Exploring the nexus between hypothyroidism and metabolic dysfunction-associated steatotic liver disease: a UK biobank cohort study

Abstract

Non-alcoholic fatty liver disease (NAFLD) is a chronic liver disease characterised by lipid deposition in liver cells. The global prevalence of NAFLD has significantly increased from 8.2% in 1990 to 30.2% in 2023, establishing it as a growing public health concern. In recent years, the name NAFLD has been replaced by metabolic dysfunction-associated fatty liver disease (MASLD). Numerous observational studies have investigated the potential association between hypothyroidism and MASLD; however, the findings remain inconsistent. In this context, a systematic analysis was conducted to examine the relationship between hypothyroidism and MASLD using data from a large cohort within the UK Biobank. Utilising prospective data from the UK Biobank, a Cox proportional hazards model supplemented with multiple sensitivity analyses was applied to investigate the association between the incidence of hypothyroidism and the onset of MASLD. In addition, stratified analyses and prognostic assessments were performed to assess potential effect modifiers. To explore the underlying mechanisms, mediation analyses were conducted, along with restricted cubic spline regression, to examine potential non-linear relationships and mediation effects within this association. The study found that after fully adjusting for multiple covariates, the risk of MASLD in hypothyroidism patients was 1.711 times that of non-hypothyroidism patients (95% CI 1.560–1.877, P < 0.001). Both subtypes of hypothyroidism, namely non-surgical related hypothyroidism (NSRH) and surgical related hypothyroidism (SRH), were associated with a markedly elevated risk of MASLD onset. For NSRH, the risk is increased by 1.710 times (95% CI 1.557–1.878, P < 0.001), and for SRH, the risk is increased by 1.763 times (95% CI 1.344–2.313, P < 0.001). Stratified analysis revealed an interaction effect between gender and BMI in relation to the risk of MASLD among individuals with NSRH. Mediation analysis revealed the critical role of specific biomarkers in elucidating the relationship between hypothyroidism and MASLD. Notably, red cell distribution width, C-reactive protein, HbA1c, and total protein were identified as significant mediators in this association. Patients with hypothyroidism exhibit a significantly increased risk of developing MASLD, with inflammatory and metabolic markers playing a mediating role in this association. These findings suggest that individuals with hypothyroidism, particularly those with elevated levels of inflammatory markers, may be at heightened risk for MASLD. As such, enhanced clinical monitoring of liver function in these patients is recommended to facilitate early detection and intervention.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-025-91221-7.

Introduction

Non-alcoholic fatty liver disease (NAFLD) is a chronic liver disease characterised by lipid deposition in liver cells. Its histological features closely resemble those of alcohol-induced liver damage, yet it occurs in individuals who do not consume, or only consume minimal amounts of, alcohol. NAFLD encompasses a broad spectrum of liver pathology, ranging from simple hepatic steatosis to non-alcoholic steatohepatitis (NASH), which may progress to advanced fibrosis, cirrhosis, and ultimately, chronic liver failure¹. In recent years, the name NAFLD has been replaced by metabolic dysfunction-associated fatty liver disease (MASLD). The results of epidemiological studies have shown that the global prevalence of MASLD has risen sharply from 8.2% in 1990 to 30.2% in 2023, becoming an emerging public health problem^2,3. At present, the prevention and diagnosis of MASLD has become a focus of medical research, and exploring the pathogenesis of MASLD and its related risk factors in depth is of considerable significance for improving clinical prognosis and developing effective preventive measures.

As research into MASLD advances, an increasing body of evidence highlights its strong association with metabolic dysfunctions, including insulin resistance (IR), central obesity, dyslipidaemia, hypertension, and hyperglycaemia. MASLD is increasingly recognised as a hepatic manifestation of metabolic syndrome (MS)⁴. Notably, thyroid gland also plays a key role in energy balance, lipid and carbohydrate metabolism, fat synthesis, and weight regulation⁵, and its hypofunction may be related to MASLD. Hypothyroidism is a disease characterised by a thyroid stimulating hormone (TSH) level above the reference range. Hypothyroidism is classified into overt and subclinical forms, depending on whether free thyroxine levels fall below the reference range⁶. While no consensus has been reached regarding the precise mechanisms, hypothyroidism-induced MASLD is increasingly recognised as a distinct pathological entity⁷.

A substantial number of observational studies have been conducted to investigate the relationship between hypothyroidism and MASLD; however, the results remain inconclusive, with conflicting findings across studies^8–13. The underlying mechanisms linking these two conditions remain poorly understood and require further exploration. Therefore, in the present study, the association between hypothyroidism and MASLD was systematically explored using a large population cohort from the UK Biobank. Further stratified, mediation, and nonlinear analyses were also performed to provide new population-based evidence for the impact of hypothyroidism on the risk of MASLD and explore the underlying potential mechanisms. These analyses offer a scientific foundation for the early screening and intervention of MASLD, promote advancements in related research, and contribute to improving both the quality of life and clinical outcomes for affected patients.

Methods

Data source

The UK Biobank is a large-scale, detailed prospective study that began in 2006, recruiting over 500,000 participants aged 40 to 69 over a period of four years. The study collects and continuously monitors a comprehensive range of phenotypic and multi-omics data, including survey responses, physical measurements, biological sample analyses, whole-genome genotyping, and a wide array of health-related outcomes, all obtained through ongoing longitudinal follow-up. Ethical approval for the UK Biobank was granted by the North West Multi-Centre Research Ethics Committee (REC reference: 11/NW/03820). All participants provided written informed consent for data collection, analysis, and linkage, with the study conducted in accordance with the principles of the Declaration of Helsinki¹⁴. Access to UK Biobank data was granted under application ID 84,347, and was approved by the Ethics Committee of the First Hospital of Jilin University. Our dataset includes data from 502,411 male and female participants. After excluding individuals with missing baseline data at recruitment (n = 115,269), the final study sample consisted of 387,142 participants. Additional details are shown in Fig. 1.

Fig. 1.

Study flowchart. MASLD metabolic dysfunction-associated steatotic liver disease, BMI body mass index, MET metabolic equivalent of task, TDI Townsend Deprivation Index.

Selection of variables, covariates, and outcomes

Peripheral venous blood samples were collected from all participants at baseline, with validation procedures conducted in accordance with the protocol of the UK Biobank study¹⁵. Data on all diagnoses, medication prescriptions, and participant deaths were recorded within a database for subsequent use and management. Participants were followed from the date of recruitment until the first occurrence of hypothyroidism diagnosis, MASLD diagnosis, study withdrawal, death, or the date of the last follow-up (whichever occurred first). Diagnoses of hypothyroidism and MASLD were identified using the 10th revision of the International Classification of Diseases (ICD-10) hospital admission codes. Specifically, “K76.0” or “K75.8” was used to identify MASLD, while “E03” or “E89.0” was used to identify hypothyroidism. Notably, “E03” denotes non-surgical hypothyroidism, whereas “E89.0” represents post-surgical hypothyroidism.

In the models developed for the present study, covariates were adjusted based on known potential risk factors for MASLD identified in previous research. The adjusted covariates included age, ethnicity (European, Asian, African, Chinese, mixed, and other), Townsend Deprivation Index (TDI), educational attainment metrics, body mass index (BMI), smoking status (current, former, never), alcohol consumption (current, former, never), and physical activity measured in metabolic equivalent of task (MET). Educational attainment indicators were derived from data on “Children and Young People” and “Adult Skills” representing educational disadvantage within a region in terms of “flow” and “stock”. The TDI, introduced in 1987, serves as a tool to measure socioeconomic deprivation, reflecting both material and social deprivation at the regional level. BMI was calculated from measured weight and height and classified as normal weight (BMI < 25.0 kg/m²), overweight (BMI 25.0–30.0 kg/m²), and obese (BMI ≥ 30.0 kg/m²)¹⁶. Further details on data extraction from the UK Biobank are available in Supplementary Table 1.

The calculation of the Dietary Inflammation Index (DII) utilized the Oxford WebQ, an online 24-hour dietary assessment tool that collects information on the consumption of 206 food items and 32 beverages over the past 24 h^17,18. Energy and nutrient intakes were calculated using the 5th edition of McCance and Widdowson’s The Composition of Foods. The data collected in the dietary assessment tool were based on the participant’s intake from the previous day, with questions such as: “Did you eat these yesterday?” or “How much of the following beverage did you consume yesterday?” In this study, the average of five 24-hour recalls was used (data were collected between April 2009 [first recall] and June 2012 [final recall], as outlined on the UK Biobank website: https://biobank.ndph.ox.ac.uk/showcase/field.cgi?id=26008). Individuals with unfeasible energy intakes, as determined by the Henry equation, were excluded. The method for calculating the DII has been previously described¹⁹.

The detailed information on genotyping, data aggregation, and quality control in the UK Biobank study has been previously described²⁰. Based on a genome-wide association study, a genetic risk score (GRS) for MASLD was constructed using 55 single nucleotide polymorphisms (SNPs) (Supplementary Table 2)²¹, and was weighted according to multivariable-adjusted risk estimates (β coefficients) for MASLD^22,23: .

The MRI imaging protocol and liver fat content analysis in the UK Biobank have been previously published²⁴. Briefly, liver MRI scans were performed using the LiverMultiScan© on a Siemens 1.5T MAGNETOM Aera, which is part of the abdominal imaging protocol of the UK Biobank. The MRI-derived proton density fat fraction (MRI-PDFF) has shown the highest accuracy for quantifying hepatic fat content compared to other non-invasive imaging modalities, and it is positively correlated with histological hepatic triglyceride content^25,26. The method used to derive MRI-PDFF has been described previously²⁷. The reference PDFF measurement of fat was calculated as the average PDFF of nine liver regions of interest, avoiding areas with heterogeneity, major blood vessels, or bile ducts during placement.

Statistical methods and software

All analyses were pre-specified before data inspection following the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines²⁸, as outlined in Supplementary Table 2. For baseline characteristics, continuous variables are presented as means (± standard deviation) and assessed for statistical differences using the Student’s t-test. Categorical variables are reported as counts (percentages) and compared using the Pearson chi-squared test. In the present study, a Cox proportional hazards model was employed to investigate the association between the incidence of hypothyroidism and the development of MASLD.

A three-model approach was employed to sequentially adjust for covariates. Model 1 adjusted for sex, age, and race; Model 2 further included adjustments for the TDI and education level, in addition to the variables in Model 1; and Model 3 added adjustments for smoking status, alcohol consumption, BMI, and MET. The proportional hazards (PH) assumption for the Cox models was tested using the Schoenfeld residual method, with all models meeting the PH assumption criteria. Sensitivity analyses excluded participants with a follow-up period of less than one year after the onset of hypothyroidism, regardless of MASLD development. Propensity score matching was also utilised, with each participant assigned a score based on baseline characteristics and a matched analysis conducted using a greedy nearest-neighbour algorithm to achieve a 1:1 match between MASLD and non- MASLD participants, ensuring a standardised mean difference of less than 10%. Additionally, subgroup analyses stratified by age, sex, race, education level, smoking status, alcohol consumption, and BMI were conducted to explore potential interactions between hypothyroidism and these variables.

A mediation analysis was also conducted. Specifically, a linear regression model was employed to investigate the associations between hypothyroidism onset and red cell distribution width, monocyte percentage, C-reactive protein, glycated haemoglobin (HbA1c), and total protein levels, adjusting for variables included in Model 3. Potential mediators and hypothyroidism were subsequently incorporated into a multivariable Cox proportional hazards model. The mediation effect was calculated as the product of the beta values from the first and second steps of the mediation analysis, with the confidence intervals and significance determined through a bootstrap resampling method conducted 1000 times. The mediation effect ratio was calculated by dividing the indirect effect by the total effect and multiplying by 100%. The indirect effect (IE) represents the impact of hypothyroidism onset on MASLD through mediator levels, while the direct effect (DE) reflects the impact of hypothyroidism on MASLD after controlling for mediator levels. A significant IE indicates the presence of a notable mediation effect. Finally, to evaluate potential nonlinear relationships between significant mediators and MASLD onset, a restricted cubic spline (RCS) analysis was performed.

All statistical analyses and visualisations were executed using R Project for Statistical Computing (version 4.2.3). All statistical tests were two-sided, with p-values < 0.05 considered statistically significant.

Results

Baseline characteristics

Table 1 presents the baseline characteristics of participants stratified by MASLD status. Among the study cohort, 53% were female, with a mean age of 56.3 ± 8.1 years. Within the MASLD group, the gender distribution was approximately balanced, and the mean age was 0.61 years older than that of non-MASLD participants. Compared to those without MASLD, participants with MASLD tended to have a higher level of education, greater TDI scores, higher rates of current and former smoking, lower rates of current alcohol consumption, higher BMI, and lower MET scores. In the hypothyroid group (see Supplementary Table 3), the proportion of females was notably higher, and the mean age was 2.68 years greater than that of non-hypothyroid participants. Hypothyroid individuals also had a higher level of education, increased TDI scores, a greater proportion of former smokers, lower rates of current alcohol consumption, higher BMI, and reduced MET scores compared to their non-hypothyroid counterparts. A total of 4,799 participants with MASLD and 21,979 participants with hypothyroidism were included in the study. Baseline data stratified by hypothyroidism and its subtypes are available in Supplementary Table 4.

Table 1.

Baseline characteristics.

Characteristic	Overall N = 387,142	Non-MASLD N = 382,343	MASLD N = 4,799	p-value
Sex, n (%)				< 0.001
Female	203,255 (53%)	200,865 (53%)	2,390 (50%)
Male	183,887 (47%)	181,478 (47%)	2,409 (50%)
Age, mean (SD) or n (%)	56.327 (8.113)	56.320 (8.116)	56.930 (7.820)	< 0.001
< 60	223,278 (58%)	220,587 (58%)	2,691 (56%)	0.024
≥ 60	163,864 (42%)	161,756 (42%)	2,108 (44%)
Race, n (%)				< 0.001
European	367,866 (95%)	363,369 (95%)	4,497 (94%)
Mixed-race	2,256 (0.6%)	2,229 (0.6%)	27 (0.6%)
Asian	6,968 (1.8%)	6,852 (1.8%)	116 (2.4%)
African	5,670 (1.5%)	5,587 (1.5%)	83 (1.7%)
Chinese	1,187 (0.3%)	1,173 (0.3%)	14 (0.3%)
Others	3,195 (0.8%)	3,133 (0.8%)	62 (1.3%)
Educational score, mean (SD)	13.614 (15.624)	13.544 (15.572)	19.167 (18.456)	< 0.001
TDI, mean (SD)	− 1.425 (3.008)	− 1.437 (3.001)	− 0.478 (3.373)	< 0.001
Smoking, n (%)				< 0.001
Never	212,343 (55%)	210,187 (55%)	2,156 (45%)
Previous	135,513 (35%)	133,578 (35%)	1,935 (40%)
Current	39,286 (10%)	38,578 (10%)	708 (15%)
Drinking, n (%)				< 0.001
Never	15,407 (4.0%)	15,118 (4.0%)	289 (6.0%)
Previous	13,238 (3.4%)	12,887 (3.4%)	351 (7.3%)
Current	358,497 (93%)	354,338 (93%)	4,159 (87%)
BMI, mean (SD) kg/m2 or n (%)	27.314 (4.704)	27.264 (4.671)	31.257 (5.559)	< 0.001
< 25	130,254 (34%)	129,774 (34%)	480 (10%)	< 0.001
25–30	165,998 (43%)	164,249 (43%)	1,749 (36%)
≥ 30	90,890 (23%)	88,320 (23%)	2,570 (54%)
MET, mean (SD)	2,654.387 (2,712.793)	2,657.845 (2,712.491)	2,378.885 (2,722.941)	< 0.001
Red blood cell distribution width, mean (SD)	13.477 (0.972)	13.475 (0.969)	13.644 (1.160)	< 0.001
Missing	14,752	14,552	200
Reticulocyte percentage, mean (SD)	1.345 (0.894)	1.341 (0.892)	1.656 (0.985)	< 0.001
Missing	21,300	21,024	276
Cholesterol, mean (SD)	5.688 (1.138)	5.691 (1.136)	5.448 (1.260)	< 0.001
Missing	21,666	21,345	321
Fasting blood glucose, mean (SD)	5.110 (1.218)	5.103 (1.199)	5.675 (2.179)	< 0.001
Missing	52,815	52,122	693
LDL-C, mean (SD)	3.553 (0.864)	3.555 (0.863)	3.411 (0.937)	< 0.001
Missing	22,346	22,006	340
Triglycerides, mean (SD)	1.742 (1.027)	1.735 (1.022)	2.273 (1.299)	< 0.001
Missing	21,922	21,588	334
Diabetes, n (%)	30,130 (7.8%)	28,332 (7.4%)	1,798 (37%)	< 0.001
Hypertension, n (%)	111,489 (29%)	108,325 (28%)	3,164 (66%)	< 0.001
Hyperlipidemia, n (%)	56,756 (15%)	54,830 (14%)	1,926 (40%)	< 0.001
Hypothyroidism, n (%)	21,979 (5.7%)	21,340 (5.6%)	639 (13%)	< 0.001

P-values were determined using different methods based on the type of variable. Specifically, Student’s t-test was used to assess statistical differences for continuous variables, while Pearson’s chi-square test was employed for categorical variables. Age and BMI are presented both as continuous data and grouped data.

MASLD non-alcoholic fatty liver disease, BMI body mass index, MET metabolic equivalent of task, LDL-C low-density lipoprotein cholesterol, SD standard deviation, TDI Townsend deprivation index.

Association between hypothyroidism and MASLD

The results of the Cox proportional hazards model (Table 2) indicate an increased risk of MASLD incidence among patients with hypothyroidism. Specifically, in Model 1, hypothyroid patients demonstrated a 2.144-fold increased risk of developing MASLD compared to non-hypothyroid patients (95% CI: 1.984–2.316, P < 0.001). The risk of MASLD in non-surgical related hypothyroidism (NSRH) patients was 2.162 times that of non-hypothyroid patients (95% CI: 1.999–2.338, P < 0.001), while surgical related hypothyroidism (SRH) patients showed a 1.896-fold risk (95% CI: 1.487–2.417, P < 0.001). In Model 2, the risk of MASLD for hypothyroid patients was 2.048 times higher compared to non-hypothyroid patients (95% CI: 1.894–2.215, P < 0.001). For non-surgical hypothyroidism (NSRH) patients, the risk was 2.061 times higher (95% CI: 1.904–2.232, P < 0.001), while for surgical hypothyroidism (SRH) patients, the risk was 1.870 times higher (95% CI: 1.467–2.385, P < 0.001). Model 3 demonstrated a 1.711-fold increased MASLD risk among hypothyroid patients compared to non-hypothyroid patients (95% CI: 1.560–1.877, P < 0.001); for NSRH patients, the risk was 1.710 times higher (95% CI: 1.557–1.878, P < 0.001), and for SRH patients, it was 1.763 times higher (95% CI: 1.344–2.313, P < 0.001).

Table 2.

Cox regression results and sensitivity analyses.

		Overall		Remove the group with disease onset within 1 year		After propensity matching
		HR 95% CI	P	HR 95% CI	P	HR 95% CI	P
Hypothyroidism	Model 1	2.144 (1.984–2.316)	< 0.001	2.570 (2.392–2.761)	< 0.001	1.295 (1.182–1.420)	< 0.001
	Model 2	2.048 (1.894–2.215)	< 0.001	2.592 (2.411–2.788)	< 0.001	1.293 (1.180–1.417)	< 0.001
	Model 3	1.711 (1.560–1.877)	< 0.001	2.605 (2.117–3.206)	< 0.001	1.289 (1.176–1.413)	< 0.001
NSRH	Model 1	2.162 (1.999–2.338)	< 0.001	2.451 (2.279–2.636)	< 0.001	1.309 (1.193–1.436)	< 0.001
	Model 2	2.061 (1.904–2.232)	< 0.001	2.467 (2.291–2.655)	< 0.001	1.306 (1.190–1.434)	< 0.001
	Model 3	1.710 (1.557–1.878)	< 0.001	2.512 (2.037–3.099)	< 0.001	1.303 (1.187–1.431)	< 0.001
SRH	Model 1	1.896 (1.487–2.417)	< 0.001	2.021 (1.854–2.204)	< 0.001	1.120 (0.853–1.469)	0.415
	Model 2	1.870 (1.467–2.385)	< 0.001	2.019 (1.850–2.205)	< 0.001	1.118 (0.852–1.466)	0.423
	Model 3	1.763 (1.344–2.313)	< 0.001	2.317 (1.828–2.937)	< 0.001	1.112 (0.848–1.460)	0.442

Model 1 was adjusted for sex, age, and race; Model 2 was additionally adjusted for the variables in Model 1, as well as for Townsend Deprivation Index and educational score; Model 3 included further adjustments for smoking, alcohol consumption, BMI, and MET.

HR hazard ratio, CI confidence interval, NSRH non-surgically related hypothyroidism, SRH surgically related hypothyroidism.

Stratified analysis of the association between hypothyroidism and MASLD

As illustrated in Fig. 2, after adjusting for covariates, stratified analysis revealed that NSRH promoted the incidence of MASLD across all age groups, genders, races, education levels, as well as regardless of smoking, alcohol consumption, and BMI. There was a significant interaction between NSRH and both gender and BMI in relation to MASLD occurrence (P-values = 0.005 and < 0.001, respectively). Among those with SRH, individuals who were female, consumed alcohol, or were overweight (BMI between 25 and 30) were more susceptible to developing MASLD, though no interaction effect was observed in these cases.

Fig. 2.

Stratified forest plot. All models were adjusted for sex, age, race, Townsend Deprivation Index, educational score, smoking, alcohol consumption, BMI, and MET. HR hazard ratio, CI confidence interval. The left panel shows the stratified analysis for the entire population, the middle panel for NSRH (non-surgically related hypothyroidism), and the right panel for SRH (surgically related hypothyroidism).

Mediation and nonlinear analyses of the association between hypothyroidism and MASLD

As illustrated in Fig. 3, mediation analysis indicated that the elevated risk of MASLD in patients with hypothyroidism was mediated through several factors. These included RDW (mediating effect proportion = 24.98%, P = 0.012), MONO% (mediating effect proportion = 56.63%, P = 0.018), CRP (mediating effect proportion = 70.35%, P = 0.039), HbA1c (mediating effect proportion = 81.50%, P = 0.062), and total protein (mediating effect proportion = 49.66%, P = 0.033). As shown in Fig. 3, these mediators exhibited nonlinear relationships in promoting MASLD. Specifically, RDW exhibited an initial steep trend within the range of 13–15%, CRP showed a rise within the 0 to 1 mg/L range, and HbA1c increased within the 30 to 50 mmol/mol range, followed by a stabilization. In contrast, total protein demonstrated a gradual increase initially, which then accelerated sharply once the value exceeded 70 g/L.

Fig. 3.

Mediation and nonlinear plots. *< 0.05, **< 0.01, ***< 0.001. MASLD metabolic dysfunction-associated steatotic liver disease, RBC red blood cells, MONO% monocyte percentage, CRP C-reactive protein, HbA1c glycated hemoglobin.

Sensitivity analyses

Sensitivity analyses showed that the increased risk of MASLD in hypothyroid patients, including those with subtypes of the condition, remained significant after excluding patients who developed MASLD within the first year of follow-up. In the propensity score-matched cohort, SRH patients exhibited a trend towards a higher risk of MASLD, but this did not reach statistical significance (in Model 3, HR 95% CI: 1.112 [0.848–1.460], P = 0.442). Baseline characteristics after matching are presented in Supplementary Table 5. In addition, as shown in Table 3, after further adjustments for hypertension, diabetes, and hyperlipidaemia, the association between hypothyroidism and increased MASLD risk remained significant. Third, we incorporated the significant interaction terms from Fig. 2 into the original model for further analysis, and found no significant changes in the results (Supplementary Table 6). Fourth, we performed an interaction analysis on the variables in the subgroup analysis, and found that sex and BMI have an interactive effect on the development of MASLD in NSRH patients (Supplementary Table 7). Fifth, to assess the potential confounding effects of dietary habits and genetic predisposition, we included the DII and the GRS for MASLD in model 3. The results remained unchanged (Supplementary Table 8). Finally, we reanalyzed the data using liver MRI-determined PDFF ≥ 5% as a replacement for the ICD-10 code to diagnose MASLD. The results showed that NSRH is more strongly associated with an increased risk of MASLD (in Model 9, HR 95% CI: 1.211[1.092– 1.343], P < 0.001). It is worth noting that the relationship between NSRH and MASLD also remains robust, but the relationship between SRH and MASLD is not statistically significant. The detail was shown in Supplementary Table 8.

Table 3.

Sensitivity analyses for additional adjusted models.

		HR 95% CI	P
Hypothyroidism	Model 4	1.423 (1.297–1.561)	< 0.001
	Model 5	1.484 (1.352–1.627)	< 0.001
	Model 6	1.477 (1.346–1.621)	< 0.001
NSRH	Model 4	1.423 (1.296–1.563)	< 0.001
	Model 5	1.480 (1.348–1.626)	< 0.001
	Model 6	1.474 (1.341–1.620)	< 0.001
SRH	Model 4	1.416 (1.079–1.858)	0.012
	Model 5	1.507 (1.149 –1.978)	0.003
	Model 6	1.483 (1.130–1.946)	0.004

All models were adjusted for sex, age, race, Townsend Deprivation Index, educational score, smoking, alcohol consumption, BMI, and MET. Model 4 was additionally adjusted for hypertension, Model 5 for diabetes, and Model 6 for hyperlipidemia.

HR hazard ratio, CI confidence interval, NSRH non-surgically related hypothyroidism, SRH surgically related hypothyroidism.

Discussion

The relationship between hypothyroidism and NASH has long been a subject of interest. As early as 2003, Liangpunsakul et al. conducted a study involving 174 NASH patients and 442 controls, which found that the prevalence of hypothyroidism in NASH patients was 15%, significantly higher than the 7.2% observed in the normal control group. This finding suggests a potential correlation between hypothyroidism and the development of NASH²⁹. In recent years, a prospective cohort study from Rotterdam found that higher free thyroxine levels were associated with a reduced risk of MASLD, and elevated TSH levels were associated with an increased risk of clinically relevant fibrosis in MASLD, further suggesting that hypothyroidism is associated with an increased risk of MASLD³⁰. Nonetheless, controversy about this issue remains. A recent cross-sectional retrospective population study conducted in Spain found that while hypothyroidism was associated with higher triglyceride levels and a higher prevalence of obesity, altered thyroid hormone levels were not linked to an increased prevalence of MASLD³¹. The findings of the present study provide strong evidence on this issue. It was found that patients with hypothyroidism had a significantly increased risk of developing MASLD. Our results are consistent with the finding of a recently published large meta-analysis that included 24 cross-sectional studies and a total of 764,244,39 adult individuals, which also assessed the association between primary hypothyroidism and MASLD³². However, the retrospective design of the cross-sectional studies cannot define the time line between the development of hypothyroidism and that of MASLD. Differently from them, we systematically explored the association between hypothyroidism and MASLD using a large, population-based prospective cohort from the UK Biobank for the first time. Moreover, we further analysed the differences between SHR and NSHR, and conducted a mediation analysis to understand the potential mechanism of the development of MASLD in patients with hypothyroidism.

In recent years, the widespread use of thyroid ultrasound screening has enabled more timely detection of thyroid tumours. As a result, the number of thyroid surgeries has increased rapidly, and SRH has gradually garnered more attention³³. Previous studies have largely focused on the overall impact of hypothyroidism without distinguishing between SRH and NSRH. However, there are important differences between the two. SRH results from the partial or complete removal of thyroid tissue, leading to a significant reduction in thyroid hormone levels, and its pathological mechanism is relatively well understood³⁴. The causes of NSRH are complex and diverse. It can result from a variety of factors, including autoimmune diseases, drug-induced effects, and nutritional deficiencies³⁵. The present study is the first to separately explore the relationship between SRH, NSRH and MASLD, filling a gap in the existing field. The results show that after adjusting for many confounding factors such as gender, age, race, TDI, education score, smoking, alcohol consumption, BMI and MET, the risk of MASLD in both SRH and NSRH patients was still significantly higher than that in non-hypothyroidism people, with risk ratios of 1.763 and 1.710, respectively. In hypothyroidism, pancreatic beta cells have fewer glucose-sensitive receptors, which causes insulin secretion to decrease, thereby reducing lipolysis in adipose tissue and increasing the transport of FFAs to liver tissue^36,37. Meanwhile, elevated TSH in hypothyroidism can also cause upregulation of SREBP-1c activity by stimulating TSH receptors on membrane of hepatocytes cells, impairing liver triglyceride metabolism and thus promoting the production of liver fat³⁸. The present findings re-emphasise the significant role of hypothyroidism (both surgically induced and primary) in the pathogenesis of MASLD.

Hierarchical analysis revealed a significant interaction between gender, BMI and non-surgically related hypothyroidism. High BMI further increases the risk of MASLD in patients with NSRH, suggesting that the interaction between obesity and hypothyroidism may play a key role in the development of metabolic syndrome. Interestingly, the finding that women are more likely to develop MASLD in the context of hypothyroidism warrants further investigation. While epidemiological studies have generally shown that the incidence of MASLD is lower in women than in men³⁹, hypothyroidism may alter this trend through various mechanisms. Specifically, metabolic disruptions caused by hypothyroidism could impact lipid metabolism and storage, thereby promoting the accumulation of fat in the liver⁴⁰. In addition, thyroid hormones play a significant role in lipid metabolism, and their deficiency may lead to a decrease in fatty acid oxidation and an increase in liver fat synthesis⁴¹. Moreover, the female-specific endocrine environment may also contribute significantly to this process. The interaction between thyroid hormones, oestrogen and fat metabolism in the liver is particularly complex. The effect of reduced thyroid hormones on hepatic fat synthesis may also interact with the estrogen metabolic pathway. Although oestrogen usually exerts a protective effect on MASLD by promoting lipid clearance and inhibiting excessive fat accumulation, reduced levels of thyroid hormones may disrupt this protective effect by affecting the metabolism and availability of oestrogen⁴². Therefore, the disruption of the delicate balance between thyroid hormones and oestrogen may be an important factor in women being more susceptible to MASLD when they have hypothyroidism. This highlights the importance of considering gender and BMI when assessing MASLD risk in clinical practice. It is worth noting that, no significant interaction was observed in patients with SRH, which may reflect the complex interplay of surgical factors on metabolic status. Compared to NSHR, SHR patients experience a sudden lack of thyroid hormones and a traumatic inflammatory response caused by surgery, so their metabolic disorders may be more severe³⁴. Especially in the early stages of hormone replacement therapy, patients may not quickly or completely restore normal thyroid hormone levels. Although long-term thyroid hormone replacement therapy helps restore some metabolic functions, differences in the degree and timing of recovery and adjustments to medication doses may cause patients to maintain a relatively low basal metabolic rate during treatment, and their metabolic status is susceptible to hormonal fluctuations, especially in female patients. Different from NSHR, the metabolic status of SHR patients is affected by the complex interaction of sharp changes in thyroid hormone levels, traumatic inflammatory responses and long-term replacement therapy, which may be the underlying mechanism for their different effects on the risk of MASLD.

In the present study, the association between hypothyroidism and MASLD was explored in-depth through mediation analysis and nonlinear research. The results of the mediation analysis show that factors such as RDW, MONO%, CRP, HbA1c and total protein played a significant mediating role in the increased risk of MASLD in patients with hypothyroidism. These findings suggest that the occurrence of MASLD in patients with hypothyroidism is not solely attributable to thyroid hormone deficiency, but rather results from a complex interaction of multiple factors, with inflammatory processes playing a crucial role. TSH has been shown to directly promote the pro-inflammatory activity of macrophages, inhibiting the efflux and exocytosis of cholesterol, which leads to the accumulation of cellular debris and the expansion of necrotic cores. Additionally, inflammatory mediators released by macrophages in plaques can exacerbate local tissue damage, further promoting inflammation and creating a vicious cycle⁴³. Moreover, due to the low availability of antioxidants in hypothyroidism, there is an increase in reactive oxygen species in the body and enhanced oxidative stress⁴⁴. Oxidative stress and chronic inflammation are also closely related to MASLD and play a crucial role in its progression to NASH and liver fibrosis^45,46. Compared with patients with simple hepatic steatosis, patients with NASH have higher levels of oxidised fatty acids and inflammatory cytokines^47,48. Conducting nonlinear analysis provides a deeper understanding of the mechanisms through which hypothyroidism influences the development of MASLD, as traditional linear regression may fail to capture the complexity of biological relationships. The study found a nonlinear relationship between RDW, CRP, HbA1c and total protein and the development of MASLD. Specifically, changes in these biomarkers appeared to have a minimal effect on MASLD risk below certain thresholds, but once these thresholds were surpassed, their impact on the risk of MASLD increased significantly. In the case of CRP, the nonlinear analysis found that below 1 mg/L, its mediating effect significantly increased with rising CRP levels, but above 1 mg/L, its mediating effect tends to stabilise. It indicates that mild inflammation plays an important role in promoting the development of MASLD in patients with hypothyroidism, and further inflammation may reach a biological “saturation” point, diminishing its impact on the progression of MASLD. This finding is of considerable clinical importance, highlighting that during the monitoring and management of hypothyroid patients, attention should be given to the dynamic changes in these biomarkers. Specifically, the rapid increase in CRP levels from low to high can serve as an early warning signal, and these biomarkers’ change can be used to identify MASLD high-risk patients at an early stage. In addition, these findings provide guidance for proactive early intervention strategies. For example, for identified high-risk patients, emphasis should be placed on the importance of adjusting the diet structure to reduce the dietary inflammatory index, appropriately increasing the amount of aerobic exercise, and using anti-inflammatory drugs if necessary to regulate the inflammatory response, thereby reducing their risk of developing MASLD and improving the prognosis.

The results of the present study reveal a significant association between hypothyroidism and MASLD, emphasising the influence of SRH and NSRH subtypes on the risk of MASLD development, as well as the interaction of individual parameters such as gender and BMI. These findings provide valuable insights for the clinical management of patients with hypothyroidism, suggesting that healthcare providers should take into account both the subtype of hypothyroidism and individual risk factors when assessing MASLD risk. Furthermore, the identification of mediating factors offers a promising avenue for future research, which could lead to more targeted prevention and intervention strategies. Timely identification and intervention of these mediating factors may enable more precise approaches to managing hypothyroidism and MASLD in clinical practice.

Although the present study strengthened the robustness of the findings through propensity score matching and sensitivity analyses, several limitations remain. First, the study utilised data from European populations, which may limit the generalisability of the results to other ethnic groups. Second, as the study was based on retrospective cohort data, while various covariates were adjusted for, the potential influence of residual confounding factors cannot be completely ruled out. Additionally, although the mediation analysis identified several key metabolic and inflammatory markers, the dynamic nature of these indicators meant that their temporal relationships could not be fully captured in this study. Future research should investigate the impact of individual differences on susceptibility to MASLD across diverse populations. A longitudinal design with a more temporal focus is necessary to better understand the role of these important mediating markers in the relationship between hypothyroidism and MASLD. Prospective studies with long-term follow-up and large multicentre cohorts will be crucial for further validating the non-linear relationships observed in the study and providing more precise guidance for clinical interventions.

Conclusion

In the present study, the association between hypothyroidism and MASLD was systematically explored through a retrospective analysis of large cohort data from the UK Biobank. The results revealed that patients with hypothyroidism have a significantly increased risk of developing MASLD. In addition, inflammatory and metabolic indicators, such as RDW, MONO%, CRP, HbA1c, and total protein, were found to play a mediating role in the pathway linking hypothyroidism to MASLD. These findings suggest that patients with hypothyroidism, particularly those with elevated levels of inflammatory markers, may be at heightened risk for MASLD. As such, clinical monitoring of liver function should be strengthened in this patient group.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Author contributions

H.W. designed and carried out the experiments and analyzed the data. C.Z. wrote the manuscript. P.W. supervised the study and revised the paper. All authors have read and agreed to the published version of the manuscript.

Data availability

The data that support the findings of this study are available from UK Biobank but restrictions apply to the availability of these data, which were used under license for the current study, and so are not publicly available, but are available from the corresponding author on reasonable request.

Declarations

Competing interests

The authors declare no competing interests.

Ethical approval

The UK Biobank project received approval from the North West Multi-centre Research Ethics Committee (approval number: 11/NW/0382), with informed consent obtained from all participants. This study was additionally approved by the Ethics Committee of the First Hospital of Jilin University.

Consent for publication

All authors agree to publish.