Multi-omics uncovers the pleiotropic genetic mechanisms linking MASLD and cardiometabolic syndromes

Abstract

Background

Metabolic dysfunction-associated steatotic liver disease (MASLD) and cardiovascular-kidney-metabolic (CKM) syndrome are interrelated conditions with shared pathophysiological features; however, the genetic architecture underlying their relationship has not been fully elucidated. Deciphering this shared genetic basis holds promise for advancing mechanistic insights and therapeutic discovery.

Methods

We performed an integrated genome-wide cross-trait analysis using GWAS summary statistics for MASLD and 38 CKM traits. Our analysis estimated genetic correlations, inferred causal relationships, and identified pleiotropic variants. Candidate causal genes and druggable targets were subsequently prioritized through integrating multi-omics data.

Results

MASLD exhibited significant genetic correlations with 16 CKM traits, especially metabolic and cardiovascular conditions. Bidirectional causal relationships were observed between MASLD and T2D, adiposity, and lipid traits. We discovered 116 pleiotropic loci, including 65 shared causal variants such as rs429358 near APOE, which exerted influence across multiple traits. Gene-based analyses prioritized 152 unique candidate pleiotropic genes, enriched in lipid and cholesterol metabolism, and highly expressed in the liver, adipose, and immune-related cell types, such as macrophages and endothelial cells. Multi-omics integration validated 131 genes using eQTL and pQTL data from multiple tissues and cohorts. Notably, FTO and APOE emerged as central pleiotropic hubs, and druggability evaluation highlighted APOE, LPL, PPARG, and GPBAR1 as established therapeutic targets for metabolic diseases.

Conclusion

This study provides a comprehensive map of the shared genetic architecture between MASLD and CKM syndrome, reveals novel causal genes and repurposable drug targets, and offers insights into precision medicine approaches for cardiometabolic and liver diseases.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12933-026-03180-6.

Research insights

What is currently known about this topic?

MASLD and CKM syndromes share pathophysiological mechanisms, yet their shared genetic architecture across the full CKM spectrum has not been systematically characterized.

What is the key research question?

What genetic mechanisms underlie the interplay between MASLD and CKM syndromes? Which pleiotropic variants, genes, and pathways mediate their bidirectional relationship?

What is new?

We identified 65 shared causal variants and 152 unique genes enriched in lipid metabolism. Multi-omics validation revealed FTO and APOE as central hubs. We prioritized four druggable targets: APOE, LPL, PPARG, and GPBAR1 for therapeutic repositioning.

How might this study influence clinical practice?

Our findings provide a genetic foundation that may inform future risk stratification approaches for MASLD patients at elevated risk of cardiometabolic complications. The genetically prioritized targets provide biological rationale for investigating integrated therapeutic strategies targeting both hepatic and cardiometabolic pathways simultaneously, potentially advancing precision medicine for multisystem disease management.

Introduction

Metabolic dysfunction-associated steatotic liver disease (MASLD), a highly prevalent metabolic disorder affecting approximately 30% of adults globally, is closely interconnected with Cardiovascular-Kidney-Metabolic (CKM) syndrome, a recently proposed framework by the American Heart Association (AHA) that encompasses a cluster of conditions, including type 2 diabetes (T2D), chronic kidney disease (CKD), and cardiovascular disease (CVD), that share common pathophysiological mechanisms [1, 2]. Epidemiological evidence reveals a strong bidirectional relationship: CKM components increase MASLD risk, while MASLD independently elevates the risk of T2D, cardiovascular events, and CKD [3–5]. However, despite the liver’s central role in metabolic homeostasis and the compelling pathophysiological and epidemiological links between MASLD and CKM syndrome, MASLD has not been formally incorporated into the CKM syndrome spectrum.

The intricate relationship of MASLD and CKM syndromes is supported by shared biological mechanisms. Key pathophysiological processes such as insulin resistance, chronic systemic inflammation, and dyslipidemia underpin both conditions [6]. Furthermore, genetic investigations have pinpointed several overlapping susceptibility loci (e.g., PNPLA3, TM6SF2, GCKR) that exert pleiotropic influences on both hepatic and cardiometabolic traits, establishing a genetic basis for their co-inheritance and reinforcing the notion that they are not merely coincidental comorbidities but arise from shared pathobiological pathways [6, 7]. Despite established heritability and growing genome-wide association studies (GWAS) evidence for both MASLD and CKM traits, no study has systematically investigated their shared genetic architecture across the full CKM spectrum [8–11].

To address this gap, we applied a multilevel analytical framework to MASLD and 38 CKM-related traits to identify shared genetic variants, causal genes, and functional pathways (Fig. 1). This approach seeks to identify high-risk subgroups and prioritize druggable targets, ultimately informing precision strategies for cardiometabolic-liver disease prevention and treatment.

Fig. 1.

Overview of study traits and analytical framework. A Schematic illustration of the traits included in this study. MASLD was analysed together with 38 CKM traits. CKM traits comprised 10 cardiovascular diseases, 4 kidney diseases or traits, 2 metabolic diseases, and 22 metabolic traits including lipids, glucose-related traits, blood pressure and adiposity indices. This graph was created in https://www.biorender.com/. B Multilevel genome-wide analytical pipeline for dissecting the shared genetic architecture betwveen MASLD and CKM traits. (1) Genetic corelation and causality: estination of genome-wide and local genetic correlations using LDSC and HDL, and bidirectional causal inference using two-sample Mendelian randomization. (2) Identification of pleiotropic variants: detection of cross-trait signals using MTAG and CPASSOC, followed by Bayesian colocalization to refine shared causal variants. (3) Mapping of pleiotropic genes: gene-based analyses integrated with transcriptome-wide analyses and functional annotation. (4) Identification of druggable targets: protein-level association analyses and drug repositioning evaluation

Methods

GWAS datasets

We obtained GWAS summary statistics for MASLD and 38 CKM traits with European ancestry. For MASLD, we incorporated two independent datasets and performed meta-analysis using the inverse variance-weighted fixed effects model in METAL to improve statistical power [12]. We applied Han and Eskin’s random effects model (RE2) for SNPs showing evidence of heterogeneity (I² ≥ 50 or P value for Cochran’s Q statistic < 0.05) implemented in METASOFT [13]. GWAS data for CKM traits including 10 cardiovascular diseases, 4 kidney diseases or traits, 2 metabolic diseases, and 22 metabolic traits comprising adiposity, glucose, lipid, and blood pressure traits, were included with sample sizes greater than 50,000. Detailed information on all contributing GWAS datasets was supplied in Supplementary Table S1.

We further performed quality control for each GWAS summary statistic by (i) matching missing rsID field based on dbSNP dataset using SumStatsRehab tool [14]; (ii) converting to GRCh37 reference genomes; (iii) excluding non-biallelic SNPs and SNPs without rsID or with duplicate rsID; (iv) keeping SNPs located within autosomes with minor allele frequency (MAF) > 0.01; and (v) removing the major histocompatibility complex (MHC) region (chr 6: 25–35 Mb).

Genome-wide genetic correlation analysis

We first estimated SNP-based heritability for each trait using single-trait linkage disequilibrium (LD) score regression (LDSC) software [15]. We then applied LDSC-SEG to quantify tissue-specific heritability enrichment across 53 GTEx v6p tissues for MASLD and CKM traits, thereby highlighting key tissues underlying their genetic architecture [16]. Multiple testing was controlled using the Benjamini–Hochberg procedure, with FDR-corrected P < 0.05 considered statistically significant.

Subsequently, to evaluate genome-wide genetic correlations between MASLD and CKM traits, we used both LDSC and high-definition likelihood (HDL) [17]. LDSC estimates genetic covariance from the relationship between LD scores and GWAS association statistics using pre-computed LD scores for well-imputed HapMap3 variants. HDL extends this framework by modelling the full LD structure across ~ 1 million HapMap3 SNPs to increase accuracy and reduce variance of correlation estimates. For both methods, we applied a Bonferroni-corrected threshold of P < 1.32 × 10⁻³ (0.05/38) to define statistically significant genetic correlations.

At the regional level, we employed the local version of HDL (HDL-L) to estimate local genetic correlations between MASLD and CKM trait pairs across 2,468 approximately LD-independent genomic regions [18]. Local correlations with FDR-adjusted P < 0.05 were considered significant.

Mendelian randomization analysis

To further investigate the causal relationship between MASLD and CKM traits, we performed bidirectional two-sample Mendelian randomization (MR) analysis. Instrumental variables (IVs) were selected as SNPs associated with each trait at genome-wide significance (P < 5 × 10⁻⁸) in the corresponding GWAS. To ensure conditional independence, we applied stringent LD clumping (r² < 0.001 within 10 Mb) using the European reference panel from the 1000 Genomes Project. The strength of each IV was assessed by calculating the F-statistic using the GWAS summary statistics of the exposure traits, and SNPs with F ≤ 10 were excluded to minimize weak-instrument bias.

We employed the inverse-variance weighted (IVW) method as the primary MR estimator [19]. To assess robustness to horizontal pleiotropy and outlier instruments, we additionally applied MR-Egger regression [20], weighted median [21], weighted mode [22], and MR-Pleiotropy Residual Sum and Outlier (MR-PRESSO) [23]. We further used radial MR to identify and exclude heterogeneity-driven outlier IVs [24]. Multiple testing corrections were performed separately for each method using the FDR procedure, and causal estimates were considered statistically significant at an FDR-adjusted P < 0.05 for the IVW method.

Sensitivity analyses included Cochran’s Q statistic to evaluate heterogeneity among IVs and the MR-Egger intercept statistic to detect directional pleiotropy. We also conducted leave-one-out analyses by sequentially removing each SNP and reapplying the IVW analysis to the remaining instruments to assess the influence of individual SNPs on the causal estimates.

Cross-trait meta-analysis

The presence of genetic correlation indicates that genetic variants may exert independent effects on both traits, known as pleiotropy, or influence one trait indirectly through their impact on the other, implying a causal relationship. Thus, to identify pleiotropic variants shared between MASLD and these CKM traits that showed genetic correlation with MASLD, we conducted cross-trait meta-analysis of GWAS summary statistics for MASLD and each CKM trait using multi-trait analysis of GWAS (MTAG) [25] and cross-phenotype association test (CPASSOC) [26] methods.

The MTAG estimator generalizes inverse-variance-weighted meta-analysis to jointly analyse multiple traits while accounting for potential sample overlap. We estimated the maximum false discovery rate (maxFDR) for each trait to evaluate potential inflation due to violation of MTAG assumptions. As a complementary approach, CPASSOC was used for sensitivity analysis, and we adopted the SHet statistic to allow for heterogeneity in SNP effects across traits.

Subsequently, we applied PLINK 1.9 software to GWAS data derived from the cross-trait meta-analysis, with the following parameters for clumping independent loci: –clump-p1 5 × 10⁻⁸ –clump-p2 1 × 10⁻⁵ –clump-r² 0.01 –clump-kb 500. LD pattern was estimated using the 1000 Genomes Project Phase 3 European reference panel. The variant with the lowest P value within each locus was designated as the index SNP. Significant pleiotropic SNPs were required to satisfy P_MTAG & P_CPASSOC < 5 × 10⁻⁸, and Psingle-trait < 1 × 10⁻³ for both traits. Among these, SNPs that were not genome-wide significant in the original single-trait GWAS and for which no neighboring SNPs within 1.0 Mb region reached P < 5 × 10⁻⁸ in the single-trait GWAS were considered novel pleiotropic variants. Detailed functional annotation for pleiotropic SNPs was conducted using ANNOVAR [27] and SNPnexus tools [28].

Bayesian colocalization analysis

We conducted colocalization analysis using the R package Coloc to determine whether the identical pleiotropic variants underlie GWAS signals for MASLD and CKM traits [29]. For each pleiotropic locus, we selected SNPs within ± 500 kb of the index variant and computed posterior probabilities for five mutually exclusive hypotheses: no association (H0), association with only one trait (H1, H2), association with both traits but with distinct causal variants (H3), and association with both traits attributable to a single shared causal variant (H4). A posterior probability PP.H4 ≥ 0.5 was considered indicative of a shared causal variant.

For pleiotropic variants shared across multiple traits, we further performed multi-trait colocalization using the R package HyPrColoc, which identifies clusters of traits sharing a common causal variant and provides posterior probabilities for each cluster [30].

Gene-based association and enrichment analyses

To identify pleiotropic genes underlying MASLD and CKM traits, we performed gene-based analyses on MTAG summary statistics using Multi-Marker analysis of Genomic Annotation (MAGMA) [31] and the multivariate set-based association test (mBAT-combo) [32]. MAGMA aggregates SNP effects within ± 5 kb of gene boundaries using multiple linear principal component regression, while mBAT-combo is designed to enhance power in the presence of LD-induced masking. LD patterns were derived from the 1000 Genomes Phase 3 European reference panel. Genes with Bonferroni-corrected P < 0.05 in both MAGMA and mBAT-combo were defined as pleiotropic genes.

Subsequently, we performed enrichment analysis of these pleiotropic genes for each pair of traits to elucidate the potential shared biological pathways underlying MASLD and CKM syndromes. Gene set enrichment analysis (GSEA) was conducted utilizing the clusterProfiler package with GO and KEGG reference gene sets [33]. Benjamini–Hochberg was applied to control for multiple hypothesis testing, with FDR < 0.05 defining significant pathway enrichment.

Tissue-specific enrichment analysis (TSEA) was performed using deTS with GTEx RNA-seq data from 47 tissues [34], and cell-type-specific enrichment analysis (CSEA) was conducted using WebCSEA with expression signatures from 1,355 tissue-cell-types (TCs) encompassing 61 general tissue types distributed among 11 human organ systems [35]. For both TSEA and CSEA, a Bonferroni-adjusted P < 0.05 was considered statistically significant.

Candidate pleiotropic genes at expression and protein level

Pleiotropic genes identified by MAGMA and mBAT-combo and located within ± 500 kb of causal pleiotropic variants were designated as candidate pleiotropic genes. To further evaluate the effects of genetically regulated gene expression on MASLD-CKM trait pairs, we conducted transcriptome-wide association studies (TWAS) using the FUSION software [36], combining MTAG summary statistics with pre-computed cis-eQTL weights from 49 GTEx v8 single tissues and an sCCA-based cross-tissue feature [37]. LD was modelled using the European reference panel from Phase 3 of the 1000 Genomes Project. Finally, we applied the aggregate Cauchy association test (ACAT) to combine results across tissues, and Bonferroni-corrected P < 0.05 was used to define significant associations.

We additionally performed Summary-data-based Mendelian Randomization (SMR) [38] to integrate GWAS and cis-eQTL data from GTEx v8 tissues and eQTLGen whole blood [39] with LD accounted for using the European reference panel from 1000 Genomes Phase 3. SMR identifies pleiotropic or potentially causal associations between gene expression and traits and uses the HEIDI-outlier test to differentiate pleiotropy from linkage. Associations with Bonferroni-corrected P < 0.05 and P value of HEIDI test ≥ 0.05 were retained.

For candidate pleiotropic genes showing risk effects at the expression level, we examined protein-level evidence using protein abundance quantitative trait loci (pQTL) datasets. We performed proteome-wide association study (PWAS) using FUSION with cis-pQTL weights for 1,348 proteins measured in 7,213 European American participants from the ARIC study [40], and SMR analyses using cis-pQTL summary data from large deCODE [41], Fenland [42] and UKB-PPP cohorts [43]. Multiple testing was controlled using FDR, and protein-trait associations with FDR-adjusted P < 0.05 were considered significant.

To facilitate biological interpretation and therapeutic prioritization, we classified candidate pleiotropic genes into three evidence-based tiers reflecting progressive validation from genomic to functional omics layers. Tier 1 genes, representing genomic evidence, were defined as those located within ± 500 kb of the shared causal variants and showing significant association in both gene-based analyses. These genes represent candidates most likely to be functionally regulated by the identified pleiotropic variants. Tier 2 genes, comprising transcriptomically validated candidates, were defined as a subset of Tier 1 genes with concordant evidence from TWAS and SMR analyses. These genes demonstrate that genetic variants influence MASLD-CKM traits through regulation of gene expression. Tier 3 genes, representing the highest-confidence candidates with proteomic validation, were defined as further validated at the protein abundance level through PWAS or SMR-pQTL analyses. These represent the highest-confidence genes where genetic regulation manifests across genomic, transcriptomic, and proteomic layers, most proximal to biological function and therapeutic intervention.

Druggability evaluation

We employed the Genome for REPositioning drugs (GREP) software to assess the enrichment of the candidate pleiotropic genes within ICD10-classified drugs categories, aiming to identify medications that could potentially be repositioned to target this gene set [44]. Pleiotropic genes associated with MASLD-CKM trait pairs at expression level were selected for target genes enrichment analysis of approved or investigated drugs curated in DrugBank and the Therapeutic Target Database.

Results

Genetic correlation

To systemically leverage shared genetics between MASLD and other related metabolic conditions, we evaluated genome-wide genetic correlations between MASLD and 38 CKM syndromes and identified significant correlations with 16 syndromes, including positive correlations with six cardiovascular diseases, eight metabolic traits and two metabolic diseases. Notably, we did not detect significant genetic correlations between MASLD and kidney-related traits. This finding contrasts with the strong genetic correlations observed for metabolic and cardiovascular traits and warrants mechanistic interpretation. Among the eight metabolic traits correlated with MASLD, we found that ISI and HDLC had a negative genetic correlation with it (Fig. 2A, Supplementary Table S2).

Fig. 2.

Genome-wide genetic correlations and bidirectional causal inference between MASLD and CKM syndromes. A Forest plot illustrating significant genome-wide genetic correlations between MASLD and 16 CKM traits. Correlations were estimated using two complementary methods, including LDSC (brown squares) and HDL (grey squares). The traits were categorized into three groups indicated by the colored vertical bars on the left: metabolic traits (MT, red), metabolic diseases (MD, blue), and cardiovascular diseases (CVD, green). The dots represented rg value and error bars represent 95% CIs. Statistical significance was defined using a Bonferroni-corrected threshold of P < 1.32 × 10⁻³ (0.05/38). B Evaluation of causal relationships using bidirectional two-sample Mendelian Randomization. The plots displayed OR value with 95% CIs for significant findings. The analyses were conducted using four methods: Inverse-Variance Weighted (IVW, red), MR Egger (blue), Weighted median (green), and Weighted mode (brown). MR estimates were derived from the IVW method

We next calculated tissue-specific heritability for MASLD and these 16 syndromes in 53 GTEx tissues to gain insights into the biological mechanisms and key tissues underlying the genetic architecture of MASLD and CKM syndromes. We found that MASLD heritability was significantly enriched in liver tissue. We also observed comparable hepatic enrichment for HDLC, TG, and T2D, pointing to a shared pathological mechanism. Furthermore, we found heritability enrichment in adipose tissue for HDLC, TG, FI, WHR, and WHRadjBMI (Supplementary Table S3, Supplementary Fig. S1).

To dissect specific genomic regions with pleiotropic effects contributing to the shared genetic basis of MASLD and CKM syndromes, local genetic correlation analysis revealed associations between MASLD and five CKM syndromes across 11 genomic regions. Most of these regions were genetically associated with TG. Three regions were shared among multiple MASLD-CKM trait pairs. For example, region 2p23.2-23.3 was implicated in pairs involving MASLD-TG and MASLD-FI; region 8q24.13 was involved in MASLD-TG and MASLD-CAD; and region 16q12.2 was correlated with MASLD-TG, MASLD-T2D, and MASLD-ISI (Supplementary Table S4).

Collectively, our genetic correlation analyses at global, local, and tissue-specific genomic levels demonstrated substantial genetic overlap between MASLD and CKM syndromes, particularly for lipid-related traits. The convergent hepatic and adipose tissue enrichment patterns highlight common metabolic pathways underlying these comorbid conditions. These analyses prioritize lipid regulatory as dominant shared genetic drivers of MASLD-CKM comorbidity and establish the foundation for identifying specific pleiotropic variants and genes underlying this genetic architecture.

Genetic causal association

While genetic correlation analysis revealed shared genetic architecture, it did not establish causality or directionality of these relationships. We therefore conducted bidirectional two-sample MR using several methods and found a bidirectional association between MASLD and several CKM syndromes. MASLD was positively associated with four CKM syndromes, including DBP, T2D, WHR, and WHRadjBMI. Furthermore, a negative association between MASLD and ISIadjBMI was observed regardless of forward or reverse analyses. Reversely, six CKM syndromes including BMI, CAD, T2D, TG, WC, WHR were referred to as causal to MASLD (Fig. 2B, Supplementary Table S5). These bidirectional MR analyses revealed reciprocal causal relationships between MASLD and key cardiometabolic traits, substantiating MASLD as an integral component of the CKM spectrum. The concordance between genetic correlation and causal inference establishes that shared genetic architecture translates into directional biological effects, motivating fine-mapping of specific pleiotropic variants driving these relationships.

Identification of pleiotropic variants

Cross-trait meta-analysis of MASLD and 16 CKM syndromes identified 116 significant independent pleiotropic variants shared between MASLD and 13 of these syndromes. (Supplementary Fig. S2). The number of shared variants ranged from 1 (between MASLD and PVD) to 23 (between MASLD and WHRadjBMI) (Supplementary Fig. S3A, Supplementary Table S6). Notably, 17 of these variants had not been previously reported in prior GWASs of MASLD or CKM syndromes (Supplementary Table S7).

Bayesian colocalization analysis was then conducted to ascertain whether these pleiotropic associations represent the same causal variant affecting both traits or distinct linked variants, thereby confirming the true shared genetic etiology. Colocalization analysis of the 116 pleiotropic variants identified 65 shared causal variants across 12 MASLD-CKM trait pairs, corresponding to 46 unique variants (Supplementary Fig. S3B-D, Supplementary Table S8). Thirteen variants were present in at least two pairs, with seven notably residing in the intron of the obesity-linked FTO gene. MASLD and TG shared the most causal variants (Fig. 3). The rs429358 variant near APOE was the most prevalent, shared by MASLD and four CKM syndromes (HDLC, T2D, WHR, WHRadjBMI) (Fig. 4A). Consistently, multi-trait colocalization highlighted four highly pleiotropic loci, rs429358, rs4709746, rs4929923, and rs15285, which jointly colocalized across MASLD and multiple cardiometabolic traits, indicating a small set of shared causal variants that coordinately influence adiposity, lipid, and cardiovascular phenotypes (Supplementary Fig. S4, Supplementary Table S9).

Fig. 3.

Landscape of shared causal variants and candidate pleiotropic genes between MASLD and CKM syndromes. The circular dendrogram illustrated the hierarchical relationship between MASLD, CKM traits, causal variants, and prioritized genes. Center and first layer: The central node represented MASLD, connecting to 12 CKM trait identified with shared causal variants. Second layer: The branches extend to 65 shared causal variants identified via colocalization (PP.H4 ≥ 0.5). Solid grey circles indicated variants shared by at least two MASLD-CKM trait pairs (n = 13). Third layer: The branches indicated the genomic regions corresponding to each causal locus. Outer layer: The outermost nodes displayed candidate pleiotropic genes located within ± 500 kb window of the causal variants. These genes were significantly associated with both MASLD-CKM trait pair in gene-based analyses. A maximum of the five nearest significant genes were displayed per locus. Solid grey gene nodes denote genes where the causal variant is located directly within the gene body

Fig. 4.

Characterization of key pleiotropic loci, genes, and biological pathways underlying MASLD-CKM comorbidity. A Regional association plots for the rs429358 locus (mapping APOE), a prominent pleiotropic variant shared across four MASLD-CKM trait pairs: MASLD-WHR, MASLD-WHRadjBMI, MASLD-T2D, and MASLD-HDLC. The x-axis represented the chromosomal position on chromosome 19, and the y-axis showed the association strength (-log₁₀P). B Heatmap displaying 46 candidate pleiotropic genes that were associated with at least two MASLD-CKM trait pairs. These genes were located within ± 500 kb of the identified causal variants and were prioritized by gene-based analyses. The color intensity represented the strength of the gene-trait association (Z-value). The bar plot on the right summarized the total number of associated trait pairs for each gene. C Pathway enrichment analysis of the pleiotropic genes. The heatmap illustrated the top 20 significant GO Biological Processes (top panel) and KEGG Pathways (bottom panel). The color intensity corresponded to the significance level (-log₁₀P). The analysis revealed a convergent enrichment in lipid metabolism and cholesterol homeostasis, providing mechanistic insights into the shared etiology

Through cross-trait meta-analysis and colocalization, we refined the genetic landscape from broad genomic correlations to 46 unique shared causal variants with high-confidence pleiotropic effects. These findings pinpoint the precise genetic variants underlying MASLD-CKM comorbidity and guide subsequent gene-level analyses.

Identification of pleiotropic genes

To translate variant-level associations into functional insights, we performed gene-based analyses to aggregate SNP effects and identify pleiotropic genes underlying MASLD and CKM syndromes. Gene-based analyses consistently identified 4,062 significant genes shared between MASLD and 16 CKM syndromes after Bonferroni correction (Supplementary Table S10, 11). Mapping these genes to regions within ± 500 kb of the 65 causal variants from 12 trait pairs yielded 241 candidate pleiotropic genes, representing 152 unique Tier 1 genes with genomic evidence for pleiotropy (Supplementary Table S12, Supplementary Fig. S5). Notably, 46 genes were shared between at least two trait pairs, with the FTO gene linked to seven MASLD-CKM traits, including BMI, HF, HTN, T2D, TG, WC, and WHR (Fig. 4B). eQTL analysis further showed that these causal variants could influence the expression of the mapped genes in multiple GTEx tissues (Supplementary Table S13).

We subsequently performed GSEA using genes that were significant in both MAGMA and mBAT-combo methods for each MASLD-CKM pair. Our analysis revealed that these risk genes were primarily involved in the biological processes of lipoprotein regulation and cholesterol metabolism (Fig. 4C, Supplementary Fig. S6, Supplementary Table S14, 15). Tissue-specific enrichment analysis demonstrated that genes from MASLD-CAD, FI, HDLC, MI, T2D, TG, WHR, and WHRadjBMI syndromes were significantly enriched in both adipose and liver tissues (Supplementary Fig. S7, Supplementary Table S16). Furthermore, cell-type-specific enrichment analysis indicated predominant enrichment analysis in reparative cell types (e.g., endothelial cell and fibroblast) and immune cells (e.g., macrophage, monocyte, T cell). Notably, hepatocytes also exhibited significant enrichment (Supplementary Fig. S8, Supplementary Table S17).

TWAS analysis confirmed 128 genes, with the FTO gene showing association with the most MASLD-CKM trait pairs, and 36 genes were validated by SMR method. Overall, 131 Tier 2 genes were validated at the transcriptional level by integrating the two methods, of which 38 were significantly associated with at least two NAFLD-CKM syndrome pairs (Fig. 5A, Supplementary Table S18-20).

Fig. 5.

Multi-omics validation and therapeutic prioritization of pleiotropic genes. A Tissue-specific transcriptional regulation of 38 candidate pleiotropic genes that were associated with at least two MASLD-CKM trait pairs. The plot integrated results from TWAS and SMR across 49 GTEx v8 tissues. Each pie chart within the grid represented a significant gene-tissue association, with the colored slices corresponding to the specific CKM traits. Genes in bold denoted high-confidence targets validated by both TWAS and SMR methods. B Validation of pleiotropic effects at the plasma protein level. The heatmap displayed associations identified by PWAS (top panel) and SMR using pQTL (bottom panel). Color intensity reflected the direction and strength of the association (Z-value). Asterisks marked statistical significance: *FDR < 0.05, **FDR < 0.01. C Prioritization of druggable targets. The sankey diagram (left) maped MASLD-CKM traits to key pleiotropic genes. The dot plot (right) illustrated the enrichment of these genes in drug-target databases, linking them to specific drugs and drug classes (ATC/ICD codes). Dot size corresponded to the OR, and color intensity represents the enrichment significance (-log₁₀P)

In summary, convergent evidence from gene-based associations and transcriptomic analyses identified 131 pleiotropic genes underlying MASLD and CKM syndromes. These genes predominantly regulate lipid metabolism and cholesterol pathways and exhibit enriched expression in liver and adipose tissues, as well as in hepatocytes and immune cells, which offers physiological understanding of the shared pathogenesis of MASLD and CKM syndrome.

Cardiovascular disease-focused genetic architecture

To provide integrated insights into cardiovascular complications, a leading cause of mortality in MASLD patients, we synthesized CVD-related findings across all analytical layers. MASLD showed significant genetic correlations with six CVD traits (CAD, MI, HF, AIS, PVD, AF). Notably, pleiotropic variants were predominantly shared with metabolic factors (87.7% of variants, 57/65) rather than CVD outcomes (12.3% of variants). Subsequently, gene-level analyses identified LPL specifically pleiotropic for MASLD-CVD, enriched in lipid metabolism pathways. This pattern indicates that MASLD-CVD genetic relationships operate predominantly through metabolic intermediates rather than direct hepato-cardiac mechanisms, supporting a model wherein systemic metabolic dysregulation drives cardiovascular risk in MASLD.

Protein level association and druggability assessment

Building on transcriptomic findings, we extended the validation to plasma protein levels, where protein abundance more directly reflects biological function and therapeutic potential. Using pQTL data from the ARIC study, the PWAS identified four genes (APOE, APOC1, PAPPA, and TIMP4) associated with the risk of five MASLD-CKM trait pairs at the plasma protein level. Complementary SMR analysis with pQTL data from multiple large-scale cohorts further revealed 20 genes across nine MASLD-CKM syndrome pairs. Integrating these analyses, we identified 20 unique Tier 3 genes validated at the proteomic level (Fig. 5B, Supplementary Table S21, 22). Building on the multi-omics validation, druggability assessment genetically prioritized four genes—APOE, LPL, PPARG, and GPBAR1—that are currently targeted by drugs approved or under investigation for metabolic diseases (Fig. 5C, Supplementary Table S23). Importantly, these findings represent genetic prioritization based on convergent multi-omics evidence rather than immediate therapeutic feasibility. Nevertheless, they provide biological rationale for drug repositioning, genetic risk stratification, and clinical trial enrichment strategies targeting patients most likely to benefit from pathway-specific interventions.

Discussion

In this comprehensive genome-wide cross-trait analysis, we systematically delineated the shared genetic architecture between MASLD and a broad spectrum of CKM syndromes. By integrating multiple analytical strategies, including genetic correlation, mendelian randomization, pleiotropy mapping, and multi-omics validation, our study provides a multi-layered atlas of the genetic crosstalk underpinning these conditions. We successfully identified significant genetic overlap, inferred robust causal relationships, and pinpointed key pleiotropic loci and genes, such as APOE and FTO, which act as central hubs in this complex disease network. These findings not only deepen our understanding of the common etiological pathways but also highlight promising therapeutic targets for the holistic management of cardiometabolic-liver disorders.

Our initial analysis revealed significant global genetic correlations between MASLD and 16 CKM traits, providing a genetic basis for the well-documented epidemiological co-occurrence. The observed positive genetic correlations with adiposity traits, T2D, and various cardiovascular diseases corroborate the central role of metabolic dysfunction in driving both MASLD and CKM pathologies [45]. Notably, the negative genetic correlations with HDLC and ISI are mechanistically informative. These findings genetically substantiate that the predisposition to lower HDLC levels and impaired insulin sensitivity are fundamental drivers of MASLD risk, aligning perfectly with the established role of dyslipidemia and insulin resistance as core pathophysiological mechanisms in MASLD development [46, 47]. Furthermore, the significant enrichment of heritability for MASLD and related metabolic traits in the liver and adipose tissue underscores that these two organs are the primary biological arenas where this shared genetic susceptibility manifests [48]. Notably, we did not detect genetic correlations between MASLD and kidney traits, despite their established clinical associations. This null finding likely reflects the fact that CKD represents a late-stage, etiologically heterogeneous outcome more strongly influenced by cumulative environmental exposures and disease progression than by shared genetic susceptibility. More importantly, our findings suggest that the MASLD-CKD relationship operates primarily through genetically influenced metabolic intermediates, including T2D, hypertension, and dyslipidemia, rather than a direct shared genetic architecture. This mediated pathway aligns with emerging concepts of metabolic dysfunction-associated kidney disease and reinforces the rationale for integrated management of cardiometabolic risk factors to prevent kidney complications in patients with MASLD, consistent with the broader CKM syndrome framework [49].

At the heart of this shared genetic landscape, our study identified 116 pleiotropic loci, with colocalization analysis confirming 46 unique shared causal variants. This fine-mapping effort moved from broad genomic regions to specific variants, providing tangible targets for functional investigation. Importantly, 17 of these pleiotropic variants represent novel loci that had not been previously reported in prior GWASs of either MASLD or CKM syndromes individually. These novel discoveries implicate genes with plausible biological roles spanning diverse pathways. Notably, rs731146 (PAPPA) emerged as a pleiotropic variant shared between MASLD and TG. PAPPA encodes pregnancy-associated plasma protein A, a metalloproteinase that regulates insulin-like growth factor (IGF) bioavailability [50]. While PAPPA has been extensively linked to atherosclerotic plaque instability and acute coronary syndromes [51], our identification of its association with MASLD and dyslipidemia extends these findings to hepatic steatosis, implicating IGF signaling as a novel pathway coordinately influencing lipid metabolism across hepatic, adipose, and vascular tissues. rs1515776, an intronic variant in phosphodiesterase 3 A (PDE3A), also showed pleiotropy between MASLD and TG. PDE3A encodes phosphodiesterase 3 A, a cyclic AMP (cAMP)-degrading enzyme that regulates cardiac contractility and metabolic signaling [52], highlighting cAMP-mediated metabolic regulation as a pathway linking hepatic lipid accumulation to systemic metabolic dysfunction. Collectively, these novel variants extend beyond the dominant lipid and cholesterol metabolism pathways to implicate additional mechanisms. The discovery of these previously unrecognized pleiotropic variants underscores the enhanced statistical power conferred by cross-trait analysis and highlights that the genetic architecture linking MASLD and CKM syndrome encompasses diverse biological processes beyond canonical pathways.

Among the previously established loci, the identification of rs429358, an exonic variant near APOE, as a pleiotropic variant influencing MASLD and multiple CKM traits, including WHR, HDLC and T2D, is a keystone finding. This variant is a missense mutation that defines the APOE ε4 allele, which has been extensively implicated in lipid metabolism and cardiovascular risk [53, 54]. Given APOE’s canonical role in cholesterol homeostasis, this finding highlights lipid dysregulation as a critical nexus in cardiometabolic-liver crosstalk and provides genetic evidence that APOE-mediated dyslipidemia may simultaneously predispose individuals to both hepatic steatosis and adverse cardiovascular-metabolic outcomes [45]. Similarly, the striking concentration of pleiotropic signals within the FTO gene, a well-established locus for obesity, genetically cements the link between adiposity pathways and the concurrent risk of MASLD and related comorbidities [55, 56]. Collectively, these findings illustrate how specific genetic variants orchestrate a cascade of pathological events across multiple organ systems.

Beyond these shared loci, our gene prioritization identified 152 unique pleiotropic genes, several of which demonstrated particularly strong evidence of shared effects. BPTF, a chromatin remodeling factor that scores highest among BMI-associated loci [57], exhibits the broadest pleiotropic effects across MASLD and cardiometabolic traits, although its specific mechanisms in MASLD pathogenesis remain to be elucidated. APOC1, which is in strong linkage disequilibrium with APOE, modulates lipoprotein metabolism by inhibiting lipoprotein lipase activity and regulating cholesteryl ester transfer protein, thereby influencing HDL metabolism in ways that may contribute to MASLD and cardiovascular risk [58, 59]. Notably, LPL demonstrated robust pleiotropic associations across multiple MASLD-CKM trait pairs. Beyond its classical role in triglyceride hydrolysis, LPL likely exerts pleiotropic effects through multiple pathways, facilitating lipoprotein remnant clearance and modulating endothelial lipase activity, which influences both atherogenic particle composition and vascular inflammation [60]. This multi-pathway mechanism positioning LPL at the intersection of lipid metabolism and cardiovascular pathology distinguishes it from single-mechanism interventions and identifies it as a promising therapeutic target for integrated management of MASLD patients with elevated cardiometabolic risk [61–63].

To translate these genetic associations into biological function, our pathway analyses consistently implicated lipid and cholesterol metabolism as the core biological processes enriched among these pleiotropic genes. This was further substantiated by tissue- and cell-type-specific analyses, which pointed not only to hepatocytes but also to adipose tissue, endothelial cells, and macrophages. The enrichment in endothelial cells and macrophages suggests that shared genetic factors may operate through pathways of vascular inflammation and immune dysregulation, providing a mechanistic link between metabolic stress and the heightened cardiovascular risk observed in MASLD patients [64–66]. These results broaden the pathogenic framework beyond simple hepatic steatosis to include systemic inflammation and endothelial dysfunction as key shared mechanisms [6, 67].

The druggability assessment identified three key therapeutic targets. Notably, PPARG demonstrated exceptional enrichment across three major disease categories: lipid-modifying agents, metabolic disorders, and renin-angiotensin system agents, highlighting its multisystem efficacy. LPL similarly showed significant enrichment in both lipid-modifying and metabolic disorder treatment categories. This convergence of therapeutic targets provides compelling genetic evidence for integrated pharmacological strategies targeting MASLD and its cardiometabolic comorbidities simultaneously [68]. These findings have profound implications for clinical trial redesign: future studies should adopt multi-system endpoints assessing hepatic and cardiometabolic outcomes simultaneously in patients stratified by shared pleiotropic variants [9]. Moreover, the robust enrichment of APOE, PPARG, and LPL across metabolic disease categories establishes MASLD as an integral component of the CKM syndrome spectrum, justifying coordinated drug development efforts that address multiple disorders through shared biological pathways [8, 69]. Importantly, these findings represent genetic prioritization of therapeutic candidates rather than immediate clinical feasibility, requiring rigorous translational validation [70, 71]. Beyond target identification, our genetic architecture findings may enable future precision medicine applications through patient stratification strategies, although prospective validation is required to establish clinical utility. Future development of polygenic risk scores incorporating the identified pleiotropic variants may potentially identify MASLD patients at highest risk for cardiometabolic complications, facilitating targeted surveillance and early intervention, though the predictive performance requires validation in prospective cohorts [72]. For instance, MASLD patients carrying APOE ε4 alleles or high-risk variants in LPL pathways might warrant intensified cardiovascular screening and preventive lipid-lowering therapy independent of conventional risk factor profiles. Furthermore, genetic stratification may potentially enhance clinical trial efficiency by enriching for participants most likely to respond to specific interventions [73]. However, we emphasize that these represent hypothesis-generating applications requiring prospective validation studies to establish predictive power and clinical utility before implementation in practice.

The main strengths of this study included a comprehensive multipronged genetic approach integrating global and local genetic correlations and pleiotropy mapping, as well as systematic multi-trait analysis across 38 CKM phenotypes with rigorous analytical criteria and multi-omics validation. We acknowledge several limitations in our study. First, our analyses were primarily based on GWAS summary statistics from individuals of European ancestry, which may limit the generalizability of our findings to other populations. Future studies must increase representation of other genetic ancestries to create a more complete global picture of MASLD-CKM genetics. Second, the diagnosis of MASLD in the source GWAS was mostly based on imaging or diagnostic codes rather than liver biopsy, the gold standard, which may introduce some level of phenotype heterogeneity. This phenotype heterogeneity capturing a mixture of subtypes with distinct genetic architectures may reduce the genetic signal in our instruments, leading to underestimation of the true causal effect of clinically significant MASLD on cardiometabolic traits. Finally, while our findings identify pleiotropic variants and genes with robust genetic evidence, the clinical utility of polygenic risk scores incorporating these variants for patient stratification and trial enrichment has not been validated. Prospective studies in independent cohorts are required to establish whether the identified genetic architecture provides sufficient predictive power for clinical decision-making and trial design.

Conclusions

This study provides a comprehensive genetic map of the complex interactions between MASLD and CKM syndrome, progressing from broad correlations to specific causal variants and genes. We identified key pleiotropic hubs and druggable targets, revealing shared mechanisms associated with lipid metabolism. These findings establish a genetic foundation for developing integrated preventive and therapeutic strategies that simultaneously target the liver, metabolic, and cardiovascular systems. While these findings provide genetic prioritization of therapeutic candidates, translational validation through biomarker-guided trials and patient stratification strategies is essential for clinical implementation.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

2hGlu

Two-hour glucose

AFib

Atrial fibrillation

AIS

Any ischemic stroke

BMI

Body mass index

BUN

Blood urea nitrogen

CAD

Coronary artery disease

CES

Cardioembolic stroke

CKD

Chronic kidney disease

CKM

Cardiovascular-kidney-metabolic

CPASSOC

Cross-phenotype association

CSEA

Cell-type specific enrichment analysis

CVD

Cardiovascular disease

DBP

Diastolic blood pressure

eGFR

Estimated glomerular filtration rate

eQTL

Expression quantitative trait loci

FG

Fasting glucose

FI

Fasting insulin

FUSION

Functional summary-based imputation

GSEA

Gene set enrichment analysis

GWAS

Genome-wide association studies

HbA1c

Glycated hemoglobin

HDL

High-definition likelihood

HDLC

High density lipoprotein cholesterol

HF

Heart failure

HTN

Hypertension

IFC

Insulin fold change during an oral glucose tolerance test

IFCadjBMI

Insulin fold change during an oral glucose tolerance test (adjusted for BMI)

ISI

Modified Stumvoll Insulin Sensitivity Index

ISIadjBMI

Modified Stumvoll Insulin Sensitivity Index (adjusted for BMI)

LAS

Large artery stroke

LDLC

Low density lipoprotein cholesterol

MI

Myocardial infarction

LDSC

Linkage disequilibrium score regression

MAF

Minor allele frequency

MAGMA

Multi-marker analysis of genomic annotation

mBAT-combo

Multivariate set-based association test

MR

Mendelian randomization

MTAG

Multi-trait analysis of GWAS

MASLD

Metabolic dysfunction-associated steatotic liver disease

nonHDLC

Non-high density lipoprotein cholesterol

PP

Pulse pressure

pQTL

Protein abundance quantitative trait loci

PVD

Peripheral vascular disease

PWAS

Proteome-wide association study

RG

Random glucose

SBP

Systolic blood pressure

SMR

Summary-data-based mendelian randomization

SVS

Small vessel stroke

T2D

Type 2 diabetes

TC

Total cholesterol

TG

Triglycerides

TSEA

Tissue-specific enrichment analysis

TWAS

Transcriptome-wide association study

UACR

Urinary albumin-to-creatinine ratio

WC

Waist circumference

WCadjBMI

Waist circumference adjusted for BMI

WHR

Waist-hip ratio

WHRadjBMI

Waist-to-hip ratio adjusted for BMI

IV

Instrumental variable

IVW

Inverse-variance weighted

Author contributions

Mohammed Eslam and Jing Ni contributed to the conceptualization and study design. Kangjia Yin and Cao Zhang conducted the statistical analysis. Kangjia Yin, Cao Zhang, and Jing Zeng drafted the manuscript. Bing Liu and Ruyun Xu provided manuscript proofreading and editing. Mohammed Eslam, Jing Ni and Jing Zeng critically reviewed the manuscript. All authors have read and approved the final version of the manuscript.

Funding

This work was supported by the Research Funds of Center for Big Data and Population Health of IHM (JKS2023008), Research Fund of Anhui Institute of Translational Medicine (2023zhyx-C07), National Natural Science Foundation of China (82100605).

Data availability

All summary statistics analysed in this study are publicly available as shown in Supplementary Table 1.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.