Abstract
Introduction. Hepatic steatosis due to non-alcoholic fatty liver disease is associated with obesity, dyslipidemia, insulin resistance, and type 2 diabetes. The Finnish Diabetes Risk Score (FINDRISC) is a prognostic screening tool to detect people at risk for type 2 diabetes without the use of any blood test. The objective of this study was to evaluate whether FINDRISC can also be used to screen for the presence of hepatic steatosis.
Patients and methods. Steatosis was determined by ultrasound. The study sample consisted of 821 non-diabetic subjects without previous hepatic disease; 81% were men (mean age 45 ± 9 years) and 19% women (mean age 41 ± 10 years).
Results. Steatosis was present in 44% of men and 10% of women. The odds ratio for one unit increase in the FINDRISC associated with the risk of steatosis was 1.30 (95% CI 1.25–1.35), similar for men and women. The area under the receiver operating characteristics curve for steatosis was 0.80 (95% CI 0.77–0.83); 0.80 in men (95% CI 0.77–0.83) and 0.83 (95% CI 0.73–0.93) in women.
Conclusions. Our data suggest that the FINDRISC could be a useful primary screening tool for the presence of steatosis.
| Abbreviations | ||
| ALT | = | alanine aminotransferase |
| AST | = | aspartate aminotransferase |
| AUDIT | = | Alcohol Using Disorders Identification Test |
| BMI | = | body mass index |
| CRP | = | C-reactive protein |
| FINDRISC | = | Finnish Diabetes Risk Score |
| FLI | = | fatty liver index |
| LAP | = | lipid accumulation product |
| METS | = | metabolic equivalents |
| NAFLD | = | non-alcoholic fatty liver disease |
| NASH | = | non-alcoholic steatohepatitis |
Key messages
Hepatic steatosis due to non-alcoholic fatty liver disease is associated with obesity, insulin resistance, and type 2 diabetes.
The Finnish Diabetes Risk Score (FINDRISC) is a useful tool to detect insulin resistance and the risk of type 2 diabetes.
FINDRISC can be used to screen for hepatic steatosis with a good discriminative power.
Introduction
Non-alcoholic fatty liver disease (NAFLD) comprises a wide group of progressive alterations in liver structure and function, ranging from hepatic steatosis and non-alcoholic steatohepatitis (NASH) to fibrosis and cirrhosis (Citation1). NAFLD is commonly associated with abdominal obesity, atherogenic dyslipidemia, and, of great importance, insulin resistance and type 2 diabetes (Citation2–4). In clinical practice steatosis is diagnosed by hepatic ultrasound scanning which presents an adequate sensitivity and specificity for this disease (Citation5). In addition, hepatic ultrasound is useful for the differential diagnosis between NAFLD and other liver and biliary diseases that alter hepatic laboratorial tests. However, there is controversy about the cost-effectiveness of the indiscriminate use of this examination, since it does not separate the more severe cases of NASH or fibrosis from more benign forms of steatosis (Citation6). Also, ultrasound is not cheap and requires skilled personnel to carry out the work. Therefore, it would be important to develop a simple and inexpensive NAFLD screening tool for primary health care in order to detect subjects that will benefit from more specific/advanced tests.
For public health purposes a risk assessment tool for type 2 diabetes was developed in Finland (Citation7–12). This Finnish Diabetes Risk Score (FINDRISC) is based on easily available information, and it uses eight parameters. The FINDRISC was shown to predict the 10-year risk of (drug-treated) type 2 diabetes with 78%–81% sensitivity and 76%–77% specificity, and it also detects reasonably well asymptomatic type 2 diabetes and abnormal glucose tolerance (Citation8). It was recently validated in the Italian population with identical results (Citation9) and was recommended by the European Society for studies in Diabetes and the European Society of Cardiology as a screening tool for type 2 diabetes (Citation10). In addition to the prediction of type 2 diabetes FINDRISC also predicts the incidence of myocardial infarction and stroke (Citation11). FINDRISC can also be used to identify subjects with higher levels of insulin resistance (Citation12).
Since there is a close association of obesity, insulin resistance, and type 2 diabetes with NAFLD (Citation1–4) we hypothesize that the FINDRISC could also be a good primary screening tool for detecting the presence of steatosis, one of its main components. Therefore, the aim of this study was to evaluate whether the FINDRISC can be used to screen for the presence of hepatic steatosis in asymptomatic subjects attending an obligatory medical check-up.
Patients and methods
The study population consisted of 991 consecutive asymptomatic men and women, who were submitted to an obligatory clinical and laboratory health evaluation paid by their employers from December 2008 to February 2009 at the Preventive Medicine Center of the Albert Einstein Hospital in São Paulo, Brazil. The examination protocol consisted of a clinical consultation, blood laboratory tests, a symptom-limited exercise stress test, and an ultrasonographic abdominal scan. All individuals provided details of their demographic, medical history, smoking status, and use of medication at the time of their clinical consultation. Quantitative alcohol consumption was measured by the Alcohol Using Disorders Identification Test (AUDIT) applied by psychologists (Citation13). Individuals with missing data and those with a previous history of liver disease, defined as a positive test for hepatitis, history of cirrhosis, biliary disease, or diabetes mellitus were excluded from the present analysis.
Information regarding medical history was obtained by a self-administered questionnaire. Diabetes was classified as being previously diagnosed by a physician or use of glucose-lowering medication. Hypertension and dyslipidemia were ascertained by a previous history of these conditions or the use of blood pressure or lipid-lowering medications. Smoking status was defined as never smoker, current smoker, or current non-smoker. During physical examinations, waist circumference was measured at the smallest diameter between the iliac crest and the costal margin using a plastic anthropometric tape held parallel to the floor. Weight (kilograms) and height (meters) were measured with standard physician's scales and a stadiometer. Obesity was defined as a body mass index > 30 kg/m2. Blood pressure was measured with a mercury sphygmomanometer as recommended by the American Heart Association (Citation14).
Blood specimens were collected after an overnight fast. Plasma lipid, glucose, and liver enzymes (alanine aminotransferase (ALT), aspartate aminotransferase (AST), and gamma glutaryl transferase (gamma GT)) levels were measured by standardized automated laboratory tests using a Vitros platform (Johnson & Johnson Clinical Diagnostics, Rochester, New York, USA). High-sensitivity C-reactive protein (CRP) levels were determined by immunonephelometry (Dade-Behring, US). All tests were performed at the Central Laboratory of the Albert Einstein Hospital.
The study subjects underwent symptom-limited treadmill exercise testing according to the Bruce and Ellestad protocols. Cardio-respiratory fitness (‘fitness’) was quantified as the maximal metabolic equivalents (METS) attained (1 MET = 3.5 mL O2 kg21 min21) determined from the final speed and grade attained (Citation15).
Hepatic steatosis was diagnosed after a minimum of 6 hours’ fast using an ACUSON XP-10 device (Mountain View, CA, USA) and was identified by the presence of an ultrasonographic pattern of a bright liver, with evident contrast between hepatic and renal parenchyma, as previously described (Citation16). All hepatic ultrasounds were performed by trained and board-certified radiologists.
The FINDRISC questionnaire was applied by registered nutritionists at the time of the routine diet interview of the health evaluation protocol. As previously described (Citation7), points were attributed for the following parameters: age, self-reported use of blood pressure medication, history of high blood glucose, physical activity at least 4 h a week, daily consumption of vegetables, fruits, and berries, and measured body mass index (BMI, kg/m2) and waist circumference.
The ethics committee of the Hospital Israelita Albert Einstein approved this study, and a waiver for informed consent was obtained.
Statistical analysis
Statistical analyses were performed with SPSS for Mac version 16.1. Differences between continuous variables were calculated using the Mann–Whitney test and those for categorical variables using the chi-square test. Risk of steatosis was assessed with logistic regression models. The goodness-of-fit was assessed by the Hosmer–Lemeshow test. Odds ratios and their respective 95% confidence intervals were calculated. Analyses were carried out first with both sexes combined and then separately in different models (model 1: univariate; models 2–6: adjusted for several covariates). The level of statistical significance was set to 0.05.
The prediction of steatosis by the FINDRISC was evaluated by receiver operating characteristic (ROC) curves. The ROC curve provided estimates of sensitivity (the probability that the test is positive for subjects who experienced the event) and specificity (the probability that the test is negative for people who did not experience the event during the follow-up) for each FINDRISC level. The ROC curves were plotted by the FINDRISC level, with sensitivity on the y-axis and the false positive rate (1 – specificity) on the x-axis. The area under the ROC curve (AUC) provides an overall estimate of the accuracy of the test. In addition, positive prediction and negative predictive vales were calculated with their respective 95% confidence intervals. Since increased adiposity is the main determinant of insulin resistance, the discriminative power of body mass index (BMI in kg/m2) and waist circumference (in cm) as markers of steatosis presence in comparison with FINDRISC was also evaluated. In addition we compared the performance of FINDRISC with two previously described indexes that predict the presence of steatosis: the lipid accumulation product (LAP), which uses waist circumference values and plasma triglycerides (Citation17), and the fatty liver index (FLI), which comprises the mathematical relation of BMI, waist circumference, gamma GT, as well as plasma triglyceride values for their calculations respectively (Citation18).
Results
The base-line characteristics of the 821 studied subjects (80.8% men) are presented in . Men had a higher mean BMI, waist circumference, and blood pressure values than did women (P values < 0. 001). The prevalence of obesity was 19% in men and 7% in women, respectively. Men had higher mean levels of total serum cholesterol, LDL cholesterol, and triglyceride than did women, whereas women had a higher mean HDL cholesterol level compared with men. The mean FINDRISC score in men was 8 and in women 6. Furthermore, men tend to have a higher AUDIT score compared with women (median 5 versus 3 points; P < 0.001). There was no statistically significant difference in smoking prevalence between men (9%) and women (6%). The prevalence of steatosis in the study population was 37%, significantly higher in men (44%) than in women (10%; P < 0.001).
Table I. Base-line characteristics of the study population.
presents the ROC information for the study population. The area under the ROC curve for steatosis was 0.80 (95% CI 0.77–0.83), 0.80 in men (95% CI 0.77–0.83) and 0.83 (95% CI 0.73–0.93) in women. shows the sensitivities, specificities, negative and predictive values, as well as the number of people presenting the best cut-off values of FINDRISC score, BMI, and waist circumference to screen for steatosis. Using the FINDRISC cut-off value of 8 points to identify undiagnosed steatosis, the sensitivity was 71% and specificity 74.5%, with positive and negative predictive values of 63% and 81%, respectively. Approximately 36% of the study population would need further testing for steatosis if this cut-off value were applied (39% of men and 25% of women). For men the sensitivity was 71%, specificity 75%, positive and negative predictive values 69% and 76%. The corresponding values in women were 72%, 73%, 22%, and 96%, respectively.
Figure 1. Receiver operating characteristics (ROC) curves for the prevalence of steatosis in the study population.
Table II. Characteristics of FINDRISC, BMI, and waist circumference for steatosis in Brazilian men and women.
The area under curve using BMI as predictor of steatosis was 0.83 (95% CI 0.81–0.86). When men and women were analyzed separately, the area under the curve was 0.81 in men (95% CI 0.78–0.84) and 0.80 (95% CI 0.68–0.92) in women. The corresponding value for waist circumference in predicting steatosis was 0.86 (95% CI 0.84–0.89). When men and women were analyzed separately, the area under the curve was 0.84 in men (95% CI 0.81–0.87) and 0.80 (95% CI 0.69–0.92) in women. Using a BMI cut-off value of 26.6 kg/m2 to identify undiagnosed steatosis resulted in a sensitivity of 74% in the study population. The corresponding specificity rate was 67%. When using a cut-off value of 93.6 cm for waist circumference, the sensitivity of predicting steatosis was 79% with a specificity of 78%.
The area under the curve using the FLI as predictor of steatosis was 0.86 (95% CI 0.85–0.87). When men and women were analyzed separately, the area under the curve was 0.83 in men (95% CI 0.82–0.84) and 0.90 (95% CI 0.87–0.93) in women (data not shown). The respective value for the LAP in predicting steatosis was 0.83 (95% CI 0.82–0.84). When men and women were analyzed separately, the area under the curve was 0.79 in men (95% CI 0.78–0.81) and 0.87 (95% CI 0.84–0.91) in women.
shows the odds ratio of steatosis for the different covariates alone and controlled for (adjusted models) in all study participants. In addition to male gender, smoking, and BMI, the FINDRISC was one of the strongest predictors of steatosis in the unadjusted model. The odds ratio for one unit increase in the FINDRISC for the risk of steatosis was 1.30 (95% CI 1.25–1.35). Smokers had significantly higher median AUDIT values than did ex-smokers and non-smokers (P < 0.001; data not shown).
Table III. Odds ratio (OR) of predictors of steatosis in Brazilians, with 95% confidence interval (CI).
After controlling for the covariates, only age, ALT, fasting glucose, waist circumference, and smoking remained statistically significantly related to steatosis. Whereas the risk increase regarding steatosis for age, fasting glucose, and ALT was with 2%–10% rather low, smoking increased the risk of steatosis by 156%.
The results of the logistic regression analysis in men showed that only age, ALT, BMI, waist circumference, and smoking were statistically significantly related with increased risk of steatosis when controlled for the covariates (). In women, however, only age and fasting glucose remained statistically significant in the multiple adjustment models.
Table IV. Odds ratio (OR) of predictors of steatosis in Brazilian men and women, with 95% confidence interval (CI).
shows the association of the FINDRISC as a continuous variable with the presence of steatosis. In the unadjusted model each unit increase in the FINDRISC questionnaire increased the odds ratios of steatosis by 1.31 in men and 1.34 in women. This association remained robust even after adjustment for blood pressure, liver enzymes, plasma lipids, smoking, and fasting glucose. However, significance was lost in women when the FINDRISC was adjusted for plasma glucose.
Table V. Odds ratios with 95% confidence intervals for the Finnish Diabetes Risk Score (FINDRISC) per unit increase for prediction of steatosis.
Discussion
The results of this study indicate that the FINDRISC questionnaire predicts well not only type 2 diabetes as previously shown (Citation7–9) but also hepatic steatosis. The area under the ROC curve for the whole study population, as well as separately for men and women, indicates a good discriminative value for the diagnosis of steatosis. The good negative predictive values for a score of 8 points suggests that in general and primary practice the FINDRISC could be a screening tool to select subjects that need a diagnostic test such as hepatic ultrasound. Thus, using the FINDRISC as the primary screening tool, it is possible to detect simultaneously the early risk of type 2 diabetes and hepatic steatosis, without any complex clinical testing. The positive FINDRISC (8 points or above) will identify the people who should be referred for further testing with more sophisticated and expensive methods.
The association of FINDRISC with steatosis remained statistically significant even after adjusting for factors associated with the presence of this disease, in particular hepatic enzymes, blood pressure, plasma lipids, smoking, or alcohol consumption measured by the AUDIT questionnaire. Despite the low prevalence of smokers in this study, smoking was independently associated with steatosis in men. Notwithstanding experimental data associating smoking to steatosis (Citation19), we have found that smokers consumed more alcohol than did non-smokers since their AUDIT scores were significantly higher. Therefore even considering the limitations of the AUDIT questionnaire (Citation20) to identify heavy drinkers, we cannot discard the possibility that these patients would also be prone to alcoholic steatosis in addition to or instead of NAFLD.
The predictive model lost its significance when adjusted for plasma glucose in women but not in men. Previous evidence has associated NAFLD with abnormalities in glucose tolerance, reflecting increased insulin resistance that accompanies both pathologic conditions (Citation21). However the importance of this latter finding for clinical practice is debatable since the idea of the FINDRISC questionnaire is to use it as a screening tool that can be applied by primary health care physicians and other health care professionals without the need of laboratory and/or imaging tests (Citation7). Identified high-risk individuals should be evaluated by special tests.
In general, the prevalence of NAFLD in Western populations ranges from 20% to 30% (Citation22). However, in obese subjects (BMI > 30 kg/m2) steatosis prevalence rises to 65%–75%, and there is evidence that NASH can be found in up to 20% of these subjects (Citation16). The prevalence of steatosis found in our study population comprising people obliged by their employers to evaluate their health status is within these figures. Currently it is believed that 10% of the cryptogenic forms of cirrhosis occur due to NAFLD (Citation23). Due to the increasing prevalence of abdominal obesity and the clustering of metabolic alterations to which it is associated, there is a great concern that NAFLD may become a common cause of liver failure in the near future (Citation6). Therefore it is important to detect NAFLD early in order to implement intensive life-style interventions/modifications, possibly combined with pharmacological therapy, for better control of NAFLD as well as type 2 diabetes and cardiovascular disease combined.
In our study BMI and waist circumference performed at least as well as the FINDRISC in predicting steatosis, and it may be argued that BMI or waist circumference might be sufficient in identifying people in need of additional tests. However, it has to be kept in mind that the FINDRISC predicts not only steatosis but also cardiovascular disease (Citation10) and diabetes risk (Citation7–9). Thus, this prognostic tool needs to be filled in once but offers simultaneously prognosis for various diseases, in contrast to BMI or waist circumference. In addition FINDRISC performed equally in predicting steatosis to other risk calculators such as the LAP (Citation17) and the FLI (Citation18). However, the FINDRISC is much easier to calculate and use in general practice since it does not need laboratory measurements and mathematical calculations.
Naturally, this study had limitations. Hepatic ultrasound cannot be used as a tool to separate steatosis from more severe forms of NAFLD, like NASH (Citation1); in addition it cannot detect earlier stages of steatosis or quantify its severity like proton magnetic resonance imaging or liver computed tomography (Citation1,Citation24). Therefore, ultrasound could have under-estimated the already high prevalence of steatosis in our population. Although magnetic resonance imaging and computed tomography are the best examinations for detecting and quantifying liver fat (Citation2), hepatic ultrasound due to its lower cost and availability is still the imaging exam of choice for diagnosing steatosis routinely. It would also be interesting to test how FINDRISC correlates with scores developed for predicting more severe forms of NAFLD like NASH, and liver fibrosis (Citation25). However, the diagnosis of NASH and fibrosis can only be confirmed by hepatic biopsy and cannot be made based on elevations of hepatic enzymes alone (Citation1,Citation2). Furthermore, this study was performed in Brazilian people, a population where the predictive power of FINDRISC for type 2 diabetes is not yet validated. However, since risk factors associated with NAFLD and insulin resistance and diabetes are common for all populations (Citation3) we assume that our findings should be reproducible in different populations, too. Even though prevalence estimates of NAFLD may be incorrect, since our target population was an opportunistic sample and not a representative population sample, the associations between various parameters studied are likely to be unbiased. It is also important to test if the FINDRISC questionnaire could be used as a screening for more severe NAFLD, diagnosed by histological evaluation.
In conclusion, the FINDRISC questionnaire presents a good tool with an adequate discriminative power for the detection of hepatic steatosis detected by ultrasound. Our data suggest that it could be a useful tool for screening of this disease in primary care. Since NAFLD, diabetes, and cardiovascular disease share common risk factors and outcome, people with a high risk of diabetes and NAFLD should be submitted to intensive life-style modification and possibly pharmacological treatment if necessary. While there is unequivocal evidence that type 2 diabetes can be prevented by life-style intervention in high-risk individuals identified by the FINDRISC (Citation26), prospective studies are necessary to test if the FINDRISC would be able to identify people at a high risk of evolution to cirrhosis and liver failure due to NAFLD.
Declaration of interest: The authors state no conflict of interest and have received no payment in preparation of this manuscript.
References
- Preiss D, Sattar N. Non-alcoholic fatty liver disease: an overview of prevalence, diagnosis, pathogenesis and treatment considerations. Clin Sci (Lond). 2008;115:141–50.
- Cortez-Pinto H, Camilo ME. Non-alcoholic fatty liver disease/non-alcoholic steatohepatitis (NAFLD/NASH): diagnosis and clinical course. Best Pract Res Clin Gastroenterol. 2004;18:1089–104.
- Bugianesi E, Gastaldelli A, Vanni E, Gambino R, Cassader M, Baldi S, . M. Insulin resistance in non-diabetic patients with non-alcoholic fatty liver disease: sites and mechanisms. Diabetologia. 2005;48:634–42.
- Gastaldelli A, Kozakova M, Hojlund K, Flyvbjerg A, Favuzzi A, Mitrakou A, . Fatty liver is associated with insulin resistance, risk of coronary heart disease, and early atherosclerosis in a large European population. Hepatology. 2009;49:1537–44.
- Joseph AEA, Saverymuttu SH, Al-Sam S, Cook MG, Maxwell JD. Comparison of liver histology with ultrasonography in assessing diffuse parenchymal liver disease. Clin Radiol. 1991;43:26–31.
- Ahmeda MH, Abua EO, Byrne CD. Non-alcoholic fatty liver disease (NAFLD): New challenge for general practitioners and important burden for health authorities? Prim Care Diabetes. 2010;4:129–37.
- Lindström J, Tuomilehto J. The diabetes risk score: a practical tool to predict type 2 diabetes risk. Diabetes Care. 2003;26:725–31.
- Saaristo T, Peltonen M, Lindström J, Saarikoski L, Sundvall J, Eriksson JG, . Cross-sectional evaluation of the Finnish Diabetes Risk Score: a tool to identify undetected type 2 diabetes, abnormal glucose tolerance and metabolic syndrome. Diab Vasc Dis Res. 2005;2:67–72.
- Franciosi M, De Berardis G, Rossi MC, Sacco M, Belfiglio M, Pellegrini F, . Use of the diabetes risk score for opportunistic screening of undiagnosed diabetes and impaired glucose tolerance: the IGLOO (Impaired Glucose Tolerance and Long-Term Outcomes Observational) study. Diabetes Care. 2005;28:1187–94.
- Rydén L, Standl E, Bartnik M, Van den Berghe G, Betteridge J, de Boer MJ, . Task Force on Diabetes and Cardiovascular Diseases of the European Society of Cardiology (ESC); European Association for the Study of Diabetes (EASD). Guidelines on diabetes, pre-diabetes, and cardiovascular diseases: The Task Force on Diabetes and Cardiovascular Diseases of the European Society of Cardiology (ESC) and of the European Association for the Study of Diabetes (EASD). Eur Heart J. 2007;28:88–136.
- Silventoinen K, Pankow J, Lindström J, Jousilahti P, Hu G, Tuomilehto J. The validity of the Finnish Diabetes Risk Score for the prediction of the incidence of coronary heart disease and stroke, and total mortality. Eur J Cardiovasc Prev Rehabil. 2005;12:451–8.
- Schwarz PE, Li J, Reimann M, Schutte AE, Bergmann A, Hanefeld M, . The Finnish Diabetes Risk Score is associated with insulin resistance and progression towards type 2 diabetes. J Clin Endocrinol Metab. 2009;94:920–6.
- Bush K, Kivlahan DR, McDonell MB, Fihn SD, Bradley KA. The AUDIT alcohol consumption questions (AUDIT-C): an effective brief screening test for problem drinking. Ambulatory Care Quality Improvement Project (ACQUIP). Alcohol Use Disorders Identification Test. Arch Intern Med. 1998;158:1789–95.
- Perloff D, Grim C, Flack J, Frohlich ED, Hill M, McDonald M, . Human blood pressure determination by sphygmomanometry. Circulation. 1993;88 (5 Pt 1):2460–70.
- Fletcher GF, Balady GJ, Amsterdam EA, Chaitman B, Eckel R, Fleg J, . Exercise standards for testing and training: a statement for healthcare professionals from the American Heart Association. Circulation. 2001;104;1694–740.
- Bellentani S, Saccoccio G, Masutti F, Crocè LS, Brandi G, Sasso F, . Prevalence of and risk factors for hepatic steatosis in Northern Italy. Ann Intern Med. 2000;132:112–7.
- Kahn SH. The ‘lipid accumulation product’ performs better than the body mass index for recognizing cardiovascular risk: a population-based comparison. BMC Cardiovasc Disord. 2005;5:26.
- Bedogni G, Bellentani S, Miglioli L, Masutti F, Passalacqua M, Castiglione A, . The Fatty Liver Index: a simple and accurate predictor of hepatic steatosis in the general population. BMC Gastroenterol. 2006;6:33.
- Bailey SM, Mantena SK, Millender-Swain T, Cakir Y, Jhala NC, Chieng D, . Ethanol and tobacco smoke increase hepatic steatosis and hypoxia in the hypercholesterolemic apoE(−/−) mouse: implications for a ‘multihit’ hypothesis of fatty liver disease. Free Radic Biol Med. 2009;46:928–38.
- Bradley KA, Boyd-Wickizer J, Powell SH, Burman ML. Alcohol screening questionnaires in women: a critical review. JAMA. 1998;280:166–71.
- Sargin M, Uygur-Bayramicli O, Sargin H, Orbay E, Yayla A. Association of nonalcoholic fatty liver disease with insulin resistance: is OGTT indicated in nonalcoholic fatty liver disease? J Clin Gastroenterol. 2003;37:399–402.
- Angulo P. Nonalcoholic fatty liver disease. N Engl J Med. 2002;346:1221–31.
- Clark JM, Diehl AM. Nonalcoholic fatty liver disease: an underrecognized cause of cryptogenic cirrhosis. JAMA. 2003;289:3000–4.
- Marques MD, Santos RD, Parga JR, Rocha-Filho JA, Quaglia LA, Miname MH, . Relation between visceral fat and coronary artery disease evaluated by multidetector computed tomography. Atherosclerosis. 2010;209:481–6.
- Shah AG, Lydecker A, Murray K, Tetri BN, Contos MJ, Sanyal AJ; Nash Clinical Research Network. Comparison of noninvasive markers of fibrosis in patients with nonalcoholic fatty liver disease. Clin Gastroenterol Hepatol. 2009;7: 1104–12.
- Lindström J, Peltonen M, Eriksson JG, Aunola S, Hämäläinen H, Ilanne-Parikka P, . Determinants for the effectiveness of lifestyle intervention in the Finnish Diabetes Prevention Study. Diabetes Care. 2008;31:857–62.