Non-alcoholic fatty liver disease fibrosis score is a useful tool for carotid atherosclerosis screening in patients with obstructive sleep apnea

Abstract

BACKGROUND

Obstructive sleep apnea (OSA) is recognized as an independent cardiovascular risk factor since this respiratory disorder induces intermittent hypoxia and systemic inflammation, which are closely related to the development and progression of atherosclerosis. In parallel, there is a strong bidirectional association between OSA and metabolic dysfunction-associated steatotic liver disease (MASLD), the latter being a hepatic manifestation of the metabolic syndrome. There is increasing clinical evidence that high blood-based MASLD predictive scores are associated with increased vascular atherosclerotic damage. The significance of these algorithms as predictors of carotid atherosclerosis in patients with OSA is poorly understood.

AIM

To assess blood-based MASLD algorithms as predictors of carotid atherosclerosis in patients with OSA.

METHODS

This cross-sectional study comprises a cardiorespiratory polygraphic study and assessment of carotid artery damage by doppler ultrasonography carried out in well-characterized patients with OSA (n = 112) and in patients without OSA (n = 32) who were considered as controls. Logistic regression models were constructed to evaluate potential associations between distinct MASLD predictive scores and clinical, analytical and polygraphic features of the study population.

RESULTS

Both the presence and volume of carotid plaques were significantly higher in patients with OSA than in controls (P = 0.035 and 0.001, respectively). None of the polygraphic variables studied were associated with the presence of carotid atherosclerosis in OSA patients. However, multivariate logistic regression analysis revealed that non-alcoholic fatty liver disease fibrosis score (NFS) (odds ratio = 1.84, P = 0.010) and fibrosis-4 index (odds ratio = 5.53, P = 0.004) were significantly associated with carotid atherosclerosis and showed good diagnostic accuracy. When these scores were tested in OSA patients without MASLD, only high values of NFS seemed to be linked to carotid atherosclerosis (P = 0.0501) and positively correlated with carotid plaque volume (r = 0.665, P = 0.041).

CONCLUSION

Elevated NFS is significantly associated with carotid plaques in OSA. This predictive score might be an initial useful tool to identify patients at high cardiovascular risk to whom in-depth evaluation must be recommended.

Key Words: Obstructive sleep apnea; Carotid atherosclerosis; Vascular risk; Liver fibrosis scores; Metabolic dysfunction-associated steatotic liver disease; Metabolic dysfunction-associated steatohepatitis

Core Tip: The non-alcoholic fatty liver disease fibrosis score, a simple non-invasive blood-based algorithm, predicts the presence of subclinical carotid atherosclerosis in patients with obstructive sleep apnea. Importantly, this association is independent of the metabolic dysfunction-associated steatotic liver disease status of the patient, suggesting that non-alcoholic fatty liver disease fibrosis score is an effective, readily available tool for cardiovascular risk stratification in the obstructive sleep apnea population, even in the absence of liver disease.

INTRODUCTION

Obstructive sleep apnea (OSA) is a highly prevalent respiratory disease characterized by throat narrowing or collapsing repeatedly during sleep, causing OSA events[1]. It is often related to metabolic disorders like metabolic dysfunction-associated steatotic liver disease (MASLD), formerly known as non-alcoholic fatty liver disease (NAFLD)[2,3], as well as to other respiratory diseases such as chronic obstructive pulmonary disease (COPD)[4] and to cardiovascular diseases (CVD)[5,6]. In previous studies, our group demonstrated that OSA is associated with insulin resistance, dyslipidemia and higher prevalence of liver steatosis, as well as metabolic dysfunction-associated steatohepatitis (MASH)[7]. Moreover, Sukahri et al[8] reported that in MASLD patients, the degree of steatosis significantly correlated with the severity of OSA and with serum levels of intracellular adhesion molecule-1 and lipoprotein-A, along with the carotid intima-media thickness (IMT), which is a widely used surrogate marker of atherosclerosis. Regarding CVD, it has been previously demonstrated in a population-based study that the severity of OSA was independently associated with subclinical systemic atherosclerosis as well as higher calcification in the ascending thoracic aorta of patients with moderate or severe OSA[9]. In another study, Ji et al[10] reported that increased inflammatory factors in OSA patients might cause the progression of atherosclerosis, and that apnea-hypopnea index (AHI) is an independent indicator of carotid atherosclerosis. However, these studies failed to characterize the presence of overlap syndrome (OS) - defined as co-existence of both COPD and OSA in the same patient - with unpredictable repercussion on results interpretation. In this regard, a meta-analysis of observational clinical studies showed that COPD patients have 2-fold higher risk of CVD than subjects without COPD, being ischemic heart disease, ischemic stroke and peripheral artery disease the most frequently observed CVD in COPD patients[11]. There is a significant burden of CVD in patients with COPD and OSA, affecting negatively quality of life and survival, and therefore, OS has an additional impact on the cardiovascular system multiplying the risk of morbidity and mortality[12]. Additionally, we have recently shown that OS is associated with higher prevalence and volume of left carotid atheromatous plaques. Indeed, both plaque prevalence and volume were increased in patients with OS, regardless of COPD severity[13]. Thus, it is imperative to discriminate between OSA and OS in order to accurately predict the potential vascular risk of each patient. CVD remains the leading cause of disease burden and mortality globally[14,15]. Notably, it represents the primary cause of death among patients with severe MASLD[16,17]. Hence, it is crucial to raise awareness of these co-morbidities among physicians caring for at-risk populations. Liver biopsy is an invasive procedure considered as the gold standard for MASH diagnosis and fibrosis staging in MASLD patients. However, liver biopsy is a costly and invasive procedure with potential serious complications and, therefore, non-invasive tools to assess advanced hepatic fibrosis, the major prognostic factor of MASLD patients[18], such as transient elastography and blood-based algorithms are emerging. Regarding the latter, Fibrosis-4 (FIB-4) Index[19,20], NAFLD Fibrosis Score (NFS)[21] and Hepamet Fibrosis Score (HFS)[22], have been developed to identify patients at risk of advanced liver fibrosis. While these clinical scores are highly effective for fibrosis staging, they often lack the specificity required to distinguish between simple steatosis and active MASH. To address this, the test OWLiver^™ was validated with excellent diagnostic accuracy[23]. Interestingly, high values of NFS and FIB-4 scores were predictors of coronary artery disease in MASLD patients[24], and have been associated with an increased risk of both cardiovascular and all-cause mortality among patients with coronary artery disease[25]. Moreover, NFS is strongly associated with carotid IMT, carotid plaque presence and arterial stiffness in patients with MASLD[26], independently of conventional cardiometabolic risk factors and insulin resistance. However, whether these scores are related to cardiovascular risk in patients with OSA has not been investigated. Given that OSA patients have a substantially higher risk of developing MASLD and CVD, we aimed to determine the potential utility of steatosis and fibrosis scores as predictors of carotid atherosclerosis in patients with OSA. We also searched for potential associations with clinical, analytical, and polygraphic features of the study population, aiming to validate these scores as rapid low-cost non-invasive markers for carotid atherosclerosis screening.

MATERIALS AND METHODS

Study population

This prospective cross-sectional cohort study included consecutive patients with clinical and polygraphic criteria of OSA (n = 112), and subjects who had sleep polygraphy within normality were included in the study and considered as controls (n = 32). Patients and controls were excluded if they drank more than 20 g/day of alcohol, had a diagnosis of asthma or cancer or any concomitant severe clinical disorder. In addition, they were also excluded if they had analytical evidence of iron overload (transferrin saturation index ≥ 55%), were seropositive for autoantibodies and/or for hepatitis B virus, hepatitis C virus, and human immunodeficiency virus as well as those taking potentially hepatotoxic drugs.

This study was performed in agreement with the Declaration of Helsinki, and with local and national laws. The Institution’s Clinical Research Ethics Committee approved the study procedures, approval No. PI/2800-16, and all participants signed a written informed consent before inclusion in the study, providing permission for their medical data to be anonymously used for research.

Demographic, clinical and biochemical assessment

Clinical examination was performed to all participants in this study including a detailed interview with special emphasis on alcohol intake, previous history of CVD - including acute myocardial infarction, stroke, chronic heart failure and atrial fibrillation - diabetes, arterial hypertension, cigarette smoking and measurements of weight and height. Body mass index was calculated as weight (kilograms) divided by height (meters) squared. After a 12-hour overnight fast, venous blood samples of each participant were obtained to test serum levels of liver enzymes, metabolic parameters and autoantibodies using routine laboratory methods. Blood-based algorithms were used to predict hepatic steatosis, Fatty Liver Index[27] and Framingham Steatosis Index (FSI)[28], as well as the absence or presence of liver fibrosis (FIB-4[19,20], NFS[21] and HFS[22]). Serum lipidomic profiling was determined to all study patients and control subjects by using the test OWLiver™ (One Way Liver SL, Derio, Spain), a commercially-available assay recently validated to distinguish simple steatosis and MASH with excellent diagnostic accuracy[23]. In addition, plasma insulin was determined by a chemiluminescent microparticle immunoassay (ARCHITECT insulin; Abbot Park, IL, United States). Insulin resistance was calculated by the Homeostatic model assessment-insulin resistance method[29]. Metabolic syndrome was defined according to the adult treatment panel III criteria[30]. Antibodies against hepatitis C virus and human immunodeficiency virus as well as hepatitis B virus surface antigen were tested by immunoenzymatic assays (Murex, Dartford, United Kingdom).

Cardiorespiratory polygraphic study

The vast majority of polygraphic studies were performed at night in the Sleep Laboratory of the Hospital Universitario de La Princesa (Madrid, Spain). However, some participants with limitations on getting the hospital admission underwent polygraphic studies at home by the usual caregivers. A previously validated cardiorespiratory polygraphy equipment (SOMNOscreen™ Plus, Randersacker, Germany) with a Domino analysis software (Domino Data Lab, San Francisco, CA, United States) was used. The presence of an AHI equal to or greater than 5 per hour was used as diagnostic criterion for certainty of OSA. Severity of OSA was classified according to the value of AHI as mild (AHI, 5-14/hour), moderate (AHI, 15-29/ hour) or severe (AHI, ≥ 30/hour).

Spirometry

To assess the diagnosis and severity of COPD, spirometry was performed to all participants by using a JAEGER™ spirometer (Vyaire Medical, Madrid, Spain), which meets all the specifications required by the Spanish Respiratory Society, the European Respiratory Society and the American Thoracic Society. All patients and control subjects underwent pre and postbronchodilator spirometric determinations.

Assessment of vascular damage

Distinct features of vascular damage were determined by using ultrasonography (Applio XG, Canon, Tokyo, Japan) to each patient as follows: (1) IMT was measured in the distal 2 centimeters (cm) of both common carotid arteries. The methodology defined in the Mannheim consensus[31] was used. Each participant had a total of three carotid IMT measures by side, which were carried out and scored for quality by two expert vascular radiologists. The radiologists performing the ultrasound assessments were blinded to the patients’ clinical history and laboratory results to minimize measurement bias. We calculated carotid IMT for each participant as the average value of all measurements that met predefined quality standards; and (2) Volume of arterial plaques was determined following internationally accepted criteria described elsewhere[32]. Arterial plaque was defined when 2 of the following criteria were met: IMT > 1.5 mm; impression in the vascular lumen; and abnormal wall texture. The plaque burden found in both carotid arteries (2 cm distal common carotid arteries and 1 cm distal internal carotid arteries) was calculated. This plaque load was expressed as the sum of the volumes of all plaques. All ultrasound measurements were performed by two expert vascular radiologists using a 7 MHz linear probe (model PLT-704SB, Tokyo, Japan) and a high frequency volumetric linear probe model PLT-1204 MV probe (Canon, Tokyo, Japan) with a 3D/4D volumetric reconstruction software model Toshiba UIMV-A500A (Canon, Tokyo, Japan).

Carotid atherosclerosis was defined as the presence of arterial plaque in either left or right carotid artery. Vascular reactivity was measured as described elsewhere[33]. Briefly, right brachial artery diameter was measured before and after 4-minute ischemia, and vascular reactivity was expressed as percentage of artery dilatation.

Statistical analysis

Sample size calculation was performed following the sample size formula for proportions developed by Cochran. To obtain a representative sample of the OSA population with an estimated carotid atherosclerosis proportion of 50%, confidence level of 95% and a margin of error of 10%, a sample of 97 subjects was required. Our study cohort (n = 112) exceeds this requirement, ensuring adequate statistical power and representing an adequate sample. The Kolmogorov-Smirnov test was applied to evaluate if variables were adjusted or not to a normal distribution. Qualitative variables are presented as absolute, n (%). Quantitative variables are expressed as measures of central tendency (mean) and dispersion (SD). Qualitative data between groups were compared by Pearson’s χ² test or Fisher exact test as appropriate. Quantitative data between groups were compared by Pearson’s or Spearmen coefficient correlations depending on their adjustment to normal distribution. The Student’s t-test was used to calculate the difference of the means in the variables that followed a normal distribution and the Mann-Whitney U test for the variables with a non-parametric distribution. Logistic regression analysis, adjusted by confounding variables was performed to identify independent polygraphic variables [AHI, oxygen desaturation index (ODI), and percentage of time with oxygen saturation below 90% during sleep (Tc90%)] and liver steatosis and fibrosis scores associated with the presence of carotid atherosclerosis in the study population. Univariate and multivariate regression models were constructed, parameters were selected by likelihood ratio test, and Box-Tidwell procedure was used for testing linearity of logit. The goodness of fit was evaluated using the Hosmer-Lemeshow statistic. Multicollinearity was formally assessed using the variance inflation factor (VIF), with values ranging between 1 and 2 for all models. This confirms that multicollinearity does not influence the stability of our models. Significance was set at a value of P < 0.05. Statistical analysis was performed using SPSS software version 26.0 (SPSS Statistics, Armonk, NY: IBM Corp, United States).

RESULTS

Patient recruitment and baseline characteristics

This prospective cross-sectional cohort study included consecutive patients with clinical and polygraphic criteria of OSA among those who attended the outpatient clinics of the Respiratory Service at Hospital Universitario de La Princesa (Madrid, Spain) during a 3-month period (n = 112). In parallel, subjects who had sleep polygraphy within normality were included in the study and considered as controls (n = 32). Spirometry was performed to all the participants, and those diagnosed as COPD were excluded. The detailed patient recruitment process and flowchart are illustrated in Figure 1.

Figure 1 Flowchart with the case selection of patients included in the study. 221 patients were enrolled because of snoring and/or obstructive sleep apnea suspicion. 74 patients were discarded due to chronic obstructive pulmonary disease and 3 due to missing relevant data. After performing night polysomnography, 112 patients were diagnosed as obstructive sleep apnea and 32 were enrolled as matched control subjects. OSA: Obstructive sleep apnea; COPD: Chronic obstructive pulmonary disease.

Vascular risk is increased in patients with OSA

Baseline characteristics of patients with or without OSA in our study cohort are detailed in Table 1. Briefly, patients with OSA were slightly older, predominantly men, and were hypertensive more frequently than controls (P = 0.016). In addition, OSA patients had higher levels of insulin, homeostatic model assessment-insulin resistance score, triglycerides and very low-density lipoprotein cholesterol, as well as increased transaminases and ferritin. Vascular damage findings of studied patients with or without OSA are shown in Figure 2. Overall, we did not find differences in the vascular reactivity or carotid IMT in patients with OSA when compared to control subjects. However, the presence of carotid plaque and its volume was significantly higher in OSA patients than in controls (P = 0.035 and 0.001, respectively).

Figure 2 Vascular characteristics of the study cohort according to obstructive sleep apnea presence. All values in control (n = 32) and obstructive sleep apnea (n = 112) groups are expressed as mean ± SD. A: Vascular reactivity expressed in percentage of brachial artery dilatation after ischemia; B: Intima-media thickness expressed as mean of left and right carotid arteries; C: Presence of carotid atheromatous plaques; D: Carotid plaque volume expressed as mean of left and right carotid plaques. OSA: Obstructive sleep apnea.

Table 1 Characteristics of the study population, n (%)/mean ± SD.

Features	Control (n = 32)	OSA (n = 112)	P value
Age (years)	54.38 ± 8.64	58.68 ± 8.64	0.014
Gender			0.015
Women	20 (62.5)	42 (37.5)
Men	12 (37.5)	70 (62.5)
Body mass index (kg/m2)	28.91 ± 5.54	30.34 ± 5.79	0.215
Arterial hypertension	9 (28.1)	59 (52.7)	0.016
Previous history of CVD	1 (3.1)	12 (10.7)	0.298
Current smoking	10 (31.2)	24 (21.4)	0.345
Diabetes mellitus type 2	2 (6.3)	16 (14.3)	0.363
Glucose (mg/dL)	97.38 ± 13.37	105.60 ± 29.92	0.071
Insulin levels (μU/L)	11.54 ± 13.26	17.19 ± 13.26	0.001
HOMA-IR score	2.93 ± 2.27	5.13 ± 8.20	0.001
Glycated Hb (%)	5.62 ± 0.53	5.70 ± 0.74	0.668
Triglycerides (mg/dL)	106.47 ± 66.57	138.41 ± 81.74	0.006
Total cholesterol	196.81 ± 39.17	192.22 ± 33.73	0.514
HDL cholesterol (mg/dL)	57.53 ± 12.96	54.30 ± 18.65	0.361
LDL cholesterol (mg/dL)	115.22 ± 30.75	110.21 ± 32.98	0.443
VLDL cholesterol (mg/dL)	21.25 ± 13.24	27.45 ± 15.11	0.005
Creatinine (mg/dL)	0.89 ± 0.16	0.89 ± 0.18	0.847
Glomerular filtration rate (mL/minute/1.73 m2)	83.98 ± 14.81	85.46 ± 16.04	0.640
ALT (IU/L)	18.78 ± 7.34	24.71 ± 14.38	0.008
AST (IU/L)	20.28 ± 5.61	22.76 ± 11.64	0.379
GGT (IU/L)	22.19 ± 13.03	34.46 ± 25.75	0.001
Iron (μg/dL)	78.47 ± 33.50	88.28 ± 28.81	0.104
Ferritin (ng/mL)	87.41 ± 72.14	143.89 ± 96.55	0.003
Transferrin (mg/dL)	244.26 ± 31.59	249.51 ± 36.54	0.468
Alkaline phosphatase (IU/L)	69.63 ± 22.69	66.54 ± 18.74	0.435
Lactate dehydrogenase (U/L)	182.72 ± 36.12	185.38 ± 36.60	0.718
Albumin (g/dL)	4.38 ± 0.27	4.44 ± 0.27	0.115
Platelets (109/L)	226.30 ± 71.27	228.20 ± 52.87	0.892
C reactive protein (mg/L)	0.42 ± 0.59	0.37 ± 0.41	0.953

OSA: Obstructive sleep apnea; CVD: Cardiovascular disease; HOMA-IR: Homeostatic model assessment-insulin resistance; Hb: Hemoglobin; HDL: High-density lipoprotein; LDL: Low-density lipoprotein; VLDL: Very low-density lipoprotein; ALT: Alanine aminotransferase; AST: Aspartate aminotransferase; GGT: Gamma-glutamyltransferase.

Characteristics of the patients with OSA according to carotid atherosclerosis

The prevalence of carotid atherosclerosis in patients with OSA in our study cohort was 40%. Table 2 summarizes the characteristics of patients with OSA according to carotid plaque presence. Overall, those patients with carotid atherosclerosis were older and more hypertensive and had more frequently previous history of CVD. Carotid atherosclerosis was also associated with lower glomerular filtration rate, lower albumin, higher platelet count and higher lactate dehydrogenase activity; all these differences were consistent with atherosclerosis risk factors described previously.

Table 2 Characteristics of patients with obstructive sleep apnea according to carotid atherosclerosis, n (%)/mean ± SD.

Features	OSA without carotid plaque (n = 67)	OSA with carotid plaque (n = 45)	P value
Age (years)	56.30 ± 7.72	62.22 ± 8.79	> 0.001
Gender			0.237
Women	22 (32.8)	20 (44.4)
Men	45 (67.2)	25 (55.6)
Body mass index (kg/m2)	29.87 ± 5.52	31.04 ± 6.16	0.297
Arterial hypertension	25 (37.3)	34 (75.6)	> 0.001
Previous history of CVD	2 (3.0)	10 (22.2)	0.003
Current smoking	10 (14.9)	14 (31.1)	0.059
Diabetes mellitus type 2	9 (13.4)	7 (15.6)	0.788
Glucose (mg/dL)	107.43 ± 36.34	102.87 ± 16.26	0.684
Insulin levels (μU/L)	17.34 ± 14.93	16.96 ± 10.45	0.870
HOMA-IR score	5.46 ± 10.24	4.64 ± 3.45	0.959
Glycated Hb (%)	5.71 ± 0.83	5.68 ± 0.59	0.741
Triglycerides (mg/dL)	143.43 ± 88.10	130.93 ± 71.54	0.481
Total cholesterol	196.45 ± 30.34	185.93 ± 37.71	0.106
HDL cholesterol (mg/dL)	52.70 ± 16.59	56.69 ± 21.33	0.269
LDL cholesterol (mg/dL)	117.21 ± 30.02	99.78 ± 34.73	0.006
VLDL cholesterol (mg/dL)	28.30 ± 15.63	26.18 ± 14.38	0.411
Creatinine (mg/dL)	0.88 ± 0.18	0.92 ± 0.19	0.244
Glomerular filtration rate (mL/minute/1.73 m2)	88.93 ± 15.06	80.31 ± 16.22	0.005
ALT (IU/L)	25.34 ± 15.55	23.76 ± 12.57	0.846
AST (IU/L)	22.03 ± 8.54	23.84 ± 15.18	0.671
GGT (IU/L)	35.94 ± 26.64	32.24 ± 24.47	0.870
Iron (μg/dL)	88.67 ± 24.33	87.69 ± 34.72	0.200
Ferritin (ng/mL)	141.73 ± 99.48	147.14 ± 93.01	0.775
Transferrin (mg/dL)	252.55 ± 36.01	244.98 ± 37.26	0.284
Alkaline phosphatase (IU/L)	68.07 ± 19.66	64.24 ± 17.24	0.291
Lactate dehydrogenase (U/L)	179.13 ± 33.73	195.59 ± 39.17	0.023
Albumin (g/dL)	4.52 ± 0.25	4.32 ± 0.27	> 0.001
Platelets (109/L)	238.30 ± 50.51	213.00 ± 53.21	0.012
C reactive protein (mg/L)	0.38 ± 0.42	0.34 ± 0.41	0.612

OSA: Obstructive sleep apnea; CVD: Cardiovascular disease; HOMA-IR: Homeostatic model assessment-insulin resistance; Hb: Hemoglobin; HDL: High-density lipoprotein; LDL: Low-density lipoprotein; VLDL: Very low-density lipoprotein; ALT: Alanine aminotransferase; AST: Aspartate aminotransferase; GGT: Gamma-glutamyltransferase.

Carotid atherosclerosis is not related to respiratory parameters

We analyzed the potential relationship between respiratory parameters and presence of carotid atherosclerosis, which is detailed in Table 3. Neither AHI nor other respiratory parameters (ODI and Tc90%) were different in OSA patients with or without carotid plaque(s). Also, categorization of the variables according to previous clinical reports did not show any difference between the two groups.

Table 3 Respiratory features of patients with obstructive sleep apnea according to carotid atherosclerosis, n (%)/mean ± SD.

Respiratory features	OSA without carotid plaque (n = 67)	OSA with carotid plaque (n = 45)	P value
AHI (events/hour)	32.48 ± 41.05	29.28 ± 20.97	0.812
5-14	24 (35.8)	13 (28.9)	0.540
15-29	16 (23.9)	15 (33.3)	0.289
≥ 30	27 (40.3)	17 (37.8)	0.845
ODI (events/hour)	25.95 ± 18.81	30.00 ± 21.17	0.294
≥ 10	53 (79.1)	40 (88.9)	0.422
Tc90%	20.42 ± 26.24	23.13 ± 25.56	0.321
≥ 10%	29 (43.3)	23 (51.1)	0.562

OSA: Obstructive sleep apnea; AHI: Apnea-hypopnea index; ODI: Oxygen desaturation index; Tc90%: Percentage of sleep time with oxygen saturation less than 90%.

Liver steatosis and fibrosis scores are related to carotid atherosclerosis in patients with OSA

We then assessed the potential association of liver steatosis and fibrosis scores with carotid atherosclerosis in patients with OSA. As shown in Table 4, we did not find differences in simple steatosis or MASH prevalence according to OWLiver™ test or fatty liver index. However, the FSI was significantly higher in OSA patients with carotid atherosclerosis. Regarding liver fibrosis, although our cohort did not show relevant levels of fibrosis, all the scores analyzed - NFS, FIB-4 score and HFS - were higher in OSA patients with carotid atherosclerosis. Both NFS and FIB-4 score showed also different proportion of patients with low risk of advanced fibrosis according to carotid plaque(s) presence, while HFS had a bigger proportion of patients classified as low risk for advanced liver fibrosis and there were no differences regarding carotid atherosclerosis.

Table 4 Metabolic and hepatic characterization of patients with obstructive sleep apnea according to carotid atherosclerosis, n (%).

Features	OSA without carotid plaque (n = 67)	OSA with carotid plaque (n = 45)	P value
Body mass index ≥ 30	28 (41.8)	24 (53.3)	0.251
HOMA-IR score ≥ 3	41 (61.2)	28 (62.2)	1.000
Dyslipidemia	29 (43.3)	24 (53.3)	0.338
Metabolic syndrome	17 (25.4)	8 (17.8)	0.367
Simple steatosis by OWLiver™ test	20 (29.9)	15 (33.3)	0.836
Steatohepatitis by OWLiver™ test	37 (55.2)	19 (42.2)	0.247
FLI (a.u.)	66.58 ± 27.09	68.88 ± 29.99	0.486
FLI ≥ 60	41 (61.2)	32 (71.1)	0.317
FSI (a.u.)	0.34 ± 0.25	0.43 ± 0.26	0.035
FSI ≥ 0.23	37 (55.2)	36 (80.0)	0.009
NFS (a.u.)	-1.54 ± 0.95	-0.75 ± 1.10	> 0.001
NFS < -1.455	39 (58.2)	11 (24.4)	> 0.001
FIB-4 (a.u.)	1.12 ± 0.38	1.57 ± 1.05	> 0.001
FIB-4 < 1.3	50 (74.6)	21 (46.7)	0.002
HFS (a.u.)	0.06 ± 0.08	0.12 ± 0.18	0.005
HFS < 0.12	55 (82.1)	34 (75.6)	0.477

HOMA-IR: Homeostatic model assessment-insulin resistance; a.u.: Arbitrary units; FLI: Fatty liver index; FSI: Framingham steatosis index; FIB-4: Fibrosis-4; NFS: Non-alcoholic fatty liver disease fibrosis score; HFS: Hepamet fibrosis score.

NFS is an independent predictor of carotid atherosclerosis

We then performed logistic regression analysis to determine the utility of FSI and liver fibrosis scores for carotid atherosclerosis screening in patients with OSA. Univariate analysis was performed with liver scores, variables with significant differences regarding carotid atherosclerosis, and potential confounders such as sex, AHI, ODI and Tc90%. Since the different liver scores share common and relevant variables, a multivariate model was built for the analysis of each score, and significant variables in univariate analysis were selected by likelihood ratio test. Noteworthy, NFS was significantly associated with the presence of carotid plaque(s) [odds ratio = 1.84, 95% confidence interval (CI): 1.16-2.93, P = 0.010] in OSA patients besides arterial hypertension and current smoking status (Table 5). FIB-4 score also was an independent predictor (odds ratio = 5.53, 95%CI: 1.72-17.70, P = 0.004) in our cohort, along with arterial hypertension, current smoking status and albumin (Table 6). On the other hand, FSI did not resist the multivariate adjustment (Table 7).

Table 5 Univariate and multivariate analysis of non-alcoholic fatty liver disease fibrosis score and the independent variables associated with carotid atherosclerosis in patients with obstructive sleep apnea (n = 112).

Independent variables	Univariate analysis			Multivariate analysis
	OR	95%CI	P value	OR	95%CI	P value
NFS (a.u.)	2.16	1.41-3.33	< 0.001	1.84	1.16-2.93	0.010
Sex (female/male)	0.61	0.28-1.33	0.215	-	-	-
Arterial hypertension (yes/no)	5.19	2.24-12.04	< 0.001	3.67	1.46-9.20	0.006
Previous history of CVD (yes/no)	9.29	1.93-44.76	0.005	-	-	-
Current smoking (yes/no)	2.57	1.02-6.47	0.044	3.10	0.99-9.66	0.052
LDL-c (mg/dL)	0.98	0.97-0.99	0.007	-	-	-
GFR (mL/minute/1.73 m2)	0.96	0.94-0.99	0.007	-	-	-
LDH (U/L)	1.01	1.00-1.02	0.033	-	-	-
AHI (events/hour)	1.00	0.98-1.01	0.633	-	-	-
ODI (events/hour)	1.01	0.99-1.03	0.292	-	-	-
Tc90%	1.00	0.99-1.02	0.586	-	-	-

Multivariate model was built following forward introduction of the variables selected by likelihood ratio test. Hosmer-Lemshow test was performed to evaluate goodness of fit of the model (P = 0.268). OR: Odds ratio; CI: Confidence interval; NFS: NAFLD fibrosis score; a.u.: Arbitrary units; CVD: Cardiovascular disease; LDL-c: Low-density lipoprotein cholesterol; GFR: Glomerular filtration rate; LDH: Lactate dehydrogenase; AHI: Apnea-hypopnea index; ODI: Oxygen desaturation index; Tc90%: Percentage of sleep time with oxygen saturation less than 90%.

Table 6 Univariate and multivariate analysis of fibrosis-4 index and the independent variables associated with carotid atherosclerosis in patients with obstructive sleep apnea (n = 112).

Independent variables	Univariate analysis			Multivariate analysis
	OR	95%CI	P value	OR	95%CI	P value
FIB-4 (a.u.)	2.16	1.41-3.33	0.004	5.53	1.72-17.70	0.004
Sex (female/male)	0.61	0.28-1.33	0.215	-	-	-
Arterial hypertension (yes/no)	5.19	2.24-12.04	< 0.001	3.84	1.42-10.36	0.008
Previous history of CVD (yes/no)	9.29	1.93-44.76	0.005	-	-	-
Current smoking (yes/no)	2.57	1.02-6.47	0.044	3.89	1.21-12.50	0.022
Diabetes (yes/no)	1.19	0.41-3.46	0.753	-	-	-
LDL-c (mg/dL)	0.98	0.97-0.99	0.007	-	-	-
GFR (mL/minute/1.73 m2)	0.96	0.94-0.99	0.007	-	-	-
LDH (U/L)	1.01	1.00-1.02	0.033	-	-	-
Albumin (g/dL)	0.72	0.60-0.87	< 0.001	0.76	0.61-0.95	0.016
AHI (events/hour)	1.00	0.98-1.01	0.633	-	-	-
ODI (events/hour)	1.01	0.99-1.03	0.292	-	-	-
Tc90%	1.00	0.99-1.02	0.586	-	-	-

Multivariate model was built following forward introduction of the variables selected by likelihood ratio test. Hosmer-Lemshow test was performed to evaluate goodness of fit of the model (P = 0.883). OR: Odds ratio; CI: Confidence interval; NFS: NAFLD fibrosis score; a.u.: Arbitrary units; CVD: Cardiovascular disease; LDL-c: Low-density lipoprotein cholesterol; GFR: Glomerular filtration rate; LDH: Lactate dehydrogenase; AHI: Apnea-hypopnea index; ODI: Oxygen desaturation index; Tc90%: Percentage of sleep time with oxygen saturation less than 90%.

Table 7 Univariate and multivariate analysis of the framingham steatosis index and the independent variables associated with carotid atherosclerosis in patients with obstructive sleep apnea (n = 112).

Independent variables	Univariate analysis			Multivariate analysis
	OR	95%CI	P value	OR	95%CI	P value
FSI (a.u.)	4.31	0.95-19.51	0.058	-	-	-
Previous history of CVD (yes/no)	9.29	1.93-44.76	0.005	14.09	2.52-78.92	0.003
Current smoking (yes/no)	2.57	1.02-6.47	0.044	-	-	-
LDL-c (mg/dL)	0.98	0.97-0.99	0.007	-	-	-
GFR (mL/minute/1.73 m2)	0.96	0.94-0.99	0.007	-	-	-
LDH (U/L)	1.01	1.00-1.02	0.033	1.02	1.00-1.03	0.014
Albumin (g/dL)	0.72	0.60-0.87	< 0.001	0.71	0.58-0.88	0.002
Platelets (109/L)	0.99	0.98-1.00	0.016	-	-	-
AHI (events/hour)	1.00	0.98-1.01	0.633	-	-	-
ODI (events/hour)	1.01	0.99-1.03	0.292	-	-	-
Tc90%	1.00	0.99-1.02	0.586	-	-	-

Multivariate model was built following forward introduction of the variables selected by likelihood ratio test. Hosmer-Lemshow test was performed to evaluate goodness of fit of the model (P = 0.108). OR: Odds ratio; CI: Confidence interval; a.u.: Arbitrary units; FSI: Framingham steatosis index; CVD: Cardiovascular disease; LDL-c: Low-density lipoprotein cholesterol; GFR: Glomerular filtration rate; LDH: Lactate dehydrogenase; AHI: Apnea-hypopnea index; ODI: Oxygen desaturation index; Tc90%: Percentage of sleep time with oxygen saturation less than 90%.

Diagnostic accuracy of NFS to predict carotid atherosclerosis

To evaluate the diagnostic performance of the scores, receiver operating characteristic curve analyses of NFS and FIB-4 to predict carotid atherosclerosis were performed. The analysis revealed an area under the curve (AUC) = 0.701 (95%CI: 0.603-0.799) (P < 0.001) for NFS. On the other hand, FIB-4 showed mildly worse results to predict carotid atherosclerosis (Figure 3A). At the established cut off point of -1.455, NFS is able to rule out the presence of carotid plaques with 75.6% sensitivity and 78.0% of negative predictive value. Together with arterial hypertension and current smoking status, AUC of NFS increases to 0.785 (95%CI: 0.698-0.873) (P < 0.001) (Figure 3B).

Figure 3 Diagnostic accuracy of non-alcoholic fatty liver disease fibrosis score to predict carotid atherosclerosis in patients with obstructive sleep apnea. A: Receiver operating characteristic curve for non-alcoholic fatty liver disease fibrosis score and fibrosis-4 to predict carotid atherosclerosis in patients with obstructive sleep apnea (n = 112); B: Receiver operating characteristic curve for the non-alcoholic fatty liver disease fibrosis score model along with arterial hypertension and current smoking status to predict carotid atherosclerosis in patients with obstructive sleep apnea (n = 112). NFS: Non-alcoholic fatty liver disease fibrosis score; FIB-4: Fibrosis-4; AUROC: Area under the receiver operating characteristic curve.

NFS predicts carotid atherosclerosis regardless of MASLD

Since liver fibrosis scores such as NFS and FIB-4 score have been previously related to atherosclerotic outcomes in MASLD populations, we next analyzed whether these liver fibrosis scores are associated with vascular outcomes in OSA patients regardless of the presence of MASLD diagnosed by the lipidomic test termed OWLiver™ (Figure 4). Only 16 of the 112 patients with OSA in our cohort were classified as normal liver, and only 6 of them did not have carotid plaques. On the one hand, we observed a trend to higher values of NFS in OSA patients with carotid atherosclerosis compared to controls, although it did not achieve statistical significance (P = 0.051). On the other hand, NFS significantly correlated with carotid plaque volume only in those OSA patients with carotid atherosclerosis (r = 0.665, P = 0.041), while FIB-4 score did not (r = -0.077, P = 0.834).

Figure 4 Non-alcoholic fatty liver disease fibrosis score is related to carotid plaque presence and volume in patients with obstructive sleep apnea regardless of metabolic dysfunction-associated steatotic liver disease presence. All values are expressed as mean ± SD. A and B: Non-alcoholic fatty liver disease fibrosis score and fibrosis-4 scores in obstructive sleep apnea patients without metabolic dysfunction-associated steatotic liver disease according to carotid plaque absence (n = 6) or presence (n = 10); C and D: Spearman correlations for non-alcoholic fatty liver disease fibrosis score and fibrosis-4 scores and carotid plaque volume in obstructive sleep apnea patients without metabolic dysfunction-associated steatotic liver disease and carotid plaque presence (n = 10). NFS: Non-alcoholic fatty liver disease fibrosis score; FIB-4: Fibrosis-4.

DISCUSSION

CVD remains the leading cause of disease burden and mortality globally[14,15], with atherosclerosis being the primary underlying pathological mechanism[34]. While CVD mortality continues to rise, particularly in low- and middle-income countries[35], it already represents the first cause of death among patients with MASLD[16,17]. Moreover, MASLD is projected to emerge as the leading cause of end-stage liver disease in the coming decades[36]. Given this landscape, there are a number of clinical studies pointing out that high values of some blood-based algorithms, such as FIB-4 and NFS among others, are associated with increased vascular atherosclerotic damage in patients with CVD[25], as well as in MASLD patients[24,26], but little is known about the significance of these blood-based algorithms as predictors of carotid atherosclerosis in patients with OSA, a highly prevalent chronic respiratory disease commonly associated with cardiometabolic disorders such as ischemic heart disease, stroke and MASLD[2,3,5,6]. In this study, we have assessed for the first time the performance of these scores to predict vascular risk in patients with OSA. In the last years, there has been quite controversy about the presence of increased atherosclerosis and/or vascular risk in patients with OSA, with studies reporting conflicting results, probably because of the different effects of OSA depending on its severity. In this regard, several studies have reported the relationship between the severity of the apnea and the presence of atherosclerotic vascular damage, largely in moderate to severe OSA patients[9,10,37,38], while others reported lack of link between OSA severity and increased inflammation or subclinical carotid atherosclerosis[39]. The findings of the present study are in line with the latter, since we did not find any polygraphic parameter related with the prevalence of carotid atherosclerotic plaque(s) in OSA patients and this might be related to distinct features of our study cohort, either due to the significant presence of mild OSA, or to the absence of patients with OS. When metabolic characteristics were assessed, we found that liver steatosis score FSI as well as liver fibrosis scores FIB-4 and NFS were increased in OSA patients with carotid plaques compared to those without carotid plaques. We built multivariate logistic regression analysis models in order to study the association between polygraphic criteria and all the variables that differed in OSA patients with or without carotid atherosclerosis. On the one hand, multivariate analysis revealed that besides arterial hypertension and current smoking status, liver fibrosis scores FIB-4 and NFS were independent predictors of carotid atherosclerosis in patients with OSA. Receiver operating characteristic curve analysis confirmed that AUC was slightly better for NFS than for FIB-4 (0.701 vs 0.683). Additionally, at the established cut off point of -1.455, NFS is able to rule out carotid atherosclerosis with 75.6% sensitivity and 78.0% of negative predictive value. Together with arterial hypertension and current smoking status, AUC of NFS increases to 0.785 (95%CI: 0.698-0.873) (P < 0.001). Although the individual positive predictive value of this score is moderate (55%), its high negative predictive value (78%) supports its use as a first-line stratification tool to identify patients who may not require urgent ultrasound evaluation, thereby optimizing clinical resources. When FIB-4 was used to build the models, albumin was retained as a significant variable for prediction of carotid plaques, however, it was not included along with NFS because it is already included within the algorithm. When we analyzed FIB-4 and NFS only in patients without MASLD, we found that NFS showed a trend to increase in patients with OSA and carotid atherosclerosis while FIB-4 did not. Moreover, NFS also correlated with carotid plaque volume in patients with OSA and absence of MASLD, suggesting that NFS is related to carotid atherosclerosis independently of MASLD. However, it is crucial to clearly distinguish between the validated role of NFS as a predictor of liver fibrosis and its extrapolation here as a cardiovascular risk stratification tool. The association found in our study suggests that NFS may be capturing a broader phenotype of “systemic vascular aging” rather than liver fibrosis alone. This biological plausibility is supported by the individual components of the NFS algorithm, which are shared metabolic and inflammatory risk factors for atherosclerosis. Specifically, age and hyperglycemia (diabetes) are well-established drivers of arterial stiffness and endothelial dysfunction. On the other hand, regarding liver steatosis scores, only FSI was increased in OSA patients with carotid plaque(s) when compared to those without plaque. However, FSI did not resist adjustment with other variables in the multivariate models. Sample size is the main limitation of our study, and further investigation is required to confirm these results. Another important limitation is the lack of specific data on statin therapy, although our database did account for other treatments such as angiotensin-converting enzyme inhibitors, angiotensin receptor blockers, and antidiabetic agents. This lack of statin data represents a potential confounding factor by indication bias, which would explain the paradoxical inverse association observed between low-density lipoprotein cholesterol levels and carotid plaque presence in our cohort. Another potential limitation of our study is the use of the OWLiver™ test as the reference to exclude MASLD presence. Although this lipidomic-based assay offers superior diagnostic accuracy compared to traditional non-invasive scores, it remains a surrogate marker and not a histological gold standard. On the other hand, it is important to note that the majority of clinical studies assessing the vascular risk in OSA patients did not take into account the presence of COPD or OS. Additionally, several studies have been carried out without performing complete lung function tests. Thus, the main strength of the present study relies on the stringent exclusion criteria applied as well as in the detailed characterization of respiratory, vascular and hepatic parameters of our study population. Our findings provide novel evidence beyond conventional risk estimation provided by a classical comprehensive assessment of traditional risk factors such as arterial hypertension or cigarette smoking. Vascular risk prediction may differ in patients with OSA, largely due to the high prevalence of metabolic comorbidities. Thus, new approaches need to be validated to take care of high-risk populations.

CONCLUSION

NFS is an algorithm very easy to calculate in daily clinical practice, because it is based on simple laboratory parameters, it does not require expensive ultrasound equipment and does not need the presence of a radiologist. Hence, we propose NFS as a useful initial tool to screen OSA patients in clinical practice in order to identify those patients at high risk of carotid atherosclerosis to whom an in-depth evaluation in specialized hospital units must be recommended.