Causal associations between circulating metabolites and risk of neuroblastoma: a bidirectional Mendelian randomization study

Abstract

Background

The current understanding of the relationship between circulating metabolites and neuroblastoma (NB) risk is insufficient. Herein, bidirectional Mendelian randomization (MR) studies were utilized to examine the potential causal associations of metabolites and the risk of NB.

Methods

Significant single-nucleotide polymorphisms (SNPs) associated with circulating metabolites were obtained from the genome-wide association study (GWAS) in the European population (N = 7824). Summary statistics for NB were aggregated from prior GWAS studies encompassing 1,627 patients and 3,254 cancer-free children. The causality was assessed primarily by inverse-variance weighted (IVW), along with MR-Egger regression, weighted median estimator, and weighted mode method.

Results

Among the 486 metabolites analyzed, a genetically determined elevation in blood butyrylcarnitine concentration (log₁₀) was significantly correlated with a 51.5% decrease in NB risk [IVW odds ratio (OR) = 0.485, 95% confidence interval (CI): 0.285–0.826, P = 0.008]. In the reverse MR analyses, an increase in log odds of NB risk demonstrated a significant correlation with a decrease in guanosine [IVW Beta ± SE: -0.042 ± 0.015 log₁₀ change, P = 0.004] and an elevation in 3-methyl-2-oxobutyrate [IVW Beta ± SE: 0.010 ± 0.004 log₁₀ change, P = 0.011] levels. Besides, no significant causal effect of NB on butyrylcarnitine was observed.

Conclusions

This research identifies bidirectional causal relationships between circulating metabolites and NB. Increased levels of blood butyrylcarnitine are associated with a reduced risk of NB, whereas lower levels of guanosine and higher levels of 3-methyl-2-oxobutyrate are indicative of NB patients. Further investigation is required for the mechanisms.

Supplementary Information

The online version contains supplementary material available at 10.1007/s12672-025-04172-w.

Introduction

Neuroblastoma (NB) is an embryonal cancer that arises from neural crest tissues in the sympathetic nervous system. It is the most common extracranial solid tumor in children, with a median age of diagnosis at 17 months [1]. NB exhibits significant biological and clinical diversity and can arise in different locations within the sympathetic nervous system [2]. The pathogenesis of NB is linked to developmental disorders and neuronal differentiation retardations [3], while clinical outcomes are affected by occurrences of spontaneous and complete regressions. The 5-year event-free survival rate for high-risk neuroblastoma patients is approximately 50% in the United States, indicating a concerning prognosis [4]. Multiple risk factors are involved in the categorization of individuals into high-risk NB groups, including age, histological type, MYCN gene amplification, chromosome alterations, DNA ploidy, and other relevant factors [5].

Metabolic dysregulation is a key characteristic of tumor cells, which adapt their metabolic states to adjust to the extracellular microenvironment. This adaptive response results in abnormal changes in cellular metabolite levels, simultaneously affecting the expression of various genes and altering cell phenotype [6]. Fatty acids serve as alternative energy reservoirs, metabolizing to produce adenosine triphosphate (ATP) at a rate double that of carbohydrates [7]. Numerous studies highlight the importance of fatty acid metabolism, especially mitochondrial fatty acid β-oxidation (FAO), in promoting the survival and advancement of cancer cells [8]. Notably, increased expression of lipogenic enzymes is associated with poor prognosis in multiple cancer types [9]. While robust evidence underscores the pivotal role of fatty acid metabolism in driving high-risk NB cases characterized by MYCN amplification [10], the causal relationship among diverse circulating metabolites in the mechanistic underpinnings of NB remains unclear, necessitating systematic and comprehensive analysis.

The possibility of a reverse causal pathway, in which the presence of NB affects circulating metabolite levels, is biologically plausible. NB cells engage in metabolic reprogramming by utilizing nutrients from and excreting metabolites into the systemic circulation. This may result in noticeable changes in the host’s metabolic profile. For instance, the “Warburg effect”, characterized by elevated glucose uptake and lactate production in cancer metabolism, can systematically alter circulating levels of associated metabolites [11]. Additionally, NB may directly release particular metabolites or signaling molecules that affect systemic metabolism [12, 13]. Therefore, investigating the reverse direction is crucial to determine whether observed associations are driven by metabolites influencing NB risk or are a consequence of the tumor’s presence, which is essential for accurate result interpretation.

Advances in genomic analysis create opportunities to explore the causality of risk factors on disease using large-scale genome-wide association studies (GWAS). Mendelian randomization (MR) analysis is a novel technique utilized in the field of epidemiological studies that leverages random allocation of genetic variants to investigate the causality between exposures and outcomes [14]. Unlike conventional observational analyses, MR ​can reduce the impact​ of unmeasured confounders on causal estimations and ​help mitigate​ reverse causal effects, ​as it utilizes​ single nucleotide polymorphisms (SNPs) which are randomly assigned at conception, before the onset of disease. In this present study, we employed bidirectional MR research involving 486 blood metabolites to elucidate the causal relationships between multiple metabolite levels and the risk of NB, integrating metabolomics with genomics to provide novel insight into the metabolism-related etiology and diagnostics of NB.

Methods

MR study design and assumptions

Making use of publicly available comprehensive GWAS statistics encompassing a large population dataset detailing metabolite levels and NB, we conducted thorough bidirectional two-sample MR analyses. The objective was to identify potential causal relationships between circulating metabolites and NB by utilizing instrumental variables (IVs) that represent the genetic variants associated with risk factors. Herein, MR analysis is based on three crucial assumptions: first, genetic variants should be strongly correlated with the exposure; second, the selected genetic variants should be independent of other confounders associated with exposure and outcome; lastly, the genetic variants employed should influence the risk of outcome solely through the exposure, rather than via alternative pathways.

Data sources and genetic instrument selection for metabolites

The Cooperative Health Research in the Region of Augsburg (KORA) platform comprises a series of population-based surveys conducted in southern Germany. Between 1985 and 2001, four independent cross-sectional surveys (S1-S4) were performed, each involving random samples of inhabitants of German nationality aged 25–74 years from the Augsburg region, totaling approximately 18,000 participants. The cohort, predominantly of Caucasian ancestry, has been followed longitudinally through repeated questionnaires, clinical examinations, and outcome ascertainment, providing a well-characterized resource for epidemiological research.​.

TwinsUK is a national adult twin registry in the United Kingdom, established primarily to investigate the genetic and environmental determinants of common diseases, especially those with higher prevalence in women. It comprises a large cohort of volunteer twins, recruited from the general population without specific screening criteria. The participants, predominantly female, have contributed to a wide array of data collections, including detailed clinical assessments, cardiovascular measurements, biological samples (such as blood, urine, and fecal samples), and food frequency questionnaires (FFQ), forming a rich resource for genetic and epidemiological studies [15].

The genetic data for 486 blood metabolites is one of the most comprehensive metabolite studies so far, originating from the above two European population studies comprising a total of 7,824 cases [16]. These identified metabolites were allocated into eight groups: amino acids, carbohydrates, cofactors and vitamins, energy compounds, lipids, nucleotides, peptides, and exotic organisms [17]. The concentrations of butyrylcarnitine were detected and log-transformed using ultrahigh performance liquid-phase chromatography separation coupled with tandem mass spectrometry in either plasma or serum samples. Genetic instruments for butyrylcarnitine were selected as follows (Supplementary Fig. 1). First, we identified SNPs associated with the exposure at a significance threshold of P < 5 × 10^{− 5} and performed linkage disequilibrium (LD) clumping under a strict threshold (r² < 0.001 within a 10 Mb window), resulting in ​91 independent SNPs. The strength of these instrumental variables was confirmed by calculating the F-statistic; all values were greater than 10, indicating little risk of weak-instrument bias. We then extracted the associations for these 91 SNPs from the NB GWAS dataset. The corresponding data for ​19 SNPs were successfully obtained. After harmonizing the effect alleles between the exposure and outcome datasets and excluding palindromic SNPs with ambiguous strand orientation, we conducted MR Steiger filtering to verify the correct direction of causality [18]. Finally, all 19 IVs passed the filtering and were retained for subsequent MR study.

Data sources and genetic instrument selection for NB

GWAS summary-level data for NB was obtained from the IEU OpenGWAS project (https://gwas.mrcieu.ac.uk/datasets/ieu-a-816/). Specifically, the GWAS data was derived from previous studies related to NB conducted by Capasso and colleagues, including 1,627 NB patients registered in the North American-based Children’s Oncology Group (COG) and 3,254 cancer-free children of self-reported Caucasian ancestry with 468,788 SNPs. The definition of NB cases is patients diagnosed as NB or ganglioneuroblastoma and registered in COG [19]. Finally, we obtained six SNPs associated with NB that met the criteria for significance (P < 5 × 10^{− 7}) and independence (r² < 0.001, kb >10000). All the IVs passed the filtrations of Steiger test and harmonizing analysis.

Statistical analysis and sensitivity analysis

In this bidirectional MR analysis, we primarily employed the inverse variance weighted (IVW) approach to estimate the causal relationship between metabolites and the risk of NB. IVW analyzed the SNP-specific Wald ratio estimates between exposure and outcome utilizing a random-effects inverse variance method with the assumption that no directional pleiotropic effect exists in all SNPs [20]. The MR-Egger regression, weighted-median estimator (WME), and weighted mode method served as additional estimation analyses with different assumptions. MR-Egger regression can provide a valid causal estimate even when all genetic variants exhibit pleiotropy, but it assumes that the instrument strength is independent of the pleiotropic effects (no measurement error in the SNP-exposure effects). Its intercept term provides a test for directional pleiotropy. The WME can provide a consistent causal estimate if at least 50% of the weight in the analysis comes from valid instrumental variables. The weighted mode method yields a consistent estimate under the assumption that the largest cluster of genetic variants (based on a similarity metric) consists of valid instruments.​ Horizontal pleiotropy is defined as the presence of alternative causal pathways through which genetic instruments affect the outcome, aside from the exposure of interest. The presence of potential horizontal pleiotropy was assessed based on the MR-Egger intercept and the MR pleiotropy residual sum and outlier (MR-PRESSO) test [21]. The robustness of various genetic IVs was assessed using Cochran’s Q-statistic test [22]. Leave-one-out sensitivity analysis was applied to determine the stability of the individual SNP in this MR study by excluding IVs one by one [23]. IVs that did not pass the sensitivity analyses will be removed, and the MR causal effect estimations will be re-evaluated. The MR results of metabolites on NB were presented in odds ratios (95% confidence intervals) per log₁₀ units genetically predicted increase in each metabolite, and results of NB on metabolites were presented as Beta ± SE per each unit higher in log odds of NB for convenience of interpretation. A cutoff of P < 0.0125 (0.05/4) was considered to indicate statistical significance after Bonferroni correction, and P < 0.05 but above Bonferroni correction threshold was defined as suggestive significant relationship [24]. All MR analyses in this study were performed in R software (4.2.0) with the TwoSampleMR and MRPRESSO packages.

Results

Bidirectional MR estimates between blood metabolites and NB

We conducted MR analyses to investigate potential associations between the risk of NB and 486 types of blood metabolites (the detailed accessions for all 486 metabolites are provided in Supplementary Table 1). The MR estimates revealed that butyrylcarnitine was the only exposure significantly associated with the risk of NB compared to other metabolites.

Additionally, we conducted a reverse MR analysis to evaluate the causal effects of NB risk on 486 metabolic exposures, aiming to determine whether NB influences the levels of candidate metabolites. The analysis revealed a causal association between the development of NB and 19 metabolites identified through the IVW method, including 12 known metabolites and 7 metabolites with unidentified chemical identities (Fig. 1, Supplementary Table 2). One unit increase in the log odds of NB risk could elevate 9 metabolites levels represented in log-transformed relative concentration values: phenylacetylglutamine [IVW Beta ± SE: 0.018 ± 0.009, P = 0.037], X-12,510–2-aminooctanoic acid [IVW Beta ± SE: 0.014 ± 0.007, P = 0.044], isobutyrylcarnitine [IVW Beta ± SE: 0.014 ± 0.007, P = 0.044], indoleacetate [IVW Beta ± SE: 0.013 ± 0.005, P = 0.016], 3-(4-hydroxyphenyl)lactate [IVW Beta ± SE: 0.012 ± 0.006, P = 0.043], 4-methyl-2-oxopentanoate [IVW Beta ± SE: 0.011 ± 0.005, P = 0.021], 3-methyl-2-oxobutyrate [IVW Beta ± SE: 0.010 ± 0.004, P = 0.011], 3-methyl-2-oxovalerate [IVW Beta ± SE: 0.009 ± 0.004, P = 0.019], and 2-linoleoylglycerophosphocholine [IVW Beta ± SE: 0.012 ± 0.006, P = 0.035]. One unit increase in the log odds of NB risk could decrease the levels of 3 metabolites represented in log-transformed relative concentration values: bilirubin [IVW Beta ± SE: – 0.023 ± 0.010, P = 0.019], X-14,205–alpha-glutamyltyrosine [IVW Beta ± SE: – 0.024 ± 0.011, P = 0.028], and guanosine [IVW Beta ± SE: – 0.042 ± 0.015, P = 0.004]. Of the 12 identified metabolites, guanosine and 3-methyl-2-oxobutyrate were found to be significantly correlated with NB (P < 0.0125). Cochran’s Q statistics (P > 0.05) and the MR-Egger intercept (P > 0.05) indicated an absence of heterogeneity and horizontal pleiotropy.

Fig. 1.

Mendelian randomization and sensitivity results for NB and metabolites in the form of forest plot. NB neuroblastoma, SNP single-nucleotide polymorphism, Beta the effect size of the exposure on outcome, CI confidence interval

Causal relationship of butyrylcarnitine on NB

We further analyzed the causality between butyrylcarnitine levels and the risk of NB with four MR methods, including IVW, MR-Egger, WME, and weighted mode (Fig. 2a, b). In the forward MR analyses, the IVW method reveals a significant causal relationship between butyrylcarnitine and NB risk in which a genetically determined increase in blood butyrylcarnitine concentration (log₁₀) was associated with a 51.5% lower risk of NB [odds ratio (OR) = 0.485, 95% confidence interval (CI): 0.285–0.826, P = 0.008]. Similar results were observed using the WME [OR = 0.509, 95% CI 0.277–0.934, P = 0.029] and weighted mode [OR = 0.522, 95% CI: 0.285–0.955, P = 0.049]. The MR-Egger analysis [OR = 0.599, 95% CI: 0.286–1.255, P = 0.192] demonstrated consistent direction and magnitude in the causal estimates (Table 1). The insignificance in the MR-Egger method may stem from its high false-positive rate of false-negative results, leading to susceptibility to outlier variants influence and undetected potential pleiotropy. The forest plot presented the causality assessments of each genetic IV on the risk of NB using different MR methods (Fig. 3).

Fig. 2.

Causal relationship of butyrylcarnitine on NB. a Causal effect of each SNP of butyrylcarnitine on the risk of NB in the form of forest plot. The red lines are the result of MR-Egger and IVW. b Causal effect of butyrylcarnitine levels on the risk of NB in the form of scatter plot. The slope of each line represents the causality estimated by different methods. c The overall heterogeneity in the causal effect estimates of butyrylcarnitine levels on NB in the form of funnel plot. d Causal effect of butyrylcarnitine levels on the risk of NB when deleting SNPs in sequence in the form of leave-one-out plot. NB neuroblastoma, SNP single-nucleotide polymorphism, MR Mendelian randomization, IVW inverse-variance weighted.

Table 1.

Bidirectional MR analyses of the causal relationship between butyrylcarnitine and NB

Directions	MR methods	No. of SNPs	Beta	SE	OR (95% CI)	Pval
Butyrylcarnitine→NB	Inverse variance weighted	19	– 0.723	0.272	0.485(0.285–0.826)	0.008
	MR Egger	19	– 0.512	0.377	0.599(0.286–1.255)	0.192
	Weighted median	19	– 0.675	0.310	0.509(0.277–0.934)	0.029
	Weighted mode	19	– 0.650	0.308	0.522(0.285–0.955)	0.049
NB→Butyrylcarnitine	Inverse variance weighted	6	– 0.009	0.006	0.991(0.980–1.001)	0.090
	MR Egger	6	0.018	0.029	1.018(0.963–1.077)	0.561
	Weighted median	6	– 0.013	0.007	0.988(0.974–1.001)	0.075
	Weighted mode	6	0.015	0.009	0.985(0.968–1.003)	0.155

MR Mendelian randomization, NB neuroblastoma, SNP single nucleotide polymorphisms, Beta the effect size of the exposure on outcome, SE standard errors, OR odds ratio, CI confidence interval

Fig. 3.

Causal effect of the butyrylcarnitine on NB estimated by four MR methods. NB neuroblastoma, MR Mendelian randomization, SNP single nucleotide polymorphisms, OR odds ratio, CI confidence interval, IVW inverse-variance weighted

Sensitivity tests of butyrylcarnitine on NB

Sensitivity analyses were conducted to validate the causal effect of butyrylcarnitine on the risk of NB. Cochran’s Q test revealed no significant heterogeneity among SNPs using IVW and MR-Egger methods (P = 0.918, Table 2). Additionally, the MR-Egger intercepts (P = 0.430) and MR-PRESSO Global Test (P = 0.916) indicated no statistically significant evidence of horizontal pleiotropy (Table 2). The symmetric results observed in the funnel plot further supported the stability of the research (Fig. 2c). The F-statistics for all IVs exceeded 10, indicating the significant relevance of the chosen genetic instruments in relation to butyrylcarnitine. The leave-one-out analysis did not identify any SNP that could affect the causality when eliminated sequentially (Fig. 2d). Consequently, the estimated causal effects of butyrylcarnitine levels on the risk of NB are robust and unaffected by any single genetic instrument. Directionality analysis conducted by Steiger (P < 0.001) confirmed the absence of reverse causality (Table 2).

Table 2.

Sensitivity analyses of the causal relationship between butyrylcarnitine and NB

Directions	Heterogeneity		Pleiotropy			Directionality
	Q	Q_Pval	Intercept	Intercept_Pval	PRESSO_Pval	Steiger_Pval
Butyrylcarnitine→NB	10.403	0.918	– 0.013	0.430	0.916	7.96E–121
NB→Butyrylcarnitine	1.764	0.881	– 0.008	0.381	0.878	6.86E–26

NB neuroblastoma

Causal relationship of NB on butyrylcarnitine

In the reverse MR analysis, NB was utilized as the exposure, with butyrylcarnitine as the outcome. For one unit increase in the log odds of NB risk, the results showed no evidence of a significant causal effect on butyrylcarnitine, represented in log-transformed relative concentration values, in IVW [Beta ± SE: – 0.009 ± 0.006, P = 0.090], MR-Egger [Beta ± SE: 0.018 ± 0.029, P = 0.561], WME [Beta ± SE: – 0.013 ± 0.007, P = 0.075], and weighted mode [Beta ± SE: – 0.015 ± 0.009, P = 0.155] (Fig. 4a, b; Table 1). Cochran’s Q test identified no significant heterogeneity (P = 0.881) among SNPs using the IVW and MR-Egger methods (Table 2). The assessment of symmetry in the funnel plot is challenging due to the limited number of IVs (Fig. 4c). To address the constraints of visual observation, we applied MR-Egger intercepts (P = 0.381) and the MR-PRESSO Global Test (P = 0.878), finding no statistically significant evidence of horizontal pleiotropy (Table 2). The leave-one-out analysis verified the robustness of SNPs when eliminated sequentially (Fig. 4d). No evidence of reverse causality was observed in the directionality analysis conducted by Steiger (P < 0.001, Table 2).

Fig. 4.

Causal relationship of NB on butyrylcarnitine. a Causal effect of each SNP of NB on butyrylcarnitine levels in the form of forest plot. The red lines are the result of MR-Egger and IVW. b Causal effect of NB on butyrylcarnitine levels in the form of scatter plot. The slope of each line represents the causality estimated by different methods. c The overall heterogeneity in the causal effect estimates of NB on butyrylcarnitine levels in the form of funnel plot. d Causal effect of NB on butyrylcarnitine levels when deleting SNPs in sequence in the form of leave-one-out plot. NB neuroblastoma, SNP single-nucleotide polymorphism, MR Mendelian randomization, IVW inverse-variance weighted

Discussion

Our study comprehensively analyzed the individual causal effects between a wide array of circulating metabolites and the risk of NB for the first time. To our knowledge, this is also the first study to explore the causal association between butyrylcarnitine and NB using the bidirectional two-sample MR approach. The findings indicate that elevated blood butyrylcarnitine levels significantly lower the risk of NB. In addition, the development of NB can significantly reduce guanosine levels while increasing 3-methyl-2-oxobutyrate levels in circulation.

Carnitine and its derivatives, such as butyrylcarnitine, play fundamental roles in cellular energy metabolism, primarily by transporting fatty acids into mitochondria for β-oxidation and modulating the acyl-CoA/CoA ratio. Given that metabolic reprogramming is a hallmark of cancer, alterations in acylcarnitine levels may reflect underlying mitochondrial dysfunction and energy metabolism abnormalities common to many malignancies, including NB [25]. ​Our finding of a putative causal effect of butyrylcarnitine on NB risk is supported by broader evidence implicating carnitine metabolism in cancer biology. For instance, a large-scale pan-cancer analysis reported an inverse association between butyrylcarnitine levels and the risk of numerous cancer types, suggesting it may serve as a general biomarker of cancer risk [26]. ​Furthermore, the relevance of this pathway to NB is underscored by experimental evidence showing that aggressive NB cells upregulate the enzyme CPT1A to enhance FAO for survival and chemoresistance [27]. Genetically predisposed higher circulating butyrylcarnitine may reflect relative inefficiency in the FAO pathway downstream of CPT1A. This inherent metabolic state, characterized by the accumulation of intermediates, might create a less favorable environment for the high-energy demands of tumor initiation, thereby reducing NB risk.​​ Therefore, our results align with the growing recognition that perturbations in FAO and mitochondrial function, as potentially indicated by acylcarnitine levels, are of significance in NB oncogenesis. Future research should investigate the specific pathways through which butyrylcarnitine influences NB pathogenesis.

Carnitine has been implicated in tumor treatment for glioblastoma through its promotion of cell senescence [28]. Another research has demonstrated that acylcarnitine mitigates the neurotoxicity of NB cells induced by cisplatin treatment without interfering with its anti-tumor efficacy, thereby supporting the use of cisplatin in NB therapy [29]. Butyrylcarnitine, in contrast to other acylcarnitines, has the capacity to inhibit the activity of short-chain acyl-CoA dehydrogenase, which is the most susceptible enzyme in the mitochondrial fatty acid β-oxidation pathway. Increased concentrations of butyrylcarnitine in the bloodstream of hyperuricemia rats may disrupt fatty acid oxidation [30]. Consequently, the increased levels of butyrylcarnitine may inhibit the β-oxidation process and give rise to fatty acid oxidation energy pathway dysfunction, resulting in an insufficient energy supply for cancer cells as well as increased mortality, thereby reducing the risk of NB. Furthermore, palmitoylcarnitine has been previously shown to promote the differentiation of NB-2 A cells and drive the formation of neural networks [31]. By providing acyl groups, palmitoylcarnitine interacts with and regulates the activity of essential proteins in the signaling cascade, thereby stimulating NB cell differentiation and inhibiting the proliferation of transformed cells. Acetyl-L-carnitine (ALC) has also been acknowledged to provide high-energy acetyl groups to metabolic pathways, thereby improving the overall energy status of the brain and modifying the biosynthesis patterns of specific neurotransmitters [32]. Acyl groups of acylcarnitine may participate in regulating protein and gene expression. Therefore, butyrylcarnitine may serve as a protective factor in NB by interacting with proteins that regulate the differentiation and proliferation of NB cells. Given the absence of relevant studies on butyrylcarnitine in NB, the mechanisms underlying these hypotheses require further verification.

Guanosine serves as a precursor in nucleotide synthesis. The transformation of tumor cells involves increased nucleotide synthesis, which promotes cell survival and accelerates cell growth [33]. The uncontrolled proliferation of NB cells may give rise to a marked reduction in guanosine levels. Branched-chain amino acid (BCAA) metabolism reprogramming is essential in various human tumors [34]. For this reason, the levels of 3-methyl-2-oxobutyrate, an intermediate of BCAA catabolism, may rise markedly in the development of NB.

This MR study implies that increased blood butyrylcarnitine levels in clinical practice may causally decrease the incidence rate of NB, while a deficiency in butyrylcarnitine could increase the risk of NB, providing a novel perspective on NB pathogenesis. The therapeutic efficacy of butyrylcarnitine deserves further investigation through clinical trials. Moreover, metabolites identified as being causally influenced by NB in this study, including guanosine and 3-methyl-2-oxobutyrate, hold promise as potential risk predictors. Their role includes improving disease diagnosis, supporting therapeutic monitoring, and advancing standards for prognosis assessment.

This study represents a comprehensive investigation into the role of butyrylcarnitine levels in NB development and the metabolic biomarker alterations associated with this condition. MR analysis utilizing a large sample size of aggregated GWAS data minimized the confounding interference and provided critical genetic evidence for precise causality estimates, addressing the limitations of prior observational and experimental studies. The application of thorough sensitivity and heterogeneity analyses confirmed the robustness of the causal estimates. The comprehensive MR screenings of 486 blood metabolites and NB additionally substantiated the accuracy and significance of the causality that increased levels of butyrylcarnitine significantly reduce the risk of NB.

Despite the above advantages, this study is subject to certain limitations. First, the study population was limited to European individuals due to the design principles of MR studies, which necessitate further verifications in other ethnic groups. Secondly, we chose SNPs with a less stringent significance threshold of 5 × 10^{− 5} to ensure an adequate number of matching matching IVs. The F-statistic was calculated to assess the risk of weak genetic instrument bias, with all results exceeding 10. Third, a potential limitation of this study is the lack of repeated measurements. However, previous studies indicate that blood metabolite levels are relatively stable, suggesting that a single measurement may be sufficient to capture medium-term exposure [26]. Finally, the causal relationship between circulating metabolites and NB was predicted by the genetic prediction method, rather than by molecular mechanisms, necessitating further validation of function and mechanism.

To sum up, our bidirectional MR research provides genetic support for the perspective that elevated levels of butyrylcarnitine causally reduce the risk of NB, positioning butyrylcarnitine as a potential protective factor for NB. The mechanism of how butyrylcarnitine inhibits the pathogenesis of NB needs further verification. Moreover, guanosine and 3-methyl-2-oxobutyrate, the metabolites affected by the causal effects of NB, may hold significance for NB diagnosis, efficacy evaluation, prognostic monitoring, and potential treatment targets after further study and assessment in the future.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

C.Y. and Q.D. contributed to concept acquisition and experimental design. H.W. and Z.Z. contributed to data analysis and manuscript preparation. C.C and Q.P. contributed to manuscript revision.

Funding

This study was financially supported by the Open Project of Jiangsu Key Laboratory of Experimental & Translational Non-coding RNA Research [202270] and the Science Foundation of Kangda College of Nanjing Medical University [KD2024KYJJ217].

Data availability

The dataset for NB (OpenGWAS ID: ieu-a-816) was obtained from the IEU Open GWAS project (https://gwas.mrcieu.ac.uk/datasets/ieu-a-816/), and genetic associations for 486 blood metabolites were collected from the Metabolomics GWAS Server (https://metabolomips.org/gwas/index.php? task=download). The complete list of metabolite accessions is presented in Supplementary Table 1.

Declarations

Ethics approval

This study only used publicly available data and hence no ethics approval was required. Details of ethical approval for each of the studies that contributed to the GWAS can be found in the original publications.

Consent to participate

Details of participant consent for each of the studies that contributed to the GWAS can be found in the original publications.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.