Causal link between circulating beta-hydroxybutyrate and myocardial infarction: Evidence from Mendelian randomization

Abstract

This study investigates the potential causal relationship between circulating ketone bodies, specifically beta-hydroxybutyrate (β-OHB), and the risk of myocardial infarction (MI), addressing limitations of confounding and reverse causation inherent in previous observational studies. We employed both univariable Mendelian randomization and multivariable Mendelian randomization (MVMR) analyses using publicly available genome-wide association study summary statistics. Genetic variants robustly associated with serum β-OHB levels (P < 5 × 10⁻⁸) served as instrumental variables. Primary analysis used the inverse-variance weighted method, supplemented by sensitivity analyses (MR-Egger, weighted median, weighted mode, MR-Lasso) and tests for pleiotropy/heterogeneity (Cochran’s Q, MR-Egger intercept, MR-Pleiotropy RESidual Sum and Outlier, leave-one-out). The MVMR model adjusted for key cardiometabolic confounders: type 2 diabetes, alcohol intake, body mass index, smoking, hypertension, and blood lipids. univariable Mendelian randomization analysis indicated a significant positive association: a one-unit increase in natural log-transformed β-OHB concentration was associated with higher MI risk (IVW odds ratio [OR] = 1.407, 95% confidence interval (CI): 1.1660–1.6980, P = .0003691). Sensitivity analyses (weighted median: OR = 1.260, 95% CI: 1.0022–1.585, P = .0478442; MR-Egger: directionally consistent) supported this finding. MVMR analysis adjusting for confounders yielded divergent results. While conventional IVW showed no significant independent association (OR = 0.9787, 95% CI: 0 0.7410–1.2928, P = .8797), the MR-Lasso method revealed a significant independent positive association between β-OHB and MI risk (OR = 1.2366, 95% CI: 1.0397–1.4707, P = .0163751). This Mendelian randomization study suggests that genetically predicted, chronically elevated levels of β-OHB may confer an increased risk of MI.

1. Introduction

Acute myocardial infarction (MI) represents a leading cause of death globally. The risk of developing acute MI escalates with age, imposing a substantial burden on both global health and economies.^[1,2] Evidence indicates that modifiable risk factors, including smoking, dyslipidemia, hypertension, diabetes, and obesity, collectively drive a significant proportion of global MI incidence.^[3–5]

Ketone bodies, comprising beta-hydroxybutyrate (β-OHB), acetoacetate, and acetone, serve as crucial alternative energy substrates for the brain during fasting or carbohydrate scarcity. β-OHB is the predominant circulating ketone. Circulating β-OHB levels rise in various physiological states such as fasting, postexercise, and during low-carbohydrate diets.^[6–10] Primarily synthesized in the liver, ketone bodies function not only as energy sources for the heart and vasculature but also as signaling molecules influencing cardiovascular function.^[11] Some work suggests enhanced ketone bodies utilization may boost myocardial blood flow, with β-OHB potentially exerting cardioprotective effects, and some scholars have attempted to use ketogenic diets to improve metabolic cardiovascular diseases.^[12–14] Another study observed a perioperative decline in circulating ketone bodies in patients undergoing cardiopulmonary bypass surgery, inversely correlating with the extent of myocardial injury.^[15] However, elevated ketone levels have also been linked to worse disease severity and adverse outcomes in conditions like heart failure and arrhythmogenic right ventricular cardiomyopathy, as well as in ST-elevation MI patients exhibiting increased 24-hour ketone flux: associations include larger infarct size and reduced left ventricular ejection fraction (LVEF).^[16–19] Observational studies further identify elevated serum β-OHB as an independent predictor of cardiovascular events, including MI, in hemodialysis patients.^[20] Nuclear magnetic resonance based investigations report ketone elevation post-acute MI, with β-hydroxybutyrate potentially implicated.^[21,22] The precise nature of the relationship between ketone bodies and MI remains clinically ambiguous, and these observational findings are susceptible to reverse causation and confounding.

Mendelian randomization (MR) offers a novel analytical approach to interrogate causal relationships. By leveraging genetic variants as instrumental variables (IVs) to proxy exposures, MR mitigates limitations inherent in traditional observational studies-namely confounding, reverse causation, and measurement error-providing a more robust framework to elucidate potential causal links between β-OHB and MI.^[23,24]

2. Methods

2.1. Data sources

Summary statistics from genome-wide association studies (GWAS) for ketone bodies (β-hydroxybutyrate) were sourced from the IEU OpenGWAS project^[25] (https://gwas.mrcieu.ac.uk/datasets/met-d-bOHbutyrate/). GWAS summary statistics for MI were obtained from the FinnGen Consortium, Release 12 (https://www.finngen.fi/en),^[26] which aims to decipher genotype-phenotype associations by analy-sing genomic and health data from Finnish biobank participants. Data for additional traits (lipoproteins, diabetes, hypertension, smoking, alcohol intake, body mass index [BMI]) were also acquired from the IEU OpenGWAS database.^[25] All utilized GWAS datasets are publicly available and received ethical approval from relevant committees; consequently, additional ethical approval for this analysis was not required. Detailed GWAS metadata are presented in Table 1.

Table 1.

Detailed information of data sources.

Explore or outcome	Ref	Ieu id	Consortium	Ancestry	Participants
3-Hydroxybutyrate	NA	met-d-bOHbutyrate	NA	European	113,595 individuals
Myocardial infarction	NA	NA	FinnGen	European	31,666 cases/ 416,171 controls
Type 2 diabetes	NA	NA	UK Biobank	European	2292 cases/ 358,849 controls
Alcoholic drinks per wk	30,643,251	ieu-b-73	GSCAN	European	335,394 individuals
Cigarettes smoked per d	NA	ieu-b-4826	Within family GWAS consortium	European	24,784 individuals
BMI	30,124,842	ieu-b-40	GIANT	European	681,275 individuals
HBP	NA	NA	Neale Lab	European	1237 cases/ 359,957 controls
HDL-C	32,203,549	ieu-b-109	UK Biobank	European	403,943 individuals
LDL-C	32,203,549	ieu-b-110	UK Biobank	European	440,546 individuals
TG	32,203,549	ieu-b-111	UK Biobank	European	441,016 individuals

BMI = body mass index, GIANT = genetic investigation of anthropometric traits, GSCAN = GWAS and sequencing consortium of alcohol and nicotine use, HBP = high blood pressure, HDL-C = high-density lipoprotein cholesterol, LDL-C = low-density lipoprotein cholesterol, NA = not applicable, Ref = reference (PubMed id), TG = triglyceride.

2.2. Mendelian randomization analysis

We employed both univariable Mendelian randomization (UVMR) and multivariable Mendelian randomization (MVMR) to investigate the potential causal effect of ketone bodies on MI. Selection of IVs for β-OHB adhered to 3 core MR assumptions^[27]: Strong association with the exposure (β-OHB); Independence from confounders; Exerting effects on the outcome (MI) solely via the exposure (no horizontal pleiotropy). MVMR extends assumption 1 to require genetic variants associated with one or more exposures, while the other assumptions hold. Within the MVMR framework, we adjusted for key cardio metabolic risk factors (type 2 diabetes, alcohol intake, BMI, smoking, hypertension, and blood lipids) to isolate the direct effect of β-OHB on MI.

Single nucleotide polymorphisms (SNPs) significantly associated with β-OHB at genome-wide significance (P < 5 × 10⁻⁸) were initially selected. These SNPs were clumped (linkage disequilibrium pruning) based on linkage disequilibrium thresholds (r² < 0.001, window size = 10,000 kb). The strength of the selected IVs was confirmed using the F-statistic (F > 10),^[28] calculated as

$${\displaystyle F=\frac{R^2(n-2)}{(1-R^2)},\mathrm{w}\mathrm{h}\mathrm{e}\mathrm{r}\mathrm{e}{R}^2\approx 2\times EAF\times (1-EAF)\times {\beta }^2}$$

(EAF: effect allele frequency; n: sample size). Effect alleles for exposure and outcome datasets were harmonized to ensure consistent directionality.

For UVMR, the inverse-variance weighted (IVW) method served as the primary analytical approach.^[29,30] Complementary sensitivity analyses included MR-Egger^[31] (accounting for directional pleiotropy), weighted median^[32] (robust to ≤50% invalid instruments), and weighted mode methods.^[33] We assessed heterogeneity using Cochran’s Q test (P < .05, suggestive of heterogeneity)^[30] and evaluated potential horizontal pleiotropy via the MR-Egger intercept test (P < .05, indicative of pleiotropy).^[31] Leave-one-out analysis tested the robustness of findings to individual SNPs.^[34] Additionally, MR-Pleiotropy RESidual Sum and Outlier testing^[35] was performed to identify and adjust for outlier SNPs contributing to pleiotropy. Figure 1.

Figure 1.

Overview of research design. IV = Instrumental variable, MVMR = multivariable Mendelian randomization, UVMR = univariable Mendelian randomization.

3. Results

3.1. Univariable MR findings

Our two-sample MR analysis suggested a potential causal relationship between elevated β-OHB levels and increased MI risk. In the primary IVW model, a one-unit increase in natural log-transformed β-hydroxybutyrate concentration was significantly associated with higher MI risk (P = .0003691; odds ratio [OR] = 1.407, 95% confidence interval (CI): 1.1660–1.6980). The MR-Egger estimate, while not reaching strict statistical significance (P = .0604743; OR = 1.710, 95% CI: 1.0475–2.790), indicated a directionally consistent effect, hinting at a potential dose-response. The weighted median approach further supported the primary IVW result (P = .0478442; OR = 1.260, 95% CI: 1.0022–1.585). The weighted mode estimate was nonsignificant (P = .1193162; OR = 1.270, 95% CI: 0.9647–1.671). (Fig. 2).

Figure 2.

Univariable Mendelian randomization association of genetically predicted β-OHB with MI.

3.2. Multivariable MR findings

We implemented MVMR to assess the independent association of β-OHB with MI risk after accounting for genetic predisposition to key cardio metabolic risk factors. After adjusting for type 2 diabetes, alcohol intake, BMI, smoking, hypertension, and blood lipids, the conventional IVW method no longer showed a significant independent association between β-OHB and MI risk (OR = 0.9787, 95% CI: 0.7410–1.2928, P = .8797). Findings from MR-Egger (OR = 0.9831, 95% CI: 0.7444–1.29821, P = .9041) and weighted median (OR = 1.1251, 95% CI: 0.8975–1.4105, P = .3067) were consistent with this null result. However, employing MR-Lasso^[36] - a method more robust to weak instruments and outliers - revealed a significant independent positive association: elevated β-hydroxybutyrate levels were linked to increased MI risk (OR = 1.2366, 95% CI: 1.0397–1.4707, P = .0163751). (Fig. 3).

Figure 3.

Multivariable Mendelian randomization.

4. Discussion

This study presents a comprehensive MR analysis exploring the causal link between circulating β-OHB and MI risk. Initial UVMR findings suggested a potential causal role for higher ketone levels in increasing MI susceptibility. Subsequent MVMR, adjusting for major cardio metabolic confounders, yielded divergent results depending on the method: while conventional IVW showed no independent association (OR = 1.022, P = .848), the more robust MR-Lasso method indicated that elevated β-OHB might independently contribute to increased MI risk (OR = 1.205, 95%CI:1.009–1.439). Collectively, these data indicate a genetically predicted association between chronically elevated β-OHB levels and MI risk.

This observed risk association presents an apparent paradox with some experimental studies. Acute β-OHB infusion or pretreatment with agents like empagliflozin (which elevates ketones) have shown protective effects, reducing infarct size and improving cardiac function in models of MI, likely through mechanisms involving metabolic optimization, ATP synthesis enhancement, NLRP3 inflammasome suppression, and AMPK/SIRT1 pathway activation.^[37–41] However, contrasting findings emerge under conditions of sustained ketonemia. Mice chronically fed a ketogenic diet, exhibiting significantly elevated plasma β-OHB, demonstrated lower LVEF 4 weeks post-MI.^[42] High-fat low-carbohydrate diets increased the risk of arrhythmic death and pump failure following ischemia-reperfusion.^[43] Clinically, higher β-OHB levels independently correlated with larger infarct size and reduced LVEF in ST-elevation myocardial infarction patients.^[17,19] This suggests that while acute ketosis might be protective during ischemia/reperfusion, chronically elevated β-OHB levels (e.g., via sustained dietary interventions like ketogenic diet) could adversely remodel cardiac metabolism or impair adaptive responses to hypoxia, potentially exacerbating MI outcomes.

Furthermore, the biological effects of β-OHB appear context-dependent. In heart failure, the myocardium shifts towards utilizing ketones and glycolysis for ATP production.^[44,45] During post-MI repair phases, β-OHB may promote angiogenesis (via PHD2 targeting) and improve cardiac remodeling and energetics.^[46,47] Moderate, controlled ketosis might indeed confer certain benefits.^[37,48] However, under hypoxic stress, β-OHB has been linked to increased cardiomyocyte apoptosis^[17] and enhanced free fatty acid oxidation, potentially aggravating reperfusion injury and delaying functional recovery.^[49] Imbalances in myocardial ketone metabolism may also fuel diabetic cardiomyopathy progression via NF-κB inflammatory pathways.^[50]

Key strengths of this study include the application of MR methodology to infer causality, minimizing concerns of reverse causation inherent in observational studies. The use of MVMR strengthens the case for a direct effect of β-OHB by accounting for major confounders.

Limitations warrant consideration. Despite adhering to core MR assumptions, residual pleiotropy or unmeasured confounding cannot be entirely ruled out. Additionally, the reliance on European-ancestry GWAS data limits generalizability to other populations. Larger-scale clinical studies are needed to validate these causal inferences and explore the nuances of timing, context, and concentration dependence observed in the experimental literature. Lastly, some important factors influencing circulating β-OHB levels could not be accounted for. These include fasting status, acute stress, starvation, nutritional intake, and medication use (e.g., SGLT2 inhibitors, salicylates), which were not captured in the GWAS summary statistics. Residual confounding from these unmeasured exposures may influence the observed genetically predicted associations, and future studies incorporating individual-level data are warranted to further validate these findings.

5. Conclusion

Employing Mendelian randomization analysis on large-scale GWAS summary statistics, this study provides evidence supporting genetically predicted association between elevated circulating β-OHB levels and increased risk of MI in individuals. The data collectively suggest that chronically higher β-OHB may constitute a risk factor for MI, although the precise mechanisms likely depend critically on the metabolic and temporal context.

Author contributions

Validation: Qianshi Wu.

Conceptualization: Yanqun Li.

Writing – original draft: Jidi Wu, Qianshi Wu, Yanqun Li.

Writing – review & editing: Jidi Wu, Wenwen Fan, Yulin Mou, Li Luo, Yong Xu, Wei Huang.