Metabolic dysfunction–associated steatotic liver disease and obstructive sleep apnea: A cohort analysis of prevalence and hepatic outcomes

Abstract

BACKGROUND

Metabolic dysfunction-associated steatotic liver disease (MASLD) is the most common chronic liver disease worldwide and often coexists with obstructive sleep apnea (OSA). Emerging evidence suggests that OSA may independently accelerate liver injury in MASLD. Despite this the epidemiology and clinical significance of OSA within the population with MASLD remain incompletely understood.

AIM

To estimate the prevalence of OSA among MASLD and determine the impact of OSA on hepatic complications.

METHODS

This was a large, multicenter, population-based retrospective cohort study conducted using the TriNetX Global Health Research Network from 2010 to 2024. We identified adults with MASLD with International Classification of Diseases, Tenth Revision codes and propensity score matched (PSM) them 1:1 with adults without MASLD. Patients with any record of prior liver disease other than MASLD or OSA (G47.3x) before the MASLD index date were excluded from the study.

RESULTS

After PSM there were 364283 pairs analyzed with balanced covariates. From 2010-2024 OSA incidence and prevalence in MASLD increased more than 400-fold (P < 0.001). MASLD was associated with 54% higher odds of OSA vs matched controls (odds ratio: 1.54), and patients with MASLD developed OSA earlier (median 189 days vs 358 days; P < 0.001). OSA markedly worsened hepatic outcomes. The odds of fibrosis were 1.78-fold higher at 1 year and 2.12-fold higher at 10 years while cirrhosis risk was 50%-55% higher at the 1-year and 10-year follow-ups (P < 0.001). Independent predictors of OSA in MASLD included male sex, obesity, older age, hypertension, and nicotine dependence.

CONCLUSION

MASLD was associated with a significantly higher risk of OSA, and coexisting OSA substantially amplified long-term hepatic complications. In addition, MASLD was associated with an earlier onset of OSA, highlighting the need for integrated hepatology-sleep medicine approaches and early identification and treatment of OSA in MASLD cohort may prevent the development of complications such as hepatic fibrosis and cirrhosis.

Key Words: Sleep-disordered breathing; Metabolic syndrome; TriNetX; Cirrhosis; Liver fibrosis; Intermittent hypoxia; Obstructive sleep apnea; Metabolic dysfunction-associated steatotic liver disease

Core Tip: Metabolic dysfunction-associated steatotic liver disease (MASLD) and obstructive sleep apnea (OSA) often coexist due to shared risk factors. Furthermore, OSA can lead to liver complications. In this large multicenter study, patients with MASLD had a significantly higher prevalence of OSA and an earlier onset of OSA. Coexisting OSA markedly increased the long-term risk of developing hepatic fibrosis and cirrhosis. These findings highlight the importance of routine OSA screening in patients with MASLD and support integrated metabolic and sleep-focused management to prevent hepatic disease progression.

INTRODUCTION

Metabolic dysfunction-associated steatotic liver disease (MASLD), formerly known as nonalcoholic fatty liver disease, has emerged as the most common chronic liver condition worldwide. Driven by the global epidemics of obesity and type 2 diabetes, MASLD is estimated to affect roughly one in four adults globally[1,2]. In the United States alone, up to 30% of adults (over 100 million people) meet criteria for MASLD[1]. This disorder represents the hepatic manifestation of metabolic syndrome, ranging from simple steatosis to its inflammatory form (metabolic dysfunction-associated steatohepatitis) that can progress to advanced fibrosis and cirrhosis[3-5]. Obstructive sleep apnea (OSA), a chronic sleep-related breathing disorder characterized by recurrent upper airway collapse during sleep resulting in intermittent hypoxemia and sleep fragmentation, affecting 7%-20% of the general adult population and up to 50%-70% of individuals with obesity[6].

Both MASLD and OSA frequently coexist within the spectrum of metabolic disease. Both disorders are closely linked to obesity, insulin resistance, systemic inflammation, and cardiometabolic dysfunction, and their prevalence has risen in parallel with the global increase in metabolic syndrome. MASLD represents the hepatic manifestation of metabolic dysfunction and is now the most common chronic liver disease worldwide, while OSA is the most common form of sleep-disordered breathing and a recognized contributor to adverse cardiometabolic outcomes[1,5,7,8]. A recent study reported the prevalence of MASLD in patients without OSA as 58.16%, in patients with mild-moderate OSA as 72.41%, and in patients with severe OSA as 78.00%[9].

While shared risk factors (such as obesity and insulin resistance) partly explain this overlap, mounting evidence suggests OSA can independently exacerbate liver injury in MASLD via multiple pathophysiological mechanisms[10]. The substantial overlap in their risk profiles suggests that MASLD and OSA may not merely coexist by chance but may interact in ways that influence disease onset, progression, and long-term complications. If left untreated OSA confers elevated risks of cardiovascular disease and all-cause mortality, underscoring the importance of its timely diagnosis and management[8].

Chronic intermittent hypoxia, the hallmark of OSA, triggers oxidative stress and inflammatory pathways in the liver, driving hepatic steatosis and promoting progression to steatohepatitis and fibrosis[11]. Clinical studies have accordingly observed that patients with MASLD with coexisting OSA tend to exhibit more severe liver disease and 2-to-3-fold greater odds of fibrosis than their counterparts without OSA[12]. Importantly, this impact of OSA on liver fibrosis appears to be independent of body mass index (BMI) and other metabolic confounders, implying that OSA confers a direct additive risk beyond the shared obesity alone[13].

Despite the growing recognition of an OSA-MASLD link, the epidemiology and clinical significance of OSA within the population with MASLD remain incompletely understood. The previous studies have been mostly limited to single-center cohorts or specific subgroups, leaving uncertainty about the true burden of OSA in MASLD and its contribution to adverse liver outcomes on a broader scale. To address this knowledge gap, the current study aimed to estimate the prevalence of OSA among MASLD and to determine the impact of OSA on liver-related outcomes.

MATERIALS AND METHODS

Study design and data source

This was a large, multicenter, population-based retrospective cohort study conducted using the TriNetX Global Health Research Network (Cambridge, MA, United States; https://trinetx.com). TriNetX is a federated platform that consolidates de-identified electronic health record (EHR) data from approximately 147 healthcare organizations, representing over 147 million unique patients worldwide. The network captures longitudinal, structured clinical data, including demographics, diagnoses, procedures, medications, laboratory parameters, and vital statistics, harmonized across participating institutions through standardized terminologies and quality-control algorithms.

For the analytic modules used in this study (compare-outcomes, incidence and prevalence, and Cox proportional hazards), queries were executed within the United States Collaborative Network subset of TriNetX, which comprises 65 contributing healthcare organizations depending on the specific analysis. All data within TriNetX are irreversibly de-identified in compliance with the Health Insurance Portability and Accountability Act Privacy Rule [45 CFR 164.514(b)(2)(i)(C)], and analyses are performed within the secure, access-restricted TriNetX environment[14]. Because only de-identified, aggregate-level data were accessed through the secure TriNetX web interface, this study was deemed exempt from institutional review board approval and informed consent. The study adhered to the ethical principles of the Declaration of Helsinki and followed both the Strengthening the Reporting of Observational Studies in Epidemiology and Reporting of studies conducted using Observational Routinely collected Data guidelines[15,16] (Supplementary Figure 1).

Study population and setting

The study cohort comprised adults (≥ 18 years) with a diagnosis of MASLD recorded between January 1, 2010 and December 31, 2024. The MASLD cohort was defined using validated International Classification of Diseases, Tenth Revision, Clinical Modification (ICD-10-CM) codes K75.81 (nonalcoholic steatohepatitis) and K76.0 (fatty liver, not elsewhere classified) (Supplementary Table 1). To enhance diagnostic precision, patients were required to have two or more MASLD codes (K75.81 or K76.0) recorded at least 90 days apart in any care setting. The index date was defined as the date of the first qualifying MASLD diagnosis that satisfied this case definition. To restrict the cohort to patients with MASLD rather than alternative chronic liver diseases, we excluded individuals with any record of viral hepatitis (B15-B19), alcoholic liver disease (K70.x), autoimmune hepatitis (K75.4), Wilson disease (E83.01), hemochromatosis (E83.1), alpha-1-antitrypsin deficiency (E88.01), or pregnancy-related liver disease. Patients with prior liver or multiorgan transplantation were also excluded as were those with incomplete demographic information (missing age, sex, race, or ethnicity). Patients with any record of OSA (G47.3x) before the MASLD index date were excluded from the incident OSA analyses to ensure correct temporal sequencing.

A comparator cohort of controls without MASLD was identified from the same network. It was restricted to adults without any record of MASLD codes throughout the study period. Patients were matched on age, sex, race/ethnicity, and major metabolic comorbidities to ensure comparability. Follow-up for both cohorts continued until December 31, 2024.

Exposure, outcomes, and definitions

The primary exposure of interest was the presence of MASLD as defined above. The primary outcome was the identification of OSA among individuals with MASLD, determined through ICD-10-CM code G47.3x. Procedural evidence of diagnostic polysomnography (CPT 95810 or 95811) was included to support OSA ascertainment when available although OSA was primarily defined using ICD-10-CM coding. Because patients with any OSA diagnosis before the index date were excluded from incidence analyses, all OSA events captured during follow-up represented new, incident OSA.

Secondary outcomes included: (1) Assessing the comparative risk of incident OSA in MASLD compared with a propensity score–matched non-MASLD cohort; (2) Determining hepatic outcomes, specifically hepatic fibrosis and cirrhosis, among patients with MASLD with OSA vs those without OSA; and (3) Identifying independent predictors of OSA within the MASLD cohort.

Hepatic fibrosis and cirrhosis were identified using predefined ICD-10-CM diagnostic codes recorded during inpatient or outpatient encounters. Hepatic fibrosis was defined using ICD-10-CM codes K74.0-K74.2, and cirrhosis was defined using ICD-10-CM codes K74.3-K74.6 (Supplementary Table 1). These outcomes were ascertained from structured EHR fields within the TriNetX platform and were time-stamped to permit longitudinal and time-to-event analyses[16].

Laboratory-based noninvasive fibrosis scores (such as fibrosis-4 index or nonalcoholic fatty liver disease fibrosis score), imaging modalities, and elastography-derived measurements were not incorporated into outcome definitions, as these data are not consistently available or standardized across all contributing healthcare organizations within the federated network. The use of ICD-based definitions aligns with prior large-scale epidemiologic studies using TriNetX and similar EHR databases.

Covariables and variables

Extracted covariables encompassed demographic and clinical parameters: Age (continuous and categorical by decade); sex; race/ethnicity (White, Black, Asian, Hispanic/Latino, or other); and BMI (kg/m²). Major comorbidities included type 2 diabetes mellitus (E11.x), hypertension (I10.x), hyperlipidemia (E78.x), chronic kidney disease (N18.x), heart failure (I50.x), nicotine dependence (F17.x), and alcohol abuse (F10.x). These variables were prespecified based on biological plausibility and their established association with both MASLD and OSA.

Bias mitigation and confounding control

Potential biases were mitigated through several methodological safeguards. To minimize confounding propensity score matched (PSM) was performed using a greedy nearest-neighbor algorithm with a caliper width of 0.1 pooled standard deviations. Matching variables included age, sex, race/ethnicity, BMI category (< 30, 30-35, 35-40, > 40 kg/m²), and all major metabolic comorbidities. Post-match balance was assessed by standardized mean differences (SMDs) with SMD < 0.1 considered acceptable. Restriction to incident OSA cases following MASLD diagnosis minimized reverse causality, and reliance on structured EHR fields reduced information bias.

Statistical analysis

Descriptive statistics summarized baseline characteristics as mean ± SD for continuous variables and n (%) for categorical variables. Prevalence and annual incidence proportions of OSA were computed using TriNetX’s Incidence and Prevalence analytics module. Temporal trends in OSA incidence and prevalence from 2010 to 2024 were assessed using the Mann-Kendall test, a nonparametric trend test robust to non-normality and missing data. Sen’s slope was used to estimate the annual percentage change in incidence and prevalence along with 95%CIs, providing an interpretable measure of the magnitude and direction of trends over time. Comparative OSA risk between matched cohorts was evaluated through risk ratios and risk differences. Kaplan-Meier survival curves and log-rank tests were applied to assess time-to-OSA diagnosis. Within the MASLD cohort Cox proportional hazards regression identified independent predictors of OSA, adjusting for all covariables. Model assumptions were verified using Schoenfeld residuals and log-log survival plots.

For hepatic outcomes odds ratios (ORs) for incident fibrosis and cirrhosis were calculated at 1 year, 3 years, 5 years, and 10 years. Sensitivity analyses excluded patients with extreme BMI values (≥ 60 kg/m²) or missing BMI data. When missingness was < 5%, median imputation was applied. Two-sided P < 0.05 was considered statistically significant. All analyses were conducted in R (v4.5.1) and cross-validated within the TriNetX Analytics (v6.2) environment.

Study size and power considerations

Given the population-based nature of the TriNetX network and the retrospective observational design, all eligible individuals meeting the inclusion and exclusion criteria were included in the analyses. No formal a priori sample size or power calculation was performed. Instead, statistical precision was evaluated by examining the width of CIs for key effect estimates. The very large sample size afforded by the network ensured narrow CIs and adequate power to detect clinically meaningful differences in OSA risk and hepatic outcomes.

Ethical and data-governance framework

TriNetX data governance adheres to the Health Insurance Portability and Accountability Act, the General Data Protection Regulation, and the United States’ Common Rule (45 CFR 46). All patient-level identifiers are removed or obfuscated prior to aggregation. To further mitigate re-identification risk, cell counts with fewer than 10 individuals were automatically rounded or suppressed. Because all analyses were conducted within a closed, federated environment and only de-identified, aggregated data leave the platform, no protected health information was accessed or exported by the investigators. Accordingly, this study was considered non–human subjects research and was exempt from institutional review board review and individual informed consent requirements.

RESULTS

Baseline characteristics

Overall, 403008 adults with MASLD and concomitant OSA and 1230088 patients with MASLD but without OSA were identified in the United States’ Collaborative Network between 2010 and 2024. After 1:1 PSM 364283 well-matched pairs were available for comparative analyses. Covariable balance was achieved across all baseline variables (SMD < 0.1). Although some variables retained statistically significant P-values after matching, all SMD were below 0.1, indicating satisfactory covariate balance; such discrepancies are expected in large datasets and reflect sample size–driven statistical sensitivity rather than clinically meaningful imbalance.

The mean age of the matched cohorts was approximately 55 years, and males comprised 51.7% of the MASLD + OSA group and 50.8% of the MASLD-only group. The racial distribution was similar (White: 76%, Hispanic: 9.5%, Black: 10%). The prevalence of hypertension (65.8% vs 66.3%), type 2 diabetes (35.5% vs 35.6%), and dyslipidemia (47.4% vs 47.5%) was comparable between groups, confirming successful matching (Table 1).

Table 1 Baseline characteristics before and after propensity score matching, n (%)/mean ± SD.

Characteristics	Before PSM				After PSM
	MASLD with OSA	MASLD without OSA	P value	SD	MASLD with OSA	MASLD without OSA	P value	SD
Total	403008	1222421			364283	364283		0.021
Age, years	55.6 ± 13.3	52.3 ± 15.3	< 0.001	0.229	55.3 ± 13.4	55.6 ± 13.9	< 0.001	0.014
Sex
Male	211174 (52.4)	573207 (46.9)	< 0.001	0.111	188288 (51.7)	185011 (50.8)	< 0.001	0.018
Female	191713 (47.6)	649214 (53.1)	< 0.001	0.111	175995 (48.3)	179272 (49.2)	< 0.001	0.018
Race/ethnicity
White	308039 (76.5)	866921 (70.9)	< 0.001	0.126	276739 (76.0)	278846 (76.5)	< 0.001	0.014
Hispanic or Latino	37325 (9.3)	178649 (14.6)	< 0.001	0.166	35576 (9.8)	3346 (9.2)	< 0.001	0.021
Not Hispanic or Latino	303603 (75.4)	834983 (68.3)	< 0.001	0.157	271870 (74.6)	275494 (74.6)	< 0.001	0.023
Black or African American	41535 (10.3)	110525 (9.0)	< 0.001	0.043	36865 (10.1)	37170 (10.2)	0.237	0.003
Asian	11546 (2.9)	60485 (4.9)	< 0.001	0.108	11286 (3.1)	10761 (3.0)	< 0.001	0.008
American Indian or Alaska native	2264 (0.6)	7879 (0.6)	< 0.001	0.011	2080 (0.6)	1756 (0.5)	< 0.001	0.012
Native Hawaiian or other Pacific Islander	2745 (0.7)	6543 (0.5)	< 0.001	0.019	2395 (0.7)	2326 (0.6)	< 0.001	0.002
BMI	37.5 ± 8.4	32.2 ± 7.2	< 0.001	0.671	36.9 ± 8.4	34.9 ± 7.6	< 0.001	0.026
BMI categories by kg/m2
30-35	152703 (37.9)	322388 (26.4)	< 0.001	0.249	135568 (37.2)	141766 (38.9)	< 0.001	0.035
35-40	144999 (36.0)	198160 (16.2)	< 0.001	0.462	116641 (32.0)	119757 (32.9)	< 0.001	0.018
> 40	133046 (33.0)	127239 (10.4)	< 0.001	0.57	99003 (27.2)	96295 (26.4)	< 0.001	0.017
Comorbidities
Diabetes mellitus type 2	157762 (39.2)	226779 (18.6)	< 0.001	0.467	129485 (35.5)	129766 (35.6)	0.492	0.002
Hypertensive diseases	277537 (68.9)	461297 (37.7)	< 0.001	0.657	239666 (65.8)	241545 (66.3)	< 0.001	0.011
Chronic kidney disease	48146 (12.0)	58410 (4.8)	< 0.001	0.261	37269 (10.2)	36589 (10.0)	0.008	0.006
Heart failure	47743 (11.9)	38808 (3.2)	< 0.001	0.334	32445 (8.9)	30314 (8.3)	< 0.001	0.021
Hyperlipidemia	205319 (51.0)	312041 (25.5)	< 0.001	0.542	172507 (47.4)	173106 (47.5)	0.160	0.003
Nicotine dependence	71558 (17.8)	147476 (12.1)	< 0.001	0.160	61281 (16.8)	60543 (16.6)	0.021	0.005
Alcohol abuse	14672 (3.6)	38466 (3.1)	< 0.001	2.027	12975 (3.6)	11750 (3.2)	< 0.001	0.019

Baseline demographic and clinical characteristics of adults with metabolic dysfunction-associated steatotic liver disease with and without obstructive sleep apnea before propensity score matching. Standardized differences quantify covariable imbalance with values > 0.1 indicating clinically meaningful differences. The P values represent group comparisons using χ² tests for categorical variables and t tests for continuous variables. Body mass index categories reflect World Health Organization obesity classifications. Note: After propensity score matching, covariate balance was primarily assessed using standardized mean differences, with values < 0.1 indicating adequate balance. Statistically significant P values for some variables may persist due to the very large sample size and do not necessarily indicate meaningful residual imbalance. BMI: Body mass index; MASLD: Metabolic dysfunction-associated steatotic liver disease; OSA: Obstructive sleep apnea; PSM: Propensity score matched.

Prevalence, incidence, and temporal trends of OSA in MASLD

Between 2010 and 2024 both the incidence and prevalence of OSA among patients with MASLD rose sharply and consistently across the TriNetX network. The incidence proportion increased from 0.03% to 3.24% while prevalence rose from 0.03% to 12.43%, representing more than a 400-fold expansion in overall burden (Figure 1). Mann-Kendall testing demonstrated a perfect monotonic upward trend for both metrics (τ = 1.00, P = 2.38 × 10^-7), indicating that the rise was continuous with no year-over-year reversals (Table 2). Sen’s slope analysis further quantified this trajectory, showing that OSA incidence increased by 0.20% per year (95%CI: 0.16-0.24), whereas prevalence increased even more rapidly at 0.76% per year (95%CI: 0.53-0.96, z = 5.15, P = 2.38 × 10^-7).

Figure 1 Temporal trends of obstructive sleep apnea in metabolic dysfunction-associated fatty liver disease (2010-2024). A: Incidence. Annual incidence proportion of obstructive sleep apnea (OSA) among patients with metabolic dysfunction-associated steatotic liver disease (MASLD) from 2010 to 2024. The purple line represents the observed incidence (%) for each year. The black dashed line represents the Sen’s slope estimate, demonstrating the monotonic upward trend identified by the Mann-Kendall test; B: Prevalence. Annual prevalence of OSA among individuals with MASLD from 2010 to 2024. The purple line displays observed prevalence (%) each year while the black dashed line shows the Sen’s slope trend line based on non-parametric trend estimation. Statistical significance of temporal trends was evaluated using the Mann–Kendall test. OSA: Obstructive sleep apnea; MASLD: Metabolic dysfunction-associated steatotic liver disease.

Table 2 Annual incidence and prevalence of obstructive sleep apnea among patients with metabolic dysfunction-associated steatotic liver disease (2010-2024).

Year	Incidence (%)	Incidence (n)	Prevalence (%)	Prevalence (n)
2010	0.029	13	0.029	13
2011	0.048	24	0.072	36
2012	0.113	63	0.177	99
2013	0.128	78	0.281	171
2014	0.203	136	0.450	303
2015	0.388	284	0.781	575
2016	0.678	535	1.376	1094
2017	0.824	682	2.024	1696
2018	1.097	935	2.889	2507
2019	1.266	1089	3.846	3398
2020	1.290	1098	4.806	4242
2021	1.700	1420	6.062	5297
2022	2.086	1666	7.702	6526
2023	2.519	1880	9.747	7857
2024	3.237	2151	12.425	9122

Yearly incidence and prevalence proportions of newly diagnosed obstructive sleep apnea (OSA) among patients with metabolic dysfunction-associated steatotic liver disease (MASLD) from 2010 to 2024. Incidence represents new OSA diagnoses per year divided by the MASLD population at risk; prevalence reflects all existing OSA cases each year among patients with MASLD. Both raw counts (n) and population percentages are shown. These trends were subsequently analyzed using Mann-Kendall monotonic trend testing and Sen’s slope estimates.

Comparative risk of OSA in MASLD vs non-MASLD

After matching 35235 patients per cohort were evaluated for comparative risk. Patients with MASLD demonstrated a 10.2% higher absolute risk of OSA (95%CI: 9.50%-10.93%), corresponding to 54% higher odds of OSA compared with the non-MASLD cohort (OR: 1.54, 95%CI: 1.50-1.59).

Kaplan–Meier analysis demonstrated a shorter time to OSA onset in MASLD (median 189 days vs 358 days in controls, log-rank P < 0.001), indicating that MASLD not only conferred a higher overall risk but also accelerated OSA manifestation (Figure 2). The separation of survival curves began early and widened steadily over time, indicating that MASLD is associated with both a higher risk and an accelerated onset of OSA. These findings remained robust across sensitivity analyses excluding extreme BMI outliers (≥ 60 kg/m²).

Figure 2 Kaplan-Meier curves comparing obstructive sleep apnea-free probability between metabolic dysfunction–associated steatotic liver disease and non-metabolic dysfunction-associated steatotic liver disease populations. Kaplan–Meier survival curves demonstrating obstructive sleep apnea (OSA)-free probability over time among patients with metabolic dysfunction-associated steatotic liver disease (MASLD) compared with a propensity-matched cohort without MASLD. Vertical dashed lines indicate key timepoints at which OSA-free probabilities diverge: 189 days for the MASLD cohort and 358 days for the non-MASLD cohort. The non-MASLD cohort showed consistently higher OSA-free probability across the follow-up period. The hazard ratio for the comparison was 1.152 (95%CI: 1.130-1.174) with a log-rank P < 0.0001, indicating a statistically significant difference between groups. OSA: Obstructive sleep apnea; MASLD: Metabolic dysfunction-associated steatotic liver disease.

Hepatic events in MASLD + OSA vs MASLD-alone

In a longitudinal follow-up extending to 10 years, patients with MASLD with coexisting OSA experienced substantially higher odds of hepatic disease progression (Table 3). The odds of developing hepatic fibrosis were 1.78-fold higher at 1 year (OR: 1.78, 95%CI: 1.65-1.92, P < 0.001) and increased to 2.12-fold at 10 years (OR: 2.12, 95%CI: 2.00-2.25). Similarly, the risk of cirrhosis was elevated by approximately 50%-55% at all follow-up points (1-year OR: 1.53, 10-year OR: 1.53, 95%CI: 1.47-1.60, P < 0.001) (Table 3).

Table 3 Primary outcomes comparing patients with metabolic dysfunction-associated steatotic liver disease and with and without obstructive sleep apnea after propensity score matching.

Cohort for primary outcomes	MASLD with OSA, n = 364283	MASLD without OSA, n = 364283	Odds ratio (95%CI)	P value
Hepatic fibrosis
1 year	1895	1069	1.777 (1.648-1.915)	< 0.0001
3 years	2796	1384	2.028 (1.901-2.163)	< 0.0001
5 years	3243	1544	2.110 (1.986-2.243)	< 0.0001
10 years	3675	1740	2.123 (2.005-2.248)	< 0.0001
Hepatic cirrhosis
1 year	2914	1911	1.529 (1.443-1.620)	< 0.0001
3 years	4158	2716	1.537 (1.464-1.514)	< 0.0001
5 years	4822	3149	1.538 (1.471-1.609)	< 0.0001
10 years	5506	3620	1.529 (1.466-1.595)	< 0.0001

Incidence of hepatic fibrosis and hepatic cirrhosis at the 1-year, 3-year, 5-year, and 10-year follow-up among propensity score-matched cohorts with metabolic dysfunction-associated steatotic liver disease with and without obstructive sleep apnea. Outcomes are presented as event counts and corresponding odds ratios with 95%CIs. All analyses reflect matched sample sizes of 364283 per group. Logistic regression models were used to calculate odds ratios at each time point. MASLD: Metabolic dysfunction-associated steatotic liver disease; OSA: Obstructive sleep apnea.

Predictors of OSA in MASLD

Multivariable Cox regression identified several independent predictors of OSA within the population with MASLD. Male sex [hazard ratio (HR): 1.45, 95%CI: 1.44-1.46, P < 0.001] and older age (HR: 1.01 per year, 95%CI: 1.007-1.008, P < 0.001) were the strongest demographic risk factors (Table 4). Among racial/ethnic groups, White (HR: 1.14) and Black (HR: 1.05) populations had slightly elevated risks compared with other categories while Hispanic/Latino ethnicity was associated with a modestly increased hazard (HR: 1.06, P < 0.001).

Table 4 Multivariable cox proportional hazards model assessing predictors of progression to hepatic disease.

Covariate	HR	95%CI	P value
Age at index	1.008	(1.007-1.008)	< 0.0001
Male	1.449	(1.436-1.463)	< 0.0001
Race/ethnicity
White	1.143	(1.123-1.163)	< 0.0001
Black or African American	1.054	(1.033-1.076)	< 0.0001
Asian	1.01	(0.981-1.041)	0.4884
American Indian or Alaska Native	0.968	(0.908-1.031)	0.3110
Native Hawaiian or Other Pacific Islander	1.074	(1.026-1.125)	0.0022
Hispanic or Latino	1.062	(1.040-1.084)	< 0.0001
Not Hispanic or Latino	0.939	(0.928-0.951)	< 0.0001
Comorbidities
Type 2 diabetes mellitus	0.989	(0.976-1.002)	0.0832
Hypertension	1.246	(1.232-1.26)	< 0.0001
Heart failure	1.009	(0.991-1.028)	0.3288
Chronic kidney disease	0.996	(0.977-1.016)	0.6856
Overweight and obesity	1.828	(1.809-1.847)	< 0.0001
Nicotine dependence	1.031	(1.017-1.046)	< 0.0001
Alcohol abuse	1.025	(0.995-1.055)	0.0983

Hazard ratios (HRs) with 95%CIs from a multivariable Cox proportional hazards model evaluating demographic factors, race/ethnicity, and comorbidities associated with progression to hepatic outcomes in patients with metabolic dysfunction-associated steatotic liver disease. Covariables include age, sex, race/ethnicity categories, cardiometabolic conditions, substance use, and obesity status. An HR > 1 indicates increased risk while an HR < 1 indicates reduced risk of hepatic progression. All variables were included in the fully adjusted model. HR: Hazard ratios.

Hypertension (HR: 1.25, 95%CI: 1.23-1.26, P < 0.001) and obesity (HR: 1.83, 95%CI: 1.81-1.85, P < 0.001) emerged as the most potent risk factors. Nicotine dependence modestly increased risk (HR: 1.03, 95%CI: 1.02-1.05, P < 0.001), whereas diabetes, chronic kidney disease, and alcohol abuse were not independently predictive after adjustment.

DISCUSSION

This 15-year, multicenter analysis of more than two million adults demonstrated that MASLD was strongly associated with a higher prevalence of OSA. Temporal trend analysis further revealed that patients with MASLD developed OSA earlier and more frequently than patients without MASLD, underscoring an accelerating burden of sleep-disordered breathing within this cohort. The co-occurrence of OSA and MASLD doubled the long-term risk of hepatic fibrosis and substantially increased cirrhosis rates. These associations persisted after adjusting for age, sex, BMI, diabetes, hypertension, and alcohol use, indicating that OSA acts as an independent amplifier of liver disease progression in MASLD. Male sex, hypertension, and obesity were the independent predictors of OSA in this population.

Most studies have demonstrated the increased risk of hepatic steatosis in patients with OSA. Only a handful of studies have reported the converse association: The risk of OSA in patients with MASLD. One study reported a significantly higher prevalence of OSA in patients with MASLD with fibrosis[17]. Similarly, apnea-hypopnea index was also found to be higher in patients with MASLD compared with patients without MASLD[18]. A recent population-based study reported a greater prevalence of sleep disorders in the MALSD cohort with transaminitis[19]. A recent large cohort study also reported 64.4% concurrence of MASLD and OSA in the cohort[20]. It also reported a linear relationship between the prevalence of OSA with grades of steatosis and fibrosis[21].

A study from a Korean cohort also reported greater odds of development of OSA in patients with hepatic steatosis compared with those without[21]. Our study adds to this growing literature and supports the increased risk of OSA in the MASLD cohort and is the first study to identify the predictors of OSA in the MASLD cohort. Also, our study provides strong evidence of significant worsening of hepatic events in patients with MASLD and OSA compared with those without OSA. Thus, findings of our study lend strong support to the growing interest in the association of OSA and MASLD and its impact on hepatic outcomes.

Over the last century, striking changes toward a high-fat and high-carb diet, increased consumption of refrigerated foods and red meats with the rising prevalence of obesity and sedentary lifestyle have resulted in dreaded effects on human metabolic health. The incidence and prevalence of diabetes, MASLD, obesity, and OSA have been on the rise at an alarming pace both in developed and developing nations. Typically, OSA, obesity, diabetes mellitus, and MASLD co-occur, likely attributable to metabolic and inflammation dysregulation[22-25].

Moreover, there is emerging evidence that MASLD is a multisystem ailment involving extrahepatic organs, potentially leading to additional effects[26]. The prevalence of OSA is estimated to be around 4% in the general population, increasing up to 40% in the population with diabetes and obesity[27]. There has been a growing interest in the interplay between OSA and MASLD. Mechanistic data suggest that intermittent hypoxia (the hallmark of OSA) can directly promote hepatic steatosis, inflammation, and fibrogenesis[28,29]. Conversely, MASLD itself may predispose to sleep-disordered breathing via fat deposition around the upper airway and general obesity[25]. Many studies in the past have concluded a higher prevalence of MASLD in patients with OSA, hypothesizing a link between hepatic injury and hypoxia[30,31]. A linear relationship independent of confounding factors between OSA severity and hepatic steatosis index has also been hypothesized[32]. The plausible explanation is intermittent hypoxia-mediated inhibition of the clearance of lipoproteins[33].

Despite multiple studies exploring the interplay of MASLD and OSA, many questions remain unexplored. Prior studies, limited by small sample sizes, cross-sectional designs, or special populations (such as bariatric surgery cohorts), have still been unable to estimate the true prevalence and impact of OSA in routine, diverse patient populations with MASLD. The current real-world study aimed to estimate the true prevalence of OSA in MALSD and its impact on liver-related outcomes and to identify the potential predictors of OSA. Therefore, these factors may guide clinicians to prioritize screening of OSA in patients with MASLD, especially in high clinical burden settings, and management of these intertwined conditions.

Our study had several strengths. Firstly, the study cohort captured a large, ethnically diverse population spanning two decades. This extensive dataset allowed robust estimation with adequate power for subgroup analysis. The use of PSM to balance demographic and metabolic covariables minimized confounding, and the application of time-to-event analyses added temporal validity to the observed associations. Furthermore, the inclusion of hepatic outcomes (fibrosis and cirrhosis) extended beyond prior cross-sectional studies, providing novel insights into the longitudinal hepatic burden of coexistent OSA in MASLD.

Our study carried inherent limitations characteristic of EHR-based research. Diagnoses were identified using ICD-10 codes rather than histologic or imaging confirmation, introducing the potential for misclassification. Although using at least two visits with similar ICD codes potentially mitigated this limitation. The current study did not provide data on detailed polysomnographic parameters such as apnea-hypopnea index or oxygen desaturation profiles, precluding severity stratification of OSA. Similarly, histologic staging or transient elastography data were unavailable, limiting granularity in assessing fibrosis progression. Although PSM and multivariable analyses adjusted for major confounders, residual confounding from unmeasured factors, such as physical activity, diet, or medication use, cannot be excluded. Despite these constraints the consistency of findings across sensitivity analyses supports the internal validity and robustness of the conclusions.

This large multicentric temporal analysis demonstrated high prevalence and predisposition of OSA in the MASLD population with earlier onset and increased hepatic morbidity of MASLD-related outcomes. Thus, early screening of OSA in the MASLD cohort may potentially improve the quality of life and may also mitigate the insulin resistance and systemic inflammation. Future interventional trials should evaluate whether such treatment attenuates hepatic steatosis and fibrosis progression in MASLD. Furthermore, integrating hepatologists, sleep specialists, and endocrinologists into a multidisciplinary model of care may optimize outcomes in this high-risk population.

Our study formulated a strong foundation for mechanistic studies elucidating the interaction of chronic intermittent hypoxia with hepatic lipid metabolism, mitochondrial function, and gut microbiota, and its potential effect on fibrosis. Finally, leveraging wearable technology and digital health platforms could refine population-level OSA surveillance and facilitate risk stratification within MASLD taken together, these findings highlight the critical importance of targeted OSA screening in patients with MASLD, particularly those with cardiometabolic risk factors, to enable earlier detection and potentially mitigate liver disease progression.

CONCLUSION

The current multicenter real-world study demonstrates that OSA is highly prevalent among individuals with MASLD and is independently associated with substantially worse hepatic outcomes of hepatic fibrosis and cirrhosis, even after adjustment for major metabolic comorbidities. In addition, male sex, obesity, and hypertension emerged as key predictors of OSA within the MASLD population. Collectively, these findings establish OSA as a clinically relevant and potentially modifiable comorbidity in MASLD and support the incorporation of routine OSA screening into the clinical assessment of patients with MASLD, particularly those with cardiometabolic risk factors. Early identification and treatment of OSA in MASLD cohort may prevent the development of complications such as hepatic fibrosis and cirrhosis.