who met at least 1 of the following exclusion criteria (consuming ≥20 g/day of alcohol, testing positive for HBs-Ag or, demonstrating evidence of liver disease other than viral hepatitis, and/or currently taking anti-hypertensive agents, lipid-lowing agents, or antidiabetic agents), the records of 6,370 Japanese subjects were reviewed for identification of subjects meeting the diagnostic criteria for NAFLD (evidence of hepato-renal contrast and liver brightness on abdominal ultrasound) and the variables associated with NAFLD in order to estimate ideal ALT cutoff levels. Results: The results of multivariate analysis of the 1,346 subjects (803 males and 543 females) who met the diagnostic criteria for NAFLD confirmed that the traditional markers of metabolic disease are also markers of NAFLD: high BMI, ALT, TG, HDL-c, UA, and HbA1c levels. Of the significant risk factors for NAFLD, ALT level, one the most common variables in screening for liver disease, was selected for estimation of its cutoff level in diagnosing NAFLD. The area under receiver operating characteristic curve (AUC), cutoff level, sensitivity, specificity, PPV, NPV, and diagnostic accuracy in predicting NAFLD in males were found to be 0.807, 25 IU/l, 75.2%, 71.0%, 63.2%, 81.7%, and 73.1%, respectively, whereas those for females were found to be 0.747, 17 IU/l, 65.8%, 70.5%, 24.2%, 93.5%, and 69.9%, respectively. The sensitivity, specificity, PPV, and diagnostic accuracy results indicate that using a cut-off level of 25 U/L for males and 17 U/L for females allows for diagnosis of NAFLD with a high degree of specificity and accuracy, while the results for NPV indicate the utility of using these levels as tools in the identification of healthy subjects for further research. Conclusions: ALT level was confirmed to be a surrogate marker for NAFLD in addition to markers associated with metabolic diseases. The ALT cutoff level used in NAFLD diagnosis should be revised downward to identify subjects at risk of NAFLD to prevent NAFLD progression and the development of associated diseases.
Plasma free choline (fCh) level is a useful diagnostic marker for predicting NASH. The area under the ROC curves is shown for the performance of the plasma fCh levels for predicting NASH in this figure. The vertical axis represents the sensitivity, and the horizontal axis represents the 1-specificity. Analysis of the relationship between these two factors was performed in 110 NAFLD patients.
Tu1032 Relationship Between the Plasma Concentrations of Free Choline and Parameters of Adiposity and Histological Findings in Patients With Nonalcoholic Fatty Liver Disease. a Multicenter Study in Japan Kento Imajo, Masato Yoneda, Koji Fujita, Yuichi Nozaki, Yuji Ogawa, Yoshiyasu Shinohara, Shingo Kato, Hironori Mawatari, Wataru Shibata, Hiroyuki Kirikoshi, Koichiro Wada, Satoru Saito, Atsushi Nakajima
Tu1033 High Levels of Ursodeoxycholic Acid Act as FXR Antagonist and Deplete Liver Cholesterol Due to Increased Bile Acid Synthesis in Morbidly Obese Patients Hanns-Ulrich Marschall, Michaela Mueller, Martin Wagner, Peter Fickert, Gernot Zollner, Dagmar Silbert, Karoline Lackner, Anders Thorell, Michael H. Trauner
Aim. Choline is a dietary component that is crucial for normal function of all cells. Most of choline is absorbed from small intestine and metabolized in the liver. We recently demonstrated that free choline (fCh) levels in blood reflect the level of liver phosphatidylcholine (PC) synthesis disability and are correlated with the disease onset of nonalcoholic steatohepatitis (NASH) (Hepatology 2009). Our aim was to investigate the association between plasma fCh concentrations and parameters of adiposity or histological findings and to validate the utility of this novel biomarker for NASH diagnosis. Methods. Our cohort consisted of 110 patients with biopsy proven nonalcoholic fatty liver disease (NAFLD) from four centers across the Japan and 25 age-matched healthy controls. Plasma fCh levels were measured using high-performance liquid chromatography (HPLC). Results. The plasma fCh levels were markedly increased in patients with NASH or borderline diagnosis as compared to simple steatosis or healthy control subjects (P < 0.01). An association between plasma fCh levels and parameters of adiposity (HOMA-IR, VFA, and SFA) was reflected in the significantly positive correlation present between these two variables in the patients with NAFLD (r=0.341, P<0.01, r=0.252, P<0.05, r=0.223, P<0.05, respectively), whereas an association between plasma fCh levels and lipoprotein synthesis (liver phosphatidylcholine synthesis (PC synthesis) and expression of microsomal triglyceride transfer protein (MTTP) mRNA) was reflected in the markedly negative correlation (r=0.341, P<0.01, r=0.252, P<0.05, r=0.223, P<0.05, respectively). Moreover, the risk of having steatosis and fibrosis on liver biopsy increased with increasing plasma fCh levels (P<0.05). The areas under the receiveroperating characteristic curves for NASH including borderline diagnosis were 0.811(Figure 2). Moreover, the areas under the ROC for fibrosis stage were 0.758 for ≥F2, 0.821 for ≥F3, 0.881 for ≥F4. Conclusion: Our study showed that the plasma fCh levels can be closely related to parameters of adiposity and the grade of liver steatosis and fibrosis. This biomarker appears to be a feasible, non-invasive and useful for distinguishing patients with NASH including borderline diagnosis from those with not NASH with a high accuracy.
Background: Ursodeoxycholic acid (UDCA) was shown to improve insulin resistance and steatosis in mice. The efficacy and possible modes of action of UDCA treatment in human non-alcoholic fatty liver disease (NAFLD) have been debated. We aimed to explore potential mechanism of UDCA action in patients with morbid obesity awaiting Roux-en-Y gastric bypass surgery. Methods: Forty morbidly obese patients were randomized to UDCA (20mg/ kg/day three weeks before surgery) or no treatment (controls). Serum liver function tests, lipids, bile acids and markers of insulin resistance/diabetes (OGTT, HOMA) were obtained before and after treatment. During surgery, biopsies were taken from the liver for histology and gene as well as protein expression studies. Results: Three patients dropped out; UDCA 1 (diarrhea), controls 2 (pregnancy, bleeding). Completers of both groups were well matched by gender (female, 68.4 vs. 77.7%), age (42.8±12.3 vs. 38.5±10.1 years), BMI (41.9±4.6 vs. 40.6±3.9 kg/m2), HOMA (5.1±2.5 vs. 6.6±3.9) and OGTT (IGT or T2DM, 37% vs. 50%). NAS scores were: no, 11 vs. 13; borderline, 4 vs. 4; definite, 4 vs. 1. UDCA despite significantly (p<0.05) expanding the BA pool 10.6±7.6 fold (≤ 55.3 μmol/L; UDCA >90%) increased bile acid synthesis as measured by serum C4 (7α-hydroxy-cholest-4-ene-3-one), CYP7A1 gene expression, and serum levels of primary bile acids CDCA and CA. Circulating FGF19 decreased by 18% (p=0.05). Significant increases in gene expression levels of key regulators of lipid turnover (SREBP2, SCD, HMGCR) were reflected by significantly decreased serum LDL-cholesterol and increased triglycerides. UDCA significantly decreased ALT, AST and gGT but did not affect HOMA, glucose tolerance, adiponectin and lectin. Conclusion: Changes in serum lipid and gene expression profiles in UDCA treated, morbidly obese patients indicate hepatic cholesterol depletion as a result of increased bile acid formation due to FXR-antagonistic effects of very high UDCA levels in noncholestatic livers. Tu1034 Noninvasive Hepatic Fibrosis Scores Predict Liver-Related Outcomes in Diabetic Patients Michael A. Dunn, Jaideep Behari, Michael R. O'Connell, Alessandro Furlan, Ayaz Aghayev, Serter Gumus, Melissa I. Saul, Kyongtae T. Bae There is an unmet need for easily performed noninvasive testing to predict outcomes in nonalcoholic fatty liver disease. We studied the clinical predictive value of 8 testing panels in 2437 diabetic adults followed for five years after a noncontrast CT scan assessed retrospectively for hepatic steatosis. We used a cutoff for liver fat content of 30%, measured by subtracting averaged spleen from liver attenuation. The method is insensitive for lesser degrees of steatosis but consistently indicates steatosis over 30%. We evaluated prediction of clinical outcomes using 8 test panels reported to indicate liver fibrosis—the AST:ALT Ratio (n = 1917 patients), AST: Platelet Ratio (n = 1690), AST:Platelet Ratio Index (n = 1690), FIB4 Index (n = 1688), BARD Score (n = 649), NAFLD Fibrosis Score (n =213), Cirrhosis Discriminant Score (n = 1052), and Goteborg University Cirrhosis Index (n = 1054). Scores were based on laboratory testing within 6 months of the CT scan. Outcomes included liver related deaths, liver transplants, hepatic encephalopathy, cardiovascular related deaths, and nonfatal cardiovascular events including myocardial infarction, strokes, angina, congestive heart failure, and arrhythmias. Results: Over 5 years, 430 patients died, including 22 liver related and 116 cardiovascular deaths. Nonfatal adverse liver outcomes included 76 liver transplants and 71 patients with hepatic encephalopathy. Nonfatal cardiovascular outcomes included 891 myocardial infarctions, 109 strokes, 270 occurrences of angina, 1034 occurrences of congestive heart failure, and 1043 arrhythmias. The quartile of patients with the highest fibrosis score was strongly and specifically associated with liver related deaths, liver transplants, and hepatic encephalopathy for 6 of the 8 test panels—AST:ALT Ratio, AST:Platelet Ratio, AST:Platelet Ratio Index, FIB4 Index, BARD Score, and NAFLD
Plasma free choline (fCh) levels are significantly higher in patients with borderline NASH or NASH than those in healthy volunteers or NAFL. The vertical axis represents the plasma fCh levels in mg/dL and the horizontal axis represents the patient groups. The box represents the interquartile range (box), median (the dot), range (lines) of the plasma fCh levels. The medians are 0.16 (ranging from 0.10 to 0.25 mg/dL), 0.15 (ranging from 0.10 to 0.22 mg/ dL), 0.11 (ranging from 0.06 to 0.19 mg/dL), and 0.09 (ranging from 0.08 to 0.12 mg/dL) mg/dL for the NASH (n=42), borderline NASH (n=32), NAFL (n=36), and control subjects (n=25) (P <0.01). All data are expressed as the mean ± SD.