Fax: (901) 495-3100.
Secondary chromosomal abnormalities predict outcome in pediatric and adult high-stage Burkitt lymphoma
Article first published online: 21 JUL 2006
Copyright © 2006 American Cancer Society
Volume 107, Issue 5, pages 1084–1092, 1 September 2006
How to Cite
Onciu, M., Schlette, E., Zhou, Y., Raimondi, S. C., Giles, F. J., Kantarjian, H. M., Medeiros, L. J., Ribeiro, R. C., Pui, C.-H. and Sandlund, J. T. (2006), Secondary chromosomal abnormalities predict outcome in pediatric and adult high-stage Burkitt lymphoma. Cancer, 107: 1084–1092. doi: 10.1002/cncr.22089
- Issue published online: 21 AUG 2006
- Article first published online: 21 JUL 2006
- Manuscript Accepted: 5 MAY 2006
- Manuscript Revised: 4 MAY 2006
- Manuscript Received: 22 FEB 2006
- National Cancer Institute. Grant Number: CA-21765
- American Lebanese and Syrian Associated Charities (ALSAC)
- Burkitt lymphoma;
- lymphoproliferative disorders;
Karyotypic abnormalities in sporadic Burkitt lymphoma (BL) have been described extensively. However, to the authors' knowledge, very limited studies have focused on the secondary chromosomal abnormalities in pediatric BL as compared with those of adult BL and on their prognostic impact.
A retrospective analysis was performed in all pediatric and adult patients at 2 institutions, with a morphologic diagnosis of BL, pretherapy tumor karyotype available, and t(8;14), t(8;22), or t(2;8) present.
There were 33 children and 37 adults. The majority of the patients (95%) had Stage III/IV disease. There were no statistically significant differences noted in karyotype complexity and the nature of the chromosomal abnormalities between these 2 groups. Abnormalities of chromosomes 13 (13q) and 22 (22q) had a negative impact on prognosis in children. In adults, abnormalities of chromosome 17 appeared to have a negative impact.
The current findings suggest that karyotypic information can be used for refining risk stratification in patients with BL. Cancer 2006. © 2006 American Cancer Society.
Burkitt lymphoma/leukemia (BL) is a high-grade non-Hodgkinlymphoma of mature B-cell phenotype and likely germinal center origin, characterized most commonly (70%–80%) by the presence of the t(8;14)(q24.1;q32).1 This translocation juxtaposes the c-myc oncogene (on 8q24) to the immunoglobulin heavy chain gene locus (IgH) on 14q32, leading to overexpression of c-myc,1 which plays a central role in malignant transformation in BL.2 Fewer cases contain the variant translocations, the t(2;8)(p11;q24) or t(8;22)(q24;q11.2)3, 4 that juxtapose c-myc to the immunoglobulin light chain genes, Ig kappa (on 2p11) and Ig lambda (on 22q11), respectively.1
Approximately 60% to 70% of sporadic BLs in adults have additional chromosomal abnormalities,3–9 more complex than those found in the uniformly Epstein-Barr virus (EBV)-positive endemic BL type, suggesting potentially more diverse mechanisms of malignant transformation and disease progression in the former.10 These chromosomal abnormalities likely contribute to disease biology and response to therapy, and may have an impact on prognosis. Little information is currently available on karyotypic abnormalities or their prognostic importance in pediatric BL.11
With the current chemotherapy regimens, the overall cure rates are approximately 90% in children and 50% to 60% in adults.12–15 However, patients with high-stage disease continue to have a less favorable prognosis, with cure rates of 80% to 85% in children15 and 40% to 50% in adults16 There are no known biological tumor features that reliably predict the behavior of BL. In addition, it is not entirely clear if the difference in outcome between children and adults is a result of a relatively poor tolerance to treatment in the latter or if it is at least partially attributable to differences in tumor biology.
This study describes and compares the chromosomal abnormalities identified in sporadic BL arising in 33 children and 37 adults. We find that karyotype complexity and specific chromosomal abnormalities correlate with outcome in patients with BL treated with current standard intensive chemotherapy regimens, whereas there are few differences in the karyotypes found in pediatric and adult BL.
MATERIALS AND METHODS
The pathology database at St. Jude Children's Research Hospital (SJCRH) was searched for patients with a diagnosis of BL and pretherapy karyotypes available. The database of the Cytogenetics Laboratory at the University of Texas M. D. Anderson Cancer Center (MDACC) was searched for patients with pretherapy karyotypes containing the t(8;14), t(8;22), or t(2;8) and a diagnosis of BL. The medical records of all patients were searched for age, gender, date of diagnosis, disease stage, central nervous system (CNS) involvement, serum lactate dehydrogenase level (LDH), human immunodeficiency virus (HIV) status, treatment protocol, date of complete remission (CR), date of recurrence, adverse events and date of adverse events, date of last follow-up, and date of death (when applicable). The study was approved by the Institutional Review Boards at both institutions.
Cytogenetic analysis was performed on bone marrow and tissue samples using direct preparations and overnight unstimulated cultures, followed by banding with Trypsin-Wright stain, as previously described.17, 18
At both institutions a minimum of 20 metaphases was examined. If <20 metaphases were available, all of those found on the slides were represented in the final karyotype. All the karyotype formulas were described according to ISCN 1995.19
All karyotypes were analyzed using a novel computer program created by 1 of the authors (Y.M.Z.). This program converts karyotypes (ISCN 1995 format) into a tabular form where each chromosome and each type of chromosomal abnormality present in a given case are recorded in a different column. Only abnormalities present in at least 2 metaphases (3 for monosomy) are recorded in this manner.
Chi-square or Fisher exact tests were used to examine associations between any 2 dichotomous variables. Spearman rank correlation coefficients were used to assess relations between 2 variables.
The primary endpoints for statistical survival analysis were event-free survival (EFS) (minimum interval from the date of CR to disease progression, recurrence, a second malignant neoplasm, death, or other treatment failure) and overall survival (OS) (time from study entry to death of any cause). EFS was considered as zero for patients who did not achieve CR. EFS and OS curves and probabilities were calculated by the method of Kaplan and Meier.20 Differences in long-term EFS and OS were tested using the log-rank test.21 The risk factors that were significant in univariate analysis with respect to the EFS and OS (abnormalities of chromosomes 13, 17, and 22 and complex karyotypes) were then included simultaneously in multivariable analysis by using a Cox proportional hazards model.22 The parameters included in the univariate analysis included patient age, gender, LDH level, CNS involvement (in children), karyotype complexity, the type of Burkitt-associated chromosomal translocation, and the presence of abnormalities of chromosomes 1, 2, 6, 7, 9, 13, 17, 22, or X. Karyotype complexity was assessed as the number of chromosomes with numerical or structural abnormalities, including the chromosomes involved in the Burkitt-type translocations. Karyotypes with more than 3 chromosome abnormalities were considered complex.
Pediatric patient group
There were 31 pediatric patients treated at SJCRH and 2 patients treated at MDACC, diagnosed between January 1990 and December 2004. There were 21 boys and 12 girls, ages 2 to 18 years (median, 9 years). Thirty-one of the 33 patients (94%) had Stage III or IV disease, and 2 patients had Stage II (Murphy staging system23). All patients were HIV-negative. Treatment protocols included: the institutional B-cell lymphoma protocol (SJBCII), a modification of the LMB-89 protocol14 (12 patients), the Pediatric Oncology Group (POG) 8617 protocol24 (13 patients), and various other protocols (8 patients). The tumor samples included bone marrow aspirates (20 patients) and other tissues (13 patients). Among the 31 patients with complete outcome information, the 5-year rates of both EFS and OS were 71.0% ± 9.6%, with a median follow-up time of 9.7 years (range, 1.1–13.3 years). In univariate analysis only abnormalities of chromosomes 13, 17, and 22 and the presence of complex karyotypes had an impact on survival in this patient group. On multivariable analysis, only chromosome 13 and 22 abnormalities retained prognostic significance (Table 1).
|Chromosome||Abnormalities present (All patients)||Adults||Children|
|EFS at 2 years (Estimate ± SE)%||P||EFS at 5 years (Estimate ± SE)%||P|
|1||Yes (n = 30)||44.4 ± 11.7||0.5021||66.7 ± 15.7||0.8879|
|No (n = 38)||48.6 ± 13.2||73.7 ± 11.4|
|2||Yes (n = 9)||40.0 ± 17.9||0.6614||75.0 ± 21.7||0.9630|
|No (n = 59)||47.5 ± 9.9||70.4 ± 9.9|
|6||Yes (n = 12)||25.0 ± 15.3||0.2018||50.0 ± 35.4||0.3273|
|No (n = 56)||52.9 ± 10.0||74.1 ± 9.4|
|7||Yes (n = 8)||50.0 ± 35.4||0.9052||75.0 ± 26.5||0.8871|
|No (n = 50)||46.1 ± 0.90||Not applicable|
|9||Yes (n = 9)||25.0 ± 15.3||0.5496||66.7 ± 27.2||0.8818|
|No (n = 59)||49.6 ± 9.80||71.4 ± 9.9|
|13||Yes (n = 11)||33.3 ± 19.2||0.9441||37.5 ± 21.0||0.0147|
|No (n = 57)||48.2 ± 9.60||82.6 ± 8.9|
|17||Yes (n = 12)||0||0.0045||0||0.00001|
|No (n = 56)||58.4 ± 10.1||84.6 ± 8.0|
|22||Yes (n = 15)||33.3 ± 15.7||0.6317||28.6 ± 17.1||0.0019|
|No (n = 53)||49.6 ± 10.2||83.3 ± 8.8|
|X||Yes (n = 10)||25.0 ± 21.7||0.4570||83.3 ± 17.0||0.5992|
|No (n = 58)||49.5 ± 9.4||68.0 ± 10.7|
Adult patient group
All the adult BL were diagnosed and treated at MDACC between January 1996 and December 2004. There were 27 men and 10 women, ages 19 to 77 years (median, 45 years). Seven patients were HIV-positive. At time of diagnosis, 36 patients had Ann Arbor Stage IV disease (with extensive bone marrow involvement) and 1 patient had Stage III disease. The initial therapy consisted of a hyperCVAD (hyperfractionated cyclophosphamide, vincristine, doxorubicin, and dexamethasone + high-dose methotrexate and cytarabine) regimen16 (13 patients), a hyperCVAD/rituximab regimen (20 patients), and other combination chemotherapy (4 patients). The tumor samples were bone marrow aspirates (36 patients) and a lymph node biopsy sample (1 patient). Outcome information was available in all patients. The 2-year rates of EFS and OS for the entire group were 46.4%± 12.0% and 92.2% ± 9.3%, respectively. The 5-year rates of EFS and OS were 41.3% ± 12.9% and 47.0% ± 14%, respectively. The median follow-up was 2.5 years (range, 0.1–7.1 years). In this patient group, only abnormalities of chromosome 17 were associated with an inferior outcome in univariate analysis (Table 1). There was no difference in outcome between theHIV-positive and HIV-negative patients, patients age < 60 years and those age ≥60 years, and patients treated on the hyperCVAD and the rituximab-hyperCVAD regimens.
The karyotypic findings are summarized in Figure 1. Constitutional inv(9)(p11q13) present in 1 pediatric patient and 2 adult patients was not included in the analysis.
All patients had either the t(8;14) translocation or 1 of the variants, t(8;22), seen in 9 patients (3 children, 6 adults) or t(2;8), seen in 2 adult patients. The frequency of the variant translocations was higher in adults (8/37, 21%), than children (3/33, 9%), but with no statistical significance (P = .19). None of these translocations had a demonstrable prognostic significance.
Additional abnormalities were present in 54 of 70 (77%) patients, including 27 children (81%) and 27 adults (73%). Approximately half of the tumors had complex karyotypes, with a similar incidence in children (17/33; 51%) and adults (19/37; 51%). In children, the presence of a complex karyotype was associated with a poor prognosis in univariate analysis. Both the 5-year EFS and OS were 87.5% ± 8.9% in cases with 1–3 chromosomal abnormalities, and 53.3% ± 16.3% in those with >3 chromosomal abnormalities (P = .049). This parameter did not retain prognostic significance in the multivariable analysis. Complex karyotypes did not have a demonstrable adverse impact on prognosis in the adult BL group.
When excluding the Burkitt-type translocations, the chromosomes most frequently involved in abnormalities (i.e., affected in >15% of all patients) were 1, 6, 13, 17, and 22. The association between chromosomal abnormalities occurring in at least 10% of the patients and outcome is summarized in Table 1.
Abnormalities of chromosome 1, present in 31 of 70 (44%) patients (13 children and 18 adults) were the most common. Three patients had more than 1 abnormality, resulting in a total of 35 abnormalities. The most common abnormalities included: duplication of 1q material in 16 of 35 (45%), and balanced translocations in 15 of 35 patients (42%). There were no statistically significant differences in incidence, type, or distribution of chromosome 1 abnormalities between children and adults. These abnormalities did not have an impact on outcome.
Abnormalities of chromosome 6 were found in 12 of 70 patients (17%), with some cases having more than 1 type of abnormality. Abnormalities included terminal or interstitial deletions or additions of the long arm (8 cases), i(6)(p10) (1 case), translocations t(1;6), (3 cases), and trisomy 6 (1 case). There were no statistically significant differences between children and adults in the type or number of chromosome 6 abnormalities. There was no correlation with outcome.
Abnormalities of chromosome 13 were the next most frequent group. These were found in 12 of 70 patients (17%), with 1 patient having 2 distinct abnormalities, and much more frequently in children (9/33) than in adults (3/37) (P = .03). Most of the abnormalities involved 13q (Table 2). The presence of chromosome 13 abnormalities conferred a worse prognosis in children, both in univariate and multivariate analysis (Fig. 2) (Table 1). These children were treated on LMB89 or SJBCII (3), POG8617 (3), and on other protocols (2). In adults, chromosome 13 abnormalities did not have an impact on outcome.
|Adults||3/37 (8%)||1||1||i13q10 (1)|
|Adults||7/37 (19%)||0||1||17p13 (1)||17p11 (1)||17q21 (1)|
|i17q10 (2)||17p12 (1)|
|22||Children||7/33 (21%)||1||0||22q13 (2)||0||22q11.2(4)|
|Adults||8/37 (22%)||0||1||0||0||22q11.2 (6)|
Chromosome 17 abnormalities were also present in 12 of 70 patients (17%) (Table 2). There was no significant difference in the distribution of abnormalities between children and adults. The presence of chromosome 17 abnormalities had a significant impact on prognosis in both patient groups (in univariate analysis). In multivariable analysis, chromosome 17 abnormalities did not retain prognostic significance in children. In adults, because chromosome 17 alterations were the sole prognostic variable in univariate analysis, no further analysis was performed (Table 1). Further analysis (Spearman correlation test) showed that in children the presence of alterations in chromosome 17 was highly correlated with chromosome 13 (P = .0016), and chromosome 22 (P = .0290) alterations, both of which had an adverse prognostic significance. In adults, there was no correlation between alterations of chromosome 17 and abnormalities of chromosome 13 or 22.
Chromosome 22 abnormalities [including the t(8;22)] were present in 15 of 70 (21%) patients, with 1 patient having more than 1 chromosome 22 abnormality (Table 2). In children, the presence of any chromosome 22 abnormality conferred a worse prognosis, both in univariate and multivariable analysis (Fig. 3). This finding appears to be independent of the t(8;22), which did not confer a poor prognosis by itself. These children were treated on the POG8617 protocol (2), LMB89/SJBCII (3), or other protocols (2). There was no association between chromosome 22 alterations and prognosis in adults.
Other chromosomes affected in more than 10% of the cases were 2, 7, 9, and X. There were no statistically significant differences in the frequency or types of abnormalities between children and adults. Also, there was no overt bearing on outcome.
There was no difference in the type and frequency of karyotypic abnormalities between the HIV-positive and HIV-negative adult patients.
Although pediatric BL is a highly curable disease with current intensive chemotherapy regimens, patients who present with high-stage disease continue to have a less favorable outcome. Characterization of BL from a genetic standpoint is likely to provide further insights into biological factors that contribute to tumor behavior and response to therapy. This analysis is strongly biased toward patients presenting with advanced stage disease. However, this is a limitation likely to be present in any series of BL cases, as patients with high tumor burden and/or bone marrow involvement are typically the most likely to provide fresh tumor samples sufficient for conventional cytogenetic analysis.
Although cytogenetic abnormalities encountered in BL have been described extensively,3, 4, 6–9 most of the series did not separate children and adults for further analysis. Non-Hodgkin lymphomas occurring in children are different from adults in their clinical presentation, pattern of organ involvement, treatment requirements, and response to therapy.23 These observations justify an approach such as the one taken in the present study, whereby chromosomal abnormalities of tumors occurring in children and adults are analyzed separately and then compared as to their nature and prognostic significance.
We found that karyotypic abnormalities occurring in pediatric and adult BL are largely similar, although with few notable differences. Our findings also strongly suggest that specific chromosomal alterations are associated with outcome in these patient age groups.
Recurring chromosome 13q abnormalities were found with increased frequency in pediatric BL, predominantly as 13q rearrangements, and were associated with an inferior outcome in children, but not in adults. It is possible that this effect on outcome was not observed in adults due to the low number of cases with chromosome 13 abnormalities present in this patient group. Chromosome 13 abnormalities have been previously reported in BL5, 11, 25 as the second most common recurring abnormality (3%–25%) in many of the studies.5, 7, 8, 11, 25, 26 An association between chromosome 13q abnormalities and adverse prognosis was suggested by Lones et al.11 in their report of 19 pediatric patients with B-cell lymphoma, in which they found 13q abnormalities in 2 patients with a poor outcome. Rare cases of BL have also been reported in children with congenital 13q deletion syndrome,27 suggesting a connection with lymphomagenesis. Recurrent 13q abnormalities, primarily as deletions, have also been described in many types of B-lymphoid neoplasms, often associated with aggressive tumor features and a poor prognosis.18, 28–33 Attempts to identify specific genes altered by 13q abnormalities have not been successful to date. The affected regions are variable and span large portions of 13q, ranging from 13q14 to 13q34.34–37 These regions include many expressed sequence tags (ESTs) as well as known genes encoding for transcription factors and cell cycle regulators, such as retinoblastoma (RB1), Kruppel-like factor 12 (KLF12), ret finger protein 2 (RFP2), Drosophila dachshund homolog 1 (DACH1), deleted in lymphocytic leukemia 1 and 2 (DLEU1 and DLEU2), and glypican 5 (GPC5), a gene with putative roles in the regulation of cell division and growth.38 However, none of the studies addressing these specific genes have been able to find alterations, other than the monoallelic loss associated with the 13q deletions.
Chromosome 17 abnormalities were present with similar frequencies in pediatric and adult BL and were associated with a poor prognosis in adults. Such abnormalities have been reported previously in BL, as well as in many other B-lineage lymphoproliferative disorders,18, 39–42 often associated with aggressive disease features and a poor prognosis. Involvement of chromosome 17 has been previously described in adult BL. In their review of 170 cases of BL occurring in adults, Kornblau et al.4 found that the presence of any type of chromosome 17 abnormality was associated with a poor response to therapy. The association found in children in our study is more tenuous, as the poor prognosis associated with chromosome 17 abnormalities observed in univariate analysis could be due to a strong association with cases also containing other chromosomal abnormalities that correlate with poor outcome. It is also possible that the presence of multiple genetic lesions (including chromosome 17), rather than chromosome 17 itself, contributes to poor outcome. It is not known precisely what genes are affected by the cytogenetic abnormalities of chromosome 17. Most of the lesions affect the short arm of this chromosome, which harbors several known or putative tumor suppressor genes and which is also a site prone to rearrangements due to an increased instability associated with the presence of low-copy repeats (LCR). Genes located on 17p include TP53, the tumor suppressor gene most commonly affected in human cancers, BCL6B (B cell lymphoma 6, member B),43 and ZNFN1A3 (Aiolos), a transcription factor important in lymphoid development and lymphomagenesis.44
Chromosome 22 abnormalities, including the t(8;22), were present with similar frequency in pediatric and adult BL in this study and correlated with an inferior outcome in children. Such abnormalities often occur in B-lineage neoplasms as translocations involving the lambda light chain gene on 22q11. The most common translocations juxtapose this locus to c-myc [t(8;22)(q24;q11) associated with BL, multiple myeloma and other high-grade B-lineage lymphomas], BCL6 [t(3;22) (q27;q11) associated with diffuse large B-cell lymphoma]45 and BCL2 [t(18;22)(q21;q11) associated with chronic lymphocytic leukemia].46 Abnormalities of chromosome 22 have also been reported in association with aggressive variants of mantle cell lymphoma.41 The abnormalities present in our study, in addition to the BL-associated t(8;22), also involved predominantly rearrangements of the proximal portion of 22q, suggesting that its role in pathogenesis might be related to dysregulation of other genes through juxtaposition with the lambda light chain gene.
The most common abnormalities identified in our study, both in children and in adults, involved the long arm of chromosome 1, most often as additions or duplications. These abnormalities have been documented in approximately one-third of BL cases by others.3, 4, 9, 47, 48 Chromosome 1 abnormalities also occur in approximately one-third of follicular lymphomas,31 of germinal center cell origin similar to BL, where they appear to have a negative impact on prognosis.42 The biological significance of these abnormalities is unclear. Jumping translocations resulting in duplications of chromosome 1q material may occur with increased frequency in BL,49 offering a possible explanation for their frequent presence in these tumors. Recent studies have demonstrated abnormalities of nuclear localization of the constitutive heterochromatin localized on 1q12 in cell lines derived from follicular lymphoma and BL,50 with putative effects on the expression of various transcription factors.
Other common alterations were those of chromosome 6. Such alterations, most commonly 6q deletions, represent one of the most frequent karyotypic abnormalities encountered in the B-lymphoid neoplasms (10%–40%).18, 41, 42, 51–55 While possibly correlating with an inferior outcome in follicular lymphoma,42 they have no prognostic significance in most other neoplasms. In BL, 6q deletions have been found in 10% to 20% of cases in previously reported series4, 6 and in cell lines.56 The regions of minimal common 6q deletion vary with disease type, suggesting that different tumor suppressor genes might be affected in each of these neoplasms.57–59 In BL cell lines these deletions are consistently associated with loss of heterozygosity in the 6q25–27 region.56 Genes possibly affected by 6q abnormalities include BACH2 (encoding for B-lineage specific transcription factor, a putative tumor suppressor gene), on 6q15,60, 61 and BLIMP1 (B-lymphocyte maturation promoting), on 6q21–q22.1.62
In conclusion, our study identifies distinct secondary cytogenetic abnormalities in BL, some of which correlate with outcome in pediatric or adult patients. The statistical significance of our findings is somewhat limited by the relatively low numbers of patients included in this study and by the variation in the treatment protocols employed. Nevertheless, these data could support research into areas of the genome that may contain genes responsible for the observed biological behavior of BL, may provide a basis for further refining the risk stratification of BL patients, and may also help direct targeted therapies as key molecular changes are identified.
We thank Kimberly Hayes, BSCLPCG, for assistance with generating and retrieving the karyotype data for the patients treated at the University of Texas M. D. Anderson Cancer Center
- 10Homogeneous immunophenotype and paucity of secondary genomic aberrations are distinctive features of endemic but not of sporadic Burkitt's lymphoma and diffuse large B-cell lymphoma with MYC rearrangement. J Pathol. 2004; 203: 940–945., , , et al.
- 17Cytogenetics as a diagnostic aid for childhood hematologic disorders: conventional cytogenetic techniques, fluorescence in situ hybridization, and comparative genomic hybridization. In: HanausekM, WalaszekZ, editors. Methods in Molecular Medicine. Totowa, NJ: Humana Press; 1998: 209–227., , .
- 19MittelmanF, editor. An International System for Human Cytogenetic Nomenclature. Basel: S. Karger, 1995.
- 21The Statistical Analysis of Failure Time Data. New York: John Wiley & Sons; 1980., .
- 22Analysis and Survival Data. London: Chapman and Hall; 1984., .