Anthropometric and classical biologic markers of malnutrition, such as serum albumin, are limited because they are influenced by nonnutritional factors. We propose that a biologic parameter that both predicts nutritional status and is unaffected by nonnutritional factors would facilitate the diagnosis of malnutrition in the elderly. This cross-sectional study included 179 randomized elderly patients. Nutritional status was assessed by the Mini-Nutritional Assessment (MNA) instrument; other end points included anthropometric measures and biologic parameters. Subjects were divided into 3 groups based on MNA-defined nutritional status, and end point means were compared using 2-way analyses of variance adjusted by sex. Correlations between the most accurate biologic marker in predicting malnutrition and other biologic and clinical variables were assessed using Pearson correlation test. Multiple linear regressions were then performed to relate the best biomarker of malnutrition to specific parameters. Finally, leptin levels that predict malnutrition were determined using receiver operating characteristic curve cutoff values. The well-nourished group had significantly higher leptin (P = .001), weight, body mass index, mid-arm circumference, and calf circumference (all, P < .001) compared with the malnourished group and the at risk of malnutrition group. Serum leptin was the optimal biomarker of MNA-defined malnutrition and had significant positive correlations with weight (P = .003) and with all anthropometric values (all P < .001), but no significant correlation with C-reactive protein. Sex, weight, and triglyceride were the best predictors of serum leptin (all P < .001). The optimal cutoff value of serum leptin to detect malnutrition was 4.3 ng/mL in men and 25.7 ng/mL in women. Serum leptin may be a good predictor of nutritional status in elderly patients.
Copyright 2010 Elsevier Inc. All rights reserved.