How to cite this article: Platt MY, DeLelys ME, Preffer FI, Sohani AR. Flow Cytometry Is of Limited Utility in the Early Identification Of “Double-Hit” B-cell Lymphomas. Cytometry Part B 2013; 84B: 143–148.
B-cell lymphomas with concurrent translocations of MYC and BCL2 or BCL6, also known as “double-hit” lymphomas (DHL), are rare malignancies characterized by aggressive clinical behavior and poor prognosis. Previous reports suggest that decreased CD20 and/or CD19 expression by flow cytometry is relatively common in DHL and may help to identify cases requiring additional cytogenetic analysis.
We conducted a retrospective analysis of 26 cases of DHL, and compared their flow cytometric characteristics to cases of Burkitt lymphoma (BL) and diffuse large B-cell lymphoma (DLBCL). Cases were analyzed by four-color flow cytometry, and bivariate dot-plots were reviewed for light scatter characteristics, CD19, CD20, CD45, and surface light chain.
Relatively few DHL cases showed dim expression of CD19 or CD20, and statistically significant differences were found only in the frequency of dim CD19 expression between DHL and BL or DLBCL. Although concomitant dim CD19 and CD20 expression was exclusive to DHL, it was present in only a minority of cases.
B-cell lymphomas with concurrent translocations of MYC and BCL2 (or, less commonly, MYC and BCL6) are rare lymphoid neoplasms characterized by aggressive clinical behavior and poor prognosis (1). These so-called double-hit lymphomas (DHL) often have morphologic features resembling Burkitt lymphoma (BL) or intermediate between BL and diffuse large B-cell lymphoma (DLBCL), and they are categorized as B-cell lymphoma, unclassifiable, with features intermediate between DLBCL and BL (BCLU) by the 2008 WHO Classification (2). However, many cases harboring these dual rearrangements are indistinguishable from conventional DLBCL and rare cases resemble follicular lymphoma with blastoid morphology or lymphoblastic lymphoma (1, 3). Although various clinical, laboratory, and pathologic parameters, such as presence of widespread extranodal disease, elevated serum LDH, intermediate cell size, and immunophenotype, have been proposed as means by which to select cases for further cytogenetic analysis, these features are not present in all cases of DHL and may also be seen with relative frequency in other high-grade mature B-cell neoplasms such as DLBCL and BL (1, 3). Hence, DHL may be currently under-recognized because straightforward diagnostic tools based on morphology and immunophenotype are currently lacking and cytogenetic analysis (by either conventional karyotype or fluorescence in situ hybridization [FISH]) is required for their definitive diagnosis.
In many pathology laboratories, not all lymphoma cases are routinely sent for karyotype, which requires fresh tissue taken at the time of surgery. FISH is routinely available at most large and reference laboratories and can confirm the diagnosis of DHL without the need for cell culture, but submission of all samples of high-grade B-cell lymphoma for such analysis may be labor-intensive, costly and unnecessary, given that DHL is estimated to represent only 3–5% of high-grade B-cell neoplasms (4, 5). Two recent reports indicate that DHL exhibits a common immunophenotype by flow cytometry, including decreased CD20 and/or CD19 expression, and suggest that flow cytometry could serve as an initial aid to the diagnosis of DHL and a method to identify cases for confirmatory cytogenetic studies early on (6, 7). In an attempt to confirm these findings in a larger series of patients and in a different laboratory setting, we conducted a retrospective analysis of flow cytometric characteristics of 26 specimens from 20 patients with DHL diagnosed at our institution, and studied the expression of CD19, CD20, CD45, and surface light chain, in addition to light scatter characteristics. To determine whether the immunophenotypic aberrancies seen in DHL were unique to such cases or more broadly seen in other subsets of highly aggressive, unclassifiable B-cell lymphoma, we studied a separate group of related high-grade B-cell lymphomas that fulfilled WHO diagnostic criteria for BCLU (2), but that lacked dual translocations involving MYC and BCL2 or BCL6. Such neoplasms have not previously been characterized in detail by flow cytometry, to our knowledge. Finally, we compared the findings of both these groups with control cases of DLBCL and BL identified during the same time period, to determine how reliably flow cytometric characteristics could be used to distinguish between various subgroups of aggressive B-cell lymphoma.
Identification of Cases and Controls
The Partners HealthCare Institutional Review Board granted approval for the study before its initiation. The electronic files of the Massachusetts General Hospital (MGH) Pathology Department were searched for cases of B-cell lymphoma on which flow cytometric analysis had been performed and on which cytogenetic and/or FISH analysis had revealed concurrent MYC and BCL2 or BCL6 rearrangements. In some cases, tissue had been sent for cytogenetic analysis at the time of frozen section evaluation at the discretion of the frozen section pathologist, while in other cases lacking cytogenetic analysis, FISH to detect MYC,BCL2, and/or BCL6 rearrangements had been performed at the time of diagnosis because BL or another highly aggressive B-cell lymphoma was a diagnostic consideration based on the morphology and/or immunophenotype. From 2004 to 2010, a total of 26 specimens from 20 patients with DHL were identified (Supp. Info. Fig. 1). All were positive for CD10 by flow cytometric analysis. Detailed morphologic, immunophenotypic, and cytogenetic findings of 11 patients (DHL-1 through DHL-11) had been previously published (3), although the flow cytometric findings of these cases were not previously studied in detail. Four patients (DHL-08, DHL-10, DHL-11, and DHL-20) had specimens collected from multiple sites of involvement and two CSF specimens were analyzed for DHL-07. In patients with multiple analyses, specimens that recapitulated the immunophenotypic characteristics of a prior specimen were excluded from the overall analysis. Cytogenetic analysis by karyotype and FISH was performed as previously described (3).
During the course of our search, we identified six additional patients with flow cytometry that met criteria for BCLU according to the 2008 WHO Classification (2) (Supp. Info. Fig. 1). These patients were analyzed as a separate group because they lacked concurrent MYC and BCL2 or BCL6 rearrangements, and therefore did not meet our criteria for DHL.
Control cases were identified by computer-assisted search of the MGH pathology files from 2007 to 2010. A total of 11 BL cases were identified in which a diagnosis of “Burkitt lymphoma” or “atypical Burkitt lymphoma” had been rendered and that had concurrent flow cytometry. Ten DLBCL control cases from 2009 were identified that had concurrent flow cytometry and cytogenetic analysis confirming the absence of concurrent MYC and BCL2 or BCL6 rearrangements. All DLBCL cases had one or more clinical, morphologic, or immunophenotypic characteristics that raised the possibility of DHL (3), and none had obvious plasmacytic or plasmablastic differentiation, which could be associated with decreased CD20 expression. These cases met criteria for the diagnosis of “Burkitt lymphoma” or “diffuse large B-cell lymphoma, not otherwise specified,” respectively, according to the 2008 WHO Classification (2, 8).
Flow Cytometric Analysis
Specimens were analyzed in our laboratory by four-color flow cytometry on a FACSCalibur flow cytometer (BD Biosciences, San Jose, CA) with data acquisition using CellQuest software, utilizing calibration and quality control measures previously published in this laboratory (9, 10). Directly conjugated antibodies and clones utilized were as follows: CD45-PerCP (2D1), CD19-APC (SJ25C1), CD20 (Leu-16)-PerCP (L27), κ-FITC (TB28-2), and λ-PE (1-155-2) (BD Biosciences). Analysis was performed on cells included in a “lymphocyte” gate as defined by low forward and side light scatter, which served to omit both low forward scatter debris and any large clumps and aggregates. All cases and controls were of sufficient viability to allow for clear definition of a lymphocyte gate for further clinical analysis. Bivariate dot-plots were reviewed by three of the co-authors (MYP, FIP, ARS) who were blinded to the diagnosis of each case, and consensus was reached regarding both the forward versus side light scatter characteristics and the intensity of expression for each surface antigen under study. Antigen expression was compared to that of normal polytypic B cells co-stained for kappa and lambda immunoglobulin light chains within the same sample. Reduced or increased antigen expression was qualitatively defined as differing by at least a half-log in expression level from that of normal B cells. If the sample was devoid of normal B cells, the intensity of antigen expression was compared with the expected intensity of expression with normal polytypic B cells in a sample of reactive follicular hyperplasia. In four DHL cases, CD20 expression was absent. Subsequent chart review revealed recent rituximab therapy, and these cases were excluded from the CD20 analysis.
Fisher's Exact Test was used to compare proportions between two groups (GraphPad QuickCalcs [http://www.graphpad.com/quickcalcs/], La Jolla, CA). P-values of <0.05 were considered statistically significant.
Clinical, Morphologic, and Cytogenetic Features
Clinical, morphologic, and cytogenetic characteristics of the patients with DHL and BCLU are summarized in Supporting Information Table 1. The 26 patients included 20 patients with DHL and 6 patients with BCLU. Of the patients with DHL, there were 9 women and 11 men with a median age of 63 years. The morphologic diagnosis was BCLU in 14 patients and 6 patients resembled conventional DLBCL. Conventional karyotype was available in 12/20 patients, and all but one showed a complex karyotype with three or more numerical or structural abnormalities. The patient lacking a complex karyotype was cytogenetically normal (46,XX) but was an incomplete study with only five metaphases available for analysis. Of the 20 patients, 17 had concurrent MYC and BCL2 rearrangements, two (DHL-17 and DHL-20) had concurrent MYC and BCL6 rearrangements, and one (DHL-3) had rearrangements involving MYC, BCL2, and BCL6 (i.e. “triple-hit” lymphoma). The patients with BCLU included four women and two men with a median age of 56 years. All, by definition, had high-grade morphology with features intermediate between DLBCL and BL, but lacked concurrent MYC and BCL2 or BCL6 rearrangements. Extranodal disease was present in 23/26 (88%) patients with DHL and BCLU, with bone marrow involvement in 9/26 (35%) patients and pathologically-confirmed central nervous system involvement in 4/26 (15%) patients.
Flow Cytometric Characteristics
Aberrant light scatter and immunophenotypic characteristics of DHL, BCLU, and control BL and DLBCL cases are summarized in Table 1.
Table 1. Summary of Flow Cytometric Characteristics of High-Grade B-cell Lymphomasa
BCLU, B-cell lymphoma, unclassifiable, with features intermediate between diffuse large B-cell lymphoma and Burkitt lymphoma; BL, Burkitt lymphoma; DHL, double-hit lymphoma; DLBCL, diffuse large B-cell lymphoma; FSC, forward scatter; sLC, surface light chain; SSC, side scatter.
Exclusions of cases from analyses were made as follows: One DHL case was omitted from CD45 analysis due to inability to assess CD45 expression. Four DHL cases were omitted from CD20 analysis due to recent rituximab therapy. One DLBCL case was excluded from light scatter analysis due to inability to assess light scatter characteristics. One DLBCL case was excluded from surface light chain analysis due to inability to assess surface light chain expression.
P = 0.0014, DHL versus BL; P = 0.050, BCLU versus BL.
P = 0.022, DLBCL versus BL.
P = 0.014, DHL versus BL; P = 0.025, DHL versus DLBCL.
Increased forward scatter was observed in the majority of DHL and BCLU cases in comparison to BL (P = 0.0014 and P = 0.050, respectively). Decreased CD19 expression was observed in 54% of DHL versus 9% of BL (P = 0.014) and 10% of DLBCL (P = 0.025) (Fig. 1A). CD20 expression levels on DHL and BCLU cells were variable, ranging from dim to bright in comparison to normal polytypic B cells, and only a minority of cases in each category was shown to have dim CD20 expression (36% of DHL and 17% of BCLU) (Fig. 1B, left). Bright CD20 expression was observed in 32% of DHL and 33% of BCLU (Fig. 1A, right; Fig. 1B, right). These proportions were generally greater than those of DLBCL and BL with decreased CD20 expression (20% and 9%, respectively) or bright CD20 expression (10% and 18%, respectively), although these differences did not reach statistical significance (Figs. 1C–1D). Of the eight DHL cases with decreased CD20 expression, seven showed concomitant decrease in CD19 expression (7/22 cases overall with dim CD20 and CD19 expression; 32%). The single BCLU with decreased CD20 expression also showed decreased CD19 expression (1/6, 17%) (Fig. 1B, left). There were no cases of BL or DLBCL with concomitant decrease in expression of CD19 and CD20. There were no significant differences in the frequency of aberrant expression of CD45 or surface light chain between the various groups.
Sensitivity and specificity were calculated for various flow cytometric parameters in terms of their ability to distinguish the high-grade, unclassifiable entities of DHL and BCLU from the well-defined entities of BL and DLBCL. Our findings, summarized in Table 2, show low to moderate sensitivity for aberrant CD19 and CD20 expression in various combinations, but relatively high specificity for dim CD19 or CD20 expression individually. Concomitant dim CD19 and CD20 expression was 100% specific for the diagnosis of DHL and/or BCLU, albeit with limited sensitivity ranging from 17 to 32%.
Table 2. Sensitivity and Specificity of Various Flow Cytometric Parameters for the Diagnosis of Unclassifiable High-Grade B-cell Lymphoma
BCLU, B-cell lymphoma, unclassifiable, with features intermediate between diffuse large B-cell lymphoma and Burkitt lymphoma; DHL, double-hit lymphoma.
Specificity calculations include control cases of Burkitt lymphoma and diffuse large B-cell lymphoma.
Both high-grade entities were combined because of their clinical and pathologic similarities, and because of the relatively small number of cases in each group.
CD20 aberrant (dim or bright)
Concomitant CD19 and CD20 dim
Double-hit lymphomas are defined by the presence of a MYC breakpoint at 8q24 in combination with another recurrent chromosomal abnormality, typically a t(14;18) involving BCL2; therefore, their diagnosis relies on cytogenetic analysis by conventional karyotype or FISH. Early and accurate identification of DHL is desirable due to its poor prognosis and resistance to therapies commonly used in BL and DLBCL (1). Recent reports suggest that DHL exhibits a common immunophenotype by flow cytometry, which could potentially serve as an initial aid in the diagnosis of DHL given that this test can be performed rapidly, within hours of acquisition of a diagnostic tissue sample (6, 7). In this study, we attempted to confirm these earlier findings in a larger series of patients with DHL and also characterized related cases of BCLU lacking dual rearrangements by flow cytometry. Our study was qualitative in nature and did not suggest that a more comprehensive quantitative study of antigen expression be undertaken. Although we identified certain immunophenotypic aberrancies that were more common in DHL and BCLU as compared to control cases of DLBCL and BL, these findings were seen with relatively low frequency in both the DHL and BCLU groups, limiting their potential utility in the early identification of these lymphomas.
In our study, dim CD19 expression was found to be the most discriminatory parameter when used alone, seen in about half of DHL and BCLU cases in comparison to approximately one-tenth of BL and DLBCL cases (Table 1). In contrast, dim CD20 expression was found to be a less sensitive marker for the diagnosis of DHL, present in only about one-third of cases. While dim CD20 expression was seen with highest frequency in the DHL group, differences in the frequency of such expression between DHL and BL or DLBCL did not reach statistical significance. In fact, almost as many cases of DHL expressed bright CD20 as expressed dim CD20. Grouping of dim or bright CD20 expression into a single group of CD20-aberrant cases yielded the highest sensitivity for the diagnosis of DHL and BCLU, ranging from 50 to 68% (Table 2). Concomitant decrease in CD19 and CD20 expression was a feature that was highly specific to DHL and BCLU, but present in only a minority of cases.
Our finding of dim CD20 expression by flow cytometry in only a minority of DHL cases (8/22, 36%) contrasts with recent reports of dim CD20 expression being a common immunophenotype in DHL: Wu et al. found decreased CD20 expression in 8 of 10 DHL cases (80%) and Harrington et al. (7) similarly showed 6 of 9 DHL cases to have dim CD20 expression (67%) (6, 7). Despite these differences, we were able to confirm that concomitant decrease in CD19 and CD20 expression appears to be highly specific to DHL. Harrington et al. (7) reported decreased CD19 expression in 6 of 9 DHL cases (67%), 5 of which were CD20dim+ (5/9, 56%). Although Wu et al. (6) found decreased CD19 in only 3 of 10 cases of DHL (30%), all of these cases also had dim CD20 expression.
The differences in our findings from those of prior authors may be due in part to the use of different reagents, instruments, software platforms, and controls. For instance, our study compared the expression of various antigens in lymphoma cells to resting co-mingling polytypic B cells, as did Harrington et al. (7), while Wu et al. (6) used germinal center B cells as a control population. A difference between our study and that of Harrington et al. (7) is that their control DLBCL cases were CD10+. According to the Hans classifier, these cases would be categorized as germinal center B-cell-like (GCB) DLBCL, and it would be expected to carry a better prognosis compared to non-GCB DLBCL (11). In our selection of DLBCL control cases, we did not consider CD10 expression but relied on the availability of cytogenetic analysis to identify cases lacking concurrent MYC and BCL2 or BCL6 rearrangements to rule out DHL with certainty. In our laboratory, cases are sent for cytogenetic analysis only after initial morphologic evaluation of frozen or permanent sections. Therefore, it is possible that there may be a bias among our DLBCL control cases toward more clinically aggressive lymphomas in which cytogenetic analysis was thought to be warranted for the diagnosis.
Although the aforementioned differences in methodology and selection of controls may indeed be responsible for differences between our results and those of prior authors, our data argue that the finding of dim CD20 or CD19 expression as a commonly observed immunophenotype specific to DHL is not generalizable across laboratories. In our hands, aberrant expression of either marker was neither entirely sensitive nor specific to DHL or BCLU, and therefore would not serve as a good predictor for their diagnosis. Our findings do, however, validate the observation that concomitant decrease in CD19 and CD20 expression by flow cytometry appears to be highly specific for DHL or BCLU, despite being present in relatively few cases (32% of DHL and 17% of BCLU cases vs. 0% of BL and DLBCL controls in our series). This latter finding can be added to the growing armamentarium of diagnostic tools available to identify potential cases of DHL and to select cases for confirmatory cytogenetic testing. For example, nuclear staining for MYC protein by immunohistochemistry has been recently studied as a predictor for the presence of a MYC rearrangement in DLBCL with varying levels of sensitivity and specificity (12–14). While the possibility of using a single immunohistochemical stain to select cases for cytogenetic analysis is attractive, the assessment of CD19 and CD20 expression by flow cytometry may be a more accessible tool that is simpler to interpret at the present time.
In conclusion, although a subset of DHL expresses aberrant levels of CD19 and/or CD20 by flow cytometry, these immunophenotypic characteristics are neither broadly expressed by nor unique to this neoplasm and are therefore of limited utility in identifying cases that require further analysis by cytogenetics. An exception is when CD19 and CD20 are both dimly expressed which appears to be 100% specific for the diagnosis of DHL, albeit with limited sensitivity. Until more sensitive and specific pathologic parameters can be identified and fully validated, the decision to perform cytogenetic analysis to confirm a diagnosis of DHL continues to rely on a combination of clinical, morphologic, and immunophenotypic features suggestive of high-grade, aggressive disease.
We thank Abigail Kelliher and Kathleen Leahy for kindly helping to retrieve flow cytometry reports.