Altered methylation levels in elderly acute myeloid leukaemia patients compared to elderly well individuals


Epigenetic alterations, such as aberrant DNA methylation, can result in gene silencing, particularly of tumour suppressor genes during leukaemogenesis (Dellett et al, 2010; Fabiani et al, 2010). Hyper-methylation has been associated with silencing of genes involved in cell cycle regulation and DNA repair among other functions. Promoter hyper-methylation of the CDKN2A gene has been documented in different cancer types. While the genetic aberrations occurring in acute myeloid leukaemia (AML) are fairly well understood, the associated epigenetic deregulation remains to be determined. Moreover, increasing evidence supports a role for epigenetic lesions and perturbations in the aetiology of ageing (Gravina & Vijg, 2010) which may be partly due to the accumulation of errors in DNA methylation.

Few studies have investigated methylation patterns in elderly well individuals. A study of non-cancer patients aged between 45 and 75 years showed LIRE1 (long interspersed nuclear element-1, retro-transposable element) (a proxy for global DNA methylation) methylation was stable, with some discrimination for gender and race/ethnicity (Kitkumthorn & Mutirangura, 2011; Zhang et al, 2011). In keeping with this finding, Gentilini et al (2012) demonstrated that centenarians show better preservation of DNA methylation status, suggesting that both slower cell growth/metabolism and better control of signal transmission through epigenetic mechanisms, may contribute to longevity.

Scandura et al (2011) reported lower hypo-methylation ratios of the LINE1 repetitive element in AML patients achieving complete remission (CR) compared with non-responders in an epigenetically primed cohort of younger AML patients (aged < 60 years at diagnosis) prior to intensive chemotherapy.

In this study we compared the methylation status of elderly AML patients with that of ‘elite elderly’ individuals to identify any differences in age- or disease-related changes. The ‘elite elderly’ subjects, when enlisted into the Belfast Elderly Longitudinal Free-living Ageing STudy (BELFAST) project, were community-living, cognitively intact and met the criteria of the Senieur protocol. They were consistent with the concept of the Perl ‘escaper’ or the 15% of elderly people who have reached the age of 90–100 with no overt signs of illness (Rea & Rea, 2011). The cohorts studied were 26 elderly AML patients (age range: 70–86 years at diagnosis; median: 74·5 years) and 45 ‘elite elderly’ individuals (age range: 71–98 years; median: 86 years), healthy community-living participants within the BELFAST project (Rea et al, 2009; Bennati et al, 2010). Ethical approval and written consent was obtained for both studies. DNA was extracted from bone marrow (BM) or peripheral blood (PB) from AML patients before treatment, whilst only PB was available for the BELFAST subjects. Both cohorts were analysed by pyrosequencing (PyroMark CDKN2A and LIRE1 kits; Qiagen, Crawley, UK) four CpG sites in the LIRE1 repeat element and four CpG sites in the CDKN2A gene (a tumour suppressor). The average methylation across all the CpG sites measured by the different assays, as well as the methylation levels for individual CpG sites in the different cohorts was compared between the AML and ‘elite elderly’ subjects using the anova and pairwise t-tests.

No statistical difference in methylation level distributions between BM and PB samples from AML patients was observed, allowing direct comparison of all AML samples irrespective of BM or PB status with the ‘elite elderly’ BELFAST subjects’ PB samples (data not shown).

There were highly significant differences between the AML and ‘elite elderly’ subjects for both CDKN2A (median: 9·8% and 5·5%; P < 0·002) and LIRE1 (median 73·8% and 85·0%; P < 10−13) methylation levels. Differences at individual CpG sites were observed, but we have limited our report to the analysis of the average methylation levels.

To explore age-related comparisons further, the ‘elite elderly’ subjects were divided into three age groups; 70–79, 80–89 and 90–99 years. The 70–79 year-old individuals showed statistically different CDKN2A hyper-methylation (median: 19·9%, P < 0·007) and LIRE1 hypo-methylation (median: 80·4%, P < 0·001) ranges when compared to all ‘elite elderly’ subjects over 80 years old (Fig 1). This data would suggest that there are changes in methylation status that are associated with ageing, although all BELFAST subjects, at participation, were recruited by similar stringent criteria (Rea et al, 2009). It is possible that some of the 70–79 year-olds were already harbouring epigenetic changes capable of compromising their potential to progress to the remarkably low morbidity ageing demonstrated by the 80–89 and 90–99 year-old cohorts (Table 1).

Table 1. Characteristics and methylation distribution of AML patients and ‘Elite elderly’ individuals
OverallAMLElite elderlyP value
n = 26n = 45 
  1. AML, acute myeloid leukaemia; ITD, internal tandem duplication; N.S., not significant.

Age at diagnosis/study entry, years; median (range)

n = 23

73 (70–77)

n = 6

76 (71–79)

P < 0.001

n = 3

84 (83–86)

n = 23

84 (80–89)


n = 16

93.5 (90–98)

Remission achieved
Figure 1.

Box and whisker plot comparing methylation levels in elderly AML patients and ‘elite elderly’ individuals Box and whisker plots showing methylation levels for LIRE1 (top row) and CDKN2A (bottom row). The% methylation shown is an average across four CpG sites for each marker. The plots are in four columns from left to right: (A) comparison of methylation levels in AML peripheral blood (PB) or bone marrow (BM) samples; (B) comparison of methylation levels in different age groups for AML patients [note that no AML patient material from the 80–89 year-old cohort was available for the CDKN2A analysis]; (C) comparison of the methylation levels in different age cohorts for ‘elite elderly’; and (D) comparison of methylation levels in AML patients who achieved a remission versus those that did not and against the overall ‘elite elderly’ population.

A comparison of the AML patients aged 70–79 and 80–89 years with their matched ‘elite’ age groups showed significant differences for both CDKN2A and LIRE1 (P < 0·015/P < 0·002 for CDKN2A and P < 0·002/P < 0·001 respectively). This demonstrated that the methylation status of elderly AML patients was independent of age and was significantly different from age-matched groups (Fig 1).

There was no correlation for CDKN2A methylation levels within the AML cohort between the presence and absence of a FLT3 mutation or NPM1 mutation, cytogenetic risk group, gender, or whether the patient entered remission or not. However, LIRE1 methylation levels were significantly different between patients with or without a FLT3 mutation (P = 0·035) and between those who achieved remission or not (P = 0·002). Those AML patients who did achieve remission had a significantly higher level of LIRE1 methylation (median 78·97%) than those that did not achieve remission (median 72·68%; P < 0·0021), although both of these groups had lower levels than those measured for the ‘elite’ elderly patients (median 85·03%; remission: P value = 0·00054; no remission: P value < 0·00003).

Scandura et al (2011) showed a possible link between clinical outcome and therapy-induced demethylation; in that study, the hypo-methylation ratio of pre- to post-treatment of LIRE1 and HIST1H2AA methylation levels showed that those patients who achieved a complete remission (CR) had a ratio around 0, indicating little or no change in methylation status. However, those patients with partial or no response showed an increase in ‘hypo-methylation ratio’ of around 1. This would indicate that hyper-methylation had occurred in the LIRE1 and HIST1H2AA markers. One interpretation of this is that those patients who achieved CR did not need to, or could not, undergo hyper-methylation as the LIRE1 loci were already ‘normally’ methylated. These patients could be similar to those in our study who had higher or nearer the levels of LIRE1 methylation seen in the ‘elite elderly’ individuals.

It is interesting to speculate that the degree of methylation at diagnosis or pre-treatment may be predictive of the potential of elderly AML patients to achieve a CR.


Supported by: Ireland-Northern Ireland-NCI Cancer Consortium [a Joint Research Projects in Cancer (JRPC) grant], Northern Ireland Leukaemia Research Fund (NILRF), Department of Employment and Learning, Northern Ireland Assembly, Cancer Research UK and Research and Education into Ageing and Belfast City Hospital Trust Fund.

Author contributions

MD, HAAC, and KAP performed the research; IMR and KIM designed the research study; MFM, IMR and KIM contributed essential reagents or tools; MD and KIM analysed the data; MD, HAAC, KAP, MFM, IMR and KIM wrote the paper.