Assessing metabolic-dysfunction associated steatotic liver disease in patients with obstructive sleep apnea: a cross-sectional study

Abstract

Study objectives

To assess the prevalence and severity of metabolic-dysfunction associated steatotic liver disease (MASLD) in patients with obstructive sleep apnea (OSA) initiating continuous positive airway pressure (CPAP) therapy.

Methods

In this cross-sectional study, 72 patients with polysomnography-confirmed OSA (as per American Academy of Sleep Medicine criteria) presenting to the Al-Ain Sleep Clinic were prospectively enrolled. All participants underwent MASLD assessment via vibration-controlled transient elastography (FibroScan) to quantify hepatic steatosis (Controlled Attenuation Parameter, CAP) scores and liver fibrosis (Liver Stiffness Measurement, LSM). MASLD was defined as CAP ≥ 238 dB/m, advanced steatosis as CAP ≥ 290 dB/m while fibrosis was identified if LSM ≥ 7.5 kPa, and advanced fibrosis if LSM ≥ 14 kPa.

Results

The cohort was predominantly male (83.3%) with mean BMI 33.33 ± 6.83 kg/m² and mean apnea-hypopnea index (AHI) of 38.2 ± 21.4 events/hour. MASLD prevalence was 79.2% (57/72; 95% CI: 68.5–87.4%), with advanced steatosis (S3) present in 45.8% (33/72; 95% CI: 34.5–57.4%). Significant fibrosis (≥ F2) was detected in 20.8% (15/72; 95% CI: 12.4–31.7%) with F2 in 16.7% and F3 in 4.2% of patients. In the CPAP subgroup (n = 16, 22.2%), MASLD prevalence (75.0%) did not differ significantly from CPAP-naïve patients (80.4%, p = 0.65), though this analysis had limitations due to sample size, cross-sectional study design and lack of adherence data. Limitations include the cross-sectional design precluding causal inference, recruitment from a specialized sleep clinic introducing potential selection bias, and a predominantly male cohort (83.3%) limiting generalizability to female OSA patients.

Conclusions

MASLD prevalence in OSA patients substantially exceeds general population estimates, with 20.8% having significant fibrosis, underscoring the need to integrate non-invasive liver assessment into routine OSA management.

Graphical Abstract

Nearly 80% of the patients with obstructive sleep apnea also have metabolic liver disease, with 46% showing advanced steatosis and 21% with significant fibrosis; far exceeding general population rates. This study establishes systematic FibroScan screening for these patients as clinical priorities, providing evidence-based risk-stratification tools for identifying high-risk patients.

Introduction

Metabolic-dysfunction associated steatotic liver disease (MASLD), previously known as non-alcoholic fatty liver disease (NAFLD), represents one of the most prevalent chronic liver conditions globally, affecting approximately 25–30% of the general adult population and constituting a leading indication for liver transplantation worldwide [1–3]. MASLD encompasses a spectrum of hepatic pathology ranging from simple steatosis to metabolic-dysfunction associated steatohepatitis (MASH), progressive fibrosis, cirrhosis, and hepatocellular carcinoma [4]. The condition has been intrinsically linked to the components of metabolic syndrome, including obesity, type 2 diabetes mellitus, dyslipidemia, and insulin resistance [5].

MASLD, as defined by the multi-society Delphi consensus statement adopted in 2023, requires the presence of hepatic steatosis (≥ 5% of hepatocytes affected) with at least one cardiometabolic risk factor, including overweight/obesity (BMI ≥ 25 kg/m² in Caucasians or ≥ 23 kg/m² in Asians), type 2 diabetes mellitus, hypertension, dyslipidemia, or evidence of metabolic dysregulation [6]. This nomenclature change from NAFLD emphasizes the metabolic dysfunction central to disease pathogenesis rather than merely excluding alcohol consumption. Obstructive sleep apnea (OSA) is a common sleep-related breathing disorder characterized by recurrent upper airway collapse, intermittent hypoxemia, and sleep fragmentation [7]. Severe OSA has been associated with a range of metabolic disturbances and end-organ effects, including insulin resistance and dyslipidemia, which overlap with risk factors for MASLD [8, 9]. Recent evidence from longitudinal cohorts and Mendelian randomization studies suggests a strong, potentially causal, and bidirectional relationship between OSA and MASLD [9, 10]. While shared risk factors like obesity are central, OSA-specific pathophysiology, particularly chronic intermittent hypoxia (CIH), is increasingly recognized as an independent driver of liver injury. CIH is hypothesized to act as a “second hit” that promotes hepatic de novo lipogenesis and inflammation, thereby accelerating the progression from simple steatosis to metabolic dysfunction-associated steatohepatitis (MASH) and fibrosis [10].

Epidemiologic studies indicate a high co-prevalence of OSA and MASLD [11]. A 2018 meta-analysis of 9 studies (n = 2,272) by Jin et al., reported that OSA was associated with 2.99-fold increased odds of NAFLD (OR 2.99, 95% CI: 2.24–4.00) and 2.36-fold increased odds of disease severity [12]. Mendelian randomization analyses have explored potential causality with Zhang et al. (2024) demonstrating that genetic liability to sleep apnea was associated with MASLD risk (OR 1.50, 95% CI: 1.18–1.91), although this association was attenuated post-BMI adjustment [13]. A recent large-scale epidemiological study using NHANES 2017–2020 data confirmed significant association between OSA risk and NAFLD (OR 1.86, 95% CI: 1.63–2.11) after multivariate adjustment [14]. Despite this growing evidence base, data specifically examining MASLD burden using validated transient elastography in OSA patients initiating therapy remain limited.

We aimed to quantify MASLD prevalence and severity in a cohort of OSA patients at the point of CPAP initiation, and to explore associations between OSA-related factors and liver disease parameters.

Methods

Study design and participants

This cross-sectional observational study was conducted at a university hospital sleep medicine clinic. All adult patients (≥ 18 years) diagnosed with OSA based on comprehensive overnight polysomnography (according to American Academy of Sleep Medicine criteria) who were initiating auto-CPAP therapy were considered for inclusion. Patients with confirmed OSA who consented to undergo liver assessment with FibroScan were enrolled.

Inclusion and exclusion criteria

Inclusion criteria were a confirmed OSA diagnosis by polysomnography as per American Academy of Sleep Medicine (AASM), willingness to undergo FibroScan evaluation, and provision of informed consent. Exclusion criteria comprised factors that could independently affect liver status: significant alcohol consumption (> 14 units/week for men, > 7 units/week for women), known viral hepatitis (HBV or HCV) or other chronic liver diseases (autoimmune hepatitis, etc.), drug-induced liver injury, pregnancy, or presence of any implanted devices (e.g., pacemakers) contraindicating FibroScan. Patients unable to provide informed consent were also excluded.

FibroScan assessment protocol

All participants underwent a non-invasive liver assessment using vibration-controlled transient elastography (FibroScan^®, Echosens, Paris, France) performed by trained operators following the manufacturer’s guidelines. Patients fasted for a minimum of 3 h prior to the examination. FibroScan measurements provided two primary parameters:

• Controlled Attenuation Parameter (CAP): measured in decibels per meter (dB/m), which quantifies hepatic steatosis.

• Liver Stiffness Measurement (LSM): measured in kilopascals (kPa), which reflects liver fibrosis.

Valid FibroScan measurements required at least 10 successful acquisitions with an interquartile range (IQR) < 30% of the median LSM value and a success rate > 60%. The appropriate probe (M or XL) was selected based on each patient’s thoracic circumference and skin-to-liver capsule distance to ensure measurement accuracy.

MASLD classification criteria

MASLD diagnosis and steatosis/fibrosis grading were based on established FibroScan cutoff values for NAFLD/MASLD. Steatosis grades by CAP were defined as: S0 (normal): CAP < 238 dB/m; S1 (mild): 238–260 dB/m; S2 (moderate): 260–290 dB/m; S3 (severe): ≥290 dB/m. Fibrosis stages by LSM were defined as: F0–F1 (none to mild fibrosis): LSM 2.0–7.0 kPa; F2 (moderate fibrosis): 7.5–10.0 kPa; F3 (severe fibrosis): 10.0–14.0 kPa; F4 (cirrhosis): ≥14.0 kPa. Accordingly, MASLD was diagnosed in any patient with CAP ≥ 238 dB/m (at least mild steatosis). Advanced steatosis was defined as CAP ≥ 290 dB/m (S3). Significant fibrosis was defined as LSM ≥ 7.5 kPa (≥ F2), and advanced fibrosis as LSM ≥ 14 kPa (F4).

Data collection

To ensure comprehensive characterization, the following data were systematically collected for all participants. Demographic data included age and sex. Anthropometric measurements included height and weight, measured using a calibrated stadiometer and scale, from which Body Mass Index (BMI) was calculated as kg/m².

A detailed medical history was obtained through patient interview and a thorough review of the electronic health record. The presence of metabolic comorbidities was defined based on established clinical criteria: type 2 diabetes mellitus was defined by a prior physician diagnosis, use of any anti-diabetic medication, or a documented hemoglobin A1c ≥ 6.5%; hypertension was defined by a prior physician diagnosis, current use of anti-hypertensive medication, or an average of two seated blood pressure measurements with systolic pressure ≥ 140 mmHg or diastolic pressure ≥ 90 mmHg. Dyslipidemia was defined by a physician diagnosis or current use of lipid-lowering therapy, including statins.

All participants’ OSA diagnosis was based on an overnight inpatient polysomnography (PSG) study, scored by a certified technologist according to the AASM Manual. Key PSG parameters were extracted for correlational analysis, including the Apnea-Hypopnea Index (AHI, events/hour), the Oxygen Desaturation Index (ODI, events/hour defined by a ≥ 3% desaturation), mean nocturnal oxygen saturation (SpO₂), and nadir SpO₂. The AHI (events/hour), ODI (events/hour), mean nocturnal SpO₂, and minimum SpO₂ were recorded for all participants from their diagnostic polysomnography performed within 6 months prior to enrollment in the study. For the subset of patients on CPAP therapy at the time of enrollment (n = 16), we recorded the duration of therapy in months from the medical record.

Sample size calculation

An a priori sample size calculation was performed to ensure the study was adequately powered for its primary objective: to estimate the prevalence of MASLD in a clinical population of OSA patients. Based on regional prevalence data for NAFLD of approximately 32% in the Middle East [1], we set this as the expected prevalence (P) for our calculation. Using a standard formula for prevalence studies;

we determined the sample size required to estimate this prevalence with a desired precision (d) of ± 11% at a 95% confidence level (Z = 1.96).The calculation indicated a minimum required sample size of 70 participants.

Statistical analysis

Statistical Analysis Statistical analyses were performed using IBM SPSS Statistics for Windows, version 29.0 (IBM Corp., Armonk, NY, USA). All tests were two-tailed, with statistical significance set at p < 0.05. Continuous variables were tested for normality using the Shapiro-Wilk test and visual inspection of histograms and Q-Q plots. Statistical analyses were performed using standard software, with two-tailed significance set at p < 0.05. Continuous variables are presented as mean ± standard deviation (SD) if approximately normally distributed, or median (interquartile range [IQR]) for non-normally distributed variables. Group comparisons were conducted using Student’s t-test for normally distributed continuous variables, Mann-Whitney U test for non-normally distributed continuous variables, and χ² or Fisher’s exact test for categorical variables. LSM data demonstrated right-skewed distribution (Shapiro-Wilk p = 0.003) and were therefore analyzed using non-parametric methods, while CAP scores were approximately normally distributed (Shapiro-Wilk p = 0.14) and analyzed using parametric methods. Correlations between continuous variables were evaluated using Pearson correlation coefficients for normally distributed variables and Spearman’s rank correlation for non-normally distributed variables. Multivariate logistic regression analysis was performed to identify independent predictors of MASLD presence. Variables included in the model were selected based on clinical relevance and univariate associations (p < 0.20), including age, sex, BMI, diabetes mellitus, hypertension, dyslipidemia, AHI, and nadir SpO₂. Odds ratios (OR) with 95% confidence intervals (CI) were calculated. Model fit was assessed using the Hosmer-Lemeshow goodness-of-fit test and Nagelkerke R². Additionally, multiple linear regression was performed to identify independent predictors of CAP and LSM values, with standardized β coefficients reported. An exploratory Receiver Operating Characteristic (ROC) curve analysis was performed to evaluate the performance of BMI as a standalone predictor for MASLD.

Ethical considerations

The study was approved by United Arab Emirates University Human Medical Research Ethics Committee (UAEU.HREC) on 6th September 2023, with an approval number ERH_2023_3147 _08. All procedures were conducted in accordance with the Declaration of Helsinki. Written informed consent was obtained from all participants prior to enrollment and study procedures.

Results

Study population characteristics

A total of 72 OSA patients met the inclusion criteria and completed the FibroScan assessments. Figure 1 illustrates the patient flow diagram. The mean age was 45.2 ± 12.8 years, and the cohort was predominantly male (83.3%). The mean BMI was 33.33 ± 6.83 kg/m², corresponding to Class I obesity on average. At the time of liver assessment, 22.2% of patients (16/72) were using CPAP therapy, while the remaining 77.8% (56/72) had not yet initiated CPAP. Diabetes mellitus was present in 11.1% of patients, and 31.9% had a history of hypertension. Polysomnographic assessment demonstrated predominantly severe OSA in the cohort, with mean AHI of 38.2 ± 21.4 events/hour (range: 5.1–98.3). OSA severity distribution was: mild (AHI 5–14.9) in 15.3% (n = 11), moderate (AHI 15–29.9) in 26.4% (n = 19), and severe (AHI ≥ 30) in 58.3% (n = 42). Mean ODI was 32.7 ± 19.8 events/hour, mean nocturnal SpO₂ was 91.2 ± 3.4%, and mean nadir SpO₂ was 78.4 ± 9.2%. Table 1 summarizes the baseline demographic and clinical characteristics of the study population.

Fig. 1.

CONSORT flow diagram of patient recruitment and study procedures. Notes: Flow diagram showing the progression of patients through the study from initial assessment for eligibility (n = 84) through final analysis (n = 72). Exclusions included: not meeting OSA criteria (n = 7) and declined participation (n = 5). The final cohort was divided into CPAP users (n = 16) and non-CPAP users (n = 56), all of whom completed FibroScan assessment

Table 1.

Baseline Characteristics of OSA Patients Stratified by MASLD Status

Characteristic	All Patients (n = 72)	MASLD Present (n = 57)	MASLD Absent (n = 15)	p-value
Demographics
Age (years), mean ± SD	45.2 ± 12.8	45.8 ± 12.5	43.4 ± 14.2	0.52
Male sex, n (%)	60 (83.3)	48 (84.2)	12 (80.0)	0.69
BMI (kg/m²), mean ± SD	33.33 ± 6.83	34.2 ± 6.9	30.1 ± 5.8	0.025*
OSA Treatment
CPAP therapy, n (%)	16 (22.2)	12 (21.1)	4 (26.7)	0.65
No CPAP therapy, n (%)	56 (77.8)	45 (78.9)	11 (73.3)	0.65
Comorbidities
Diabetes mellitus, n (%)	8 (11.1)	8 (14.0)	0 (0.0)	0.12
Hypertension, n (%)	23 (31.9)	20 (35.1)	3 (20.0)	0.25
Dyslipidemia, n (%)	18 (25.0)	16 (28.1)	2 (13.3)	0.22
FibroScan Results
CAP score (dB/m), mean ± SD	282.24 ± 69.27	309.8 ± 51.2	171.3 ± 28.7	< 0.001***
LSM (kPa), mean ± SD	6.36 ± 3.32	6.8 ± 3.5	4.7 ± 1.8	0.02*
OSA parameters
AHI (events/hr), mean ± SD	38.2 ± 21.4	39.8 ± 22.1	32.3 ± 17.9	0.24
ODI (events/hr), mean ± SD	32.7 ± 19.8	34.2 ± 20.4	27.1 ± 16.3	0.21
Mean nocturnal SpO₂ (%), mean ± SD	91.2 ± 3.4	90.8 ± 3.5	92.7 ± 2.8	0.05
Nadir SpO₂ (%), mean ± SD	78.4 ± 9.2	77.2 ± 9.5	83.0 ± 6.8	0.03*

*Continuous variables presented as mean ± standard deviation. Categorical variables presented as number (percentage). Statistical comparisons performed using Student’s t-test or Mann-Whitney U test for continuous variables and χ² test or Fisher’s exact test for categorical variables. *p < 0.05, **p < 0.01, **p < 0.001

BMI body mass index, CAP controlled attenuation parameter, CPAP continuous positive airway pressure, LSM liver stiffness measurement, MASLD metabolic-dysfunction associated steatotic liver disease, OSA obstructive sleep apnea, SD standard deviation, AHI Apnea-Hypopnea Index, ODI Oxygen Desaturation Index, SpO2 oxygen saturation.

MASLD prevalence and severity

MASLD was present in 57 of 72 patients, yielding an overall prevalence of 79.2% (95% CI: 68.5–87.4%). Nearly half of the entire OSA cohort (45.8%, 33/72; 95% CI: 34.5–57.4%) exhibited advanced steatosis (grade S3). This subset of patients with severe hepatic steatosis constituted 57.9% of those with MASLD. Figure 2 illustrates the distribution of steatosis grades and overall MASLD prevalence in the study population. Among patients with MASLD, the majority had at least moderate steatosis: mild steatosis (S1) was observed in 9.7%, moderate (S2) in 23.6%, and severe (S3) in 45.8% of all patients.

Fig. 2.

MASLD frequency and steatosis grade distribution in OSA patients. Notes: Pie chart showing the distribution of hepatic steatosis grades in 72 OSA patients assessed by FibroScan. MASLD was present in 79.2% of patients (57/72), with advanced steatosis (S3) found in 45.8% (33/72). Normal liver (S0): CAP < 238 dB/m; Mild steatosis (S1): 238–260 dB/m; Moderate steatosis (S2): 260–290 dB/m; Advanced steatosis (S3): ≥290 dB/m. CAP, controlled attenuation parameter; MASLD, metabolic-dysfunction associated steatotic liver disease; OSA, obstructive sleep apnea

Liver fibrosis assessment

Significant fibrosis (≥ F2 by FibroScan) was present in 15 patients, corresponding to 20.8% of the cohort (95% CI: 12.4–31.7%). Most patients with significant fibrosis had moderate fibrosis (F2), while three patients (4.2%) demonstrated severe fibrosis (F3). No patients met the FibroScan criteria for cirrhosis (F4) in this cohort. Table 2 shows the distribution of liver fibrosis stages among the study population. Figure 3 illustrates the relationship between steatosis severity and fibrosis. LSM values demonstrated significant right-skewed distribution (Shapiro-Wilk p = 0.003, skewness = 1.42) and were analyzed using non-parametric methods. Median LSM was 5.4 kPa (IQR: 4.1–7.8 kPa). In subgroup comparisons, Mann-Whitney U test was employed for LSM comparisons (CPAP vs. non-CPAP: U = 387, p = 0.45; MASLD vs. non-MASLD: U = 245, p = 0.02).

Table 2.

Distribution of Liver Steatosis and Fibrosis Stages in OSA Patients

Parameter	Classification	Cutoff Values	Count (n)	Percentage (%)	95% CI
Hepatic Steatosis (CAP)
S0 (Normal)	CAP < 238 dB/m	< 238 dB/m	15	20.8	12.4–31.7
S1 (Mild)	Mild steatosis	238–260 dB/m	7	9.7	4.0–19.0
S2 (Moderate)	Moderate steatosis	260–290 dB/m	17	23.6	14.4–35.1
S3 (Advanced)	Advanced steatosis	≥ 290 dB/m	33	45.8	34.5–57.4
MASLD (Any Grade)	CAP ≥ 238 dB/m	≥ 238 dB/m	57	79.2	68.5–87.4
Liver Fibrosis (LSM)
F0–F1 (Normal/Mild)	None to mild fibrosis	2.0–7.0 kPa	57	79.2	68.5–87.4
F2 (Moderate)	Moderate fibrosis	7.5–10.0 kPa	12	16.7	9.2–27.3
F3 (Severe)	Severe fibrosis	10.0–14.0 kPa	3	4.2	0.9–11.7
F4 (Cirrhosis)	Cirrhosis	≥ 14.0 kPa	0	0.0	0.0–5.0
Significant Fibrosis	LSM ≥ 7.5 kPa	≥ 7.5 kPa	15	20.8	12.4–31.7
Advanced Fibrosis	LSM ≥ 14.0 kPa	≥ 14.0 kPa	0	0.0	0.0–5.0

*Note: Steatosis and fibrosis staging based on established FibroScan cutoff values for MASLD. Confidence intervals calculated using Wilson score method

CAP controlled attenuation parameter, CI confidence interval, LSM liver stiffness measurement, MASLD metabolic-dysfunction associated steatotic liver disease, OSA obstructive sleep apnea

Fig. 3.

Liver fibrosis distribution. Notes: Bar chart displaying the distribution of liver fibrosis stages in OSA patients

CPAP therapy analysis

Among the 16 patients receiving CPAP therapy at enrollment, MASLD prevalence remained elevated at 75.0% (12/16), compared to 80.4% (45/56) in those not on CPAP (p = 0.65, not significant). Mean CAP and LSM values were similar between CPAP users and non-users (mean CAP 275 dB/m vs. 285 dB/m, p = 0.67; mean LSM 5.8 vs. 6.5 kPa, p = 0.45). There were no statistically significant differences in the rates of MASLD or advanced liver disease between the two groups. However, this subgroup comparison should be interpreted with caution given the small sample size (n = 16) providing limited statistical power (estimated power = 0.12 to detect a 20% difference at α = 0.05). Figure 4 compares key liver parameters (hepatic steatosis, liver stiffness) and MASLD prevalence in CPAP users versus non-users.

Fig. 4.

Comparison of liver parameters between CPAP users and non-CPAP users. Notes: Bar chart comparing liver parameters between CPAP users (n = 16) and non-CPAP users (n = 56). No statistically significant differences were observed in CAP scores, liver stiffness measurements, or MASLD prevalence between groups. Error bars represent standard error of the mean. Statistical comparisons: CAP score p = 0.67 (95% CI for difference: −42.3 to 62.3 dB/m); LSM p = 0.45 (95% CI for difference: −1.2 to 2.6 kPa); MASLD prevalence p = 0.65 (95% CI for difference: −0.27 to 0.17)

Metabolic comorbidities and MASLD

Patients with metabolic comorbidities demonstrated a higher burden of fatty liver disease. Notably, all patients with diabetes mellitus in this cohort had MASLD (100% vs. 76.6% in those without diabetes). Similarly, OSA patients with hypertension showed a higher MASLD prevalence (87.0%) compared to normotensive patients (75.5%). Table 3 details MASLD prevalence stratified by the presence or absence of key metabolic comorbidities, as well as by sex. Although males had a slightly higher MASLD prevalence than females (80.0% vs. 75.0%), the difference was not statistically significant in this sample. The clustering of metabolic risk factors (diabetes, hypertension) with MASLD in OSA patients is evident.

Table 3.

MASLD Prevalence by Patient Characteristics and Risk Factors

Factor	Subgroup	MASLD Present/Total	Prevalence (%)	95% CI	p-value
Sex
	Male (n = 60)	48/60	80.0	67.7–89.2	0.69
	Female (n = 12)	9/12	75.0	42.8–94.5
Age Groups
	< 40 years (n = 23)	17/23	73.9	51.6–89.8	0.45
	40–60 years (n = 39)	32/39	82.1	66.5–92.5
	> 60 years (n = 10)	8/10	80.0	44.4–97.5
BMI Categories
	Normal weight (< 25 kg/m²) (n = 3)	2/3	66.7	9.4–99.2	0.62
	Overweight (25–29.9 kg/m²) (n = 20)	15/20	75.0	50.9–91.3
	Class I obesity (30–34.9 kg/m²) (n = 28)	23/28	82.1	63.1–93.9
	Class II obesity (35–39.9 kg/m²) (n = 15)	12/15	80.0	51.9–95.7
	Class III obesity (≥ 40 kg/m²) (n = 6)	5/6	83.3	35.9–99.6
Diabetes Mellitus
	Present (n = 8)	8/8	100.0	63.1–100.0	0.12
	Absent (n = 64)	49/64	76.6	64.3–86.2
Hypertension
	Present (n = 23)	20/23	87.0	66.4–97.2	0.25
	Absent (n = 49)	37/49	75.5	61.1–86.7
Dyslipidemia
	Present (n = 18)	16/18	88.9	65.3–98.6	0.22
	Absent (n = 54)	41/54	75.9	62.4–86.5
CPAP Therapy
	CPAP users (n = 16)	12/16	75.0	47.6–92.7	0.65
	Non-CPAP users (n = 56)	45/56	80.4	67.6–89.8

*Statistical comparisons performed using χ² test or Fisher’s exact test for categorical variables. Confidence intervals calculated using Wilson score method

BMI body mass index, CI confidence interval, CPAP continuous positive airway pressure, MASLD metabolic-dysfunction associated steatotic liver disease

BMI stratification analysis

MASLD prevalence increased progressively with higher BMI categories, though the relationship was not strictly linear due to small sample sizes at the extremes of BMI. In normal-weight OSA patients (BMI < 25), MASLD was observed in an estimated two-thirds of patients, however, this subgroup comprised only 3 patients, severely limiting the reliability of this estimate and precluding meaningful statistical inference. These exploratory findings in normal-weight patients require confirmation in larger cohorts. In the overweight and obese categories MASLD prevalence generally exceeded 75%. The highest MASLD prevalence was seen in Class II obese patients. In Class III obesity (BMI ≥ 40), MASLD prevalence appeared slightly lower than Class II, likely reflecting the very small sample (n = 6) in this category. Figure 5 illustrates MASLD prevalence across BMI categories.

Fig. 5.

Relationship between BMI and MASLD prevalence. Notes: Stacked bar chart showing MASLD prevalence across different BMI categories in OSA patients. A progressive increase in MASLD prevalence was observed with higher BMI categories, though the relationship was not strictly linear. Sample sizes: Normal weight (n = 3), Overweight (n = 20), Class I obesity (n = 28), Class II obesity (n = 15), Class III obesity (n = 6). BMI, body mass index

Correlation analysis

There were significant positive correlations between measures of adiposity and liver steatosis/fibrosis in this cohort. BMI showed a moderate positive correlation with CAP score (r = 0.42, p < 0.001) and a weaker correlation with LSM (r = 0.28, p = 0.018). Figure 6 provides a visual correlation matrix of key clinical variables (BMI, CAP, LSM, and age), with correlation coefficients and significance levels. Notably, age was not significantly correlated with CAP or LSM in this relatively middle-aged cohort.

Fig. 6.

Correlation matrix of key clinical variables. Notes: Heat map displaying Pearson correlation coefficients between BMI, CAP score, liver stiffness measurement (LSM), and age. *p < 0.05, **p < 0.01, ***p < 0.001. Confidence intervals for key correlations: BMI-CAP r = 0.469 (95% CI: 0.26–0.64); CAP-LSM r = 0.282 (95% CI: 0.05–0.48). Strong positive correlations were observed between BMI and CAP scores (r = 0.42) and between CAP scores and LSM (r = 0.35). Color coding indicates correlation strength: Red (strong correlation, |r|≥0.5), Yellow (moderate correlation, 0.3 ≤ r < 0.5), Green (weak correlation, 0.1 ≤ r < 0.3), Gray (minimal correlation, r < 0.1)

In multivariate logistic regression analysis (Table 4), no individual variable reached statistical significance as an independent predictor of MASLD after adjustment for confounders (model Nagelkerke R²=0.18, Hosmer-Lemeshow p = 0.62). The quasi-complete separation caused by 100% MASLD prevalence in diabetic patients precluded stable estimation for this covariate. While BMI (adjusted OR 1.09 per kg/m²) and lower nadir SpO₂ (adjusted OR 1.38 per 5% decrease) showed trends toward association with MASLD, neither achieved statistical significance. The modest explanatory power of the model suggests that unmeasured factors or the high baseline MASLD prevalence in this population limited detection of independent predictors. In linear regression analysis for CAP score, BMI remained the only significant predictor (β = 0.42, p < 0.001), explaining 22% of variance (R²=0.22).

Table 4.

Multivariate Logistic Regression Analysis for Predictors of MASLD

Variable	Unadjusted OR (95% CI)	Adjusted OR (95% CI)	p-value
BMI (per 1 kg/m² increase)	1.12 (1.02–1.23)	1.09 (0.98–1.21)	0.11
Diabetes mellitus*	- (undefined)*	- (undefined)*	excluded*
Hypertension	2.19 (0.58–8.31)	1.87 (0.42–8.32)	0.41
AHI (per 10 events/hr increase)	1.18 (0.89–1.56)	1.11 (0.79–1.55)	0.54
Nadir SpO₂ (per 5% decrease)	1.48 (1.02–2.14)	1.38 (0.91–2.09)	0.13
Age (per 10-year increase)	1.09 (0.72–1.65)	0.97 (0.60–1.58)	0.91
Male sex	1.33 (0.32–5.58)	1.15 (0.23–5.72)	0.87

MASLD Metabolic dysfunction-associated steatotic liver disease, OR Odds ratio, CI Confidence interval, BMI Body mass index, AHI Apnea-hypopnea index, SpO₂ Peripheral oxygen saturation

*All diabetic patients had MASLD, resulting in quasi-complete separation; Diabetes mellitus was therefore excluded from the multivariate model due to its perfect association with MASLD

In an exploratory analysis, an ROC curve was constructed to assess the utility of BMI as a standalone screening tool for MASLD (Fig. 7). The Area Under the Curve (AUC) was 0.85 (95% CI:), indicating good discriminatory ability. The optimal cutoff point identified via the Youden index was a BMI of 32 kg/m², which yielded a sensitivity of 75.4% and a specificity of 80.0%. While BMI is a strong predictor, these values indicate that relying on it alone would still misclassify a proportion of patients, reinforcing the value of direct, non-invasive assessments like FibroScan in this high-risk population.

Fig. 7.

ROC curve analysis for MASLD prediction using BMI. Notes: Receiver operating characteristic (ROC) curve for BMI as a predictor of MASLD presence. Area under the curve (AUC) = 0.85; 95% confidence interval (95% CI) = 0.76–0.95, indicating good discriminatory ability. The optimal BMI cutoff for MASLD prediction was 32 kg/m², with sensitivity 75.4% and specificity 80.0%. The diagonal reference line (red) represents no discriminatory ability (AUC = 0.5)

Discussion

This cross-sectional study demonstrates a remarkably high prevalence of MASLD (79.2%) among patients with OSA, substantially exceeding general population estimates of ~ 25–30% [1]. Nearly half of the OSA patients (45.8%) in our cohort exhibited advanced hepatic steatosis, and one in five (20.8%) had significant liver fibrosis, underscoring a significant subclinical burden of liver disease in this population. These findings align with emerging evidence suggesting a bidirectional causal relationship between OSA and MASLD, where each condition may exacerbate the other through shared metabolic pathways [11–14].

MASLD prevalence in OSA: evidence for bidirectional causality

The observed MASLD prevalence of 79.2% in our OSA cohort is notably higher than regional NAFLD prevalence estimates for the Middle East (up to 32%) [2]. This finding aligns with recent systematic reviews and meta-analyses demonstrating consistent associations between OSA and liver disease across diverse populations [15, 16]. A 2023 Mendelian randomization study provided genetic evidence for bidirectional causality, showing that genetically predicted OSA increases MASLD risk (OR = 1.84, 95% CI: 1.20–2.83), while MASLD may predispose to OSA development [14].

The recent nomenclature changed from NAFLD to MASLD, formally adopted in 2023, emphasizes metabolic dysfunction as the central diagnostic criterion [6]. This paradigm shift aligns perfectly with our findings, as OSA represents a quintessential metabolic dysfunction characterized by chronic intermittent hypoxia (CIH), oxidative stress, and systemic inflammation. The high prevalence of advanced steatosis (45.8%) in our cohort exceeds that reported in unselected NAFLD populations, suggesting that OSA patients may experience accelerated disease progression.

Pathophysiological mechanisms: beyond shared risk factors

The strong correlation observed between BMI and hepatic steatosis severity (CAP: r = 0.469, p < 0.001) in our study underscores obesity’s central role in both conditions. However, the presence of MASLD even in normal-weight OSA patients (66.7% prevalence) suggests OSA-specific mechanisms contribute independently to liver pathology. Recent mechanistic studies have elucidated multiple pathways:

1. Chronic Intermittent Hypoxia (CIH): CIH triggers hepatic hypoxia-inducible factor (HIF) signaling, particularly HIF-2α, promoting gluconeogenesis, de novo lipogenesis, and inflammatory gene expression [17, 18]. Mesarwi et al. demonstrated that CIH increases hepatic lipid accumulation by 30–40% in murine models, independent of dietary factors [19].

2. Gut-Liver Axis Disruption: A 2024 meta-analysis revealed that OSA patients exhibit increased intestinal permeability and altered gut microbiota composition, with elevated lipopolysaccharide levels correlating with liver inflammation markers (r = 0.52, p< 0.001) [20].

3. Adipose Tissue Dysfunction: Recent evidence shows that adipocyte-derived exosomes from OSA patients carry specific microRNAs (particularly miR-455-3p) that promote hepatic steatosis through the TCONS_00039830/Smad2 axis [21].

Clinical implications and risk stratification

The clustering of MASLD with metabolic comorbidities in our OSA patients highlights opportunities for integrated management. Our finding that 100% of OSA patients with diabetes had MASLD, compared to 76.6% without diabetes, underscores the synergistic effects of metabolic dysfunction. Figure 8 illustrates a clinical decision algorithm for MASLD screening in patients with OSA.

Fig. 8.

Clinical decision algorithm for MASLD screening in OSA patients. Notes: Flowchart demonstrating a proposed clinical decision tree for OSA patients to determine who requires liver screening. OSA patients with BMI ≥ 32 kg/m², diabetes mellitus, or hypertension are recommended for liver screening with FibroScan due to high MASLD risk. Lower-risk patients receive routine follow-up with lifestyle advice and periodic liver function tests

To translate these findings into clinical practice, we developed a risk stratification tool based on our cohort data as seen in Table 5. This OSA-MASLD risk assessment tool is free available as a free online calculation from https://osa-masld-risk.netlify.app/ [22]. See Fig. 9 for details.

Table 5.

OSA-MASLD Risk Score

Risk Factor	Points	Prevalence in MASLD	Odds Ratio (95% CI)
BMI ≥ 32 kg/m²	2	68.4%	3.2 (1.8–5.7)
Diabetes Mellitus	3	100%	- (undefined)
Hypertension	1	87.0%	2.1 (1.2–3.8)
Male Gender	1	84.2%	1.3 (0.8–2.1)
Dyslipidemia	1	88.9%	2.3 (1.3–4.1)

Risk Categories:

• Low Risk (0–2 points): 40% MASLD prevalence - Routine monitoring

• Moderate Risk (3–4 points): 75% MASLD prevalence - Consider FibroScan screening

• High Risk (≥ 5 points): 95% MASLD prevalence - Urgent liver assessment recommended

Fig. 9.

MASLD Risk Calculator and Clinical Decision Tool for OSA Patients. Notes: Risk Stratification and Clinical Actions scoring system developed with 0–2 points indicates likely low risk of MASLD (< 50%). 3–4 points being Moderate Risk of MASLD (≥ 75%) and ≥ 5 points indicating high risk of MASLD (95%). The patient with moderate and high risk of MASLD requires urgent confirmation with imaging such as FibroScan and possible hepatology referral

Understanding the correlation matrix: clinical and research applications

The Pearson correlation matrix (Fig. 6) provides crucial insights into the interrelationships between anthropometric and liver parameters:

BMI-CAP correlation (r = 0.469, p < 0.001)

This moderate positive correlation indicates that BMI explains approximately 22% of the variance in hepatic steatosis (r²=0.220). While clinically meaningful, this finding suggests that other factors contribute significantly to liver fat accumulation in OSA patients.

CAP-LSM correlation (r = 0.282, p < 0.05)

The weak but significant correlation between steatosis and fibrosis indicates these are related but partially independent processes. This has important clinical implications:

• Some patients develop significant steatosis without proportional fibrosis.

• Both CAP and LSM measurements provide complementary diagnostic information.

• Sequential disease progression models may not apply uniformly to all patients.

Age independence

The lack of significant correlation between age and liver parameters suggests that OSA-related mechanisms may override age-related factors in driving liver disease, emphasizing the importance of OSA-specific interventions regardless of patient age.

CPAP therapy: limited impact and need for combination approaches

Our cross-sectional comparison revealed no significant difference in MASLD prevalence between CPAP users (75.0%) and non-users (80.4%, p = 0.65). However, this null finding must be interpreted with substantial caution given several methodological limitations. First, the small CPAP subgroup (n = 16) provided only 12% statistical power to detect a clinically meaningful difference. Second, our cross-sectional design cannot distinguish whether CPAP therapy truly has no effect or whether patients established on CPAP may have had more severe baseline disease prompting earlier treatment initiation. Third, CPAP adherence data were unavailable, and variable adherence rates (averaging 4–5 h/night in clinical practice) may dilute any potential therapeutic effect. Despite these limitations, our findings align with the landmark randomized controlled trial by Ng et al., which showed that 6 months of therapeutic CPAP did not reduce liver fat or fibrosis compared to sham CPAP [23]. Although earlier studies suggested possible benefit on MASLD with CPAP therapy in adults and children with OSA using liver injury markers and ultrasound but this has not been replicated [24–26]. Several factors may explain these observations:

1. Insufficient Treatment Duration: Liver remodeling may require longer than typical CPAP treatment periods.

2. Adherence Issues: Real-world CPAP adherence averages 4–5 h/night, potentially insufficient for metabolic benefits.

3. Established Disease: Once MASLD is established, correcting nocturnal hypoxemia alone may be inadequate.

Recent evidence supports combination of therapeutic approaches. The 2024 FDA approval of tirzepatide for OSA represents a paradigm shift, with the SURMOUNT-OSA trials demonstrating 63% AHI reduction alongside significant weight loss [27]. A meta-analysis by Liu et al. showed that combining CPAP with weight loss interventions reduced liver enzymes by 35% more than either intervention alone [28].

Emerging therapeutic paradigms

Recent therapeutic advances offer new opportunities for managing OSA patients with MASLD:

GLP-1 receptor agonists

A 2024 systematic review demonstrated that these agents reduce both AHI (mean reduction: 58.7%) and hepatic steatosis (CAP reduction: 42 dB/m) in OSA patients [29].

Resmetirom

The first FDA-approved medication for MASH achieved resolution in 29.9% of patients in the MAESTRO-NASH trial, offering targeted liver therapy for OSA patients with advanced disease [30].

SGLT-2 inhibitors

Emerging evidence suggests these agents improve both sleep-disordered breathing and MASLD through weight-independent mechanisms, including reduced oxidative stress and improved mitochondrial function [31].

Our findings contribute to the growing recent evidence which establishes OSA as a distinct driver of MASLD and aligns with a recent meta-analysis which identified OSA-related hypoxemia as a stronger predictor of hepatic steatosis than AHI, an observation consistent with our own correlation findings between oxygen desaturation parameters and liver stiffness measurements [32]. This is more pronounced in high-risk populations with recent study showing that OSA significantly increases the odds of MASLD in patients with type 2 diabetes (OR 2.41) [33]. However, this risk is not just exclusive to patients with obesity as even non-obese (normal BMI) patients with fatty liver disease share the same altered cardiovascular and metabolic profile as obese patients, suggesting that visceral adiposity and insulin resistance can drive liver injury independently of the BMI category [34]. Hence, as suggested in recent literature, MASLD must be managed as a multisystem disease as effects of MASLD extend beyond hepatic manifestations and includes cardiovascular disease, chronic kidney disease and extrahepatic malignancies [35]. Screening strategies should look beyond simple weight-based stratification and target shared metabolic drivers, regardless of obesity status, including sleep fragmentation and intermittent hypoxia, to prevent progression to fibrosis and cardiovascular complications.

Study limitations and strengths

Several limitations warrant consideration. The cross-sectional design precludes causal inference, though our findings align with mechanistic studies supporting bidirectional causality. The small CPAP subgroup (n = 16) and extremely limited normal-weight OSA subgroup (n = 3) precluded adequately powered comparisons. Post-hoc power analysis indicated only 12% power to detect a clinically meaningful 20% difference in MASLD prevalence between CPAP users and non-users. These subgroup analyses should therefore be considered hypothesis-generating, and the absence of statistically significant differences should not be interpreted as evidence of equivalence. The predominantly male cohort (83.3%) reflects OSA demographics but may limit generalizability to female patients, who may have different MASLD risk profiles especially post-menopause, which can be associated with a greater than two-fold increase in the odds of MASLD (OR 2.37), as elucidated in a recent meta-analysis [36]. The absence of CPAP adherence data also prevents assessment of dose-response relationships between OSA treatment and liver outcomes.

Another limitation is that the FibroScan is validated for MASLD assessment, but it remains inferior to the gold-standard liver biopsy for histological characterization. In addition, our sample size calculation was powered for MASLD estimation rather than multivariate analyses, limiting the reliability of regression models and we lacked data on important potential confounders including dietary habits, physical activity levels, insulin resistance markers (HOMA-IR), and sleep architecture parameters beyond standard polysomnographic indices. Finally, the absence of multiple serial FibroScan assessments prevents comprehensive evaluation of MASLD trajectory over a longer period of time.

Strengths include the use of validated FibroScan technology providing objective liver assessment, comprehensive phenotyping enabling multivariable analysis, and a well-characterized OSA population at the critical juncture of treatment initiation. Our correlation analyses provide effect size estimates valuable for future power calculations and study design.

Future research priorities

Based on our findings and current knowledge gaps, we suggest several research priorities. Longitudinal studies employing prospective cohorts to track liver outcomes with long-term CPAP therapy are essential, particularly those incorporating adherence monitoring and dose-response analyses to determine whether a therapeutic threshold exists for hepatic benefit. Combination therapy trials comparing CPAP alone versus CPAP combined with metabolic interventions such as GLP-1 receptor agonists or structured lifestyle modification programs can help establish optimal management strategies for this patient population. The development and validation of non-invasive biomarkers capable of identifying OSA patients at highest risk for MASLD progression is another research priority, which can enable targeted screening and early intervention. Precision medicine using genetic and metabolomic profiling can also identify patient subgroups most likely to benefit from specific therapeutic interventions, moving beyond the current one-size-fits-all paradigm. Finally, health economic analyses evaluating the cost-effectiveness of systematic MASLD screening within OSA clinics would provide essential data to inform clinical practice guidelines and healthcare policy decisions.

Conclusions

This study reveals an alarmingly high prevalence of MASLD (79.2%) in patients with OSA, with nearly half exhibiting advanced hepatic steatosis and one-fifth demonstrating significant liver fibrosis. The substantial burden of occult liver disease, particularly among OSA patients with metabolic comorbidities, supports systematic liver health assessment in OSA management protocols. Our risk stratification model provides a practical framework for identifying high-risk patients requiring urgent evaluation.

The correlation analyses demonstrate that while obesity is a major contributor, OSA-specific mechanisms independently drive liver pathology. The limited impact of CPAP therapy alone on liver parameters emphasizes the need for integrated care approaches targeting both sleep-disordered breathing and metabolic dysfunction. Emerging combination therapies, including GLP-1 receptor agonists and liver-targeted medications, offer promising avenues for comprehensive management.

Clinicians managing OSA patients should maintain a high index of suspicion for MASLD, particularly in those with BMI ≥ 30.2 kg/m², diabetes, or hypertension. Incorporating non-invasive liver screening such as FibroScan into routine OSA care could identify patients with advanced disease requiring hepatology referral. The development of evidence-based guidelines for dual management of OSA and MASLD represents a critical need, with potential to improve outcomes for millions of patients worldwide with these overlapping conditions.

Funding

No funding was received for this research.

Data availability

Data will be made available on reasonable request.

Declarations

Ethical approval

All procedures performed in studies involving human participants were in accordance with the ethical standards of the United Arab Emirates University Human Medical Research Ethics Committee, with an approval number ERH_2023_3147, and with the 1964 Helsinki declaration and its later amendments.

Financial disclosure

None to declare.

Non-financial disclosure

None to declare.

Conflict of interest

No Conflicts of Interest.