AIM: Cognitive dysfunction (CD) is frequently observed in cirrhotic patients. However, the biochemical profiles associated with CD remain unclear. We investigated the biochemical profiles associated with the incidence of CD in cirrhotic patients by using multivariate analyses, including a decision-tree algorithm. METHODS: In this study, 27 viral cirrhotic patients were enrolled. All subjects underwent neuropsychiatric tests; two or more abnormal results were defined as CD. A logistic regression model was used for multivariate stepwise analysis. A decision-tree algorithm was constructed, and the categorical differences based on the decision-tree model were analyzed by χ(2) -tests. RESULTS: Multivariate stepwise analysis showed the levels of total bilirubin, triglycerides and free fatty acids (FFA) as independent bioparameters associated with the incidence of CD in cirrhotic patients. The decision-tree algorithm showed that among patients with FFA of 514 mEq/L or more, 77.8% had CD. Meanwhile, among patients with FFA of less than 514 mEq/L and triglycerides of 106 mg/dL or more, 20.0% had CD. The sensitivity, specificity and accuracy for the incidence of CD using the lipid profile (FFA >514 mEq/L or triglycerides <106 mg/dL) were 85.7% (12/14), 61.5% (8/13) and 74.1% (20/27), respectively. CONCLUSION: The levels of total bilirubin, FFA and triglycerides are independently associated with the incidence of CD in cirrhotic patients. In addition, a decision-tree algorithm revealed that FFA of more than 514 mEq/L or triglycerides of less than 106 mg/dL is a profile associated with the incidence of CD. Thus, this lipid profile could be a possible screening bioparameter for CD in cirrhotic patients.
AIM: Cognitive dysfunction (CD) is frequently observed in cirrhotic patients. However, the biochemical profiles associated with CD remain unclear. We investigated the biochemical profiles associated with the incidence of CD in cirrhotic patients by using multivariate analyses, including a decision-tree algorithm. METHODS: In this study, 27 viral cirrhoticpatients were enrolled. All subjects underwent neuropsychiatric tests; two or more abnormal results were defined as CD. A logistic regression model was used for multivariate stepwise analysis. A decision-tree algorithm was constructed, and the categorical differences based on the decision-tree model were analyzed by χ(2) -tests. RESULTS: Multivariate stepwise analysis showed the levels of total bilirubin, triglycerides and free fatty acids (FFA) as independent bioparameters associated with the incidence of CD in cirrhotic patients. The decision-tree algorithm showed that among patients with FFA of 514 mEq/L or more, 77.8% had CD. Meanwhile, among patients with FFA of less than 514 mEq/L and triglycerides of 106 mg/dL or more, 20.0% had CD. The sensitivity, specificity and accuracy for the incidence of CD using the lipid profile (FFA >514 mEq/L or triglycerides <106 mg/dL) were 85.7% (12/14), 61.5% (8/13) and 74.1% (20/27), respectively. CONCLUSION: The levels of total bilirubin, FFA and triglycerides are independently associated with the incidence of CD in cirrhotic patients. In addition, a decision-tree algorithm revealed that FFA of more than 514 mEq/L or triglycerides of less than 106 mg/dL is a profile associated with the incidence of CD. Thus, this lipid profile could be a possible screening bioparameter for CD in cirrhotic patients.