Causality of Genetically Determined Metabolites on Chronic Kidney Disease: A Two-Sample Mendelian Randomization Study In Silico

Abstract

Introduction:

Chronic kidney disease (CKD) is associated with metabolic disorders. However, the evidence for the causality of circulating metabolites to promote or prevent CKD is still lacking.

Methods:

The two-sample Mendelian randomization (MR) analysis was conducted to evaluate the latent causal relationship between the genetically proxied 486 blood metabolites and CKD. Genome-wide association study (GWAS) data for exposures were derived from 7824 European GWAS on metabolite levels, which have been extensively utilized in the medical field to elucidate the mechanisms underlying disease onset and progression. The random inverse variance weighted (IVW) is the primary analysis for causality analysis while MR-Egger and weighted median as complementary analyses. For the further identification of metabolites, reverse MR and linkage disequilibrium score regression were performed for further evaluation. The drug target for N-acetylornithine was subsequently supplemented into the analysis, with MR and colocalization analysis being utilized. Key metabolic pathways were identified via MetaboAnalyst 4.0 (https://www.metaboanalyst.ca/) online website.

Results:

N-acetylornithine was identified as a reliable metabolite that increases the susceptibility to estimated glomerular filtration rate (eGFR) decrease (β = 0.047; 95% confidence interval: −0.068 to −0.026; P _IVW = 1.5E-5). The “glyoxylate and dicarboxylate metabolism” pathway showed significant relevance to CKD development (P = 6E-4), whereas the “glycine, serine, and threonine metabolism” pathway was also recognized as associated with CKD by general practitioners (P = 7E-4). Colocalization analysis revealed a robust genetic link between N-acetylornithine and both CKD and eGFR, with 85.1% and 99.4% colocalization rates, respectively. IVW-MR analysis substantiated these findings with a significant positive association for CKD (odds ratio = 1.43, P = 4.7E-5) and a negative correlation with eGFR (b = −0.04, P = 1.13E-31).

Conclusions:

MR was utilized to explore the potential causal links between 61 genetic serum metabolites and CKD. N-acetylornithine and NAT8 were further explored as a potential therapeutic target for CKD treatment.

Introduction

The main function of the kidneys is to filter waste products in the blood, and glomerular filtration rate (GFR) reflects the rate at which the kidneys filter blood per unit of time.¹ Kidney disease has a significant hereditary predisposition, and the kidneys are not only the target organ for many diseases but also drive and influence the progression of many systemic diseases.² Chronic kidney disease (CKD) arises from diverse underlying causes, manifesting clinically as a decline in kidney function and pathologically characterized by the presence of renal fibrosis.³ Owing to its high global prevalence of 10%–16% in common people and over 50% in high-risk subpopulations and lack of therapeutic options,⁴ CKD imposes heavy burden to health.^3,5

Serum metabolites, from a more nuanced micro-level perspective, offer insights into the etiology and progression of diseases and guide the investigation of potential drug targets. The 486 serum metabolites utilized in this study are derived from prior research and have been extensively applied across various medical fields, including the exploration of sepsis⁶ and anxiety disorders.⁷ Preliminary investigations within this domain have focused on the mannose molecule, yet they have not delved into the exploration of further drug targets.⁸

In this article, we want to introduce a novel instrument, Mendelian randomization (MR),⁹ to assist us in confirming the potential relationship between blood metabolites and CKD. In contrast to various other methodologies used in epidemiological studies, MR offers a unique advantage by delivering impartial assessments regarding the allocation of genetic variants at the time of conception. This approach is typically impervious to the influences of confounding elements and the issue of reverse causation.^10
–12 This research methodology, which uses innate genotypes as instrumental variables (IVs), emulates the design of randomized controlled trials (RCTs), thereby enhancing the reliability of the conclusions drawn and significantly mitigating the risk of reverse causality.^13,14

Owing to its significant analytical prowess, MR has seen extensive application over the past 10 years. It is utilized to deduce the causal relationships between risk factors and disease outcomes by leveraging publicly accessible genome-wide association study (GWAS) summary data.^15,16

Studies on biomarkers for the diagnosis and pathology of CKD using MR have been numerous.¹⁷ The use of proteomic techniques in the blood of patients with CKD established testosterone-2 as an important cornerstone of clinically relevant physiological markers of kidney disease progression and pointed to other proteins that deserve further study.¹⁸ Previous studies have noted that high-density lipoprotein is a protective factor for kidney function, in patients with CKD, whereas low-density lipoprotein is not causally associated with CKD.¹⁹ Higher baseline levels of fibroblast growth factor-23, parathyroid hormone, and phosphate were associated with an increased risk of end-stage renal disease (ESRD) or death, respectively, independent of GFR.²⁰ Serum 6-bromotryptophan levels were identified as a risk factor for the exacerbation of CKD.²¹ Advanced glycation end products include methylglyoxal (MG)-derived hydro imidazolidinone MG-H1, Nε-carboxymethyl-lysine, and glucosinolates advance CKD.²² The single nucleotide polymorphism (SNP) rs13538, located within the NAT8 gene, has been identified as associated with N-acetylornithine levels in African American patients with CKD, further implicating its connection with CKD progression and estimated glomerular filtration rate (eGFR). We have taken note of this intriguing finding and are exploring its potential as a genetic target for CKD therapy.²³

We hypothesized that the genetically determined metabolites (GDMs) atlas could serve as a tool for deducing the causal influence of GDMs on CKD. To explore this, we used a two-sample MR strategy with the following objectives: (1) to evaluate the potential causal impacts of circulating human serum metabolites on the development of CKD, (2) to pinpoint specific GDMs exerting causal effects across four distinct GWAS related to CKD, (3) to uncover possible metabolic pathways that could shed light on the underlying mechanisms of anxiety disorders, and (4) to investigate drug targets for the prevention and delayed progression of CKD.

Methods

GWAS data for 486 blood metabolites

Genomic data for serum metabolites were obtained from the GWAS server. (http://metabolomics.helmholtz-muenchen.de/gwas/). By performing a GWAS study on the untargeted metabolome, Shin et al. obtained the most comprehensive serum metabolome GWAS data to date, screening 486 genetically influenced serum metabolites.²⁴ The researchers recruited 7824 participants from two European population cohorts, including 1768 participants in Germany and 6056 participants in the United Kingdom. Both studies were approved by local ethics committees, and all participants voluntarily signed informed consent forms prior to the study. After a rigorous screening process, 486 genetically influenced serum metabolites were ultimately screened, of which 309 are known, whereas the remaining 177 are unknown. Known metabolites are classified as peptides, nucleotides, amino acids, energy, cofactors and vitamins, lipids, carbohydrates, and xenobiotics, based on the Kyoto Encyclopedia of Genes and Genomes (KEGG) database. Ultimately, this GWAS incorporated 2.1 million SNPs.^24,25 More details can be found in Table 1. The flowchart of the study is shown in Fig. 1.

FIG. 1.

The overview of the study.

Table 1.

Basic Information About GWAS Datasets Involved

Phenotype	Resource	Diagonis	Sex	Sample size	Population	PMID
CKD	CKDgen	eGFR < 60 mL/min/1.73 m²	Males and females	133,814	European	26831199
CKD	ebi-a-GCST008026	eGFR < 60 mL/min/1.73 m²	Males and females	20,920	Hispanic or Latin American	31217584
eGFR	CKDgen	NA	Males and females	133,814	European	26831199
UACR	CKDgen	NA	Males and females	54,450	European	26631737

CKD, chronic kidney disease; eGFR, estimated glomerular filtration rate; UACR, urine albumin-to-creatinine ratio.

CKD genetics Consortium data

We analyzed the data gathered from the CKD genetics (CKDGen) Consortium to obtain effect estimates on three different parameters: quantitative eGFR, dichotomous eGFR < 60 mL/min/1.73 m², and albuminuria for the selected variants. Kidney function was assessed using eGFR, a quantitative measure adjusted for sex and age. Out of 133,814 participants, the average eGFR was 87.12 ± 23.09 mL/min/1.73 m². Albuminuria, quantified by the urine albumin-to-creatinine ratio (UACR) from spot urine samples, was assessed in a cohort of 54,450 participants. The analysis was performed using sex- and age-adjusted residuals following the natural logarithm transformation. Patients with diabetes were designated as the case group, whereas non-diabetic individuals served as the control group.²⁶ A cohort of 118,147 individuals with an eGFR below 60 mL/min/1.73 m² were designated as the case group for CKD.^20,27 More details can be found in Table 1.

Instruments selection

In order to satisfy the most important assumption of relevance in MR, we screened the IVs with strict criteria. We adjusted the significance threshold to P < 1 × 10⁻⁵ to select SNPs related to metabolites. Then, we clumped SNPs by removing linkage disequilibrium (LD) (R ² > 0.01) within 500 kb. The above criteria have proven to be reliable in the literature in the past.^28,29 To remove bias from faulty instruments, we computed R ² and F statistics for each SNP.²⁰

SNPs with an F value less than 10 were considered inadequate and subsequently excluded.³⁰ We then extracted the previously obtained SNPs in the outcome and excluded the SNPs associated with the outcome (P < 1E-5). Furthermore, we harmonized the exposure with the data in the outcome and synergized the effect alleles. Then to fulfil hypothesis 3, we excluded SNPs directly associated with the outcome (P < 1E-5).³¹

Estimation of causal effect and sensitivity analyses

The primary approach used for assessing the causal impact of potentially influential metabolites on eGFR, CKD, and UACR was the inverse variance weighted (IVW) analysis. This method aimed to evaluate the relationship between these metabolites and the outcome. To further investigate the presence of heterogeneity and pleiotropy effects, Cochran’s Q statistic and the corresponding P values were computed.³²

In order to effectively address the issue of horizontal pleiotropy, three complementary and rigorous sensitivity analyses were performed. First, the weighted median-based method was used, which yields a reliable estimate by considering valid IVs with a weight of at least 50%.³³ Second, the mode-based method was utilized, which offers a consistent effect estimate by assuming that the majority of the genetic variants are valid instruments, especially as the sample size increases.³⁴ Lastly, MR-Egger regression was used, providing a robust approach to test for potential horizontal pleiotropy.³⁵ These sophisticated analytical techniques collectively strengthen the validity and reliability of our findings. To validate any potential horizontal pleiotropic outliers that could significantly impact the estimation, we used two robust methods: Leave-one-out analysis and the MR pleiotropy residual sum and outlier (MR-PRESSO) method.³² These analytical approaches provide a rigorous assessment of the data, allowing us to identify and address any influential outliers that may compromise the accuracy of our estimation. By incorporating these comprehensive validation techniques, we further enhance the credibility and robustness of our findings.³⁶ In essence, we conducted a meticulous screening of blood metabolites for their potential causal effects on CKD, using a range of rigorous criteria: (1) The P value derived from the primary analysis demonstrated statistical significance (P < 0.05, IVW method). (2) Consistency was observed in both the direction and magnitude of effects across all three MR methods utilized. (3) No evidence of heterogeneity or horizontal pleiotropy was found in the MR results, ensuring robustness. (4) The MR estimates remained largely unaffected by the presence of a single SNP, indicating their stability and reliability. MR results of CKD, eGFR, and UACR can be found in Table 2, Table 3, and Table 4, respectively. SNPs selected for mannose and N-acetylornithine can be found in Table 8.

Table 2.

Results of Primary and Sensitivity Analysis About CKD

Metabolite	Super pathway	Outcome	NSNP	b_IVW	se_IVW	Pval_IVW	b _W m	se_Wm	Pval_Wm	b_MR-Egger	se_MR.Eggei	Pval_MR.Egger	Egger_intercept	Se	Pval	Or
2-Methoxyacetaminophen sulfate*	Xenobiotics	CKD	86	−0.038	0.013	0.004	−0.026	0.019	0.166	−0.057	0.026	0.035	0.005	0.006	0.411	0.963
4−Acetamidobutanoate	Amino acid	CKD	30	0.651	0.308	0.035	1.029	0.480	0.032	1.043	0.814	0.211	−0.004	0.008	0.607	1.917
ADpSGEGDFXAEGGGVR*	Peptide	CKD	5	0.471	0.192	0.014	0.306	0.253	0.227	0.357	0.672	0.632	0.005	0.030	0.870	1.601
Betaine	Amino acid	CKD	21	−1.007	0.333	0.003	−0.958	0.386	0.013	−1.976	0.961	0.054	0.015	0.014	0.296	0.365
Citrulline	Amino acid	CKD	44	0.814	0.280	0.004	0.673	0.402	0.094	1.659	1.145	0.155	−0.008	0.011	0.451	2.258
Gamma-glutamylglutamine	Peptide	CKD	21	0.794	0.360	0.027	1.010	0.485	0.037	1.483	1.207	0.234	−0.010	0.016	0.556	2.213
Glycerate	Carbohydrate	CKD	14	−0.766	0.390	0.050	−0.713	0.518	0.168	−3.429	1.975	0.108	0.036	0.026	0.194	0.465
Glycine	Amino acid	CKD	22	0.808	0.287	0.005	1.234	0.238	0.000	1.145	0.447	0.019	−0.009	0.010	0.336	2.243
Hydroxyisovaleroyl carnitine	Amino acid	CKD	5	−0.837	0.355	0.018	−0.936	0.420	0.026	−1.440	1.851	0.493	0.016	0.047	0.762	0.433
Athosterol	Lipid	CKD	15	−0.393	0.152	0.010	−0.388	0.211	0.066	−0.503	0.441	0.274	0.004	0.016	0.795	0.675
Malate	Energy	CKD	15	−0.800	0.397	0.044	−0.671	0.477	0.159	−2.965	2.103	0.182	0.032	0.030	0.314	0.449
Mannose	Carbohydrate	CKD	17	0.550	0.228	0.016	0.535	0.292	0.067	0.643	0.485	0.205	−0.002	0.011	0.831	1.733
N-acetylornithine	Amino acid	CKD	19	0.219	0.082	0.008	0.252	0.089	0.004	0.275	0.104	0.017	−0.005	0.005	0.396	1.245
Stearate (18:0)	Lipid	CKD	37	0.624	0.279	0.025	0.649	0.411	0.114	0.016	0.827	0.985	0.006	0.008	0.440	1.867
X-01911		CKD	16	−0.451	0.125	0.000	−0.406	0.175	0.021	−0.884	0.691	0.222	0.016	0.026	0.535	0.637
X-10346		CKD	10	0.203	0.098	0.039	0.170	0.132	0.197	0.207	0.193	0.313	0.000	0.013	0.978	1.225
X-12040		CKD	9	0.107	0.044	0.015	0.149	0.059	0.012	0.154	0.082	0.103	−0.009	0.013	0.516	1.113
X-12189		CKD	14	0.119	0.048	0.013	0.126	0.067	0.062	0.156	0.085	0.091	−0.005	0.010	0.603	1.126
X-12524		CKD	24	−1.207	0.512	0.018	−1.221	0.716	0.088	−4.703	3.294	0.167	0.033	0.030	0.294	0.299
X-13549		CKD	16	−0.971	0.486	0.046	−1.119	0.707	0.113	−1.253	1.683	0.469	0.003	0.016	0.864	0.379

IVW, inverse variance weighted; NSNP, Number of SNP.

Table 3.

Results of Primary and Sensitivity Analysis About eGFR

Metabolite	Super pathway	Outcome	NSNP	b_IVW	se_IVW	Pval IVW	b_Wm	se_Wm	Pval Wm	b MR.Egger	se_MR.Egger	Pval_MR.Egger	Egger_intercept	se	Pval
X-04499-3,4-dihydroxybutyrate	Amino acid	eGFR	21	−0.117	0.036	0.001	−0.068	0.049	0.168	−0.131	0.144	0.374	0.000	0.002	0.920
Valine	Amino acid	eGFR	4	−0.397	0.149	0.008	−0.471	0.193	0.014	−0.599	0.896	0.572	0.001	0.006	0.840
Urea	Amino acid	eGFR	13	−0.126	0.050	0.012	−0.071	0.067	0.294	−0.006	0.175	0.972	−0.002	0.003	0.490
Kynurenine	Amino acid	eGFR	37	−0.164	0.053	0.002	−0.026	0.053	0.628	−0.155	0.155	0.326	0.000	0.002	0.948
N-acetylornithine	Amino acid	eGFR	19	−0.047	0.011	0.000	−0.048	0.011	0.000	−0.052	0.014	0.002	0.000	0.001	0.537
Homocitrulline	Amino acid	eGFR	7	−0.089	0.041	0.029	−0.089	0.047	0.058	−0.228	0.104	0.080	0.004	0.003	0.212
Mannose	Carbohydrate	eGFR	17	−0.103	0.036	0.004	−0.099	0.042	0.019	−0.091	0.079	0.266	0.000	0.002	0.864
Glycerate	Carbohydrate	eGFR	14	0.150	0.058	0.009	0.134	0.070	0.055	0.155	0.317	0.633	0.000	0.004	0.987
Mannito	Carbohydrate	eGFR	10	−0.033	0.015	0.031	−0.038	0.019	0.051	−0.083	0.060	0.204	0.003	0.004	0.413
1,5-Anhydroglucitol (1,5-AG)	Carbohydrate	eGFR	29	−0.045	0.022	0.040	−0.063	0.032	0.049	−0.067	0.054	0.227	0.000	0.001	0.662
Erythronate*	Carbohydrate	eGFR	47	−0.090	0.044	0.042	−0.063	0.052	0.225	−0.069	0.197	0.728	0.000	0.002	0.912
Choline	Lipid	eGFR	22	−0.160	0.068	0.019	−0.100	0.096	0.293	−0.418	0.320	0.206	0.002	0.003	0.420
1-Linoleoylglycerol (1-monolinolein)	Lipid	eGFR		−0.043	0.019	0.020	−0.047	0.026	0.074	−0.082	0.068	0.263	0.001	0.003	0.576
Dehydroisoandrosterone sulfate (DHEA-S)	Lipid	eGFR	19	−0.042	0.020	0.035	−0.016	0.027	0.557	−0.038	0.067	0.580	0.000	0.002	0.949
1-Linoleoylglycerophosphoethanolamine*	Lipid	eGFR	13	0.053	0.027	0.047	0.035	0.037	0.350	0.017	0.070	0.806	0.001	0.002	0.587
5-Dodecenoate (12:1n7)	Lipid	eGFR	14	−0.069	0.028	0.012	−0.031	0.038	0.408	−0.073	0.083	0.393	0.000	0.002	0.961
1-Oleoylglycerophosphoethanolamine	Lipid	eGFR	10	0.075	0.037	0.045	0.054	0.049	0.269	0.066	0.220	0.772	0.000	0.004	0.970
Urate	Nucleotide	eGFR	19	−0.120	0.046	0.009	−0.103	0.060	0.087	−0.113	0.085	0.201	0.000	0.001	0.922
X-12244-N-acetyIcarnosine	Peptide	eGFR	22	−0.083	0.035	0.017	−0.062	0.043	0.156	−0.035	0.110	0.753	−0.001	0.002	0.652
Gamma-glutamylvaline	Peptide	eGFR	14	−0.103	0.047	0.029	−0.059	0.066	0.370	−0.049	0.264	0.856	−0.001	0.004	0.838
ADpSGEGDFXAEGGGVR*	Peptide	eGFR	5	−0.072	0.032	0.023	−0.042	0.032	0.187	−0.035	0.113	0.776	−0.002	0.005	0.754
X-06350		eGFR	12	0.121	0.042	0.003	0.124	0.055	0.025	0.226	0.212	0.311	-0.002	0.003	0.625
X-08766		eGFR	14	0.097	0.047	0.038	0.131	0.056	0.020	0.050	0.199	0.807	0.001	0.003	0.811
X-11261		eGFR	17	0.044	0.017	0.008	0.036	0.022	0.104	0.017	0.038	0.667	0.001	0.001	0.449
X-11787		eGFR	28	−0.117	0.039	0.003	−0.222	0.057	0.000	−0.198	0.067	0.007	0.001	0.001	0.154
X-12093		eGFR	9	0.043	0.018	0.018	0.046	0.017	0.007	0.056	0.039	0.196	−0.001	0.002	0.708
X-12405		eGFR	8	−0.055	0.026	0.038	−0.061	0.035	0.084	−0.020	0.089	0.829	−0.001	0.003	0.697
X-13477		eGFR	7	−0.146	0.073	0.045	−0.110	0.077	0.155	−0.482	0.307	0.178	0.006	0.005	0.313

NSNP, Number of SNP.

Table 4.

Results of Primary and Sensitivity Analysis About UACR

Metabolite	Super pathway	Outcome	NSNP	b_IVW	se_IVW	Pval_IVW	b Wm	se Wm	Pval_Wm	b MR.Egger	se_MR.Egger	Pval_MR.Egger	Egger_intercept	Se	Pval
X-12442-5,8-tetradecadienoate	Lipid	UACR	11	0.163	0.074	0.027	0.222	0.097	0.022	0.474	0.199	0.041	−0.011	0.007	0.127
1-Eicosatrienoylglycerophosphocholine*	Lipid	UACR	15	−0.245	0.100	0.014	−0.343	0.133	0.010	−0.673	0.318	0.054	0.010	0.007	0.182
Asparagine	Amino acid	UACR	39	−0.285	0.109	0.009	−0.115	0.162	0.477	−0.192	0.321	0.553	−0.001	0.003	0.761
Bilirubin (Z,Z)	Cofactors and vitamins	UACR	10	−0.089	0.031	0.005	−0.072	0.036	0.045	−0.049	0.053	0.381	−0.004	0.004	0.385
Glucose	Carbohydrate	UACR	32	−0.358	0.168	0.034	−0.287	0.241	0.235	−1.149	1.259	0.369	0.007	0.011	0.531
Mannose	Carbohydrate	UACR	17	−0.320	0.111	0.004	−0.437	0.132	0.001	−0.655	0.225	0.011	0.008	0.005	0.114
Salicyluric glucuronide*	Xenobiotics	UACR	5	−0.075	0.031	0.014	−0.071	0.039	0.065	−0.096	0.122	0.490	0.003	0.015	0.873
X-11799		UACR	8	0.102	0.040	0.011	0.093	0.049	0.059	0.067	0.083	0.451	0.003	0.006	0.644
X-11876		UACR	15	−0.185	0.085	0.028	−0.264	0.117	0.024	−0.082	0.381	0.832	−0.003	0.009	0.786
X-12116		UACR	8	−0.217	0.079	0.006	−0.312	0.106	0.003	−0.460	0.296	0.171	0.009	0.010	0.426
X-12206		UACR	10	0.418	0.143	0.003	0.299	0.183	0.102	0.543	0.620	0.407	-0.003	0.012	0.840
X-12253		UACR	7	−0.175	0.088	0.046	−0.131	0.122	0.283	−1.804	1.100	0.162	0.046	0.031	0.197
X-13671		UACR	13	−0.430	0.143	0.003	−0.430	0.193	0.026	−0.823	0.782	0.315	0.005	0.010	0.620

NSNP, Number of SNP.

Evaluation of genetic correlation and directionality

However, MR estimates may violate causal effects in cases where there is a genetic correlation between exposure and the outcome of interest.^37,38 Although SNPs related to CKD were excluded in the selection of IVs, there is still a possibility that SNPs with no direct relation to CKD may mediate the genetic pathways of CKD. To address this concern and ensure that the causal effects are not confounded by the coheritability of the exposure with the outcome, we utilized linkage disequilibrium score (LDSC) regression and performed a reverse MR analysis.³⁹ That is, MR analyses were performed using CKD, eGFR, and UACR as exposures and 486 serum metabolites as end points. LDSC regression estimates the coinheritance of two traits based on SNP data by performing chi-squared statistics. Therefore, LDSC was implemented in this study to examine the genetic correlation between the screened metabolites and CKD. More detailed results of LDSC can be found in Table 5.

Table 5.

LDSC Results of 61 Associations to Outcomes Respectively

Exposure	Outcome	RG	RG SE	RGP
M01303.metal.pos.txt.gz	CKD	−0.138	0.312	0.659
MO1358.metal.pos.txt.gz	CKD	−0.208	0.119	0.079
MO1558.metal.pos.txt.gz	CKD	0.201	0.132	0.127
M01572.metal.pos.txt.gz	CKD	0.180	0.364	0.622
M02132.metal.pos.txt.gz	CKD	0.237	0.125	0.058
M02730.metal.pos.txt.gz	CKD	−0.166	0.420	0.693
M15630.metal.pos.txt.gz	CKD	0.577	0.512	0.260
M32388.metal.pos.txt.gz	CKD	−0.029	0.464	0.950
M32735.metal.pos.txt.gz	CKD	0.290	0.240	0.227
M33391.metal.pos.txt.gz	CKD	−0.415	0.253	0.101
M33488.metal.pos.txt.gz	CKD	0.215	0.201	0.284
M33610.metal.pos.txt.gz	CKD	−0.298	0.210	0.157
M33782.metal.pos.txt.gz	CKD	0.172	0.265	0.515
M34106.metal.pos.txt.gz	CKD	0.020	0.866	0.981
M34327.metal.pos.txt.gz	CKD	0.246	0.366	0.502
M35327.metal.pos.txt.gz	CKD	−1.522	0.679	0.025
M36399.metal.pos.txt.gz	CKD	0.101	0.377	0.788
M01572.metal.pos.txt.gz	eGFR	0.301	0.228	0.187
M01604.metal.pos.txt.gz	eGFR	−0.425	0.169	0.012
MO1649.metal.pos.txt.gz	eGFR	−0.057	0.109	0.597
M15140.metal.pos.txt.gz	eGFR	−0.228	0.087	0.009
M15335.metal.pos.txt.gz	eGFR	−0.353	0.264	0.180
M15506.metal.pos.txt.gz	eGFR	−0.632	0.277	0.022
M15630.metal.pos.txt.gz	eGFR	−0.792	0.354	0.025
M16822.metal.pos.txt.gz	eGFR	−0.044	0.150	0.770
M20675.metal.pos.txt.gz	eGFR	−0.156	0.115	0.177
M22032.metal.pos.txt.gz	eGFR	−0.026	0.569	0.963
M22138.metal.pos.txt.gz	eGFR	−0.149	0.163	0.361
M32393.metal.pos.txt.gz	eGFR	−0.317	0.216	0.141
M32425.metal.pos.txt.gz	eGFR	−0.012	0.222	0.957
M32578.metal.pos.txt.gz	eGFR	−0.256	0.254	0.312
M33477.metal.pos.txt.gz	eGFR	−0.028	0.077	0.719
M33508.metal.pos.txt.gz	eGFR	−0.232	0.369	0.529
M33666.metal.pos.txt.gz	eGFR	−0.619	0.237	0.009
M33833.metal.pos.txt.gz	eGFR	−0.349	0.299	0.244
M35628.metal.pos.txt.gz	eGFR	−0.194	0.211	0.358
M20489.metal.pos.txt.gz	UACR	0.031	0.173	0.855
M27716.metal.pos.txt.gz	UACR	0.064	0.605	0.916
M33144.metal.pos.txt.gz	UACR	−0.132	0.267	0.620
M34283.metal.pos.txt.gz	UACR	−0.113	0.117	0.336
M35464.metal.pos.txt.gz	UACR	−0.152	0.148	0.303

SE, Standard Error; RG, Regression; RGP, Regression p value.

Colocolization and drug target analysis

Bayesian colocalization, leveraging the “coloc” package with default settings, was applied to assess the probability of overlapping causal variants for two traits. Detailed methodology and package source code are available at the GitHub repository (https://github.com/chr1swallace/coloc). This analysis quantifies the posterior probability for five hypotheses, with a focus on Hypothesis 3, which assumes independent variants link both protein levels and molecular signatures to a common genomic region, and Hypothesis 4 (PPH4), which proposes shared variants for both associations. We used both the coloc.abf and coloc.susie algorithms, with colocalization evidence defined by a gene-based PPH4 exceeding 80%, as indicated by either algorithm.⁴⁰

We further developed an instrumental variable by identifying SNPs within a 100 kb window of the NAT8 gene, which are genome-wide significantly (P < 5E-8) associated with N-acetylornithine levels, to serve as proxies for the exposure to potential target drugs. To consolidate the findings, we utilized the IVW-MR approach to synthesize effect estimates derived from genetic variants linked to N-acetylornithine levels. In addition, we validated our results using external GWAS databases, specifically the ebi-a-GCST008026 and ieu-a-104 datasets.⁴¹ More detailed results of colocolization and drug target analysis can be found in Table 6.

Table 6.

Colocalization Results and Drug Target Analysis

Outcome	Exposure	Egger_intercept	se	Pval	h0.abf	h1.abf	h2.abf	h3.abf	h4.abf	Correct_causal_direction	Steiger pval
CKD	N-acetylornithine	−0.013	0.042	0.787	0.000	0.138	0.000	0.010	0.851	TRUE	0.000
eGFRcrea	N-acetylornithine	0.000	0.002	0.951	0.000	0.000	0.000	0.006	0.994	TRUE	0.000

Metabolic pathway analysis

To explore the function of the identified serum metabolites, we used the MetaboAnalyst 4.0 (https://www.metaboanalyst.ca/) online website to perform the analyses. In order to obtain comprehensive and reliable results, all metabolites associated with CKD, UACR, and eGFR at P < 0.05 were included in the analysis. More detailed results of pathway enriched can be found in Table 7.

Table 7.

Primary Enrichment Results

Pathway	Outcome	FDR	Impact
Glyoxylate and dicarboxylate metabolism	CKD	0.028	0.185
Glycine, serine, and threonine metabolism	CKD	0.028	0.320
Arginine biosynthesis	CKD	0.074	0.228
Glycerolipid metabolism	CKD	1.000	0.093
Citrate cycle (TCA cycle)	CKD	1.000	0.044
Pentose phosphate pathway	CKD	1.000	0.000
Pyruvate metabolism	CKD	1.000	0.031
Glutathione metabolism	CKD	1.000	0.089
Porphyrin and chlorophyll metabolism	CKD	1.000	0.000
Biosynthesis of unsaturated fatty acids	CKD	1.000	0.000
Arginine and proline metabolism	CKD	1.000	0.000
Steroid biosynthesis	CKD	1.000	0.059
Primary bile acid biosynthesis	CKD	1.000	0.008
Aminoacyl-tRNA biosynthesis	CKD	1.000	0.000
Glycolysis/Gluconeogenesis	UACR	1.000	0.000
Alanine, aspartate, and glutamate metabolism	UACR	1.000	0.000
Porphyrin and chlorophyll metabolism	UACR	1.000	0.053
Aminoacyl-tRNA biosynthesis	UACR	1.000	0.000
Arginine biosynthesis	eGFR	0.130	0.000
Glycine, serine, and threonine metabolism	eGFR	0.363	0.024
Purine metabolism	eGFR	0.748	0.000
Valine, leucine, and isoleucine biosynthesis	eGFR	0.748	0.000
Glycerolipid metabolism	eGFR	1.000	0.093
Pantothenate and CoA biosynthesis	eGFR	1.000	0.000
Pentose phosphate pathway	eGFR	1.000	0.000
Glyoxylate and dicarboxylate metabolism	eGFR	1.000	0.079
Glycerophospholipid metabolism	eGFR	1.000	0.026
Valine, leucine, and isoleucine degradation	eGFR	1.000	0.000
Tryptophan metabolism	eGFR	1.000	0.094
Aminoacyl-tRNA biosynthesis	eGFR	1.000	0.000

FDR, False discovery rate.

Results

IVs selection

The range of instruments for each metabolite varied from 4 to 86, whereas two metabolites lacking the minimum requirement of three instruments were excluded from the MR analysis. The minimum F statistic for instrumental SNPs exceeded 10 for each metabolite, demonstrating the substantial strength of the instruments. The primary impact of these SNPs on metabolites is observed at the blood level, specifically at the translational level, with no investigation conducted into their functional implications.

Causal effects on 3 datasets related to CKD phenotype

In the IVW model, a total of 61 significant metabolite associations (IVW, P < 0.05) with three CKD-related phenotypes were successfully identified. Out of these associations, 39 were linked to known metabolites, whereas 22 associations were found for previously unidentified metabolites. 14,19,6 associations of known metabolites and 6,7,6 associations of unknown metabolites for CKD, eGFR, and UACR are shown in Fig. 2. MR results of CKD, eGFR and UACR are detailed in Figs. 3–5, respectively.

FIG. 2.

Mendelian randomization associations of known metabolites on the risk of the three different GWAS datasets of CKD, eGFR, and UACR (derived from IVW analysis). CKD, chronic kidney disease; eGFR, estimated glomerular filtration rate; GWAS, genome-wide association study; IVW, inverse variance weighted; UACR, urine albumin-to-creatinine ratio.

FIG. 3.

A forest plot about causal associations between 20 causal metabolites and CKD showing NSNPs, OR, OR (95% CI) and P value. CI, confidence interval; OR, odds ratio.

FIG. 4.

A forest plot about causal associations between 28 causal metabolites and eGFR showing NSNPs, OR, OR (95% CI) and P value.

FIG. 5.

A forest plot about causal associations between 13 causal metabolites and UACR showing SNPs, OR, OR (95% CI) and P value.

Following multiple-testing correction, a single robust association (IVW, P < 1.03E-4) was identified for eGFR, specifically with N-acetylornithine (P _IVW = 1.51E-5). Notably, the IVW analysis indicated that N-acetylornithine significantly heightened the risk of eGFR reduction [β = 0.047; 95% confidence interval (CI): −0.068 to −0.026; P _IVW = 1.5E-5]. Furthermore, alternative methods such as weighted median analysis (β = 0.048, 95% CI: −0.068 to −0.027, P = 5.77E-6) and MR-Egger analysis (β = 0.052, 95% CI: −0.079 to −0.025, P = 1.5E-3) consistently supported these findings. In addition, a reverse MR analysis was conducted, revealing no causal link between eGFR and N-acetylornithine. Consequently, the study’s findings offer compelling evidence for a causal association between N-acetylornithine and CKD, thereby confirming the reliability of the results.

However, by intersecting the serum metabolites derived from the three analyses, a metabolite of potential significance across all three phenotypes was identified, which is mannose. The relationship between mannose and 3 CKD phenotypes are shown in Fig. 6.

FIG. 6.

Venn plot for the crossover of three phenotypic metabolites-mannose.

Sensitivity

Although the IVW method is highly effective in establishing causal relationships between an exposure and a disease outcome, it is susceptible to weak instrument bias. To ensure the robustness of the findings, additional sensitivity and pleiotropy analyses were conducted.

The sensitivity analyses result for N-acetylornithine on the eGFR is shown in Tables 2–4. In the case of the IVW method, the Cochran’s Q statistic yielded a value of 22.0 (P = 0.232). Similarly, for the MR-Egger method, the Cochran’s Q statistic was 21.49 (P = 0.205). These results from the Cochran’s Q test indicate minimal heterogeneity within the data.

Upon examining the relationship between IVs and results, no significant intercept was found [intercept = 0.0005, Standard Error (SE) = 0.0008, P = 0.54]. In addition, the MR-PRESSO analysis revealed no evidence of horizontal pleiotropy, as demonstrated by the absence of any noticeable asymmetry in the funnel plot. These findings indicate the absence of heterogeneity and horizontal pleiotropy in our MR study, reinforcing the reliability of the causal effect of N-acetylornithine on eGFR. Notably, all metabolites that successfully passed all sensitivity analyses (P < 0.05). Furthermore, the MR-Egger regression analysis, which aimed to assess the presence of horizontal pleiotropy for the 61 associated metabolites, revealed no remarkable intercept. Consistent with these results, the Cochran’s Q test indicated minimal heterogeneity. Details can be found in Table 8.

Table 8.

Information About SNPs of N-Acetylornithine and Mannose

Evaluation of genetic correlation and directionality

Regarding the LDSC, owing to limitations of the exposure and outcome database, some metabolite scores could not be calculated. In addition, a reverse MR analysis had been previously conducted. Taking into account the results from both methods, it is suggested that the LDSC results may serve as supplementary references rather than as definitive evidence. LDSC-based estimates noted little genetic correlation observed between our blood metabolites and three CKD phenotypes except for CKD and X-13549 (RG = −1.52, SE = 0.68, P = 0.03), eGFR and urate (RG = −0.43, SE = 0.17, P = 0.01), kynurenine (RG = −0.23, SE = 0.09, P = 0.01), choline (RG = −0.63, SE = 0.28, P = 0.02), N-acetylornithine (RG = −0.79, SE = 0.35, P = 0.03), X-12244–N-acetylcarnosine (RG = −0.62, SE = 0.24, P = 0.01). The MR Steiger directionality test results confirmed the accuracy of our causal inference (P < 0.001). Details can be found in Table 5.

Colocolization and drug target analysis

The colocalization analysis results are predominantly contingent upon PPH4, which indicates the likelihood of shared genetic predispositions at the SNP level for both exposure and outcome. Specifically, for N-acetylornithine, the PPH4 posterior probability was 85.1% for its association with CKD and 99.4% for eGFR, suggesting a strong genetic correlation.

Four common SNPs [minor allele frequency (MAF) > 1%] in low linkage disequilibrium (R ² < 0.30), significantly associated with N-acetylornithine, were identified within ±100 kb of the NAT8 region (P < 5E-8). Furthermore, IVW-MR analysis revealed a positive association between NAT8-mediated N-acetylornithine and CKD [odds ratio (OR) = 1.43, 95% CI: 1.20–1.70, P = 4.7E-5) and eGFR (β = −0.04, 95% CI: −0.02 to −0.06, P = 1.13E-31), underscoring the importance of NAT8 in kidney function regulation.

These findings were robust to sensitivity analyses and free from pleiotropy, confirming the reliability of the results. Results are shown in Fig. 7.

FIG. 7.

(A) Colocalization result showing shared SNPs between N-acetylornithine and CKD. (B) Colocalization result showing shared SNPs between N-acetylornithine and eGFR. (C) A forest plot about drug target NAT8 associated with N-acetylornithine level to CKD. (D) A forest plot about drug target NAT8 associated with N-acetylornithine level to eGFR. SNP, single nucleotide polymorphism.

Metabolic pathway analyses

The metabolic pathway analyses revealed the identification of two crucial metabolic pathways that are primarily involved in the development of CKD. The findings indicated that the “Glyoxylate and dicarboxylate metabolism” pathway displayed potential relevance to the development of CKD (P = 0.0006, FDR = 0.028). In addition, the “glycine, serine and threonine metabolism” pathway was identified as being associated with CKD by general practitioners (P = 0.0007, FDR = 0.028). Results are shown in Fig. 8.

FIG. 8.

Enriched significant metabolic pathways of CKD.

Discussion

In this two-sample MR study, 61 causal relationships for three CKD phenotypes were identified, all of which passed sensitivity analyses. One relationship showed multiple statistically significant tests even after correction and was considered the most reliable marker. Another relationship was observed in all three phenotypes simultaneously, suggesting its crucial role in CKD onset and progression. In addition, two metabolic pathways associated with CKD were identified. The research approach was enhanced by incorporating colocalization methods and expression quantitative trait locus analyses to connect metabolites with genetic variants, thus advancing the identification of potential drug targets. Previous articles have been mainly limited to methodological constraints that do not allow the demonstration of a causal relationship between genetically significantly influenced serum metabolites and the progression of CKD, or to lipid²⁰ or proteomic studies.¹⁹ The study offers novel insights into the interaction between genes and the environment in the development of CKD, thereby deepening the understanding of its etiology. The findings indicate potential metabolic targets and the gene target NAT8 ²³ as therapeutic targets for CKD, suggesting that the condition could be prevented and improved with long-term treatment.

In the amino acid or peptide pathway, 13 amino acids and three peptides have been identified as causally associated with the development of CKD. Among them, three are considered protective factors: betaine, hydroxyisovaleroyl carnitine, and asparagine. The remaining amino acids and peptides are classified as risk factors for CKD. These include N-acetylornithine, 4-acetamidobutanoate, glycine, citrulline, γ-glutamylglutamine, valine, urea, kynurenine, homocitrulline, 3,4-dihydroxybutyrate, N-acetylcarnosine, γ-glutamylvaline, and ADpSGEGDFXAEGGGVR*. It is worth noting that in the African American population, N-acetylornithine has been identified as an important factor in the development and progression of CKD. This was demonstrated through RCTs showing a reduction in GFR.²³ In a cross-sectional study, it was observed that valine and asparagine were significantly elevated in patients with diabetic nephropathy. Valine was predicted to be associated with a rapid decline in eGFR and was identified as an independent risk factor for CKD. In contrast, the exploration of the role of asparagine was limited owing to constraints inherent in the study methodology itself.⁴² In patients with hepatorenal dysfunction in cirrhosis, it has been observed that 4-acetamidobutanoate is significantly elevated. However, its specific role in CKD requires further in-depth study and investigation.⁴³ The decreased renal function has been found to be associated with alterations in microbial composition, glycine metabolism, and polyamine metabolism, which can contribute to renal fibrosis. In addition, hypertension has shown a strong positive correlation with dysregulation of glycine metabolism and microbiota.⁴⁴ Previous literature has confirmed that citrulline serves as a marker of renal metabolism and is associated with the development of CKD. This association has been established through urinary isotope dilution studies. However, the literature has not yet been able to identify the specific antecedents or precursors of citrulline in relation to CKD.⁴⁵ Betaine has been observed to have beneficial effects in improving renal fibrosis in patients with CKD. Its mechanisms of action are attributed to its anti-inflammatory and antioxidant properties.⁴⁶ In an MR study, it was found that an obesity-induced increase in blood urea levels is a risk factor for CKD. This suggests that elevated blood urea, which can be influenced by obesity, contributes to the development of CKD.⁴⁷ The kynurenine pathway, which is the primary metabolic pathway for tryptophan degradation, has been found to be positively associated with the development of new-onset CKD and poor renal prognosis. This pathway is considered a risk factor for CKD, likely because of its promotion of renal fibrosis. These findings have been consistently observed in three large-cohort longitudinal studies.^48
–50 In metabolomics studies, N-acetylcarnosine, homocitrulline, and γ-glutamylvaline have been identified as risk factors for CKD progression. However, it is important to note that these studies have not established a causal relationship between these metabolites and the progression of CKD. Although they have shown an association, further research is needed to determine the underlying mechanisms and establish causation.^51
–53 Studies of hydroxyisovaleroyl carnitine, γ-glutamylglutamine, 3,4-dihydroxybutyrate, and ADpSGEGDFXAEGGGVR* in CKD are still lacking and need to be initiated.

In the lipid pathway, 10 lipids were identified as being causally associated with the development of CKD. Four of them are protective factors: lathosterol, 1-linoleoylglycerophosphoethanolamine*, 1-oleoylglycerophosphoethanolamine, and 1-eicosatrienoylglycerophosphocholine*. The others are risk factors, which include stearate (18:0), choline, 1-linoleoylglycerol (1-monolinolein), dehydroisoandrosterone sulfate (DHEA-S), 5-dodecenoate (12:1n7), and X-12442–5,8-tetradecadienoate. Lathosterol is a marker of cholesterol synthesis. In two case–control studies, it was found that populations with lower eGFR had lower levels of lathosterol, and lathosterol was associated with higher eGFR. However, neither of these studies discussed the relationship between lathosterol and the development of CKD.^54,55 Deposition of lipids, such as stearate (18:0), in the kidneys has been implicated in the progression of inflammation in the kidneys.⁵⁶ Choline, which is produced by the gut microbiota, has been found to be associated with the onset and progression of CKD. However, many aspects of this association still need to be further investigated and clarified.⁵⁷ It has been found in a network pharmacology study that 1-monolinolein was identified as one of the key targets for the treatment of diabetic nephropathy. However, its role in the progression of CKD development has not yet been extensively studied or focused on.⁵⁸ The observation that patients with CKD often have lower levels of DHEA-S and are associated with increased mortality may appear contradictory to our findings. However, the existing literature does not extensively discuss or study the relationship between DHEA-S and the development of CKD in healthy individuals. Therefore, the causal relationship between DHEA-S and the development of CKD remains unclear and requires further investigation.^59,60 The role played by 1-linoleoylglycerophosphoethanolamine, 1-oleoylglycerophosphoethanolamine, 1-eicosatrienoylglycerophosphocholine*, 5-dodecenoate (12:1n7), and X-12442–5,8-tetradecadienoate in the development and progression of CKD has not been addressed by studies.

In the carbohydrate pathway, six carbohydrates and one energy-related component have been identified as causally associated with the development of CKD. Among them, three are identified as protective factors: glycerate, malate, and glucose. In addition, there are three risk factors: mannitol, 1,5-anhydroglucitol (1,5-AG), and erythronate*. The effect value of mannose differed among the three outcomes and was a bidirectional factor for CKD. The fact that malate is a protective factor in CKD seems to contradict the previous finding in the literature that tricarboxylic acid pathway metabolites, such as malate, are elevated in patients with CKD. However, the previous literature is deficient in that only correlations can be made, and no analytical studies have been carried out to argue for causality.^61,62 Glucose has been shown to be a high risk factor for kidney failure in diabetic nephropathy, thus seemingly contradicting our results. However, since not all of our patients with CKD were because of diabetic nephropathy, the results are still valid, and further stratification of the sample is needed to explore the role in renal diseases other than diabetic nephropathy as well as in CKD. Another possible explanation is that the relationship between glucose and outcome is nonlinear and therefore within a certain range our results are worth referring to, as glucose is the main substance for tubular energy supply. Mannose is analyzed in eGFR and UACR with opposite effects on CKD, whereas it is an important substance for complement activation and may have a dual role in the progression of CKD, which is worth exploring.^63,64 Mannitol dilates renal blood vessels and is associated with contrast-induced nephropathy, but the exact mechanism remains unexplored.⁶⁵ Recent studies have shown that 1,5-AG can improve the performance of patients with CKD by inhibiting oxidative stress and inflammation to attenuate renal fibrosis.⁶⁶ Glycerat has been shown to reduce the risk of CKD.⁵² Erythronate can be used as a serum metabolite for estimating GFR, but the association with CKD is not clear.⁶⁷

In addition, it was found that urate, part of the nucleotide pathway, acts as a risk factor, whereas salicyluric glucuronide, which belongs to Xenobiotics, serves as a protective factor. In addition, bilirubin (Z,Z), belonging to the cofactors and vitamins pathway, is also identified as a protective factor. Urate has been recognized as a risk factor for CKD in observational studies. However, an MR article demonstrated that there does not seem to be a causal relationship between urate and CKD. To further investigate this matter, this article aims to explore the association between urate and decreased eGFR, which clarifies the inconsistent conclusions found in experimental studies and previous MR studies.^68,69 Previous observational and MR studies have consistently identified bilirubin as a protective factor against the development and progression of CKD. The findings correspond with prior studies, underscoring the necessity for additional research to elucidate the precise mechanisms through which bilirubin provides its protective effects in CKD. Understanding these mechanisms in depth will provide valuable insights into potential therapeutic interventions for CKD.^70
–72 Salicyluric glucuronide* in the development and progression of CKD has not been addressed by studies.

Moreover, 10 protective factors and 9 risk factors were found but are unknown material which deserve to be explored in depth.

The strength of our study is that it provides a more comprehensive summary of the relationship between 486 genetically influenced blood metabolites and the onset and development of CKD. Many findings in the literature support our findings, which adds much credibility to the article. These metabolites are potential CKD risk factors or protective factors, especially several metabolites that are directly causally related to the development of CKD. Their mechanisms in the development of CKD deserve to be further explored, and they can also be tried as predictors of CKD development. Those associated with eGFR and UACR, in contrast, are likely to play an important role in the development of CKD and deserve further exploration for their value as indicators of CKD progression and prognosis.

However, this study has several limitations. First, the GWAS data were not stratified based on the primary cause of CKD, despite multiple causes being involved. Second, the sample primarily consisted of a European population, necessitating validation through expansion to other ethnicities. Third, the effect values obtained through MR do not indicate the direct benefits from interventions as the genetic variant’s impact on exposure is a long-term effect. Fourth, owing to limited data availability, the study used metabolites with uncorrected p values for pathway analysis, which may introduce biases or incomplete understanding of underlying mechanisms. Lastly, although the study identified metabolites associated with the risk of CKD, further research is required to elucidate their specific roles in the pathogenic mechanisms of CKD.

In recent years, the role of chronic low-grade inflammation in the development and progression of CKD has gained recognition. Serum metabolites, such as peptides, lipids, and glycans, have been discovered to serve as either proinflammatory or anti-inflammatory mediators within the realm of chronic inflammation.^73
–75 Furthermore, CKD is typically considered irreversible and tends to worsen over time. To address this, advanced MR methods were utilized to identify both risk and protective factors associated with the onset of CKD. MR studies have emerged as a burgeoning trend in drug target identification, offering enhanced preclinical preparation for clinical pharmacological research.⁷⁶ This approach ensures that resources are allocated to potential drug targets with a higher likelihood of success. In an RCT, a site mutation in NAT8 was linked to adverse outcomes in African American patients with CKD.²³ The study, which included a European population, supports the potential efficacy of NAT8 as a drug target, thus paving the way for further investigation. MR studies targeting drug interventions for CKD are relatively scarce. Current examples include GLUT9 inhibitors,⁷⁷ which protect the kidneys by lowering uric acid levels, and SLC13A5 inhibitors, where plasma calcium and citrate serve as biologically plausible target biomarkers.⁷⁶

The MR research conducted in this study identified 61 causal relationships that are potentially linked to the pathogenesis of CKD. Significantly, N-acetylornithine has been identified as having a substantial influence on CKD. Consequently, colocalization analyses were conducted, and the effects of NAT8-mediated N-acetylornithine on CKD and eGFR were examined to strengthen the evidence supporting its potential as a drug target. The study also highlighted two metabolic pathways that are likely relevant to the pathology of CKD. These findings offer valuable insights into the use of certain metabolites as potential biomarkers for exploring targeted drugs in the treatment of human diseases. However, further research is warranted to elucidate the specific mechanisms by which these metabolites contribute to either adverse or favorable outcomes, and the associated drug research is deemed worthy of investment.

Footnotes

Acknowledgment

The authors would like to thank Mr. Guozi for his support in programming skills and Mr. Yiyu for his advice on code and essay writing.

Authors’ Contributions

Research idea and study design: Z.Z. and B.C.; data acquisition: Z.Z.; data analysis/interpretation: Z.Z., B.C., and Q.W.; statistical analysis: Z.Z., B.C., and Q.W.; supervision or mentorship: Z.Z. and Q.W. Each author contributed important intellectual content during article drafting or revision and accepts accountability for the overall work by ensuring that questions pertaining to the accuracy or integrity of any portion of the work are appropriately investigated and resolved.

Study Approval Statement

No more ethical approval was required for this study because it was a secondary analysis based on publicly available datasets. The original studies have obtained local ethical approvals and informed consent from the participants.

Data Availability Statement

Publicly available datasets were analyzed in this study. These data can be found here: () [CKDGEN Meta-Analysis datasets (uni-freiburg.de)] and [IEU OpenGWAS project (mrcieu.ac.uk)].

Author Disclosure Statement

No conflicting financial interests exist.

Consent to Participate Statement

Patient consent was not required as this study was based on publicly available data. All participants provided informed consent in their corresponding original studies, as described in the Methods section.

Funding Information

This study was not supported by any sponsor or funder.

References

Eckardt

, Coresh

, Devuyst

, et al. Evolving importance of kidney disease: From subspecialty to global health burden. Lancet, 2013; 382(9887):158–169; doi: 10.1016/s0140-6736(13)60439-0

White

, Polkinghorne

, Atkins

, et al. Comparison of the prevalence and mortality risk of CKD in Australia using the CKD Epidemiology Collaboration (CKD-EPI) and Modification of Diet in Renal Disease (MDRD) Study GFR estimating equations: The AusDiab (Australian Diabetes, Obesity and Lifestyle) Study. Am J Kidney Dis, 2010; 55(4):660–670; doi: 10.1053/j.ajkd.2009.12.011

Global, regional, and national burden of chronic kidney disease, 1990-2017: A systematic analysis for the Global Burden of Disease Study 2017. Lancet, 2020; 395(10225):709–733; doi: 10.1016/s0140-6736(20)30045-3

Sanchez-Niño

, Sanz

, Ramos

, et al. Clinical proteomics in kidney disease as an exponential technology: Heading towards the disruptive phase. Clin Kidney J, 2017; 10(2):188–191; doi: 10.1093/ckj/sfx023

Hill

, Fatoba

, Oke

, et al. Global prevalence of chronic kidney disease—A systematic review and meta-analysis. PLoS One, 2016; 11(7):e0158765; doi: 10.1371/journal.pone.0158765

Shang

, Qian

, Zhang

, et al. Human blood metabolites and risk of sepsis: A Mendelian randomization investigation. Eur J Clin Invest, 2023; 54(4):e14145; doi: 10.1111/eci.14145

Xiao

, He

, Liu

, et al. Causality of genetically determined metabolites on anxiety disorders: A two-sample Mendelian randomization study. J Transl Med, 2022; 20(1):475; doi: 10.1186/s12967-022-03691-2

Yin

, Shan

, Han

, et al. Causal effects of human serum metabolites on occurrence and progress indicators of chronic kidney disease: A two-sample Mendelian randomization study. Front Nutr, 2023; 10:1274078; doi: 10.3389/fnut.2023.1274078

Rabbani

, Thornalley

. Advanced glycation end products in the pathogenesis of chronic kidney disease. Kidney Int, 2018; 93(4):803–813; doi: 10.1016/j.kint.2017.11.034

10.

Sekula

, Del Greco

, Pattaro

, et al. Mendelian randomization as an approach to assess causality using observational data. J Am Soc Nephrol, 2016; 27(11):3253–3265; doi: 10.1681/asn.2016010098

11.

Pierce

, Burgess

. Efficient design for Mendelian randomization studies: Subsample and 2-sample instrumental variable estimators. Am J Epidemiol, 2013; 178(7):1177–1184; doi: 10.1093/aje/kwt084

12.

Burgess

, Scott

, Timpson

, et al. Using published data in Mendelian randomization: A blueprint for efficient identification of causal risk factors. Eur J Epidemiol, 2015; 30(7):543–552; doi: 10.1007/s10654-015-0011-z

13.

Richmond

, Davey Smith

. Mendelian randomization: Concepts and scope. Cold Spring Harb Perspect Med, 2022; 12(1); doi: 10.1101/cshperspect.a040501

14.

Hemani

, Zheng

, Elsworth

, et al. The MR-base platform supports systematic causal inference across the human phenome. Elife, 2018; 7; doi: 10.7554/eLife.34408

15.

Wang

, Liu

, Tian

, et al. Is hearing impairment causally associated with falls? Evidence from a two-sample Mendelian randomization study. Front Neurol, 2022; 13:876165; doi: 10.3389/fneur.2022.876165

16.

Zeng

, Ge

, Xu

, et al. Type 2 diabetes is causally associated with reduced serum osteocalcin: A genomewide association and Mendelian randomization study. J Bone Miner Res, 2021; 36(9):1694–1707; doi: 10.1002/jbmr.4330

17.

Levey

, Atkins

, Coresh

, et al. Chronic kidney disease as a global public health problem: Approaches and initiatives—a position statement from kidney disease improving global outcomes. Kidney Int, 2007; 72(3):247–259; doi: 10.1038/sj.ki.5002343

18.

Zhang

, Parikh

. Biomarkers of acute and chronic kidney disease. Annu Rev Physiol, 2019; 81:309–333; doi: 10.1146/annurev-physiol-020518-114605

19.

Matías-García

, Wilson

, Guo

, et al. Plasma proteomics of renal function: A transethnic meta-analysis and Mendelian randomization study. J Am Soc Nephrol, 2021; 32(7):1747–1763; doi: 10.1681/asn.2020071070

20.

Lanktree

, Thériault

, Walsh

, et al. HDL cholesterol, LDL cholesterol, and triglycerides as risk factors for CKD: A Mendelian randomization study. Am J Kidney Dis, 2018; 71(2):166–172; doi: 10.1053/j.ajkd.2017.06.011

21.

Scialla

, Astor

, Isakova

, et al. Mineral metabolites and CKD progression in African Americans. J Am Soc Nephrol, 2013; 24(1):125–135; doi: 10.1681/asn.2012070713

22.

Tin

, Nadkarni

, Evans

, et al. Serum 6-bromotryptophan levels identified as a risk factor for CKD progression. J Am Soc Nephrol, 2018; 29(7):1939–1947; doi: 10.1681/asn.2017101064

23.

Luo

, Surapaneni

, Zheng

, et al. NAT8 variants, N-acetylated amino acids, and progression of CKD. Clin J Am Soc Nephrol, 2020; 16(1):37–47; doi: 10.2215/cjn.08600520

24.

Shin

, Fauman

, Petersen

, et al. An atlas of genetic influences on human blood metabolites. Nat Genet, 2014; 46(6):543–550; doi: 10.1038/ng.2982

25.

Suhre

, Shin

, Petersen

, et al. Human metabolic individuality in biomedical and pharmaceutical research. Nature, 2011; 477(7362):54–60; doi: 10.1038/nature10354

26.

Teumer

, Tin

, Sorice

, et al. Genome-wide association studies identify genetic loci associated with albuminuria in diabetes. Diabetes, 2016; 65(3):803–817; doi: 10.2337/db15-1313

27.

Pattaro

, Teumer

, Gorski

, et al. Genetic associations at 53 loci highlight cell types and biological pathways relevant for kidney function. Nat Commun, 2016; 7:10023; doi: 10.1038/ncomms10023

28.

Choi

, Chen

, Stein

, et al. Assessment of bidirectional relationships between physical activity and depression among adults: A 2-sample Mendelian randomization study. JAMA Psychiatry, 2019; 76(4):399–408; doi: 10.1001/jamapsychiatry.2018.4175

29.

Yang

, Yan

, Zhao

, et al. Assessing the causal effects of human serum metabolites on 5 major psychiatric disorders. Schizophr Bull, 2020; 46(4):804–813; doi: 10.1093/schbul/sbz138

30.

Burgess

, Butterworth

, Thompson

. Mendelian randomization analysis with multiple genetic variants using summarized data. Genet Epidemiol, 2013; 37(7):658–665; doi: 10.1002/gepi.21758

31.

Gill

, Brewer

, Monori

, et al. Effects of genetically determined iron status on risk of venous thromboembolism and carotid atherosclerotic disease: A Mendelian randomization study. J Am Heart Assoc, 2019; 8(15):e012994; doi: 10.1161/jaha.119.012994

32.

Chen

, Hong

, Gao

, et al. Causal relationship between physical activity, leisure sedentary behaviors and COVID-19 risk: A Mendelian randomization study. J Transl Med, 2022; 20(1):216; doi: 10.1186/s12967-022-03407-6

33.

Bowden

, Davey Smith

, Haycock

, et al. Consistent estimation in mendelian randomization with some invalid instruments using a weighted median estimator. Genet Epidemiol, 2016; 40(4):304–314; doi: 10.1002/gepi.21965

34.

Hartwig

, Davey Smith

, Bowden

. Robust inference in summary data Mendelian randomization via the zero modal pleiotropy assumption. Int J Epidemiol, 2017; 46(6):1985–1998; doi: 10.1093/ije/dyx102

35.

Burgess

, Thompson

. Interpreting findings from Mendelian randomization using the MR-Egger method. Eur J Epidemiol, 2017; 32(5):377–389; doi: 10.1007/s10654-017-0255-x

36.

Verbanck

, Chen

, Neale

, et al. Detection of widespread horizontal pleiotropy in causal relationships inferred from Mendelian randomization between complex traits and diseases. Nat Genet, 2018; 50(5):693–698; doi: 10.1038/s41588-018-0099-7

37.

O’Connor

, Price

. Distinguishing genetic correlation from causation across 52 diseases and complex traits. Nat Genet, 2018; 50(12):1728–1734; doi: 10.1038/s41588-018-0255-0

38.

Reay

, Kiltschewskij

, Geaghan

, et al. Genetic estimates of correlation and causality between blood-based biomarkers and psychiatric disorders. Sci Adv, 2022; 8(14):eabj8969; doi: 10.1126/sciadv.abj8969

39.

Nolde

, Holtfreter

, Kocher

, et al. No bidirectional relationship between depression and periodontitis: A genetic correlation and Mendelian randomization study. Front Immunol, 2022; 13:918404; doi: 10.3389/fimmu.2022.918404

40.

Lin

, Zhou

, Xu

. Potential drug targets for multiple sclerosis identified through Mendelian randomization analysis. Brain, 2023; 146(8):3364–3372; doi: 10.1093/brain/awad070

41.

Huang

, Xiao

, Ji

, et al. Association of lipid-lowering drugs with COVID-19 outcomes from a Mendelian randomization study. Elife, 2021; 10; doi: 10.7554/eLife.73873

42.

Zhu

, Bai

, Xie

, et al. Impaired amino acid metabolism and its correlation with diabetic kidney disease progression in type 2 diabetes mellitus. Nutrients, 2022; 14(16); doi: 10.3390/nu14163345

43.

Mindikoglu

, Opekun

, Putluri

, et al. Unique metabolomic signature associated with hepatorenal dysfunction and mortality in cirrhosis. Transl Res, 2018; 195:25–47; doi: 10.1016/j.trsl.2017.12.002

44.

Feng

, Cao

, Chen

, et al. Microbiome-metabolomics reveals gut microbiota associated with glycine-conjugated metabolites and polyamine metabolism in chronic kidney disease. Cell Mol Life Sci, 2019; 76(24):4961–4978; doi: 10.1007/s00018-019-03155-9

45.

Rhee

, Clish

, Ghorbani

, et al. A combined epidemiologic and metabolomic approach improves CKD prediction. J Am Soc Nephrol, 2013; 24(8):1330–1338; doi: 10.1681/asn.2012101006

46.

Alvarenga

, Ferreira

, Kemp

, et al. The role of betaine in patients with chronic kidney disease: A narrative review. Curr Nutr Rep, 2022; 11(3):395–406; doi: 10.1007/s13668-022-00426-z

47.

Kjaergaard

, Teumer

, Witte

, et al. Obesity and kidney function: A two-sample Mendelian randomization study. Clin Chem, 2022; 68(3):461–472; doi: 10.1093/clinchem/hvab249

48.

Liu

, Miao

, Deng

, et al. Gut microbiota-derived tryptophan metabolism mediates renal fibrosis by aryl hydrocarbon receptor signaling activation. Cell Mol Life Sci, 2021; 78(3):909–922; doi: 10.1007/s00018-020-03645-1

49.

Wee

, Liu

, Ching

, et al. The kynurenine pathway in acute kidney injury and chronic kidney disease. Am J Nephrol, 2021; 52(10-11):771–787; doi: 10.1159/000519811

50.

Curran

, Kopp

. Aryl hydrocarbon receptor mechanisms affecting chronic kidney disease. Front Pharmacol, 2022; 13(:782199; doi: 10.3389/fphar.2022.782199

51.

Sekula

, Goek

, Quaye

, et al. A metabolome-wide association study of kidney function and disease in the general population. J Am Soc Nephrol, 2016; 27(4):1175–1188; doi: 10.1681/asn.2014111099

52.

Kim

, Yu

, Li

, et al. Serum metabolomic signatures of plant-based diets and incident chronic kidney disease. Am J Clin Nutr, 2022; 116(1):151–164; doi: 10.1093/ajcn/nqac054

53.

Wen

, Zheng

, Surapaneni

, et al. Metabolite profiling of CKD progression in the chronic renal insufficiency cohort study. JCI Insight, 2022; 7(20); doi: 10.1172/jci.insight.161696

54.

Sonoda

, Shoji

, Kimoto

, et al. Kidney function, cholesterol absorption and remnant lipoprotein accumulation in patients with diabetes mellitus. J Atheroscler Thromb, 2014; 21(4):346–354; doi: 10.5551/jat.20594

55.

Shoji

, Akiyama

, Fujii

, et al. Association of kidney function with serum levels of cholesterol absorption and synthesis markers: The CACHE study CKD analysis. J Atheroscler Thromb, 2022; 29(12):1835–1848; doi: 10.5551/jat.63311

56.

Mitrofanova

, Merscher

, Fornoni

. Kidney lipid dysmetabolism and lipid droplet accumulation in chronic kidney disease. Nat Rev Nephrol, 2023; 19(10):629–645; doi: 10.1038/s41581-023-00741-w

57.

Arias

, Arboleya

, Allison

, et al. The relationship between choline bioavailability from diet, intestinal microbiota composition, and its modulation of human diseases. Nutrients, 2020; 12(8); doi: 10.3390/nu12082340

58.

Fang

, Wang

, Zheng

, et al. Network pharmacology study of Yishen capsules in the treatment of diabetic nephropathy. PLoS One, 2022; 17(9):e0273498; doi: 10.1371/journal.pone.0273498

59.

Kakiya

, Shoji

, Hayashi

, et al. Decreased serum adrenal androgen dehydroepiandrosterone sulfate and mortality in hemodialysis patients. Nephrol Dial Transplant, 2012; 27(10):3915–3922; doi: 10.1093/ndt/gfs162

60.

van der Burgh

, Khan

, Neggers

, et al. The role of serum testosterone and dehydroepiandrosterone sulfate in kidney function and clinical outcomes in chronic kidney disease: A systematic review and meta-analysis. Endocr Connect, 2022; 11(6); doi: 10.1530/ec-22-0061

61.

Liu

, Liu

, Gurung

, et al. Urine tricarboxylic acid cycle metabolites predict progressive chronic kidney disease in type 2 diabetes. J Clin Endocrinol Metab, 2018; 103(12):4357–4364; doi: 10.1210/jc.2018-00947

62.

Titan

, Venturini

, Padilha

, et al. Metabolomics biomarkers and the risk of overall mortality and ESRD in CKD: Results from the Progredir cohort. PLoS One, 2019; 14(3):e0213764; doi: 10.1371/journal.pone.0213764

63.

Bernard

, Zhou

, Surapaneni

, et al. Serum metabolites and kidney outcomes: The atherosclerosis risk in Communities Study. Kidney Med, 2022; 4(9):100522; doi: 10.1016/j.xkme.2022.100522

64.

, Chen

, Du

, et al. Metabolomic markers of ultra-processed food and incident CKD. Clin J Am Soc Nephrol, 2023; 18(3):327–336; doi: 10.2215/cjn.0000000000000062

65.

Majumdar

, Kjellstrand

, Tymchak

, et al. Forced euvolemic diuresis with mannitol and furosemide for prevention of contrast-induced nephropathy in patients with CKD undergoing coronary angiography: A randomized controlled trial. Am J Kidney Dis, 2009; 54(4):602–609; doi: 10.1053/j.ajkd.2009.03.024

66.

Zhou

, Wu

, Si

, et al. The gut microbe Bacteroides fragilis ameliorates renal fibrosis in mice. Nat Commun, 2022; 13(1):6081; doi: 10.1038/s41467-022-33824-6

67.

Peng

, Liu

, Ieong

, et al. A Metabolomics study of metabolites associated with the glomerular filtration rate. BMC Nephrol, 2023; 24(1):105; doi: 10.1186/s12882-023-03147-9

68.

Jordan

, Choi

, Verbanck

, et al. No causal effects of serum urate levels on the risk of chronic kidney disease: A Mendelian randomization study. PLoS Med, 2019; 16(1):e1002725; doi: 10.1371/journal.pmed.1002725

69.

Yanai

, Adachi

, Hakoshima

, et al. Molecular biological and clinical understanding of the pathophysiology and treatments of hyperuricemia and its association with metabolic syndrome, cardiovascular diseases and chronic kidney disease. Int J Mol Sci, 2021; 22(17); doi: 10.3390/ijms22179221

70.

Mortada

. Hyperbilirubinemia, hypertension, and CKD: The links. Curr Hypertens Rep, 2017; 19(7):58; doi: 10.1007/s11906-017-0756-8

71.

, Li

, Liu

, et al. Relationship between serum bilirubin levels s and the progression of renal function in patients with chronic kidney disease and hyperuricemia. Clin Chim Acta, 2018; 486:156–161; doi: 10.1016/j.cca.2018.07.045

72.

Park

, Lee

, Kim

, et al. Serum bilirubin and kidney function: A Mendelian randomization study. Clin Kidney J, 2022; 15(9):1755–1762; doi: 10.1093/ckj/sfac120

73.

Brennan

, Kantharidis

, Cooper

, et al. Pro-resolving lipid mediators: Regulators of inflammation, metabolism and kidney function. Nat Rev Nephrol, 2021; 17(11):725–739; doi: 10.1038/s41581-021-00454-y

74.

Nakagawa

, Sanchez-Lozada

, Andres-Hernando

, et al. Endogenous fructose metabolism could explain the Warburg effect and the protection of SGLT2 inhibitors in chronic kidney disease. Front Immunol, 2021; 12:694457; doi: 10.3389/fimmu.2021.694457

75.

Baek

, He

, Afshinnia

, et al. Lipidomic approaches to dissect dysregulated lipid metabolism in kidney disease. Nat Rev Nephrol, 2022; 18(1):38–55; doi: 10.1038/s41581-021-00488-2

76.

Gill

, Zagkos

, Gill

, et al. The citrate transporter SLC13A5 as a therapeutic target for kidney disease: Evidence from Mendelian randomization to inform drug development. BMC Med, 2023; 21(1):504; doi: 10.1186/s12916-023-03227-5

77.

Ueda

, Fukui

, Kamatani

, et al. GLUT9 as a potential drug target for chronic kidney disease: Drug target validation by a Mendelian randomization study. J Hum Genet, 2023; 68(10):699–704; doi: 10.1038/s10038-023-01168-8