Obstructive sleep apnoea and nonalcoholic fatty liver disease: risk factor or just coincidence?


Obesity has a high prevalence in many countries and is closely related to several comorbidities, including nonalcoholic fatty liver disease (NAFLD) and obstructive sleep apnoea syndrome (OSAS) (1). OSAS is characterized by repeated episodes of upper airway obstruction during sleep, usually associated with interrupted sleep and chronic intermittent hypoxia (CIH) (2). The main risk factor for OSAS is obesity, being present in approximately 70% of obstructive sleep apnoea patients (3, 4). The diagnosis is made by means of polysomnography, and continuous positive airway pressure is considered the gold standard therapy (2).

In this issue of Liver International, Younossi and colleagues have described the relationships between OSAS and NAFLD in obese patients who underwent bariatric surgery (5). This study was designed to evaluate the association of polysomnography parameters and histological subtypes of NAFLD as well as hepatic fibrosis.

The association of OSAS and NAFLD has been the target of growing attention. Ischaemic hepatitis due to severe hypoxaemia resulting from OSAS in obese patients has been reported, and several studies have examined the association between these two diseases (6–8). Most of these investigations found a positive association but others failed to do so probably because of methodological differences and deficiencies. One of the limitations in these investigations has to do with the evaluation of the existence or not of NAFLD/nonalcoholic steatohepatitis (NASH).

Some authors (9, 10) studied this association by means of liver enzyme levels and polysomnography, indicating that OSAS treatment improved liver function tests in the majority of the patients. Histological data were not available in these studies, but the decrease in liver enzymes may be related to improvement in histological lesions.

Tannéet al. (11) suggested that severe OSAS could also be an independent risk factor for NASH. They investigated 163 patients who had polysomnography performed for suspected OSAS. Patients with severe OSAS presented greater peripheral insulin resistance and displayed a higher percentage of hepatic steatosis, as well as necrosis and fibrosis scores, despite a similar body mass index. In this study, liver enzymes were measured in all patients but histological confirmation was available in only half of the 32 patients with abnormal liver tests.

Tatsumi and Saibara (12) evaluated nonobese patients with OSAS to see whether they were prone to developing hepatic steatosis, and whether intermittent hypoxaemia contributed to the progression of steatohepatitis. Eighty-three patients with OSAS and 41 controls without the disease underwent polysomnography, abdominal computerized tomography and liver enzyme determinations. Type III procollagen (P-III-P) was used as a marker of liver fibrosis. They showed a negative correlation between the P-III-P levels and mean O2 saturation, suggesting that oxygen desaturation during sleep is a risk for developing steatohepatitis, especially in patients with underlying hepatic steatosis.

Another study assessing OSAS, glucose intolerance (IGT) and hepatic steatosis showed that there were no significant differences between the groups with and without OSAS regarding abnormal liver enzymes (13). The authors assessed 45 nondiabetic, obese women with an oral glucose tolerance test, polysomnography and abdominal ultrasonography. The patients with OSAS had more frequent IGT (40 vs 12%; P=0.041) than patients without it. Among the tested variables, IGT was the only predictor of grade III steatosis. The authors concluded that OSAS was an independent risk factor for IGT and this, in turn, was associated with hepatic steatosis in the studied sample.

Nonalcoholic fatty liver disease is clinically silent in the majority of patients, and the diagnosis is based on laboratory, imaging and histological findings. Biochemical abnormalities usually consist of elevated aspartate aminotransferase, alanine aminotransferase (ALT), γ-glutamyltransferase and alkaline phosphatase; however, significant liver disease may be present with normal levels of liver enzymes. Ultrasonography, computed tomography and magnetic resonance imaging play an important role in diagnosing NAFLD, but all three techniques cannot distinguish benign steatosis from steatohepatitis, severity of inflammation and grade and degree of fibrosis or stage of disease (14). The value of liver biopsy in establishing the diagnosis of NAFLD in routine clinical practice remains controversial, but it is the only available procedure that can accurately grade and stage NAFLD, and it also excludes other causes of liver disease (15). In the above studies, the diagnosis was based on liver enzymes and imaging technique abnormalities, but the absence of biopsy data raises questions about the association of the degree of OSAS and the severity of NAFLD/NASH.

Another methodological limitation of these studies is the diagnosis of OSAS. This condition can be suspected on the basis of history and physical examination, but overnight polysomnography is needed to confirm the diagnosis (2). In these studies, the authors used different cut-off points to define OSAS (apnoea hypopnoea index ≥5; 10 or 15 events/h of sleep), making it difficult to compare their results.

Singh et al. (16) published a study that casts some doubt on the relationship between OSAS and NAFLD. One hundred and ninety patients with a biochemical diagnosis of NAFLD were studied (116 with steatosis and 74 with steatohepatitis), of whom 50 were subjected to liver biopsy. OSAS was assessed by means of the Modified Berlin Sleep Apnea Questionnaire. The authors did not find a difference between the prevalence of OSAS symptoms in the patients with steatosis and steatohepatitis (45 vs 49%; P=0.66); however, they did not perform liver biopsy in all patients and the diagnosis of OSAS was based on a ‘sleep questionnaire’, which is not a reliable method of diagnosing sleep apnoea.

Some authors evaluated obese patients after bariatric surgery, with polysomnography and liver biopsy finding discrepant results (17–19). Jouët et al. (17) studied 62 obese patients and found that OSAS was a risk factor for elevated liver enzymes but not for histological NASH. Kallwitz et al. (18) concluded that, in obese patients with NAFLD, OSAS was associated with elevated ALT levels and a trend towards histological evidence of progressive liver disease.

We reported a study (19) that showed OSAS to be one of the predictive factors of fibrosis in NAFLD, and that OSAS could contribute to the severity of hepatic disease in severely obese individuals. The logistic regression analysis identified the following independent predictors of fibrosis: OSAS [odds ratio (OR): 5.61; 95% confidence interval (CI): 1.74–18.07], dyslipidaemia (OR: 3.65; 95% CI: 1.16–11.44) and hepatocellular ballooning in histology (OR: 12.29; 95% CI: 2.08–72.83).

Younossi and colleagues (5) described the relationship between oxygen desaturation and NASH, diagnosed by liver biopsy. They reported that subjects with histological NASH had lower oxygen saturation and a higher respiratory disturbance index compared with non-NASH controls, suggesting that nocturnal hypoxic episodes in NAFLD patients may be a risk factor for developing NASH.

Studies in severely obese patients deserve some attention. Morbid obesity is predictive of severe OSAS (4) and advanced liver fibrosis (14). Two questions must be answered in this regard: (i) Is this patient group of morbidly obese patients representative of the general NAFLD population? and (ii) Can these findings be applicable to NAFLD in general?

Although there are some limitations of the literature in this area, the biggest unanswered question has to do with the pathophysiological mechanisms, particularly how OSAS is associated with liver fibrosis. A ‘two-hit’ hypothesis has been proposed to explain the pathogenesis of NAFLD. The first hit represents the accumulation of excess fat in the liver secondary to peripheral insulin resistance (20). Recent studies have shown that obstructive sleep apnoea is associated with insulin resistance irrespective of obesity (21–22). Obstructive apnoea episodes are associated with an adrenergic discharge, which activates the hypothalamic–pituitary–adrenal axis, releasing cortisol and aggravating insulin resistance. Other factors that may contribute to insulin resistance in OSAS patients are the promotion of a pro-inflammatory state with increased tumour necrosis factor-α and interleukin-6, and the increased levels of leptin, which in turn could contribute to insulin resistance (23–25).

The second hit, which is thought to result in progression of NAFLD to a necro-inflammatory hepatitis, remains unclear, but oxidative stress, lipid peroxidation and/or cytokine-mediated injury have been implicated (20). Some studies have shown that intermittent hypoxia, characteristic of obstructive sleep apnoea, can lead to oxidative stress, lipid peroxidation and may exacerbate pre-existing fatty liver of obesity via upregulation of pathways of lipid biosynthesis in the liver (26–28). None of the studies reviewed here investigated these mechanisms.

In conclusion, even though there is a proven association between OSAS and liver fibrosis, a causal role of OSAS in liver fibrosis has not been clearly proven. The major weakness of all studies, including that of Younossi and colleagues (5), is that they are only descriptive with virtually no mechanistic insight. Additional mechanistic studies are needed to determine whether and how hypoxia may contribute to liver fibrosis. It would also be important to examine the impact of the effective treatment of OSAS on the histological outcome of patients with OSAS and NAFLD.