SEARCH

SEARCH BY CITATION

Keywords:

  • chronic kidney disease;
  • DNA methylation;
  • homocysteine;
  • inflammation;
  • interleukin-6

Abstract.

  1. Top of page
  2. Abstract.
  3. Introduction
  4. Patients and methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Conflict of interest statement
  9. Acknowledgements
  10. References

Objective.  The lifespan of dialysis patients is as short as in patients with metastatic cancer disease, mainly due to cardiovascular disease (CVD). DNA methylation is an important cellular mechanism modulating gene expression associated with ageing, inflammation and atherosclerotic processes.

Design.  DNA methylation was analysed in peripheral blood leucocytes from three different groups of chronic kidney disease (CKD) populations (37 CKD stages 3 and 4 patients, 98 CKD stage 5 patients and 20 prevalent haemodialysis patients). Thirty-six healthy subjects served as controls. Clinical characteristics (diabetes mellitus, nutritional status and presence of clinical CVD), inflammation and oxidative stress biomarkers, homocysteine and global DNA methylation in peripheral blood leucocytes (defined as HpaII/MspI ratio by the Luminometric Methylation Assay method) were evaluated. CKD stage 5 patients (n = 98) starting dialysis treatment were followed for a period of 36 ± 2 months.

Results.  Inflamed patients had lower ratios of HpaII/MspI, indicating global DNA hypermethylation. Analysis by the Cox regression model demonstrated that DNA hypermethylation (HpaII/MspI ratio <median) was significantly associated with both all-cause (RR 5.0; 95% CI: 1.7–14.8; P < 0.01) and cardiovascular (RR 13.9; 95% CI: 1.8–109.3; P < 0.05) mortality, even following the adjustment for age, CVD, diabetes mellitus and inflammation.

Conclusion.  The present study demonstrates that global DNA hypermethylation is associated with inflammation and increased mortality in CKD.


Introduction

  1. Top of page
  2. Abstract.
  3. Introduction
  4. Patients and methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Conflict of interest statement
  9. Acknowledgements
  10. References

The lifespan of chronic kidney disease (CKD) patients is reduced mainly as a consequence of premature deaths due to cardiovascular disease (CVD) [1]. Despite improvements in dialysis technology, the majority of dialysis patients die within a 5-year period – a survival similar, or even worse, than that of patients with various metastatic cancer diseases. As conventional (Framingham) risk factors alone cannot explain the exceedingly high mortality rate in this population, much interest has focused on nontraditional risk factors. Amongst several novel and commonly observed risk factors, oxidative stress [2], hyperhomocysteinaemia [3] and persistent inflammation [4], have attracted much interest. Inflammatory biomarkers, such as C-reactive protein (CRP) and interleukin (IL)-6, are strong independent predictors of poor outcome in this population [5, 6], whereas hyperhomocysteinaemia, high blood pressure and hypercholesterolaemia are associated with better outcome in CKD [3, 7].

Epigenetics (i.e. properties of the genome that are not explained by the primary DNA sequence, but the result of modifications of DNA and/or associated proteins due to, e.g. acetylation, methylation, ubiquitination, sumoylation and phosphorylation) may be particularly important for patients suffering from uraemic toxicity. Changes in genomic DNA methylation (addition of methyl groups to the 5′-position of cytosine rings within the context of CpG dinucleotides) have important regulatory functions in normal and pathological cellular processes, and contribute to normal development and differentiation of cells and tissues. However, epigenetic modifications are crucial also for various disease-related processes, such as tumorigenesis [8] and atherogenesis [9]. As the epigenetic state, such as DNA methylation, may be responsible for changes contributing to the development of a variety of cancers [10, 11], which share many similarities in its pathogenesis with atherosclerosis (such as ageing, smoking and aberrant cell proliferation), it is possible that epigenetics may be of particular importance in the high incidence of CVD in uraemia. Recently, we reported a new method to analyse global genomic DNA methylation using a luminometric technology (Luminometric Methylation Assay, LUMA) to quantitate methylation-sensitive restriction digestions [12, 13]. The method is based on digestion of genomic DNA by isoschizomers HpaII and MspI. The target sequence for both enzymes is CCGG. HpaII, however, is not able to cut if the internal cytosine is methylated, whilst MspI is insensitive to this modification. By using a separate internal digestion control recognizing a sequence devoid of the CpG dinucleotide (EcoRI), an accurate HpaII/MspI ratio can be obtained from experimental samples. Thus, DNA methylation is defined as the HpaII/MspI ratio. If DNA were completely unmethylated the HpaII/MspI ratio would be 1.0, and if DNA is 100% methylated the HpaII/MspI ratio would approach 0. As the LUMA is performed in 96-well plates, it is now possible to perform large-scale accurate analyses to examine global epigenotype-phenotype-outcome relationships in clinical materials. Given the high prevalence of inflammation and hyperhomocysteinaemia in combination with a high cardiovascular mortality rate in CKD patients, this population may be particularly well suited for studies associating phenotype with epigenetic alterations and clinical outcome.

The aim of the current study was to evaluate peripheral blood cell DNA methylation (defined as HpaII/MspI ratios in LUMA) in clinically well-defined cohorts of renal patients. By this approach, it was possible to evaluate the association between renal function, hyperhomocysteinaemia, inflammation and aberrant DNA methylation. In addition, the independent impact of DNA methylation on all-cause and cardiovascular survival was assessed by multivariate regression analysis in incident patients starting renal replacement therapy.

Patients and methods

  1. Top of page
  2. Abstract.
  3. Introduction
  4. Patients and methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Conflict of interest statement
  9. Acknowledgements
  10. References

Chronic kidney disease patients (Groups 1 and 2)

Altogether, three different groups of CKD patients were enrolled in the cross-sectional part of the study. Characteristics of the patients are shown in Table 1. The patients were divided into groups according to the degree of renal failure using the K/DOQI Guidelines staging [14]. The Ethics Committee of the Karolinska Institutet approved the study and informed consent was obtained from the patients.

Table 1.   Clinical characteristics, renal function, inflammatory and oxidative stress biomarkers and DNA methylation markers in controls (Group 4) and patients with chronic kidney disease (Groups 1–3)
 Group 1Group 2AGroup 2BGroup 3P-valuec 
Stages 3 and 4Stage 5Stage 5HDGroup 4
NoninflamedaNoninflamedaInflamedaInflamedbControls
  1. aBased on one CRP measurement.

  2. bBased on the mean of weekly CRP measurements during 3 months.

  3. cKruskal–Wallis (continuous variables) or chi-square (dichotomized variables) between Groups 1–3.

  4. dMedian and range.

  5. •not applicable

N37564220 36
Age (years)58 ± 254 ± 258 ± 260 ± 3NS63 ± 2
Males (%)78635550NS69
Smokers/ex-smokers (%)52546032NS54
BMI (kg m−2)26.7 ± 0.724.8 ± 0.624.7 ± 0.722.7 ± 1.1<0.0525.1 ± 0.5
Dialysis time (months)d28 (4–269)
Diabetes mellitus (%)27303345NS0
Wasting (SGA ≥2; %)5184360<0.00010
Cardiovascular disease (%)24275265<0.010
S-creatinine (μmol L−1)264 ± 18746 ± 37680 ± 39678 ± 47<0.000181 ± 2
GFR (mL min−1)29 ± 26 ± 16 ± 1<0.000185 ± 3
S-albumin (g L−1)37 ± 135 ± 130 ± 133 ± 1<0.000139 ± 1
Haemoglobin (g L−1)129 ± 2108 ± 2104 ± 2118 ± 3<0.0001128 ± 2
Cholesterol (mmol L−1)5.3 ± 0.24.9 ± 0.25.0 ± 0.24.1 ± 0.2<0.015.1 ± 0.1
Glucose (mmol L−1)5.9 ± 0.36.4 ± 0.46.6 ± 0.4NS5.3 ± 0.2
hsCRP (mg L−1)d2.8 (0.3–8.7)2.6 (0.2–9.1)20.5 (11.0–218.0)22.9 (10.5–128.9)<0.00011.2 (0.2–9.1)
IL-6 (pg mL−1)d3.4 (0.8–8.8)3.9 (0.8–16.9)11.3 (3.2–112.0)9.6 (3.0–91.5)<0.00011.9 (0.4–10.9)
Leucocytes (109 mL−1)7.6 ± 0.37.3 ± 0.210.5 ± 0.510.5 ± 0.8<0.00016.3 ± 0.3
Fibrinogen (g L−1)3.7 ± 0.14.4 ± 0.16.2 ± 0.24.5 ± 0.3<0.00012.9 ± 0.1
TNF-α (pg mL−1)d8.4 (4.7–27.8)10.8 (5.4–21.1)11.8 (3.3–49.2)13.0 (7.7–51.9)<0.013.7 (1.3–10.6)
8-OHdG (ng mL−1)d0.4 (0.2–1.2)0.7 (0.3–3.2)0.9 (0.5–3.8)1.1 (0.7–2.8)<0.00010.3 (0.1–1.0)
Folic acid (nmol L−1)11 ± 214 ± 114 ± 2NS20 ± 2
B-12 (pmol L−1)363 ± 29435 ± 24494 ± 38<0.05327 ± 25
Homocysteine (μmol L−1)19.6 ± 1.341.5 ± 3.926.8 ± 2.222.9 ± 3.2<0.00018.7 ± 0.5
HpaII/MspI ratio0.61 ± 0.020.60 ± 0.010.54 ± 0.010.55 ± 0.02<0.00010.60 ± 0.01

Group 1 consisted of 37 noninflamed (CRP: <10 mg L−1) CKD stages 3 and 4 (GFR: 15–60 mL min−1) patients (58 ± 2 years). Patients with CKD stages 3 and 4 were recruited from the hospital renal outpatient clinic. The study exclusion criteria were age <18 or >80 years, clinical signs of acute infection, active vasculitis at the time of evaluation or unwillingness to participate in the study. The causes of CKD in this group were chronic glomerulonephritis (n = 7), diabetic nephropathy (n = 1), polycystic kidney disease (n = 4), nephrosclerosis (n = 5), other renal diseases (n = 5) and unknown aetiologies (n = 6). Many patients were on antihypertensive medications [ACE inhibitors (ACEI) and/or angiotensin II receptor blocker (ARB), n = 34]; β-blockers (n = 25); calcium channel blockers (CCB; n = 17). Statins were used in 10 (27%) patients. Presence of clinical CVD was present in nine (24%) patients and defined by medical history, clinical symptoms and/or findings of cardiac, cerebrovascular (stroke), peripheral vascular disease and/or other atherosclerotic manifestation.

Group 2 consisted of CKD stage 5 patients enrolled at the initiation of renal replacement therapy (RRT) in the renal programme of the Karolinska University Hospital Huddinge between 1998 and 2005, included as participants of an ongoing prospective study [15]. Sampling was performed close to start of RRT. The patients were subdivided into two groups based on CRP levels. Group 2A consisted of 56 noninflamed (CRP: <10 mg L−1) whereas Group 2B consisted of 42 inflamed (CRP: ≥10 mg L−1). Characteristics of the patients in Group 2 are shown in Table 1. The study exclusion criteria were age <18 or >70 years, clinical signs of acute infection, active vasculitis or liver disease at the time of evaluation, or unwillingness to participate in the study. The causes of CKD in these patients were chronic glomerulonephritis in 19 patients, diabetic nephropathy in 31 patients, polycystic kidney disease in nine patients, nephrosclerosis in three patients, other renal diseases in 14 patients and unknown aetiologies in 22 patients.

Thirty-seven (38%) patients had a clinical history or signs of cardiovascular (acute myocardial infarction, n = 9; ischaemic heart disease, n = 21), cerebrovascular (n = 9), peripheral vascular disease (n = 10) and/or other atherosclerotic manifestations, such as aortic aneurysm and/or atherosclerotic renal artery stenosis (n = 4) at the start of the study. Forty-three of the patients started peritoneal dialysis (PD), whereas 55 patients started HD. Haemodialysis was performed three times a week (4–5 h per session) using bicarbonate dialysate and a high-flux synthetic membrane, whilst PD patients received glucose and/or polyglucose-based solutions with four daily exchanges. Most patients were on antihypertensive medications (ACEIs and/or ARBs; n = 56); β-blockers (n = 55); CCB (n = 32). Statins were used in 30 (31%) patients. During the observation period (36 ± 2 months; range: 3–94) 26 patients died; 13 due to complications related to CVD (including sudden death), whereas 13 patients died of noncardiovascular or unknown, causes (cancer, n = 1; unknown cause, n = 4; infectious complications, n = 4; wasting, n = 4). Patients were followed until death or the end of the observation period (3 March 2006). Twenty-seven patients (27%) received a kidney transplant subsequent to entering the study (14 ± 3 months; range: 4–51) and were followed in the same way as those that did not receive a renal transplant.

Prevalent haemodialysis patients (Group 3)

As the presence of inflammation in Groups 1 and 2 was based on one single CRP measurement, we also included 20 constantly inflamed (mean CRP of consecutive weekly measurements during 3 months >10 mg L−1) prevalent HD patients (60 ± 3 years) in Group 3 to evaluate the relation between persistent inflammation and aberrant DNA methylation. Only HD patients unwilling to participate or with active HIV/hepatitis infection were excluded. Median dialysis vintage was 28 months (range: 4–269). In this selected group HD patients, 13 (65%) patients had a clinical history or signs of cardiovascular (acute myocardial infarction, n = 8 and ischaemic heart disease, n = 10), cerebrovascular (n = 6) and/or peripheral vascular disease (n = 4). Haemodialysis was performed three times a week (4–5 h per session) using bicarbonate dialysate and a high-flux synthetic membrane. Protein catabolic rate was 1.04 ± 0.06 g kg−1 day−1 and Kt/V 1.61 ± 0.05 per session. Many patients were on antihypertensive medications (ACEIs and/or ARBs; n = 10); β-blockers (n = 10); CCB (n = 6). Statins were used in five (25%) patients.

Control subjects (Group 4)

A population-based randomly selected group of 36 control subjects (63 ± 2 years; 69% males) were used for comparative analyses of biochemical and metabolic parameters. Characteristics of the controls (Group 4) are shown in Table 1. The control subjects (randomly selected by the Statistics Bureau of Sweden) were investigated according to a similar protocol as the renal patient groups.

Measurement methods

After an overnight fast, venous blood samples were drawn and stored at −70 °C for biochemical analyses. GFR was estimated by the mean of creatinine and urea clearance, calculated from 24-h urinary samples collected from the CKD stage 5 patients, or as iohexol clearance in CKD stages 3 and 4 patients and controls. No evaluation of GFR was performed in the prevalent HD patients (Group 3) as the majority were anuric. Plasma IL-6 was analysed on an Immulite Automatic Immunoassay Analyzer (DPC, Los Angeles, CA, USA) with an assay manufactured for this analyzer. TNF-α was analysed in serum by an immunometric assay on an Immulite Analyzer (DPC). Plasma total homocysteine (n = 153) was determined with high-performance liquid chromatography using fluorescence detection [16]. High sensitivity (hs)-CRP concentration was measured by nephelometry. Serum 8-hydroxydeoxyguanosine (8-OHdG) was determined (n = 134) by a competitive ELISA kit (Japanese Institute for the Control of Aging, Fukuroi, Shizuoka, Japan). Subjective global nutritional assessment (SGA) was used to evaluate the protein-energy nutritional status [17]. Malnutrition for the purpose of this study was defined as an SGA ≥2.

Measurements of DNA methylation by LUMA

From a 5 mL EDTA sample of peripheral blood DNA was extracted using QIAamp® DNA kit. Restriction enzymes (HpaII, MspI and EcoRI), were purchased from New England Biolabs (Beverly, MA, USA). PSQTM 96 SNP reagents for pyrosequencing were purchased from Biotage AB (Uppsala, Sweden). DNA quantification was performed using the RediPlateTM 96 PicoGreen® kit from Molecular Probes (Eugene, OR, USA). Luminometric Methylation Assay was run as described elsewhere in detail [12, 13]. Briefly, genomic DNA (200–500 ng) was cleaved with HpaII + EcoRI or MspI + EcoRI in two separate reactions and was run in a 96-well format. The digestion reactions were run in a PSQ96TM MA system (Biotage AB). Peak heights were calculated using the PSQ96TM MA software. The HpaII/EcoRI and MspI/EcoRI ratios were calculated as (dGTP + dCTP)/dATP for the respective reactions. The HpaII/MspI ratio was defined as (HpaII/EcoRI)/(MspI/EcoRI). To validate reproducibility of LUMA results, 30 blood samples showing different level of DNA methylation were randomly selected and assayed four times by two independent technicians.

Statistical analysis

Results are expressed as mean and standard error of mean (normally distributed variables) or median and range (non-normal distribution) unless otherwise indicated, with P-value of <0.05 indicating significance. The Mann–Whitney U-test or Kruskal–Wallis method were used for nonparametric comparisons, and Student's t-test or one-way anova for parametric comparisons. The General Linear Models (GLM) procedure with least square mean values was used to identify significant interactions between CKD stage and inflammation. When significant interactions were found between factors, these were identified with the simple main effect tests. We assessed proportions using the chi-square test. Spearman's rank correlation was used to determine correlations between two variables.

Survival was analysed by Kaplan–Meier statistics and compared between the groups by the log-rank test. Survival over the whole observation period was studied by Cox's proportional hazards analysis to identify factors independently associated with clinical outcome. The Cox proportional hazards model (the PHREG procedure) was used to examine the effects of baseline and follow-up variables on the outcome variables. The relative risks for mortality were determined by univariate and multivariate Cox regression analysis and presented as hazard ratio (HR; 95% confidence intervals). Survival was measured from the day of examination until death or censoring, which was made at the end of the follow up (3 March 2006). No patient was lost to follow up. The statistical analysis was performed using statistical software sas version 9.1 (SAS Campus Drive, Cary, NC, USA 27513). The authors have full access to the data and take responsibility for its integrity.

Results

  1. Top of page
  2. Abstract.
  3. Introduction
  4. Patients and methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Conflict of interest statement
  9. Acknowledgements
  10. References

DNA methylation in relation to clinical characteristics

Clinical characteristics of the patient groups are presented in Table 1. There was no correlation between HpaII/MspI ratio and storage time of DNA in the freezer. No significant correlations were observed between age and HpaII/MspI ratios in either group (Groups 1–3). Moreover, no correlations were observed between estimated GFR, BMI, cholesterol and HpaII/MspI. However, when 97 males and 58 female patients were compared, the HpaII/MspI ratio was slightly higher (0.59 ± 0.01 vs. 0.56 ± 0.01; P < 0.05) in males. No differences in HpaII/MspI (0.58 ± 0.01 vs. 0.57 ± 0.01) ratios were observed between 52 diabetic and 103 nondiabetic patients. Similarly, no differences in the HpaII/MspI (0.58 ± 0.01 vs. 0.58 ± 0.01) ratio were observed between nonsmokers versus smokers/ex-smokers. On the other hand, a significantly lower ratio of the HpaII/MspI (0.55 ± 0.01 vs. 0.60 ± 0.01; P < 0.01) ratio, indicating DNA hypermethylation, was observed in 58 patients with CVD vs. 97 patients without clinical signs of CVD. One hundred and seven well-nourished patients (SGA 1) had a significantly higher ratio of HpaII/MspI (0.59 ± 0.01 vs. 0.55 ± 0.01; P < 0.01) than 43 malnourished patients (SGA ≥2). Data on SGA was missing in five of the patients.

DNA methylation in relation to inflammation and oxidative stress biomarkers

Inflamed patients (n = 62) had a significantly lower ratio of HpaII/MspI (0.54 ± 0.01 vs. 0.61 ± 0.01; P < 0.0001) compared with 93 noninflamed patients, which indicates DNA hypermethylation in their peripheral blood cells. Notably, control subjects had HpaII/MspI ratios similar to those observed in noninflamed CKD patients (Fig. 1). When we used GLM to adjust for the influence of inflammation, we found that the HpaII/MspI ratios were still significantly (P < 0.05) lower in inflamed than noninflamed patients. Significant correlations were observed between hsCRP and HpaII/MspI (Rho = −0.30; P < 0.001) ratios. Importantly, these correlations remained strong even if persistently inflamed prevalent HD patients (Group 3) were excluded from analysis (Rho = −0.27; P < 0.01). Whereas inverse correlations (Fig. 2) were observed between the HpaII/MspI ratio and IL-6 (Rho = −0.19; P < 0.05), as well as leucocytes (Rho = −0.32; P < 0.0001), and serum albumin (Rho = 0.21; P < 0.05), the correlations between HpaII/MspI and circulating levels of both fibrinogen (data not shown) and TNF-α did not attain statistical significance.

image

Figure 1.  Box plots showing high sensitivity C-reactive protein levels (top), HpaII/MspI ratios (bottom) in CKD stages 3 and 4 patients (Group 1), CKD stage 5 patients without (Group 2A) or with (Group 2B) inflammation, haemodialysis patients with persistent inflammation (Group 3) and controls (Group 4).

Download figure to PowerPoint

image

Figure 2.  Correlations (Spearman rank) between HpaII/MspI ratios and high sensitivity C-reactive protein, interleukin-6, leucocytes and serum albumin levels in 155 patients with advanced renal disease (Groups 1–3).

Download figure to PowerPoint

Markedly elevated serum levels of 8-OHdG were observed in Groups 1–3 compared with controls (Table 1). As expected, inflamed CKD 5 and inflamed HD patients had the highest levels of this surrogate marker of oxidative stress (Table 1). Whereas no significant correlations were observed between 8-OHdG and HpaII/MspI ratios in the combined patient group (Groups 1–3) significant positive correlations were observed between 8-OHdG and both IL-6 (Rho = 0.47; P < 0.0001), hsCRP (Rho = 0.48; P < 0.0001), serum albumin (Rho = −0.34, P < 0.001) and leucocytes (Rho = 0.26; P < 0.01).

DNA methylation in relation to homocysteine and folic acid

Compared with controls, markedly elevated serum levels of homocysteine were observed in renal patients (Fig. 3). However, inflamed CKD stage 5 (Group 2B) and inflamed HD (Group 3) patients had significantly lower serum levels of homocysteine compared with noninflamed CKD stage 5 (Group 2A) patients (Fig. 3). As expected, a strong positive correlation was observed between serum homocysteine and serum albumin (Rho = 0.29; P < 0.001). Well-nourished (SGA 1) CKD patients had higher (32.7 ± 2.3 vs. 23.3 ± 2.3 μmol L−1; P < 0.05) serum homocysteine levels than malnourished (SGA ≥2) patients. Inflamed patients had lower homocysteine levels (25.6 ± 1.8 vs. 32.6 ± 2.6 μmol L−1; P < 0.05) than noninflamed patients. Whereas inverse correlations between homocysteine levels and estimated GFR (Rho = −0.35; P < 0.0001) as well as IL-6 (Rho = −0.20; P < 0.05) were observed, the correlations between homocysteine and folic acid (Rho = −0.16) and hsCRP (Rho = −0.12) did not attain statistical significance. Estimations of GFR and folic acid measurements were missing in the prevalent HD patients (Group 3). In the whole patient population, the correlations between homocysteine levels and HpaII/MspI ratio (Rho = 0.08; NS) did not attain statistical significance. However, in CKD 5 patients (Group 2) a weak, but significant (Rho = 0.21; P < 0.05), positive correlation was observed between homocysteine and HpaII/MspI (Fig. 4). The correlations between folic acid and the HpaII/MspI ratio, were not significant in both the combined patient group and the CKD 5 subgroup.

image

Figure 3.  Box plot showing serum homocysteine levels in CKD stages 3 and 4 patients (Group 1), CKD stage 5 patients without (Group 2A) or with (Group 2B) inflammation, haemodialysis patients with persistent inflammation (Group 3) and controls (Group 4).

Download figure to PowerPoint

image

Figure 4.  Correlation (Spearman rank) between HpaII/MspI ratio and homocysteine in CKD 5 patients indicating that hyperhomocysteinaemia is associated with global DNA hypomethylation.

Download figure to PowerPoint

DNA methylation in relation to clinical outcome

All CKD stage 5 patients (Group 2) were followed prospectively for a period of 36 ± 2 months following the initiation of dialysis treatment. Significantly lower ratios of the HpaII/MspI (0.54 ± 0.01 vs. 0.59 ± 0.01; P < 0.01) ratio was observed in 26 nonsurvivors (mortality rate 9% per year). Moreover, a significantly lower HpaII/MspI ratio (0.53 ± 0.02 vs. 0.58 ± 0.01; P < 0.05) was observed in 13 patients who died of cardiovascular causes. By using Kaplan–Meier survival curves we assessed the association between DNA methylation and mortality (HpaII/MspI ratios divided into median), and noted a significantly increased all-cause (log-rank 14.4; P < 0.001) and CVDs (log-rank 10.5; P < 0.01) mortality in CKD stage 5 patients with HpaII/MspI ratio <median. When patients were divided according to median HpaII/MspI, the adjustment for potential confounding factors (age, DM, inflammation and CVD) using Cox proportional hazards model revealed a significant (likelihood ratio 38.3; P < 0.001) difference in survival between the two groups (Fig. 5).

image

Figure 5.  Unadjusted (a) and adjusted (b) all-cause mortality in 98 CKD stage 5 patients starting renal replacement therapy.

Download figure to PowerPoint

The Cox proportional hazard model was used to adjust both all-cause and cardiovascular mortality for age (≥58 years), gender (male versus female), DM, CVD, mode of dialysis (PD versus HD), inflammation (CRP: ≥10 mg L−1) and DNA hypermethylation (HpaII/MspI <median). By using univariate analysis all variables, except gender and mode of dialysis treatment, were significantly associated with all-cause mortality as shown in Table 2. Inflammation was closely associated (P = 0.06) with CVD mortality in univariate analysis. When we performed a multivariate Cox regression analysis, including only the variables that were significant in the univariate analysis, HpaII/MspI ratio <median remained significantly associated with both all-cause (P < 0.01) and CVD mortality (P < 0.05), whereas CVD lost its association with outcome (Table 2). In fact, global DNA hypermethylation was the strongest independent risk factor for CVD mortality.

Table 2.   Unadjusted and adjusted hazard ratios (HR) for all-cause and cardiovascular mortality in 98 CKD stage 5 patients starting dialysis treatment (Group 2)
VariableUnadjusted HR (95% CI)P-valueAdjusted HR (95% CI)P-value
  1. PD, peritoneal dialysis; HD, haemodialysis. • not applicable.

All-cause mortality
 Age (>median 58 years)8.6 (2.9–24.9)<0.0016.0 (1.9–18.5)<0.01
 DNA hypermethylation (HpaII/MspI <median)6.1 (2.1–17.7)<0.0015.0 (1.7–14.8)<0.01
 Cardiovascular disease3.0 (1.4–6.5)<0.011.1 (0.4–2.6)NS
 Diabetes mellitus2.6 (1.2–5.7)<0.052.5 (1.1–5.5)<0.05
 Inflammation (CRP ≥10 mg L−1)2.3 (1.0–5.3)<0.051.2 (0.5–2.8)NS
 Mode of dialysis (PD versus HD)1.5 (0.7–3.3)NS
 Male gender0.8 (0.4–1.7)NS
Cardiovascular mortality
 DNA hypermethylation (HpaII/MspI <median)13.4 (1.7–103.1)<0.0513.9 (1.8–109.3)<0.05
 Age (>median 58 years)7.7 (1.7–34.6)<0.014.9 (0.99–23.2)NS
 Diabetes mellitus4.1 (1.3–12.5)<0.054.6 (1.4–15.0)<0.05
 Cardiovascular disease3.2 (1.1–9.7)<0.051.0 (0.3–3.6)NS
 Inflammation (CRP ≥10 mg L−1)3.1 (0.99–10.0)NS
 Mode of dialysis (PD versus HD)1.4 (0.4–4.2)NS
 Male gender0.8 (0.3–2.4)NS

Reproducibility of LUMA

To investigate the reproducibility of the LUMA method, randomly selected DNA samples (n = 30) from peripheral blood leucocytes were assayed on four independent occasions. Although DNA methylation levels were different between individual samples, for all cases, SEM of the assay was below 10%, suggesting a significant robustness of the method (Fig. 6).

image

Figure 6.  Reproducibility of the Luminometric Methylation Assay (LUMA). Genomic DNA extracted from normal peripheral blood leucocytes of 30 healthy individuals assayed four times with the LUMA method. Mean values of HpaII/MspI are plotted as squares with error bars denoting ±SEM.

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Abstract.
  3. Introduction
  4. Patients and methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Conflict of interest statement
  9. Acknowledgements
  10. References

In the present study we investigate, for the first time, the effect of declining renal function and inflammation on global DNA methylation and its relation to outcome in well-characterized CKD patients. A main finding of the present study is the strong association between surrogate markers of inflammation and DNA methylation in peripheral blood leucocytes. Whereas CKD patients without signs of overt inflammation had HpaII/MspI ratios comparable with age- and gender-matched controls, inflamed CKD stage 5 patients and prevalent HD patients exhibited DNA hypermethylation.

Although we have yet only began to scratch the surface of the roles and interactions of abnormal DNA methylation in relation to inflammation and immunity, evidence suggests that inflammation may cause aberrant DNA methylation [18]. Chronic inflammation has been reported to be one factor associated with increased DNA methylation in both chronic gastritis [19] and gastric cancer [20]. Further support for the concept that inflammation drives hypermethylation may be derived from an in vitro study by Hodge et al. [21] who demonstrated that the inflammatory cytokine IL-6 might exert an impact on epigenetic changes in cells via regulation of a DNA methyltransferase gene. Subsequently, the same group demonstrated in vitro that IL-6 is capable of maintaining promotor methylation [22]. In accordance, we observed a significant increase in global DNA methylation when human lymphocytes were treated with IL-6 (1 ng mL−1) for 24 h (data not shown). However, given the cross-sectional nature of our study, we cannot exclude that DNA hypermethylation may induce an inflammatory response. Indeed, as suppressors of cytokine signalling (SOCS) are inactivated by hypermethylation [23] it could be speculated that aberrant DNA methylation, via inhibition of SOCS, contributes to exaggerated cytokine signalling. Thus, persistent inflammation may cause aberrant DNA methylation, which via an inhibition of SOCS further exaggerates IL-6 signalling, perpetuating a vicious circle. Clearly, longitudinal studies evaluating DNA methylation in response to various nutritional and pharmacological anti-inflammatory treatment strategies are needed to resolve the complex interactions between inflammation and aberrant DNA methylation.

Markedly elevated serum levels of homocysteine were observed in CKD patients and there was a strong inverse correlation between homocysteine and GFR, confirming previous results [3]. Moreover, as we found significantly lower homocysteine levels in inflamed and wasted CKD patients our results support the concept that the presence of an inflammation-wasting syndrome contributes to the well-documented counter-intuitive association between low homocysteine levels and worse clinical outcome in this patient group [3, 24]. As homocysteine is bound to serum albumin, lower levels of homocystein are usually found in clinical states characterized by hypoalbuminaemia, such as persistent inflammation [3]. It has previously been demonstrated that hyperhomocysteinaemia is associated with DNA hypomethylation through an increase in S-adenosylhomocysteine levels [25]. Moreover, Ingrosso et al. [26] demonstrated that hyperhomocysteinaemia is associated with DNA hypomethylation in HD patients. In accordance with the present study hyperhomocysteinaemia was associated with higher ratios of HpaII/MspI (indicating DNA hypomethylation) in CKD 5 patients. Whereas we studied a large cohort of both incident CKD and prevalent HD patients with and without inflammation and systemic diseases. Ingrosso et al. [26] studied a selected group of 32 prevalent HD patients without systemic disease and/or diabetes mellitus. Thus, the results of these two studies may not be readily compared.

A novel finding of the present study was the association between DNA hypermethylation on both all-cause and cardiovascular mortality in incident CKD stage 5 patients. As ageing and atherosclerosis are considered major independent risk factors for cardiovascular morbidity and mortality both in the nonrenal and renal patient population, we found it remarkable that the association between CVD and mortality in univariate analysis was lost when we adjusted for the impact of DNA methylation (Table 2). This observation may suggests that aberrant DNA methylation is a determinant, or marker, of atherosclerotic CVD in this population. Consequently, epigenetic modifications may be more important in the pathogenesis of atherosclerosis than genetic polymorphisms [9]. It could be speculated that aberrant DNA methylation may contribute to accelerated atherosclerosis by upregulation of atherosclerosis-susceptible and downregulation of atherosclerosis-protective genes. Indeed, methylation-associated inactivation of the oestrogen receptor-α gene, has been demonstrated in vitro to play a role in atherogenesis and ageing of the vascular system [27]. Furthermore, Zhu et al. [28] recently demonstrated that DNA methylation might modify monocarboxylate transport (inadequate monocarboxylate transport may affect the development of atherosclerotic lesions) by suppressing monocarboxylate transporter expression. Thus, this is yet another mechanism by which aberrant DNA methylation may regulate smooth muscle cell function and contribute to atherosclerosis. Finally, a study in the genetically atherosclerosis-prone ApoE null mice demonstrate that alterations in DNA methylation profiles, including both hypermethylation and hypomethylation, were present in aortas before atherosclerotic lesions were present [29]. Taken together, the results of the present and previous studies indicate that aberrant DNA methylation is an important process through which novel factors for atherosclerosis, including proinflammatory cytokines and hyperhomocysteinaemia, affect the function of the genome.

Some limitations of the present study should be addressed. At first, in the CKD stages 3 and 4 and the CKD stage 5 patient groups, the definition of inflammation was based on only one single CRP determination. As inflammatory biomarkers vary with time [30], a better definition based on several biomarkers should be established. However, as we found DNA hypermethylation also in persistently (during 3 months) inflamed HD patients our findings support the concept that inflammation and DNA hypermethylation are associated. Although, some may argue that the cut-off point for defining inflammation (CRP: >10 mg L−1) is high, we previously used receiver operating characteristics (ROC) to demonstrate that this specific CRP cut-off is the best for differentiating between survivors and nonsurvivors in a group of 663 CKD patients [31]. In addition, as the cardiovascular risk factor profile may be different in patients with advanced CKD compared with the general population [7], a similar study should be carried out on cardiovascular risk factors in other patient groups before the present results are applied to the general population. We are currently addressing this point. Furthermore, as the number of patients and events in the outcome substudy was limited, the significant association between global hypermethylation status and outcome (Table 2) must be interpreted with caution and needs to be confirmed in larger cohorts of patients. Finally, as DNA was only taken from peripheral blood cells our study does not reveal any information about regional tissue methylation status in relation to inflammation and CVD.

Conclusions

  1. Top of page
  2. Abstract.
  3. Introduction
  4. Patients and methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Conflict of interest statement
  9. Acknowledgements
  10. References

We show, for the first time, that global DNA hypermethylation is correlated with systemic inflammation in CKD patients. As our findings in CKD 5 patients support the concept that hyperhomocysteinaemia may be associated with global hypomethylation further studies are needed to evaluate if inflammation and hyperhomocysteinaemia have opposing effects on the DNA methylation process. Moreover, our result demonstrates an association between global DNA hypermethylation and outcome in patients starting dialysis treatment. As epigenetic DNA modifications are potentially reversible, there is a possibility that future therapies directed at modifying the epigenome may have favourable effects on cardiovascular outcomes in this patient group.

Acknowledgements

  1. Top of page
  2. Abstract.
  3. Introduction
  4. Patients and methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Conflict of interest statement
  9. Acknowledgements
  10. References

This study was supported by unconditional grants from Söderberg Foundation (PS), Swedish Medical Research Council (PS), Swedish Cancer Foundation (TJE), Baxter Healthcare (MS), Amgen (PS), Swedish Society of Medicine (PS and MS), the Karolinska Institutet and Hospital (PS), the Swedish Labor Market Insurance AFA (TJE) and a conditional salary grant from the Stockholm County Council (PS).

References

  1. Top of page
  2. Abstract.
  3. Introduction
  4. Patients and methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Conflict of interest statement
  9. Acknowledgements
  10. References
  • 1
    Foley RN, Parfrey PS, Sarnak MJ. Clinical epidemiology of cardiovascular disease in chronic renal failure. Am J Kidney Dis 1998; 32 (Suppl. 5): S1129.
  • 2
    Himmelfarb J, Stenvinkel P, Ikizler TA, Hakim RM. The elephant of uremia: oxidative stress as a unifying concept of cardiovascular disease in uremia. Kidney Int 2002; 62: 152438.
  • 3
    Suliman ME, Barany P, Kalantar-Zadeh K, Lindholm B, Stenvinkel P. Homocysteine in uraemia – a puzzling and conflicting story. Nephrol Dial Transpl 2005; 20: 1621.
  • 4
    Stenvinkel P, Ketteler M, Johnson RJ et al. Interleukin-10, IL-6 and TNF-α: important factors in the altered cytokine network of end-stage renal disease – the good, the bad and the ugly. Kidney Int 2005; 67: 121633.
  • 5
    Zimmermann J, Herrlinger S, Pruy A, Metzger T, Wanner C. Inflammation enhances cardiovascular risk and mortality in hemodialysis patients. Kidney Int 1999; 55: 64858.
  • 6
    Pecoits-Filho R, Barany B, Lindholm B, Heimbürger O, Stenvinkel P. Interleukin-6 and its receptor is an independent predictor of mortality in patients starting dialysis treatment. Nephrol Dial Transpl 2002; 17: 16848.
  • 7
    Kalantar-Zadeh K, Block G, Humphreys MH, Kopple JD. Reverse epidemiology of cardiovascular risk factors in maintenance dialysis patients. Kidney Int 2003; 63: 793808.
  • 8
    Nephew KP, Huang TH. Epigenetic gene silencing in cancer initiation and progression. Cancer Lett 2003; 190: 12533.
  • 9
    Dong C, Yoon W, Goldsmith-Clermont PJ. DNA methylation and atherosclerosis. J Nutr 2002; 132: 2406S9S.
  • 10
    Erlinger T, Platz E, Rifai N, Helzlsouer KJ. C-reactive protein and the risk of incident colorectal cancer. JAMA 2004; 291: 58590.
  • 11
    Helzlsouer KJ, Erlinger TP, Platz EA. C-reactive protein levels and subsequent cancer outcomes: results from a prospective cohort study. Eur J Cancer 2006; 42: 7047.
  • 12
    Karimi M, Johansson S, Ekström TJ. Using LUMA: a luminometric based assay for global DNA-methylation. Epigenetics 2006; 1: 458.
  • 13
    Karimi M, Johansson S, Stach D et al. LUMA (LUminometric Methylation Assay) – a high throughput method to the analysis of genomic DNA methylation. Exp Cell Res 2006; 312: 198995.
  • 14
    National Kidney Foundation. KDOQI clinical practice guidelines for nutrition in chronic renal failure. Am J Kidney Dis 2000; 35 (Suppl. 2): S1S140.
  • 15
    Stenvinkel P, Heimbürger O, Paultre F et al. Strong association between malnutrition, inflammation, and atherosclerosis in chronic renal failure. Kidney Int 1999; 55: 1899911.
  • 16
    Ubbink JB, Hayward Vermaak WJ, Bissbort S. Rapid high-performance liquid chromatographic assay for total homocysteine levels in human serum. J Chromatogr 1991; 565: 4416.
  • 17
    Detsky AS, McLaughlin JR, Baker JP et al. What is subjective global assessment of nutritional status? J Parenter Enterol Nutr 1987; 33: 6503.
  • 18
    Teitell M, Richardson B. DNA methylation in the immune system. Clin Immunol 2003; 109: 25.
  • 19
    Kang GH, Lee HJ, Hwang KS, Lee S, Kim JH, Kim JS. Aberrant CpG island hypermethylation of chronic gastritis, in relation to aging, gender, intestinal metaplasia, and chronic inflammation. Am J Pathol 2003; 163: 15516.
  • 20
    Jang TJ, Kim DI, Shin YM, Chang HK, Yang CH. p16(INK4a) promoter hypermethylation of non-tumorous tissue adjacent to gastric cancer is correlated with glandular atrophy and chronic inflammation. Int J Cancer 2001; 93: 62934.
  • 21
    Hodge DR, Xiao W, Clausen PA, Heidecker G, Szyf M, Farrar WL. Interleukin-6 regulation of the human DNA metyltransferase (HDNMT) gene in human erythroleukemia cells. J Biol Chem 2001; 276: 3950811.
  • 22
    Hodge DR, Peng B, Cherry JC et al. Interleukin 6 supports the maintenance of p53 tumor suppressor gene promotor methylation. Cancer Res 2005; 65: 467382.
  • 23
    Galm O, Yoshikawa H, Esteller M, Osieka R, Herman JG. SOCS-1, a negative regulator of cytokine signalling, is frequently silenced by methylation in multiple myeloma. Blood 2003; 101: 27848.
  • 24
    Zoccali C, Mallamaci F. Homocysteine and risk in end-stage renal disease: a matter of context. Kidney Int 2006; 69: 2046.
  • 25
    Yi P, Melnyk S, Pogribna M, Pogrinby IP, Hine RJ, James SJ. Increase in plasma homocysteine associated with parallel increases in plasma S-adenosylhomocysteine and lymphocyte DNA hypomethylation. J Biol Chem 2000; 275: 2931823.
  • 26
    Ingrosso D, Cimmino A, Perna AF et al. Folate treatment and unbalanced methylation and changes of allelic expression induced by hyperhomocysteinemia in patients with uremia. Lancet 2003; 361: 16939.
  • 27
    Post WS, Goldschmidt-Clermont PJ, Wilhide CC et al. Methylation of the estrogen receptor gene is associated with aging and atherosclerosis in the cardiovascular system. Cardiovasc Res 1999; 43: 98591.
  • 28
    Zhu S, Goldsmith-Clermont PJ, Dong C. Inactivation of monocarboxylate transporter MCT3 by DNA methylation in atherosclerosis. Circulation 2005; 112: 135361.
  • 29
    Lund G, Andersson L, Lauria M et al. DNA methylation polymorphisms precede any histological sign of atherosclerosis in mice lacking apolipoprotein E. J Biol Chem 2004; 279: 2914754.
  • 30
    Kaysen GA, Dublin JA, Müller HG, Rosales LM, Levin NW. The acute-phase response varies with time and predicts serum albumin levels in hemodialysis patients. Kidney Int 2000; 58: 34652.
  • 31
    Stenvinkel P, Wanner C, Metzger T et al. Inflammation and outcome in end-stage renal failure: does female gender constitute a survival advantage? Kidney Int 2002; 62: 17918.