AbstractCancer cells undergo massive alterations in their DNA methylation patterns which result in aberrant gene expression and malignant phenotypes. Abnormal DNA methylation is a prognostic marker in several malignancies, but its potential prognostic significance in adult T-cell acute lymphoblastic leukemia (T-ALL) is poorly defined. Here, we performed methylated DNA immunoprecipitation to obtain a comprehensive genome-wide analysis of promoter methylation in adult T-ALL (n=24) compared to normal thymi (n=3). We identified a CpG hypermethylator phenotype that distinguishes two T-ALL subgroups and further validated it in an independent series of 17 T-lymphoblastic lymphoma. Next, we identified a methylation classifier based on nine promoters which accurately predict the methylation phenotype. This classifier was applied to an independent series of 168 primary adult T-ALL treated accordingly to the GRAALL03/05 trial using methylation-specific multiplex ligation-dependent probe amplification. Importantly hypomethylation correlated with specific oncogenic subtypes of T-ALL and identified patients associated with a poor clinical outcome. This methylation-specific multiplex ligation-dependent probe amplification based methylation profiling could be useful for therapeutic stratification of adult T-ALL in routine practice. The GRAALL-2003 and -2005 studies were registered at http://www.clinicaltrials.gov as #NCT00222027 and #NCT00327678, respectively.
T-cell acute lymphoblastic leukemias (T-ALL) are aggressive and heterogeneous malignancies which are predominated by the 10-39-year age group where they account for 20% of acute lymphoblastic leukemias (ALL).1 T-ALL is associated with a wide range of acquired genetic abnormalities that contribute to developmental arrest and abnormal proliferation of malignant lymphoid progenitors.32 Despite the diversity of observed mutations and deletions, genome wide expression64 assays led to the identification of few oncogenic T-ALL subgroups, namely the immature/early thymic precursor (ETP) (LyL1, MEF2C), late cortical (TAL1), early cortical (TLX1/3 and NKX2.1) and HOXA clusters. Although cancer is typically considered a genetic disease, epigenetic aberrations also play important roles in tumor potentiation, initiation, and progression.7 Epigenetics is defined as changes in gene expression that are not due to changes in gene sequence, and include DNA methylation, histone modifications, microRNA (miRNA) and nucleosome positioning. Unlike genetic alterations, epigenetic changes are reversible by enzymatic activity and pharmacological treatment with small molecule inhibitors, like those targeting enzymes involved in DNA methylation or chromatin modifications. Altered epigenetic states are a common feature of all cancer types and the most studied epigenetic modification in primary cancer samples is DNA methylation, which is known to display characteristic changes in malignant cells compared to normal tissue. These include diffuse hypomethylation and focal hypermethylation changes at discrete loci potentially associated with repression of specific genes related to cancer pathogenesis.
In the field of ALL, DNA methylation studies have mostly focused on pediatric B-cell precursor ALL (BCP-ALL) describing promoter hypermethylation and specific methylation signatures according to the cytogenetic subgroup.8 In pediatric T-ALL, DNA methylation was analyzed by Infinium 27 K and 450 K arrays and two distinct CpG island methylator phenotype (CIMP) groups were identified. Patients with a CIMP-negative profile displayed a significantly higher cumulative incidence of relapse (CIR) compared to CIMP-positive patients suggesting a prognostic relevance of aberrant DNA methylation profiles in T-ALL.109 Furthermore, it has more recently been shown in a pediatric series that CIMP status correlates with known oncogenic subgroups, for instance, with higher expression of TAL1 in a CIMP-negative subgroup (11). However such data for adult T-ALL are still lacking. In this work, we report genome-wide promoter methylation profiling by methylation-dependent immunoprecipitation (MeDIP) in a cohort of adult T-ALL. Subsequently, a nine-promoter classifier was applied to a large series of 168 adult T-ALL included in the GRAALL 03/05 trial that distinguished two subgroups with highly significant differences in the clinical outcome. Thus, MeDIP profiling is a potential candidate for risk stratification of adult T-ALL and could provide important information in treatment decision making and therapeutic targeting.
Patients and treatments
Adult patients (15-60 years old) included in two successive French ALL cooperative group trials (GRAALL-2003 and GRAALL-2005) with T-ALL, and defined according to the 2008 World Health Organisation classification, were analyzed. The GRAALL-2003 protocol was a multicenter phase 2 trial, which enrolled 76 adults with T-ALL between November 2003 and November 2005 of whom 50 had sufficient diagnostic tumor material available.12 The multicenter randomized GRAALL-2005 phase 3 trial was very similar to the GRAALL-2003 trial, with the addition of a randomized evaluation of an intensified sequence of hyperfractionated cyclophosphamide during induction and late intensification.13 Between May 2006 and May 2010, 337 adults with T-ALL were randomized in the GRAALL-2005, of which 185 had available diagnostic material. All samples contained >80% blasts. Phenotypic and oncogenetic characteristics were as described.1614 Informed consent was obtained from all patients at enrollment. All trials were conducted in accordance with the Declaration of Helsinki and approved by local and multicenter research ethical committees.
Global DNA methylation was assessed by a MeDIP assay on an initial series of 24 T-ALL and three human thymi and a second (confirmatory series) of 17 T-lymphoblastic lymphomas (T-LBL) and three human thymi. Briefly, methylated DNA was immunoprecipitated as described previously17 using 2 μg of sonicated genomic DNA. MeDIP samples were directly subjected to labeling and hybridization to previously described custom human promoter arrays (Agilent, Santa Clara, CA, USA) covering either 17,970 promoters17 (T-ALL series) or 25,490 promoters18 (T-LBL series), following the manufacturer’s instructions. The median-normalized log2 enrichment ratios (MeDIP/Input) were calculated for each probe using the CoCAS software19 and visualized using the IGB tool (http://bioviz.org/igb). Finally, a methylation score was computed for each promoter by calculating the median enrichment ratio of overlapping probes. A summary of the methylation scores per promoter in T-ALL and T-LBL samples is provided in the Online Supplementary Tables S1-2, respectively.
Clustering of methylation profiles
Hierarchical clustering analysis (Average Linkage) based on the methylation signal of the top 5% genes with highest variance was performed with the TIGR MeV v. 4.9.0 program,20 using the -1 Spearman rank correlation method. Analysis of the differential methylation signal between the groups was performed using the significant analysis of microarrays (SAM) algorithm (threshold value: FDR<0.121 and delta=2.144). The graphical clustering representation of the clustering was done with the GenePattern software.21 The list of differentially methylated promoters in T-ALL and T-LBL is provided in Online Supplementary Table S3.
Validation of DNA methylation signature
Direct methylation levels were analyzed by methylation-specific multiplex ligation-dependent probe amplification (MS-MLPA) with custom probes (Online Supplementary Table S4) and, SALSA MLPA P200 Reference-1 probemix and EK1 reagent kits from MRC-Holland (Amsterdam, the Netherlands), according to manufacturer’s recommendations. Data were analyzed with the Coffalyser software (MRC-Holland, Amsterdam, the Netherlands). In addition, the promoter methylation patterns were verified by quantitative PCR (qPCR) analysis of MeDIP samples and by bisulfite sequencing using specific primers for the MEIS1 gene promoter.
DNA methylation signatures in T-ALL/T-LBL
Global promoter regions DNA methylation by MeDIP-array was performed in a training series of 24 adult T-ALL. Unsupervised hierarchical clustering defined two major groups (group 1 and group 2) with distinct methylation profiles (Figure 1A). The supervised signature of differential methylation (FDR<0.121) between these two groups resulted in 300 unique differentially methylated gene promoters with a vast majority of hypermethylated (hyperM) promoters (297 of 300) in the so-called hyperM group. The second group displayed an intermediate methylation (interM) profile compared to the normal thymic tissue (Figure 1B and Online Supplementary Figure S1). Interestingly, all the TLX cases without exception (including six TLX1 and two TLX3 cases) clustered in the hyperM group. Conversely, the two SIL-TAL1 cases belonged to the interM group; suggesting a role of oncogenetic abnormalities in the observed methylation profiles.
A very similar differential methylation signature (Figure 2A-B) was observed in an independent series of 17 T-LBL. One TLX1+ and five TLX3 T-LBL, as in T-ALL, were clustered in the group with a hyperM promoter profile (253 of 255 hyperM gene promoters). T-ALL and T-LBL promoter methylation signatures displayed a highly significant overlap (P<0.0001) with 97 common gene promoters differentially methylated (Figure 2C and Online Supplementary Table S2). Among them, the differential methylation of MEIS1 promoter was confirmed with two different targeted methods, MeDIP-QPCR (Online Supplementary Figure S2A) and bisulfite sequencing (Online Supplementary Figure S2B).
Driver oncogenes defined distinct aberrant methylation profiles
In an effort to explore the DNA methylation profiles in a larger T-ALL series, a minimal robust signature able to predict the methylation state was defined with a remaining error risk inferior to 0.05. This predictor contained the following nine gene promoters: BMP4, HOXB7, KCNA1, LHX1, MEIS1, PROX1, PSD3, RUNX2, SEMA6A (Figure 3A). A MS-MLPA panel was designed to explore the methylation status of these nine gene promoters and a methylation ratio corresponding to the methylation average of these nine differentially methylated regions (DMR) was calculated. As expected, this predictor allowed the separation of hyperM and interM T-ALL from the training cohort (P=0.0016) (Figure 3B-C). We then performed this analysis on a series of 168 primary adult T-ALL uniformly treated according to the GRAALL03-05 trial (Figure 3D). The methylation ratio was widely ranged (mean=0.62, min=0.04, max=1.1) and major oncogenetic drivers (TLX1, TLX3, SIL-TAL1, HOXA overexpression) defined distinct methylation profiles. TLX1 and TLX3 T-ALL displayed significantly hyperM promoters compared to the HOXA subgroup (P=0.03 and P=0.02 respectively), to the SIL-TAL1 subgroup (P<0.0001) or the others T-ALL (P<0.0001). Conversely, SIL-TAL1 cases and others T-ALL expressing high level of TAL1 had significantly hypomethylated (hypoM) promoters (P<0.0001) compared to TLX1/3, HOXA or others T-ALL. Unlike oncogenetic status, immature early thymic progenitor-ALL (ETP-ALL) lacked a significant distinct methylation signature compared to non-ETP-ALL (Figure 3E).
Low level of promoter methylation predicted a poor outcome subgroup of adult T-ALL
T-ALL patients with the lowest methylation level (Q1, n=42/168) were significantly more men, were younger, and had a higher white blood cell (WBC) count at diagnosis than patients with higher methylation levels (Table 1). Moreover, hypoM T-ALL demonstrated a significantly more frequent mature phenotype (TCRαβ) and were associated with SIL-TAL1 rearrangement. They were also significantly associated with a low rate of NOTCH1 pathway mutations and a high risk NOTCH1/FBXW7/RAS/ PTEN molecular classifier.22 In detail, we observed a significantly lower incidence of NOTCH1/FBXW7 mutations and also a greater incidence of PTEN alterations (mutation and/or deletion) in the hypoM subgroup (CIMP-neg) as compared to the Int/High methylated cases (Online Supplementary Table S5). Despite a better bone marrow response at D8 (M1 status) in patients with low methylation, we did not observe any impact of methylation on complete remission (CR) rate or post-induction minimal residual disease (MRD) level. In univariate analysis, patients with low methylation levels had higher CIR (sub-distribution hazard ratio [SHR] 1.87, 95% CI: 1.03-3.38, P=0.04; Table 2 and Figure 4A) and a shorter overall survival (OS) (hazard ratio [HR] 1.78, 95% CI: 1.06-2.98, P=0.03; Table 2 and Figure 4B). In multivariate analysis for CIR, the only prognostic factor to be significantly associated with a reduced CIR was the NOTCH1/FBXW7/RAS/PTEN molecular classifier. However, in multivariate analysis for OS, including age, WBC at diagnosis, central nervous system (CNS) involvement, prednisone response, the molecular classifier, and the methylation level as covariates, a low methylation was still independently associated with a higher risk of death (HR 1.79, 95% CI: 1.00-3.19, P=0.05; Table 2).
Despite recent insights into the molecular and cellular mechanisms responsible for T-ALL onset and progression, survival rates remain around 50% in adults, justifying the search for novel therapeutic options or more adapted/personalized regimens. The present study focused on promoter DNA methylation in a large series of adult T-ALL. As previously reported in pediatric T-ALL,119 we showed that DNA methylation status is also a prognostic factor in adult T-ALL. Similarly, patients with a hypoM profile dis play an unfavorable outcome compared to hyperM patients. Importantly, even if hypoM status is associated with the molecular high-risk classifier,22 methylation level remains an independent prognostic factor. Moreover, methylation status does not seem to influence the initial clinical response to therapy since there were no significant differences regarding the glucosteroids and initial chemotherapy responses (chemosensitivity or MRD) between hypoM and hyperM patients. Methylation status could therefore represent a relevant additional prognostic factor for adult T-ALL. Nevertheless, further validation by another independent series is needed. Moreover, it would be interesting to study the prognostic impact of this methylation signature in T-LBL, which displayed similar methylation distortion patterns.
We used the relatively new methodology of methylation specific-multiplex ligation-dependent probe amplification (MS-MLPA) to evaluate the promoter DNA methylation level. MS-MLPA is a powerful and easy-to-perform PCR-based technique and we demonstrated that MS-MLPA could provide an attractive alternative way to assess methylation classification compared to array analysis. This approach permits methylation analysis of multiple targets in a single experiment and has been successfully used to evaluate the diagnostic relevance of different markers in several tumor types including lung,23 rectal,24 breast,25 bladder,26 prostate,27 and adrenocortical cancer.28 Additionally, MS-MLPA has the advantage of requiring little DNA and does not require DNA bisulfite conversion or immunoprecipitation. MS-MLPA is readily compatible with clinical routine and could enhance prognostication and precision medicine.
However, array analysis or methylation analysis at the whole genome level would be relevant in T-ALL to gain information and investigate how aberrant methylation patterns are involved in leukemogenesis. We have observed that aberrant methylation profiles were mostly associated with the driver oncogene involved. In particular, a hypoM subgroup with unfavorable outcome is mainly enriched in SIL-TAL1+ cases and also in cases negative for the main oncogenes TLX1, TLX3, SIL-TAL1 and HOXA. Deciphering the molecular mechanism of aberrant methylation and the relationship with driver oncogenes could identify new deregulated pathways for adapted-therapy.
We thank all French, Swiss and Belgian participants, clinicians, biologists and clinical research assistants, in the LALA-94 and GRAALL 2003-2005 trials, for collecting and providing data and samples.
- Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/106/6/1575
- FundingMeDIP experiments were supported by the INCA-project PAIR-Lymphomes. Work in SS laboratory was supported by Ligue contre le Cancer (Equipe labellisée). Work in the Necker hematology department and SS laboratory was supported by grants from ITMO (Institut Thématique Multi-Organisme) épigénétique et cancer. The GRAALL was supported by grants P0200701 and P030425/AOM03081 from Le Programme Hospitalier de Recherche Clinique, Ministère de l’Emploi et de la Solidarité, France and the Swiss Federal Government in Switzerland. AT was supported by a grant from the French program CARPEM (CAncer Research for PErsonalized Medicine).
- Received April 4, 2019.
- Accepted September 19, 2019.
- Chessells JM, Hall E, Prentice HG. The impact of age on outcome in lymphoblastic leukaemia; MRC UKALL X and XA compared: a report from the MRC Paediatric and Adult Working Parties. Leukemia. 1998; 12(4):463-473. PubMedhttps://doi.org/10.1038/sj.leu.2400959Google Scholar
- Teitell MA, Pandolfi PP. Molecular genetics of acute lymphoblastic leukemia. Ann Rev Pathol. 2009; 4:175-198. Google Scholar
- Van Vlierberghe P, Pieters R, Beverloo HB, Meijerink JP. Moleculargenetic insights in paediatric T-cell acute lymphoblastic leukaemia. Br J Haematol. 2008; 143(2):153-168. PubMedhttps://doi.org/10.1111/j.1365-2141.2008.07314.xGoogle Scholar
- Ferrando AA, Neuberg DS, Staunton J. Gene expression signatures define novel oncogenic pathways in T cell acute lymphoblastic leukemia. Cancer Cell. 2002; 1(1):75-87. PubMedhttps://doi.org/10.1016/S1535-6108(02)00018-1Google Scholar
- Homminga I, Pieters R, Langerak AW. Integrated transcript and genome analyses reveal NKX2-1 and MEF2C as potential oncogenes in T cell acute lymphoblastic leukemia. Cancer Cell. 2011; 19(4):484-497. PubMedhttps://doi.org/10.1016/j.ccr.2011.02.008Google Scholar
- Soulier J, Clappier E, Cayuela JM. HOXA genes are included in genetic and biologic networks defining human acute T-cell leukemia (T-ALL). Blood. 2005; 106(1):274-286. PubMedhttps://doi.org/10.1182/blood-2004-10-3900Google Scholar
- Flavahan WA, Gaskell E, Bernstein BE. Epigenetic plasticity and the hallmarks of cancer. Science. 2017; 357(6348)Google Scholar
- Nordlund J, Backlin CL, Wahlberg P. Genome-wide signatures of differential DNA methylation in pediatric acute lymphoblastic leukemia. Genome Biol. 2013; 14(9):r105. PubMedhttps://doi.org/10.1186/gb-2013-14-9-r105Google Scholar
- Borssen M, Haider Z, Landfors M. DNA methylation adds prognostic value to minimal residual disease status in pediatric T-cell acute lymphoblastic leukemia. Pediatr Blood Cancer. 2016; 63(7):1185-1192. Google Scholar
- Borssen M, Palmqvist L, Karrman K. Promoter DNA methylation pattern identi fies prognostic subgroups in childhood T-cell acute lymphoblastic leukemia. PLoS One. 2013; 8(6):e65373. PubMedhttps://doi.org/10.1371/journal.pone.0065373Google Scholar
- Haider Z, Larsson P, Landfors M. An integrated transcriptome analysis in T-cell acute lymphoblastic leukemia links DNA methylation subgroups to dysregulated TAL1 and ANTP homeobox gene expression. Cancer Med. 2019; 8(1):311-324. Google Scholar
- Huguet F, Leguay T, Raffoux E. Pediatric-inspired therapy in adults with Philadelphia chromosome-negative acute lymphoblastic leukemia: the GRAALL-2003 study. J Clin Oncol. 2009; 27(6):911-918. PubMedhttps://doi.org/10.1200/JCO.2008.18.6916Google Scholar
- Maury S, Chevret S, Thomas X. Rituximab in B-lineage adult acute lymphoblastic leukemia. N Engl J Med. 2016; 375(11):1044-1053. PubMedGoogle Scholar
- Asnafi V, Buzyn A, Le Noir S. NOTCH1/FBXW7 mutation identifies a large subgroup with favorable outcome in adult T-cell acute lymphoblastic leukemia (T-ALL): a Group for Research on Adult Acute Lymphoblastic Leukemia (GRAALL) study. Blood. 2009; 113(17):3918-3924. PubMedhttps://doi.org/10.1182/blood-2008-10-184069Google Scholar
- Bergeron J, Clappier E, Radford I. Prognostic and oncogenic relevance of TLX1/HOX11 expression level in T-ALLs. Blood. 2007; 110(7):2324-2330. PubMedhttps://doi.org/10.1182/blood-2007-04-079988Google Scholar
- Bond J, Marchand T, Touzart A. An early thymic precursor phenotype predicts outcome exclusively in HOXA-overexpressing adult T-cell acute lymphoblastic leukemia: a Group for Research in Adult Acute Lymphoblastic Leukemia study. Haematologica. 2016; 101(6):732-740. PubMedhttps://doi.org/10.3324/haematol.2015.141218Google Scholar
- Jia J, Pekowska A, Jaeger S. Assessing the efficiency and significance of Methylated DNA Immunoprecipitation (MeDIP) assays in using in vitro methylated genomic DNA. BMC Res Notes. 2010; 3:240. PubMedhttps://doi.org/10.1186/1756-0500-3-240Google Scholar
- Cornen S, Guille A, Adelaide J. Candidate luminal B breast cancer genes identified by genome, gene expression and DNA methylation profiling. PLoS One. 2014; 9(1):e81843. PubMedhttps://doi.org/10.1371/journal.pone.0081843Google Scholar
- Benoukraf T, Cauchy P, Fenouil R. CoCAS: a ChIP-on-chip analysis suite. Bioinformatics. 2009; 25(7):954-955. PubMedhttps://doi.org/10.1093/bioinformatics/btp075Google Scholar
- Saeed AI, Sharov V, White J. TM4: a free, open-source system for microarray data management and analysis. Biotechniques. 2003; 34(2):374-378. PubMedhttps://doi.org/10.1016/S0076-6879(06)11009-5Google Scholar
- Reich M, Tabor T, Liefeld T. The GenePattern Notebook environment. Cell Syst. 2017; 5(2):149-151. Google Scholar
- Trinquand A, Tanguy-Schmidt A, Ben Abdelali R. Toward a NOTCH1/FBXW7/RAS/PTEN-based oncogenetic risk classification of adult T-cell acute lymphoblastic leukemia: a group for research in adult acute lymphoblastic leukemia study. J Clin Oncol. 2013; 31(34):4333-4342. PubMedhttps://doi.org/10.1200/JCO.2012.48.5292Google Scholar
- Castro M, Grau L, Puerta P. Multiplexed methylation profiles of tumor suppressor genes and clinical outcome in lung cancer. J Transl Med. 2010; 8:86. PubMedhttps://doi.org/10.1186/1479-5876-8-86Google Scholar
- Molinari C, Casadio V, Foca F. Gene methylation in rectal cancer: predictive marker of response to chemoradiotherapy¿. J Cell Physiol. 2013; 228(12):2343-2349. Google Scholar
- Moelans CB, de Groot JS, Pan X, van der Wall E, van Diest PJ. Clonal intratumor heterogeneity of promoter hypermethylation in breast cancer by MS-MLPA. Mod Pathol. 2014; 27(6):869-874. PubMedhttps://doi.org/10.1038/modpathol.2013.207Google Scholar
- Garcia-Baquero R, Puerta P, Beltran M. Methylation of tumor suppressor genes in a novel panel predicts clinical outcome in paraffin-embedded bladder tumors. Tumour Biol. 2014; 35(6):5777-5786. Google Scholar
- Gurioli G, Salvi S, Martignano F. Methylation pattern analysis in prostate cancer tissue: identification of biomarkers using an MS-MLPA approach. J Transl Med. 2016; 14(1):249. Google Scholar
- Jouinot A, Assie G, Libe R. DNA Methylation is an independent prognostic marker of survival in adrenocortical cancer. J Clin Endocrinol Metab. 2017; 102(3):923-932. Google Scholar