CD19 is expressed on most B-cell lymphomas; however, the frequency and types of B-cell lymphomas with low-level expression of CD19 are not well characterized.
CD19 is expressed on most B-cell lymphomas; however, the frequency and types of B-cell lymphomas with low-level expression of CD19 are not well characterized.
We reviewed flow cytometric histograms specifically for decreased CD19 expression on 349 cases analyzed by the Flow Cytometry Laboratory at University Hospitals of Cleveland (Cleveland, Ohio). Results of flow cytometry were correlated with the morphologic diagnosis.
Of the cases reviewed, 125 (36%) showed a visible decrease in CD19 expression compared with normal B lymphocytes. Decreased CD19 expression was noted in 79% of follicular lymphomas (27 of 34), 36% of small lymphocytic lymphomas/chronic lymphocytic leukemias (82 of 228), 31% of mantle cell lymphomas (4 of 13), 24% of diffuse large B-cell lymphomas (8 of 33), and 13% of marginal zone B-cell lymphomas/lymphoplasmacytoid lymphomas (4 of 30) and was not observed in any Burkitt lymphoma (0 of 5) or hairy cell leukemia (0 of 6). Decreased CD19 expression was significantly more frequent in follicular lymphomas than in other lymphoma subtypes (P < 0.001). No significant difference was observed in the frequency of decreased CD19 expression based on histologic grade of follicular lymphoma.
Diminished expression of CD19 expression occurs frequently in B-cell lymphomas, in particular follicular lymphoma, and may be helpful in identifying B-cell lymphoma cells in complex cell mixtures such as bone marrow specimens. © 2004 Wiley-Liss, Inc.
CD19 is expressed throughout B-cell development until terminal plasma cell differentiation and therefore is expressed in most mature B-cell lymphomas (1). The presence of this antigen usually indicates a B-cell lineage neoplasm, but it is less useful for subclassifying B-cell lymphomas. With the advent of three-color and four-color flow cytometry, gating strategies using CD19 in conjunction with forward and side scatter are useful to document clonality of B-cell lymphomas because CD19-expressing cells can be electronically isolated from other cells within the sample as an aid to determine light chain restriction (2, 3).
Surprisingly, alterations in the expression of CD19 have not been well characterized on B-cell malignancies. Several studies have indicated that the number of CD19 molecules expressed on some B-cell lymphomas, in particular chronic lymphocytic leukemia/small lymphocytic lymphoma (CLL/SLL) cells, is smaller than that on normal B lymphocytes by quantitative analysis (4–6). However, studies of this type have been limited to analysis of B-cell lymphomas of the blood and have not addressed the expression of CD19 in tissue biopsies. In addition, quantitative analysis is not often performed on samples in a clinical setting, making a numerical measurement of CD19 molecules less useful for diagnostic purposes.
A more clinically relevant and simpler tool is the ability to visually detect differences in the expression of a marker, such as CD19, between normal and neoplastic cell populations. In some instances, the intensity or level of expression of an antigen can help in classifying a lymphoma and aid in identifying an abnormal cell population in a complex mixture of cells. Low-level expression of CD20 in small CLL/SLL cells is one parameter that can help distinguish CLL/SLL from phenotypically similar mantle cell lymphomas (7, 8), and decreased expression of CD10 is a feature more often observed in grade 3 follicular lymphomas than in other follicular lymphoma subtypes (9). The diminished expression of CD3, typical of mycosis fungoides/Sézary cells, has been used to identify mycosis fungoides/Sézary cells interspersed with benign T cells in the peripheral blood and enhances the sensitivity of detecting these cells by flow cytometry (10). Knowledge of the frequency of altered CD19 expression in various B-cell lymphomas is needed to determine the usefulness of using decreased CD19 to identify low-level or minimal residual B-cell lymphoma cells in peripheral blood and bone marrow specimens.
Therefore, we evaluated a large number of B-cell lymphomas for a decrease in CD19 expression by using CD19 expressed on normal B lymphocytes as a comparison. Our results indicated that a visible decrease in CD19 expression is observed in many B-cell lymphomas and is particularly common in follicular lymphomas, regardless of grade. An example of the usefulness of a gating strategy that uses CD19-dim lymphocytes to identify minimal residual disease is presented. Knowledge of this relatively frequent but unappreciated alteration should be useful to the practicing cytometrist to increase the sensitivity of detecting B-cell lymphoma cells in samples composed of complex cell mixtures.
Cases were identified from the files of the Flow Cytometry Laboratory of University Hospitals of Cleveland (Cleveland, OH). Only cases with diagnostic flow cytometry at presentation and subsequent morphologic/immunohistochemical confirmation with peripheral blood, bone marrow, and/or tissue examination were included, and 349 cases were examined in this study. Over a 3-year period, from 2001 through 2003, by review of files of the Flow Cytometry Laboratory of University Hospitals of Cleveland, 121 lymphoma cases were identified, consisting of 34 follicular lymphomas, 13 mantle cell lymphomas, 33 diffuse large B-cell lymphomas (DLBCLs), 30 marginal zone B-cell lymphomas/lymphoplasmacytoid lymphomas, 5 Burkitt lymphomas, and 6 hairy cell leukemias (Tables 1 and 2). An additional database of 228 CLL/SLL cases, primarily peripheral blood samples, culled from over a 5-year period (1998 through 2003) and analyzed by the same laboratory, was also included in the study (Tables 1 and 2). Diagnoses of CLL/SLL and marginal zone B-cell lymphoma/lymphoplasmacytoid lymphoma were derived primarily from flow cytometric analysis and peripheral blood/bone marrow morphology. Diagnosis of the remaining cases was verified by review of histologic sections. The sample distribution of the lymphomas for each lymphoma subtype is also presented in Table 1. Control peripheral blood samples were attained from normal donor blood samples used for daily quality control and peripheral blood samples sent for enumeration of T, B, and natural killer cells in which no abnormality was identified. Analysis of CD19 expression on lymphoid cells in bone marrow specimens was attained from samples that were negative lymphoma staging marrows, patients who had benign hematologic disorders, or patients who had acute leukemia in remission.
|Lymphoma typea||Specimen type||With CD19 internal control|
|Lymph node||Blood||Bone marrow||Tissue|
|Diffuse large B||10||2||6||15||69.7% (23/33)|
|Hairy cell||0||2||4||0||50% (3/6)|
|Lymphoma type||Total||CD19 intensity|
|Dim (%)||Normal (%)|
|SLL/CLLa||228||82 (36%)||146 (64%)|
|Follicular||34||27 (79%)||7 (21%)|
|Mantle cell||13||4 (31%)||9 (69%)|
|Diffuse large B cell||33||8 (24%)||25 (76%)|
|Burkitt||5||0 (0%)||5 (100%)|
|Marginal/ lymphoplasmacytoid||30||4 (13%)||26 (87%)|
|Hairy cell leukemia||6||0 (0%)||6 (100%)|
|Total non-SLL/CLL||121||43 (36%)||78 (64%)|
Peripheral blood and bone marrow aspirates were collected in heparin tubes, and lymph node and tissue specimens were placed in RPMI medium. All specimens were processed within 24 h of receipt. For staining, aliquots (usually 100 μl) of blood and bone marrow samples were used directly. Cells from lymph nodes and tissue specimens were harvested by using mechanical dispersion and resuspended in RPMI before staining. Specific cell-dispersing enzymes were not used. Staining was performed by standard methods as directed by the manufacturer (BD Biosciences, San Diego, CA) with red cell lysis. For light chain analysis, cells were washed once in phosphate buffered saline, placed in RPMI plus 10% fetal calf serum, incubated at 37°C for 30 min, and washed again before staining. Four-color immunophenotyping was performed by using combinations of antibodies labeled with fluorescein isothiocyanate (FITC), phycoerythrin (PE), or peridinin chlorophyll protein (PerCP), and allophycocyanin (APC).
All antibodies were purchased from BD Biosciences. Antibodies pertinent to this study included kappa (clone TB 28-2, FITC conjugate), lambda (clone 1-155-2, PE conjugate), CD45 (clone 2D1, PerCP conjugate), CD10 (clone H10a, FITC conjugate), and CD19 antibody (clone SJ25C1, APC conjugate). Antibody staining was performed as directed by the manufacturer in the standard manner, and samples were analyzed on a FACSCalibur flow cytometer (BD Biosciences).
Alignment of forward scatter, side scatter, FL-1, FL-2, and FL-3 channels on the flow cytometers was checked by daily calibration with yellow-green beads (Polysciences Inc., Warrington, PA) and maintained within the coefficients of variation recommended by the manufacturer. Alignment of the FL-4 channel was assessed daily with blue beads (Spherotech, Libertyville, IL) as suggested by the manufacturer. Linearity checks and compensation were performed daily with Calibrite beads (BD Biosciences) according to the manufacturer's recommendation, and compensation was rechecked with CD45-stained lymphocytes every other day.
For analysis, 100,000 events were acquired, and cells were analyzed with CellQuest software (BD Biosciences). A linear scale was used for forward and right-angle light scatter, to evaluate fluorescence in all channels according to our routine laboratory procedure, and to enhance detection of differences in antigen intensities. Flow cytometric histograms of lymphoma cases were specifically reviewed for decreased CD19 expression by visual inspection. Expression of CD19 in the abnormal clonal B-cell population was compared with that of residual polyclonal normal B-cells in the kappa-FITC, lambda-PE, CD45-PerCP, and CD19-APC tubes by two of three independent reviewers (H.M., N.A., and J.P.), and a consensus was reached on each specimen. In cases in which a benign B-cell population was difficult to identify, CD19 expression was evaluated based on comparison with benign tissue samples run during the same period on the same instrument. The percentages of cases with available CD19 internal intensity controls for each lymphoma subtype are presented in Table 1. It should be noted that only 108 of the 228 CLL/SLL cases were specifically evaluated for the presence of a normal B-cell population.
The CD19 channel values on normal B lymphocytes was determined with CellQuest quadrant statistics, with cuts based on isotype controls after CD45 and side scatter gating. CD19-dim events were defined as CD19-positive events that fell between CD19 expressed on normal B lymphocytes and a CD19-negative cell population after CD45 and side scatter gating.
Statistical analysis was performed chi-square test and Excel software (Microsoft, Redmond, WA).
A total of 349 B-cell lymphoma cases was reviewed for decreased CD19 expression in the neoplastic cell population. Patients' median age at the time of diagnosis was 67.7 years (range 10–101). Only two patients were younger than 20 years, and both were diagnosed with Burkitt lymphoma. An example of decreased CD19 expression in a B-cell lymphoma is shown in Figure 1. Fluorescence shifts of CD19 of approximately 100 to 200 channel values made the distinction between neoplastic and benign B-cell populations plainly visible. This degree of altered CD19 expression was typically seen in follicular lymphomas. Smaller shifts were more difficult to identify, especially if a normal B-cell population was absent, as was the case for most CLL/SLL cells. When detected, shifts of only 20 to 30 channel values were commonly seen in these lymphomas (Fig. 2). Only cases in which the fluorescence shift was clearly visible were considered to have low-level CD19 expression.
Decreased CD19 expression, as defined above, was observed in 125 of 349 lymphomas examined (36%; Table 2). Of the lymphomas studied, follicular lymphomas showed low-level CD19 expression significantly more often than did other B-cell lymphoma subtypes, with 79% (27 of 34) of cases having diminished CD19 expression (P < 0.001, chi-square test). This was followed by CLL/SLL cells (36%), mantle cell lymphomas (31%), diffuse large cell lymphomas (24%), and marginal zone B-cell lymphomas/lymphoplasmacytoid lymphomas (13%). Decreased CD19 expression was not observed in any of the five cases of Burkitt lymphoma or in any of the six cases of hairy cells leukemia, although only a few of these cases were examined. Tissue source did not affect on the percentage of cases with decreased CD19 expression within a lymphoma subtype (data not shown).
Low-level CD19 expression on follicular lymphomas was further analyzed according to the histologic grade of the corresponding tissue biopsy specimens. No significant difference was noted in the frequency of decreased CD19 level among the different grades of follicular lymphomas (Table 3). Interestingly, low-level CD19 expression was identified in 40% of the CD10-positive DLCBLs, whereas only 11% of the CD10-negative cases showed diminished CD19 expression (P = 0.05, chi-square test; Table 4).
|Follicular lymphoma subtype||CD19 dim (%)||CD19 normal (%)||Total|
|All||27 (79%)||7 (21%)||34|
|Grade 1||6 (75%)||2 (25%)||8|
|Grade 2||12 (80%)||3 (20%)||15|
|Grade 3||7 (78%)||2 (22%)||9|
|Ungradeable||2 (100%)||0 (0%)||2|
|CD19 dim (%)||CD19 normal (%)||Total|
|CD10+||6 (40%)*||9 (60%)||15|
|CD10−||2 (11%)*||16 (89%)||18|
|All||8 (24%)||25 (76%)||33|
Surface light chain-negative B-cell lymphomas occasionally occur (11). The inability to document clonality by light chain restriction limits the sensitivity of detecting these lymphoma cells in bone marrow or blood specimens. To determine the usefulness and sensitivity of using low-level CD19 expression in the analysis of surface light chain-negative B-cell lymphomas, we examined CD19 expression on 32 peripheral blood and 21 bone marrow samples uninvolved by lymphoma or leukemia and stained with the CD19-APC antibody. By using CD45 and side scatter to gate the lymphocytes, the average channel value shifts of CD19 on normal B lymphocytes were 572 over the negative control population in the peripheral blood and 545 over the negative control population in the bone marrow. The average numbers of events that fell between the CD19-positive cells and the negative control population were 0.04% in the peripheral blood and 0.03% in the bone marrow. Therefore, the expected sensitivity of bone marrow or peripheral blood analysis for identifying a malignant B-cell population by using only a CD19-dim and CD45 side scatter gating strategy (as in surface light chain-negative cases) is somewhat higher than 0.03% to 0.04%. Hematogones (CD45-dim, CD19-positive cells) were specifically excluded from these analyses.
Enhanced sensitivities would be expected in surface light chain-positive CD19-dim B-cell lymphomas. An illustration of the use of diminished CD19 expression to detect low-level bone marrow involvement by a surface light chain-positive follicular lymphoma is shown in Figure 3. In the example shown, the patient was known to harbor a follicular lymphoma with decreased CD19 expression. In the tube stained with anti–kappa-FITC, anti–lambda-PE, anti–CD45-PerCP, and anti–CD19-APC labeled antibodies, 100,000 total events were acquired and the lymphocytes were gated by using CD45 and side scatter in the standard manner. Analysis of the total CD19-positive cells within the lymphocyte gate did not show clear evidence of clonality as assessed by the kappa/lambda ratio (kappa/lambda = 68/29; Figs. 3B and 3E). However, when CD19-dim cells were gated, marked skewing of the kappa/lambda ratio (kappa/lambda = 87/4) was readily apparent and consistent with clonality (Figs. 3C and 3F). In this particular sample, approximately 0.07% of the bone marrow cells were clonal kappa-positive B-cell lymphoma cells.
Our results show that the intensity of CD19 expression on lymphoma cells is visibly decreased in 36% of all B-cell lymphomas with low-level CD19 significantly more common in follicular lymphomas than other lymphoma types, regardless of grade, occurring in 79% of cases. Therefore, diminished expression of CD19 may be one helpful feature in distinguishing follicular lymphomas from other B-cell lymphomas and in recognizing lymphoma cells when interspersed among other cell types. Due to its frequency, gating strategies that distinguish CD19 expression on normal cells from that on neoplastic lymphoid cells is particularly useful to detect marrow involvement by B-cell lymphomas and to monitor minimal residual disease.
Decreased CD19 expression was more common in CD10-positive DLBCLs (40%) than in CD10-negative DLBCLs (11%), which may relate to the follicle center cell origin of the former, although the difference was only at the level of statistical significance (P = 0.05). The frequency of diminished CD19 expression on CD10-positive DLBCLs (40%) was less than that observed on follicular lymphomas (79%; P = 0.008 chi-square test), even though the frequency of low-level CD19 expression in grade 3 follicular lymphomas (78%) was as common as in follicular lymphomas of lower grades. Further, none of the five Burkitt lymphomas, which are germinal center cell neoplasms, showed decreased CD19 expression. Therefore, although the alteration is common in follicular lymphomas, the abnormality appears to be less frequent in more aggressive lymphomas of germinal center cell origin. Additional analyses incorporating molecular data would be of interest to determine whether a relation exists between a specific genetic abnormality, translocation, or gene expression profile and CD19 expression.
The lymphoma dataset analyzed in this study included a large number of CLL/SLL cases. Although previous studies have indicated that CD19 expression is slightly decreased in CLL/SLL compared with normal B cells by quantitative analysis, the percentage of cases with this decrease had not been determined (4–6). A large data pool of CLL/SLL cases was felt to be helpful to decrease the error in categorizing these lymphomas by visual inspection. Nonetheless, the frequency of lymphomas with a visually detectable decrease in CD19 expression was essentially the same for cases of CLL/SLL and non-CLL/SLL.
It should be noted that many of CLL/SLL cases contained few normal B lymphocytes, making the visual comparison with normal B lymphocytes difficult. Nevertheless, 61% of cases contained an identifiable, albeit small, normal B-cell population that served as an internal control for CD19 intensity. In these cases and those of other lymphomas, which lacked an internal control for CD19 positivity, we attempted to compare the CD19 expression level with that of normal B lymphocytes run on the same instrument on the same day. This limited our sensitivity for detecting changes of surface CD19 expression on these lymphomas, and it is likely that we underestimated the frequency of decreased CD19 expression on CLL/SLL cases in particular, because CD19 expression in CLL/SLL is usually only slightly depressed (roughly 20 to 30 channel values). To be useful, the alteration of CD19 expression needs to be clearly discernible visually from normal CD19 expression on B lymphocytes. As such, we found that the percentage of CLL/SLL cases that had decreased CD19 intensity necessary to visually discriminate these cells from normal B lymphocytes was only 36%.
Analysis of the CD19 expression on normal B lymphocytes in bone marrow and peripheral blood showed that approximately 0.03% to 0.04% of all events show low-level CD19 expression after gating on lymphocytes. Therefore, the expected sensitivity for identifying a CD19-dim malignant B-cell lymphoma cell using CD19 expression alone to identify the neoplastic cells is slightly higher than 0.04%. Some of the normal CD19-dim events that lack surface light chain and are likely granulocytic or monocytic elements within the lymphocyte gate or T or natural killer cells with high CD19 background staining. In typical practice, however, light chain staining is used to verify clonality and therefore neoplasia.
By using methods that use low-level CD19-expressing lymphocytes coupled with an additional forward scatter subgate for cell size, we have found that sensitivities in the range of 0.01% to 0.02% can be reached with the acquisition of 100,000 total events, assuming that a minimum of 10 to 20 light chain-restricted cells is adequate to document clonality. Obviously, the acquisition of more cells would enhance the sensitivity of the analysis. It should be noted that, in occasional cases, surface light chain is not expressed on B-cell lymphoma cells (11). In those instances, the identification of a population of CD19-dim cells within the lymphocyte gates that constitutes greater than 0.04% of all events should raise the suspicion of tissue involvement by malignant B-cell lymphoma cells, particularly if the lymphoma is known to express low levels of CD19.
We used only a single CD19 antibody in this study, an APC conjugate of clone SJ25C1 from BD Immunocytometry Systems. We have observed similar findings of CD19-dim lymphoma cells with other fluorescent conjugates of this same clone. The reactivity of different CD19 monoclonal antibodies in our B-cell lymphoma cases was not examined because this study was a retrospective analysis. Therefore, we cannot be absolutely sure that other CD19 monoclonal antibodies would produce similar findings and that some idiosyncratic reactivity or unusual behavior by the SJ25C1 CD19 monoclonal antibody can be totally excluded. This antibody is widely used in clinical laboratories, and atypical staining patterns have not been observed by or reported to the manufacturer (personal communication from BD Immunocytometry Systems). In addition, at least 15 published studies have used this particular monoclonal antibody (12–25). Moreover, we used multiple lots of the reagent over the course of the study period without any noticeable change in the reactivity pattern. Therefore, we believe it is unlikely that this antibody is unique in identifying lymphoma cells with low-level CD19 expression. However, only a prospective study incorporating a head-to-head comparison of SJ25C1 with other CD19 monoclonal antibodies in an analysis similar to ours can confirm this assumption.
The frequency with which low-level expression of CD19 on B-cell lymphomas is encountered raises the question of mechanisms that lead to its decreased surface expression. Pax-5 appears to mediate transcriptional control of CD19 (26). However, immunohistochemical studies of pax-5 expression in B cell lymphomas have indicated that the protein is strongly expressed in virtually all B-cell lymphomas, suggesting that other mechanisms may play a role in decreasing CD19 expression (27, 28). A reported perturbation that results in altered expression of CD19 is the absence of CD81, a tetraspanin that normally associates with CD21 and CD19 on the surface of B lymphocytes (29–31). In studies in mice, the targeted disruption of CD81 has resulted in decreased surface expression of CD19, likely due to disruption of the normal trafficking or membrane stability of CD19 (29–31).
CD19 lowers the threshold required for activation through the B-cell antigen receptor; CD19-deficient mice have impaired antibody responses to thymus-dependent antigens, decreased germinal center formation, and impaired affinity maturation of serum antibodies, and CD81-deficient animals show similar characteristics (29–34). The relation between these two antigens is intriguing, particularly because CD81 appears to be the receptor for hepatitis C virus, which is known to be involved in lymphomagenesis (35, 36). Surprisingly, CD81 expression on B-cell lymphomas has not been systematically characterized, although CD81 is decreased on peripheral blood B cells in patients who have hepatitis C virus associated with lymphoproliferative disorders (37). Therefore, the role of CD81, if any, in the decreased CD19 expression that we have observed in B-cell lymphomas, in particular follicular lymphomas, should be investigated.
Differences in the expression of an antigen may be helpful in determining the type of lymphoma and identifying an abnormal cell population. Previous analyses examining the expression of CD19 have focused on quantitative analyses of the molecule on B-cell lymphoproliferative disorders of the blood (4–6). Studies of this type, although informative, do not provide an indication of the practical value of identifying low-level CD19 expression in lymphomas because most laboratories discriminate normal from abnormal populations by visual inspection of histograms rather than by deriving a numerical measurements of surface molecule expression. In addition, the diminished expression of CD19 in B-cell lymphomas in nodal specimens has not been previously investigated. Our study is a simple evaluation of CD19 in various B-cell lymphomas to determine the prevalence of the diminished expression of this antigen. A detailing of this alteration in the B-cell lymphoma subtypes should be useful to the practicing cytometrist in the evaluation of hematopoietic neoplasms and in identifying B-cell lymphoma cells in complex cell mixtures such as bone marrow aspirates.