Classification of Diabetes Volunteers Based on Plasma Amino Acids Metabolic Profiling

Abstract

To study the relation of plasma amino acids metabolic profiling and diabetes blood sugar level by using metabolomics, an on-line pre-column o-phthalic dicarboxaldehyde ( OPA) derivatization HPLC method for the relative quantitative determination of 21 amino acids in plasma samples was developed. The derivatization procedure was accomplished by an online auto-injector and the signals were detected with a UV detector. Mobile phase A was 10 mmol/L Na(2)HPO(4)-Na(2)B(4)O(7) buffer at pH 7.95, and B was acetonitrile-methanol-water (45:45:10 by volume). The elution program was 5% B at the start and 46% B at the end in 30 min. The 21 amino acids were satisfactorily separated in 27 min. Fifty-one plasma samples from diabetes volunteers were analyzed and were pre-divided into blood sugar ( B. S.) 9. 5 group. By SPSS and Matlab software platform, relation analysis was performed by using linear discriminant analysis, stepwise discriminant analysis, and scores plot models. By correlating the levels of blood sugar( B. S.) to 20 amino acids peak areas except glutamine, the results showed that the B. S. 9.5 group with a total hit ratio of 94.1%, stepwise discriminant analysis ( SDA) model indicated that 7 out of 20 amino acids,i.e., Arg, Cit, Leu, Asp, Asn, Thr and Trp essentially contributed to the above two groups division with the total hit ratio of 90.2%, which maybe have more means as potential marker groups for the early diagnosis of diabetes and in-depth research.