Frequency and prognostic impact of CEBPA proximal, distal and core promoter methylation in normal karyotype AML: a study on 623 cases.

The clinical impact of aberrant CEBPA promoter methylation (PM) in AML is controversially discussed. The aim of this study was to clarify the significance of aberrant CEBPA PM with regard to clinical features in a cohort of 623 cytogenetically normal (CN) de novo AML. 555 cases had wild-type CEBPA, 68 cases harbored CEBPA mutations. The distal promoter was methylated in 238/623 cases (38.2%), the core promoter in 8 of 326 cases (2.5%), whereas proximal PM was never detected. CEBPA PM and CEBPA mutations were mutually exclusive. CEBPA distal PM positive cases were characterized by reduced CEBPA mRNA expression levels and elevated white blood cell counts. CEBPA distal PM was less frequent in patients with mutations in FLT3, NPM1 and TET2 and more frequent in cases with RUNX1 and IDH2R140 mutations. Overall, no association of methylation to prognosis was seen. However CEBPA distal PM was associated with inferior outcome in cases with low FLT3-ITD ratio or TET2 mutations. A distinct gene expression profile of CEBPA distal PM positive cases compared to CEBPA mutated and CEBPA distal PM negative cases was observed. In conclusion, the presence of aberrant CEBPA PM is associated with distinct biological features but impact on outcome is weak.

Introduction

The CCAAT/enhancer binding protein α (CEBPA) is a transcription factor with critical roles in tissue specific gene expression and proliferation arrest. In the hematopoietic system, CEBPA expression is restricted to myelomonocytic cells and is specifically up-regulated during granulocyte differentiation [1]. Loss of CEBPA function is known to result in a block of granulopoiesis. CEBPA has gained interest in the AML field, as it has been shown that down-regulation of CEBPA protein through mutations, posttranscriptional modifications and protein-protein interactions with fusion proteins such as RUNX1-RUNX1T1 or CBFB-MYH11 plays a key role in leukemic transformation.

Mutations in the CEBPA gene have been described for approximately 5–10% of all AML patients and are most common in CN-AML (15%) [2], [3]. In addition to genetic mutations, in recent years, epigenetic modifications, such as DNA promoter hypermethylation have gained increasing interest as additional mechanisms for transcriptional regulation of cancer-related genes. Hence, inactivation of gene expression by abnormal hypermethylation of CpG islands in promoter regions of tumor suppressor genes has been described for many cancer entities [4].

Studies of the CEBPA promoter revealed three regions important for promoter function. The core promoter region (−141 to +103 upstream from transcription start site) contains the TATA box and several regulatory factors necessary for CEBPA expression [5]. The upstream promoter region (−1422 to −896 upstream from transcription start site) has been shown to interact with MBD2 and MeCP2 methyl-CpG binding proteins and contains binding sites for the transcriptional factors USF−1/−2 and Sp1 suggesting that methylation decreases the cis-activity of these factors, leading to lower CEBPA expression. According to methylation levels, the upstream promoter region can be divided into a highly methylated distal region (−1422 to −1121 upstream from transcription start site) and a lowly methylated proximal region (−1121 to −896 upstream from transcription start site) [6].

Recent reports have shown that epigenetic modification of the distal CEBPA promoter region resulted in the down regulation of CEBPA expression in lung cancer [6], head and neck squamous cell carcinoma [7] and pancreatic cancer cells [8]. Additionally, several studies document epigenetic modification of CEBPA in AML. Chim et al. [9] found aberrant methylation in the CEBPA core promoter (−141 to −15 from transcription start site) in 2/70 unselected AML patients (2.8%). Wouters et al. [10] found a correlation between silenced CEBPA and frequent CEBPA core promoter hypermethylation in six of 285 patients with de novo AML (1.4%). Hackanson et al. [11] have observed methylation in the distal CEBPA promoter region in 20 of 39 (51%) AML patients carrying the recurrent cytogenetic aberrations inv(16), t(8;21), t(15;17), t(9;11) or complex karyotype. Lin et al. [12] evaluated the methylation status of the CEBPA core, proximal and distal promoters in a total cohort of 193 unselected patients with de novo AML. They found heterogeneous methylation in the distal promoter region, but not in the proximal or core promoter regions. In the total cohort of 193 patients, high CEBPA PM was correlated with better treatment response and in a subcohort of 25 CN-AML patients without CEBPA and NPM1 mutations, cases with high CEBPA PM had longer overall survival (OS) compared to cases with low CEBPA PM. Due to the differences in these studies with respect to selected patient cohorts and the examined CEBPA promoter regions, the clinical implications of CEBPA methylation in AML remain unclear.

In the present study, we analyzed the methylation status of the CEBPA promoter region including core, proximal and distal promoters in 555 de novo CN-AML with wt CEBPA to clarify the frequency and the significance of aberrant CEBPA PM with regard to clinical features. To exclude coincidence of CEBPA PM with CEBPA mutations, we also analyzed 68 CN-AML cases with CEBPA mutations for CEBPA PM. In addition, we addressed the question, whether CEBPA PM positive AML constitutes a distinct entity in CN-AML. Therefore, we performed global gene expression profiling (GEP) comparing CEBPA PM positive cases to CEBPA mutated and CEBPA unmutated/unmethylated cases.

Materials and Methods

Patients

We analyzed a total cohort of 623 de novo AML patients that were referred to our laboratory for first diagnosis of AML between August 2005 and October 2010 (Table 1). AML was diagnosed according to the FAB and WHO classifications [13], [14]. 294 patients were female, 329 male and the median age was 63.9 years (range 20.0–89.6 years). Bone marrow blast percentages ranged from 20 to 99% (median: 67.5%) in 604 patients with non-M6 AML. 19 Patients with AML M6 subtype had blast percentages below 20% (3–17%, median: 12%), as characteristic for the AML M6 subtype. 555 of the 623 cases had CN-AML and wild-type CEBPA. Data on other molecular markers was available in: NPM1: n = 551, FLT3-ITD: n = 552, FLT3-TKD: n = 447, MLL-PTD: n = 552, RUNX1: n = 467, ASXL1: n = 420, IDH1R132: n = 382, IDH2R140: n = 344, and IDH2R172: n = 345, TET2: n = 113 and DNMT3A: n = 119, respectively. For comparison 68 patients with CEPBA mutations (38 monoallelic, 20 biallelic, 10 homozygous) were analyzed in addition. Clinical follow up data was available in 435 patients. Patients received standard induction and consolidation chemotherapies such as “7+3”, TAD or HAM. All patients gave written informed consent for scientific studies, e.g. molecular analyses. The study was approved by the Internal Review Board and adhered to the tenets of the Declaration of Helsinki.

Table 1

Patient cohorts.

	Cases	Male	Female	Median age, years
	n = 623	n = 329	n = 294	(range)
1. CEBPA wt	555	293	262	63.9 (20.0–89.6)
2. CEBPA mut	68	36	32	64.7 (15.7–84.6)
monoallelic	38	22	16	65.1 (22.3–74.3)
biallelic	20	9	11	58.9 (15.7–84.6)
homozygous	10	5	5	62.2 (28.5–78.3)

wt: wild-type; mut: mutated.

Cytomorphology, Cytogenetics, Immunophenotyping

Cytomorphologic assessment was based on May-Grünwald-Giemsa stains, myeloperoxidase reaction, and non-specific esterase using alpha-naphtyl-acetate as described before and was performed according to the criteria defined in the FAB and the WHO classifications [14]–[16]. Cytomorphology was performed in our laboratory in 526/555 of the CEBPAwt cases. In addition, 4 cases were identified as AML by immunophenotyping according to blast cell counts. 25 cases were defined as AML by the sent diagnostic report of the clinical centers. Cytogenetic studies were performed after short-term culture. Karyotypes, analyzed after G-banding, were described according to the International System for Human Cytogenetic nomenclature [17]. Prognostic classification into “favorable”, “intermediate” and “adverse” groups was performed according to the refined MRC classification [18]. Cytogenetic results were available for all patients in the study. Immunophenotyping was performed in our laboratory in 284 cases as described previously [19].

Isolation and Bisulfite Treatment of Nucleic Acids

DNA was extracted according to a standard procedure from fresh bone marrow or peripheral blood cells after Ficoll separation of mononucleated cells. Bisulfite treatment of genomic DNA was performed using the DNA Methylation Gold Kit (Zymo Research, Orange, CA, USA). Bisulfite treated DNA was used in subsequent DNA methylation analyses which were performed either by methylation-specific polymerase chain reaction (MSP) or bisulfite sequencing.

Methylation-specific Polymerase Chain Reaction (MSP) and Bisulfite Sequencing

Primers used for methylation-specific PCR, bisulfite sequencing and PCR conditions are summarized in Table S1 and were described previously [12], [20]. Locations of individual primers are shown in Figure 1. The PCR products for bisulfite sequencing were purified using the Sephadex® PCR purification system (Sigma-Aldrich, St. Louis, MO, USA), and the methylation patterns were determined using the BigDye Terminator v3.1 Cycle Sequencing kit (Applied Biosystems, Life Technologies, Darmstadt, Germany) on an automated ABI 3730 Genetic Analyzer (Applied Biosystems, Life Technologies, Darmstadt, Germany). Bisulfite treated CpGenome universal methylated DNA (Chemicon, Temecula, CA, USA) was used as a methylation-positive control. We analyzed 24 individual CpG dinucleotides in the distal promoter region (−1423 to −1121 upstream from transcription start site) and 20 individual CpG dinucleotides in the proximal promoter region (−1121 to −896 upstream from transcription start site). To evaluate methylation ratio, we counted the amount of cytosines with methylation and the level of methylation for each cytosine by Sanger Sequencing analysis [21]. The percentage of methylation was calculated as the peak height of cytosine vs the peak height of thymine for each CpG site. A single cytosine at the corresponding CpG site was considered as 100% methylation, a single thymine as no methylation and overlapping cytosine and thymine as partial methylation (15–100%). Cytosines were counted as methylated if the methylation intensity was 15% or higher. (Figure 1B/Table S2). Because the PCR products were not cloned, this analysis represents an approximation of the “average” methylation status at each CpG residue for each patient.

Figure 1

DNA methylation analysis of the CEBPA core promoter and upstream promoter regions.

A) Scheme of the CEBPA promoter region. The areas of DNA methylation analysis are shown below. Red: CEBPA-promoter outer primers; grey: CEBPA distal PM primers, black: CEBPA proximal PM primers; blue: CEBPA-U and CEBPA-M primers; BS: Bisulfite Sequencing; MSP: Methylation specific PCR. B) Bisulfite sequencing results for the distal CEBPA promoter region for 3 individual cases compared to positive control. Boxes indicate individual CpG sites analyzed. CG indicates a methylated CpG site; TG indicates an unmethylated CpG site. The distal CEBPA promoter region of patient 1 is fully methylated (level of methylation: 75–100%), patient 2 is partly methylated (level of methylation: 20–50%) and patient 3 is unmethylated (level of methylation: <15%). C) MSP for CEBPA core promoter with CEBPA-U and CEBPA-M primers. Upper: N: Non-template control; UM1/UM2: positive controls with unmethylated DNA; M: positive control with methylated DNA. 1–4 samples of patients showing positive amplification in UMS-PCR but not in MS-PCR. Lower: 5–8: samples of patients showing positive amplification in MS-PCR but not in UMS-PCR; D) Frequency and distribution of methylated CpG islands within the distal CEBPA promoter region. Bar chart showing the frequency and distribution of methylated CpG islands within the distal CEBPA promoter region for 238 cases with CEBPA distal promoter methylation.

RNA Isolation, Reverse Transcription and Quantitative Real-time PCR

Isolation of mononucleated cells, mRNA extraction, and random primed cDNA synthesis was performed as described previously [22]. RQ-PCR was performed by the use of the Applied Biosystems 7500 Fast Real Time PCR System with the application of specific CEBPA TaqMan Gene Expression Assay (Assay ID: HS00269972_S1). Amplification was performed after initial incubation at 95°C for 1 minute in a 3-step cycle procedure (denaturation 95°C, 20 seconds, ramp rate 20°C/s, annealing temperature 60°C, 45 seconds, ramp rate 20°C/s, and extension 72°C, 26 seconds, ramp rate 2°C/s) for 50 cycles. The expression of CEBPA was normalized against the expression of the control gene ABL1 to adjust for variations in mRNA quality and efficiencies of cDNA synthesis. The CEBPA expression levels are given as: %CEBPA/ABL1.

Global Gene Expression Profiling

To detect underlying common differences in their gene expression profiles (GEP) we investigated 9 CEBPA methylated (unmutated), 8 CEBPA single-mutated, 10 CEBPA double-mutated, and 10 non-methylated/non-mutated cases (Affymetrix HG-U133 Plus 2.0 microarrays; Santa Clara, CA). All cases analyzed were taken from the subcohorts described above. The microarray sample preparation assay was performed as previously reported [23]. Gene expression raw data was processed according to the manufacturer’s recommendations. After quality control, raw data was normalized for visualization and interpretation using the Robust Multichip Average (RMA) algorithm as implemented in the R-package affy version 1.18.0. Detection calls, i.e. present, marginal, or absent expression, were determined by default parameters. Probe sets were filtered out by the genefilter package. Probe intensities were considered if the normalized signal was above 100 (unlogged data). Significantly regulated genes were determined using the LIMMA toolbox. A gene was determined as significantly regulated if the p-value was <0.05 after multiple testing correction by Benjamini-Hochberg procedure [24]. Clustering of expression data was performed using Manhattan distance function and complete clustering. To find significant associated biological processes, we performed a Gene Ontology term enrichment analysis, which was carried out with the R package GOstats [PMID: 15461798]. Raw data is available at GEO with accession number GSE34733.

Statistical Analyses

Survival curves were calculated for overall survival (OS), event free survival (EFS) and relapse free survival (RFS) according to Kaplan-Meier and compared using the two-sided log rank test. OS was the time from diagnosis to death or last follow-up. EFS was the time from diagnosis to treatment failure, relapse, death, or last follow-up in complete remission. RFS was the time from achievement of complete remission to relapse, death, or last follow-up in complete remission. Complete remission and relapse were defined according to Cheson et al. [25]. Cox regression analysis was performed for OS and EFS with different parameters as covariates. Median follow-up was calculated taking into account the respective last observations in surviving cases and censoring non-surviving cases at the time of death. Parameters which were significant in univariable analyses were included into multivariable analyses. Dichotomous variables were compared between different groups using the χ2-test and continuous variables by Student’s T-test and Spearman’s rank correlation. Results were considered significant at p≤0.05. All reported p-values are two-sided. No adjustments for multiple comparisons were performed. SPSS version 19.0 (IBM Corporation, Armonk, NY) was used for statistical analysis.

Results

Distribution of CEBPA Promoter Methylation

We evaluated the CEBPA promoter methylation status in a total cohort of 623 patients with de novo CN-AML using methylation-specific PCR and bisulfite sequencing methods. This cohort comprised of 555 cases with CEBPA wild-type (wt) and 68 CEBPA mutated (mut) status.

Methylation specific PCR analysis revealed CEBPA core PM in only 8 of the first 326 cases analyzed (2.5%) (Figure 1C/Table 2). Because of this low frequency we did not continue with this analysis. The total cohort of 623 cases was subsequently analysed using semiquantitative bisulfite sequencing. In the cohort of 555 CN-AML cases with CEBPAwt we identified 238 of 555 cases (42.9%) with methylated CpG sites in the distal promoter region (CEBPA dPMpos) (Table S2). The amount of methylated cytosines ranged from 2 to 24 and the methylation levels ranged from 15% to 100% compared to positive control (Figure 1 B). The cytosines of the first 11 CpG sites of the distal CEBPA promoter were more often methylated than the 13 cytosines of the C-terminal CpG sites (Figure 1D). We next defined a threshold among the 238 CEBPA dPMpos cases by forming a ratio of the methylation intensity and the amount of methylated cytosines (see Materials and Methods). Mean methylation ratio was 541. Cases with a ratio less than 541 were defined as lowly methylated (CEBPA dPMlow), cases with a ratio higher than 541 as highly methylated (CEBPA dPMhigh). According to this, 144/238 (60.5%) cases were CEBPA dPMlow and 94/238 (39.5%) cases were CEBPA dPMhigh. Methylation of the 20 individual CpG dinucleotides in the proximal promoter was not detected (Table 2).

Table 2

Frequency and distribution of CEBPA promoter methylation.

	Cases	CEBPA	CEBPA	CEBPA
		Core	Proximal	Distal
	n =	Promoter	Promoter	Promoter
CEBPA wt	555	8/326 (2.5%)	0/572	238/555 (42.9%)
CEBPA mut	68
monoallelic	38	0	0	0
Biallelic	20	0	0	0
homozygous	10	0	0	0

wt: wild-type; mut: mutated.

One single patient with AML M0 subtype carried methylation throughout both the distal and the core CEBPA promoter.

None of the 68 CEBPA mutated cases harbored methylation in any promoter region analyzed and thus aberrant CEBPA PM and mutation status were mutually exclusive (Table 2).

Cytomorphology and Immunophenotyping

According to the FAB classification system, of the 526/555 CEBPAwt cases with cytomorpholocical data 28 were AML M0, 162 AML M1, 183 AML M2, 114 AML M4, 13 AML M5, 17 AML M6 and one AML M7 in FAB subgroups, respectively. In 8 cases FAB classification was not possible. 284 cases were analyzed by immunophenotyping in addition. Cases with CEBPA distal PM as compared to those without revealed a more immature phenotype with stronger expression of CD34 (mean positive cells 33±29% vs. 27±26%, p = 0.038) and CD133 (mean positive cells 26±29% vs. 17±23%, p = 0.030) and a weaker expression of CD64 (mean positive cells 33±24% vs. 43±27%, p = 0.002). Furthermore, we observed that the T-lymphoid marker CD7 was significantly stronger expressed in cases with CEBPA distal PM (mean±SD positive cells 29±23% vs. 20±19%; p = 0.001).

Influence of CEBPA Distal PM on CEBPA Expression

To determine whether aberrant DNA methylation in the distal promoter affects CEBPA expression, we analyzed CEBPA expression levels in 120/555 cases by quantitative real-time RT-PCR. Median CEBPA expression level was 134.7 (range: 2.7–637.0). We correlated CEBPA expression levels to CEBPA methylation levels by Spearman’s rank correlation and found a limited but significant inverse correlation (Spearman correlation coefficient = −0.201, p = 0.023; Figure 2). We conclude that the DNA methylation in the CEBPA distal promoter region correlates at least in part with the downregulation of CEBPA expression in CN-AML patients and that also other causes for DNA methylation must be considered.

Figure 2

CEBPA expression correlates with promoter methylation levels.

Spearman’s rank correlation of CEBPA expression levels to CEBPA methylation levels shows a significant inverse correlation (Spearman correlation coefficient = -0,201, p = 0.023).

Correlation to Clinical Features

There was no significant difference in median age between the CEBPA dPMpos compared to CEBPA dPM negative (CEBPA dPMneg) cases (64.3 vs. 64.4 years). The same holds true for platelet count (median 97.7 vs. 95.8×109/L, n.s.) and hemoglobin level (median 9.6 vs. 9.2×g/dL, n.s.) (Table 3). Bone marrow blast percentage also did not differ between CEBPA dPMpos and CEBPA dPMneg cases (median 58.0% vs. 68.5%; n.s.). Furthermore, there was no correlation between bone marrow blast percentage and CEBPA dPM threshold (data not shown). Solely, the white blood cell count (WBC) was significantly lower in CEBPA dPMpos compared to CEBPA dPMneg cases (median 34.6 vs. 50.9×109/L, p = 0.003). To analyze whether the CEBPA dPM threshold is decisive for an elevated WBC count, we compared CEBPA dPMlow to CEBPA dPMhigh cases. There was no significant difference in the WBC counts in these cases (median 36.6 vs. 31.3×109/L; n.s.), indicating that the CEBPA dPM status per se and not CEBPA dPM ratio impacts on the WBC count.

Table 3

Patient characteristics of CEBPA distal promoter methylation positive cases.

	Total n = 555	CEBPA distal PM negative, n = 317	CEBPA distal PM positive, n = 238	p =
Sex
Female	262 (47.2%)	151/262 (57.6%)	111/262 (42.3%)	n.s.
Male	293 (52.8%)	165/293 (56.3%)	128/293 (43.7%)	n.s.
Age [years] median (range)	63.9 (20.0–89.6)	64.4 (20.9–89.6)	63.3 (20.0–87.6)	n.s.
BM blasts, % median (range)	65 (3–99)	68.5 (4–99)	58 (3–99)	n.s.
WBC count [×109/L] median (range)	43.9 (0.6–400.0)	50.9 (0.6–400.0)	34.6 (0.7–187.1)	0.003
Hb level [g/dL] median (range)	9.3 (2.8–16.3)	9.2 (2.8–16.3)	9.6 (5.0–14.2)	n.s.
Platelet count [×109/L] median (range)	96.6 (3.0–950.0)	95.8 (3.0–950.0)	97.7 (17.0–363.0)	n.s.

n.s.: not significant; BM: bone marrow; WBC: white blood cell; Hb: hemoglobin.

CEBPA core PM positive cases showed no differences in age, WBC count, platelet count and hemoglobin levels compared to CEBPA core PM negative cases (data not shown). Due to the limited number of cases, we did not perform further analysis on CEBPA core PM positive cases.

Association with Other Mutations

To determine, whether CEBPA dPM correlates with mutations frequently reported in AML we analyzed FLT3-ITD and MLL-PTD as well as mutations in NPM1, FLT3-TKD, RUNX1, ASXL1, DNMT3A, IDH1R132, IDH2R140, IDH2R172 and TET2 in correlation to CEBPA dPM status. Cases positive for FLT3-ITD with ratio <0.5 were grouped together with the FLT3-ITD negative cases, as it has been shown that only a FLT3-ITD ratio >0.5 has a significant adverse prognostic impact [26]. Thus, FLT3-ITD negative patients and patients with FLT3-ITD ratio <0.5 are combined and designated as FLT3-ITD/FLT3wtratio<0.5. NPM1 mutations (90/237, 38.0% vs. 160/314, 51.0%, p = 0.003), FLT3-ITD/FLT3wtratio<0.5 (32/238, 13.4% vs. 66/314, 21.0%; p = 0.024) and TET2 mutations (10/58, 17.2% vs. 21/55, 38.2%; p = 0.02) were less frequent in the CEBPA dPMpos compared to CEBPA dPMneg cases while IDH2R140 mutations (41/153, 26.8% vs. 28/191, 14.7%; p = 0.007) were significantly more frequent. NPM1 mutated cases and TET2 mutated cases showed significantly lower CEBPA dPM ratios compared to the respective wt cases while cases with a FLT3-ITD ratio <0.5 and those with IDH2R140 mutations showed significantly higher CEBPA dPM ratios as compared to the respective control cases. Furthermore, we observed a positive correlation of CEBPA dPM ratio with RUNX1 mutations (Table 4). Moreover, we analyzed, whether a CEBPA dPM threshold is important for the correlation to the above described mutations. Comparison of CEBPA dPMhigh versus CEBPA dPMlow cases revealed that solely the frequency of NPM1 mutations was significantly higher in CEBPA dPMlow compared to CEBPA dPMhigh cases (62/143, 43.4% vs. 28/94, 29.8%; p = 0.04). In contrast, there was no correlation of CEBPA dPM threshold to FLT3-ITD ratio, TET2, RUNX1 or IDH2R140 mutations (data not shown).

Table 4

Correlation of CEBPA distal promoter methylation ratio to molecular mutations.

Mutation (n = cases analyzed)	CEBPA distal PM ratio ± s.d.	p-value
NPM1 (n = 551)
wt (n = 301)	295.4±434.9	<0.001
mut (n = 250)	158.2±295.2
FLT3 -ITD (n = 552)
Negative and ratio<0.5 (n = 454)	248.6±393.4	0.026
ratio>0.5 (n = 98)	163.8±326.8
FLT3 -TKD (n = 447)
wt (n = 399)	237.1±397.4	n.s.
mut (n = 48)	268.3±364.7
MLL -PTD (n = 552)
neg (n = 477)	226.7±384.1	n.s.
pos (n = 75)	271.7±380.6
RUNX1 (n = 467)
wt (n = 356)	216.8±366.9	0.049
mut (n = 111)	308.7±444.1
ASXL1 (n = 420)
wt (n = 353)	226.3±388.3	n.s.
mut (n = 67)	270.7±416.6
TET2 (n = 113)
wt (n = 82)	290.2±380.1	0.034
mut (n = 31)	131.9±252.5
IDH1 R132 (n = 382)
wt (n = 347)	229.9±394.6	n.s.
mut (n = 35)	216.1±341.6
IDH2 R140 (n = 344)
wt (n = 275)	213.2±377.5	0.025
mut (n = 69)	354.9±480.3
IDH2 R172 (n = 345)
wt (n = 335)	230.8±388.6	n.s.
mut (n = 10)	270.0±357.4
DNMT3A (n = 119)
wt (n = 74)	352.3±499.0	<0.001
mut (n = 45)	156.3±290.4

wt: wild-type; mut: mutated.

Prognostic Relevance of CEBPA Promoter Methylation

First, we analyzed, whether CEBPA dPM status has impact on prognosis. CEBPA dPMpos cases did not significantly differ from CEBPA distal PMneg cases with regard to OS (24.3 months vs. median n.r.; n.s.) and EFS (median 14.4 months vs. 14.9 months; n.s.). (Figure 3 A+B). Applying the CEBPA dPM threshold also revealed no significant impact of CEBPA dPMhigh versus CEBPA dPMlow and CEBPA dPMneg status on OS (median 32.1 months vs. 24.3 months vs. median n.r. respectively; n.s.) and EFS (median 14.4 months vs. 14.9 months vs. 14.9 months, respectively; n.s.) (Figure 3 C+D). As for the observed association of CEBPA dPM with other molecular markers, we performed further subcohort analysis. No prognostic impact of CEBPA dPM was seen in subcohorts defined by age, NPM1 mutations or IDHR140 mutations (data not shown). However, OS in patients with FLT3wt/FLT3-ITDratio<0.5 was significantly worse in cases with additional CEBPA dPM compared to those without (32.6 months vs. median n.r., p = 0.02). Moreover, the threshold of CEBPA dPM seemed to be of importance for patients with FLT3wt/FLT3-ITDratio<0.5, as CEBPA dPMhigh cases showed significantly worse OS compared to CEBPA dPMneg cases (16.9 months vs. median n.r.; p = 0.03) (Figure 3 E+F). Furthermore, patients with TET2mut and CEBPA dPMpos had significantly worse OS (median 9.9 months vs. 20.3 months, p = 0.003) and EFS (median 4.7 months vs. 10.7 months; p = 0.035) compared to those TET2mut patients with CEBPA dPMneg (Figure 3 G+H). With regard to CEBPA dPM threshold, survival analysis of TET2 mutated cases was not valid, as patient numbers were too small. In the TET2wt subcohort however, we observed significantly worse EFS for CEBPA dPMhigh compared to CEBPA dPMneg cases (12.1 months vs. median n.r.; p = 0.018) (Figure 3 I). We also analyzed outcome according to CEBPA methylation status compared to CEBPA mutation status and found no significant impact for the methylation status (Figure S1).

Figure 3

Kaplan Meier survival analysis according to CEBPA distal promoter methylation status.

Survival within the total cohort of 470 patients with CN-AML and CEBPAwt. Kaplan Meier plot showing A) Overall and B) Event-free survival of CEBPA distal promoter methylation positive (red) compared to CEBPA distal promoter methylation negative cases (grey). C) and D) Overall survival and event-free survival within the total cohort of 470 patients according to CEBPA distal PM threshold. Kaplan Meier plot of CEBPA distal promoter methylation high cases (red) compared to CEBPA distal promoter methylation low (black) and CEBPA distal promoter methylation negative cases (dark grey). E) Survival Analysis within the cohort of 388 cases with CN-AML with FLT3-ITD ratio <0.5. Kaplan Meier plot showing overall survival according to CEBPA distal promoter methylation status of CEBPA distal promoter methylation positive (red) compared to CEBPA distal promoter methylation negative cases (grey). F) Overall survival within the cohort of 388 cases with CN-AML with FLT3-ITD ratio <0.5.according to CEBPA distal promoter methylation threshold of CEBPA distal promoter methylation high cases (red) compared to CEBPA distal promoter methylation low (black) and CEBPA distal promoter methylation negative cases (dark grey). G) Survival analysis according to CEBPA distal promoter methylation status within the cohort of 30 cases with CN-AML and TET2 mutations. Kaplan Meier plot showing overall survival and H) Event-free survival of CEBPA distal promoter methylation positive (red) compared to CEBPA distal promoter methylation negative cases (grey). I) Event-free survival within the cohort of 80 patients with CN-AML and TET2 wild-type according to CEBPA distal promoter methylation threshold. Kaplan Meier plot of CEBPA distal promoter methylation high cases (black) compared to CEBPA distal PM low (red) and CEBPA distal promoter methylation negative cases (dark grey).

Uni- and Multivariable Analysis

In univariable analysis, the following parameters were associated with worse EFS: higher age (p<0.001), higher WBC count (p<0.001), FLT3-ITD/wt ratio higher than 0.5 (p<0.001), and RUNX1 mutations (p = 0.003). NPM1 mutations (p = 0.011) were associated with better EFS. CEBPA dPM status or CEBPA dPM ratio had no significant influence on EFS. In multivariable analysis, only age (P<0.001), WBC count (p<0.001) and FLT3-ITD/wt ratio <0.5 (p = 0.001) maintained their relevance for EFS. Investigating OS, age (p<0.001) and WBC (p<0.001), the FLT3-ITD ratio >0.5 (P = 0.014), NPM1 mutations (p<0.001) and RUNX1 mutations (p = 0.001) were prognostically relevant in univariable analysis. CEBPA dPM status or CEBPA dPM ratio had no significant impact on OS in univariable analysis. In multivariable analysis, age (p<0.001), WBC count (p<0.001), FLT3-ITD/wt ratio >0.5 (p = 0.012) and NPM1 mutations (p = 0.007) retained their prognostic impact (Table 5).

Table 5

Influence of different biological and leukemia-associated parameters on OS and EFS in 555 CN-AML patients in uni- and multivariable analysis.

Parameter	EFS univariable		EFS multivariable		OS univariable		OS multivariable
	P	RR	P	RR	P	RR	P	RR
Age	<0.001	1.34*	<0.001	1.41*	<0.001	1.49*	<0.001	1.54*
Gender	NS	–	–	–	NS	–	–	–
WBC count	<0.001	1.06#	<0.001	1.06#	<0.001	1.06#	<0.001	1.08#
CEBPA dPM status	NS	–	–	–	NS	–	–	–
CEBPA dPM ratio	NS	–	–	–	NS	–	–	–
FLT3-ITD/wt ratio (≥0.5)	<0.001	1.74	0.001	1.91	0.014	1.55	0.012	1.93
NPM1mut	0.011	0.72	NS	0.79	<0.001	0.52	0.007	0.51
RUNX1mut	0.003	1.56	NS	–	0.001	1.75	NS	–

Abbreviations: EFS:event-free survival; NS: not significant; OS: overall survival; RR: relative risk.

Per 10 years of increase;

Per 10×109/L.

Age, peripheral blood cell counts and CEBPA dPM ratio were considered as continuous parameters.

We compared the gene expression profiles of 9 CEBPA dPM samples with 8 CEBPA single-mutated, 10 CEBPA double-mutated and 10 CEBPA non-methylated/non-mutated cases. For the multicomparison, significant differential expression was detected for a total of 727 genes.

We identified 548 genes for the comparison of CEBPA methylated and non-methylated/non-mutated samples. Comparison of CEBPA methylated with double-mutated samples revealed 298 significantly differentially expressed genes. Analyzing these pairwise comparisons, we identified overlapping 119 genes, which were significantly differentially expressed for both sets (Tables S4/S5). We identified no significantly differentially expressed genes for CEBPA methylated and single-mutated samples, as CEBPA single-mutated cases showed a strong heterogeneous expression pattern (compare Figure S2).

To further study the expression profiles we then applied clustering algorithms. CEBPA methylated samples showed a unique pattern compared to the remaining samples (CEBPA single-mutated, CEBPA double-mutated and CEBPA non-methylated/non-mutated cases). By Gene Ontology analysis, significantly differentially expressed genes for the multicomparison were associated with a function for myeloid cell differentiation and hematopoietic development, e.g. RUNX1 was upregulated and Kruppel-like factor 1 (KLF1) was downregulated (Table S3).

Discussion

In the present study, we investigated the frequency and the clinical relevance of CEBPA PM in 623 de novo CN-AML and showed that aberrant DNA methylation in the promoter of CEBPA is very heterogeneously spread across the core, proximal and distal promoter regions. A distinct pattern of aberrant DNA methylation was mainly restricted to the distal promoter region of CEBPA (42.9%), whereas methylation of the CEBPA core promoter seems to be a rare event in AML (2.5%). Methylation of the CEBPA proximal promoter was not observed in any case. These findings are in line with previous reports [11], [12], [27]. Coincidence of CEBPA PM and CEBPA mutations was never observed, indicating that these two events are mutually exclusive.

Aberrant CEBPA PM has also been described in lung cancers and head and neck cancers. The core promoter was not affected by epigenetic silencing in these entities [6], [7]. It is noteworthy that the DNA methylation patterns within the CpG islands showed tumor-type specificity with CEBPA methylation being restricted to the distal promoter region in head and neck cancer [24], whereas in lung cancer also the proximal promoter region was differentially methylated [23]. In contrast, in AML CEBPA methylation could be observed in the distal promoter region as well as in the core promoter region. A possible explanation for this finding could be that different regulatory regions are used in different tissues, and epigenetic mechanisms interrupt the interaction of the relevant binding proteins with these regions through chromatin conformation changes.

Data regarding influence of CEBPA PM on CEBPA expression in AML is heterogeneous (an overview is given in Table S6). This is probably due to the variability in the selected cohorts as well as the CEBPA promoter region analyzed. Hollink et al [27] analyzed the relevance of CEBPA core PM in 237 unselected cases of pediatric AML and found it to be a rare event, as it occurred in only three cases (1.3%). Furthermore, CEBPA gene expression was down regulated in these cases. Wouters et al. [10] showed in a cohort of 285 unselected AML patients that CEBPA silencing is not associated with CEBPA hypermethylation, suggesting a possible yet unknown mechanism of CEBPA mRNA repression. Lin et al. [12] correlated the methylation levels in the distal CEBPA promoter region with its transcript levels in leukemic cells prepared from 12 unselected AML patients and observed a negative correlation. They conclude that the DNA methylation in the distal CEBPA promoter region correlates with the down regulation of CEBPA expression in patients with AML. We were able to confirm this data, as we also found CEBPA expression to be negatively correlated with CEBPA distal PM in CN-AML cases in the present study.

Another interesting finding was the aberrant expression of the T-cell marker CD7 of CEBPA dPM positive cases, which has already been reported by Wouters et al. [10]. In contrast, we were not able to confirm a mixed myeloid/T-lymphoid phenotype, as we did not detect an increased expression of myeloid markers like CD13 and CD33.

To date, there is only one study by Lin et al. regarding prognostic relevance of CEBPA distal PM in AML [12]. This report describes favorable prognosis for AML patients with CEBPA PM in a subcohort of 59 cases after excluding patients with favorable karyotypes, NPM1 mutations and CEBPA mutations. Furthermore a survival advantage for patients with CEBPA promoter hypermethylation was seen within a subcohort of 25 CN-AML patients with wt CEBPA and wt NPM1. However, these results were based on a relatively small number of cases analyzed. Furthermore, in multivariable analysis, they found high CEBPA PM to be an independent prognostic factor for disease-free survival. In another paper Hollink et al. [27] performed unsupervised cluster analysis in 237 unselected cases of pediatric AML and identified five cases with silenced CEBPA, including three cases with aberrant CEBPA PM. Four of these cases experienced relapse indicating poor outcome for patients with silenced CEBPA.

With regard to prognosis, our survival data of 470 patients show for the first time that CEBPA distal PM per se is not a prognostic factor in the overall cohort of CN-AML. Furthermore, the survival analysis of a subcohort of 260 CN-AML patients with wt NPM1 and wt CEBPA revealed no prognostic impact of CEBPA distal PM (data not shown) and thus is in contrast to the study of Lin et al. [12]. However, we observed an adverse impact of high CEBPA distal PM on OS in the subset of cases with FLT3wt/FLT3-ITDratio<0.5. But not only in this more favorable group a prognostic effect could be shown. Also in the more adverse subgroup with TET2 mutations the CEBPA dPMpos had significantly worse OS and EFS compared to those with TET2 mutations and CEBPA dPMneg. However, this result is based on only a limited number of patients and needs to be validated in a larger cohort. In multivariable analysis, CEBPA dPM had no significant impact on OS and EFS in our series, which is again in contrast to the study of Lin et al.

Taken together, our data is in line with the concept that CEBPA PM does not directly influence prognosis. We rather assume that the negative prognostic effect of CEBPA PM which was observed in certain subgroups is caused by the CEBPA PM induced down regulation of CEBPA expression. This concept is supported by the study of Figueroa et al. [28] which showed that CEBPA silenced cases (n = 8) had a considerably worse outcome compared to CEBPA mutated cases (n = 8) (5-year overall survival 25% vs. 88%; log-rank test P<0.003). Furthermore, Barjesteh et al. were also able to show an unfavorable prognosis for six patients with intermediate-risk karyotype AML and low CEBPA expression [29].

Data on gene expression profiles of CEBPA methylated AML are heterogeneous. In the study of Hollink et al [27], unsupervised cluster analysis of the total cohort of 237 cases showed that CEBPA mutated cases predominantly clustered together with CEBPA hypermethylated cases. Figueroa et al. [28] performed unsupervised cluster analysis and also found a similar gene expression profile of CEBPA silenced AML and CEBPA mutated AML. In contrast to these studies, our gene expression analysis of CEBPA PMpos cases showed a highly distinct clustering of CEBPA methylated cases compared to CEBPA mutated cases, emphasizing the relevance of the aberrant CEBPA distal PM. Our data is affirmed by our recently published study on gene expression profiling in AML [30]. In this study gene expression signatures for 30 CEBPA mutated cases were compared with the profiles of 204 CEBPA wt cases. CEBPA mutated cases and CEBPA wt cases showed a highly distinct gene expression signature and did not cluster together. As in the present study all CEBPA distal PM positive cases are CEBPA unmutated, it is feasible that they show a distinct gene expression profile from CEBPA mutated cases.

It can be speculated that CEBPA promoter methylation is not a focal, targeted event and that it may perhaps more likely occur in the context of global hypermethylation as observed by Figueroa [28]. Such analysis should be in the focus of further evaluation.

In conclusion, we demonstrate that aberrant methylation of the distal CEBPA promoter can be found in a substantial proportion of CN-AML patients. It is positively correlated to genotypes with RUNX1mut and IDH2R140mut and negatively correlated with NPM1mut, FLT3-ITD, TET2mut and DNMT3Amut. An effect of epigenetic modifications of the CEBPA promoter on survival was not found in the total cohort of 555 CN-AML patients. However, we detected an adverse effect in the subsets with FLT3wt/FLT3-ITDratio<0.5 and those with TET2mut but only in univariable analysis. Furthermore, aberrant methylation of the distal CEBPA promoter was closely correlated to reduced CEBPA expression and a distinct gene expression profile. This change in underlying gene expression profile suggests a contribution of CEBPA methylation to leukemic transformation. However, CEBPA distal PM is a negligible prognostic marker, as we show that its influence on outcome in CN-AML is strongly dependent on other markers.

Gene expression profiling. Heatmap visualizing the gene expression profiles of 9 CEBPA methylated/unmutated 10 CEBPA double-mutated, 8 CEBPA single-mutated and 10 non-methylated/unmutated cases. The black symbols indicate the CEBPA methylated cases, the red symbols CEBPA double-mutated cases, the blue symbols CEBPA single-mutated cases and the grey symbols the CEBPA non-methylated/wild-type cases.

(EPS)

Click here for additional data file.

Kaplan Meier survival analysis according to mutation- and distal promoter methylation status. Survival within the total cohort of 555 patients with CN-AML. Kaplan Meier plot showing A) Overall and B) Event-free survival of CEBPAbi mutated (grey) and CEBPAmono+homo mutated cases (red) compared to CEBPA distal promoter methylation high cases (black) compared to CEBPA distal promoter methylation low (purple) and CEBPA distal promoter methylation negative cases (blue).

(EPS)

Click here for additional data file.

Primers and PCR conditions.

(DOC)

Click here for additional data file.

Methylation status at each CpG residue for each patient.

(XLS)

Click here for additional data file.

Significantly expressed genes with function in regulation of cellular and component organization identified by Gene Ontology.

(DOC)

Click here for additional data file.

Significantly expressed genes with a function for myeloid cell differentiation and hemopoietic development identified by Gene Ontology.

(DOC)

Click here for additional data file.

Significantly expressed genes in Gene Ontology.

(DOC)

Click here for additional data file.

Overview of promoter methylation studies.

(DOC)

Click here for additional data file.