BACKGROUND: Recent studies have shown that 43% of patients with NAFLD improve with vitamin E therapy. Currently, this treatment response can only be evaluated by liver puncture. Therefore, there is an urgent need for biomarkers to evaluate and predict the therapeutic response to vitamin E. OBJECTIVE: To use metabolites to perform the following studies: 1. Screening of biomarkers for the determination of the efficacy of vitamin E. 2. Screening of biomarkers for the prediction of the efficacy of vitamin E. METHODS: Blood specimens were retained at baseline as well as at the treatment endpoint from three groups of patients enrolled in the PIVENS study: 1. vitamin E effective group (16 patients); 2. vitamin E ineffective group (15 patients); 3. placebo effective group (15 patients). Effective was defined as meeting the primary treatment endpoint of the trial. Metabolites were analyzed using GC/MS and LC/MS platforms. Categorical data were analyzed for continuous variables using chi-square test, Tukey’s post hoc test combined with ANOVA/covariance analysis. Baseline predictors of responsiveness to vitamin E treatment were analyzed using logistic regression. Patients with NAFLD were evaluated separately for steatosis, ballooning, and intralobular inflammatory conditions in addition to overall status. RESULTS: The three groups of patients were comparable in terms of demographics, clinical characteristics, and baseline histology. Biological markers of treatment response included a significant decrease in glutamyl amino acids (including leucine and valine) in the vitamin E effective group compared to patients who did not respond to vitamin E treatment (0.82, P=0.02 and 0.80, P=0.03, respectively). Sphingosine levels were also decreased in the histologically responsive patients (0.64-fold decrease, p=0.02). Also these two biomarkers were markers of the placebo effective group (glutamyl amino acids 0.85, p=0.05 and sphingosine 0.63, p=0.02). Additional markers were mainly related to individual histological characteristics: ballooning (low bilirubin 0.55, p=0.04 and high succinate 1.28, p=0.03); inflammation (high hydrocinnamic acid 1.35, p=0.05 and high succinate 1.28, p=0.03). Baseline predictors of vitamin E efficacy included: 3-phenylpropionate (OR: 29.4, 95% CI: 1.23-707.0), menadione (OR:20.2, 95% CI: 1.2-338.6) indolepropionate (OR:16.2, 95% CI: 1.45-180.7). CONCLUSION: In patients with nonalcoholic fatty liver, the decrease in markers of oxidative stress response after treatment with vitamin E or after natural improvement was strongly associated with improved histology. The decrease in sphingomyelinase activity was associated with responsiveness to treatment. The pre-treatment efficacy predictors were all derived from intestinal microecological metabolites (indolepropionate and phenylpropionate) suggesting that vitamin E may act in part by altering the specific bacterial community that utilizes the Stikolam response. This study suggests that it is feasible to use metabolites to determine as well as predict the efficacy of vitamin E in the treatment of patients with NAFLD.