γ-Butyrobetaine, L-carnitine and acylcarnitines and their association with incident atrial fibrillation among men and women with suspected chronic coronary syndrome

Abstract

Aims

Carnitine-metabolites play a key role in cell energy metabolism and have been implicated in the pathogenesis of atrial fibrillation (AF). However, any associations between L-carnitine and the precursor γ-butyrobetaine (γBB) with new-onset AF are not established. We therefore examined the relationship between serum γBB, L-carnitine and selected acylcarnitines with incident AF.

Methods and results

We studied 3815 patients who underwent coronary angiography due to suspected chronic coronary syndrome and followed up for incident CVD events, including new-onset AF. Risk associations between the biomarkers and incident AF were explored by logistic regression for the total population and stratified according to sex. The majority (71.6%) of the patients were men, the median age was 61 years. Over a median follow-up of 7.3 years, 412 (10.8%) were reported with new-onset AF. In a multivariate model higher serum γBB, L-carnitine and palmitoylcarnitine [OR per 1SD (95% CI): 1.15 (1.01–1.31), P = 0.04; 1.14 (1.01–1.27), P = 0.03 and 1.15 (1.03–1.28), P = 0.01] were associated with increased risk of incident AF. These associations were particularly stronger among women (P for interaction < 0.04, all).

Conclusion

In this large cohort of patients with suspected chronic coronary syndrome elevated circulating levels of γBB, L-carnitine and palmitoylcarnitine were associated with increased risk of incident AF, with stronger associations among women. Our results motivate further studies examining the relationship between carnitine metabolites and cardiac arrhythmias and highlight the need for sex specific analyses.

Graphical Abstract

Graphical Abstract.

Introduction

Atrial fibrillation (AF) is the most common clinically significant arrhythmia, with its incidence and prevalence increasing alongside AF-related healthcare costs.¹ Hence, it is of great interest to identify novel risk markers of incident AF.^1,2 Circulating biomarkers currently play a minor role in risk prediction. However, increasing evidence implicates disturbances in myocardial energy metabolism in the pathogenesis of AF, particularly in pathways governing fatty acid utilization and mitochondrial function.³ The carnitine pathway is a central regulator of fatty acid transport and oxidation and has recently emerged as a potential link between metabolic dysregulation and AF risk.

Carnitine is an amine that facilitates transport of fatty acids (FA) into the mitochondrial matrix for β-oxidation, and is synthesized endogenously from trimethyllysine (TML) with γ-butyrobetaine (γBB) as an intermediate metabolite. It circulates freely as L-carnitine or esterified to FAs of varying chain lengths as acylcarnitines.^4,5 Further, L-carnitine may be converted to trimethylamine (TMA) and trimethylamine-n-oxide (TMAO) in processes influenced by the gut microbiome.⁶

We have previously published positive associations between both TML and TMAO with incident AF from the same cohort as currently investigated.^7,8 Further, γBB is associated with increased risk of cardiovascular mortality⁹; and several acylcarnitines have been linked to incident heart failure (HF) and AF.¹⁰ In addition, L-carnitine has been suggested as a useful supplement in prevention of arrhythmia after acute myocardial infarction (AMI).¹¹ No study has so far reported any relationships between L-carnitine or γBB and new-onset AF.

We examined potential relationships between γBB, L-carnitine as well as acylcarnitines with incident AF, in a large Norwegian cohort of patients evaluated for chronic coronary syndrome. As there is an increasing acknowledgment of sex differences in the serum metabolome, including sex specific variations in both serum L-carnitine and several acylcarnitines, we also conducted sex stratified analyzes.^12,13

Methods

Study population

The study cohort has been described in detail previously.¹⁴ To summarize, 4166 patients with suspected chronic coronary syndrome were examined with coronary angiography at two Norwegian university hospitals from 2000 to 2004. Approximately two-thirds of the patients were included in the Western Norway B-Vitamin Intervention Trial (WENBIT), a randomized controlled trial investigating B-vitamin treatment as secondary CVD prevention.¹⁵ Patients with a pre-existing diagnosis of AF, or AF at baseline, as well as those without a valid measurement of the biomarkers in question were excluded, leaving 3815 patients for the final analyses.

All patients provided written informed consent. The study was conducted in accordance with the Declaration of Helsinki and was approved by the Regional Committee for Medical and Health Research Ethics, Norwegian Health Region West.

Clinical and biochemical data

Baseline data collection has been described in detail elsewhere.¹⁴ The patients provided anamnestic information which was checked against medical records when available. Study specific anthropometric and clinical data were collected by trained study personnel.

Venous blood samples were collected 1–2 days prior to or immediately after angiography. Standard laboratory analyses were carried out on fresh blood samples at each recruiting hospital. Study-specific analyses were carried out at the laboratory of Bevital AS (www.bevital.no) on samples stored at −80°C and later thawed before analysis by personnel blinded to the study outcomes. Serum metabolite concentrations were determined by liquid chromatography-tandem mass spectrometry as previously described.^16,17

Endpoint data

The end-point of interest was receiving a diagnosis of AF (according to the International Classification of Diseases 10th edition I48) during hospitalization or death due to AF. Endpoint data were retrieved from the Cardiovascular Disease in Norway project (CVDNOR; https://cvdnor.w.uib.no/) and the Norwegian cause of death registry (https://www.fhi.no/en/ch/cause-of-death-registry/). A 11-digit personal identification number, unique to each Norwegian resident, was used to link endpoint data to each patient.

Statistical analyses

Continuous and categorical data are presented as median (25th-75th percentiles) and counts (%), respectively. Between groups differences were tested by Mann–Whitney U test and Chi square test for continuous and categorical variables, respectively.

We used logistic regression to obtain odds ratios (ORs) [95% confidence intervals (CIs)] for incident AF per one standard deviation (SD) increase in log-transformed serum concentrations of the biomarkers in question. We applied one unadjusted model as well as model 1 adjusted for age and sex. Further, we also created a model 2 additionally adjusted for hypertension, diabetes, smoking, anamnestic HF, body-mass index (BMI), estimated glomerular filtration rate (eGFR) and plasma TML and TMAO. We also created a model 3 which additionally incorporated fasting status.

Potential effect modification by sex was examined by introducing interaction terms in model 2. Further, improvement in model fit was evaluated using a likelihood ratio test (LRT) comparing models with and without the interaction term. For biomarkers that showed significant interaction with sex according to later AF we calculated sex-specific odds ratios (ORs) [95% confidence intervals (CIs)] for incident AF, with multivariable adjustment as reported for model 2. Further, any non-linear relationships between the biomarker and future AF risk were explored by constructing a generalized additive model (GAM) smoothed spline for the entire population and in sex-specific subgroups.

Reclassification of patients and model discrimination were explored by adding the metabolite in question to a multivariate logistic regression model containing the same covariates as model 2 and calculating the continuous net reclassification index (NRI>0) and the Receiver Operating Curve Area Under the Curve (ROC-AUC). We also calculated the E-values of Model 2 to investigate the possibility of unmeasured confounding.¹⁸

All tests were two-sided and the significance level set to 0.05. We used the software IBM SPSS Statistics for Windows, Version 26.0. Armonk, NY: IBM Corp., and R for Windows (R Core Team 2016, Vienna, Austria, utilizing the packages #survival, #Hmisc, #PredictABEL, #ROCR, #ggplot2 and #ICC).

Results

Baseline characteristics

Baseline characteristics for the entire population and according to sex are presented in Table 1. The majority (72%) was men and median (25th–75th percentile) age at inclusion was 61 (54–69) years. The women were somewhat older [63 (56–71) vs. 61 (54–68) years, P < 0.001] and had a higher prevalence of hypertension (50.3% vs. 44.3%, P < 0.001). Men had more often suffered a previous AMI (45.5% vs. 28.2%, P < 0.001), were more often smokers (34.1% vs. 26.7%, P < 0.001) and reported more often anamnestic heart failure (6.8% vs. 4.4%, P < 0.01). BMI and prevalence of diabetes were similar among both sexes.

Table 1.

Baseline characteristics

Demographics and risk factors	Total population (n = 3815)	Male participants (n = 2733)	Female participants (n = 1082)	P-value
Age, years	61 (54–69)	61 (54–68)	63 (56–71)	< 0.001
BMI, kg/m2	26 (24–28)	26 (24–28)	26 (23–29)	0.029
Diabetes Mellitus, n%	441 (11.6%)	316 (11.6)	125 (11.6)	0.99
Smoking, n%	1223 (32.1%)	932 (34.1)	291 (26.7)	< 0.001
Hypertension, n%	1754 (46.0%)	1210 (44.3)	544 (50.3)	< 0.001
Previous AMI, n%	1548 (40.6%)	1243 (45.5)	305 (28.2)	< 0.001
Anamnestic Heart failure, n%	233 (6.1%)	185 (6.8)	48 (4.4)	< 0.01
Ejection fraction, %	66 (60–70)	65 (60–70)	70 (60–70)	< 0.001
Biomarkers and metabolites
γBB, μmol/L	1.02 (0.89–1.19)	1.08 (0.93–1.23)	0.88 (0.77–1.01)	< 0.001
L-Carnitine, μmol/L	39.1 (34.3–44.0)	39.6 (35.2–44.6)	37.0 (31.8–42.2)	< 0.001
Acetylcarnitine, μmol/L	5.89 (4.74–7.43)	5.8 (4.7–7.3)	6.1 (4.9–7.7)	< 0.001
Octanoylcarnitine, μmol/L	0.13 (0.09–0.20)	0.17 (0.09–0.19)	0.14 (0.09–0.21)	0.014
Palmitoylcarnitine, μmol/L	0.08 (0.07–0.10)	0.09 (0.07–0.12)	0.08 (0.06–0.11)	< 0.001
Valerylcarnitine, μmol/L	0.11 (0.09–0.15)	0.10 (0.07–0.15)	0.10 (0.07–0.12)	< 0.001
Propionylcarnitine, μmol/L	0.43 (0.34–0.54)	0.45 (0.36–0.55)	0.40 (0.31–0.49)	< 0.001
TML, μmol/L	0.85 (0.69–1.06)	0.89 (0.72–1.10)	0.75 (0.62–0.92)	< 0.001
TMAO, μmol/L	5.6 (3.6–9.5)	5.86 (3.69–10.10)	5.11 (3.35–8.45)	< 0.001
CRP, mg/L	1.74 (0.86–3.50)	1.73 (0.85–3.36)	1.78 (0.88–3.78)	0.07
eGFR, mL/min/1.73 m2	91 (79–100)	92 (81–101)	88 (74–97)	< 0.001
ApoB, g/L	0.87 (0.73–1.04)	0.86 (0.73–1.04)	0.88 (0.74–1.06)	0.01
ApoA1, g/L	1.30 (1.14–1.48)	1.25 (1.10–1.42)	1.45 (1.26–1.62)	< 0.001
Triglycerides, mmol/L	1.50 (1.08–2.15)	1.55 (1.10–2.24)	1.39 (1.0–1.92)	< 0.001
Endpoints
New-onset AF	412 (10.8%)	313 (11.5%)	99 (9.2%)	0.04
Continuous variables are presented as median (25th-75th percentile).
Categorical variables are presented as no. (%).

γBB, gamma butyrobetaine; AF, Atrial fibrillation; AMI, acute myocardial infarction; BMI, body mass index; TMAO, trimethylamine n-oxide; TML, trimethyllysine.

Men had higher serum concentrations of all the studied biomarkers except for acetylcarnitine which was higher among women.

The association between serum biomarkers and incident atrial fibrillation

Over a median (25th–75th percentile) follow-up of 7.3 years (6.3–8.6), 412 (10.8%) patients were registered with AF. The incidence of AF was slightly higher in men than in women (11.5% vs. 9.2%, P = 0.04). In the unadjusted model all biomarkers showed positive associations with incident AF; however, only γBB, L-carnitine and palmitoylcarnitine remained significantly associated in model 2 (Table 2). Further adjusting for fasting status in model 3 did not alter the risk estimates significantly (data not shown).

Table 2.

Risk associations between γBB, L-carnitine, acylcarnitines and AF

	Unadjusted		Model 1a		Model 2b
	OR (95% CI)	P-value	OR (95% CI)	P-value	OR (95% CI)	P-value	Pi	LRT X2(1)(P)
Plasma γBB	1.33 (1.20–1.48)	0.01	1.16 (1.04–1.30)	0.01	1.15 (1.01–1.31)	0.04	0.04	4.3 (0.04)
Plasma L-Carnitine	1.12 (1.01–1.24)	0.04	1.14 (1.02–1.27)	0.02	1.14 (1.01–1.27)	0.03	0.03	5.0 (0.03)
Acetylcarnitine	1.24 (1.12–1.37)	0.01	1.14 (1.03–1.27)	0.01	1.09 (0.97–1.22)	0.15	<0.01	11.4 (<0.01)
Octanoylcarnitine	1.17 (1.06–1.29)	0.01	1.04 (0.94–1.16)	0.43	1.01 (0.91–1.12)	0.82	0.36	0.83 (0.36)
Palmitoylcarnitine	1.32 (1.19–1.46)	0.01	1.18 (1.06–1.32)	< 0.01	1.15 (1.03–1.28)	0.01	<0.01	7.4 (<0.01)
Valerylcarnitine	1.06 (0.96–1.17)	0.26	1.02 (0.92–1.14)	0.69	0.94 (0.84–1.06)	0.32	0.41	0.68 (0.41)
Propionylcarnitine	1.09 (0.98–1.21)	0.1	1.04 (0.94–1.16)	0.47	0.99 (0.89–1.12)	0.98	0.33	0.96 (0.33)

CI, confidence interval; values per 1 SD log transformed; LRT, Likelihood ratio test; OR indicates odds ratio; Pi, P-value for gender interaction; SD, standard deviation.

^aAdjusted for sex and age.

^bAdjusted for sex, age, DM, HT, smoking, anamnestic heart failure, BMI, eGFR, TML and TMAO.

Interaction analyses showed significant effect modification by sex for γBB, L-carnitine, palmitoylcarnitine and acetylcarnitine. LRT indicated improved model fit for these four biomarkers and in subgroup analyses the ORs were substantially higher in women than in men (Table 2 and Figure 1). Interestingly the risk relationships were essentially linear and much stronger in women compared with men (see Supplementary material online, Figure S1).

Figure 1.

Forest plot showing the risk association between Butyrobetaine, L-Carnitine, Acetylcarnitine and Palmitoylcarnitine and new-onset AF in subgroups based on gender. Squares represent ORs and horizontal lines represent CIs. Blue indicates men and red women. The values are calculated via logistic regression in a multivariate model adjusted for age, hypertension, diabetes, smoking, anamnestic heart failure, body-mass index, eGFR, TML and TMAO and with the log-transformed biomarker in question per 1 SD.

Reclassification, model discrimination and E-values

When adding serum γBB, L-carnitine, palmitoylcarnitine or acetylcarnitine to model 2 only γBB significantly improved the NRI>0. Neither of the biomarkers improved ROC-AUC (see Supplementary material online, Table S1).

The E-values for the OR and for the lower CI of Model 2 were 1.57 and 1.11 for γBB, 1.54 and 1.11 for L-carnitine and 1.57 and 1.20 for palmitoylcarnitine.

Discussion

In this large cohort of patients evaluated for chronic coronary syndrome, higher levels of serum γBB, L-carnitine and palmitoylcarnitine were associated with increased risk of receiving a diagnosis of AF during long-term follow-up. The relationship was stronger among women, however only γBB enhanced risk prediction beyond conventional AF risk factors.

A growing body of evidence supports atrial cardiomyopathy as a precursor to AF, where diverse atrial pathological changes lead to impaired atrial function.¹⁹ γBB as well as its precursor TML and L-carnitine are essential for maintaining mitochondrial function, while long-chain acylcarnitines levels increase rapidly in ischemic tissue and promote electrophysiological derangements conceivably contributing to the development of atrial cardiomyopathy via metabolic remodelling.^3,20,21

We previously reported a positive association between plasma TML and incident AF among community dwellers.⁷ Further downstream metabolism of γBB yields L-carnitine via γBB-hydroxylase.²² Interestingly, an animal study found that during reperfusion after cardiac ischaemia, inhibition of γBB-hydroxylase decreased the susceptibility of ventricular fibrillation and lowered cardiac concentrations of long-chain acylcarnitines.²³ Moreover, our findings of a positive association between the long-chained palmitoylcarnitine and later AF risk are supported by prior studies linking medium- and long-chain acylcarnitines to AF risk.^10,24,25 Notably long-chain acylcarnitines in general and palmitoylcarnitine in particular has been shown to promote electrophysiological changes in cardiomyocytes promoting arrythmia.^26,27

Complex, metabolic pathways cannot be resolved by our research design; however these findings suggest that accumulation of downstream metabolites in the carnitine pathway may be indicative of a proarrhythmic state, potentially involving mitochondrial function and energy metabolism.

Notably, γBB may also be derived from the microbiome via degradation of dietary L-carnitine,⁶ and L-carnitine supplementation has been reported to lower postoperative AF risk.^28–30 Further, blood levels of acylcarnitines have been found to depend upon fasting status.³¹ We were not able to account for dietary intake in the present study; however adjusting for fasting status as well as plasma TMAO, which is also linked to dietary L-carnitine intake and incident AF,⁸ did not alter risk estimates in the current study. This implies that our findings primarily are related to endogenous rather than dietary factors.

An intriguing novel finding in our study was the stronger risk relationships between γBB, L-carnitine and palmitoylcarnitine among women than in men. Although direct mechanistic evidence is limited, our finding is biologically plausible given established sex differences in mitochondrial function, fatty acid utilization, and acylcarnitine metabolism, partly influenced by sex hormones.³² Interestingly, systemic L-carnitine levels have been reported to be lower in women than in men, being in line with our findings, but concentrations appear to have greater age-dependent increase in women.^33,34

Moreover, γBB and acylcarnitines correlate with inflammatory markers and activate proinflammatory pathways, consistent with previous findings in our cohort where patients who developed AF had higher levels of CRP and neopterine, however, there was no significant difference in CRP levels among men and women in our cohort.^8,35,36

Taken together, γBB, L-carnitine and acylcarnitines may contribute to inflammation, electrophysiological and structural changes that collectively are considered to play important roles in the development of atrial cardiomyopathy and AF.^37–39

The stronger association between carnitine-related biomarkers, and in particular γBB, with AF risk in women than in men may be clinically relevant. Women generally experience more symptoms, worse quality of life and more cerebral stroke then their male counterparts.^40,41 Together, this suggests that biomarkers improving early risk stratification could be of value and opens up the possibility of sex-specific risk scores as has been established for coronary artery disease.⁴²

Strengths and limitations

Strengths of this study include its prospective design, long-term follow-up, large sample size and endpoints obtained from national registries. Still, it is not without limitations. First, as with any observational study, there is a risk of residual confounding. However, E-values suggest that substantial unmeasured confounding would be required, which is unlikely given our extensive covariate adjustment, however, residual confounding cannot fully be ruled out. Second, while γBB did improve risk reclassification, none of the other biomarkers improved reclassification or ROC-AUC this suggests limited incremental value beyond established risk factors by the biomarkers in question, and most likely limits translation into clinical use regarding risk prediction. Third, data on red meat intake, which contains carnitine, was not available. Fourth, competing risk from death is likely during long-term follow-up, but could not be fully accounted for because the exact onset of AF could not be assessed. Fifth, the cohort consisted primarily of Caucasian individuals from Western Norway and the results require replication in more diverse populations.

Conclusions

In this large cohort of patients with suspected chronic coronary syndrome, higher levels of γBB, L-carnitine and palmitoylcarnitine were associated incident AF and the associations were stronger in women. These novel sex-specific findings highlight carnitine metabolism as a field that requires further exploration on cardiac arrhythmias and underscore the need for sex-stratified analyses in cardiovascular biomarker research.

Lead author biography

Mads M. Svenningssnon, MD, is a consultant cardiologist and researcher at the Department of Cardiology at Haukeland University Hospital. His research focuses primarily on cardiovascular epidemiology, biomarkers and atrial fibrillation.

Data availability

The data from WECAC that support the findings of this study are available from the corresponding author upon reasonable request.

Supplementary material

Supplementary material is available at European Heart Journal Open online.

Funding

The current project was funded by the Department of Heart Disease at Haukeland University Hospital.