Xia Shen1, Can Wang2, Ningning Liang3, Zhen Liu2, Xinde Li2, Zheng-Jiang Zhu4, Tony R Merriman5, Nicola Dalbeth6, Robert Terkeltaub7, Changgui Li2, Huiyong Yin1. 1. ShanghaiTech University, Chinese Academy of Sciences Key Laboratory of Nutrition, Metabolism, and Food Safety, Shanghai Institute of Nutrition and Health, Chinese Academy of Sciences, Shanghai, China, and University of Chinese Academy of Sciences, Beijing, China. 2. Shandong Provincial Key Laboratory of Metabolic Diseases, Qingdao Key Laboratory of Gout, Affiliated Hospital of Qingdao University Medical College, and Institute of Metabolic Diseases, Qingdao University, Qingdao, China. 3. Chinese Academy of Sciences Key Laboratory of Nutrition, Metabolism, and Food Safety, Shanghai Institute of Nutrition and Health, Chinese Academy of Sciences, Shanghai, China, and University of Chinese Academy of Sciences, Beijing, China. 4. Interdisciplinary Research Center on Biology and Chemistry, Shanghai Institute of Organic Chemistry, Chinese Academy of Sciences, Shanghai, China. 5. Institute of Metabolic Diseases, Qingdao University, Qingdao, China, University of Alabama at Birmingham, and University of Otago, Dunedin, New Zealand. 6. University of Auckland, Auckland, New Zealand. 7. VA San Diego Healthcare System and University of California, San Diego.
Abstract
OBJECTIVE: To systematically profile metabolic alterations and dysregulated metabolic pathways in hyperuricemia and gout, and to identify potential metabolite biomarkers to discriminate gout from asymptomatic hyperuricemia. METHODS: Serum samples from 330 participants, including 109 with gout, 102 with asymptomatic hyperuricemia, and 119 normouricemic controls, were analyzed by high-resolution mass spectrometry-based metabolomics. Multivariate principal components analysis and orthogonal partial least squares discriminant analysis were performed to explore differential metabolites and pathways. A multivariate methods with Unbiased Variable selection in R (MUVR) algorithm was performed to identify potential biomarkers and build multivariate diagnostic models using 3 machine learning algorithms: random forest, support vector machine, and logistic regression. RESULTS: Univariate analysis demonstrated that there was a greater difference between the metabolic profiles of patients with gout and normouricemic controls than between the metabolic profiles of individuals with hyperuricemia and normouricemic controls, while gout and hyperuricemia showed clear metabolomic differences. Pathway enrichment analysis found diverse significantly dysregulated pathways in individuals with hyperuricemia and patients with gout compared to normouricemic controls, among which arginine metabolism appeared to play a critical role. The multivariate diagnostic model using MUVR found 13 metabolites as potential biomarkers to differentiate hyperuricemia and gout from normouricemia. Two-thirds of the samples were randomly selected as a training set, and the remainder were used as a validation set. Receiver operating characteristic analysis of 7 metabolites yielded an area under the curve of 0.83-0.87 in the training set and 0.78-0.84 in the validation set for distinguishing gout from asymptomatic hyperuricemia by 3 machine learning algorithms. CONCLUSION: Gout and hyperuricemia have distinct serum metabolomic signatures. This diagnostic model has the potential to improve current gout care through early detection or prediction of progression to gout from hyperuricemia.
OBJECTIVE: To systematically profile metabolic alterations and dysregulated metabolic pathways in hyperuricemia and gout, and to identify potential metabolite biomarkers to discriminate gout from asymptomatic hyperuricemia. METHODS: Serum samples from 330 participants, including 109 with gout, 102 with asymptomatic hyperuricemia, and 119 normouricemic controls, were analyzed by high-resolution mass spectrometry-based metabolomics. Multivariate principal components analysis and orthogonal partial least squares discriminant analysis were performed to explore differential metabolites and pathways. A multivariate methods with Unbiased Variable selection in R (MUVR) algorithm was performed to identify potential biomarkers and build multivariate diagnostic models using 3 machine learning algorithms: random forest, support vector machine, and logistic regression. RESULTS: Univariate analysis demonstrated that there was a greater difference between the metabolic profiles of patients with gout and normouricemic controls than between the metabolic profiles of individuals with hyperuricemia and normouricemic controls, while gout and hyperuricemia showed clear metabolomic differences. Pathway enrichment analysis found diverse significantly dysregulated pathways in individuals with hyperuricemia and patients with gout compared to normouricemic controls, among which arginine metabolism appeared to play a critical role. The multivariate diagnostic model using MUVR found 13 metabolites as potential biomarkers to differentiate hyperuricemia and gout from normouricemia. Two-thirds of the samples were randomly selected as a training set, and the remainder were used as a validation set. Receiver operating characteristic analysis of 7 metabolites yielded an area under the curve of 0.83-0.87 in the training set and 0.78-0.84 in the validation set for distinguishing gout from asymptomatic hyperuricemia by 3 machine learning algorithms. CONCLUSION: Gout and hyperuricemia have distinct serum metabolomic signatures. This diagnostic model has the potential to improve current gout care through early detection or prediction of progression to gout from hyperuricemia.