|
|
Background The distinction between Burkitt's lymphoma and diffuse large-B-cell lymphoma is unclear. We used transcriptional and genomic profiling to define Burkitt's lymphoma more precisely and to distinguish subgroups in other types of mature aggressive B-cell lymphomas.
Methods We performed gene-expression profiling using Affymetrix U133A GeneChips with RNA from 220 mature aggressive B-cell lymphomas, including a core group of 8 Burkitt's lymphomas that met all World Health Organization (WHO) criteria. A molecular signature for Burkitt's lymphoma was generated, and chromosomal abnormalities were detected with interphase fluorescence in situ hybridization and array-based comparative genomic hybridization.
Results We used the molecular signature for Burkitt's lymphoma to identify 44 cases: 11 had the morphologic features of diffuse large-B-cell lymphomas, 4 were unclassifiable mature aggressive B-cell lymphomas, and 29 had a classic or atypical Burkitt's morphologic appearance. Also, five did not have a detectable IG-myc Burkitt's translocation, whereas the others contained an IG-myc fusion, mostly in simple karyotypes. Of the 176 lymphomas without the molecular signature for Burkitt's lymphoma, 155 were diffuse large-B-cell lymphomas. Of these 155 cases, 21 percent had a chromosomal breakpoint at the myc locus associated with complex chromosomal changes and an unfavorable clinical course.
Conclusions Our molecular definition of Burkitt's lymphoma clarifies and extends the spectrum of the WHO criteria for Burkitt's lymphoma. In mature aggressive B-cell lymphomas without a gene signature for Burkitt's lymphoma, chromosomal breakpoints at the myc locus were associated with an adverse clinical outcome.
The distinction between Burkitt's lymphoma and diffuse large-B-cell lymphoma is not reliably reproducible with the use of the current criteria of morphology, immunophenotype, and genetic abnormalities.6,7 The Burkitt's translocation or its variants, which juxtapose the locus of the myc oncogene and one of the three immunoglobulin (IG) loci, are present in almost all Burkitt's lymphomas.8,9 Nevertheless, myc translocations are not specific for Burkitt's lymphoma since they also occur in other lymphomas, including diffuse large-B-cell lymphoma. In the latter, chromosomal breakpoints at the myc locus are recurrently associated with non-IG partner loci and complex chromosomal alterations.10,11,12,13,14,15,16,17
The imprecise distinction between Burkitt's lymphoma and diffuse large-B-cell lymphoma on diagnosis may lead to the inadequate treatment of some patients with a mature aggressive B-cell lymphoma. Studies involving gene-expression profiling indicate that diffuse large-B-cell lymphomas comprise two or more main biologic subgroups with different clinical behaviors.18,19,20,21 Until now, however, to our knowledge there has been no signature of gene expression that distinguishes Burkitt's lymphoma from diffuse large-B-cell lymphoma.
Our aim was to establish a molecular definition of Burkitt's lymphoma and to search for other clinically relevant subgroups of diffuse large-B-cell lymphoma. For this purpose, we performed gene-expression and genomic profiling of 220 mature aggressive B-cell lymphomas that had been diagnosed by a panel of expert hematopathologists.
Methods
This study was conducted from July 2003 through November 2005. It was approved by the local ethics commission (Charité University Hospital, Berlin). The data discussed are available from the Gene Expression Omnibus of the National Center for Biotechnology Information (www.ncbi.nlm.nih.gov/geo/) through GEO accession number GSE4475 [NCBI GEO] .
Gene Expression and Genetic Analyses
RNA and DNA were extracted from frozen sections (Qiagen). Affymetrix U133A GeneChip hybridization of all 220 specimens was performed in accordance with the manufacturer's recommendations, involving 5 µg of total RNA. We also performed array-based comparative genomic hybridization in 185 cases, applying a BAC/PAC array containing 2799 DNA fragments.22,23 The number of imbalances per case was determined as an indicator of genetic complexity (explained in detail in the Supplementary Appendix, available with the full text of this article at www.nejm.org).
Interphase fluorescence in situ hybridization was performed on frozen or paraffin-embedded tissues from 217 cases with the use of probes for IGH, IGK, IGL, myc, BCL6 and BCL2 loci24,25,26 (see the Supplementary Appendix). Tumor-biopsy specimens in which myc was fused to IGH, IGK, or IGL were referred to as "IG-myc"; lymphomas with myc breakpoints without fusion of myc to an IG locus were called "non–IG-myc."
Analysis of Microarray Data
Probe intensities were normalized according to a variance-stabilization method.27 Gene-expression levels were estimated by fitting an additive model according to Irizarry et al.28 Germinal-center and activated B-cell–like lymphomas were diagnosed according to the method of Wright et al.29 (Further details and algorithms underlying the core-group extension are given in the Supplementary Appendix.) The raw gene-expression data are available at www.ncbi.nlm.nih.gov/geo.
Statistical Analysis
The Mann–Whitney U test, chi-square test, Fisher's exact test, and log-rank test were used to test for differences among groups. Survival was calculated from the day of diagnosis until death or until the end of follow-up. The Cox proportional-hazards model was used to analyze prognostic factors. Since not all variables of the international prognostic index were available for all patients, the age at diagnosis and Ann Arbor stage were used to adjust for known prognostic factors.30
Results
Patients and Diagnoses
Biopsy specimens of 220 mature aggressive B-cell lymphomas (i.e., classic Burkitt's lymphomas, atypical Burkitt's lymphomas, and diffuse large-B-cell lymphomas), in which at least 70 percent of all cells were tumor cells, were included in this retrospective study. All specimens were reviewed by a panel of expert hematopathologists using the criteria of the World Health Organization (WHO) (provided in the Supplementary Appendix).
Clinical data were available for 146 patients (median year of collection, 1994) who had received a variety of primary treatments. The median follow-up was 60 months (range, 0 to 209). Characteristics of the tumors and patients are given in Table 1 and Table 2, respectively, and in Tables S1, S2, and S3 of the Supplementary Appendix.
|
|
To derive a molecular signature of Burkitt's lymphoma, we devised a computational algorithm called "core-group extension." Given a predefined core group of expression profiles, the algorithm identifies additional cases that have a similar pattern of gene expression and provides a gene-expression signature for the extended group. The method is described in detail in the Supplementary Appendix.
The core group consisted of eight cases that satisfied the WHO criteria for Burkitt's lymphoma (a consensus histologic classification of classic or atypical Burkitt's lymphoma [Figure 1A], CD20+, BCL6+, CD10+, BCL2–, CD5–, Ki-67 score 
95 percent, IG-myc+). We applied core-group extension to a training set of 105 lymphomas and identified 58 genes that constituted the molecular Burkitt's lymphoma (mBL) signature (see the Supplementary Appendix). Each case was assigned an mBL-signature index score between 0 and 1, with a higher score reflecting a greater similarity of gene expression in the sample to that in the core group. Cases with an index score greater than 0.95 were classified as mBL, and those with an index score of less than 0.05 were designated non–molecular Burkitt's lymphoma (non-mBL). The remaining cases were considered intermediate (Figure 2). By rerunning the core-group extension algorithm 1000 times with random perturbations of the core-group data (bootstrapping), we evaluated the stability of the mBL signature (Figure 2). Although the mBL signature was stable at the extremes, the index scores between mBL cases and non-mBL cases were continuous (Figure 2). Results obtained in an independent test set of 107 cases were consistent with the training set with regard to the histologic and genetic characteristics of the mBL and non-mBL cases and the size and stability during bootstrap validation (Figure 2).
|
|
We identified 36 lymphomas with an mBL-signature index score of greater than 0.95 in addition to the 8 core Burkitt's lymphomas, for a total of 44 mBL cases. The 8 core cases were similar to the additional 36 cases with regard to age distribution and genetic features (Table 1, and Table S1 in the Supplementary Appendix), as well as clinical course. Of the additional 36 mBL cases, 21 were categorized as atypical Burkitt's lymphomas because of their Burkitt-like morphology or their deviant immunophenotype. It is important to note that 11 of these 36 cases had the distinctive morphologic appearance of diffuse large-B-cell lymphoma (Figure 1B). The remaining four mBL cases had the morphologic appearance of mature aggressive B-cell lymphoma but could not be further classified histologically. With regard to immunophenotype, CD10 and BCL6 were consistently expressed in the 42 and 39 mBL cases, respectively, that could be evaluated. BCL2 was detected at a low level in seven of these mBL cases and at a high level in two.
Features of Mature Aggressive B-Cell Lymphomas without the mBL Signature
Of all 220 lymphomas, 176 had an mBL-signature index score of less than 0.95. Of these 176 cases, 128 had an mBL-signature index score of less than 0.05 and were thus assigned to the non-mBL group. The remaining 48 cases had an mBL-signature index score between 0.05 and 0.95 and thus could not be assigned unambiguously to the mBL or non-mBL group. These cases were assigned to the intermediate group, representing the transition zone between the mBL and non-mBL groups.
With few exceptions (12 cases), the histologic diagnosis in the non-mBL cases was diffuse large-B-cell lymphoma (Table 1). The histologic diagnosis of 39 (81 percent) of the intermediate cases was also diffuse large-B-cell lymphoma. Non-mBL and intermediate cases showed strong concordance regarding age distribution, immunophenotype, growth fraction (Ki-67 score), and chromosomal complexity (Table 1).
Genetic Aberrations and Gene Expression
We were able to evaluate 43 mBL cases for the presence of myc translocations by using fluorescence in situ hybridization. All but five cases (88 percent) carried an IG-myc fusion and one of these five had both non–IG-myc and IGH-BCL2 fusions. In the 38 mBL cases with IG-myc fusion, IGH-BCL2 fusion and BCL6 breakpoints were absent. The average chromosomal complexity score was low in the 38 mBL cases with IG-myc fusion but was high in the 5 mBL cases without IG-myc fusion (median complexity score, 2 vs. 9; P<0.001).
The frequency of myc breakpoints (regardless of translocation partner) was significantly lower in the intermediate and non-mBL groups (present in 35 of the 171 cases that could be evaluated [20 percent]) than in the mBL group (present in 39 of 43 cases [91 percent], P<0.001) (Table 1). It is remarkable that myc breakpoints were common in the intermediate group (26 of 48 cases [54 percent]), whereas they were uncommon in the non-mBL group (9 of 123 cases [7 percent]). Non-IG partners were frequently involved in myc translocation in both the intermediate group (10 of 26 cases [38 percent]) and the non-mBL group (4 of 9 cases [44 percent]). Among the 35 myc-positive intermediate and non-mBL cases, 16 (46 percent) had a concurrent IGH-BCL2 fusion, BCL6 breakpoint, or both. The chromosomal complexity score was significantly higher in the intermediate and non-mBL groups than in the mBL group (median complexity score, 8.5 vs. 2; P<0.001) regardless of the presence of myc breakpoints (median complexity score, 7.5) or absence of myc breakpoints (median complexity score, 9).
On the basis of these data, we distinguished three main cytogenetic groups within the mature aggressive B-cell lymphomas. The first is called "myc-simple": lymphomas with IG-myc fusions and a low chromosomal complexity score (<6) that do not have IGH-BCL2 fusions and BCL6 breakpoints. The second is called "myc-complex": all lymphomas with non–IG-myc fusions or all lymphomas with IG-myc fusions that have a high chromosomal complexity score (6), an IGH-BCL2 fusion, or BCL6 breakpoint, or any combination of these. The third is called "myc-negative," comprising myc-negative lymphomas. The mBL group predominantly consisted of myc-simple lymphomas; the non-mBL group predominantly consisted of myc-negative lymphomas. In contrast, the intermediate group contained most of the myc-complex cases but also occasional myc-simple and several myc-negative cases (Table 1).
Correlation of Molecular and Clinical Features
No significant differences were observed regarding the morphologic characteristics, immunophenotype, or gene-expression pattern between the 146 patients with survival data and the 74 patients without or with incomplete clinical information (Table S2 and Figure S1 in the Supplementary Appendix). The clinical data available for the 146 patients (Figure S1 in the Supplementary Appendix) suggest that the tissue specimens were obtained during a period of relapse in less than 10 percent; this percentage was assumed to be representative of those without clinical information.
Patients with lymphomas classified as mBL or myc-simple had a significantly better five-year survival rate than patients with non-mBL or intermediate lymphomas (75 percent vs. 39 percent, P=0.003 [Figure 3]) or with myc-complex or myc-negative lymphomas (70 percent vs. 41 percent, P=0.005 [Figure S3 in the Supplementary Appendix]). However, the results of multivariate regression analysis showed that even if we ignored the different treatments received by these patients, the favorable outcome among patients with mBL or myc-simple lymphomas could largely be explained by the onset of the disease at a young age and the limited stage of the disease (Table 2 and Table 3).
|
|
When we applied the signature for activated B-cell–like or germinal-center B-cell–like large-B-cell lymphomas described by Rosenwald et al.19 to our non-mBL and intermediate cases, we found that lymphomas with a germinal-center B-cell–like signature were associated with a significantly better five-year survival rate than activated B-cell–like lymphomas (51 percent vs. 12 percent, P=0.003) (Figure S4 in the Supplementary Appendix). The hazard ratio for death for lymphomas carrying a germinal-center B-cell–like signature as compared with activated B-cell–like signature (1.79; 95 percent confidence interval, 0.94 to 3.42; P=0.08) was in the same range as that reported by Rosenwald et al.19 (Table 3).
Discussion
The distinction between Burkitt's lymphoma and diffuse large-B-cell lymphoma is clinically important, because these lymphomas are treated with different chemotherapeutic protocols and differ in their outcome.31 Even with the use of current diagnostic criteria, the distinction is not precise; agreement among expert hematopathologists on the pathological diagnosis of classic Burkitt's lymphoma, atypical Burkitt's lymphoma, and diffuse large-B-cell lymphomas is only 53 percent.6,7 Our results provide a molecular definition of Burkitt's lymphoma that reliably and reproducibly distinguishes it from other mature aggressive B-cell lymphomas.
Of the 220 mature aggressive B-cell lymphomas studied, we identified 44 with a consistent pattern of gene expression that was characteristic of mBL. The distinctive mBL signature consisted of 58 genes, including several target genes of the nuclear factor-
B pathway (i.e., BCL2A1, FLIP, CD44, NFKBIA, BCL3, and STAT3) that are known to distinguish activated B-cell–like or germinal-center B-cell–like lymphomas.19,29 It is notable that these genes were expressed at lower levels in mBL cases than in cases of germinal-center B-cell–like diffuse large-B-cell lymphoma.
The mBL signature extends the WHO definition of Burkitt's lymphoma1 to cases with the morphologic characteristics of diffuse large-B-cell lymphoma and expression of BCL2. Yet not all cases with morphologic or immunophenotypical features of Burkitt's lymphoma were classified as mBL. Parallel analyses of genetic features — by means of interphase fluorescence in situ hybridization and array-based comparative genomic hybridization — strongly support the biologic relevance of the mBL signature. Of the 38 mBL cases for which we had genetic data, 29 (76 percent) were classified as myc-simple, irrespective of their morphologic appearance. However, in four mBL cases, no IG-myc fusion was detectable, and one case was found to have a non–IG-myc translocation.
Clinically, all patients with mBL had a favorable prognosis (five-year survival rate, 75 percent). There was no significant difference in survival between patients whose tumors had a morphologic appearance of classic or atypical Burkitt's lymphoma and those with a morphologic appearance of diffuse large-B-cell lymphoma, irrespective of the presence or absence of a myc breakpoint (data not shown). Our results suggest that a molecular diagnosis of Burkitt's lymphoma yields a more precise definition of this entity than do current diagnostic criteria.
Of the 220 mature aggressive B-cell lymphomas, 176 had an mBL-signature index below 0.95 and thus were not classified as mBL. Most of these 176 cases had the morphologic appearance of diffuse large-B-cell lymphoma. In line with a previous report,19 there was a difference in five-year survival between cases with a germinal-center B-cell–like gene signature and an activated B-cell–like gene signature (51 percent vs. 12 percent). More striking, the presence of breakpoints at the myc locus was strongly associated with an unfavorable five-year survival rate, as compared with the absence of such breakpoints (15 percent vs. 44 percent). This association was independent of stage, age, and whether the tumor was classified as activated B-cell–like or germinal-center B-cell–like. In addition, 46 percent of cases with IG-myc and non–IG-myc breakpoints had concurrent BCL2 and BCL6 translocations, suggesting that the myc breakpoint arose during clonal evolution and conferred clinical aggressiveness.32
In the 208 cases that could be evaluated genetically, the pattern of chromosomal aberrations allowed us to define three cytogenetic groups. The myc-simple group comprises lymphomas with an IG-myc fusion and a low number of chromosomal imbalances (complexity score, less than 6). This group included cases in which the Burkitt's translocation is most likely to constitute the primary oncogenic event. The myc-simple cases largely overlapped with the mBL cases and were associated with a favorable clinical outcome. In contrast, myc-complex status was associated with a poor outcome, independently of age and clinical stage (five-year survival rate, 21 percent). The myc-complex cases were common in the intermediate group. We speculate that in these intermediate cases, a primary genetic aberration induced a gene-expression profile that was subsequently shifted toward the mBL profile by a myc translocation.
In summary, we have devised a molecular definition of Burkitt's lymphoma by global gene-expression and genomic profiling. Mature aggressive B-cell lymphomas with both the mBL signature and the cytogenetic myc-simple status have a favorable outcome and can be regarded as biologic Burkitt's lymphoma. Mature aggressive B-cell lymphomas with myc breakpoints but without an mBL signature are clearly distinct from biologic Burkitt's lymphoma and are associated with a poor clinical outcome. The few cases that met only one of the two criteria for biologic Burkitt's lymphoma require further investigation. The distinction between biologic Burkitt's lymphoma and other mature aggressive B-cell lymphomas with myc breakpoints or without them is a prerequisite for planning clinical trials of the treatment of mature aggressive B-cell lymphomas. Such trials should clarify whether Burkitt's lymphomas and aggressive B-cell lymphomas (with myc breakpoints or without them) should be treated differently.
Supported by a grant from the Deutsche Krebshilfe (70-3173-Tr3) and funding from the Schweizerische Arbeitsgemeinschaft für Klinische Krebsforschung. No potential conflict of interest relevant to this article was reported.
This article is dedicated to the memory of Reza M. Parwaresch.
Source Information
Mr. Bentink and Drs. Berger, Klapper, and Wessendorf contributed equally to this article.
Address reprint requests to Dr. Stein at Charité Universitätsmedizin Berlin, Campus Benjamin Franklin, Institute of Pathology, Hindenburgdamm 30, D-12200 Berlin Germany, or at Harald.Stein{at}charite.de.
References
The authors' affiliations are as follows: the Institute of Pathology, Campus Benjamin Franklin, Charité Universitätsmedizin, Berlin (M.H., M.K., D.L., H.S.); the Max Planck Institute for Molecular Genetics, Department of Computational Molecular Biology, Computational Diagnostics Group, Berlin (S.B., R. Spang); the Institute for Medical Informatics, Statistics, and Epidemiology, Universität Leipzig, Leipzig (H.B., D.H., M.L.); the Institute of Hematopathology and Lymph Node Registry (W.K., R.M.P., M.S., H.-H.W.), the Institute of Human Genetics (L.H., J.I.M.-S., R. Siebert), and the Second Medical Department (C.P., H.T.), Universitätsklinikum Schleswig-Holstein, Campus Kiel, Kiel; Cytogenetic and Molecular Diagnostics, Internal Medicine III (S.W., C.S.), and the Institute of Pathology (T.F.E.B., P.M.), Universitätsklinikum Ulm, Ulm; the Institute of Pathology, Universitätsklinikum Schleswig-Holstein, Campus Lübeck, Lübeck (H.-W.B., A.C.F.); the Department of Oncology and Hematology, Universitätsklinikum Hamburg-Eppendorf, Hamburg (J.D.); the Institute of Pathology, Universitätsklinikum Frankfurt, Frankfurt (M.-L.H.); the Institute of Pathology, Universität Würzburg, Würzburg (E.H., H.-K.M.-H., G.O., A.R.); the German Cancer Research Center, Heidelberg (P.L.); the Interdisciplinary Center for Bioinformatics, Leipzig (M.R.); and the Department of Hematology and Oncology, Georg-August Universität, Göttingen (B.S., L.T.) — all in Germany; and the Institute of Pathology, Kantonsspital St. Gallen, St. Gallen, Switzerland (S.B.C.).
The members of the Molecular Mechanisms in Malignant Lymphomas Network Project are as follows: Writing Committee — M. Hummel, M. Loeffler, A. Rosenwald, R. Siebert, H. Stein, L. Trümper; Steering Committee — M. Loeffler, R. Siebert, H. Stein, L. Trümper; Reference Pathology Panel — T.F.E. Barth, H.-W. Bernd, S.B. Cogliatti, A.C. Feller, M.-L. Hansmann, W. Klapper, P. Möller, H.-K. Müller-Hermelink, G. Ott, R.M. Parwaresch, T. Rüdiger, H. Stein, H.-H. Wacker; Gene Expression — E. Berg, M. Hummel, M. Kühn, H. Lammert, D. Lenze, R. Zollinger; Tissue Processing and RNA and DNA Extraction and Analysis — E. Berg, C. Burek, M. Frank, J. Helfrich, A. Juhl, H. Lammert, R. Liebertz, S.W. Popov, C. Pott, E. Sevecke-Wessels, M. Szczepanowski, H. Trautmann; Interphase Cytogenetics — T.F.E. Barth, P. Behrmann, C. Becher, H. Brückner, M. Buck, J. Dierlamm, I. Eichelbrönner, E. Haralambieva, L. Harder, S. Hartmann, K. Hinz, J.-I. Martin-Subero, E.M. Murga Penas, G. Ott, S. Pries, M. Ratjen, A. Rosenwald, D. Schuster, R. Siebert, R. Zühlke-Jenisch; Array-Based Comparative Genomic Hybridization — M. Bentz, Z. Keresmann, P. Lichter, S. Ruf, C. Schwaenen, S. Wessendorf; Bioinformatics and Biostatistical Analyses — H. Berger, S. Bentink, D. Hasenclever, D. Kostka, M. Loeffler, M. Rosolowski, R. Spang; and Clinical Data Collection and Interpretation — H. Berger, S. Höller, M. Loeffler, C. Pott, U. Schönwiese, B. Stürzenhofecker, L. Trümper.
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||
Related Letters:
Genomic Diagnosis of Burkitt's Lymphoma
Lin B. T., Dave S. S., Staudt L. M., Hummel M., Stein H., Siebert R., the Molecular Mechanisms in Malignant Lymphomas Network Project of the Deutsche Krebshilfe
Extract |
Full Text |
PDF
N Engl J Med 2006;
355:1064-1065, Sep 7, 2006.
Correspondence
This article has been cited by other articles:
HOME | SUBSCRIBE | SEARCH | CURRENT ISSUE | PAST ISSUES | COLLECTIONS | PRIVACY | HELP | beta.nejm.org Comments and questions? Please contact us. The New England Journal of Medicine is owned, published, and copyrighted © 2008 Massachusetts Medical Society. All rights reserved. |
Previous
