Skip to main content
Advertisement
Browse Subject Areas
?

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

  • Loading metrics

Metabolic-dysfunction associated steatotic liver disease-related diseases, cognition and dementia: A two-sample mendelian randomization study

  • Yao-Shuang Li,

    Roles Conceptualization, Data curation, Formal analysis, Investigation, Project administration, Visualization, Writing – original draft, Writing – review & editing

    Affiliations Tianjin Union Medical Center, Tianjin Medical University, Tianjin, China, Department of Endocrinology, Tianjin Union Medical Center, Nankai University Affiliated Hospital, Tianjin, China

  • Yu-Ge Xia,

    Roles Data curation, Formal analysis

    Affiliation Geriatric Department, The Second Affiliated Hospital of Anhui University of Chinese Medicine, Hefei, Anhui, China

  • Yan-Lan Liu,

    Roles Data curation, Formal analysis

    Affiliation Department of Endocrinology, Tianjin Union Medical Center, Nankai University Affiliated Hospital, Tianjin, China

  • Wei-Ran Jiang,

    Roles Investigation, Methodology

    Affiliation Eastman Institute for Oral Health, University of Rochester Medical Center, Rochester, New York, United States of America

  • Hui-Na Qiu,

    Roles Data curation, Formal analysis

    Affiliation Department of Endocrinology, Tianjin Union Medical Center, Nankai University Affiliated Hospital, Tianjin, China

  • Fan Wu,

    Roles Data curation, Formal analysis

    Affiliation Department of Endocrinology, Tianjin Union Medical Center, Nankai University Affiliated Hospital, Tianjin, China

  • Jing-Bo Li ,

    Roles Supervision, Validation

    13207628978@163.com (JNL); doctorljb@aliyun.com (JBL)

    Affiliation Department of Endocrinology, Tianjin Union Medical Center, Nankai University Affiliated Hospital, Tianjin, China

  • Jing-Na Lin

    Roles Funding acquisition, Supervision, Validation, Writing – review & editing

    13207628978@163.com (JNL); doctorljb@aliyun.com (JBL)

    Affiliation Department of Endocrinology, Tianjin Union Medical Center, Nankai University Affiliated Hospital, Tianjin, China

Abstract

Background

The results of current studies on metabolic-dysfunction associated steatotic liver disease (MASLD)-related diseases, cognition and dementia are inconsistent. This study aimed to elucidate the effects of MASLD-related diseases on cognition and dementia.

Methods

By using single-nucleotide polymorphisms (SNPs) associated with different traits of NAFLD (chronically elevated serum alanine aminotransferase levels [cALT], imaging-accessed and biopsy-proven NAFLD), metabolic dysfunction-associated steatohepatitis, and liver fibrosis and cirrhosis, we employed three methods of mendelian randomization (MR) analysis (inverse-variance weighted [IVW], weighted median, and MR-Egger) to determine the causal relationships between MASLD-related diseases and cognition and dementia. We used Cochran’s Q test to examine the heterogeneity, and MR-PRESSO was used to identify outliers (NbDistribution = 10000). The horizontal pleiotropy was evaluated using the MR-Egger intercept test. A leave-one-out analysis was used to assess the impact of individual SNP on the overall MR results. We also repeated the MR analysis after excluding SNPs associated with confounding factors.

Results

The results of MR analysis suggested positive causal associations between MASLD confirmed by liver biopsy (p of IVW = 0.020, OR = 1.660, 95%CI = 1.082–2.546) and liver fibrosis and cirrhosis (p of IVW = 0.009, OR = 1.849, 95%CI = 1.169–2.922) with vascular dementia (VD). However, there was no evidence of a causal link between MASLD-related diseases and cognitive performance and other types of dementia (any dementia, Alzheimer’s disease, dementia with lewy bodies, and frontotemporal dementia). Sensitivity tests supported the robustness of the results.

Conclusions

This two-sample MR analysis suggests that genetically predicted MASLD and liver fibrosis and cirrhosis may increase the VD risk. Nonetheless, the causal effects of NAFLD-related diseases on VD need more in-depth research.

1 Introduction

Because of its close association with metabolic abnormalities, the term "non-alcoholic fatty liver disease" has recently been replaced by "metabolic-dysfunction associated steatotic liver disease (MASLD)" [1]. MASLD is characterized by the abnormal accumulation of fat in the liver tissue in the absence of excessive intake of alcohol or other special causes. It ranges from simple hepatic steatosis, metabolic dysfunction-associated steatohepatitis (MASH), liver fibrosis, and hepatocellular carcinoma, depending on the stage of disease development. In recent years, the global prevalence of MASLD has continued to increase, and the adult prevalence was about 37% in 2019 [2], which poses a great threat to public health. MASLD is often accompanied by obesity and abnormal metabolism of blood lipids and glucose [3, 4], and has also been shown to be a risk factor for diabetes, hypertension, and cardiovascular and cerebrovascular diseases [5, 6]. Recently, it has been found that MASLD may also be associated with cognitive impairment and dementia. Since both MASLD and dementia are strongly associated with obesity, diabetes, and cardiovascular and cerebrovascular events [7, 8], it may be reasonable to speculate that there is an association between the two. Indeed, current research on MASLD-related diseases and cognitive impairment and dementia is limited and the results obtained are not consistent.

A longitudinal analysis from the Rotterdam Study showed that MASLD and fibrosis were not associated with either cognitive impairment or the risk of dementia [9], but results from another four-year longitudinal study supported a positive association between MASLD and risk of cognitive impairment [10]; the results of these two large, recently published longitudinal studies are conflicting. In addition, a recent meta-analysis of seven studies involving 891,562 subjects showed that MASLD was associated with an increased risk of cognitive impairment (OR = 1.44; 95%CI: 1.17–1.78; p = 0.001), but not with all-cause dementia (p = 0.341) or Alzheimer’s Disease (AD, p = 0.489) [11]. Overall, the current evidence is insufficient to elucidate the relationship between MASLD-related diseases and cognitive impairment and dementia. Moreover, due to the limitations of observational studies, a causal relationship between the two cannot be established and confounding factors are inevitable.

Mendelian randomization (MR) is an epidemiological research method that uses genetic variants to explore the causal relationship between exposure and outcome. Compared with traditional observational studies, MR can minimize the bias caused by confounding factors [12]. Several previous studies using MR have identified causal relationships between MASLD and some risk factors [13, 14], but so far no causal relationship between MASLD-related diseases and cognitive performance or dementia has been established.

Based on the Genome wide association study (GWAS) data from several large cohorts, this study aimed to elucidate the effects of MASLD-related diseases (MASLD, MASH, liver fibrosis, and cirrhosis) on global cognitive performance and risk of several types of dementia (any dementia, AD, vascular dementia [VD], dementia with lewy bodies [DLB], and frontotemporal dementia [FTD]) using two-sample MR analysis.

2 Methods

Data used for analysis in this study were collected from pooled GWAS data that had been published, and therefore no additional ethical approval or informed consent was required. Fig 1 summarizes the flow of this study.

thumbnail
Fig 1. Flowchart of the Mendelian randomization study exploring the causal relationships between MASLD-related diseases, cognition, and dementia.

MASLD, metabolic-dysfunction associated steatotic liver disease; MASH, metabolic dysfunction-associated steatohepatitis; SNPs, single-nucleotide polymorphisms; LD, linkage disequilibrium; T2D, type 2 diabetes.

https://doi.org/10.1371/journal.pone.0297883.g001

2.1 Data source for MASLD-related diseases and selection of instrumental variables (IVs)

Data on MASLD were obtained from a large-scale GWAS recently published by Vujkovic et al. [15], which referred to MASLD in terms of chronically unexplained elevated serum alanine aminotransferase (cALT). Specifically, an increase in ALT > 40 U/L in men or > 30 U/L in women at two time points at least 6 months apart within 2 years after exclusion of other liver diseases. The study reported 77 single-nucleotide polymorphisms (SNPs) that were significantly associated with cALT at the genome-wide level (p < 5×10−8) in an interracial discovery cohort (n = 218,595, 75.1% European Americans, 17.1% African Americans, 6.9% Hispanic Americans, and 0.9% Asian Americans). These SNPs were subsequently validated with two external cohorts (imaging-assessed liver fat [n = 44,289] and biopsis-proven MASLD [n = 63,969]), with 22 and 36 SNPs successfully replicated, respectively. Finally, there were 17 SNPs that were replicated in both validation cohorts. Therefore, four sets of IVs related to MASLD were used in the MR analysis: (1) 77 CALT-associated SNPs; (2) 22 SNPs associated with cALT in the imaging cohort; (3) 36 SNPs associated with cALT in the biopsy cohort; (4) 17 SNPs associated with cALT in both cohorts.

The SNPs associated with MASH and liver fibrosis and cirrhosis were derived from the FinnGen (https://www.finngen.fi/en/access_results), which contains aggregated GWAS data for a large number of European populations [16]. The summary data of MASH included 157 cases and 377,120 controls, and the summary data of liver fibrosis and cirrhosis included 1,841 cases and 366,450 controls. In order to obtain more IVs, we relaxed the correlation thresholds of MASH and liver fibrosis and cirrhosis to 5×10−6.

SNPs with linkage disequilibrium (LD, r2 > 0.001), incompatible SNPs and palindromic sequences with moderate minor allele frequency were excluded. F-statistics were also calculated to assess the strength of the IVs. F-statistics > 10 means that IVs can effectively represent exposure factors [17].

2.2 Data sources for cognitive performance and dementia

Summary statistics on cognitive performance was obtained from the Social Science Genetic Association Consortium (SSGAC). This dataset meta-analyzed GWAS data from the UK Biobank (n = 225,056, European ethnicity) and the Cognitive Genomics Consortium (COGENT, n = 35,298, European ethnicity) [18, 19] (S1 Table in S2 File). The ’g’ index was used here to assess overall cognitive function, which was calculated by integrating all cognitive assessment test scores for each individual [19]. Previous studies have shown that the comprehensive index ’g’ is basically consistent with the results of the original cognitive tests and can accurately reflect the global cognitive performance of subjects [20, 21].

Pooled GWAS data for AD was obtained from the Psychiatric Genomics Consortium (PGC). The data comes from a recent large-scale meta-analysis of GWAS from 13 cohorts by Wightman et al. [22] (S1 Table in S2 File). This study contained 90,338 late-onset AD (LOAD) cases (46,613 proxies) and 1,036,225 (318,246 proxies) controls of European ethnicity. Late age of onset poses difficulties in recruiting LOAD cases, so relatively young individuals were included in the study as proxy cases or controls, determined on the basis of their parents’ age and LOAD-related conditions.

The GWAS data of any dementia, VD, DLB, and FTD could be downloaded from https://gwas.mrcieu.ac.uk/. Participants in the any dementia included 5,933 cases and 212,859 controls of European ancestry, and participants in the VD included 98 cases and 211,300 controls of European ancestry. The GWAS data for DLB included 2,591 cases and 4,027 controls (European ethnicity) from 44 consortia [23]. The summary statistics on FTD included 515 cases and 2,509 controls (European ethnicity) from 45 clinical centers [24]. An overview of all the GWAS data included in this study is provided in Table 1.

thumbnail
Table 1. Overview of the data used in the mendelian randomization analysis.

https://doi.org/10.1371/journal.pone.0297883.t001

2.3 Statistical analysis

Three MR analysis methods (random-effect models of inverse-variance weighted [IVW], weighted median, and MR-Egger) were used to determine the causal relationships between MASLD-related diseases, cognitive performance, and dementia. IVW represents the primary analysis results, and this analysis is characterized by regression that does not consider the intercept term and uses the inverse of the squared outcome variance as the weight for fitting. As a supplementary estimation result, the assumption of MR-Egger analysis is that all IVs are pleiotropic, that is, the existence of intercept terms is considered in the regression, and the weighted median considers that half of the genetic variation is pleiotropic [25, 26]. Therefore, the combination of the three MR analysis methods is able to provide more robust outcome estimates.

Before MR analysis, we used Cochran’s Q test to determine the heterogeneity of IVs, and MR-PRESSO was used to identify potential outliers (NbDistribution = 10000). If outliers were present (p < 0.05), they were removed and then subjected to MR analysis [27]. Next, we also used multiple sensitivity analyses to ensure the reliability of our results. Horizontal pleiotropy was assessed using the MR-Egger intercept test. The presence of significant horizontal pleiotropy (p < 0.05) indicates that the genetic variants do not meet the basic condition for being IVs [28]. Leave-one-out analysis was used to evaluate the effect of individual SNP on the overall MR estimation results. We also calculated the overall statistical power (https://shiny.cnsgenomics.com/mRnd/), and the recommended value is > 80% [29].

Finally, in order to eliminate the effects of potential confounding factors, we also examined whether each tool SNP had genome-wide level association (p < 5×10−8) with key risk factors for cognitive impairment and dementia on the PhenoScanner (www.phenoscanner.medschl.cam.ac.uk) [30, 31], including cerebrovascular disease, hypertension, heart disease (atrial fibrillation, arrhythmia and cardiac arrest), atherosclerosis, cerebral amyloid angiopathy, type 2 diabetes (T2D), hyperinsulinemia, traumatic brain injury, epilepsy, depression, respiratory diseases, anemia, hearing disorders, sleep disorders, brain-derived neurotrophic factor, homocysteine, educational attainment, physical activity, diet, smoking, and alcohol drinking [32, 33].

TwoSampleMR (0.4.9) and MR-PRESSO (1.0) packages of R software (4.3.0) were used to perform all analyses. Two-sided p values of less than 0.05 were considered to indicate statistical significances.

3 Results

3.1 Causal relationship between MASLD-related diseases and cognitive performance

The cALT-related SNPs in LD were removed and then merged and harmonized with the outcome data of cognitive performance, resulting in 45 IVs with F-statistics ranging from 35 to 167 that could be used for further analysis (S2 Table in S2 File). The following Cochran’s Q test suggested the presence of heterogeneity: p of IVW = 1.02e-07. Consistent with this finding, MR-PRESSO test also suggested the presence of outliers (p < 1e-04, rs10883451, rs174535 and rs5117). Thus, we performed MR analysis after removing the three outliers (S2 Table in S2 File), and the results suggested no causal relationship between genetically predicted MASLD and global cognitive performance (p of IVW = 0.847, Table 2). Similarly, we performed these MR analyses with the use of three other sets of IVs representing MASLD (S2 Table in S2 File). According to the results of heterogeneity test and MR-PRESSO analysis, we performed MR analysis after removing outliers (rs5117 in imaging-assessed cohort, rs10883451 and rs5117 in biopsy-confirmed cohort, and rs5117 in these two cohorts). The results were consistent with our previous findings, and all suggested no causal relationship between genetically predicted MASLD and global cognitive performance (all p values were greater than 0.05, Table 2).

thumbnail
Table 2. Mendelian randomization analysis of MASLD-related diseases and cognitive performance.

https://doi.org/10.1371/journal.pone.0297883.t002

After data processing, we obtained 3 MASH IVs and 12 liver fibrosis and cirrhosis IVs that could be used for MR analysis (S2 Table in S2 File). Cochran’s Q test showed no heterogeneity: p of IVW = 0.102 or 0.060. MR-PRESSO tests also showed no outliers. The results of the subsequent MR analysis showed that neither genetically predicted MASH nor liver fibrosis and cirrhosis was associated with global cognitive performance (all p values were greater than 0.05, Table 2). In addition, MR-Egger intercept tests showed no horizontal pleiotropy (p > 0.05; Table 2). No single SNP strongly violated the overall effect of MASLD-related diseases on cognitive performance in leave-one-out sensitivity analyses (S1-S3 Figs in S1 File).

Finally, in order to exclude the possible influence of potential confounders, based on the results searched online (www.phenoscanner.medschl.cam.ac.uk), we further excluded the SNPs associated with T2D (rs13389219, rs17036160, rs2943652, rs56094641 and rs58542926) and fasting insulin (rs4841132) in the c-ALT IVs. Results of MR analysis still showed no causal relationship between MASLD and cognitive performance (p of IVW = 0.996, p of MR-Egger = 0.375, p of weighted median = 0.958). Similar results were obtained after removing these confounding SNPs from the other three sets of IVs. We also found a T2D-related SNP (rs1802295) in the IVs of liver fibrosis and cirrhosis. After excluding this SNP, the results of the MR analysis were consistent (p of IVW = 0.846, p of MR-Egger = 0.454, p of weighted median = 0.475).

3.2 Causal relationship between MASLD-related diseases and any dementia

After removing SNPs in LD and merging and harmonizing them with the outcome data on any dementia, the numbers of MASLD-related diseases IVs that were available for further analyses were 52, 16, 28, 13, 3, and 14, respectively (S3 Table in S2 File). Cochran’s Q test suggested no heterogeneity, and MR-PRESSO analysis showed no outliers (p > 0.05). MR analysis showed no causal association between genetically predicted MASLD-related diseases and any dementia (all p values were greater than 0.05, Table 3). Sensitivity analysis showed no horizontal pleiotropy (p > 0.05, Table 3) and no single SNP significantly affected overall MR results (S4-S6 Figs in S1 File).

thumbnail
Table 3. Mendelian randomization analysis of MASLD-related diseases and any dementia.

https://doi.org/10.1371/journal.pone.0297883.t003

In addition, by searching for the second phenotype, we identified SNPs associated with T2D (rs13389219, rs17036160, rs2943652, rs56094641 and rs58542926), fasting insulin (rs4841132) and hypertension (rs7653249) in the MASLD cohort, and SNPs associated with T2D (rs1802295 and rs8100204) in the liver fibrosis and cirrhosis cohort. After excluding these potentially confounding SNPs, the results of the MR analysis still indicated no causal association (p > 0.05).

3.3 Causal relationship between MASLD-related diseases and AD

The numbers of MASLD-related diseases IVs that were available for MR analyses were 38, 14, 22, 11, 2, and 12, respectively (S4 Table in S2 File). Neither Cochran’s Q test nor MR-PRESSO analysis suggested heterogeneity or outliers in these IVs (p > 0.05). When MR analyses were performed using the first set of IVs (associated with cALT, n = 38), the third set (associated with cALT in the biopsy cohort, n = 22), and the fourth set (associated with cALT in both the biopsy and imaging cohorts, n = 11), the p values of IVW were greater than 0.05, which represented the results of the primary analysis (25), as well as for MR-Egger. Therefore, despite the small p values of weighted median (p = 0.026, 0.039 and 0.039, respectively), the overall results of MR analysis suggested that there was no causal relationship between MASLD and AD (Table 4). However, when MR analysis was conducted using the second set of IVs (associated with cALT in the imaging cohort, n = 14), the results indicated a significant association between genetically predicted MASLD and AD (p of IVW = 0.049, p of MR-Egger = 0.181, p of weighted median = 0.048, Table 4). On the other hand, since the number of IVs related to MASH was less than 3, we could only conduct IVW analysis, and the result showed that there was no causal relationship between MASH and AD (p of IVW = 0.095, Table 4). There was also no evidence of a causal relationship between liver fibrosis and cirrhosis and AD (p of IVW = 0.179, Table 4). MR-Egger intercept test did not suggest any pleiotropy (p > 0.05, Table 4). Leave-one-out analysis also indicated that no single SNP strongly violated the global effect of MASLD-related diseases on AD (S7 and S8 Figs in S1 File).

thumbnail
Table 4. Mendelian randomization analysis of MASLD-related diseases and AD.

https://doi.org/10.1371/journal.pone.0297883.t004

We then repeated MR analysis with the exclusion of five T2D-related SNPs (rs13389219, rs17036160, rs2943652, rs56094641 and rs58542926) in the MASLD IVs and one T2D-related SNP (rs1802295) in the liver fibrosis and cirrhosis IVs, and all results showed no causal relationship between MASLD-related diseases and risk of AD (p > 0.05). This meant that the relationship between image-verified MASLD and AD became insignificant (p of IVW = 0.124, Table 4), indicating that the previous positive results were due to the confounding factor. MR-Egger intercept test showed no significant horizontal pleiotropy (p > 0.05, Table 4).

3.4 Causal relationship between MASLD-related diseases and VD

After processing the data, there were 52, 16, 28, 13, 3, 14 MASLD-related diseases IVs that could be used for MR analysis. The Cochran’s Q test suggested the presence of heterogeneity in the c-ALT cohort: p of IVW = 0.039, and MR-PRESSO test also suggested one outlier (p = 0.030, rs1658943). Thus, we performed MR analysis after excluding this outlier (S5 Table in S2 File). The IWV, which represents the primary MR result, showed no causal association between MASLD and MASH for VD (p > 0.05, Table 5), although the P-values obtained by other analyses were less than 0.05. However, the MR analysis indicated a significant association between genetically predicted liver fibrosis and cirrhosis and VD (p of IVW = 0.009, p of MR-Egger = 0.429, p of weighted median = 0.021, Table 5). Sensitivity analysis showed no significant intercept (p > 0.05; Table 5) and no single SNP significantly affected overall MR estimates (S9-S11 Figs in S1 File).

thumbnail
Table 5. Mendelian randomization analysis of MASLD-related diseases and VD.

https://doi.org/10.1371/journal.pone.0297883.t005

Next, we repeated MR analysis with the exclusion of SNPs associated with T2D (rs13389219, rs17036160, rs2943652, rs56094641 and rs58542926), fasting insulin (rs4841132) and hypertension (rs7653249) in the MASLD IVs and two T2D-related SNPs (rs1802295 and rs8100204) in the liver fibrosis and cirrhosis IVs. Unlike before, we found that the causal association between MASLD confirmed by liver biopsy and VD became significant (p of IVW = 0.020, p of MR-Egger = 0.022, p of weighted median = 0.006, Table 5). MR-Egger intercept test also proved that the result was reliable (p > 0.05, Table 5). The associations between the other three sets of MASLD-related IVs and VD were also more significant than before, although still not statistically significant (p of IVW = 0.056, 0.051, and 0.050; S6 Table in S2 File). The MR results for MASH and liver fibrosis and cirrhosis and VD remained the same as before.

3.5 Causal relationship between MASLD-related diseases and DLB

The numbers of MASLD-related diseases IVs that were available for further MR analyses were 49, 16, 27, 12, 3, and 13, respectively (S7 Table in S2 File). Cochran’s Q test suggested no heterogeneity, and MR-PRESSO analysis showed no outliers (p > 0.05). MR analysis showed no causal association between genetically predicted MASLD-related diseases and DLB (all p values were greater than 0.05, Table 6). Sensitivity analysis showed no horizontal pleiotropy (p > 0.05, Table 6) and no single SNP significantly affected overall MR results (S12-14 Figs in S1 File).

thumbnail
Table 6. Mendelian randomization analysis of MASLD-related diseases and DLB.

https://doi.org/10.1371/journal.pone.0297883.t006

We also conducted MR analysis with the exclusion of SNPs associated with T2D (rs13389219, rs17036160, rs2943652, rs56094641 and rs58542926), fasting insulin (rs4841132) and hypertension (rs7653249) in the MASLD IVs and two T2D-related SNPs (rs1802295 and rs8100204) in the liver fibrosis and cirrhosis IVs. The results were consistent with the previous results and still showed no significant association between MASLD-related diseases and DLB.

3.6 Causal relationship between MASLD-related diseases and FTD

After processing the data, there were 4, 1, 1, 2 MASLD-related diseases IVs that could be used for MR analysis of FTD. Neither Cochran’s Q test nor MR-PRESSO analysis suggested heterogeneity or outliers in these IVs (p > 0.05). Since the numbers of IVs in the imaging, biopsy and liver fibrosis and cirrhosis cohorts were less than 3, we could only conduct wald ratio or IVW analysis, and the result showed that there was no causal relationship between MASLD-related diseases and FTD (all p values were greater than 0.05, Table 7). Sensitivity analysis suggested that there was no significant intercept (p > 0.05, Table 7) and no single SNP significantly affected overall MR results (S15 Fig in S1 File). After removing one T2D-related SNP (rs1802295) in the liver fibrosis and cirrhosis cohort, the MR analysis still showed no significant correlation.

thumbnail
Table 7. Mendelian randomization analysis of MASLD-related diseases and FTD.

https://doi.org/10.1371/journal.pone.0297883.t007

4 Discussion

By using the genes of MASLD-associated diseases as IVs, the results of MR analysis suggest causal associations between MASLD confirmed by liver biopsy and liver fibrosis and cirrhosis with VD. Genetically predicted MASLD and liver fibrosis and cirrhosis may increase the risk of VD. However, we did not find a causal association between MASLD-related diseases and cognitive performance and other types of dementia.

The MR analysis showed that MASLD (based on cALT in the imaging cohort) was a protective factor for AD, but this association disappeared when the confounder T2D was excluded. We speculated that this result was mainly driven by T2D. To verify the conjecture, we performed a leave-one-out analysis. The result showed that there were three confounding SNPs that might have great impacts on the overall MR result: rs2943652, rs56094641 and rs58542926. The result of MR analysis was no longer significant after the removal of these SNPs (S16 Fig in S1 File). In fact, extra copies of the effecting alleles in the three SNPs were associated with a reduced risk of T2D [34]. So overall, our results did not prove a causal link between MASLD (based on cALT in the imaging cohort) and AD.

The current findings suggest that MASLD and liver fibrosis and cirrhosis are associated with VD risk, which support some recent studies. Evidence from a recent large matched cohort study involving more than 30,000 people showed that MASLD was associated with increased risk of VD. The diagnosis of MASLD is based on the International Classification of Diseases code from the National Patient Registry in Sweden [35]. Another large cross-sectional study from rural China also suggested that moderate and severe MASLD, as estimated by abdominal ultrasonography, was associated with VD [36]. Nevertheless, another prospective cohort study with a median follow-up of 15.5 years showed that MASLD, as estimated by fatty liver index, was not associated with either the risk of incident dementia or cognitive decline [9]. The different diagnostic methods of MASLD (e.g., fatty liver index, abdominal ultrasound) and different cognitive function measures (e.g., various cognitive function tests) used in these studies may account for the inconsistent results. In addition, even with a prospective study design, there will inevitably be inverse associations and confounding factors that can interfere with the results. Therefore, the MR design of our study can effectively overcome the limitations of observational studies to reveal whether there is a causal relationship between MASLD and cognitive ability and risk of dementia. Compared with previous MASLD-related MR analysis [37], the SNPs associated with MASLD at the genome-wide level used in our study were collected from gene summary data of a larger population with sufficient statistical strength. In addition, we used four sets of IVs associated with different phenotypes of MASLD, including liver biopsy-proven MASLD, which is the gold standard for MASLD diagnosis, thereby ensuring that our MR results are more reliable and robust. To ensure the rigor of the study, we also searched the website for SNPs associated with more than a dozen confounding factors and performed sensitivity analysis after excluding confounding SNPs.

Our results suggest that MASLD-associated diseases are associated with VD, but not with other types of dementia (including AD, DLB and FTD). This may be because the association between MASLD and dementia is primarily mediated by vascular lesions. In fact, metabolic disorders in MASLD patients can lead to a variety of vascular damage, such as atherosclerosis, vascular endothelial dysfunction, reduced blood flow [38]. The clinical manifestations of various vascular diseases (stroke, coronary heart disease, hypertension, etc.) are also closely related to MASLD [39]. In addition, there is evidence that patients with MASLD have decreased middle cerebral artery blood flow, and long-term cerebral hypoperfusion can induce microvascular lesions and accelerate the occurrence of cognitive decline and dementia [40]. Besides, a study has shown that biomarkers of fibrosis are better predictors of dementia risk than traditional cardiometabolic risk factors [35]. In severe hepatic steatosis or fibrosis, the systemic chronic inflammatory state caused by adipokine dysfunction may be one of the main causes of dementia [41]. Previous studies have shown that inflammation can significantly accelerate cognitive decline [42, 43].

To the best of our knowledge, this is the first study to explore the causal relationship between genetically predicted MASLD and global cognitive performance and dementia risk using the two-sample MR analysis. The main advantage of this study is that the MR design can effectively simulate the randomized controlled trial and minimize the influence of confounding bias, because the SNPs used as IVs in the study are randomly assigned to the offspring. At the same time, MR studies can avoid the possible reverse causality in observational studies. Given the high prevalence of MASLD in the population and the significant disease burden from cognitive impairment and dementia, our findings may have implications for primary care policy making to the extent that enhanced screening for VD in patients with genetically determined MASLD may be useful.

However, there are some limitations to this study. First, the cALT we used is not a perfect biomarker for MASLD [44], and therefore some MASLD susceptibility genes unrelated to cALT may be missed. But we also used imaging-based and biopsy-proven MASLD-associated SNPs as IVs, which also ensured the reliability of the current study method. Second, due to the limitations of the summary-level dataset, we were unable to further analyze the causal relationship between MASLD-related diseases and function in different cognitive domains.

5 Conclusions

This two-sample MR analysis suggests that genetically predicted MASLD and liver fibrosis and cirrhosis may increase the VD risk. Anyway, more prospective studies are needed to prove this link in the future.

Supporting information

S1 File. Leave-one-out analysis for MASLD-related diseases and cognitive performance and dementia.

https://doi.org/10.1371/journal.pone.0297883.s001

(PDF)

S2 File. Description of outcome data and construction of instrumental variables.

https://doi.org/10.1371/journal.pone.0297883.s002

(XLSX)

Acknowledgments

We would like to thank the SSGAC, PGC, Finngen, and IEU OpenGWAS project that provided the GWAS data for our MR analyses.

References

  1. 1. Lekakis V, Papatheodoridis GV. Natural history of metabolic dysfunction-associated steatotic liver disease [published online ahead of print, 2023 Nov 6]. Eur J Intern Med. 2023;S0953–6205(23)00397-7. pmid:37940495
  2. 2. Le MH, Yeo YH, Li X, Li J, Zou B, Wu Y, et al. 2019 Global NAFLD Prevalence: A Systematic Review and Meta-analysis. Clin Gastroenterol Hepatol. 2022 Dec;20(12):2809–2817.e28. pmid:34890795
  3. 3. Li L, Liu DW, Yan HY, Wang ZY, Zhao SH, Wang B. Obesity is an independent risk factor for non-alcoholic fatty liver disease: evidence from a meta-analysis of 21 cohort studies. Obes Rev 2016;17(6):510–519. pmid:27020692
  4. 4. Young S, Tariq R, Provenza J, Satapathy SK, Faisal K, Choudhry A, et al. Prevalence and Profile of Nonalcoholic Fatty Liver Disease in Lean Adults: Systematic Review and Meta-Analysis. Hepatol Commun 2020;4(7):953–972. pmid:32626829
  5. 5. Mantovani A, Byrne CD, Bonora E, Targher G. Nonalcoholic Fatty Liver Disease and Risk of Incident Type 2 Diabetes: A Meta-analysis. Diabetes Care 2018;41:372–382. pmid:29358469
  6. 6. Targher G, Byrne CD, Lonardo A, Zoppini G, Barbui C. Non-alcoholic fatty liver disease and risk of incident cardiovascular disease: A meta-analysis. J Hepatol 2016;65:589–600. pmid:27212244
  7. 7. Love S, Miners JS. Cerebrovascular disease in ageing and Alzheimer’s disease. Acta Neuropathol. 2016 May;131(5):645–58. pmid:26711459
  8. 8. Li X, Song D, Leng SX. Link between type 2 diabetes and Alzheimer’s disease: from epidemiology to mechanism and treatment. Clin Interv Aging. 2015 Mar 10;10:549–60. pmid:25792818
  9. 9. Xiao T, van Kleef L, Ikram MK, de Knegt J, Ikram MA. Association of Nonalcoholic Fatty Liver Disease and Fibrosis With Incident Dementia and Cognition: The Rotterdam Study. Neurology. 2022 Aug 9;99(6):e565–e573. pmid:35618435
  10. 10. Liu Q, Liu C, Hu F, Deng X, Zhang Y. Non-alcoholic Fatty Liver Disease and Longitudinal Cognitive Changes in Middle-Aged and Elderly Adults. Front Med (Lausanne). 2022 Jan 17;8:738835. pmid:35111769
  11. 11. Wang L, Sang B, Zheng Z. Risk of dementia or cognitive impairment in non-alcoholic fatty liver disease: A systematic review and meta-analysis. Front Aging Neurosci. 2022 Sep 20;14:985109. pmid:36204558
  12. 12. Davey Smith G, Hemani G. Mendelian randomization: genetic anchors for causal inference in epidemiological studies. Hum Mol Genet. 2014 Sep 15;23(R1):R89–98. pmid:25064373
  13. 13. Xie J, Huang H, Liu Z, Li Y, Yu C, Xu L, et al. The associations between modifiable risk factors and nonalcoholic fatty liver disease: A comprehensive Mendelian randomization study. Hepatology. 2023 Mar 1;77(3):949–964. pmid:35971878
  14. 14. Ren Z, Simons PIHG, Wesselius A, Stehouwer CDA, Brouwers MCGJ. Relationship between NAFLD and coronary artery disease: A Mendelian randomization study. Hepatology. 2023 Jan 1;77(1):230–238. pmid:35441719
  15. 15. Vujkovic M, Ramdas S, Lorenz KM, Guo X, Darlay R, Cordell HJ, et al. A multiancestry genome-wide association study of unexplained chronic ALT elevation as a proxy for nonalcoholic fatty liver disease with histological and radiological validation. Nat Genet. 2022 Jun;54(6):761–771. pmid:35654975
  16. 16. Kurki MI, Karjalainen J, Palta P, Sipilä TP, Kristiansson K, Donner KM, et al. FinnGen provides genetic insights from a well-phenotyped isolated population. Nature. 2023 Jan;613(7944):508–518. pmid:36653562
  17. 17. Pierce BL, Ahsan H, Vanderweele TJ. Power and instrument strength requirements for Mendelian randomization studies using multiple genetic variants. Int J Epidemiol. 2011 Jun;40(3):740–52. pmid:20813862
  18. 18. Becker J, Burik CAP, Goldman G, Wang N, Jayashankar H, Bennett M, et al. Resource profile and user guide of the Polygenic Index Repository. Nat Hum Behav. 2021 Dec;5(12):1744–1758. pmid:34140656
  19. 19. Trampush JW, Yang ML, Yu J, Knowles E, Davies G, Liewald DC, et al. GWAS meta-analysis reveals novel loci and genetic correlates for general cognitive function: a report from the COGENT consortium. Mol Psychiatry. 2017 Mar;22(3):336–345. pmid:28093568
  20. 20. Williams CM, Labouret G, Wolfram T, Peyre H, Ramus F. A General Cognitive Ability Factor for the UK Biobank. Behav Genet. 2023 Mar;53(2):85–100. pmid:36378351
  21. 21. Panizzon MS, Vuoksimaa E, Spoon KM, Jacobson KC, Lyons MJ, Franz CE, et al. Genetic and Environmental Influences of General Cognitive Ability: Is g a valid latent construct? Intelligence. 2014;43:65–76. pmid:24791031
  22. 22. Wightman DP, Jansen IE, Savage JE, Shadrin AA, Bahrami S, Holland D, et al. A genome-wide association study with 1,126,563 individuals identifies new risk loci for Alzheimer’s disease. Nat Genet. 2021 Sep;53(9):1276–1282. pmid:34493870
  23. 23. Chia R, Sabir MS, Bandres-Ciga S, Saez-Atienzar S, Reynolds RH, Gustavsson E, et al. Genome sequencing analysis identifies new loci associated with Lewy body dementia and provides insights into its genetic architecture. Nat Genet. 2021 Mar;53(3):294–303. pmid:33589841
  24. 24. Van Deerlin VM, Sleiman PM, Martinez-Lage M, Chen-Plotkin A, Wang LS, Graff-Radford NR, et al. Common variants at 7p21 are associated with frontotemporal lobar degeneration with TDP-43 inclusions. Nat Genet. 2010 Mar;42(3):234–9. pmid:20154673
  25. 25. Bowden J, Davey Smith G, Burgess S. Mendelian randomization with invalid instruments: effect estimation and bias detection through Egger regression. Int J Epidemiol. 2015 Apr;44(2):512–25. pmid:26050253
  26. 26. Ong JS, MacGregor S. Implementing MR-PRESSO and GCTA-GSMR for pleiotropy assessment in Mendelian randomization studies from a practitioner’s perspective. Genet Epidemiol. 2019 Sep;43(6):609–616. pmid:31045282
  27. 27. Burgess S, Bowden J, Fall T, Ingelsson E, Thompson SG. Sensitivity Analyses for Robust Causal Inference from Mendelian Randomization Analyses with Multiple Genetic Variants. Epidemiology. 2017 Jan;28(1):30–42. pmid:27749700
  28. 28. Verbanck M, Chen CY, Neale B, Do R. Detection of widespread horizontal pleiotropy in causal relationships inferred from Mendelian randomization between complex traits and diseases. Nat Genet. 2018 May;50(5):693–698. pmid:29686387
  29. 29. Brion MJ, Shakhbazov K, Visscher PM. Calculating statistical power in Mendelian randomization studies. Int J Epidemiol. 2013 Oct;42(5):1497–501. pmid:24159078
  30. 30. Staley JR, Blackshaw J, Kamat MA, Ellis S, Surendran P, Sun BB, et al. PhenoScanner: a database of human genotype-phenotype associations. Bioinformatics. 2016 Oct 15;32(20):3207–3209. pmid:27318201
  31. 31. Kamat MA, Blackshaw JA, Young R, Surendran P, Burgess S, Danesh J, et al. PhenoScanner V2: an expanded tool for searching human genotype-phenotype associations. Bioinformatics. 2019 Nov 1;35(22):4851–4853. pmid:31233103
  32. 32. Edwards Iii GA, Gamez N, Escobedo G Jr, Calderon O, Moreno-Gonzalez I. Modifiable Risk Factors for Alzheimer’s Disease. Front Aging Neurosci. 2019 Jun 24;11:146. pmid:31293412
  33. 33. Zhang YR, Xu W, Zhang W, Wang HF, Ou YN, Qu Y, et al. Modifiable risk factors for incident dementia and cognitive impairment: An umbrella review of evidence. J Affect Disord. 2022 Oct 1;314:160–167. pmid:35863541
  34. 34. Gaulton KJ, Ferreira T, Lee Y, Raimondo A, Mägi R, Reschen ME, et al; DIAbetes Genetics Replication And Meta-analysis (DIAGRAM) Consortium. Genetic fine mapping and genomic annotation defines causal mechanisms at type 2 diabetes susceptibility loci. Nat Genet. 2015 Dec;47(12):1415–25. pmid:26551672
  35. 35. Shang Y, Widman L, Hagström H. Nonalcoholic Fatty Liver Disease and Risk of Dementia: A Population-Based Cohort Study. Neurology. 2022 Aug 9;99(6):e574–e582. pmid:35831178
  36. 36. Wang Y, Li Y, Liu K, Han X, Dong Y, Wang X, et al. Nonalcoholic fatty liver disease, serum cytokines, and dementia among rural-dwelling older adults in China: A population-based study. Eur J Neurol. 2022 Sep;29(9):2612–2621. pmid:35608965
  37. 37. Miao Z, Garske KM, Pan DZ, Koka A, Kaminska D, Männistö V, et al. Identification of 90 NAFLD GWAS loci and establishment of NAFLD PRS and causal role of NAFLD in coronary artery disease. HGG Adv. 2021 Aug 24;3(1):100056. pmid:35047847
  38. 38. Muzurović E, Mikhailidis DP, Mantzoros C. Non-alcoholic fatty liver disease, insulin resistance, metabolic syndrome and their association with vascular risk. Metabolism. 2021 Jun;119:154770. pmid:33864798
  39. 39. Fargion S, Porzio M, Fracanzani AL. Nonalcoholic fatty liver disease and vascular disease: state-of-the-art. World J Gastroenterol. 2014 Oct 7;20(37):13306–24. pmid:25309067
  40. 40. VanWagner LB, Terry JG, Chow LS, Alman AC, Kang H, Ingram KH, et al. Nonalcoholic fatty liver disease and measures of early brain health in middle-aged adults: The CARDIA study. Obesity (Silver Spring). 2017 Mar;25(3):642–651. pmid:28169509
  41. 41. Marra F, Bertolani C. Adipokines in liver diseases. Hepatology. 2009 Sep;50(3):957–69. pmid:19585655
  42. 42. Rosenberg GA, Bjerke M, Wallin A. Multimodal markers of inflammation in the subcortical ischemic vascular disease type of vascular cognitive impairment. Stroke 2014;45(5):1531–1538. pmid:24692476
  43. 43. Takeda S, Sato N, Morishita R. Systemic inflammation, blood-brain barrier vulnerability and cognitive/non-cognitive symptoms in Alzheimer disease: relevance to pathogenesis and therapy. Front Aging Neurosci 2014;6:171. pmid:25120476
  44. 44. Mofrad P, Contos MJ, Haque M, Sargeant C, Fisher RA, Luketic VA, et al. Clinical and histologic spectrum of nonalcoholic fatty liver disease associated with normal ALT values. Hepatology. 2003 Jun;37(6):1286–92. pmid:12774006