中文摘要：

采用核磁共振方法检测肝硬化、肝细胞癌（简称肝癌）患者和健康人血清中代谢物，研究3组血清代谢物组的差异。利用偏最小二乘法-判别式分析（partial least square-discriminant analysis，PLS—DA）对NMR谱数据进行模式识别分析，探讨利用基于^1H NMR代谢组学技术诊断肝癌的可行性。结果表明，与健康人相比，肝硬化、肝癌患者血清中脂质（低密度脂蛋白和极低密度脂蛋白）、胆碱、乙酰乙酸等含量减少，谷氨酰胺、丙酮酸、苯丙氨酸、酪氨酸等含量增加。PLS—DA分析结果显示肝癌患者可与健康人、肝硬化患者鉴别开来，肝癌诊断灵敏度达92．1％，假阳性率为5．7％，优于血清甲胎蛋白（alpha—fetoprotein，AFP）检测。研究结果表明，基于^1H NMR代谢组学技术结合PLS—DA的方法具有灵敏、准确、重复性好等优点，有助于肝癌早期诊断。

英文摘要：

A comparison for the metabonome in serum samples from patients of LC（liver cirrhosis）, HCC（hepatocellular carcinoma） and healthy volunteers were studied with the method of NMR（nuclear magnetic resonance spectroscopy）. To explore the potential of applying ^1H NMR-based metabonomics to diagnose HCC, PLS-DA （partial least square-discriminant analysis） was employed to perform the pattern recognition analysis of NMR spectra. In comparison with healthy volunteers, the result show that patients of LC and HCC were characterized by depleted levels of lipids （LDL, low-density lipoprotein and VLDL, very low-density lipoprotein）, choline, acetoacetate, etc. and by elevated levels of glutamine, pyruvate, phenylalanine, tyrosine, etc.. The results of PLS-DA analysis shows that HCC patients could be discriminated from healthy volunteers and LC patients with diagnositic sensitivity of 92.1% and false positive rate of 5.7% These were better than the results of serum AFP （alpha-fetoprotein） detection. The results also show that ^1H NMR-based metabonomics couple with PLS-DA analysis is useful for clinical diagnosis of HCC, and the method has the superiorities of sensitiveness, accuracy and good reproducibility.