Digestive Diseases and Sciences

, Volume 55, Issue 11, pp 3200–3206

Protective Role of Coffee in Non-alcoholic Fatty Liver Disease (NAFLD)

Authors

  • Daniela Catalano
    • Clinica di Medicina Interna e Terapia Medica, Dipartimento di Medicina Interna, Facoltà di Medicina e ChirurgiaUniversità di Catania
  • Giuseppe Fabio Martines
    • Clinica di Medicina Interna e Terapia Medica, Dipartimento di Medicina Interna, Facoltà di Medicina e ChirurgiaUniversità di Catania
  • Antonia Tonzuso
    • Clinica di Medicina Interna e Terapia Medica, Dipartimento di Medicina Interna, Facoltà di Medicina e ChirurgiaUniversità di Catania
  • Clara Pirri
    • Clinica di Medicina Interna e Terapia Medica, Dipartimento di Medicina Interna, Facoltà di Medicina e ChirurgiaUniversità di Catania
  • Francesca M. Trovato
    • Clinica di Medicina Interna e Terapia Medica, Dipartimento di Medicina Interna, Facoltà di Medicina e ChirurgiaUniversità di Catania
    • Clinica di Medicina Interna e Terapia Medica, Dipartimento di Medicina Interna, Facoltà di Medicina e ChirurgiaUniversità di Catania
Original Article

DOI: 10.1007/s10620-010-1143-3

Cite this article as:
Catalano, D., Martines, G.F., Tonzuso, A. et al. Dig Dis Sci (2010) 55: 3200. doi:10.1007/s10620-010-1143-3

Abstract

Aim

The benefits of coffee on abnormal liver biochemistry, cirrhosis and hepatocellular carcinoma have been reported, but there is a lack of satisfactory explanation. Thus, this study aims to investigate if coffee use has any relationship with bright liver, measured by ultrasound bright liver score (BLS), in patients with non-alcoholic fatty liver disease (NAFLD), and which relationship, if any, is present with BMI and insulin resistance.

Methods

This study was performed on 245 patients, 137 with NAFLD and 108 controls. Coffee drinking was defined according to the absolute number of cups of coffee (only espresso coffee), and also graded as 1 (0 cups of coffee/day), 2 (1–2 cups of coffee/day) 3 (≥3 cups of coffee/day). Insulin resistance was assessed by homoeostasis model-insulin resistance index (HOMA).

Results

Less fatty liver involvement is present in coffee vs. non-coffee drinkers. Odds ratios show that obesity, higher insulin resistance, lower HDL cholesterol, older age and arterial hypertension are associated with a greater risk of more severe BLS; to the contrary, coffee drinking is associated with less severe BLS. In the multiple logistic regression (MLR) model, number of cups of coffee, HOMA and BMI account for 35.8% of the variance to BLS. Coffee use is inversely associated with the degree of bright liver, along with insulin resistance and obesity, which, to the contrary, are directly associated with greater likelihood and severity of bright liver appearance.

Conclusions

A possible opposite, if not antagonistic, role of coffee with regard to overweightness and insulin resistance, similar to that reported in hepatocarcinoma and cirrhosis, is envisaged in the natural history of NAFLD.

Keywords

CoffeeNAFLDInsulin resistanceUltrasoundHOMALife-styles

Introduction

Coffee drinking has an inverse and graded association with the risk of liver cancer [1] and a greater consumption of coffee may reduce the risk of hepatocarcinoma, i.e. an increase in consumption of two cups of coffee per day was associated with a 43% reduced specific risk [2]. The hypothesis is that some ingredient in coffee could protect against cirrhosis and especially alcoholic cirrhosis [3, 4]. Moreover, coffee, but not other beverages containing caffeine, may inhibit the onset of alcoholic and non-alcoholic liver cirrhosis [5].

It was reported that an increase in body weight is the strongest determinant for increases in serum γ-GT and that coffee drinking may be associated with a reduced risk of the development of increased serum γ-GT levels [6]. Non-alcoholic fatty liver disease (NAFLD) is an increasingly prevalent condition affecting adults and children, leading to significant morbidity, and it is sometimes associated with the metabolic syndrome [7]. Increased transaminases [8] and the ultrasound (US) bright liver pattern are among the most considered clues of NAFLD. The US bright liver pattern and the derived bright liver score (BLS) are acceptable non-invasive yardsticks when other tools are unavailable or excessively invasive and requiring repetition [9]. Even if, currently, liver biopsy is the gold standard for diagnosis, histological lesions of liver assessed by fine needle biopsy are unevenly distributed throughout the parenchyma; therefore, sampling error of liver biopsy can result in substantial misdiagnosis and staging inaccuracies, even in quite diffuse abnormalities such as liver steatosis [10]. Non-invasive US imaging allows a sufficient and extensive visualization of the liver, with major advantages for diagnosis and follow-up [11].

Many genetic and environmental risk factors could be variously involved in the natural history of disease risk progression [1216]. It was reported that coffee use is associated with a lower insulin resistance both in overweight and in non-obese women [17]. Despite the information that, even in a small group of patients, a direct involvement of caffeine in the establishment of insulin resistance was described [18], elsewhere, components of metabolic syndrome are significantly and inversely related to coffee consumption [19]. Moreover, among persons at high risk of liver injury, the use of coffee, and especially caffeine, was associated with a lower risk of elevated ALT transaminase activity [20].

Increased body weight is a relevant factor in many diseases. In this regard bioimpedance assessment (BIA) of body composition provides detailed information on body compartments, notably fat and lean mass, useful in the definition of overweight–obesity features and, much more, in the dietary prescription strategy and follow-up [21]. Effective treatment, in the absence of proven therapies, is directed toward weight loss and co-morbidity management [22] despite the fact that it has been reported that present dietary recommendations may worsen NAFLD histopathology [23].

In some population groups most adult subjects are actually non-alcohol drinkers or very light alcohol beverage drinkers. We find that this provides a favorable opportunity to assess if caffeine is associated with any special feature of liver laboratory results in the general population, without the confounding effect of different alcohol intakes. The aim of the study was to investigate if coffee drinking, and degree of coffee use, has any relationship with bright liver, measured as ultrasound (US) bright liver score (BLS), in patients with no present or recent use of alcohol, i.e. with clinical diagnosis of NAFLD. Moreover, we investigate which relationship, if any, is present within the measured variables involved.

Patients and Methods

A total of 157 NAFLD (men 77; women 80) consecutive patients were studied in a gastroenterology and nutrition unit operating as an autonomous outpatient clinic and day hospital (DH). The clinic is equipped with four US machines, and an echographic assessment (abdomen, thyroid) is included in the first step of the medical residents’ approach to all DH patients. Patients were referred by their family doctors for evaluation and nutritional counseling due to minor gastro-intestinal ailments, i.e. symptoms that were not preliminarily attributed by the referring physician to active liver and/or digestive disease or secondary to other definite causes. A control group of 153 non-NAFLD patients (men 70; women 83) was built-up among the patients referred to the same day hospital clinic. This control group was chosen according to the same eligibility and with values below the detailed exclusion criteria, provided that bright liver was not observed.

Diagnosis of NAFLD was non-invasively performed according to the presence of US appearance of fatty liver (assessed as BLS ≥1). Patients were followed thereafter for at least 6 months, and were included in the study provided that no exclusion criterion was present. All patients with clinical/echographic signs of severe chronic liver disease, apart from the lone finding of bright liver, with congestive heart failure, renal failure (GFR below 60 ml/min/1.73 m2), oncological disease, and thyroid disease were preliminarily excluded from this study. Exclusion criteria were known factors of liver steatosis, i.e. diabetes (fasting glucose level ≥126 mg/dl), alcohol history (above 20 g/day in the last 5 years), previous HBV and/or HCV infections. Personalized diets were allowed, with daily recommendations derived from the software used (Dietosystem TM, Milan, Italy). All subjects had a traditional Mediterranean diet, which is characterized by a high intake of vegetables, legumes, fruits, nuts, and cereals, a high intake of olive oil and a low (or no) intake of saturated lipids, as well as a moderately high intake of fish, a low-to-moderate intake of dairy products (mostly in the form of cheese or yogurt), and a low intake of meat and poultry. Subjects on habitual unhealthy diets, related to unbalanced composition of nutrients (excessive intake or exclusion of protein, lipids and carbohydrates) according to national and WHO recommendations, were excluded by the analysis of data.

Since the study attempts to reduce some of the most powerful confounding factors such as alcohol, diabetes and other alimentary factors inducing transaminase increase, a preliminary eligibility criterion was the existence of normal aminotransferase levels at the beginning of the study, defined as ALT <30 IU/l in men and <19 IU/l in women [18]. Subsequent increase of transaminases ALT and/or AST above upper normal values at the two subsequent consecutive controls, 1 month apart, throughout a 6-month follow-up, were further exclusion criteria. For data analysis, the initially observed values, when ALT levels were considered normal as required by the eligibility criteria, were used.

Arterial hypertension was defined as >140 mmHg of systolic and >90 mmHg of diastolic blood pressure. Well-controlled arterial hypertension patients were not excluded from the analysis of this study.

Daily users of cola beverages and tea were preliminarily excluded if an estimated habitual daily intake of caffeine with these beverages above 100 mg (the content of a cup of coffee) was reported; in this regard, measurements were estimated according to dietary content tables [15].

Coffee drinking habits were defined according to the average of the absolute number of coffee cups (only espresso coffee), and also graded as 1 (0 cups of coffee/day), 2 (1–2 cups of coffee/day), or 3 (≥3 cups of coffee/day), reported during the 6-month follow-up. No attempt was made in order to modify coffee drinking habits. Smoking was assessed according to the number of cigarettes/day and also on a yes–no previous smoker-status basis, according to present habit.

Weight (BW) was measured in light clothing, without shoes, in kg, and height (H) was measured in meters, using a scale-integrated stadiometer. Body mass index was calculated as BW/H2 and patients were categorized as normal weight (<25.0 kg/m2), overweight (≥25.0 and ≤29.9 kg/m2) and obese (>30.0 kg/m2). Waist–hip ratio was assessed as well.

Insulin resistance (IR) was assessed by the homoeostasis model-insulin resistance index (HOMA), according to the formula ‘(fasting insulin value × fasting blood sugar level)/405’. HOMA correlates with the IR index measured by the hyperinsulinaemic euglycaemic clamp method, even though it has a day-to-day variability [24]. For this reason, the average of three measurements from 3 months of follow-up are considered in this study for statistical analysis [21].

Bioimpedance assessment (BIA) of body composition was performed by a single-frequency BIA device with 50 kHz and 800 mA (model BIA 101 RJL; Akern, Firenze, Italy) according to the standard tetrapolar technique; to avoid artifacts, the electrodes were placed on the feet, ankles, hands and wrists. The body composition was calculated from BIA measurements (resistance and reactance) and anthropometric variables (body weight and height) using the software provided by the Akern/RJL Systems. This predictive model allows for calculation of total body water (TBW), body fat mass (FM), fat-free mass (FFM), body cell mass (BCM) and extracellular water (ECW). All measurements were expressed in kilograms and in relative percentages [21].

Ultrasound examination used for data analysis was performed by the same senior echographist (DC) to exclude interobserver variability. The examiner was unaware of clinical details at the time of the procedure. A GE echo-color-doppler machine (GE Logiq 5 Expert US; GE Medical Systems, Milwaukee, Wisconsin, USA) with high resolution and real-time sectional scan transducers was used. The liver was assessed for size, contour, echogenicity, structure, portal vein dimensions, flow velocity, and penetration of the US beam, i.e. posterior beam attenuation and portal vessel wall distinction; these last criteria were assessed in the report of the degree of steatosis. The grades of bright liver, i.e. bright liver score (BLS), considered for statistical evaluation were classified by unifying three subscale-grades focused on echogenicity, beam penetration and portal vessel wall distinction. Grade 0 was normal or slightly reduced in comparison with right kidney echogenicity. Grade 1 was slightly increased in comparison with right kidney echogenicity. Grade 2 was clearly increased, with a fall in echo amplitude (i.e. posterior beam attenuation owing to the high reflectivity of the fat liver parenchyma). Grade 3 was markedly increased with a fall in echo amplitude and impaired portal vessel wall distinction, i.e. loss of echoes from the walls of the portal vein [9, 21].

Statistical Analysis

Descriptive results of continuous variables are expressed as mean (±standard deviation). A preliminary comparison of differences between the two groups of patients (NAFLD vs. non NAFLD) was performed by Student’s t-test and χ2, as suitable in the specific subsets. Comparison was performed by ANOVA and χ2 in NAFLD patients, according to their coffee drinking habits (coffee use) and degree of BLS. Odds ratio to bright liver severity was calculated. Multiple linear regression was aimed to recognize predictive effects to BLS of coffee drinking habits and/or of any other of the measured variables, including γ-GT.

A two sided p-value <0.05 was considered statistically significant. All analyses were performed using SPSS 13.0 for Windows (SPSS, Chicago, IL, USA).

Results

Anthropometric and biochemical characteristics of the subjects, NAFLD and controls, are shown in Table 1. There were no significant differences in regard to age and coffee drinking habits (number of cups of coffee/day). The differences found between NAFLD and controls were higher γ-GT, alkaline phosphatase, triglycerides and lower HDL-cholesterol. Moreover, higher HOMA, insulin, and fat mass assessed by BIA were observed (Table 1). Liver dimensions were greater in NAFLD patients, all with US bright liver, in comparison to the control group without bright liver.
Table 1

Comparison of NAFLD vs. control patientsa

Characteristic

NAFLD (n = 157)

Control (n = 153)

p

Age (years)

49.67 ± 13.52

47.82 ± 10.39

0.180

Cups of coffee/day

2.25 ± 1.59

2.05 ± 1.71

0.282

Cigarettes smoked/day

12.94 ± 17.22

7.83 ± 11.17

0.004*

BMI (kg/m2)

31.99 ± 5.52

24.49 ± 3.57

<0.0001*

Fat mass (%)

41.23 ± 12.22

36.13 ± 17.68

0.009*

Blood glucose (mg/dl)

97.55 ± 19.79

88.92 ± 16.17

<0.0001*

Insulin (µIU/ml)

14.49 ± 12.77

6.98 ± 3.73

<0.0001*

HOMA-IR

3.60 ± 3.41

1.55 ± 0.87

<0.0001*

Total cholesterol (mg/dl)

200.90 ± 39.44

198.66 ± 38.90

0.614

HDL cholesterol (mg/dl)

49.02 ± 11.52

57.87 ± 18.74

<0.0001*

Triglycerides (mg/dl)

132.04 ± 61.91

108.84 ± 82.30

0.005*

LDL cholesterol (mg/dl)

125.61 ± 35.49

119.00 ± 34.78

0.100

γGT (U/l)

40.50 ± 45.58

27.60 ± 31.22

0.004*

AST (U/l)

24.13 ± 9.07

21.53 ± 7.08

0.005*

ALT (U/l)

19.43 ± 6.28

17.42 ± 5.45

0.029*

Alkaline phosphatase (U/l)

73.09 ± 23.82

66.27 ± 21.80

0.009*

Longitudinal right liver (mm)

160.71 ± 18.75

137.86 ± 20.30

<0.0001*

Longitudinal left liver (mm)

98.45 ± 17.92

88.47 ± 14.23

<0.0001*

Caudate liver lobe (mm)

42.37 ± 9.11

39.43 ± 8.63

0.004*

Portal vein diameter (cm)

1.02 ± 0.09

1.04 ± 0.58

0.588

Portal vein blood flow (cm/s)

22.52 ± 2.97

23.76 ± 18.13

0.399

Women

80

83

0.532b

Men

77

70

Values given as n or mean ± standard deviation as appropriate

NAFLD non-alcoholic fatty liver disease, BMI body mass index, HOMA homoeostasis model-insulin resistance index, HDL high-density lipoprotein, LDL low-density lipoprotein, γGT γ-glutamyltransferase, AST aspartateaminotransferase, ALT alanineaminotransferase

* Significant difference

aStudent’s t test

bPearson's chi-square

The NAFLD patients (n = 137) did not show any significant difference with regard to coffee (n = 29) vs. non-coffee user groups (n = 108), apart from slightly higher levels of serum total cholesterol. More specifically, no significant differences were seen for smoking, assessed as daily number of cigarettes, age, BMI, blood pressure, triglycerides or HOMA.

Significant direct linear correlations (Table 2) between number of cups of coffee vs. total cholesterol and cigarette smoking and inverse linear correlation vs. ultrasound fatty liver appearance, assessed as BLS, were found. No correlation between number of cups of coffee and insulin resistance assessed by HOMA was present, in any group, despite a significant relationship of HOMA vs. BLS (r = 0,377; p < 0.0001). A significant inverse correlation with caudate liver lobe linear measurements and coffee was present only in NAFLD patients, along with a significant relationship between liver dimensions and BLS (Table 2).
Table 2

Correlations

Characteristic

Cups of coffee/day

NAFLD (n = 157)

Controls (n = 153)

r

p

r

p

Age

−0.047

0.562

−0.153

0.059

Cups of coffee/day

1

1

Cigarettes/day

0.124

0.148

0.197

0.023*

BP systolic

0.074

0.366

0.073

0.373

BP diastolic

0.062

0.448

0.042

0.606

BMI

0.023

0.785

−0.070

0.390

Fat mass

−0.053

0.560

0.015

0.864

Blood glucose

−0.140

0.081

−0.021

0.801

Insulin

−0.051

0.527

−0.035

0.672

HOMA-IR

−0.083

0.301

−0.026

0.745

Blood urea

0.010

0.904

−0.045

0.579

Creatinine

0.075

0.353

−0.002

0.977

Total cholesterol

0.174

0.030*

0.095

0.242

HDL cholesterol

0.077

0.338

0.030

0.715

Triglycerides

−0.059

0.463

0.026

0.752

LDL cholesterol

0.186

0.020*

0.080

0.330

γGT

−0.044

0.587

0.034

0.682

AST

−0.091

0.259

0.162

0.046

ALT

0.128

0.326

−0.033

0.733

Alkaline phosphatase

−0.158

0.049*

−0.116

0.154

Longitudinal right liver

−0.141

0.078

−0.009

0.910

Longitudinal left liver

−0.213

0.007**

0.030

0.709

Caudate liver lobe

−0.186

0.019*

−0.021

0.738

BLS

−0.253

0.001**

NAFLD non-alcoholic fatty liver disease, BP blood pressure, BMI body mass index, HOMA homoeostasis model-insulin resistance index, HDL high-density lipoprotein, LDL low-density lipoprotein, γGT γ-glutamyltransferase, AST aspartateaminotransferase, ALT alanineaminotransferase, BLS bright liver score

p < 0.05; ** p < 0.01; *** p < 0.001

Graded lower fatty liver involvement (BLS) is present in coffee drinkers in comparison with non-coffee drinkers (Table 3), a finding that is significant by chi-square analysis (χ2= 15.986; p = 0.003).
Table 3

Pearson’s chi-square in 157 NAFLD patients

Group

BLS 1

BLS 2

BLS 3

Total

Pearson chi-square

Coffee group 1

9

12

9

30

21.370

Coffee group 2

21

37

7

65

Coffee group 3

29

33

0

62

Total

59

82

16

157

Coffee group 1: no coffee; coffee group 2: <3 cups of coffee/day; coffee group 3: ≥3 cups of coffee/day

BLS 1: light bright liver; BLS 2: moderate bright liver; BLS 3: severe bright liver

p < 0.0001

Odds ratios to NAFLD, i.e. to more severe BLS, are shown in Fig. 1. Obesity, higher insulin resistance, lower HDL cholesterol, older age and arterial hypertension are associated with a greater risk of more severe grades of BLS (>1); coffee drinking is associated with a lower risk of severe BLS.
https://static-content.springer.com/image/art%3A10.1007%2Fs10620-010-1143-3/MediaObjects/10620_2010_1143_Fig1_HTML.gif
Fig. 1

Odds ratios to more severe bright liver scores (BLS) are shown. Obesity, higher insulin resistance, lower HDL cholesterol, older age and arterial hypertension are associated with a greater risk of more severe grades of BLS (>1); coffee drinking is associated with a lower risk of severe BLS

By multiple regression analysis, BLS is explained by number of cups of coffee, HOMA and BMI, which together account for 37.1% of the variance (Table 4). The analysis process ultimately excluded any transaminases, including γ-GT, explaining effect; also no effect of cholesterol and triglycerides was present. Due to the significant correlation between age and coffee drinking habits, the model was weighted by age. In this model, coffee use was inversely associated with the degree of bright liver, while insulin resistance and obesity, conversely, were directly associated with greater likelihood and severity of bright liver appearance.
Table 4

Stepwise linear regressions to BLS in NAFLD patients

Characteristic

β

p

CI 95%

BMI

4.142

<0.0001

0.019–0.055

Longitudinal right liver

3.267

<0.001

0.003–0.013

HOMA-IR

2.922

0.004

0.018–0.092

Cups of coffee/day

−2.585

0.011

−0.133 to −0.018

BLS bright liver score, NAFLD non-alcoholic fatty liver disease, BMI body mass index, HOMA homoeostasis model-insulin resistance index

R = 0.622, R2 = 0.387, F = 22.053, Sig = <0.0001

The complete model includes: cups of coffee/day, BMI, HOMA, total cholesterol, HDL cholesterol, triglycerides, longitudinal right liver length, longitudinal left liver length, caudate liver lobe length

Weighted least squares regression—weighted by age

Discussion

In our study we challenge the hypothesis that a very prevalent liver condition, fatty liver, when not associated with alcohol use and diabetes, and which is conceivably mainly related to insulin resistance and lifestyles inducing obesity could be affected by the subject’s habit of coffee drinking. The model designed, aimed to explain the degree of BLS, i.e. the intensity of US fatty liver appearance, by BMI, HOMA and the number of coffee cups, demonstrates a graded protective effect of the quantity of coffee. Conversely, BMI and HOMA, together with obesity and insulin resistance, account for greater BLS. No independent effect of fat mass percentage, assessed by BIA, toward the severity of bright liver was found.

In this model transaminase, γ-GT, cholesterol and triglycerides do not contribute significantly to explain the possible presence of the degree of bright liver, assessed by US BLS and not as a single morphological element. On the other hand, HOMA, included as a putative hallmark of insulin resistance, has a significant explicatory effect along with BMI, i.e. the degree of obesity and of insulin resistance, as expected, account for the presence of US bright liver. Conversely, the quantity of habitual coffee intake, assessed as a continuous variable, has an inverse relationship to the presence and the degree of bright liver.

Coffee intake, conjecturally, appears to exert a sort of “protective” effect toward the disadvantages of overweightness and insulin resistance.

We did not find any correlation between number of cups of coffee habitually taken and insulin resistance assessed by HOMA in the NAFLD nor the control group patients (Table 2). This finding does not support any likely direct metabolic effect of coffee drinking habits through an insulin resistance mechanism and, despite the high IR baseline levels and a possible protective coffee effect on fatty liver appearance, coffee and IR, not sharing complementary relationships, appear to have independent effects on NAFLD.

The concurrent effect of coffee habits, obesity grade and insulin resistance could also be conceivably important in clinical trials with drugs and/or therapeutic regimens addressed to modify degree of bright liver and/or metabolic profiles in NAFLD, as a confounding factor, and it is reasonable to suggest that this issue must be taken into account in study design and evaluation.

The beneficial effects of coffee on abnormal liver biochemistry, cirrhosis and hepatocellular carcinoma have been reported by several previous studies. There are many putative mechanisms explaining how coffee drinking may protect against liver damage with evidence both supporting and refuting a caffeine-mediated and a diterpene-mediated hypothesis [25]. Steatosis alone is considered to be relatively innocuous, is usually reversible, and it is the development of liver cell ballooning and inflammation (steatohepatitis) that determines whether a patient progresses to irreversible liver damage and fibrosis. Different mechanisms may be operating, but NAFLD and insulin resistance link is presently well established [26]. There is also an impact of steatosis on HCV treatment response, even with controversial interpretations [27]. Also, hepatocellular carcinoma is a possible complication of NAFLD, but whether it is related to frequently associated metabolic disorders (e.g., overweightness, diabetes) or to underlying cirrhosis, as a sequel itself of NAFLD, is not definite [1, 2]. This can be due to a possible role of lipid metabolism in pathogenesis of HCV infection, which includes liver injury and hepatocarcinogenesis mechanisms.

The observation that coffee drinking has an inverse and graded association with the risk of liver cancer and that high serum γ-GT is associated with an increased risk of liver cancer [1] suggests that different links can exist among all these factors. Long-term use of coffee might also result in acquired tolerance to caffeine, which could account for a different metabolic effect compared with short-term use of caffeine [28]. Excessive coffee drinking was associated with other variables that suggest an unhealthy lifestyle; however, the most proposed antidiabetogenic effects of coffee seem to involve improved insulin sensitivity rather than improved beta-cell function [29], a feature that is not confirmed by our study.

Conclusion

Coffee drinking in patients with NAFLD is inversely associated with the degree of bright liver involvement and appearance, with a trend opposite to overweightness. It is shown that this last factor, assessed by BMI, along with insulin resistance, assessed by HOMA, increases the likelihood and severity of bright liver appearance. A possible favorable and/or preventive role of coffee use in the natural history of NAFLD, similar to that reported in hepatocarcinoma and cirrhosis, and not mediated by insulin resistance, is envisaged.

Copyright information

© Springer Science+Business Media, LLC 2010