Rapid and accurate differential diagnosis between Burkitt lymphoma (BL) and CD10+ diffuse large B-cell lymphoma (DLBCL) is imperative because their treatment differs. Recent studies have characterized several antigens differentially expressed in these 2 types of lymphoma. Our goal was to determine whether use of these markers would aid in the differential diagnosis of BL vs CD10+ DLBCL by flow cytometric immunophenotyping (FCI). Twenty-three cases of CD10+ B-cell lymphomas with available cryopreserved samples were identified (13 BL and 10 CD10+ DLBCL). Multiparameter FCI was performed using the following antibodies: CD18, CD20, CD43, CD44, and CD54 and isotype controls. Expression of CD44 and CD54 was detected at a significantly lower level in BL compared with CD10+ DLBCL (P = .001 and P = .01, respectively). There was not a significant difference in expression of CD18 and CD43. Our data show that expression of CD44 and CD54 differs significantly between BL and CD10+ DLBCL.
Burkitt lymphoma (BL) and diffuse large B-cell lymphoma (DLBCL) are distinct subtypes of mature B-cell lymphomas.1,2 The former is believed to arise from germinal center B cells and consists of monomorphic intermediate-sized B lymphocytes with a very high proliferation index and chromosomal translocations involving the MYC oncogene.2 The latter comprises a more heterogeneous group of lymphomas that typically exhibit larger cells but may overlap morphologically with BL.2,3 Gene expression profiling studies have subdivided DLBCL into 2 molecular subgroups: germinal center B cell–like and activated B cell–like.3,4 Despite generally exhibiting a lower proliferation rate and less cellular monomorphism than BL, some DLBCLs may have a germinal center phenotype with very high proliferation index and a “starry-sky” appearance, mimicking BL.5–11
BL is common in children but also occurs in adults, in whom distinction from DLBCL may pose a problem. The development of brief, very intensive chemotherapy regimens has led to a very high cure rate in children with BL. The use of these regimens in adults, often in combination with the antibody rituximab (Rituxan), has also made the cure of a majority of adults possible.12 BL in adults cannot be treated effectively with the common regimens used for DLBCL such as CHOP-R (cyclophosphamide, doxorubicin, vincristine [Oncovin], prednisone, and rituximab). Prompt diagnosis and initiation of appropriate therapy with attention to the possibility of tumor lysis syndrome are necessary for optimal results.
Differentiation between BL and DLBCL is critical for patient management and requires a combination of morphologic evaluation, clinical history, immunophenotyping, and cytogenetic and/or fluorescence in situ hybridization (FISH) studies. Multiparameter flow cytometric immunophenotyping (FCI) is a well-established tool for phenotyping of hematolymphoid neoplasms. Given the phenotypic overlap between BL and germinal center B cell–like DLBCL, however, antibody panels routinely used in clinical laboratories cannot definitively differentiate between these 2 entities.
Previous studies have demonstrated down-regulated or absent expression of CD44 in BL relative to DLBCL by immunohistochemical analysis and gene expression profiling studies.7–9,13–17 Differences in CD43 expression have also been observed by immunohistochemical analysis, with BL showing increased expression compared with DLBCL.8,10,18 Other adhesion molecules have also been shown to be differentially expressed in the 2 entities, with BL cases exhibiting lower expression of CD18 and CD54.9,19 Several other studies have documented the generally low expression of adhesion molecules in BL,20–23 and some have suggested that this may provide a means of escape from immunosurveillance.9–11,13–21 To date, however, only Attarbaschi and colleagues24 have reported data demonstrating lack of CD44 expression by FCI in BL; their study was restricted to pediatric BL.
In this study, we designed a novel antibody panel with CD18, CD43, CD44, and CD54 to differentiate BL from CD10+ DLBCL by FCI. Our results show that some adhesion molecules are differentially expressed between BL and CD10+ DLBCL, and the new antibody panel may prove to be valuable for rapid initial assessment of this highly proliferative tumor and for evaluation of minimal residual disease.
Materials and Methods
CD10+ B-cell leukemias and lymphomas were identified by searching Emory University Department of Pathology (Atlanta, GA) electronic archives. Samples were identified from between 2002 and 2007 that had available excess patient sample cryopreserved in the Tissue Bank of the Winship Cancer Institute, Emory University. The original diagnoses were made using standard World Health Organization criteria in use at that time.1 Slides, pathology reports, and clinical histories were reviewed by 3 of us (S.D.S., S.L., and K.P.M.) in conjunction with ancillary cytogenetic and FISH data, when available, to confirm the diagnoses. Approval from the Emory University Institutional Review Board was obtained.
Flow Cytometric Immunophenotyping
Results from multiparameter 4-color FCI performed as part of the routine diagnostic workup were reviewed. FCI had been performed as described previously.25 All were run on a FACSCalibur or FACSCanto I (Becton Dickinson, San Jose, CA) using our institution’s standard panel that includes the following: CD2, CD3, CD4, CD5, CD7, CD8, CD10, CD11b, CD11c, CD13, CD14, CD15, CD16, CD19, CD20, CD22, CD23, CD25, CD33, CD34, CD36, CD38, CD45, CD56, CD103, CD117, FMC7, HLA-DR, κ, and λ and isotype controls (BD Biosciences, San Jose, CA).
For this study, cryopreserved samples were thawed and viability was assessed by trypan blue exclusion. FCI was performed as described previously25 using the following antibody-fluorochrome combinations: CD18-allophycocyanin (APC), CD20-peridinin chlorophyll protein (PerCP), CD54-phycoerythrin (PE), CD43-fluorescein isothiocyanate (FITC) (tube 1); CD44-PE, CD43-FITC, CD20-PerCP (tube 2); and IgG 1+2-FITC, IgG 1+2-PE, IgG 1-APC, and CD20-PerCP (tube 3, isotype controls) (BD Biosciences) Table 1. The CD44 antibody (clone 515) binds the 85-kDa standard CD44 molecule. Data were analyzed using CellQuest software, version 3.3 (Becton Dickinson). Appropriate staining of these antigens was assessed by analysis of 5 peripheral blood, 5 bone marrow aspirate, and 5 lymph node samples, all of which lacked phenotypically abnormal populations.
In the test samples, CD20 vs right-angle light scatter gating was used to highlight the lymphoma population. In the single BL case lacking CD20 expression, forward vs right-angle light scatter gating was used. The expression of antigens examined was assessed relative to isotype controls. The difference between mean fluorescence intensity (MFI; channel values using a linear scale) and isotype controls (ΔMFI) was calculated for each antibody in the experimental panels and classified as follows: low, 25 to 150; moderate, 151 to 300; and high, greater than 300. In some cases, isotype controls demonstrated higher nonspecific staining than the specific antibodies, resulting in negative ΔMFI.
Fluorescence In Situ Hybridization
As part of the original diagnostic workup, FISH studies for MYC (6 cases) and BCL2 (4 cases) were performed using LSI IGH/MYC/CEP 8 tricolor dual-fusion probes; LSI IGH/BCL2 dual-color, dual-fusion probes; and LSI MYC dual-color, break-apart probes (Vysis, Abbott Molecular, Abbott Park, IL). Fixed cell preparations from 24-hour cultures were analyzed in most cases. Additional interphase FISH analysis was performed using paraffin-embedded tissue sectioned into 2-μm slices in 7 cases. We scored 200 for each probe analyzed from cultured cells, and 40 interphase cells were scored for paraffin-embedded tissue. The normal cutoff level for both fusion probes was 5% or greater abnormal signals, and the cutoff for the break-apart signal pattern was 4% or greater abnormal signals.
The ΔMFI for the novel antibodies was compared between BL and CD10+ DLBCL by using 2-sample t tests. The resulting P values indicate whether intensity of antigen expression differed significantly between the 2 disease categories. The positive predictive value (PPV) and negative predictive value (NPV) were determined for selected antibodies.
Searching the pathology database yielded 14 cases of BL and 12 cases of CD10+ DLBCL with excess patient sample cryopreserved in the Tissue Bank, Winship Cancer Institute, Emory University. Low cellularity and viability precluded further analysis in 1 case of DLBCL. One BL case was excluded because slides and cytogenetic/FISH results were not available for review. Slides were available and slide review was performed in 23 cases: 12 BL and 11 CD10+ DLBCL. Diagnosis was confirmed in 22 cases, with 1 case of DLBCL reclassified as BL based on confirmation of the morphologic and immunophenotypic impression of BL by positive FISH studies for the MYC translocation. Representative cases of BL and DLBCL are shown in Image 1. Molecular genetic characteristics of cases are shown in Table 2. Two BL cases did not have confirmed MYC translocations. One case occurred in the tonsils and kidneys of a 4-year-old girl. Morphologic examination of a fine-needle aspiration specimen from the kidney showed classic BL morphologic features of monomorphic, intermediate-sized, round cells with basophilic vacuolated cytoplasm. The second case occurred in the breasts of a 28-year-old pregnant woman. A needle core biopsy showed monomorphic, intermediate-sized cells with an MIB-1 proliferation index of more than 99%. Both patients were treated with Burkitt regimens and showed no evidence of disease at the 2-year follow-up.
Demographic and clinical data are summarized in Table 2. The mean age at diagnosis was similar for BL (47.7 years) and CD10+ DLBCL (51.5 years). The sole sample from a pediatric patient had a diagnosis of BL. The male/female ratio for the 2 entities was similar: 8:5 in BL and 3:2 in CD10+ DLBCL, with males more commonly affected in both groups. Of 13 patients with BL, 6 (46%) were HIV+, whereas 2 (20%) of 10 patients with CD10+ DLBCL were HIV+. Seronegativity for HIV was documented in 5 cases of BL and 3 cases of CD10+ DLBCL. For 7 patients, testing for HIV was not performed. Two-year follow-up data were available for 19 cases (10 BL and 9 CD10+ DLBCL). Outcome in BL was not significantly different from that for CD10+ DLBCL, with 5 (42%) of 12 patients with BL alive without evidence of disease compared with 5 (56%) of 9 patients with CD10+ DLBCL.
The majority of cases of BL expressed CD10, CD19, CD20, CD22, CD38, FMC7, HLA-DR, and CD45 as assessed by routine FCI using the standard panel used at our institution (see the “Materials and Methods” section). One exceptional BL case failed to express CD20 and FMC7; no history of chemotherapy, including rituximab treatment, was noted. CD10+ DLBCL had a similar immunophenotype, with expression of CD10, CD19, CD20, CD22, CD38, FMC7, HLA-DR, and CD45.
Based on quantitative assessment of the ΔMFI relative to isotype controls, expression of CD44 and CD54 was detected at a significantly lower level in BL compared with CD10+ DLBCL (P = .001 and P = .01, respectively). Representative scatter plots of CD44 and CD54 expression are shown in Image 2. The expression of CD44 and CD54 is summarized in Figure 1. BL cases generally failed to express CD44 (n = 11), although 1 case with low and 1 case with high levels of expression were observed. In both of the BL cases with CD44 expression, the patients were HIV+. One case exhibited a t(2;8) translocation, and the other showed the classic t(8;14). CD10+ DLBCL cases showed a range of intensity of expression of CD44 (low in 2 cases, moderate in 4, high in 3, and absent in 1). The DLBCL case lacking CD44 expression showed evidence of IGH/BCL2 translocation by polymerase chain reaction and initially manifested as a subcutaneous mass on the patient’s arm. Assessment based on positive vs negative expression of CD44 gives a PPV of 74% for CD10+ DLBCL and an NPV of 92% for BL.
BL cases largely showed low or absent CD54 expression (n = 4 and n = 3, respectively), with 6 cases exhibiting moderate expression. Most CD10+ DLBCL cases showed high or moderate CD54 expression (n = 3 and n = 4, respectively); however, 2 cases showed low expression and 1 case failed to express CD54. The latter case had a documented history of follicular lymphoma, exhibited a t(14;18), and also failed to express CD18. Assessment based on positive vs negative expression of CD44 gives an NPV of 75% for BL. Combining CD44 and CD54 gives the best PPV and NPV for CD10+ DLBCL and BL. If both antibodies are expressed, the PPV for CD10+ DLBCL is 80%. When neither antigen is expressed, it is most consistent with a diagnosis of BL (NPV of 100% for CD10+ DLBCL).
CD18 and CD43 were expressed at similar levels in BL and CD10+ DLBCL (P = .38 and P = .40, respectively). BL cells predominantly showed absent (n = 6) to low (n = 6) expression of CD18; 1 case showed moderate expression. CD10+ DLBCL cases showed heterogeneity in CD18 expression, with absent (n = 3), low (n = 4), moderate (n = 2), and high (n = 1) levels observed. BL cases showed expression of CD43 at low (n = 12) to moderate (n = 1) intensity. CD10+ DLBCL cases also generally showed low (n = 9) CD43 expression, with 1 case failing to express CD43.
Expression of CD18 and CD54 was low to moderate in normal control CD20+ lymphocytes, which appropriately failed to express CD43 in all but 1 case; levels of CD44 expression were high in this population.
We evaluated a novel immunophenotypic panel including the antigens CD18, CD43, CD44, and CD54 with the goal of distinguishing between BL and CD10+ DLBCL by FCI. Our studies demonstrated that there is a significant difference in expression of CD44 and CD54 between BL and CD10+ DLBCL, with significantly higher expression in CD10+ DLBCL. CD18 and CD43 are expressed at similar levels in both entities. This differential expression can be exploited for purposes of rapidly discriminating between the 2 entities, particularly in the case of CD44, which shows a PPV of 74% for CD10+ DLBCL and an NPV of 92% for BL. Analysis of CD54 expression could also prove useful, with an NPV of 75% for BL. Combining CD44 and CD54 gives the best PPV and NPV: 80% and 100%, respectively. The addition of CD18 and CD43, on the other hand, would most likely prove unhelpful for distinguishing BL and CD10+ DLBCL.
CD18, CD43, CD44, and CD54 are antigens that have been reported to be overexpressed or underexpressed in BL relative to other non-Hodgkin lymphoma subtypes, particularly DLBCL.7–10,15–24 These antigens are adhesion molecules with important functions in normal lymphocytes and postulated roles in the pathogenesis and spread of lymphomas and other malignant neoplasms.15,21,26
We found that the majority of BL cases lacked expression of CD44 as judged by FCI. All but 2 cases failed to express CD44, and 1 of the positive cases showed very low levels of expression. Our data are consistent with prior observations.8,9,24 CD44, or lymphocyte homing receptor, functions in lymphocyte activation and in cell-cell and cell-matrix adhesion. Prior studies have shown evidence for its role in lymphoma dissemination, and high serum levels and tissue expression of certain isoforms have been linked to worse prognosis in lymphoma.27,28 Of note, there was no obvious morphologic, clinical, or genetic distinguishing feature of the BL case that strongly expressed CD44. The robust nature of CD44 expression in CD10+ DLBCL, together with its relatively reliable absence or deficiency in BL, makes CD44 expression an excellent candidate for use in rapid immunophenotypic discrimination of BL and DLBCL.
We also demonstrated a statistically significant difference in expression of CD54, with decreased expression in BL. However, the relatively small difference between the generally low CD54 expression in BL and the generally moderate to high expression seen in DLBCL may limit the practical application of this finding because quantitative assessment of fluorescence intensity is not routinely performed in clinical laboratories. CD54 (intercellular adhesion molecule [ICAM]-1), has a broad array of functions, and interacts with lymphocyte function–associated antigen 1 to mediate cell-cell and cell-endothelium adhesion. Of interest, CD44 and CD54 expression are NFκB dependent.29 Recent gene expression profiling studies have shown that NFκB target genes are down-regulated in BL.13
Our data show that CD18 expression is generally absent to low in BL, which agrees with the results of some previous studies.19,20,22 However, the heterogeneous expression of CD18 observed in CD10+ DLBCL limits the usefulness of this finding for the purpose of discriminating BL and CD10+ DLBCL. We found that CD43 expression was generally low in both BL and CD10+ DLBCL, differing little from the isotype control. Only 1 case exhibiting moderate expression was identified. This differed from the findings of prior studies, which have shown that in immunohistochemical analysis, CD43 is expressed in most cases of BL.10,18 The reason for this discordance is uncertain, but the discrepancy may indicate that CD43 expression is more accurately assessed by immunohistochemical methods than by FCI. Of note, the CD43 antibodies used in major immunohistochemical studies10,18 targeted the 115-kDa glyco-form of CD43 (Ly-48, leukosialin), whereas the anti-CD43 conjugate used in this study reacts with the major 95-135 kDa sialoglycoprotein.
Many of the patients in this study had a history of HIV/AIDS, which is a widely recognized risk factor for these lymphoma subtypes. Applicable data on the incidence of HIV infection in patients with BL or DLBCL are lacking; however, at our medical center, this figure may be higher than would be expected elsewhere in the United States owing to a high regional HIV/AIDS prevalence of approximately 0.7%.30,31 The male predominance, age at diagnosis, and propensity for extranodal manifestations of both BL and DLBCL are essentially in agreement with findings of prior studies.
Patient survival in our cohort did not differ significantly between BL and DLBCL. The reasons for the relatively low survival rates seen are uncertain but may be secondary to the relatively small sample. In addition, the patients were treated at a tertiary care referral center, and, thus, many may have been referred because of significant preexisting medical conditions or failure of prior treatment or may have sought care at advanced stages of disease. Nearly half of the BL cases were HIV-related, which presents additional treatment challenges. At least 1 of the CD10+ DLBCL cases with a fatal outcome represented progression from a known history of follicular lymphoma. The other patient with DLBCL who died of his disease was HIV+.
Our data demonstrate differential expression of CD44 and CD54 between BL and DLBCL, as measured by FCI, and we suggest that assessing expression of these 2 antigens, in addition to routine FCI, may be useful for distinguishing the 2 entities. Although gene expression profiling has been shown to clearly define these non-Hodgkin lymphoma subtypes, this technique is impractical for use in routine settings.13,14 Other groups have set forth algorithms for differentiating BL and DLBCL based on immunohistochemical panels and results of karyotyping or FISH studies.6–8,11 The advantages of FCI over immunohistochemical studies are significant; most important, the turnaround time can be as short as 2 hours, compared with 1 or 2 days for immunohistochemical analysis. In addition, objective quantitation of antigen expression is readily performed, without the need for additional specialized equipment. The primary disadvantage is the requirement for fresh tissue or fluid samples for analysis.
Phenotypic assessment of adhesion molecule expression by flow cytometry may be useful to establish a provisional diagnosis on which to base initial treatment of BL and CD10+ DLBCL. Prospective studies are needed to further assess the usefulness of CD44 and CD54 as FCI markers for distinguishing BL from DLBCL.