FMC7(-).


All 20 cases were CD23(-) and FMC7(+). These cases comprised cases of lymph node-based disease with blood or marrow involvement and cases of classical splenic MZL.


Of 24 cases, 17 (71%) were CD23(-) and FMC7(+). The remaining seven cases (29%) were FMC7(+) with a minor subpopulation of CD23 (±) cells.


All 14 cases were CD23(-) and FMC7(+).

Diffuse LCL

The largest variation of patterns was seen in this group. Unlike that found in the CLL group, the pattern spread was due to variability in the expression of both CD23 and FMC7. These six cases showed three different patterns of CD23 and FMC7 coexpression. Two cases were CD23(-) and FMC7(+), two cases were CD23(±) with a subpopulation of FMC7(+), and the remaining two cases lacked detectable expression of both CD23 and FMC7 (dual negative).


Both cases were CD23(-) and FMC7(+).

Various studies (12–14) have reported a relationship between CD20 expression and the FMC7 antigen epitope. In particular, these studies raise technical issues with regard to simultaneous multicolor immunophenotypic staining with FMC7 and anti-CD20 and raise questions as to the diagnostic usefulness of the FMC7 antigen. Our immunophenotypic panel for the study of our 218 cases of B-cell LPD utilized an anti-CD20 reagent (clone L27, peridinin chlorophyll protein labeled; Becton Dickinson) in a separate tube (along with anti-immunoglobin light chain antibodies) from that of the FMC7. Thus, we compared the CD20 expression with that of FMC7 expression without concern for steric hindrance of antibody binding. We observed no reliably consistent correlation between CD20 and FMC7 expression among our cases (Fig. 2).

Figure 2.

Expression of CD20 does not parallel FMC7 expression in LPDs. Eleven cases of various types of LPD are shown. They demonstrate CD20 and FMC7 antigen density, but no consistent relationship of the expression level or pattern is observed.


Utilization of flow cytometry in the diagnosis and subclassification of B-cell LPD is very sensitive and specific (1, 15). Immunophenotypic characterization of certain subgroups of B-cell LPD is well documented and generally well accepted as the major defining feature of some specific LPD entities (1, 8–10, 15–25). Specific LPD entities defined by immunophenotype with minimal morphologic correlation include CLL (CD5+, CD10-, CD23+, FMC7-), CLL/PL (CD5+, CD10-, CD23+, FMC7±), MCL (CD5+, CD10-, CD23-, FMC7+), MZL (CD5-, CD10-, CD23-, FMC7+), and HCL (CD5-, CD10 -/+, CD23-, FMC7+, CD25+, CD11c+, CD103+). Our findings indicate that the coexistent expression pattern of CD23 and FMC7 can add further information toward the reproducible subclassification of LPD in blood and bone marrow specimens, similar to that previously reported for diagnostic tissue biopsies (10).

After showing the monoclonality for kappa or lambda light chains, the expression of CD5 on B cells is a most important step in the subclassification of B-cell LPD. The B-cell LPD that expresses CD5 includes CLL and it variant and MCL. The presence of CD5 expression in B-cell CLL is well documented and is recognized in a recent classification system (18–22). Our findings closely parallel recent studies indicating that CD5- CLL, if truly an entity, constitutes a very minor population of patients with this form of low-grade B-cell leukemias (20). Most of these reported CD5- B-cell CLLs show more aggressive clinical behavior (20, 26). In our study, none of 121 cases of B-cell CLL were truly negative for CD5 expression, all expressed CD23, and only 15 cases show a subpopulation that expresses FMC7. CLL/PL and PLL are considered variants related to CLL, although the true origin of PLL remains controversial. Both of these variants are different from B-cell CLL by their clinical behavior and immunophenotypical properties (26). When B-cell CLL is being transformed into CLL/PL, the percentage of FMC7+ cells increases as does the expression of surface immunoglobulin. The expression of CD5 remains on these neoplastic cells during this transformation. In our study, we observe the same pattern of persistent CD5 expression with FMC7 in cases of CLL/PL (n = 8) and PLL (n = 2). Only one of eight cases of CLL/PL was negative for FMC7. Both our cases of PLL were positive for FMC7. Our findings indicate the pattern of CD23(+) and FMC7(±) to be very characteristic of CLL/PL, thus providing a means of indicating patients with a potentially more aggressive clinical course and worthy of closer clinical follow-up. A minor subset of CLL cases (15 of 121) also had coexistent expression of CD23 and FMC7, which we speculate to be cases in the process of early transformation into CLL/PL. As the number of PLL cases is limited in this study, we hesitate to generalize from our experience of only two cases, other than to confirm previous observations of PLL being FMC7+ (19, 22). Another subgroup in B-cell LPD that expresses CD5 is mantle cell leukemia/lymphoma (21, 23). Recognition of this subgroup is of clinical importance due to the relatively poor outcome and potentially different clinical management for patients with MCL (26, 27). Expression of CD23 in B-cell CLL and coexpression of CD23 and FMC7 in CLL/PL and PLL distinguish these more indolent LPD subtypes from MCL (CD23- and FMC7+) as initially described by Stein et al. (28). There are few reported cases of MCL with CD23 expression (23). All cases (N = 15) in our study of MCL were CD23(-) and FMC7(+).

In cases of CD5- B-cell LPD, the diagnostic possibilities includes MZL, FCL, HCL, WM, LCL, and BL. Although MZL is typically CD5-, there are rare cases reported that show CD5 expression (29, 30). Most of these cases presented with more advanced disease and frequent relapses, which is unusual for this low-grade B-cell LPD. All of our cases of MZL were CD5-, CD23-, and FMC7+. Follicular center cell lymphoma is clonal B lineage with a common expression of CD10 and/or CD38, which is typically CD5- (31). In our study, all our cases were CD5-, CD10+, FMC7+. The expression of CD23 was variable, as 17 cases were negative and 7 cases showed a positive subpopulation of abnormal cells. These two patterns of coexistent expressions were similar to our previous observations of FCL in tissue biopsies (10). Interestingly, although only 24 cases of FCL were studied in these blood and bone marrow specimens, the majority of cases lacked significant CD23 expression. In our previous study of 62 cases of FCL from tissue biopsy specimens (10), 45 of 64 cases exhibited significant CD23 expression The significance of this apparent anatomical difference is unknown, but one might speculate this phenotypic difference could relate to either clonal evolution in advanced stage disease or a predilection for FCL cells lacking CD23 to enter the peripheral blood circulation and seed the bone marrow.

Previous studies by Hubl et al. (12) suggested that FMC7 antigen expression paralleled that of CD20, implying the information to be redundant. However, their methodology paired FMC7 in the same staining antibody cocktail with CD20, which has been reported to block potentially the FMC7 binding epitope (13, 14). Recent evidence by Serke et al. (13) indicated that the FMC7 antibody is in fact binding to a CD20-related epitope. However, the level of expression of the FMC7 epitope and CD20 expression, at least in neoplastic LPD cells, is far from a perfect correlation. Our results of CD20 expression measured separately from the sample studied for FMC7 showed an imperfect relationship and indicated that FCM7 expression provides diagnostic information incremental to that of CD20 expression levels. The histograms in Figure 2 were selected to show the spectrum of interrelationships in the binding of the CD20 and FMC7 clones used in our study and do not represent an exhaustive study into the relationship of these two B-cell epitopes. The difference in our findings from those of Hubl et al. (12) might be due to clone selection or to the fact that the expression of these two antigenic sites was measured in separate tubes. Although the purpose of this study was not focused on the relationship between CD20 clones and FMC7 epitope expression, our observations seemingly refute the concept that FMC7 expression can be predicted simply by other reagents directed to CD20.

In summary, we found that simultaneous staining of B-cell neoplasms for expression of CD23 and FMC7 is a useful tool to assist in the reproducible subclassification of LPD in blood and bone marrow, particularly in those instances where it is clinically relevant for prognostic and therapeutic purposes. The CD23/FMC7 pattern of coexpression, when used in combination with other monoclonal antibodies, has a similar diagnostic utility as the CD5/CD20 expression pattern. We believe that CD23, FMC7, and CD5 expression in clonal B-cell populations provides a diagnostic algorithm (Fig. 3) that can be utilized adjunctively with other data. In our experience, this resulted in the accurate classification of >95% of B-cell LPD. This study on blood and bone marrow specimens yielded similar patterns in various subtypes of LPD to those observed in our previous study on tissue biopsies (10), thus making the proposed diagnostic algorithm useful in all types of diagnostic specimens. We conclude that the simultaneous CD23/FMC7 coexistent expression pattern, as defined by flow cytometric immunophenotyping in conjunction with other antibodies, is a valuable contribution toward accurate and reproducible classification of B-cell LPD. The prognostic significance of the CD23 and FMC7 expression pattern in lymphoma requires more extensive study, but may be worthwhile, especially in low-grade LPD, where a heterogeneity of CD23 and FMC7 antigen coexistent expression patterns occurs.

Figure 3.

Antigenic algorithm for classification of B-cell LPDs. After identification of a monoclonal B-cell population, the above approach using the CD23 and FMC7 pattern provides an antigenic subclassification of the various subtypes of LPD. Along with morphologic and other immunophenotypic data, this apptoach can provide reproducible categorization of at least 95% of cases.


We are grateful to Katharine A. Steel, Cindy Sounart-Miscovitch, and Carole A. Ceckowski for their expert technical assistance with flow cytometric analysis.