Abstract
Objectives
Obstructive sleep apnea (OSA) is an independent risk factor for non-alcoholic fatty liver disease. This study was planned to assess proportion of patients with OSA that have hepatic steatosis and fibrosis, as measured by transient elastography, to explore variables influencing their development and to find out the polysomnography parameters that predict the need for transient elastography screening in OSA.
Methods
Consecutive participants having polysomnography proven OSA were included in the study after screening for eligibility criteria. Data of the polysomnography were scored manually following standard criteria. Participants were subjected to transient elastography (Fibroscan®) and serum investigations after diagnostic polysomnography. The polysomnography, fibroscan®, and laboratory data were tabulated and analyzed.
Results
A total of 71 participants were enrolled. 16.9% participants had mild OSA, 28.2% had moderate OSA, and remaining participants had severe OSA. Liver steatosis was found in 63.4% participants while hepatic fibrosis was noted in 9.9%. Oxygen desaturation index (ODI), apnea-hypopnea index (AHI), and percentage of sleep spent below 90% oxygen saturation (T90) were significantly associated with the presence of hepatic steatosis and fibrosis. Receiver operating curve (ROC) showed that at the cut-offs of 73 events/hr, 13% and 72.2 events/hr, AHI, T90 and ODI, predicted hepatic fibrosis with area under ROC of 0.960, 0.944, and 0.933, respectively (P < 0.001).
Conclusions
Patients with OSA are at increased risk for development of hepatic steatosis and fibrosis. ODI, AHI, and T90 during polysomnography predict the presence of underlying hepatic fibrosis.
Keywords: obstructive sleep apnea, polysomnography, non-alcoholic fatty liver disease, hepatic fibrosis, transient elastography
Non-alcoholic fatty liver disease (NAFLD) has evolved into global epidemic and the leading cause of chronic liver disease in many regions of the world. Non-alcoholic fatty liver (NAFL), non-alcoholic steatohepatitis (NASH), fibrosis, cirrhosis, and hepatocellular carcinoma constitute the NAFLD disease spectrum. Pathogenesis of NAFLD and NASH has not been completely understood. A number of factors have been identified that enhance the risk of NAFLD. Modifiable risk factors include obesity, insulin resistance, dyslipidemia, intestinal microbiota, oxidative stress, and metabolic syndrome, while non-modifiable factors constitute age, gender, ethnicity, and familial predisposition.1
Obstructive sleep apnea, also known as OSA, is a prevalent sleep disorder that has been linked to a cluster of metabolic conditions that are grouped together under the umbrella term of “metabolic syndrome.” These metabolic conditions include hypertension, diabetes, obesity, and dyslipidemia.2 It is, therefore, unsurprising that OSA has been found to be associated with NAFLD. OSA has been postulated to consequentially lead to the complicated metabolic abnormalities observed in NAFLD.3 Chronic intermittent hypoxia (CIH), a characteristic of OSA, leads to tissue hypoxia, which consequentially amounts to the production of reactive oxygen species. This leads to the activation of NF-κB pathway, thereby resulting in oxidative stress and inflammation. The occurrence of hypoxia also leads to stimulation of carotid body receptors, sympathetic nervous system activation, and lipolysis, which ultimately results in hepatic steatosis.3,4 CIH has also been demonstrated to cause insulin resistance, malfunction of crucial processes in hepatic lipid metabolism, as well as hepatic steatosis and fibrosis.3,4 Each of these factors contribute to the development and progression of NAFLD.4 Figure 1 depicts the current understanding of the pathogenesis of NAFLD in OSA.
Figure 1.
Pathogenesis of non-alcoholic fatty liver disease in obstructive sleep apnea.
Few studies have been conducted assessing the relationship between OSA and liver disease but each has their own limitations.5, 6, 7, 8 Earlier studies exclusively enrolled morbidly obese patients, excluding subjects with other factors causing OSA such as oro-mandibulo-facial abnormalities.5, 6, 7 Metabolic syndrome, being a key driver of NAFLD pathogenesis, stood as an important confounder in these studies. Multiple previous studies relied on clinical- and questionnaire-based diagnosis of OSA, which has poor diagnostic value in reliably detecting the presence of OSA. Prior research also employed the use of liver biopsies to look for histopathological evidence of NAFLD.6,7 Liver biopsy has been the gold standard reference for assessing liver fibrosis but has its own inherent shortcomings viz., invasiveness, sampling variability, suboptimal interobserver reproducibility, high cost, and high risk of complications which restricts its frequent use.9 Later studies utilized fibrosis-4 index (FIB-4), NAFLD fibrosis score, aspartate aminotransferase to platelets ratio index, Body Mass Index, Aspartate aminotransferase/ Alanine aminotransferase ratio, Diabetes Mellitus (BARD) score, and ultrasonography to counter problem of invasiveness but fell short in accurately quantifying liver fibrosis since these tools have poor sensitivities and specificities juxtaposed to transient elastography, a novel method of assessing hepatic steatosis and fibrosis.10
However, Trzepizur et al. assessed liver stiffness using transient elastography in patients with OSA and found a dose response relationship between severity of OSA and liver stiffness values. The limitation posed by their study was the inclusion of subjects with clinical suspicion of OSA without polysomnography (PSG) proven diagnosis, and with the presence of at least one criterion of metabolic syndrome, which renders the study findings applicable to solely presumptive OSA patients with metabolic comorbidities.8 However, most of the studies so far have been conducted among Caucasian population. Due to the genetic, phenotypic, and lifestyle differences, it is possible that Indian population may differ with regards to the development of NAFLD in association with OSA.11 Only one study having small sample size (23 subjects) studied the association between OSA and NAFLD in Indian population and found a significant link between severity of OSA and development of hepatic fibrosis.12 Lastly, it is unknown if PSG characteristics aid in screening patients for NAFLD and, if so, at what thresholds.
The present study was devised to overcome methodological limitations of previous studies. The principal objective of this observational study is to find out the proportion of participants with OSA who have NAFLD, as measured by transient elastography, to find out variable associated to the development of hepatic steatosis and fibrosis, and lastly, to identify the predictive ability of polysomnographic variables for detecting hepatic steatosis and fibrosis.
PARTICIPANTS AND METHODS
Patient Consent Statement
Written informed consent was obtained from participants prior to enrollment in the study.
Ethics Approval Statement
This study was approved by the Institutional Ethics Committee, All India Institute of Medical Sciences, Rishikesh (Reference no. 509/IEC/PGM/2020). The study conforms to ethical standards according to Declaration of Helsinki. Written informed consent was obtained from participants prior to enrollment in the study.
Study Setting and Oversight
A single center, cross-sectional, observational study was conducted over a period of 12 months. The study was approved by the Institutional Ethics Committee, All India Institute of Medical Sciences, Rishikesh (Reference no. 509/IEC/PGM/2020). Written informed consent was obtained from participants prior to enrollment in the study.
Participants
Consecutive adult (≥18 years of age) participants with PSG-proven (based on International Classification of Sleep Disorders – 3 criteria) OSA were enrolled from Division of Sleep Medicine.13 However, participants having any other risk factors for chronic liver disease apart from NAFLD, the presence of secondary causes of NAFLD (such as drugs, jejunoileal bypass or extensive small bowel resection or total parenteral nutrition), individuals consuming more than 20 g of alcohol per day (as per history confirmed by two relatives) and non-consenting individuals were excluded.
Sample Size Calculation
To determine the appropriate sample size for our study, we used the formula for calculating sample size in a proportion estimation study with a 10% margin of error and 95% confidence level. According to one prior study in India, the prevalence of NAFLD in OSA patients is 91.3%. Therefore, the formula is:
n = (Z2 ∗ P ∗ (1-P))/E2where: n = required sample size; Z = Z-score (1.96 for a 95% confidence level); P = prevalence of NAFLD in OSA patients (0.913, according to older studies); E = margin of error (0.10).
Plugging in the values, we have:
n = (1.96ˆ2 ∗ 0.913 ∗ (1-0.913))/0.10ˆ2
n ≈ (3.8416 ∗ 0.913 ∗ 0.087)/0.01
n ≈ 30.7819.
Since the sample size must be a whole number, we'll round up to the nearest integer:
n ≈ 31.
So, a sample size of 31 patients is sufficient to achieve a 95% confidence level with a 10% margin of error in the study.
Study Procedure
A comprehensive clinical history was obtained, as well as anthropometric data such as height, weight, body mass index (BMI), and waist circumference.14 All of these were documented in accordance with conventional guidelines.15 Fasting venous blood was drawn from antecubital vein following aseptic procedure for the assessment of hemogram (Coulter 750 Hematology Analyzer, Beckman Coulter Inc, USA), liver function tests including aspartate aminotransferase (AST), alanine aminotransferase (ALT), gamma-glutaryl transferase, and alkaline phosphatase (ALP) (AU 680 Clinical Chemistry Analyzer, Beckman Coulter Inc, USA), fasting blood sugar, fasting plasma insulin (ADVIA Centaur CP Immunoassay System, SIEMENS, USA), lipid profile, HbA1c (TOSOH Automated Glycohemoglobin Analyzer (HLC-723G8), TOSOH Corporation, Japan), and markers of hepatitis B (Virucheck HbsAg, Viola Diagnostic Systems, India), hepatitis C (Qualpro HCV, Qualpro Diagnostics, India), and human immunodeficiency virus (Qualpro HIV, Qualpro Diagnostics, India). National Cholesterol Education Program-Adult Treatment Panel III criteria modified for Asians was used to diagnose metabolic syndrome.16 All patients were assessed for the presence of metabolic syndrome and also underwent a FibroScan® evaluation to look for the presence of hepatic steatosis and fibrosis, which constitute the spectrum of NAFLD. Patients were graded according to severity of steatosis as follows: S0, no steatosis (<262 dB/m); S1, mild steatosis (262 – <295 dB/m); S2, moderate steatosis (295– <324 dB/m); S3, severe steatosis (≥324 dB/m), and on the basis of liver stiffness into three groups—no significant fibrosis (Liver stiffness measurement (LSM) <6.0 kPa), significant fibrosis (LSM ⩾8.2 kPa), and advanced fibrosis (LSM ⩾9.7 kPa) as per the latest Indian national association for study of liver (INASL) guidelines.17
Polysomnography
Overnight attended in-laboratory PSG diagnostic study was performed using the Somnoscreen (SOMNOmedics GmBH, Germany) after giving time to the participants for acclimatization to the sleep laboratory. All parameters recommended by the manual for scoring of PSG data were recorded. Six channels of electroencephalogram (F3-M2, F4-M1, C3-M2, C4-M1, O1-M2, O2-M1), two channels of electrooculogram (right and left), and two channels of submental electromyogram (right and left) were recorded for the ascertainment of sleep stages. The sensitivity of channels, sampling rates, and filters for all channels were set according to prevailing guidelines.18 Cardiac parameters included lead II of electrocardiogram, and for the leg movement, surface electromyogram was recorded from both sides of anterior tibialis muscle. Respiratory parameters included measurement of nasal airflow through pressure transducer and thermistor, respiratory efforts using thoracic and abdominal respiratory inductance plethysmography belts, and overnight pulse oximetry using average of 3 s. Body position was assessed using a body position sensor. Scoring of the data was done manually according to standard guidelines by a certified sleep physician and trained sleep technician.18 All indicators and indices recommended by the manual for the scoring of PSG data were calculated. Of all sleep-related parameters obtained after the scoring of data, four parameters, i.e. apnea-hypopnea index (AHI: defined as number of apnea and hypopneas per hour of sleep), percentage of time with oxyhemoglobin saturation percentage below 90% (T90), oxygen desaturation index (ODI defined as number of times the blood oxygen desaturation fell below 3% from the baseline per hour of sleep) and sleep efficiency (percentage of time spent in sleep/time spent in bed during recording) were considered as predictor variables for the steatosis and hepatic fibrosis.8,19
FibroScan®
The procedure was performed using an Echosens FibroScan® 502 Touch device (Paris, France) by a manufacturer-certified FibroScan® technician. The patients were instructed to fast for at least 2 h. The choice of probe was made using an automatic probe selection tool that takes into account the patient's morphology and automatically measures the skin–liver capsule (SCD) distance. As the patient lay supine, the ultrasonic probe was pushed against the skin (intercostal area) overlaying the liver. The probe produces a vibration and measures the velocity of the shear wave as it spreads through the liver. The device then computes the median values of continuous attenuation parameter (in decibels per meter) and liver stiffness (in kilopascals, kPa). The goal is to acquire ten quantitative measurements, with at least a 60% success rate and an interquartile range of <30% of the median value.
Statistical Analysis
The data were tabulated in Excel sheet (Microsoft Corporation. Microsoft Excel [Internet], 2018, available from: https://office.microsoft.com/excel) and analyzed with the help of SPSS software (IBM Corp. Released 2011. IBM SPSS Statistics for Windows, Version 28.0. Armonk, NY: IBM Corp). Descriptive statistics were computed. A P-value of <0.05 was considered significant. The presence of steatosis and fibrosis were the primary dependant variables. The primary independent variables were PSG parameters—AHI, ODI, T90, and sleep efficiency. Other independent variables included serum investigations, presence of diabetes, presence of metabolic syndrome, homeostasis model assessment for insulin resistance (HOMA-IR), and age. Binary logistic regression was performed to study the independent variables influencing the development of hepatic fibrosis after accounting for all confounding variables—age, gender, diabetes, metabolic syndrome, HOMA-IR. Various tests of significance were used as per data in reference. A receiver operating curve (ROC) analyses was performed to study the predictive value of PSG parameters for predicting hepatic fibrosis.
RESULTS
Characteristics of Patients
A total of 139 PSG-proven OSA patients were screened, out of which 71 met the inclusion criteria and were subjected to evaluation (Figure 2). 68 participants were excluded—16 subjects did not give consent, 12 subjects were below 18 years of age, 36 subjects had history of significant alcohol consumption, and 4 subjects had history of chronic liver disease due to other causes. Male participants outnumbered female participants (63.4% males vs. 36.6% females). 66 (93%) subjects were above the normal BMI range for the Asians.14 The subjects enrolled predominantly consisted of those with severe OSA (54.9%). Other characteristics of participants are depicted in Table 1.
Figure 2.
Flowchart of patient flow in the study.
Table 1.
Characteristics of Study Subjects.
| Mean ± SD | |
|---|---|
| Age (Years) | 48.77 ± 12.25 |
| BMI (kg/m2) | 29.81 ± 6.32 |
| n (%) | |
| Gender | |
| Male | 45 (63.4%) |
| Female | 26 (36.6%) |
| BMI | |
| <18.5 kg/m2 | 1 (1.4%) |
| 18.5–22.9 kg/m2 | 4 (5.6%) |
| 23.0–24.9 kg/m2 | 12 (16.9%) |
| 25.0–29.9 kg/m2 | 24 (33.8%) |
| 30.0–34.9 kg/m2 | 17 (23.9%) |
| 35.0–39.9 kg/m2 | 6 (8.5%) |
| 40.0–44.9 kg/m2 | 6 (8.5%) |
| >45.0 kg/m2 | 1 (1.4%) |
| Severity of OSA | |
| Mild | 12 (16.9%) |
| Moderate | 20 (28.2%) |
| Severe | 39 (54.9%) |
| Median (IQR) | |
| Apnea-hypopnea Index (AHI) | 32.60 (18.35–58.45) |
| ODI | 32.80 (20.15–65.30) |
| T90 (%) | 2.80 (0.20–11.80) |
| Sleep Efficiency (%) | 75.70 (66.40–86.80) |
AHI, Apnea-hypopnea Index; BMI, Body mass index; IQR, Inter-quartile range; ODI, Oxygen desaturation index; OSA, Obstructive sleep apnea.
FibroScan® Results
Liver FibroScan® variables and outcomes have been mentioned in Table 2. 63.4% subjects were observed to have steatosis with most harboring S3 grade steatosis. Fibrosis was present in 7 (9.9%) subjects.
Table 2.
FibroScan® Variables and Outcomes.
| n (%) | |
|---|---|
| Steatosis (Present) | 45 (63.4%) |
| Steatosis Grade | |
| S0 | 26 (36.6%) |
| S1 | 15 (21.1%) |
| S2 | 9 (12.7%) |
| S3 | 21 (29.6%) |
| Fibrosis Grade | |
| No significant fibrosis | 64 (90.1%) |
| Significant fibrosis | 4 (5.6%) |
| Advanced fibrosis | 3 (4.2%) |
| Fibrosis (Present) | 7 (9.9%) |
Association of Hepatic Steatosis and Fibrosis With Other Variables
In the univariate analysis, BMI, metabolic syndrome, AHI, ODI, T90 (%), fasting plasma insulin, HOMA-IR, total cholesterol (mg/dL), and total LDL (mg/dL) levels were associated with hepatic steatosis (Table 3) while BMI, metabolic syndrome, diabetes, severity of OSA, AHI, ODI, T90 (%), fasting plasma insulin, HOMA-IR, total cholesterol (mg/dL), and total LDL (mg/dL) levels were associated with hepatic fibrosis (Table 4).
Table 3.
Association Between Hepatic Steatosis and Other Parameters.
| Parameters | Steatosis |
P value | |
|---|---|---|---|
| Present (n = 58) | Absent (n = 13) | ||
| Age (Years) | 48.76 ± 12.06 | 48.81 ± 12.81 | 0.9871 |
| Gender | 0.4373 | ||
| Male | 27 (60.0%) | 18 (69.2%) | |
| Female | 18 (40.0%) | 8 (30.8%) | |
| BMI (kg/m2)∗∗∗ | 31.34 ± 6.45 | 27.16 ± 5.21 | 0.0094 |
| Metabolic Syndrome ∗∗∗ | 32 (55.2%) | 2 (15.4%) | 0.0091 |
| Diabetes | 18 (31.0%) | 1 (7.7%) | 0.1622 |
| Severity of OSA | 0.0793 | ||
| Mild | 5 (11.1%) | 7 (26.9%) | |
| Moderate | 11 (24.4%) | 9 (34.6%) | |
| Severe | 29 (64.4%) | 10 (38.5%) | |
| AHI∗∗∗ | 47.25 ± 27.01 | 28.75 ± 18.27 | 0.0054 |
| ODI∗∗∗ | 51.13 ± 29.09 | 30.48 ± 19.20 | 0.0054 |
| T90 (%)∗∗∗ | 14.95 ± 22.48 | 3.24 ± 5.40 | 0.0034 |
| Sleep Efficiency (%) | 75.76 ± 13.63 | 74.84 ± 13.15 | 0.7811 |
| AST (U/L)∗∗∗ | 50.73 ± 38.56 | 31.85 ± 13.85 | 0.0074 |
| ALT (U/L) | 56.84 ± 38.71 | 38.92 ± 15.91 | 0.1094 |
| ALP (U/L)∗∗∗ | 122.47 ± 45.52 | 100.54 ± 43.61 | 0.0384 |
| GGT (U/L) | 35.87 ± 18.84 | 30.46 ± 17.74 | 0.3334 |
| Fasting plasma insulin (mU/L)∗∗∗ | 20.99 ± 12.34 | 10.27 ± 6.81 | <0.0014 |
| HOMA-IR∗∗∗ | 133.54 ± 84.76 | 61.71 ± 41.09 | 0.0014 |
| Fasting blood glucose (mmol/dL) | 7.97 ± 12.40 | 5.98 ± 0.80 | 0.7244 |
| Total Cholesterol (mg/dL)∗∗∗ | 213.31 ± 55.22 | 176.42 ± 34.19 | 0.0054 |
| Total HDL (mg/dL) | 47.72 ± 9.41 | 47.81 ± 7.99 | 0.9691 |
| Total LDL (mg/dL)∗∗∗ | 138.73 ± 37.92 | 118.23 ± 29.05 | 0.0214 |
| Serum Triglyceride (mg/dL) | 163.98 ± 69.56 | 161.73 ± 43.21 | 0.8954 |
∗∗∗Significant at P < 0.05.
AHI, Apnea-hypopnea Index; ALP, Alkaline phosphatase; ALT, Alanine aminotransferase; AST, Aspartate aminotransferase; BMI, Body mass index; GGT, Gamma-glutaryl transferase; HDL, High densoty lipoprotein; HOMA-IR, Homeostasis model assessment-estimated insulin resistance; LDL, Low density lipoprotein; ODI, Oxygen desaturation index; OSA, Obstructive sleep apnea.
t-test.
Fisher's Exact Test.
Chi-Squared Test.
Wilcoxon-Mann-Whitney U Test.
Table 4.
Association Between Hepatic Fibrosis and Other Parameters.
| Parameters | Fibrosis |
P value | |
|---|---|---|---|
| Present (n = 9) | Absent (n = 62) | ||
| Age (Years) | 48.86 ± 12.84 | 48.77 ± 12.29 | 0.7801 |
| Gender | 1.0002 | ||
| Male | 5 (71.4%) | 40 (62.5%) | |
| Female | 2 (28.6%) | 24 (37.5%) | |
| BMI (kg/m2)∗∗∗ | 35.84 ± 7.56 | 29.15 ± 5.87 | 0.0261 |
| Metabolic Syndrome ∗∗∗ | 6 (88.9%) | 26 (41.9%) | 0.0112 |
| Diabetes∗∗∗ | 6 (66.7%) | 13 (21.0%) | 0.0092 |
| Severity of OSA∗∗∗ | 0.0492 | ||
| Mild | 0 (0.0%) | 12 (18.8%) | |
| Moderate | 0 (0.0%) | 20 (31.2%) | |
| Severe | 7 (100.0%) | 32 (50.0%) | |
| AHI∗∗∗ | 81.59 ± 7.73 | 35.98 ± 22.75 | <0.0011 |
| ODI∗∗∗ | 84.27 ± 14.85 | 39.11 ± 24.97 | <0.0011 |
| T90 (%)∗∗∗ | 40.87 ± 26.90 | 7.36 ± 14.80 | <0.0011 |
| Sleep Efficiency (%) | 67.46 ± 10.32 | 76.29 ± 13.44 | 0.0701 |
| AST (U/L) | 80.29 ± 68.40 | 39.83 ± 24.46 | 0.3581 |
| ALT (U/L) | 75.71 ± 60.28 | 47.49 ± 28.37 | 0.9921 |
| ALP (U/L) | 127.43 ± 33.59 | 113.02 ± 46.90 | 0.2511 |
| GGT (U/L) | 42.29 ± 26.49 | 32.97 ± 17.46 | 0.5891 |
| Fasting plasma insulin (mU/L)∗∗∗ | 34.47 ± 8.35 | 15.16 ± 10.53 | <0.0011 |
| HOMA-IR∗∗∗ | 213.77 ± 47.69 | 95.58 ± 73.63 | 0.0011 |
| Fasting blood glucose (mmol/dL) | 18.09 ± 31.27 | 6.05 ± 0.98 | 0.0711 |
| Total Cholesterol (mg/dL)∗∗∗ | 236.29 ± 34.93 | 195.81 ± 51.66 | 0.0431 |
| Total HDL (mg/dL) | 45.43 ± 7.32 | 48.01 ± 9.02 | 0.5361 |
| Total LDL (mg/dL)∗∗∗ | 166.57 ± 24.56 | 127.36 ± 35.18 | 0.0061 |
| Serum Triglyceride (mg/dL) | 136.57 ± 46.57 | 166.06 ± 61.87 | 0.1861 |
∗∗∗Significant at P < 0.05.
AHI, Apnea-hypopnea Index; ALP, Alkaline phosphatase; ALT, Alanine aminotransferase; AST, Aspartate aminotransferase; BMI, Body mass index; GGT, Gamma-glutaryl transferase; HOMA-IR, Homeostasis model assessment-estimated insulin resistance; LDL, Low density lipoprotein; ODI, Oxygen desaturation index; OSA, Obstructive sleep apnea.
Wilcoxon-Mann-Whitney U Test.
Fisher's Exact Test.
Adjustment for Confounders
Three models were developed to ascertain the association of hepatic fibrosis with variables that reached statistical significance during univariate analysis (Table 5). Each PSG parameter which was significantly associated with hepatic fibrosis was studied along with other confounder variables associated with hepatic fibrosis as divulged in this study and previous studies, including HOMA-IR, metabolic syndrome and diabetes. Model A studied AHI along with other variables and correctly predicted hepatic fibrosis in 94.4% subjects (P value < 0.001) and found AHI to be significantly associated with hepatic fibrosis independent of other variables, with 76.4% variance in the results (Nagelkerke R square 0.764). Model B studied ODI along with other variables and correctly predicted hepatic fibrosis in 91.5% subjects (P value < 0.001) and found ODI to be significantly associated with hepatic fibrosis after adjusting for confounders, with 56.5% variance in the results (Nagelkerke R square 0.565). Model C studied T90 along with other variables and correctly predicted hepatic fibrosis in 94.4% subjects (P value < 0.001) and found none of the variables to be significantly associated with hepatic fibrosis independent of other variables, with 71.7% variance in the results (Nagelkerke R square 0.717).
Table 5.
Binary Logistic Regression of PSG Variables With Potential Confounders.
| Model | Variables | B | S.E. | Wald | dF | Sig. | Odds Ratio |
|---|---|---|---|---|---|---|---|
| A | AHI | 0.136 | 0.060 | 5.182 | 1 | 0.023 | 1.146 |
| Metabolic Syndrome | 2.505 | 2.671 | 0.880 | 1 | 0.348 | 12.247 | |
| Diabetes | −0.117 | 1.821 | 0.004 | 1 | 0.949 | 0.889 | |
| HOMA-IR | 0.034 | 0.018 | 3.656 | 1 | 0.056 | 1.035 | |
| Constant | −19.398 | 8.415 | 5.313 | 1 | 0.021 | 0.000 | |
| B | ODI | 0.053 | 0.027 | 3.826 | 1 | 0.050 | 1.054 |
| Metabolic Syndrome | 1.986 | 1.671 | 1.413 | 1 | 0.235 | 7.287 | |
| Diabetes | −0.871 | 1.281 | 0.462 | 1 | 0.497 | 0.419 | |
| HOMA-IR | 0.015 | 0.009 | 2.951 | 1 | 0.086 | 1.015 | |
| Constant | −9.197 | 3.011 | 9.329 | 1 | 0.002 | 0.000 | |
| C | T90 (%) | 0.110 | 0.057 | 3.770 | 1 | 0.052 | 1.116 |
| Metabolic Syndrome | 2.569 | 2.459 | 1.092 | 1 | 0.296 | 13.058 | |
| Diabetes | −1.216 | 1.841 | 0.436 | 1 | 0.509 | 0.297 | |
| HOMA-IR | 0.057 | 0.033 | 2.931 | 1 | 0.087 | 1.059 | |
| Constant | −17.275 | 9.423 | 3.361 | 1 | 0.067 | 0.000 |
AHI, Apnea-hypopnea Index; HOMA-IR, Homeostasis model assessment-estimated insulin resistance; ODI, Oxygen desaturation index.
Predictive Ability of PSG Parameters for Hepatic Steatosis and Fibrosis
The area under the ROC curve (AUROC) for various PSG variables predicting the presence of hepatic fibrosis was analyzed (Figure 3). While comparing the various parameters, there was no significant difference in the diagnostic performance of ODI and T90 (%) (DeLong's Test P = 0.740), T90 (%) and AHI (DeLong's Test P = 0.0.553), and ODI and AHI (DeLong's Test P = 0.0.289). The diagnostic performance of ODI (Area under curve (AUC) = 0.933), T90 (%) (AUC = 0.944), and AHI (AUC = 0.960) was significantly better than that of sleep efficiency (%) (AUC = 0.711) with DeLong's Test P values of 0.012, 0.007/and 0.005, respectively. Although PSG variables exhibited excellent diagnostic performance in predicting fibrosis, their AUROC were low when assessed for prediction of steatosis, with poor diagnostic utility (Table 6).
Figure 3.
ROC Curve Analysis Showing Diagnostic Performance of apnea hypopnea index (AHI), Oxygen desaturation index (ODI), Percentage of sleep time spent below 90 percent saturation (T90) and Sleep efficiency (SE) in Predicting Hepatic Fibrosis.
Table 6.
Predictive Value of PSG Variables in Assessing Hepatic Steatosis.
| Variable | AUROC | Sensitivity | Specificity | PPV | NPV |
|---|---|---|---|---|---|
| AHI (Cutoff: 45.3 by ROC) | 0.700 | 57.8% (42–72) | 84.6% (65–96) | 86.7% (69–96) | 53.7% (37–69) |
| ODI (Cutoff: 52.7 by ROC) | 0.702 | 44.4% (30–60) | 92.3% (75–99) | 90.9% (71–99) | 49.0% (34–64) |
| T90 (%) (Cutoff: 3.6 by ROC) | 0.710 | 55.6% (40–70) | 84.6% (65–96) | 86.2% (68–96) | 52.4% (36–68) |
| Sleep Efficiency (%) (Cutoff: 73.8 by ROC) | 0.517 | 60.0% (44–74) | 80.8% (61–93) | 84.4% (67–95) | 53.8% (37–70) |
AHI, Apnea-hypopnea Index; NPV, Negative predictive value; ODI, Oxygen desaturation index; PPV, Positive predictive value; ROC, Receiver operating curve.
DISCUSSION
In this study, the prevalence of steatosis and fibrosis in subjects with OSA were 63.4% and 9.9%, respectively. A multitude of variables were associated with the presence of hepatic steatosis and fibrosis as mentioned in the results section. The present study demonstrated that patients with greater severity of OSA are at higher risk of developing hepatic fibrosis. AHI, ODI, and T90 were the PSG variables with the strongest association with hepatic fibrosis. This finding is consistent with previously published clinical studies and fundamental research demonstrating that intermittent hypoxia plays a crucial role in the hepatic effects of OSA.7,20
This study highlights the high prevalence of significant liver abnormalities, notably hepatic fibrosis, in OSA patients, especially in those with severe OSA with higher values of AHI, ODI, and T90. On histology, 81–100% of morbidly obese OSA patients were shown to have NAFLD in earlier investigations.5,6 In one such study, histopathological examination indicated NAFLD lesions, NAFLD activity score, and fibrosis to be consistently more severe in individuals with the higher ODI values.7 Trezipur et al. reported that the association between OSA and LSM values remained to be quite significant even after adjustment for such metabolic comorbidities and also divulged AHI and ODI as factors with the strongest independent association with liver stiffness measurements.8 Sookian and Pirola reported a greater prevalence of NAFLD in OSA patients, identified obesity as a confounding variable independently linked with both OSA and NAFLD, and conducted a metaregression analysis that revealed the association between OSA and NAFLD was independent of obesity.21 These results suggest that, in addition to the appropriate therapy for OSA, a considerable number of non-obese OSA patients also require the opinion of hepatologists for specific management. The study conducted on Indian population also divulged similar percentages with NAFLD being noted in 91.3% OSA patients (21 out of 23 subjects), with AHI attributed as the single important independent predictor of significant fibrosis in patients with NAFLD.12
Similar to previous studies, the presence of hepatic fibrosis in our patient cohort also significantly correlated with underlying comorbidities such as metabolic syndrome, diabetes, BMI, and HOMA-IR. This was clinically expected since metabolic syndrome is an umbrella with diabetes, NAFLD, and OSA being driven by similar interrelated pathophysiological mechanisms as previously described. Interestingly, it was observed that PSG variables were significantly associated with both hepatic steatosis and fibrosis but were excellent predictors only because of the presence of hepatic fibrosis. Also, once metabolic comorbidities were studied with PSG variables in regression analysis, only AHI and ODI were found to be significantly associated with hepatic fibrosis. These findings could probably be explained by the influence of other metabolic parameters like diabetes, HOMA-IR, and metabolic syndrome on the pathogenesis of NAFLD overall. OSA patients with NAFLD in our study also demonstrated the occurrence of dyslipidaemia with significantly higher values of serum total cholesterol, and serum LDL levels. In the present study, no significant correlation between liver enzymes (AST and ALT levels) and the presence of fibrosis was found, albeit older studies exhibiting higher mean values of AST and ALT in OSA patients.5,12
A novel finding from our study was the performance of various PSG parameters for predicting hepatic fibrosis. AHI (cut-off: 73 events/hr) was demonstrated to be the best parameter in terms of AUROC. AHI, ODI (cut-off: 72.2 events/hr), T90 (cut-off: 13%), and sleep efficiency (cutoff: 80.1%) had highest sensitivities and negative predictive value, while AHI (cut-off: 73 events/hr) displayed highest specificity and positive predictive value in predicting hepatic fibrosis in OSA subjects. These cut-off values for PSG variables may assist a physician in identifying high-risk OSA patients with propensity for the development of hepatic fibrosis. Further studies are warranted to validate these thresholds in a larger population of OSA patients for routine use in clinical practice and inclusion in guidelines.
An exponential increase in liver related mortality is witnessed with increasing fibrosis stage.22 Therefore, quantification of liver fibrosis in patients with NAFLD is a vital part of patient evaluation. Strength of this study is the use of transient elastography, the findings of which substantially correlate with hepatic steatosis and fibrosis, and have recently been demonstrated to be prognostic indicators in NAFLD patients.23 This is a compelling argument for the employment of this investigation in routine practice to determine the severity of NAFLD and identify the subset of patients who warrant further evaluation and treatment for NAFLD. Although TE has been said to have minimal inter-observer variability, LSM failure and inaccurate LSM values have been observed to occur in 3.1% and 15.8% of examinations, respectively.24,25 These events have been predominantly linked to high BMI (>30 kg/m2) and operator experience of fewer than 500 examinations.25 To solve this constraint, the Automatic Probe Selection tool, which automatically measures SCD and advises the probe to be used to acquire accurate LSM data, was employed in our work to solve this constraint. This tool is a recent addition to the device's software. The XL probe is recommended for usage in overweight/obese people since it measures more deeply into the skin surface (3.5–7.5 cm vs. 2.5–6.5 cm) and operates at a lower central frequency (2.5 MHz vs. 3.5 MHz for the standard M probe).26 Another strength of the present study is the selection of the participants. We included subjects with PSG-proven OSA, whilst most previous studies hugely relied on clinical and questionnaire based diagnosis of OSA. This makes our data more reliable and applicable in the context of OSA patients. Another strength is inclusion of patients with causes of OSA other than obesity such as oro-mandibulo-facial abnormalities, although only 3 patients in our study had OSA attributable to such causes.
The current study had certain limitations. First, all our subjects were symptomatic PSG proven OSA patients; therefore, it is difficult to comment on the presence of NAFLD in asymptomatic OSA patients. Second, the occurrence of referral bias was inevitable being a hospital based study. Ultimately, due to the study's cross-sectional design, it is impossible to identify whether OSA preceded the development of increased liver stiffness in our group of participants. Prospective studies are required to investigate the causality relationship between OSA and NAFLD.
Continuous positive airway pressure treatment has recently been shown to be effective in reducing the activity of NAFLD in OSA patients, but the selection of patients who may benefit from this treatment strategy is yet to be protocolized.27 This study, therefore, aims to fill this lacunae in literature. The routine employment of transient elastography will help in non-invasively recognizing a subset of OSA patients who are at-risk for and harbor NAFLD changes. This will assist physicians in the decision making process for further intervention and therapy.
OSA is associated with increased prevalence of hepatic steatosis and fibrosis, which constitute the spectrum of NAFLD. AHI and ODI are significantly associated with the presence of hepatic fibrosis even after adjusting for confounders. AHI, ODI, and T90 are excellent predictors of hepatic fibrosis and may be used to guide clinicians to select patients warranting further workup for NAFLD and referral to a hepatologist.
Author contribution statement
Ashwin Parchani: Writing – original draft preparation, Data Curation, Resources, Investigation. Ravi Gupta:Conceptualisation, Methodology, Formal Analysis. Ravi Kant: Conceptualisation, Supervision, Writing – Editing and Review. Lokesh Kumar Saini: Visualisation, Project administration, Investigation. Rohit Gupta: Conceptualisation, Supervision, Writing – Editing and Review.
Conflicts of interest
The authors declare no potential conflicts of interest with respect to the research, authorship or publication of this study.
Acknowledgments
We would like to thank Dr. Shiana Singh, Junior resident, AIIMS, Rishikesh for her contribution to the article in the form of medical illustration. We would also like to thank the healthcare staff and technicians at sleep lab and transient elastography suite for their support in the study.
Funding statement
The authors received no financial support for this study.
Data availability statement
The data used to support the findings of this study are available on request to the author.
References
- 1.Younossi Z., Anstee Q.M., Marietti M., et al. Global burden of NAFLD and NASH: trends, predictions, risk factors and prevention. Nat Rev Gastroenterol Hepatol. 2018 Jan;15:11–20. doi: 10.1038/nrgastro.2017.109. [DOI] [PubMed] [Google Scholar]
- 2.Tasali E., Ip M.S.M. Obstructive sleep apnea and metabolic syndrome: alterations in glucose metabolism and inflammation. Proc Am Thorac Soc. 2008 Feb 15;5:207–217. doi: 10.1513/pats.200708-139MG. [DOI] [PubMed] [Google Scholar]
- 3.Mesarwi O.A., Loomba R., Malhotra A. Obstructive sleep apnea, hypoxia, and nonalcoholic fatty liver disease. Am J Respir Crit Care Med. 2019 Apr 1;199:830–841. doi: 10.1164/rccm.201806-1109TR. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Aron-Wisnewsky J., Clement K., Pépin J.L. Nonalcoholic fatty liver disease and obstructive sleep apnea. Metabolism. 2016 Aug;65:1124–1135. doi: 10.1016/j.metabol.2016.05.004. [DOI] [PubMed] [Google Scholar]
- 5.Kallwitz E.R., Herdegen J., Madura J., Jakate S., Cotler S.J. Liver enzymes and histology in obese patients with obstructive sleep apnea. J Clin Gastroenterol. 2007 Dec;41:918–921. doi: 10.1097/01.mcg.0000225692.62121.55. [DOI] [PubMed] [Google Scholar]
- 6.Daltro C., Cotrim H.P., Alves E., et al. Nonalcoholic fatty liver disease associated with obstructive sleep apnea: just a coincidence? Obes Surg. 2010 Nov;20:1536–1543. doi: 10.1007/s11695-010-0212-1. [DOI] [PubMed] [Google Scholar]
- 7.Aron-Wisnewsky J., Minville C., Tordjman J., et al. Chronic intermittent hypoxia is a major trigger for non-alcoholic fatty liver disease in morbid obese. J Hepatol. 2012 Jan;56:225–233. doi: 10.1016/j.jhep.2011.04.022. [DOI] [PubMed] [Google Scholar]
- 8.Trzepizur W., Boursier J., Le Vaillant M., et al. Increased liver stiffness in patients with severe sleep apnoea and metabolic comorbidities. Eur Respir J. 2018 Jun;51 doi: 10.1183/13993003.00601-2018. [DOI] [PubMed] [Google Scholar]
- 9.Sumida Y., Nakajima A., Itoh Y. Limitations of liver biopsy and non-invasive diagnostic tests for the diagnosis of nonalcoholic fatty liver disease/nonalcoholic steatohepatitis. World J Gastroenterol. 2014 Jan 14;20:475–485. doi: 10.3748/wjg.v20.i2.475. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.tian Cao Y., lan Xiang L., Qi F., Zhang Y juan, Chen Y., qiao Zhou X. Accuracy of controlled attenuation parameter (CAP) and liver stiffness measurement (LSM) for assessing steatosis and fibrosis in non-alcoholic fatty liver disease: a systematic review and meta-analysis. eClinicalMedicine. 2022 Sep;51 doi: 10.1016/j.eclinm.2022.101547. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.De A., Duseja A. Nonalcoholic fatty liver disease: Indian perspective. Clin Liver Dis. 2021 Sep;18:158–163. doi: 10.1002/cld.1141. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Agrawal S., Duseja A., Aggarwal A., et al. Obstructive sleep apnea is an important predictor of hepatic fibrosis in patients with nonalcoholic fatty liver disease in a tertiary care center. Hepatol Int. 2015 Apr;9:283–291. doi: 10.1007/s12072-015-9615-3. [DOI] [PubMed] [Google Scholar]
- 13.American Academy of Sleep Medicine, editor. International Classification of Sleep Disorders. 3. 2014. p. 383. (Darien, Ill: American Acad. Of Sleep Medicine). [Google Scholar]
- 14.Appropriate body-mass index for Asian populations and its implications for policy and intervention strategies. Lancet. 2004 Jan;363:157–163. doi: 10.1016/S0140-6736(03)15268-3. [DOI] [PubMed] [Google Scholar]
- 15.Ma W.Y., Yang C.Y., Shih S.R., et al. Measurement of Waist Circumference: midabdominal or iliac crest? Diabetes Care. 2013 Jun;36:1660–1666. doi: 10.2337/dc12-1452. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Tan C.E., Ma S., Wai D., Chew S.K., Tai E.S. Can we apply the national cholesterol education Program adult treatment Panel definition of the metabolic syndrome to Asians? Diabetes Care. 2004 May;27:1182–1186. doi: 10.2337/diacare.27.5.1182. [DOI] [PubMed] [Google Scholar]
- 17.Duseja A., Singh S.P., De A., et al. Indian national association for study of the liver (INASL) guidance paper on nomenclature, diagnosis and treatment of nonalcoholic fatty liver disease (NAFLD) J Clin Exp Hepatol. 2023 Mar 1;13:273–302. doi: 10.1016/j.jceh.2022.11.014. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Rb B. 2015. AASM Scoring Manual. [Google Scholar]
- 19.Takahashi A., Anzai Y., Kuroda M., et al. Effects of sleep quality on non-alcoholic fatty liver disease: a cross-sectional survey. BMJ Open. 2020 Oct 29;10 doi: 10.1136/bmjopen-2020-039947. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Savransky V., Nanayakkara A., Vivero A., et al. Chronic intermittent hypoxia predisposes to liver injury. Hepatol Baltim Md. 2007 Apr;45:1007–1013. doi: 10.1002/hep.21593. [DOI] [PubMed] [Google Scholar]
- 21.Sookoian S., Pirola C.J. Obstructive sleep apnea is associated with fatty liver and abnormal liver enzymes: a meta-analysis. Obes Surg. 2013 Nov;23:1815–1825. doi: 10.1007/s11695-013-0981-4. [DOI] [PubMed] [Google Scholar]
- 22.Dulai P.S., Singh S., Patel J., et al. Increased risk of mortality by fibrosis stage in nonalcoholic fatty liver disease: systematic review and meta-analysis. Hepatology. 2017;65:1557–1565. doi: 10.1002/hep.29085. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Boursier J., Vergniol J., Guillet A., et al. Diagnostic accuracy and prognostic significance of blood fibrosis tests and liver stiffness measurement by FibroScan in non-alcoholic fatty liver disease. J Hepatol. 2016 Sep;65:570–578. doi: 10.1016/j.jhep.2016.04.023. [DOI] [PubMed] [Google Scholar]
- 24.Fraquelli M., Rigamonti C., Casazza G., et al. Reproducibility of transient elastography in the evaluation of liver fibrosis in patients with chronic liver disease. Gut. 2007 Jul;56:968–973. doi: 10.1136/gut.2006.111302. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Castéra L., Foucher J., Bernard P.H., et al. Pitfalls of liver stiffness measurement: a 5-year prospective study of 13,369 examinations. Hepatol Baltim Md. 2010 Mar;51:828–835. doi: 10.1002/hep.23425. [DOI] [PubMed] [Google Scholar]
- 26.Berger A., Shili S., Zuberbuhler F., et al. Liver stiffness measurement with FibroScan: use the right probe in the right conditions. Clin Transl Gastroenterol. 2019 Apr 17;10 doi: 10.14309/ctg.0000000000000023. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Hirono H., Watanabe K., Hasegawa K., Kohno M., Terai S., Ohkoshi S. Impact of continuous positive airway pressure therapy for nonalcoholic fatty liver disease in patients with obstructive sleep apnea. World J Clin Cases. 2021 Jul 6;9:5112–5125. doi: 10.12998/wjcc.v9.i19.5112. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
The data used to support the findings of this study are available on request to the author.