基于1H-NMR的模式识别方法在慢性心力衰竭患者血清代谢组学中的应用

摘要

目的 探讨基于1H-NMR的模式识别方法应用于慢性心力衰竭(CHF)血清代谢组学研究的可行性.方法 应用1H-NMR技术对9例CHF患者和6例健康人血清进行检测,对所得数据分别采用主成分分析法(PCA)和正交信号校正偏最小二乘法(OPLS)进行模式识别分析,确定CHF患者与健康人血清差异代谢物,比较两种模式识别方法判别能力.结果 CHF患者与健康人血清1H-NMR图谱有明显差异.其中PCA模式识别法未能将CHF组和健康对照组区分开,而用OPLS模式识别方法两组可明显区分.结论 1H-NMR技术是研究CHF代谢组学的有效技术手段,CHF患者血清代谢组与健康人存在明显差异.OPLS模式识别法明显优于PCA法,能够去除非实验因素的影响,提高分类效果.%Objective To investigate the feasibility of applying 'H-NMR-based pattern recognition in the studies of serum metabonomics in chronic heart failure (HF). Methods 'H-NMR technique was applied for examination of the serum samples from 9 patients with chronic heart failure and 6 healthy individuals. The data were analyzed for pattern recognition through principal component analysis (PCA) and Orthogonal Partial Least Square (OPLS) to determine the differences in serum metabolites between the two groups. The recognition ability of the two analysis methods were compared. Results The serum 'H-NMR spectra of heart failure patients and healthy individuals were significantly different. The PCA method failed to distinguish the patterns between the two groups, but OPLS clearly differentiated the two groups. Conclusions 'H-NMR technique is effective in the study of serum metabolomics in chronic heart failure. The serum metabonomics of patients with chronic heart failure and the healthy individuals are significantly different. OPLS pattern recognition method is superior to PCA method in that the former can remove the influence of non-experimental factors and provide an improved characterization.