It has been generally acknowledged that the diagnosis, treatment and prognosis evaluation of leukemia largely rely on an adequate identification of genetic abnormalities. A systemic analysis of genetic aberrations was performed in a cohort of 1346 patients with newly diagnosed acute lymphoblastic leukemia (ALL) in China. The pediatric patients had higher incidence of hyperdiploidy and t(12;21) (p13;q22)/ETV6–RUNX1 than adults (P<0.0001); in contrast, the occurrence of Ph and Ik6 variant of IKZF1 gene was much more frequent in adult patients (all P<0.0001). In B-ALL, the existence of Ik6 and that of BCR–ABL were statistically correlated (P<0.0001). In comparison with Western cohorts, the incidence of t(9;22) (q34;q11)/BCR–ABL (14.60%) in B-ALL and HOX11 expression in T-ALL (25.24%) seemed to be much higher in our group, while the incidence of t(12;21) (p13;q22)/ETV6–RUNX1 (15.34%) seemed to be lower in Chinese pediatric patients. The occurrence of hyperdiploidy was much lower either in pediatric (10.61% vs 20–38%) or adult patients (2.36% vs 6.77–12%) in our study than in Western reports. In addition, the frequencies of HOX11L2 in adult patients were much higher in our cohort than in Western countries (20.69% vs 4–11%). In general, it seems that Chinese ALL patients bear more adverse prognostic factors than their Western counterparts do.
Identification of genetic abnormalities, the common characteristics of cancer, is essential in diagnosis, treatment and prognostic evaluation of leukemia, because it can help to stratify this heterogeneous disease into different groups. Over the last several decades, a growing number of genetic aberrations have been documented as prognostic and predictive markers in acute lymphoblastic leukemia (ALL), including recurring chromosomal translocations, gene mutations, deletions and amplifications and other molecular alterations.1, 2, 3 Of note, the molecular characterization of cytogenetic abnormalities has not only provided insights into the mechanisms of leukemogenesis, but also led to the establishment of new tailored treatment strategies, as exemplified by the success of tyrosine kinase inhibitor in the treatment of ALL with t(9;22) (q34;q11) and its associated BCR–ABL fusion gene.4
Children and adult ALL manifest distinct clinical behaviors, disease features and treatment outcomes, which are nowadays believed to be determined by different genetic aberration spectrums between these two groups of patients. The incidences and frequencies differ in terms of important genetic alterations with prognostic value between children and adult ALL. For example, BCR–ABL fusion gene and 11q23-associated MLL gene rearrangements related to poor prognosis seem to have lower incidence in pediatric ALL as compared with adult one. With the development of high-resolution genomic profiling technology, more genetic defects have been found which may contribute to explaining the underlying physiopathology of ALL and to clarifying the differences in somatic genetics background between children and adult cases, such as t(9;22) (q34;q11)/BCR–ABL, t(1;19) (q23;p13)/TCF3–PBX1, t(4;11) (q21;q23)/MLL–AFF1, t(11;19) (q23;p13)/MLL–MLLT1, t(8;14) (q24;q32)/IGH@–MYC, t(12;21) (p13;q22)/ETV6–RUNX1, CRLF2 overexpression, IKZF1 transcript variants in B-ALL, and del(1p32)/SIL–TAL1, t(10;11) (p13;q21)/CALM–AF10 fusion genes, HOX11/HOX11L2 expression, and NOTCH1 mutations in T-ALL. Using these molecular markers, it might be feasible to further stratify the treatment of patients according to different mechanisms in ALL pathogenesis and prognosis.
Large-scale analyses of the spectrum and incidence of genetic abnormalities were carried out in pediatric and adult patients in Western countries by different groups. Considering the potential ethnic background difference between Western and Chinese populations, such kind of investigation is of important value for Chinese ALL patients. Hence, we performed this study to systematically examine the previously known genetic events, newly established molecular markers and clinical profile of 1346 patients with ALL from all over China while treated in the hospitals affiliated to Shanghai Institute of Hematology (SIH). In particular, we intended to compare the differences in both incidence and spectrum of genetic aberration between Chinese and Western cohorts.
Patients and methods
A total of 1452 patients with de novo ALL were diagnosed and/or treated in the SIH-based hospital network between November 1989 and December 2010. French–American–British criteria was used to define the ALL morphologic subtypes (L1 through L3, with a few cases not classifiable according to the French–American–British nomenclature).5 Morphological, immunophenotypical and cytogenetic assessments were performed at the time of diagnosis and the molecular studies were carried on once the biomarkers were available. Samples from 1346 out of the 1452 patients were available and subject to cytogenetic and/or molecular analysis. All the data were re-reviewed by the institutional evaluation group. This study was approved by the ethical board of the participating centers. All patients gave informed consent for both treatment and cryopreservation of bone marrow (BM) and peripheral blood according to the Declaration of Helsinki.
Cytological examinations were performed on the BM and peripheral blood smear using Wright–Gimesa's staining for each patient at diagnosis. Periodic acid Schiff reaction, naphthol AS-D chloroacetate esterase staining and neutrophil alkaline phosphatase staining tests were performed.
Immunophenotyping was performed by identifying the CD45-positive population in fresh BM or peripheral blood blast cells. In the assessment of the different surface antigens (CD45, CD34, CD19, CD20, CD22, CD10, CD13, CD33, human leukocyte antigen-DR, CD2, CD3, CD4, CD5, CD7, CD8 and CD56), the samples containing a cell population larger than 20% expressing certain antigens were considered as positive for such antigens. Positivity for terminal deoxynucleotide transferase and cytoplasmic antigens (CyCD3, CyIgM, CD79a) was rendered to the samples with >10% of cells exhibiting nuclear or intracytoplasmic fluorescence. These phenotyping experiments were accomplished with immunohistochemistry (Dako, Glostrup, Denmark) before 2001, and with flow cytometry analysis (BD, Franklin Lakes, NJ, USA) after 2001. The immunologic definition in B- and T-lineage ALL rested on the European Group for the Immunological Characterization of Leukemia criteria.6, 7
Cytogenetic studies were performed centrally in SIH. Chromosomes were R-banded and/or G-banded on unstimulated BM cells after a 24-h culture. The R- and/or G-banded karyotype was analyzed according to the International System for Human Cytogenetic Nomenclature (2005)8 and most cases were confirmed with relevant molecular markers. For multiplex fluorescence in situ hybridization analysis, the multi-color probe kit and the assay procedure were provided by MetaSystems (Hamberg, Germany).
Detection of gene fusions, abnormal expressions and mutations
BM samples were collected at diagnosis and mononuclear cells were enriched by density gradient centrifugation with Ficoll solution. Genomic DNA and total RNA were extracted using TRIzol reagent (Invitrogen, Carlsbad, CA, USA). BCR–ABL, ETV6–RUNX1, TCF3–PBX1, MLL–MLLT1, MLL–AFF1 and SIL–TAL1 fusion genes were assessed by the multiplex-nested reverse transcriptase (RT)-PCR amplification according to Pallisgaard et al.9 with modification. Expression of HOX11 and HOX11L2, and NOTCH1 mutations were assessed according to Zhu et al.10 The CALM–AF10 fusion gene was detected by using the published protocol.11
Detection of transcript variants and genomic deletion of IKZF1
After reverse transcription, complementary DNA was used for the detection of expression of IKZF1 transcripts as previously described.12 Briefly, RT-PCR analysis for IKZF1 was performed with the primer pair of F1 5′-CCCCTGTAAGCGATACTCCAGATG-3′ and R1 5′-GATGGCTTGGTCCATCACGTGGGA-3′, of which the products could include all IKZF1 transcripts. Ik6 variant, defined as lacking of the exon 4–7 (Δ4–7), was the major abnormal transcript variant of IKZF1 gene that we detected. To identify the genomic lesions of IKZF1, we examined DNA samples from 50 cases with Ik6 by long-range genomic PCR and sequencing analysis. It was found that the genomic deletion was consistent with the RT-PCR data. To further confirm and fine-map the focal IKZF1 deletions, DNA samples from six cases with Ik6 were analyzed by Affymetrix NspI 250K and SNP6.0 arrays (Singapore) according to the manufacturer's instructions. As a result, the recurrent DNA copy number anomalies in IKZF1 gene and the deletion of exons 4 to 7 with breakpoints occurring in introns 3 and 7 on chromosome 7p12 were identified.13 The details were shown in Supplementary Figure 1.
Identification of CRLF2 overexpression
The expression of CRLF2 was detected according to Cario et al.14 Briefly, real-time RT-PCR was performed by a SYBR Green PCR Mastermix (Qiagen, Hilden, Germany). QuantiTect Primer Assays were used (CRLF2 (QT00210987), Qiagen). The glyceraldehyde 3-phosphate dehydrogenase was co-amplified as an endogenous control to standardize the amount of the sample RNA added to the reaction. Each sample was tested in triplicate. The cut-off value of CRLF2 overexpression was identified according to Yoda et al.15 Genomic rearrangement was confirmed by fluorescence in situ hybridization for the detection of breakpoints of IGH@ locus (LSI IGH@ Dual Color Break-Apart Rearrangement Probe, Abbott Diagnostics, Chicago, IL, USA) and involvement of CRLF2 on chromosome X,16 and genomic PCR for the PAR1 deletion resulting in fusion of P2RY8–CRLF2 transcript, which was also confirmed by RT-PCR.17, 18 (Supplementary Figure 2).
Fisher's exact P-test was performed to compare the differences of gender, immunophenotype and cytogenetic abnormalities between adults and children ALL patients, as well as the cytogenetic/molecular spectrums between our group and Western cohorts. Two-side P-values <0.05 were considered statistically significant. All the above statistical procedures were performed with the SPSS statistical software package, version 13.0 (SPSS Inc., Chicago, IL, USA).
Characteristics of the patients
From November 1989 to December 2010, a total of 1346 patients with newly diagnosed ALL were included into the study. Among them, 400 (29.72%) were adults (18 years) and 946 (70.28%) were children (<18 years). The characteristics of the entire patient cohort are summarized in Table 1.
There were 284 childhood patients aged from 3 to 5 years, which accounted for 21.10% of the entire and 30.02% of childhood patients, respectively.
Distributions and frequencies of cytogenetic and/or molecular abnormalities
After cytogenetic and molecular assays, the genetic profiles were obtained in 1346 (92.7%) out of 1452 ALL patients, including 946 children and 400 adults. Cytogenetic and/or molecular abnormalities were identified in 922 (68.50%) among these 1346 ALL patients.
In general, the numerical abnormalities were mainly observed in children, while the structural abnormalities were more predominant in adults (Table 2). The incidence of hyperdiploidy (defined as bearing 51–65 chromosomes) was obviously higher in the pediatric patients than in adults (10.57% vs 2.00%, P<0.0001), and most of the patients of hyperdiploidy were with B-lineage ALL (77.78%). In addition, near-tetraploidy (84–94 chromosomes) was observed in six cases. There were no significant differences in other numerical abnormalities between adults and children ALL (P=0.4781).
Table 2 shows the incidences and frequencies of cytogenetic and/or molecular aberrations relatively specific for B-ALL such as t(9;22)/BCR–ABL, t(1;19)/TCF3–PBX1, t(4;11)/MLL–AFF1, t(11;19)/MLL–MLLT1, t(8;14)/IGH@–MYC, t(12;21)/ETV6–RUNX1, CRLF2-overexpression, and variant transcripts of IKZF1 gene and for T-ALL such as del(1p)/SIL–TAL1, t(10;11)/CALM–AF10, HOX11 and HOX11L2 expression and NOTCH1 mutations. Interestingly, two patients of t(9;22)/BCR–ABL showed T-ALL immunophenotypic profile (Supplementary Table 1). The abnormalities of t(9;22)/BCR–ABL were more frequent in adult ALL than in childhood one (P<0.0001), as in the case of the distribution of Ik6 variant of IKZF1 gene (P<0.0001). On the contrary, t(12;21)/ETV6–RUNX1 was of significantly higher incidence in pediatric patients than in adult ones (P<0.0001). More involvement of t(4;11)/MLL–AFF1 in adults (P=0.0022) and t(1;19)/TCF3–PBX1 in children (P=0.0004) was also observed. In addition, although statistical significance was not reached (P=0.0628), a tendency of high prevalence of NOTCH 1 mutations was observed in adult patients than in pediatric parallels.
Other simple translocations were involved in 30 patients. Among them, several chromosomal translocations with at least one known gene involved had already been reported in the literatures, such as t(3;9) (q26;p23)/EVI1–?,19 t(12;17) (p13;q11)/ZNF384–TAF15,20 t(9;14) (q34;q32)/ABL1–EML1,21 t(10;14) (q24;q11)/HOX11–TRD,22 t(8;14) (q24;q11) /MYC–TCR23 in children, and t(1;5) (q23;q33)/?–PDGFRB,24 der(9)t(7;9) (q11;p13)/PAX5–ELN,25 t(8;14) (q11;q32)/CEBPD–IGH@,26 t(3;11) (q29q13;p15)del(3) (q29)/NUP98–IQCG27 in adults. We also discovered 20 new simple translocations that were revealed by R/G banding, and confirmed by chromosome painting and multiplex fluorescence in situ hybridization (Supplementary Table 2).
Complex translocation karyotypes were observed in 15 patients, most of them being derived from the progression of t(9;22)/BCR–ABL (9 patients) and t(12p;?)/ETV6–? (6 patients) (Supplementary Table 3). Multiplex fluorescence in situ hybridization data were available in four patients with complex translocations, among them, one patient (case 11) carried two different clones as indicated in Supplementary Figure 3. In addition, 10 cases with co-existence of two different fusion genes were identified as insidious complex translocations.
In addition, there was a group of 137 patients (50 adults and 87 children) bearing non-translocation structural changes.
Co-existence of cytogenetic and/or molecular alterations in ALL
Figure 1 represents the co-existence status of cytogenetic changes, gene mutations and gene expression levels among distinct immunophenotypic ALL. The details of the data could be seen in Supplementary Table 4.
The phenomenon of overlapping cytogenetic and/or molecular events were observed in either B- or T-lineage ALL and some of them were even statistically correlated. In B-ALL, CRLF2 overexpression (P=0.124) and Ik6 variant of IKZF1 gene (P<0.0001) were overlapped with t(9;22)/BCR–ABL. Of importance, in the patients with Ph chromosome, the frequency of Ik6 variant could be as high as 43.64% (72/165), which was much greater than that in patients without Ph (12.77%, 66/517) (P<0.0001). Meanwhile, HOX11 and HOX11L2 expression and NOTCH1 mutations were overlapped to some extent in T-ALL.
Association of cytogenetic and/or molecular abnormalities with age
Figure 2 displayed the comparison of distribution of different cytogenetic and/or molecular aberrations in different age groups. With the exception of infant ALL, the occurrence of t(9;22)/BCR–ABL increased with the growth of age, and this tendency was also observed in ALL with 11q23/MLL-related chromosomal translocations. In contrast, the frequency of hyperdiploidy seemed decreased with age. In addition, t(12;21)/ETV6–RUNX1 fusion was mainly observed in patients aged below 19 years. The frequencies and distributions of cytogenetic and/or molecular abnormalities of adolescent patients (15–19 years) tended to be very similar to that of elder children patients (10–14 years), although the former group showed a slight increase in non-translocation structural change, other simple translocations and complex translocations, and a relative decrease in numeric changes as compared with the latter group.
In addition, we found that non-chromosomal translocation structural abnormalities were significantly differently distributed in children and adult patients: chromosome (12p)/(containing ETV6 gene), chromosome (1q)/(including PBX1 gene), chromosome (9p)/(comprising PAX5 and CDKN2A genes) and other related genetic translocations, deletions and interruptions, as well as i (17q) (fragment of deletion including p53 gene), dup (5q31)/(containing duplication of AFF4 gene) seemed to be more involved in children ALL; on the contrary, i(7q)/(lack of IKZF1 gene) was more implicated in adult ALL patients. The detailed data were shown in Figure 3.
Comparison of the cytogenetic and/or molecular spectrums between Chinese and Western cohorts
We have compared the incidences and frequencies of the cytogenetic and/or molecular spectrums between our series and Western groups referenced in Tables 3 and 4 for pediatric and adult ALL, respectively. The Western cohorts with the number of the patients >1000 and 200 were selected for pediatric and adult B-ALL, respectively. Besides, we also enrolled two Asian reports including one from Taiwan, China for comparison.28, 29 The incidence of t(9;22)/BCR–ABL (14.60%) in B-ALL seemed to be higher in our series. In contrast, the incidence of t(12;21)/ETV6–RUNX1 (15.34%) seemed to be lower in pediatric patients among Chinese cohorts, and the difference reached statistical significance (P<0.05) (Table 3). The occurrence of hyperdiploidy was significantly lower either in pediatric (10.61% vs 20–38%, all P<0.05) or adult patients (2.36% vs 6.77–12%, all P<0.05) in our study than in Western reports.3, 30, 31, 32, 33, 34, 35 The frequency of HOX11L2 expression in our cohort was similar to that in children T-ALL in Western countries (25.24% vs 23.6–24.1%); however, it seemed much higher in adults in our group (20.69% vs 4–11%).36, 37, 38, 39, 40 Moreover, the HOX11 expression seemed to be higher in pediatric (25.24% vs 7.1–8.33%, P<0.05) but lower in adult patients (17.24% vs 18–31%) in our cohort as compared with Western reports (Table 4).36, 37, 38, 39, 40 No significant differences were observed in terms of other markers.
It has been long appreciated that genetic abnormalities were the most important factors that determine the clinical behavior, treatment response and outcome of patients with ALL either in children or in adults. With the accumulation of more data of newly identified cytogenetic and/or molecular events, the in depth investigation of basic mechanism of the pathogenesis of ALL may be feasible. In this study, we present the incidences and patterns of cytogenetic and/or molecular aberrations in both children and adult ALL in a large cohort of 1346 Chinese patients, and compared differences in terms of genetic markers between Western and Chinese ALL settings.
Co-existence of gene abnormalities previously considered to be involved in distinct molecular pathways were observed, such as the t(9;22)/BCR–ABL and Ik6 variant of IKZF1 gene (P<0.001). Interestingly, as indicated in a sequential study, both alterations were associated with a poor prognosis of the ALL patients. IKZF1 gene has several splice variations, and it is thought that these splice variants could influence IKZF1 function. More recent studies suggested that, in vitro, BCR–ABL could affect IKZF1 splicing, leading to the selection of non-DNA-binding isoform of Ik6, which might be the molecular basis of the correlation of BCR–ABL and Ik6.41 Although the interaction between BCR–ABL and Ik6 seems fascinating, the detailed mechanism is still unknown, and further research is warranted to investigate the potential cross-talk of these genetic events.
As the main part of this study, we compared the spectrum of cytogenetic/molecular aberrations between pediatric and adult patients, and between Western and Chinese cohorts. We believe that the substantial difference of the prognosis between pediatric and adult patients is dependent on the different cytogenetic/molecular alterations underlying the disease mechanism, which makes ALL in children and that in adults almost two different disease groups. In this study, we identified a number of genetic characteristics in Chinese childhood and adult ALL: hyperdiploidy (51–65 chromosomes), t(12;21)/ETV6–RUNX1, t(1;19)/TCF3–PBX1 and HOX11 expression occurring more frequently in children, whereas Ph chromosome, t(4;11)/MLL–AFF1, Ik6 variant of IKZF1 gene, and NOTCH1 mutations being more common in adults; with regard to non-translocation structural changes, some being only involved in children, such as dup(1q32), del(1q21q31), del(2p24), dup(5q31), add(9p22), i(9q), del(12p13) and i(17q), while i(7q) occurring exclusively in adults. Taking the above data as a whole, an interesting pattern emerged (also see Supplementary Table 5): the molecular events of childhood ALL seem more likely associated with the abnormalities of transcription factors or co-factors regulating embryonic or hematopoietic development and also affecting cell differentiation such as ETV6 (12p13.1), PBX1 (1q23), MEF2D(1q22), HOX11 (10q24), PAX5 (9p13) and AFF4 (5q31); whereas those of adult ALL were more involved in signal transduction pathways giving rise to proliferative advantages to leukemia clones. It is reasonable to assume that a number of genes implicated in the developmental processes are more active in children. Moreover, the regulation of these genes should be very complicated as some of them are only expressed at early stages (such as PAX5) in B-cell differentiation;42, 43 and others are downregulated with the growing of age (such as AFF4).44 In contrast, in adult patients, the most frequent ones in B-ALL are t(9;22)/BCR–ABL that activates the tyrosine kinase signaling pathway, Ik6 variant of IKZF1 gene, and CRLF2 overexpression that is a cytokine receptor activating the JAK/STAT pathway.14, 15 Interestingly, CRLF2 overexpression and IKZF1 splicing deletion, which are significantly more involved in adults than in children, are both related to the poor outcome in B-ALL and closely associated with JAK mutations.16, 18, 45 Analyzed by gene expression profiling, the three abnormal transcriptional signatures of CRLF2 overexpression, IKZF1 aberration and BCR–ABL demonstrated a high degree of similarity, suggesting the presence of similar pathogenic mechanisms with interference of the transcriptional regulations or signaling pathways that control the proliferation, survival and self-renewal of hematopoietic stem cells.18, 45 In parallel, we observed that MLL gene was more frequently involved in adults than in children, supporting the point of view that epigenetic regulatory pathway, which has an important role in the pathogenesis of leukemia may contribute more to the onset of adult ALL than to that of in pediatric ones.46, 47, 48 On the basis of these data, we conclude that the cytogenetic/molecular abnormalities in the pathogenesis of pediatric and adult ALL might be substantially different. In the ‘borderline’ age groups between adult and pediatric patients (that is, adolescent patients and elder childhood patients), a tendency of similar frequencies and distributions of cytogenetic and/or molecular changes was observed, although adolescent patients showed more chromosomal structural changes and less numeric changes. Given the similar spectrum of genetic changes, more intensive treatment regimen favored in childhood patients might be appropriate for adolescent ones in order to improve clinical outcome.
In this survey, we identified that the incidence of Ph chromosome in Chinese adult patients was similar to that of Western cohorts;32, 34, 49, 50 while in children it was much higher in our series than in Western reports (14.60% vs 1.5–3%).3, 30, 31, 51, 52, 53 In the pediatric B-ALL patients, the frequency of t(12;21)/ETV6–RUNX1 was much lower in our cohorts than that in Western countries3, 30, 31, 51, 52, 53 and some other Asian countries (15.34% vs 18.8–25%).28 However, our results are very similar to the reports of the subgroup of childhood ALL of Chinese origin reported by Ariffin et al.28 (13.3%) and by Liang et al.29 (14.1%). In addition, the incidence of hyperdiploidy was obviously lower in Chinese either in pediatric or in adult patients,3, 30, 31, 32, 33, 34, 35 which was substantiated by Liang's report.29 The occurrence of HOX11 expression in our series was also different from that of Western reports, evidenced by much more frequencies in children (25.24% vs 7.1–8.33%), and less frequencies in adults (17.24% vs 18–31%) in Chinese cases.36, 37, 38, 39, 40 On the other hand, although the frequency of HOX11L2 expression in childhood T-ALL revealed in this study was similar to that in Western cohorts, it seemed much higher in the Chinese adult ALL patients than in Western ones (20.69% vs 4–11%).36, 37, 38, 39, 40 In general, when compared with data from Western countries, the frequency of cases with adverse prognostic factors seems to be higher in Chinese population than in Western cohorts, including t(9;22)/BCR–ABL in B-ALL and HOX11L2 expression in T-ALL. In contrast, the incidence of cytogenetic changes bearing favorable prognostic value such as t(12;21)/ETV6–RUNX1 and hyperdipoidy (51–65 chromosomes) seems to be lower in Chinese pediatric ALL cohorts. Similar tendency is also observed in adult patients. As summarized in Tables 3 and 4, the ALL patients in Chinese population, either in children or in adults, seem to be more risky than in Western cohorts.
In conclusion, the frequencies and distributions of genetic abnormalities of ALL patients between children and adult ones, and Chinese and Western ones show significant differences in this study, and the potential impact of these differences on clinical behavior and treatment outcome is urgently needed to be investigated to further discriminate the nuance of the ALL subgroups.
This work was supported in part by the Mega project of Ministry of Science and Technology (2008ZX09312-026, 2009ZX09103-431), the National Basic Research Program (973, 2010CB529200), the National Natural Science Foundation of China (30871106, 30772744, 30821063) and Science and Technology Commission Foundation of Shanghai (09dZ1974500).
About this article
Supplementary Information accompanies the paper on the Leukemia website (http://www.nature.com/leu)