ABSTRACT
Although children with acute lymphoblastic leukemia (ALL) generally have a good outcome, some patients do relapse and survival following relapse is poor. Altered DNA methylation is highly prevalent in ALL and raises the possibility that DNA methylation-based biomarkers could predict patient outcome. In this study, genome-wide methylation analysis, using the Illumina Infinium HumanMethylation450 BeadChip platform, was carried out on 52 diagnostic patient samples from 4 genetic subtypes [ETV6-RUNX1, high hyperdiploidy (HeH), TCF3-PBX1 and dic(9;20)(p11-13;q11)] in a 1:1 case-control design with patients who went on to relapse (as cases) and patients achieving long-term remission (as controls). Pyrosequencing assays for selected loci were used to confirm the array-generated data. Non-negative matrix factorization consensus clustering readily clustered samples according to genetic subgroups and gene enrichment pathway analysis suggested that this is in part driven by epigenetic disruption of subtype specific signaling pathways. Multiple bioinformatics approaches (including bump hunting and individual locus analysis) were used to identify CpG sites or regions associated with outcome. However, no associations with relapse were identified. Our data revealed that ETV6-RUNX1 and dic(9;20) subtypes were mostly associated with hypermethylation; conversely, TCF3-PBX1 and HeH were associated with hypomethylation. We observed significant enrichment of the neuroactive ligand-receptor interaction pathway in TCF3-PBX1 as well as an enrichment of genes involved in immunity and infection pathways in ETV6-RUNX1 subtype. Taken together, our results suggest that altered DNA methylation may have differential impacts in distinct ALL genetic subtypes.
Subject(s)
Basic Helix-Loop-Helix Transcription Factors/genetics , Core Binding Factor Alpha 2 Subunit/genetics , DNA Methylation/genetics , DNA-Binding Proteins/genetics , Epigenomics , Precursor Cell Lymphoblastic Leukemia-Lymphoma/genetics , Proto-Oncogene Proteins c-ets/genetics , Proto-Oncogene Proteins/genetics , Repressor Proteins/genetics , Child , Child, Preschool , CpG Islands/genetics , Genome, Human , Humans , Infant , Oncogene Proteins, Fusion/genetics , Pre-B-Cell Leukemia Transcription Factor 1 , Precursor Cell Lymphoblastic Leukemia-Lymphoma/classification , Precursor Cell Lymphoblastic Leukemia-Lymphoma/pathology , Prognosis , Recurrence , ETS Translocation Variant 6 ProteinABSTRACT
BACKGROUND: Chromosomal abnormalities in childhood acute lymphoblastic leukaemia are well established disease markers and indicators of outcomes. However, the long-term prognosis and independent prognostic effect of some abnormalities has been questioned. Also, little is known about the association between cytogenetics and the characteristics of relapse (eg, time and site of relapse) that are known to predict outcome after relapse. METHODS: We analysed cytogenetic data from 1725 children with B-cell precursor acute lymphoblastic leukaemia who were included in the UK Medical Research Council ALL97/99 study and followed up for a median time of 8.2 years. Univariate and multivariate analysis were done to examine risk of relapse, event-free survival, and overall survival associated with 21 chromosomal abnormalities and three cytogenetic risk groups constructed from these data. FINDINGS: Two chromosomal abnormalities were associated with a significantly better outcome (ETV6-RUNX1, hazard ratio [HR] 0.51, 95% CI 0.38-0.70 and high hyperdiploidy, 0.60, 0.47-0.78), whereas five abnormalities were associated with an increased risk of relapse (intrachromosomal amplification of chromosome 21 [iAMP21], 6.04, 3.90-9.35; t(9;22), 3.55, 2.21-5.72; MLL translocations, 2.98, 1.71-5.20; abnormal 17p, 2.09, 1.30-3.37; and loss of 13q, 1.87, 1.09-3.20). Multivariate analysis incorporating age, white-cell count, and treatment parameters showed that six cytogenetic abnormalities (ETV6-RUNX1, high hyperdiploidy, iAMP21, t(9;22), loss of 13q, and abnormal 17p) retained their significance for effect on relapse risk. Based on these data, patients were classified into good, intermediate, and poor cytogenetic risk groups. Slow early treatment response correlated with cytogenetic risk group: 34 of 460 (7%) in the good-risk group, 22 of 211 (10%) in the intermediate-risk group, and 27 of 95 (28%) in the poor-risk group had a slow response (p<0.0001). Additionally, the proportion of patients with a very early (<18 months) relapse varied by cytogenetic risk group: eight of 129 (6%) patients in the good-risk group had a very early relapse, compared with 24 of 98 (24%) in the intermediate-risk group, and 37 of 82 (45%) in the poor-risk group (p<0.0001). However, there was no difference in the site of relapse by cytogenetic risk group. INTERPRETATION: Individual chromosomal abnormalities are strong independent indicators of outcome, especially risk of relapse. Diagnostic cytogenetics identifies patients with a higher rate of relapse and those who are likely to have a high-risk relapse. FUNDING: Leukaemia and Lymphoma Research (LLR).
Subject(s)
Chromosome Aberrations , Precursor Cell Lymphoblastic Leukemia-Lymphoma/genetics , Adolescent , Child , Child, Preschool , Humans , Infant , Precursor Cell Lymphoblastic Leukemia-Lymphoma/therapy , Prognosis , Survival AnalysisABSTRACT
BACKGROUND: 6-mercaptopurine has been a standard component of long-term continuing treatment for childhood lymphoblastic leukaemia, whereas 6-thioguanine has been mainly used for intensification courses. Since preliminary data have shown that 6-thioguanine is more effective than 6-mercaptopurine, we compared the efficacy and toxicity of the two drugs for childhood lymphoblastic leukaemia. METHODS: Consecutive children with lymphoblastic leukaemia diagnosed in the UK and Ireland between April, 1997, and June, 2002, were randomly assigned either 6-thioguanine (750 patients) or 6-mercaptopurine (748 patients) during interim maintenance and continuing therapy. All patients received 6-thioguanine during intensification courses. We analysed event-free and overall survival on an intention-to-treat basis. We obtained toxicity data using an adverse-event reporting system, with follow-up questionnaires to seek detailed information for specific toxicities. This trial is registered with the International Standard Randomised Controlled Number 26727615 with the name ALL97. FINDINGS: After a median follow up of 6 years, there was no difference in event-free or overall survival between the two treatment groups. Although 6-thioguanine conferred a significantly lower risk of isolated CNS relapse than did 6-mercaptopurine (odds ratio [OR] 0.53, 95% CI 0.30-0.92, p=0.02), the benefit was offset by an increased risk of death in remission (2.22, 1.20-4.14, p=0.01), mainly due to infections during continuing therapy. Additionally, 95 patients developed veno-occlusive disease of the liver. Of these, 82 were randomly assigned 6-thioguanine, representing 11% of all 6-thioguanine recipients. On long-term follow-up, about 5% of 6-thioguanine recipients have evidence of non-cirrhotic portal hypertension due to periportal liver fibrosis or nodular regenerative hyperplasia. INTERPRETATION: Compared with 6-mercaptopurine, 6-thioguanine causes excess toxicity without an overall benefit. 6-mercaptopurine should remain the thiopurine of choice for continuing therapy of childhood lymphoblastic leukaemia.