Does non-alcoholic fatty liver disease predispose patients to type 2 diabetes in the absence of obesity?


Professor Geoffrey C Farrell, Gastroenterology and Hepatology Unit, The Canberra Hospital, Australian National University Medical School, ACT 2605, Australia. Email:


See article in J. Gastroenterol. Hepatol. 2010; 25: 352–356

Type 2 diabetes (T2D) and non-alcoholic fatty liver disease (NAFLD) are associated with each other more frequently than expected by chance. NAFLD is present in approximately 80% of patients with T2D, and T2D is present in 21–45% of patients with NAFLD, with glucose intolerance present in up to 70% if an oral glucose tolerance test (OGTT) is performed.1–3 T2D may cause metabolic fatty liver disease (so-called NAFLD) and, like diabetes,NAFLD is now considered a manifestation of metabolic syndrome (MetS).1 Insulin resistance, the primary pathophysiological disorder leading to T2D and MetS is so often found in NAFLD that this form of liver disease may be regarded as similar to or a complication of ‘pre-diabetes’, thereby indicating the high future risk for onset of diabetes as well as cardiovascular disease.1,3 In several studies, NAFLD diagnosed by ultrasonography together with unexplained elevation of liver enzymes predicted diabetes risk, independent of obesity and other components of MetS.4–11 Thus, the concept has arisen that NAFLD may signify more than just the presence of a liver disease; it may also be an early mediator of T2D and MetS.

Although histological examination remains the gold standard for diagnosis of NAFLD, pathological definition is often not possible in community-based epidemiological studies. Alternatively, in subjects without substantial alcohol consumption or other causes of liver disease, persistent elevation of alanine aminotransferase (ALT) and γ-glutamyltransferase (GGT) is regarded as a surrogate marker of NAFLD.1,12 In 1998, a longitudinal study examined the association of elevated liver enzymes with incident diabetes.4 Since then, high values of ALT and GGT, even within the normal range, have been reported to predict incident diabetes and MetS; some studies demonstrated stronger association between GGT and diabetes than ALT, while other studies reported the opposite.4–6 In a meta-analysis of results from prospective population-based studies fully adjusted for other diabetes risk factors (albeit variably adjusted), 1 U/L increase of loge ALT was associated with 85% increase in diabetes risk, and 1 U/L increase of loge GGT with 92% increase.4 This indicates that elevations in liver enzymes attributable to NAFLD increase incident diabetes rate independently of commonly measured diabetes risk factors. Recently, Adams et al. found subjects with elevated liver enzymes attributed to NAFLD were at increased risk of developing metabolic complications at 11 years follow up; they were threefold more likely to develop diabetes and 50% more likely to develop MetS compared with the age-matched population.5 Multivariate modeling showed that the increased risk of metabolic complications could be explained by associated visceral obesity and subsequent insulin resistance, which almost invariably accompanies patients with NAFLD. In contrast to this high risk of diabetes, only a small minority of subjects with NAFLD develop cirrhosis over 10 years, with an even smaller proportion dying from liver disease during this period of follow up.5

A limitation of these studies is that subjects with NAFLD and normal liver enzymes were not included in the case population.4–6 Liver enzyme changes are neither highly sensitive nor specific to accumulation of fat in the liver and related liver damage. Further, only a minority of patients with T2D have abnormal liver enzymes, while the entire histological spectrum of NAFLD can be seen in patients with normal liver enzymes.7,8 Thus, normal liver enzymes is not a perfect criterion to exclude NAFLD, and patients with alterations in glucose metabolism and insulin resistance despite normal ALT should also be considered in selecting cases of possible NAFLD for hepatic imaging and/or histological assessment.8 Ultrasonography can estimate the severity of the hepatic steatosis relatively accurately, even though it cannot differentiate between the histological entities of simple steatosis and non-alcoholic steatohepatitis (NASH).12 The presence of NAFLD by ultrasound correlated significantly with the number of MetS components.13 Compared with overall obesity (body mass index, BMI) and abdominal obesity (waist circumference), ultrasound-diagnosed fatty liver had the highest positive predictive value and most attributable risk as a percentage for detecting clustering of cardiovascular risk factors as MetS.2 Therefore, NAFLD defined by ultrasound may be a better diabetes predictor than liver enzymes.

In order to determine the association between ultrasound-diagnosed NAFLD and risk of development of diabetes, Shibata et al. conducted an observational cohort study among middle-aged male workers in a Japanese company from 1997–2005.9 Workers who had a daily alcohol consumption of more than 20 g and those with impaired glucose tolerance by 75 g OGTT were excluded. The remaining 3189 workers were classified into fatty liver and non-fatty liver groups based on the findings of liver ultrasonography. Both groups were followed for development of T2D. Hazard ratio (HR) was determined in a Cox proportional hazard analysis, and a nested case–control study was conducted to determine the odds ratio (OR). The average age of participants was 48 years at entry, and mean follow up was 4 years. The incidence of diabetes in the fatty liver group was 2073/100 000 person-years (65 cases), whereas it was 452/100 000 person-years (44 cases) in the non-fatty liver group. The age- and BMI-adjusted HR of diabetes associated with fatty liver was 5.5 (95% confidence interval [CI] = 3.6–8.5). In the nested case–control analysis, the OR adjusted for age and BMI was 4.6 (95% CI = 3.0–6.9). These findings are similar to those of Fan et al. who recently found Chinese patients with ultrasound-diagnosed NAFLD had a threefold increase in incidence of diabetes than age-, sex- and occupation-matched controls over a 6-year follow-up period, although this study did not adjust fully for metabolic factors other than obesity.10 In addition, several registry-based cohort studies have demonstrated that a significant proportion of patients with biopsy-proven NAFLD develop impaired fasting glucose (IFG) or T2D soon after initial diagnosis of NAFLD.1

In this issue of the Journal, Yamada et al. explore the impact of ultrasound-diagnosed fatty liver on the incidence of IFG or T2D in Japanese people undergoing a health checkup.11 A total of 12 375 individuals (6799 men and 5576 women) without hyperglycemia or T2D at baseline were re-assessed after 5 years. IFG and T2D were newly diagnosed in 7.6% and 1% of men, and 3.8% and 0.5% of women, respectively, within the study period. In both sexes, the prevalence of newly diagnosed IFG and T2D was significantly higher in the participants with fatty liver than among those without fatty liver, and after adjustment for the other risk factors, fatty liver remained an independent risk factor for IFG and/or T2D. The impact of fatty liver on incidence of IFG and T2D was stronger among participants with a lower BMI. Therefore, the presence of fatty liver may be a better predictor for development of T2D than obesity itself, and it can be considered to be an early predictor of T2D.

In general, this is a well written and concise manuscript with a clear message, and answers a question that is important and significant in public health. Furthermore, it has the strength of studying a large number of individuals. Some limitations of this study include the lack of liver enzymes and OGTT at baseline, the use of a single result of fasting plasma glucose (FPG) for diagnosis of diabetes at follow up, and lack of rigorous exclusion of other etiologies of fatty liver, making it difficult to draw conclusions regarding the metabolic risk among subjects with NAFLD. It should be noted that regional guidelines recommend an OGTT be performed at diagnosis of NAFLD when FPG is more than 5.6 mmol/L12, and several studies now report a much higher prevalence of glucose intolerance and established T2D at diagnosis of NAFLD when OGTT is routinely performed.14–16

Both T2D and hepatogenous diabetes (complicating cirrhosis) are associated with increased liver-related morbidity and mortality in cirrhotic patients regardless of etiology.1 Unlike the hepatogenous diabetes attributed to cirrhosis, T2D in NAFLD is more frequently associated with risk factors such as age, BMI and family history of T2D.1,4–11 It constitutes a risk factor for NASH, for fibrotic progression to cirrhosis and ultimately hepatocellular carcinoma. The finding of diabetes is thus associated with an increased risk of all-cause death and liver-related mortality in patients with NAFLD, and diabetic and cardiovascular risk may compete with liver-related complications in dictating the final outcome.1,3

The biological mechanisms by which NAFLD contributes to a higher risk of developing T2D are not fully understood. However, the fatty liver could contribute in the same way as visceral adipose tissue to insulin resistance, systemic inflammation and oxidative stress, while decreased serum adiponectin concentrations might also be part of the mechanism.1,3,17 Interestingly, among subjects with NAFLD, only insulin resistance was predictive of diabetes development, whereas among subjects without NAFLD, obesity in addition to insulin resistance was associated with subsequent diabetes.1,3,4 Thus, development of NAFLD may be an important predisposing step in overweight and abdominally obese individuals towards development of T2D.

In summary, subjects with ultrasound-diagnosed NAFLD and/or unexplained liver enzymes elevation have a high incidence of T2D and metabolic complications in the near future. FPG and possibly OGTT should be performed at diagnosis of NAFLD, and patients would benefit from being screened regularly thereafter for development of diabetes.12,18 This could be of particular importance in apparently lean individuals whose only evidence of central adiposity may be fatty liver. Furthermore, identification of NAFLD provides a point of early intervention for advice about lifestyle modifications, including curbing energy excess, restituting nutritional imbalances and increasing physical activity to a minimum equivalent of 140 min fast walking/week. Interventions to prevent the development of diabetes among the vast population of overweight and obese individuals may be more efficacious if targeted at those with highest risk, among which concomitant NAFLD should now be recognized.