Association between metabolic dysfunction-associated steatotic liver disease and pulmonary function: a population-based and two-sample mendelian randomization study

Abstract

Background

Hepatic steatosis and its related complications are risk factors for multiple respiratory diseases; however, the causal relationship between metabolic dysfunction-associated steatotic liver disease (MASLD) and pulmonary function remains controversial. We aimed to identify it using a national cohort and Mendelian randomization (MR).

Methods

We enrolled 30,442 participants from the 2007 to 2012 National Health and Nutrition Examination Survey. Demographics, pulmonary function indices (forced expiratory volume in 1 s [FEV1], forced vital capacity [FVC]), and variables used to calculate the liver fat score (LFS) were collected. A two-sample MR analysis employing the summary data of genome-wide association studies on MASLD and FEV1/FVC, chronic obstructive pulmonary disease (COPD), and asthma from the Finngen Biobank and Medical Research Council Integrative Epidemiology Unit was performed.

Results

A total of 3,462 participants, 1,335 of whom had MASLD (LFS > -0.640), were finally included in the study. The FEV1 (3,204.7 vs. 3,262.5 ml, P = 0.061), FVC (4,089.1 vs. 4,143.8 ml, P = 0.146), FEV1/FVC ratio (78.5% vs. 78.8%, P = 0.233), and FEV1/predicted FEV1 ratio (146.5% vs. 141.7%, P = 0.366) were not significantly different between people with MASLD and those without. Additionally, the MR analysis suggested no causal correlation between MASLD and FEV1/FVC (P = 0.817), MASLD and COPD (P = 0.407), and MASLD and asthma (P = 0.808). Reverse MR studies showed no causal relationships yet (all P > 0.05).

Conclusion

Our study provides convincing evidence that there is no causal association between MASLD and pulmonary function.

Background

Metabolic dysfunction-associated steatotic liver disease (MASLD) is the most prevalent chronic liver disease worldwide, affecting more than one-third of the general population. MASLD is characterized by hepatic fat accumulation induced by metabolic stress. Along with simple steatosis, MASLD may progress to steatohepatitis, fibrosis, or hepatocellular carcinoma, with varying extents of liver inflammation and necrosis. Furthermore, it has been increasingly recognized as a multisystemic disease with extrahepatic comorbidities, such as endocrine diseases (type 2 diabetes mellitus, hypothyroidism, and osteoporosis), cardiovascular diseases, chronic kidney disease, and respiratory diseases (obstructive sleep apnea syndrome [OSA], chronic obstructive pulmonary disease [COPD], asthma, and lung cancer) [1]. Therefore, early screening for comorbidities associated with MASLD may facilitate the management of these complications.

Existing evidence supports that MASLD is related to chronic respiratory diseases, including COPD [2] and asthma [3], since they both share harmful factors, such as an inflammatory state, increased visceral fat, oxidative stress, obesity, and insulin resistance [4]. In a previous meta-analysis summarizing six observational studies (five cross-sectional and one longitudinal) including 133,707 subjects (27.8% with MASLD), incident MASLD was independently associated with a decreased predicted forced expiratory volume in 1 s (FEV1) and forced vital capacity (FVC) [5]. However, in a more recent individual data meta-analysis including 2120 patients (75% steatosis determined by the Hepatic Steatosis Index), the severity of OSA/COPD was associated with an increased risk of fibrosis rather than steatosis presence [6]. Therefore, this contradicting evidence raises concerns about whether MASLD is causally associated with worsening pulmonary function and the causal association between MASLD and respiratory diseases remains unclear under a complicated bias setting.

We aimed to validate the relationship between MASLD and pulmonary function using a United States national cohort from the 2007–2012 National Health and Nutrition Examination Survey (NHANES). Furthermore, we aimed to demonstrate whether MASLD has a direct causal role in lung diseases, including both COPD and asthma, using Mendelian randomization (MR).

Methods

Source of data

A total of 30,442 participants from the NHANES 2007–2012 were enrolled in the study as spirometry was continuously conducted during that period. Exclusion criteria were as follows: (1) age less than 18 years old, (2) minority races, (3) pregnancy, (4) missing data of spirometry, alanine aminotransferase (ALT), platelet, body mass index (BMI), insulin, fasting blood glucose (FBG), blood pressure (BP), waist circumference, or aspartate aminotransferase (AST), (6) pulmonary diseases manifesting as cough or expectoration. The study was approved by the institutional review boards of The First Affiliated Hospital of Guangdong Pharmaceutical University, The First Affiliated Hospital of Sun Yat-sen university, and The Fifth People’s Hospital of Shunde District, Foshan City, respectively [2023018].

Data collection

Demographics data, and history of surgery, diabetes, and pulmonary disease as cough or expectoration lasting at least 3 months were routinely collected. Information on anthropometrics, BP (systolic [SBP], diastolic [DBP]), pulmonary function indices (FEV1, FVC), routine blood test, cholesterol (triglycerides [TG], total cholesterol [TC], high-density lipoprotein cholesterol [HDL-C], low-density lipoprotein cholesterol [LDL-C]), glucose (FBG, 2-h plasma glucose for oral glucose tolerance test [OGTT 2 h], insulin, Hemoglobin A1c [HbA1c]) and biochemistry data were carefully recorded. The predicted FEV1, FEV1/FVC ratio, FEV1/predicted FEV1 ratio were calculated according to the NHANES III Eq. [7].

We calculated liver fat score (LFS) for defining MASLD when LFS > -0.640: LFS = -2.89 + 1.18 × metabolic syndrome (MS) [present = 1, no = 0] + 0.45 × diabetes [present = 1, no = 0] + 0.15 × insulin (mU/L) + 0.04 × AST (U/L) − 0.94 × AST/ALT [8]. MS is commonly defined as at least three of the following traits: (1) waist circumference > 102 cm, (2) HDL-C ≤ 1.03 mmol/L, (3) TG ≥ 1.7 mmol/L, (4) SBP ≥ 130 mmHg or DBP ≥ 85 mmHg, (5) FBG ≥ 5.6 mmol/L [9]. Diabetes was diagnosed based on the history or if at least one trait appears: (1) FBG ≥ 7.0 mmol/L, (2) OGTT 2 h ≥ 11.1 mmol/L, (3) HbA1c ≥ 6.5% [10]. MASLD fibrosis score (NFS) was calculated to evaluate the fibrosis level of liver as follows: NFS = − 1.675 + (0.037 × age) + (0.094 × BMI) + (1.13 × diabetes [present = 1, no = 0]) + (0.99 × AST/ALT) – (0.013 × platelet [10⁹/L]) – (0.66 × albumin [g/dl]) [11]. NFS < − 1.455 represents a low risk for advanced fibrosis. Prediabetes was diagnosed based on one or more of the following features: (1) FBG ≥ 5.6 and ≤ 6.9 mmol/L, (2) OGTT 2 h ≥ 7.8 and ≤ 11.0 mmol/L, (3) HbA1c ≥ 5.7% and ≤ 6.4% [12]. Impaired glucose tolerance (IGT) was comprised of diabetes and prediabetes.

Instrumental variable selection

MR is an advanced epidemiological approach using single nucleotide polymorphisms (SNPs) from Genome-Wide Association Study (GWAS) data results to infer potential causal association between exposure and outcome [13]. The causal association was investigated between MASLD and pulmonary function using bi-directionally two-sample MR analysis. FEV1/FVC, COPD and asthma were used to stand for pulmonary function. The GWAS aggregated datasets of MASLD (finn-b-NAFLD), FEV1/FVC (ieu-b-4856), COPD (finn-b-COPD_MODE), and asthma (finn-b-ASTHMA_MODE) were collected from FinnGen consortium and Medical Research Council-Integrative Epidemiology Unit, and included 218,792, 51,396, 142,436, and 155,386 European individuals with 16,380,466, 9,703,538, 16,379,910, and 16,380,170 SNPs, respectively.

SNPs were selected as instrumental variables (IVs) if satisfying both criteria as follows: (1) at the genome-wide research significance (P < 5 × 10^− 7); (2) without linkage disequilibrium (r² threshold = 0.001, window size = 10000 kb) [14]. The F statistic of the IVs was calculated to quantify the strength of the IVs and evaluate if the IVs had weak bias. F > 10 indicates that IV deviations unlikely exist.

Mendelian randomization analysis

Inverse variance weighted (IVW), weighted mode, and MR Egger were utilized for MR analysis. IVW was used as the primary analytical method to estimate a causal association and others for sensitivity analysis. Meanwhile, MR Egger introduces an intercept term to measure horizontal pleiotropy. If value of intercept was close to 0 and P > 0.05, it implied that the IVs had no directional horizontal multiplicity. Moreover, Cochran’s Q statistics by IVW and MR Egger was performed to assess the heterogeneity across the IVs and MR pleiotropy residual sum and outlier test was used to remove IVs with horizontal pleiotropy. In addition, the leave-one-out method, sequentially removing one SNP and calculating the estimated value of the remaining SNPs which was compared with the total estimate, was exploited to evaluate the stability of the MR results.

Statistical analysis

The continuous variables were presented as means and standard deviations or medians and interquartile ranges. The categorical variables were expressed as numbers and percentages. Student’s t-test, chi-square test, or nonparametric test was utilized to compare between two groups, as appropriate. Linear correlation between pulmonary function indices and LFS was evaluated. IBM SPSS software, version 20 (IBM Corporation, Armonk, NY, USA) was applied to analyze the data. All images were generated by Adobe Illustrator 2021. MR analysis was performed using the “TwoSampleMR” and “MR-PRESSO” packages in R software version 3.6.3. Odds ratio (OR) and 95% confidence intervals (CI) were calculated. The P value for statistical significance was set at a two-sided value of 0.05.

Results

A total of 3,462 participants were eligible among the 30,442 participants from NHANES 2007–2012 based on inclusion criteria in the study (Fig. 1). Moreover, 1,335 participants were recognized as having MASLD (LFS > -0.640). Demographic characteristics, anthropometric and spirometric data, and blood test results were shown in Table 1. Compared to non-MASLD participants, participants with MASLD were older in age (45.3 vs. 42.2 years, P < 0.001), with a higher proportion of males (59.6% vs. 46.1%, P < 0.001), and Mexican Americans (26.7% vs. 19.6%, P < 0.001). As expected, the MASLD group showed more serious metabolic abnormalities, including a higher BMI, waist circumference, BP, TG, TC, LDL-C, FBG, OGTT 2 h, and HbA1c levels, in addition to lower HDL-C levels (all P < 0.001). The incidence rates of diabetes (12.4% vs. 2.4%), MS (62.6% vs. 3.2%), and IGT (73.3% vs. 40.2%) were significantly higher in patients with MASLD (all P < 0.001). Furthermore, participants with MASLD had a higher NFS score (-1.20 vs. -2.03, P < 0.001), indicating a higher risk for liver fibrosis. Pulmonary parameters, such as FEV1 (3,204.7 vs. 3,262.5 ml), FVC (4,089.1 vs. 4,143.8 ml), FEV1/FVC ratio (78.5% vs. 78.8%), and FEV1/predicted FEV1 ratio (146.5% vs. 141.7%), showed no significant differences between MASLD and non-MASLD individuals (all P > 0.05). In addition, both groups showed a similar rate of current smokers (P = 0.290).

Fig. 1

Schematic for participant enrollment in this study

Table 1 Characteristics of participants

The 1,335 MASLD participants were then categorized into subgroups according to NFS score, MASLD without fibrosis (NFS < − 1.455) group and MASLD with fibrosis (H-MASLD, NFS ≥ − 1.455) group. The MASLD with fibrosis group was much older (52.7 vs. 35.2 years, P < 0.001), and more likely to be non-Hispanic white (55.9% vs. 45.2%, P < 0.001) (Table 2). However, the MASLD without fibrosis group performed better in spirometry tests, exhibiting a much larger FVC (4,409.2 vs. 3,854.8 ml, P < 0.001), and FEV1/predicted FEV1 ratio (191.3% vs. 113.8%, P < 0.001).

Table 2 Characteristics of the MASLD participants with or without fibrosis

To further investigate the relationship between pulmonary function and MASLD, a linear association analysis was performed; no significant association was found between pulmonary function indices and LFS in the entire study population. In the MASLD group, a much weak correlation was only observed between FEV1/FVC ratio and LFS (r² = 0.079, P = 0.004). However, there was a greatly weak positive association between pulmonary function (FVC: r² = 0.082, P = 0.022; FEV1: r² = 0.133, P < 0.001; FEV1/FVC ratio: r² = 0.127, P < 0.001; FEV1/Predicted FEV1: r² = 0.138, P < 0.001) and LFS in the MASLD with fibrosis group (Table 3; Fig. 2). After removing the extreme values (three for the total population, three for the MASLD group, two for the MASLD with fibrosis group) and reexaming the association, the association between pulmonary function (FEV1/FVC ratio: r² = 0.711, P < 0.001; FEV1/Predicted FEV1: r² = 0.452, P < 0.001) and LFS in the MASLD with fibrosis group still exist and became stronger (Supplementary Table 2).In the MR analysis evaluating the causal association between MASLD and FEV1/FVC, four significant and independent SNPs (rs738408, rs78305825, rs75025168, rs8100204) related to MASLD were selected from the GWAS database. No evidence of heterogeneity between IVs was detected using Cochran’s Q test (IVW: P = 0.533, MR-Egger: P = 0.470). The F statistic ranged from 25.36 to 102.49. The IVW results demonstrated that MASLD had no significant effect on FEV1/FVC (OR = 1.002; 95% CI: 0.984–1.021; P = 0.817) (Table 4). Similar results were obtained in MR-Egger regression (P = 0.148) and weighted mode (P = 0.481). In addition, MR-Egger regression did not detect any evidence of horizontal pleiotropic (intercept = -0.017, P = 0.495). Furthermore, leave-one-out sensitivity analysis showed that the results were reliable as the estimated comprehensive effects of excluding any SNP were consistent (Fig. 3AB).

Table 3 Linear correlation between pulmonary function and LFS score in participants

Fig. 2

Linear association between pulmonary function and the liver fat score [LFS]. The forced vital capacity [FVC], forced expiratory volume in 1 s [FEV1], FEV1/FVC ratio, FEV1/predicted FEV1 ratio) and LFS in the whole participants (A–D), metabolic dysfunction-associated steatotic liver disease [MASLD] participants (E–H), and the MASLD with fibrosis group [H-MASLD] (I–L)

Table 4 The estimates of MR analysis for genetically predicting causal association

Fig. 3

The scatter plot and leave-one-out plot for Mendelian randomization [MR] analyses of causal associations. The plots between metabolic dysfunction-associated steatotic liver disease [MASLD] and FEV1/FVC(AB), chronic obstructive pulmonary disease [COPD] (CD), or asthma (EF)

Meanwhile, five SNPs (rs8100204, rs78305825, rs738408, rs72882094, rs75025168) related to MASLD without heterogeneity (Cochran’s Q: IVW: P = 0.562, MR-Egger: P = 0.472) and deviations (F > 10) were selected from the GWAS database. The IVW results showed that MASLD had no causal association with COPD (OR = 0.98; 95% CI: 0.93–1.03; P = 0.407) (Table 4). No evidence of horizontal pleiotropy was detected by MR-Egger regression (intercept = 0.044, P = 0.548) and the results were proven stable by leave-one-out analysis (Fig. 3CD). Moreover, no causal association between MASLD and asthma was demonstrated using MR analysis (IVW: OR = 1.00; 95% CI: 0.96–1.03; P = 0.808) with five selected SNPs (Table 4, Fig. 3EF).

Furthermore, reverse MR studies between FEV1/FVC and MASLD (IVW: OR = 1.548; 95% CI: 0.483–4.966; P = 0.463), COPD and MASLD (IVW: OR = 0.931; 95% CI: 0.721–1.203; P = 0.586), asthma and MASLD (OR = 0.821; 95% CI: 0.600–1.124; P = 0.218), were conducted, and no causal relationship was found (Table 4; Fig. 4).

Fig. 4

The scatter plot and leave-one-out plot for Mendelian randomization (MR) analyses of causal associations. The plots between FEV1/FVC and metabolic dysfunction-associated steatotic liver disease [MASLD] (AB), chronic obstructive pulmonary disease [COPD] and MASLD (CD), or asthma and MASLD (EF)

Discussion

The present study utilized the two-sample MR method to test the causal relationships among FEV1/FVC, two major chronic lung diseases, and MASLD. Our findings indicate that MASLD does not have a direct causal association with pulmonary function or related fibrosis. Neither of our bidirectional MR analyses revealed a significant causal relationship between FEV1/FVC, COPD or asthma, and MASLD, which is in contrast to the findings of previous studies.

Several epidemiological, observational, and experimental studies have reported that MASLD results in impaired lung function [15, 16]. From an investigation on 2,119 Korean men based on abdominal sonographic steatosis grading and pulmonary function, MASLD-associated hepatic steatosis severity independently determined a lower FVC (OR = 0.988, 95% CI: 0.978–0.998) and FEV (OR = 0.990, 95% CI: 0.981–0.998) than their non-steatotic counterparts after adjusting for age, BMI, smoking status, serum glucose, TC, hypertension, diabetes, TG, and HDL-C [15]. Another survey based on NHANES III (1988–1994) also concluded that steatosis severity by ultrasound examination increased with the gradual loss of lung function. Similar trends between MASLD and lung function have been observed in a population of middle-aged and elderly Chinese individuals [17]. Although these surveys showed positive results, owing to the inevitable confounding factors or other potential limitations of observational studies, such as study design, sample size, or regional variances, it remains inconclusive whether a relationship exists when the new criteria for MASLD are proposed. We further validated the association between MASLD and lung function using the same database as the NHANES III, but with an updated period of 2007–2012. Furthermore, we performed a subgroup analysis of fibrosis severity in MASLD instead of steatosis, the latter of which better represents disease progressive form or poor prognosis [18, 19]. However, we did not observe negative associations between LFS scores or NFS levels and the lung function parameters FEV1 and FVC. These findings contradict the conclusions of previous studies. In this study, 91.3% of individuals had at least one cardiometabolic risk factor, and ruling out alcohol-related fatty liver disease (ALD), metabolic (met)-ALD, cryptogenic steatotic liver disease (SLD), or other specific etiology SLD led to the distinction of individuals, with previous studies using the old diagnostic criteria of MASLD. Therefore, the possible impact of these new criteria may strengthen the importance of the number of metabolic perturbations in SLD development.

The association between decreased lung function and MASLD has been extensively reported; however, previous reports were mainly based on cross-sectional studies, overlooking any causality assumptions and lacking exploration of its associations with potential causes. Our results suggested that no associations between MASLD and FEV1/FVC, COPD, or asthma were retained in the MR analysis. Furthermore, the reverse MR analysis did not support a causal association between genetically predicted FEV1/FVC, COPD or asthma and MASLD. There are several explanations for the potential bidirectional link between MASLD and COPD. Patients with COPD or asthma often have low physical activity [20], and a sedentary lifestyle has been identified as an important factor in the development of MASLD [21]. Moreover, a systemic inflammatory status due to an increase in visceral fat and liver fat accumulation may induce lipotoxicity, oxidative stress, and inflammatory cytokine release, which contribute to the escalation of COPD/asthma. Our results might help to clarify that these associations should be evaluated more carefully. Genetically and environmentally, predisposing different extents in an individual with MASLD to suffer these respiratory diseases, and its possible varied comorbidities could lead to these observed associations.

The limitations of our study are as follows: (1) Our GWAS information was derived from studies of European ancestry, which may limit the generalizability of these conclusions to other races. (2) The GWAS of FEV1/FVC/COPD/asthma did not distinguish its subtypes; therefore, this might restrict the evaluation of their respective relationships. (3) Because of insufficient data, it might not be possible to evaluate the association between FEV1/FVC/COPD/asthma and the histological progression of MASLD. (4) The NFS scores indicated the usefulness of this tool for screening MASLD, but its grading steatosis severity value was low. Thus, magnetic resonance imaging derived proton density fat fraction or magnetic resonance spectroscopy-based liver fat content assessments remain necessary in the future. (5) The datasets for MR study are mostly from FinnGen consortium, which is not representative of European populations given its particular genetic characteristics from past historical events. (6) We adopted a less strict threshold of 5 × 10 − 7 to select SNPs instead of the genome-wide threshold (5 × 10 − 8) in the IVs selection during MR study.

Conclusion

This validation study with MR analysis indicated that MASLD had no direct causal association with decreased lung function, even with a high risk of fibrosis progression. Our findings provide genetic evidence for the absence of direct causal effects of FEV1/FVC/COPD/asthma on MASLD. Further research should focus on assessing the effects of MASLD subtypes on pulmonary function to validate our conclusions.