Ceruloplasmin and atrial fibrillation: evidence of causality from a population-based Mendelian randomization study

Authors


Abstract

Objectives

Inflammatory diseases and inflammatory markers secreted by the liver, including C-reactive protein (CRP) and ceruloplasmin, have been associated with incident atrial fibrillation (AF). Genetic studies have not supported a causal relationship between CRP and AF, but the relationship between ceruloplasmin and AF has not been studied. The purpose of this Mendelian randomization study was to explore whether genetic polymorphisms in the gene encoding ceruloplasmin are associated with elevated ceruloplasmin levels, and whether such genetic polymorphisms are also associated with the incidence of AF.

Design

Genetic polymorphisms in the ceruloplasmin gene (CP) were genotyped in a population-based cohort study of men from southern Sweden (Malmö Preventive Project; = 3900). Genetic polymorphisms associated with plasma ceruloplasmin concentration were also investigated for association with incident AF (= 520) during a mean follow-up of 29 years in the same cohort. Findings were replicated in an independent case–control sample (The Malmö AF cohort; = 2247 cases, 2208 controls).

Results

A single nucleotide polymorphism (rs11708215, minor allele frequency 0.12) located in the CP gene promoter was strongly associated with increased levels of plasma ceruloplasmin (= 9 × 10−10) and with AF in both the discovery cohort [hazard ratio 1.24 per risk allele, 95% confidence interval (CI) 1.06–1.44, = 0.006] and the replication cohort (odds ratio 1.13, 95% CI 1.02–1.26, = 0.02).

Conclusions

Our findings indicate a causal role of ceruloplasmin in AF pathophysiology and suggest that ceruloplasmin might be a mediator in a specific inflammatory pathway that causally links inflammatory diseases and incidence of AF.

Abbreviations
AF

atrial fibrillation

CI

confidence interval

CRP

C-reactive protein

BMI

body mass index

HR

hazard ratio

IL-6

interleukin-6

LDL

low-density lipoprotein

NFκB

nuclear factor kappa B

OR

odds ratio

SNP

single nucleotide polymorphism

Introduction

Relationships between inflammatory diseases, raised plasma concentrations of inflammatory markers and incidence of atrial fibrillation (AF) have been documented in several prospective studies [1-8]. These associations could potentially result either from inflammatory processes in the atria in inflammatory conditions or from arrhythmogenic effects of inflammatory mediators, which are increased in blood in inflammatory states.

The first inflammatory marker to be associated with AF, C-reactive protein (CRP) [4], was investigated for a causal relation with AF in a large Mendelian randomization study, that is, to determine whether genetic variants that confer increased CRP levels also confer increased risk of AF; the results from this study did not support a causal relationship [5]. We recently found in the Malmö Preventive Project that amongst a panel of seven inflammatory markers, only ceruloplasmin was independently associated with increased AF risk [9]. Ceruloplasmin is an acute-phase reactant that is involved in the metabolism of iron and is an important carrier of circulating copper. Both oxidative and anti-oxidative functions have been reported for ceruloplasmin, and it may induce LDL oxidation as well as protect against free radical proteolysis [10, 11]. The findings of several studies indicate that oxidative stress might play a central role in the initiation and maintenance of AF mediated by electrophysiological remodelling of the atrium [12-14]. It remains unclear whether ceruloplasmin acts as a passive marker of inflammation or as a causal mediator in the development of AF.

The purpose of this study was to further explore the relationship between ceruloplasmin and AF risk using Mendelian randomization methods. We investigated whether genetic polymorphisms in the gene encoding ceruloplasmin are associated with elevated ceruloplasmin levels, and whether such genetic polymorphisms are also associated with incidence of AF.

Methods

Discovery cohort

Between 1974 and 1984, a total of 22 444 men aged 26–61 years participated in a screening programme for the detection of individuals at high risk of cardiovascular diseases in Malmö, Sweden (The Malmö Preventive Project). The participation rate was 71%. As a part of the programme, plasma levels of ceruloplasmin were determined for a random sample of 6455 men [9, 15, 16]. Of these men, 3907 were later re-examined in follow-up studies including DNA sampling [17]; the mean age at baseline was 46.3 ± 3.6 years. After excluding seven subjects with a previous hospital diagnosis of AF, 3900 men remained in the cohort.

The health service authority of Malmö approved and funded the screening programme. The regional ethics committee approved the data linkage with the Swedish population and hospital registers (No. 85/2004, LU 244-02). All participants provided written informed consent.

Baseline examination

Blood samples were drawn after an overnight fast. Levels of serum cholesterol were determined using standard methods at the laboratory of Malmö University Hospital. Diabetes was defined as fasting whole-blood glucose concentration ≥6.1 mmol L−1, 2-h postglucose load ≥10.0 mmol L−1 or self-reported diabetes. Body mass index (BMI) was calculated as body weight divided by height squared. Blood pressure (mm Hg) was measured twice in the right arm using a sphygmomanometer after resting for 10 min. Data on smoking habits were collected from a questionnaire.

Plasma levels of ceruloplasmin were assessed by means of an electroimmunoassay [18]. The analysis was performed consecutively at the time of study entry. The lower detection limit for ceruloplasmin was 20 mg L−1, and the coefficient of variation was <5% [18].

Follow-up

All men were followed from the baseline examination until first hospitalization due to AF, death, emigration from Sweden or 31 December 2008. AF was defined as a diagnosis of either atrial fibrillation or atrial flutter, given the close inter-relationship between these disorders, as in previous studies [6, 19]. Subjects were considered to have AF if diagnosed with a primary or contributory hospital discharge code according to the International Classification of Diseases 8th revision (ICD-8) 427.92, ICD-9 427D or ICD-10 I48. Because a relationship between ceruloplasmin and AF could in theory be mediated via an increased risk of other types of heart disease, we also performed an analysis in which subjects with a hospital diagnosis of myocardial infarction or heart failure prior to or at the time of the diagnosis of AF were censored. In this analysis, the follow-up time was considered only up to the time of the myocardial infarction or heart failure event. The Swedish Hospital Discharge Register was used for case retrieval [20].

Selection, genotyping and quality control of single nucleotide polymorphisms

We selected 17 tag single nucleotide polymorphisms (SNPs) across the ceruloplasmin gene (CP) and genotyped them using IPLEX on a MassARRAY platform (Sequenom, San Diego, CA, USA) according to the manufacturer's standard protocols. Twenty per cent of the samples were run in duplicate without any inconsistencies. All genotypes were called by two investigators.

Replication in independent samples

SNPs significantly associated with AF in the discovery analyses were tested for replication in a case–control study, the Malmö AF cohort. This cohort consisted of 4826 unique subjects (62% male) drawn from the Malmö Diet and Cancer Study [19] and re-examinations from the Malmö Preventive Project as described previously [21]. Cases were subjects with a diagnosis of AF before 31 December 2006; controls were matched according to sex, age (±1 year), time of baseline examination (±1 year) and follow-up time. A total of 371 subjects (166 cases and 205 controls) from the Malmö Preventive Project who were also included in the discovery analyses were excluded from the replication study. After exclusions, there were 4455 subjects (59% male) in the replication analysis.

We also attempted to perform in silico replication of our findings using data from the AFGen study, a meta-analysis of genomewide association studies of individuals of European ancestry [21]. However, neither the SNP of interest nor any appropriate proxy was available in this data set.

Statistical analysis

Univariate analysis of variance (anova) was used to compare plasma levels of ceruloplasmin across genotypes and also to compare risk factor distributions for the different SNPs. Associations between SNPs and AF were examined using Cox proportional hazards regression, in which the additive effect per copy of the minor allele on AF risk was tested. The analysis was adjusted for age, sex, smoking, diabetes, BMI and systolic blood pressure. The Cox proportional hazard assumption was confirmed by plotting incidence rates over time.

Logistic regression analysis was used to examine the association between ceruloplasmin and SNPs in the case–control cohort. An additional sensitivity analysis was also performed after censoring individuals with incident myocardial infarction and/or heart failure. The Hardy–Weinberg equilibrium was calculated using a Web-based calculator [22]. All analyses were performed using pasw statistics 18 (SPSS Inc, Chicago, IL, USA) or stata/ic 12.1 (StataCorp, College Station, TX, USA).

Results

Genetic determinants of plasma levels of ceruloplasmin

The baseline characteristics of the discovery study cohort are presented in Table 1. Baseline levels of ceruloplasmin differed significantly (< 0.05) between genotypes in nine of 17 SNPs (two degrees of freedom). The strongest association between levels of ceruloplasmin and genotype was observed for rs13075891 (= 2 × 10−10) and rs11708215 (= 9 × 10−10). These two SNPs also showed the strongest pairwise correlation (r2 = 0.56; Table S1). When the SNPs were fitted in an additive model (one degree of freedom), six SNPs remained significantly associated with ceruloplasmin levels (Table 2). The models were adjusted for age, smoking, cholesterol, systolic blood pressure, diabetes and BMI. Age, cholesterol and smoking were also significantly and positively associated with plasma levels of ceruloplasmin.

Table 1. Baseline characteristics of the study cohorts
 Malmö Preventive ProjectMalmö AF cohort
Incident AFAF casesControls
YesNo
  1. AF, atrial fibrillation; BMI, body mass index.

  2. Values are presented as means ± standard deviation unless otherwise stated.

n 520338022472208
Age (years)47 ± 346 ± 465 ± 865 ± 8
Male sex (%)10010059.658.8
Current smoker (%)42.341.842.442.6
Diabetes (%)2.93.37.75.0
Systolic blood pressure (mmHg)131 ± 16127 ± 15147 ± 20148 ± 21
BMI (kg m−2)25 ± 325 ± 326 ± 427 ± 4
Table 2. Associations between single nucleotide polymorphisms (SNPs) and plasma levels of ceruloplasmin
SNPNumber of subjectsMinor allele frequencyP-value for relationship with ceruloplasmin (two degrees of freedom)aP-value for relationship with ceruloplasmin (one degree of freedom)a r 2
  1. a

    Adjusted for age, smoking, cholesterol level, systolic blood pressure, diabetes and body mass index.

rs1170821538310.129 × 10−100.0010.051
rs1170971438300.498 × 10−51 × 10−50.046
rs1171400037640.072 × 10−90.120.051
rs1307589137940.102 × 10−100.0020.052
rs1309576438160.050.0150.860.043
rs1686157938330.200.0170.950.043
rs1686158237840.300.0070.0160.043
rs1719550538060.040.0050.0260.043
rs1783883138340.153 × 10−60.0060.047

Genetic variants and AF

During a mean follow-up of 28.8 years, 520 men were hospitalized with AF. One SNP, rs11708215, showed a significant association with incidence of AF, with a hazard ratio (HR) of 1.24 [95% confidence interval (CI) 1.06–1.44] per copy of the minor C allele (= 0.006; Table 3). The relationship remained significant after adjustment for potential covariates (age, smoking, cholesterol level, systolic blood pressure, diabetes and BMI) and was even stronger if individuals with incident heart failure or myocardial infarction prior to the diagnosis of AF were censored in the analysis (HR 1.38 per risk allele, 95% CI 1.18–1.63, = 9.7 × 10−5). Allele characteristics for rs11708215 are shown in Table 4.

Table 3. Associations between single nucleotide polymorphisms (SNPs) and incidence of atrial fibrillation (AF) in three models
SNPHR for AFaHR for AFbHR for AFc
  1. HR, hazard ratio; CI, confidence interval.

  2. a

    HR (95% CI) and P-value per allele, crude model.

  3. b

    HR (95% CI) and P-value per allele, adjusted for age, smoking, diabetes, body mass index, cholesterol level, systolic blood pressure and ceruloplasmin level.

  4. c

    HR (95% CI) and P-value per allele, after adjustments for above risk factors; subjects with a hospital diagnosis of heart failure or myocardial infarction before incidence of AF were censored (= 126).

rs117082151.24 (1.06–1.44); = 0.0061.23 (1.06–1.43); = 0.0071.38 (1.18–1.63); = 9.7 × 10−5
rs117097141.07 (0.95–1.21)  
rs130758911.21 (1.00–1.46); = 0.051.18 (0.98–1.43); = 0.09 
rs168615821.12 (0.98–1.28)  
rs171955050.86 (0.61–1.20)  
rs178388311.15 (0.98–1.36)  
Table 4. Allele characteristics for the single nucleotide polymorphism rs11708215
 GenotypeP trend
TTCTCC
  1. AF, atrial fibrillation; CI, confidence interval.

  2. a

    Adjusted for age, smoking, body mass index, diabetes, cholesterol level, systolic blood pressure and plasma level of ceruloplasmin.

n 3080581170 
Plasma level of ceruloplasmin, μmol L−13073243250.001
AF, n (%)389 (12.6)93 (16.0)30 (17.7) 
Incidence (95% CI) of AF, hazard ratio1.01.27 (1.02–1.60)1.47 (1.02–2.14)0.006
Incidence (95% CI) of AF, adjusted hazard ratioa1.01.25 (0.99–1.57)1.50 (1.03–2.17)0.007

Replication in an independent sample

The relationship between rs11708215 and ceruloplasmin remained significantly associated with AF in the replication study (the Malmö AF cohort; Table 5) with an odds ratio (OR) of 1.13 (95% CI 1.02–1.26) per minor allele (= 0.017). After excluding 219 individuals with prior myocardial infarction or heart failure and adjustment for potential risk factors, the OR was 1.16 (95% CI 1.04–1.29, = 0.008).

Table 5. Association between single nucleotide polymorphism (SNP) rs11708215 and atrial fibrillation (AF) in the Malmö AF cohort
SNPOR (95% CI) of AF per alleleOR (95% CI) of AF per alleleaOR (95% CI) of AF per allelea,b
  1. OR, odds ratio; CI, confidence interval.

  2. P-values per allele.

  3. a

    Adjusted for age, sex, smoking, diabetes, body mass index and systolic blood pressure.

  4. b

    Subjects with a history of myocardial infarction and/or heart failure before AF diagnosis were excluded (= 219).

rs117082151.13 (1.02–1.26); = 0.021.14 (1.03–1.27); = 0.021.16 (1.04–1.29); = 0.008

Discussion

In this Mendelian randomization study, we showed that an SNP (rs11708215) in the promoter region of CP is associated with increased levels of the gene product in blood and with increased risk of AF. The results were replicated and confirmed in an independent cohort in which the relationships remained significant after adjusting for potential covariates and also after exclusion of individuals with a history of heart failure or myocardial infarction.

The results indicate that the SNP rs11708215 in the CP gene might affect the risk of AF by modifying plasma levels of ceruloplasmin. However, adjustments for plasma levels of ceruloplasmin did not eliminate the significant association between the SNP and AF, suggesting that an additional mechanistic pathway might be present. Another explanation, which has also been put forward in other Mendelian randomization experiments with larger effect estimates than expected [23], might be that the association between rs11708215 and AF represents a lifelong genetic risk, whereas the association between ceruloplasmin and AF represents a risk that is present over a more limited time frame.

It is also possible that the relationships between genetic variations in the CP gene, ceruloplasmin and incidence of AF could be explained by other genes in the same chromosomal region. However, this seems less likely as the SNP associated with ceruloplasmin, and AF is located in the promoter region upstream of the CP gene at chromosome 3q23-24. This region is strongly associated with binding of several transcription factors including nuclear factor kappa B (NFκB). NFκB is thought to be involved in altering ion channel transcription, thereby causing electrical remodelling of the heart [24]. A recent study has also shown that common genetic variants of the gene encoding the receptor for interleukin-6 (IL-6) are reproducibly associated with AF risk [25]. Of note, NFκB is activated by cytokines, including by binding of IL-6 to its receptor [26, 27].

Ceruloplasmin is a plasma protein that is considered to be an acute-phase reactant with levels increased by two to threefold in inflammatory conditions [15, 16, 28]. A high level of ceruloplasmin is a risk factor for several cardiovascular disorders including myocardial infarction, arteriosclerosis, unstable angina, abdominal aortic aneurysm, vasculitis and peripheral arterial disease [29]. The protein is thought to function as a copper transporter as it accounts for approximately 95% of the serum copper in healthy adults [30]. It has also been proposed that ceruloplasmin is involved in iron homeostasis, coagulation, angiogenesis and immune function [29].

The findings of several studies also suggest that ceruloplasmin may have effects on heart muscle, which might be mediated through copper. It was shown that left ventricular hypertrophy, commonly observed in combination with AF, was related to accumulation of loosely bound copper in patients with type-2 diabetes [31]. Patients with genetically elevated copper levels (Mb Wilson) also show signs of cardiomyopathy, possibly due to deposition of copper in the heart [32-34], and a relatively high frequency of atrial arrhythmias has been described [35]. By contrast, loss-of-function mutations in the CP gene (aceruloplasminaemia) result in neurodegeneration and accumulation of iron in several organs including the liver and brain, but no reported effects on copper metabolism or cardiac disease [36, 37].

Ceruloplasmin has been reported to posses both oxidative and anti-oxidative functions. It was shown that ceruloplasmin catalyses the oxidation of Fe2+ to Fe3+ [38, 39], as well as the oxidation of Cu+ to Cu2+ [40]. The reaction reduces O2 to H2O without releasing superoxide or hydrogen peroxide. This so-called ferroxidase activity is essential for iron homeostasis and is thought to be responsible for the ability of ceruloplasmin to block free radical-induced proteolysis and DNA damage. The ferroxidase activity is increased, for example, during inflammation and infection [11]. By contrast, ceruloplasmin has been reported to possess oxidative effects and may induce LDL oxidation [10]. Experiments have shown that ceruloplasmin increases lipid oxidation by a process requiring superoxide released from vascular smooth muscle cells and endothelial cells [41, 42]. Removal of one of the seven copper atoms completely blocked oxidant activity [10]. In addition, removal of loosely bound copper from ceruloplasmin may be induced by reactive oxygen species. The free copper may then catalyse the formation of more reactive oxygen species and may also lead to direct effects on signal transduction and transcription [43]. Thus, the role of ceruloplasmin as a protective or pathological protein might be dependent on the oxidative status. However, it remains unclear whether an elevated level of ceruloplasmin is a cause of oxidative stress or instead a compensatory reaction against oxidative stress.

Oxidative stress is thought to play a central role in the initiation and continuation of AF, and several mechanisms might be involved in this process. Reactive oxygen species may cause ectopic firing as well as atrial electrical and structural remodelling by altering ion channels [44, 45]. Reactive oxygen species have been linked to abnormal Ca2+ homeostasis and gap junctions [44, 45]. If ion channels are altered and action potential is decreased by reactive oxygen species, this could cause inexcitable areas which could promote re-entry arrhythmias [44].

We were able to replicate our findings for the rs11708215 SNP in an independent case–control cohort, which further strengthens our hypothesis that ceruloplasmin may cause AF.

There are some potential limitations of our study that should be considered. First, all diagnoses in this study were made during hospitalization, and the cases should therefore be valid. Furthermore, a validation study of cases retrieved from the Swedish national registers showed that the validity of the diagnosis was very high [19]. However, some cases of AF might only be treated in primary care and are therefore not included in this study. Nevertheless, the estimates of prevalence and incidence in a recent validation study of this population were largely comparable with estimates in other epidemiological studies of AF [19]. Secondly, electrocardiographic information was not available at baseline, suggesting that some cases may have had AF already when entering the study. However, because the mean age was only 47 years, and AF is strongly related to age, the number of cases with AF at baseline is assumed to be small. Finally, there was a significant time-lapse between baseline examinations and blood collection for genotyping, which may have lead to a bias as patients with AF may have died before genotyping. However, this would result in an underestimation of the actual risk associated with the minor allele of rs11708215.

In conclusion, our results show that genetic polymorphisms in the promoter of the CP gene are associated with elevated plasma levels of ceruloplasmin. One of these polymorphisms is also reproducibly associated with increased risk of AF. Our findings indicate a causal relationship between ceruloplasmin and AF, but additional work is necessary to elucidate the causal pathway.

Conflict of interest statement

No conflict of interest to declare.

Acknowledgements

This work was supported by grants from the Swedish Heart and Lung Foundation, the Swedish Research Council (2011-3891; SFO EpiHealth), the Region Skåne, Skåne University Hospital Foundation, the Swedish Academy of Pharmaceutical Sciences and the Ernhold Lundström Foundation.

Ancillary