Association of type 2 diabetes mellitus and other glycemic traits with the risk of NAFLD

doi:10.21203/rs.3.rs-2377438/v1

Download PDF

Article

Association of type 2 diabetes mellitus and other glycemic traits with the risk of NAFLD

https://doi.org/10.21203/rs.3.rs-2377438/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

Extensive observational evidence has proposed an association between Non-Alcoholic Fatty Liver Disease (NAFLD) and type 2 diabetes mellitus (T2DM). However, the causal relationship between these two diseases requires further investigation. Therefore, we assessed the potential causal relationship between NAFLD and T2DM and other glycaemic traits using two-sample Mendelian randomization (MR).We selected fasting insulin, T2DM, fasting glucose and Hemoglobin A1c(Hb1Ac) from meta-analyses of four genome-wide association studies (GWASs). GWAS data for NAFLD were obtained from the FinnGen consortium. MR analysis was accomplished using the inverse variance weighted (IVW) method, weighted median (WM) method, and MR-Egger regression analysis. Sensitivity analyses were conducted to evaluate heterogeneity and pleiotropy.The results showed that only fasting insulin and T2DM were significantly causally related to the risk of NAFLD when combined with the FinnGen consortium [fasting insulin, odds ratio (OR): 9.54; 95% confidence interval (CI) 2.66–34.16; P = 0.0005, T2DM :OR: 1.20; 95% CI 1.00- 1.45; P = 0.0497]. No significant association was observed between fasting glucose and HbA1c and the risk of NAFLD.This study supports the independent causal role of fasting insulin and T2DM in the pathogenesis of NAFLD, both of which are risk factors for the development of NAFLD.

Health sciences/Endocrinology

Health sciences/Gastroenterology

Health sciences/Risk factors

Non-Alcoholic Fatty Liver Disease (NAFLD) and Type 2 diabetes mellitus (T2DM) are pathologies whose prevalence continues to increase worldwide¹. Both diseases are precipitated by an excessive caloric intake, which promotes insulin resistance and fatty liver². NAFLD has become the most common chronic liver disease affecting approximately 25% of the global population^3–5. It comprises a wide spectrum of pathologies ranging from non-alcoholic fatty liver(NAFL), characterized by simple steatosis without inflammation, to non-alcoholic steatohepatitis (NASH), characterized by steatosis of the liver accompanied by inflammation and hepatocyte ballooning⁶, which can lead to advanced fibrosis^7,8, cirrhosis⁹ and hepatocellular carcinoma(HCC)^10,11. All these complications of NASH can pose significant health, economic, and patient-experience burdens to the patients, their families, and society. In recent years, researchers have recognized that metabolic dysfunction is strongly associated with hepatic steatosis and, as a result, an international consensus of experts has recommended changing the abbreviation for the disease from NAFLD to metabolic (dysfunctional) associated fatty liver or "MAFLD". This change goes well beyond a mere semantic revision and may be a first step in facilitating the process to better conceptualize the disease for health promotion, patient targeting, case identification, ongoing clinical trials, and the provision of medical care^12,13.

Diabetes is one of the fastest-growing diseases worldwide, projected to affect 693 million adults by 2045 ^1,14. Since 1980 the age-standardized prevalence of T2DM in adults has increased substantially¹⁵. T2DM is closely associated with other metabolic diseases associated with excessive calorie intake and a sedentary lifestyle, such as obesity^16,17. The liver plays a central role in lipid and glucose metabolism, and hepatic insulin resistance (IR) is a hallmark of T2DM and NAFLD¹⁸. Thus, a growing number of studies have shown that NAFLD is closely associated with T2DM, which appears to be the most important risk factor for the development of NAFLD, including NASH, and the most important clinical predictor of liver-related morbidity and mortality^2,19,20.A meta-analysis investigating the global epidemiology of NAFLD and NASH in people with T2DM showed that T2DM is an important risk factor for NAFLD and that T2DM appears to accelerate the progression of NAFLD liver disease. The study also highlights that the prevalence of NAFLD is more than two times higher in patients with T2DM than in the general population²¹. However, given the complex bidirectional relationship between NAFLD, IR, and chronic hyperglycaemia, and the fact that observational studies can be confounded by potential confounding factors and reverse causality, it is difficult to distinguish whether NAFLD is a cause or a consequence of IR or T2DM ^22,23. The causal relationship between the two remains controversial.

Mendelian randomization (MR) is a type of instrumental variables (IVs) analysis that uses genetic variation as an IV to detect and quantify causal relationships²⁴. Due to MR has been increasingly used in observational studies in recent years because of its ability to overcome the effects of potential confounding and reverse causality^25,26. Early MR studies were usually conducted in small sample populations and used only a small amount of genetic variation, which made MR studies less valid²⁷. However, there has been a revolution in the field with the discovery of many genetic variants that are strongly associated with specific traits in biology, and with the public release of hundreds of thousands of aggregated data on the association of exposures and diseases with genetic variants in many large sample genome-wide association studies (GWAS). These pooled data have enabled researchers to estimate genetic associations in large samples of data, thus facilitating the development of MR research^28,29.

In this study, we aimed to assess the causal relationship between T2DM and NAFLD using 2-sample MR analysis. In addition, other glycaemic characteristics associated with diabetes, including fasting insulin, Hemoglobin A1c (HbA1c) and fasting glucose were also used to assess the potential causal relationship with the risk of NAFLD.

All methods were performed in accordance with the relevant guidelines and regulations.Based on summarized data from GWASs consortia that was made publically available, this two-sample MR analysis was carried out. The chosen genetic variants must adhere to three essential presumptions to undergo MR analyses: The following assumptions are made: (1) relevance assumption: genetic variants are closely related to the exposure of interest; (2) independence assumption: genetic variants are unrelated to any confounding factors that affect the exposure-outcome association; and (3) exclusion restriction: genetic variants do not affect outcomes unless they also affect the exposure of interest. Study design overview and assumptions of the MR design is presented in Fig. 1.

Genetic instrument selection

Glycemic traits are used to diagnose and monitor T2DM, and cardiometabolic health. To date, most genetic studies of glycemic traits have focused on individuals of European ancestry. In our study, fasting insulin, T2DM, fasting glucose and HbA1c levels were used as exposure tools in this study to infer a causal relationship between genetically predicted glycemic traits and NAFLD. Fasting insulin reflects a combination of insulin secretion and insulin resistance, both components of T2DM, and insulin clearance. Fasting glucose and HbA1c are glycemic traits used to diagnose diabetes³⁰. HbA1c, used to monitor and diagnose diabetes, is influenced by average glycemia over a 2-to 3-month period³¹. In addition, HbA1c is the most commonly used biomarker to monitor glucose control in patients with diabetes³². Collectively, all four glycemic traits are useful to better understand T2DM pathophysiology and cardiometabolic outcomes^33,34.

We selected genetic variants for fasting insulin and fasting glucose from the large GWAS data³⁵, which included 281,416 individuals without diabetes. In these individuals, ~ 30% of whom were of non-European ancestry (13% East Asian, 7% Hispanic, 6% African American, 3% South Asian, and 2% sub-Saharan African). For HbA1c, SNPs were obtained from a trans-ancestral GWAS meta-analysis³⁶, which included 46,368 nondiabetic adults of European descent. For T2DM, we selected a meta-analysis of GWAS with ~ 16 million genetic variants in 62,892 T2DM cases and 596,424 controls of European ancestry³⁷. All studies had been approved by a relevant ethical review board and participants had given informed consent. Table 1 shows the detailed clinical information of these GWASs. To ensure that the selected single nucleotide polymorphisms (SNPs) met the assumption of association of instrumental variables, we selected only SNPs that reached genome-wide significance (P < 5 × 10^− 8). To remove SNPs with linkage disequilibrium(LD), we set the parameters r² = 0.001 and kb = 10000, i.e. removing SNPs with r² > 0.001 with the most significant SNPs in the 10000kb range³⁸.

Table 1

Details of the GWASs included in the Mendelian randomization.
Phenotype	PubMed ID	SNPs available	Sample size	Author	Ancestry
Fasting insulin	34059833	29,664,438	151,013	Chen J et al	European
Type 2 diabetes mellitus	30054458	5,030,727	655,666	Xue A et al	European
Fasting glucose	34059833	31,008,728	200,622	Chen J et al	European
HbA1c	20858683	2,529,804	46,368	Soranzo N et al	European

Outcome data sources

Genetic data for NFALD were retrieved from the FinnGen online database ( https://r4.finngen.fi/ ), a public-private partnership combining genotype data from the Finnish Biospecimen Bank and digital health record data from the Finnish Health Registry. The FinnGen study included 894 NFALD patients and 217,898 non-NFALD patients. The International Classification of Diseases diagnosis code used to define NFALD is K76.0. The code change can be found at https://risteys.finngen.fi/ via ICD codes.

Statistical Analyses

The random effects Inverse-variance weighted (IVW) method is the main MR method used in our study³⁹. However, as traditional IVW methods are susceptible to imbalance-level pleiotropy, they may give biased estimates if there are invalid instrumental variables. Therefore, we used additional MR methods such as the weighted median(WM) method⁴⁰, MR-Egger regression analysis⁴¹ and MR-PRESSO method⁴² to further validate causal effects and performed multiple sensitivity analyses to check whether the selected SNPs violated the three key assumptions of the instrumental variables. Regardless of the kind of pleiotropy, the WM method believes that even if only half of the SNPs are genuine instrumental factors, it will still offer a reliable estimate⁴³. MR-Egger regression analysis can identify and control for bias due to pleiotropy⁴⁴. The method assumes that MR-Egger regression analysis can produce valid estimates even if all SNPs are invalid, provided that the association of individual SNPs with exposure is not related to the corresponding pleiotropy⁴⁵. In addition, the intercept term of the MR-Egger regression analysis can be used as a measure of unbalanced pleiotropy, with the intercept term being statistically significantly different from zero. If P < 0.05, this indicates the presence of pleiotropy. The results from the WM method and MR-Egger regression analysis are less precise than the IVW method due to lower statistical power, but if the estimates from all three methods are in the same direction, this suggests that the estimates from the MR analysis are reliable. The MR-PRESSO method⁴² is used to detect and correct for any outliers in SNPs that reflect potential pleiotropy bias. “MR-PRESSO global test” detects the presence of pleiotropy, the “MR-PRESSO outlier test” removes outliers from selected SNPs and provides a corrected result (which removes possible pleiotropy), and the “MR-PRESSO distortion test” checks for differences between pre-corrected and post-corrected results. Finally, a sensitivity analysis was carried out. Sensitivity analyses included tests of heterogeneity and leave-one-out. As a measure of overall multiplicity, we tested for heterogeneity using a modified Cochran's Q statistic and concluded that P < 0.05 indicated significant heterogeneity⁴³. Leave-one-out analysis was also performed to evaluate whether the MR estimate was driven or biased by a single SNP⁴¹.If the MR results estimated from other SNPs after excluding a particular SNP are significantly different from the total results, it means that the MR results are sensitive to that SNP. We used the odds ratio (OR) as the main effect indicator, including the 95% confidence interval (CI).

All analyses were conducted in R (version 4.1.2.) MR analyses were performed using “Mendelian Randomization”, “TwoSampleMR” and “MR-PRESSO” R packages^46,47.

Ethical issues

Additional ethical approval was not required, since this study was based on MR analysis and depended on summary GWASs data rather than individual-level data, and the databases used in this analysis had been published or shared before.

The causal associations between fasting insulin and NFALD

Based on the selection criteria of the IVs, we finally identified 38 SNPs associated with fasting insulin levels. The characteristics of the SNP loci associated with fasting insulin levels are listed in Supplementary materials Table S1. The causal effects between fasting insulin and NAFLD were estimated using the IVW method (random effects model), the MR-Egger method, and the WM method, as described in the methodology section. the results of the MR correlation analysis between NAFLD and Fasting insulin showed IVW (OR: 9.54; 95% CI 2.66–34.16; P = 0.0005), WM method (OR: 7.35; 95% CI 1.29–42.07; P = 0.0249), and MR-Egger regression method (OR: 10.90; 95% CI 0.18-673.92; P = 0.2636), where the IVW and WM methods showed P > 0.05 and the results were statistically significant (Table 2). Therefore, fasting insulin may be positively associated with NAFLD, the scatter plot as shown in Supplementary materials Fig. 1A. However, most of the estimates obtained by these two methods were slightly attenuated due to lower statistical efficacy, but the direction of the results was consistent with the main IVW method. The modified Cochran's Q statistic suggested no significant heterogeneity (P = 0.0870) (Table 3), and the intercept term of the MR-Egger method found no evidence of significant pleiotropy (P = 0.9469) (Table 4). The estimates obtained after removing each SNP in turn in the leave-one-out sensitivity test did not change significantly (i.e., the position of the entire line did not change significantly), suggesting that there were no individual SNP that unduly influenced the overall causal effect estimates and that the findings have a high degree of confidence, the forest plots for the MR leave-one-out analysis as shown in Supplementary materials Fig. 2A.

Table 2

Mendelian randomization results of the applied methods.
exposure	Number of SNPs	MR methods	Beta	SE	Pval	OR (95% CI)
Fasting insulin	38	MR Egger	2.3891	2.1041	0.2637	10.9(0.18-673.92)
		Weighted median	1.9953	0.8898	0.0249	7.35(1.29–42.07)
		Inverse variance weighted	2.2551	0.6510	0.0005	9.54(2.66–34.16)
Type 2 diabetes mellitus	114	MR Egger	0.0540	0.2382	0.8210	1.06(0.66–1.68)
		Weighted median	0.0435	0.1186	0.7135	1.04(0.83–1.32)
		Inverse variance weighted	0.1847	0.0941	0.0497	1.20(1.00-1.45)
Fasting glucose	68	MR Egger	-0.5961	0.7194	0.4103	0.55(0.13–2.26)
		Weighted median	-0.7220	0.4966	0.1460	0.49(0.18–1.29)
		Inverse variance weighted	0.1050	0.3814	0.7830	1.11(0.53–2.35)
HbA1c	11	MR Egger	0.4866	1.3697	0.7306	1.63(0.11–23.84)
		Weighted median	0.4411	0.6969	0.5268	1.55(0.40–6.09)
		Inverse variance weighted	0.0779	0.5402	0.8853	1.08(0.37–3.12)

Table 3

The results of heterogeneity test.
exposure	method	Q	df	Pval
Fasting insulin	Inverse variance weighted	48.0124	37	0.1061
Type 2 diabetes mellitus	Inverse variance weighted	185.8782	113	0.0000
Fasting glucose	Inverse variance weighted	101.2882	67	0.0043
HbA1C	Inverse variance weighted	2.9500	10	0.9826

Table 4

The results of pluripotency test.
exposure	Intercept	SE	Pval for intercept
Fasting insulin	-0.0023	0.0345	0.9469
Type 2 diabetes mellitus	0.0101	0.0169	0.5513
Fasting glucose	0.0187	0.0163	0.2549
HbA1c	-0.0156	0.0480	0.7529

The causal associations between T2DM and NFALD

We finally identified 114 SNPs associated with T2DM levels. The characteristics of the SNP loci associated with T2DM levels are listed in Supplementary materials Table S3.The results of the MR correlation analysis between NAFLD and T2DM showed IVW (OR: 1.20; 95% CI 1.00- 1.45; P = 0.0497), WM method (OR: 1.04; 95% CI 0.83–1.32; P = 0.0497), and MR-Egger regression method (OR: 1.06; 95% CI 0.66–1.68; P = 0.8210), where the IVW and WM methods showed P > 0.05 and the results were statistically significant(Table 2). Therefore, T2DM may be positively associated with NAFLD, the scatter plot as shown in Supplementary materials Figure S1B. Although the modified Cochran's Q statistic suggested significant heterogeneity (P < 0.05) in some of the analyses (Table 3), no significant outliers were found in the selected SNPs by the MR-PRESSO method, and the intercept term of the MR-Egger method found no evidence of significant pleiotropy (P = 0.5513) (Table 4). Finally, the forest plots for the MR leave-one-out analysis shows the result was unstable. (Supplementary materials Figure S2B).

The causal associations between fasting glucose and HbA1c level and NFALD

We identified 68 and 11 SNPs associated with fasting glucose and HbA1c level, respectively. The characteristics of the SNP loci associated with fasting glucose and HbA1c level are listed in Supplementary materials Table S3-S4. The results indicated that genetically predicted fasting glucose and HbA1c level both were not associated with NAFLD using the MR methods. The results of the MR correlation analysis between NAFLD and fasting glucose showed IVW method (OR = 1.11, 95% CI, 0.53–2.35, P = 0.157), WM method (OR = 0.49, 95% CI, 0.18–1.29, P = 0.1460),and MR-Egger method (OR = 0.55, 95% CI, 0.13–2.26, P = 0.4103).And the MR correlation analysis between NAFLD and HbA1c level showed IVW method (OR = 1.03, 95% CI, 0.37–3.12, P = 0.8853), WM method (OR = 1.55, 95% CI, 0.40–6.09, P = 0.5268), and MR-Egger method (OR = 1.63, 95% CI, 0.11–23.84, P = 0.7306) (Table 2), and the scatter plot as shown in Supplementary materials Figure S1C-D. Modified Cochran's Q statistic suggests significant heterogeneity in fasting glucose (P < 0.05), no significant heterogeneity was found in HbA1c level (P = 0.9826) (Table 3). In addition,the intercept term of the MR-Egger method found no evidence of significant pleiotropy in both fasting glucose (P = 0.2549) and HbA1c level (P = 0.7529) (Table 4). The leave-one-out sensitivity test showed in Supplementary materials Figure S2C-D. Figure 2. shows the OR plot for NAFLD and T2DM and the other 3 glycemic traits.

We have conducted one of the more comprehensive studies to date on the effects of multiple glycemic traits and the risk of NAFLD. We found that higher fasting insulin levels increased the risk of NAFLD. In addition, genetic liability to T2DM also appeared to increase the risk of NAFLD. However, no significant association was observed between fasting glucose and HbA1c and the risk of NAFLD. Sensitivity analyses with different statistical models showed almost similar results.

With the pandemic of western lifestyles, NAFLD is gradually becoming a metabolic disease that continues to increase in prevalence worldwide¹. The pathogenesis of NAFLD is not fully understood and epidemiological, biochemical, and therapeutic evidence suggests that the main pathophysiological mechanism in most patients with NAFLD is changes in lipid metabolism due to insulin resistance. Insulin resistance occurs when the tissue response to the hormone insulin is inadequate. In adipose tissue, the normal ability of insulin to inhibit lipolysis is impaired, resulting in an excessive release of free fatty acids which, like fat obtained from the diet, are subsequently absorbed by the liver. The inability of the tissue to respond appropriately to insulin results in abnormally high glucose levels, which leads to the production and secretion of insulin levels in excess of those required to maintain glucose levels in the blood⁴⁸. Both hyperglycaemia and hyperlipidaemia stimulate new fat production, which further increases the accumulation of liver fat^49,50. Under normal conditions, appropriate hepatic lipid levels would be maintained by increased lipoprotein secretion for outward transport; however, in patients with NAFLD, lipoprotein secretion is not sufficiently increased⁵¹. Conversely, the accumulation of specific lipids, such as diacylglycerols or ceramides, can impair hepatic insulin signaling, leading to a pathological increase in hepatic gluconeogenesis⁴⁸. A longitudinal study found an association between randomized plasma glucose levels and the risk of developing NAFLD, even in individuals without diabetes⁵². In recent years, several observational studies have reported an association between NAFLD and T2DM, and a large meta-analysis of observational studies from 20 countries estimated the global prevalence of NAFLD in patients with T2DM to be ~ 56%²¹. In addition, T2DM is an established risk factor for the rapid progression of NAFLD to NASH, cirrhosis, or HCC^21,53−55. Angulo et al⁵⁶. found that a baseline diagnosis of T2DM was positively associated with overall mortality and liver-related outcomes in patients with NAFLD. Circulating ALT and AST are markers of NAFLD and in a study using Mendelian randomization, by including up to 64,094 T2DM cases and 607,012 controls, it was found that increased circulating ALT and AST were associated with a higher risk of T2DM, and that higher fasting insulin was associated with higher circulating ALT. This study provides some support that insulin resistance and fasting insulin increase the risk of developing NAFLD⁵⁷. In addition, NAFLD also contributes to diabetes. NAFLD, especially NASH with varying degrees of hepatic fibrosis, exacerbates hepatic insulin resistance and leads to the release of several pro-inflammatory mediators and pro-diabetic hepatic factors (e.g. fetuin-A, fetuin-B, fibroblast growth factor 21, retinol-binding protein 4 and selenoprotein P) that may promote the development of diabetes mellitus^58,59.

In addition, this study found that higher fasting insulin levels increased the risk of NAFLD. This is consistent with the results of a study that measured intact insulin molecule levels by mass spectrometry to predict NAFLD. The study included 180 patients and found that insulin secretion during oral glucose tolerance test (OGTT) was increased approximately 4-fold in NAFLD patients without diabetes compared to all other subgroups (p = 0.008). This study concluded that accurate measurement of fasting intact insulin levels by mass spectrometry in non-diabetic patients is a simple, non-invasive strategy for predicting the onset of NAFLD⁶⁰.The threshold level at which HbA1c increases the risk of developing NAFLD has rarely been determined by longitudinal analysis. A study from Yoo et al⁶¹. showed for the first time an independent association between HbA1c variability and the risk of NAFLD in patients with diabetes, with elevated mean HbA1c levels increasing the risk of NAFLD, even in normal glucose tolerance (NGT) subjects. Furthermore, in the study by Alexopoulos et al⁶², the relationship between glycaemic control represented by HbA1c and liver injury was explored by including patients with biopsyproven NAFLD/NASH (n = 713). This study found that higher mean HbA1c levels were associated with higher levels of steatosis and hepatocyte ballooning. The data showed that every 1% increase in mean HbA1c was associated with a 15% increase in the odds of progression to the fibrotic phase (OR, 1.15; 95% ci, 1.01, 1.31). Glycaemic control was generally associated with increased severity of hepatocellular ballooning (OR: 1.74; 95% CI, 1.01–3.01; P = 0.048) and liver fibrosis significantly (OR: 4.59; 95% CI, 2.33–9.06; P < 0.01) compared with good glycaemic control. All the above studies show that blood glucose control predicts the severity of hepatocyte ballooning and liver fibrosis in NAFLD/NASH, so optimizing glycaemic control may be a means of improving the risk of NASH-related fibrosis progression.

Our study has several significant strengths. First, we have conducted one of the more comprehensive studies to date on the effects of multiple glycaemic traits and the risk of NAFLD. Secondly, using the MR design, our study can mimic a randomized controlled trial (RCT) in an observational setting. studies in MR using genetic variation as IV may avoid some of the limitations of observational studies (confounding, reverse causality, regression dilution bias) and RCTs (representativeness and feasibility issues) in making causal inferences⁶³. Finally, this study helps to provide a basis and new targets for screening for fasting insulin levels, fasting glucose and HB1Ac in patients with NAFLD. However, several limitations should be noted in our study. First, the present study did not find a significant association between fasting glucose and HbA1c and HAFLD, which is inconsistent with some of the current studies that have found an association with the risk of developing NAFLD in patients with fasting hyperglycaemia and high HbA1c levels. On the one hand, this may be related to the limited sample size of NAFLD selected for this study, and on the other hand, the interaction between genes and the environment cannot be assessed at the pooled level data to be assessed, with different lifestyle and metabolic factors and NAFLD varying by age or gender. We should be aware of the diversity of patients with NAFLD and in the future may consider subgroup analyses of confounding factors such as age, gender, and body mass index (BMI) in patients with NAFLD. Second, most participants in this study were European. While this largely avoids population heterogeneity, MR results should be further validated in other populations to verify generalizability when more GWAS data are available in future studies. Third, there was heterogeneity in some of the MR analyses performed in this study. Although the results obtained by several MR methods were similar, we cannot completely exclude the possibility of bias in the estimation of derived defects due to heterogeneity or pleiotropy defects in the genetic tools.

In conclusion, both NAFLD and T2DM can be treated with lifestyle interventions such as a low-calorie diet and moderate physical activity, or with medications that lower circulating insulin levels, to reduce the risk of developing glycaemia and related complications on the one hand, and to significantly improve liver steatosis or liver injury on the other. The results of this study therefore further emphasize the important role of glycaemic management in the prevention of NAFLD and help to provide a basis and new targets for the screening of fasting insulin levels, fasting glucose and additional glycaemic traits such as 2-hour fasting glucose and haemoglobin glycation index (HGI) in patients with NAFLD.

NAFLD: Non-alcoholic fatty liver disease; T2DM: Type 2 diabetes mellitus; NAFL: Non-alcoholic fatty liver.

NASH: Non-alcoholic steatohepatitis; CI: Confidence interval; GWAS: Genome-wide association study; HbA1c: Hemoglobin A1c; BMI: Body mass index; NGT: Normal glucose tolerance; OGTT: Oral glucose tolerance test; HGI: Haemoglobin glycation index; RCT: Randomized controlled trial; IVW: Inverse-variance weighted; MR: Mendelian randomization; SNP: Single nucleotide polymorphism; EA: Effect allele; OA: Other allele; SE: Standard error; NA: Not applicable.

Data availability statement

Our study was based on summarized GWAS data which all were from publicly available studies. Additional ethical approval was not required. This data can be found here: https://r4.finngen.fi/; https://gwas.mrcieu.ac.uk/. All codes and data sets used in the research are available at https://github.com/biostatYu/MRcode/.

We thank all the participants and investigators of the FinnGen consortium and related GWAS.

Author contributions

G.Z.W.: Conception; Methodology; Analyzing; Writing the manuscript;Z.J.X.Y.Z.J.:Methodology; Analyzing; K.D.M.,D.H.B.: Reviewing; Y.Y.A.,L.X.K.: Conception; Supervision; Revising the manuscript.

Conflict of interest

The authors declare that they have no competing interests.

Additional information

Supplementary material The Supplementary Material for this article can be found online.

Funding This work was supported by the National Science and Technology Major Project “13th Five-Year” (2018ZX10725505), the National Natural Science Foundation of China (NSFC Grant No. 82174341) and the New teacher Start-up Fund Project of the Beijing University of Chinese Medicine (2022-JYB-XJSJJ-050).

Correspondence and requests for materials should be addressed to Y.Y.A.

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and

institutional afliations.

Nawrot, M., Peschard, S., Lestavel, S. & Staels, B. Intestine-liver crosstalk in Type 2 Diabetes and non-alcoholic fatty liver disease. Metabolism: clinical and experimental 123, 154844, doi:10.1016/j.metabol.2021.154844 (2021).
Kechagias, S., Nasr, P., Blomdahl, J. & Ekstedt, M. Established and emerging factors affecting the progression of nonalcoholic fatty liver disease. Metabolism: clinical and experimental 111s, 154183, doi:10.1016/j.metabol.2020.154183 (2020).
Younossi, Z. M. et al. Global epidemiology of nonalcoholic fatty liver disease-Meta-analytic assessment of prevalence, incidence, and outcomes. Hepatology (Baltimore, Md.) 64, 73–84, doi:10.1002/hep.28431 (2016).
Younossi, Z. et al. Global Perspectives on Nonalcoholic Fatty Liver Disease and Nonalcoholic Steatohepatitis. Hepatology (Baltimore, Md.) 69, 2672–2682, doi:10.1002/hep.30251 (2019).
Powell, E. E., Wong, V. W. & Rinella, M. Non-alcoholic fatty liver disease. Lancet (London, England) 397, 2212–2224, doi:10.1016/s0140-6736(20)32511-3 (2021).
Ren, Z., Simons, P., Wesselius, A., Stehouwer, C. D. A. & Brouwers, M. Relationship between NAFLD and coronary artery disease: A Mendelian randomization study. Hepatology (Baltimore, Md.), doi:10.1002/hep.32534 (2022).
Ekstedt, M. et al. Fibrosis stage is the strongest predictor for disease-specific mortality in NAFLD after up to 33 years of follow-up. Hepatology (Baltimore, Md.) 61, 1547–1554, doi:10.1002/hep.27368 (2015).
Dulai, P. S. et al. Increased risk of mortality by fibrosis stage in nonalcoholic fatty liver disease: Systematic review and meta-analysis. Hepatology (Baltimore, Md.) 65, 1557–1565, doi:10.1002/hep.29085 (2017).
EASL-EASD-EASO Clinical Practice Guidelines for the management of non-alcoholic fatty liver disease. Journal of hepatology 64, 1388–1402, doi:10.1016/j.jhep.2015.11.004 (2016).
Feldstein, A. E. et al. The natural history of non-alcoholic fatty liver disease in children: a follow-up study for up to 20 years. Gut 58, 1538–1544, doi:10.1136/gut.2008.171280 (2009).
Younossi, Z. M. et al. Association of nonalcoholic fatty liver disease (NAFLD) with hepatocellular carcinoma (HCC) in the United States from 2004 to 2009. Hepatology (Baltimore, Md.) 62, 1723–1730, doi:10.1002/hep.28123 (2015).
Fouad, Y. et al. What's in a name? Renaming 'NAFLD' to 'MAFLD'. Liver international: official journal of the International Association for the Study of the Liver 40, 1254–1261, doi:10.1111/liv.14478 (2020).
Eslam, M., Sanyal, A. J. & George, J. MAFLD: A Consensus-Driven Proposed Nomenclature for Metabolic Associated Fatty Liver Disease. Gastroenterology 158, 1999–2014.e1991, doi:10.1053/j.gastro.2019.11.312 (2020).
Cole, J. B. & Florez, J. C. Genetics of diabetes mellitus and diabetes complications. Nature reviews. Nephrology 16, 377–390, doi:10.1038/s41581-020-0278-5 (2020).
Worldwide trends in diabetes since 1980: a pooled analysis of 751 population-based studies with 4.4 million participants. Lancet (London, England) 387, 1513–1530, doi:10.1016/s0140-6736(16)00618-8 (2016).
Scully, T., Ettela, A., LeRoith, D. & Gallagher, E. J. Obesity, Type 2 Diabetes, and Cancer Risk. Front Oncol 10, 615375, doi:10.3389/fonc.2020.615375 (2020).
Ahlqvist, E. et al. Novel subgroups of adult-onset diabetes and their association with outcomes: a data-driven cluster analysis of six variables. The lancet. Diabetes & endocrinology 6, 361–369, doi:10.1016/s2213-8587(18)30051-2 (2018).
Haas, J. T., Francque, S. & Staels, B. Pathophysiology and Mechanisms of Nonalcoholic Fatty Liver Disease. Annual review of physiology 78, 181–205, doi:10.1146/annurev-physiol-021115-105331 (2016).
Hossain, N. et al. Independent predictors of fibrosis in patients with nonalcoholic fatty liver disease. Clinical gastroenterology and hepatology: the official clinical practice journal of the American Gastroenterological Association 7, 1224–1229, 1229.e1221-1222, doi:10.1016/j.cgh.2009.06.007 (2009).
Fracanzani, A. L. et al. Risk of severe liver disease in nonalcoholic fatty liver disease with normal aminotransferase levels: a role for insulin resistance and diabetes. Hepatology (Baltimore, Md.) 48, 792–798, doi:10.1002/hep.22429 (2008).
Younossi, Z. M. et al. The global epidemiology of NAFLD and NASH in patients with type 2 diabetes: A systematic review and meta-analysis. Journal of hepatology 71, 793–801, doi:10.1016/j.jhep.2019.06.021 (2019).
Valenti, L., Bugianesi, E., Pajvani, U. & Targher, G. Nonalcoholic fatty liver disease: cause or consequence of type 2 diabetes? Liver international: official journal of the International Association for the Study of the Liver 36, 1563–1579, doi:10.1111/liv.13185 (2016).
Targher, G., Corey, K. E., Byrne, C. D. & Roden, M. The complex link between NAFLD and type 2 diabetes mellitus - mechanisms and treatments. Nature reviews. Gastroenterology & hepatology 18, 599–612, doi:10.1038/s41575-021-00448-y (2021).
Lawlor, D. A., Harbord, R. M., Sterne, J. A., Timpson, N. & Davey Smith, G. Mendelian randomization: using genes as instruments for making causal inferences in epidemiology. Statistics in medicine 27, 1133–1163, doi:10.1002/sim.3034 (2008).
Burgess, S., Timpson, N. J., Ebrahim, S. & Davey Smith, G. Mendelian randomization: where are we now and where are we going? International journal of epidemiology 44, 379–388, doi:10.1093/ije/dyv108 (2015).
Willer, C. J. et al. Discovery and refinement of loci associated with lipid levels. Nature genetics 45, 1274–1283, doi:10.1038/ng.2797 (2013).
Zheng, J. et al. Recent Developments in Mendelian Randomization Studies. Current epidemiology reports 4, 330–345, doi:10.1007/s40471-017-0128-6 (2017).
Emdin, C. A., Khera, A. V. & Kathiresan, S. Mendelian Randomization. Jama 318, 1925–1926, doi:10.1001/jama.2017.17219 (2017).
Bowden, J. & Holmes, M. V. Meta-analysis and Mendelian randomization: A review. Research synthesis methods 10, 486–496, doi:10.1002/jrsm.1346 (2019).
Goodarzi, M. O. et al. Fasting insulin reflects heterogeneous physiological processes: role of insulin clearance. American journal of physiology. Endocrinology and metabolism 301, E402-408, doi:10.1152/ajpendo.00013.2011 (2011).
Mortensen, H. B. & Christophersen, C. Glucosylation of human haemoglobin a in red blood cells studied in vitro. Kinetics of the formation and dissociation of haemoglobin A1c. Clinica chimica acta; international journal of clinical chemistry 134, 317–326, doi:10.1016/0009-8981(83)90370-4 (1983).
Dimas, A. S. et al. Impact of type 2 diabetes susceptibility variants on quantitative glycemic traits reveals mechanistic heterogeneity. Diabetes 63, 2158–2171, doi:10.2337/db13-0949 (2014).
Udler, M. S., McCarthy, M. I., Florez, J. C. & Mahajan, A. Genetic Risk Scores for Diabetes Diagnosis and Precision Medicine. Endocrine reviews 40, 1500–1520, doi:10.1210/er.2019-00088 (2019).
Chen, J. et al. The trans-ancestral genomic architecture of glycemic traits. Nature genetics 53, 840–860, doi:10.1038/s41588-021-00852-9 (2021).
Soranzo, N. et al. Common variants at 10 genomic loci influence hemoglobin A₁(C) levels via glycemic and nonglycemic pathways. Diabetes 59, 3229–3239, doi:10.2337/db10-0502 (2010).
Xue, A. et al. Genome-wide association analyses identify 143 risk variants and putative regulatory mechanisms for type 2 diabetes. Nature communications 9, 2941, doi:10.1038/s41467-018-04951-w (2018).
Clarke, L. et al. The 1000 Genomes Project: data management and community access. Nature methods 9, 459–462, doi:10.1038/nmeth.1974 (2012).
Hartwig, F. P., Davey Smith, G. & Bowden, J. Robust inference in summary data Mendelian randomization via the zero modal pleiotropy assumption. International journal of epidemiology 46, 1985–1998, doi:10.1093/ije/dyx102 (2017).
Bowden, J., Davey Smith, G., Haycock, P. C. & Burgess, S. Consistent Estimation in Mendelian Randomization with Some Invalid Instruments Using a Weighted Median Estimator. Genetic epidemiology 40, 304–314, doi:10.1002/gepi.21965 (2016).
Burgess, S. & Thompson, S. G. Interpreting findings from Mendelian randomization using the MR-Egger method. European journal of epidemiology 32, 377–389, doi:10.1007/s10654-017-0255-x (2017).
Verbanck, M., Chen, C. Y., Neale, B. & Do, R. Detection of widespread horizontal pleiotropy in causal relationships inferred from Mendelian randomization between complex traits and diseases. Nature genetics 50, 693–698, doi:10.1038/s41588-018-0099-7 (2018).
Mordi, I. R. et al. Type 2 Diabetes, Metabolic Traits, and Risk of Heart Failure: A Mendelian Randomization Study. Diabetes care 44, 1699–1705, doi:10.2337/dc20-2518 (2021).
Burgess, S., Bowden, J., Fall, T., Ingelsson, E. & Thompson, S. G. Sensitivity Analyses for Robust Causal Inference from Mendelian Randomization Analyses with Multiple Genetic Variants. Epidemiology (Cambridge, Mass.) 28, 30–42, doi:10.1097/ede.0000000000000559 (2017).
Bowden, J., Davey Smith, G. & Burgess, S. Mendelian randomization with invalid instruments: effect estimation and bias detection through Egger regression. International journal of epidemiology 44, 512–525, doi:10.1093/ije/dyv080 (2015).
Hemani, G. et al. The MR-Base platform supports systematic causal inference across the human phenome. Elife 7, doi:10.7554/eLife.34408 (2018).
Yavorska, O. O. & Burgess, S. MendelianRandomization: an R package for performing Mendelian randomization analyses using summarized data. International journal of epidemiology 46, 1734–1739, doi:10.1093/ije/dyx034 (2017).
Petersen, M. C. & Shulman, G. I. Mechanisms of Insulin Action and Insulin Resistance. Physiological reviews 98, 2133–2223, doi:10.1152/physrev.00063.2017 (2018).
Smith, G. I. et al. Insulin resistance drives hepatic de novo lipogenesis in nonalcoholic fatty liver disease. The Journal of clinical investigation 130, 1453–1460, doi:10.1172/jci134165 (2020).
Lambert, J. E., Ramos-Roman, M. A., Browning, J. D. & Parks, E. J. Increased de novo lipogenesis is a distinct characteristic of individuals with nonalcoholic fatty liver disease. Gastroenterology 146, 726–735, doi:10.1053/j.gastro.2013.11.049 (2014).
Fujita, K. et al. Dysfunctional very-low-density lipoprotein synthesis and release is a key factor in nonalcoholic steatohepatitis pathogenesis. Hepatology (Baltimore, Md.) 50, 772–780, doi:10.1002/hep.23094 (2009).
Pang, Y. et al. Diabetes, Plasma Glucose, and Incidence of Fatty Liver, Cirrhosis, and Liver Cancer: A Prospective Study of 0.5 Million People. Hepatology (Baltimore, Md.) 68, 1308–1318, doi:10.1002/hep.30083 (2018).
Chalasani, N. et al. The diagnosis and management of nonalcoholic fatty liver disease: Practice guidance from the American Association for the Study of Liver Diseases. Hepatology (Baltimore, Md.) 67, 328–357, doi:10.1002/hep.29367 (2018).
Wong, V. W. et al. Asia-Pacific Working Party on Non-alcoholic Fatty Liver Disease guidelines 2017-Part 1: Definition, risk factors and assessment. Journal of gastroenterology and hepatology 33, 70–85, doi:10.1111/jgh.13857 (2018).
Vilar-Gomez, E. et al. Fibrosis Severity as a Determinant of Cause-Specific Mortality in Patients With Advanced Nonalcoholic Fatty Liver Disease: A Multi-National Cohort Study. Gastroenterology 155, 443–457.e417, doi:10.1053/j.gastro.2018.04.034 (2018).
Angulo, P. et al. Liver Fibrosis, but No Other Histologic Features, Is Associated With Long-term Outcomes of Patients With Nonalcoholic Fatty Liver Disease. Gastroenterology 149, 389–397.e310, doi:10.1053/j.gastro.2015.04.043 (2015).
De Silva, N. M. G. et al. Liver Function and Risk of Type 2 Diabetes: Bidirectional Mendelian Randomization Study. Diabetes 68, 1681–1691, doi:10.2337/db18-1048 (2019).
Meex, R. C. R. & Watt, M. J. Hepatokines: linking nonalcoholic fatty liver disease and insulin resistance. Nature reviews. Endocrinology 13, 509–520, doi:10.1038/nrendo.2017.56 (2017).
Petersen, M. C. & Shulman, G. I. Roles of Diacylglycerols and Ceramides in Hepatic Insulin Resistance. Trends Pharmacol Sci 38, 649–665, doi:10.1016/j.tips.2017.04.004 (2017).
Bril, F. et al. Intact Fasting Insulin Identifies Nonalcoholic Fatty Liver Disease in Patients Without Diabetes. The Journal of clinical endocrinology and metabolism 106, e4360-e4371, doi:10.1210/clinem/dgab417 (2021).
Yoo, J. H. et al. Mean and visit-to-visit variability of glycated hemoglobin, and the risk of non-alcoholic fatty liver disease. Journal of diabetes investigation 12, 1252–1262, doi:10.1111/jdi.13455 (2021).
Alexopoulos, A. S. et al. Glycemic Control Predicts Severity of Hepatocyte Ballooning and Hepatic Fibrosis in Nonalcoholic Fatty Liver Disease. Hepatology (Baltimore, Md.) 74, 1220–1233, doi:10.1002/hep.31806 (2021).
Chen, X. et al. Depression and prostate cancer risk: A Mendelian randomization study. Cancer Med 9, 9160–9167, doi:10.1002/cam4.3493 (2020).

No competing interests reported.

Download PDF

Version 1

posted

You are reading this latest preprint version

Association of type 2 diabetes mellitus and other glycemic traits with the risk of NAFLD

Status:

Version 1

Abstract

Figures

Introduction

Methods And Materials

Genetic instrument selection

Outcome data sources

Statistical Analyses

Ethical issues

Results

Discussion

Abbreviations

Declarations

References

Additional Declarations

Supplementary Files

Status:

Version 1