基于蛋白芯片的慢性肾衰舌苔上清液中蛋白研究

Study on Proteinum in Clear Supernatant Liquid of Tongue Coating of Chronic Renal Failure Based on Protein Chip

目的:通过比较慢性肾衰(CRF)患者与正常对照组人群舌苔上清液中蛋白表达谱的差异,筛选慢性肾衰舌苔蛋白标志物并建立预测模型,探讨其在慢性肾衰诊断中的意义。方法:对67例慢性肾衰患者和38例正常对照组人群舌苔上清液样本,运用SELDI-TOF-MS蛋白芯片技术筛选慢性肾衰舌苔蛋白标志物,经生物信息学分析建立预测模型并进行验证。结果:①慢性肾衰组67例和正常对照组38例舌苔样本经SELDI-TOF-MS技术测定,质荷比1000～20000范围内共检测到242个蛋白峰,经生物信息学统计分析,有13个差异质谱峰有统计学意义(P<0.01),其中m/z 1092.68、m/z 1508.26等7个差异质谱峰在慢性肾衰组中呈高表达;m/z 13302.5、m/z 14330.7等6个差异质谱峰在慢性肾衰组中呈低表达。②利用层次聚类算法进行层次聚类分析和主成分分析(PCA),结果显示,慢性肾衰组与正常对照组样本之间区分较明显,但均存在部分重叠。③经生物信息学分析建立慢性肾衰预测模型,最终得到m/z 1049.61、m/z 1076.94、m/z 15295.7等7个差异质谱峰组成的生物标记物可以将慢性肾衰组和正常对照组样品较好的分类(最终预测模型的灵敏度为61.4%,特异度为57.3%,预测正确率为64.4%)。结论:该研究运用SELDI-TOF-MS蛋白芯片技术,初步筛选出了慢性肾衰舌苔蛋白标志物并建立了预测模型,为慢性肾衰的诊断研究提供客观依据。

This study aimed to screen proteinum markers of tongue coating related to chronic renal failure （CRF） and establish the predictive model by comparing differences of protein spectrum expression in clear superuatant liquid of tongue coating between CRF patients and normal controls in order to explore its significance in the diagnosis of CRF. Clear supernatant liquid of tongue coating samples of 67 CRF patients and 38 normal controls were used in the study. Proteinum markers of tongue coating were selected according to CRF with technique of SELDI-TOF-MS. The predictive model was established and verified by bioinformatics analysis. Results showed that tongue coating samples of 67 CRF patients and 38 normal samples in the control group were determined by technique of SELDI- TOF-MS. All 242 proteinum peaks have been detected at 1000-20000 e/m. And 13 distinct mass spectrum peaks have been analyzed by bioinformatics with statistical significance （P〈 0.01）. Seven distinct mass spectrum peaks, such as m/z 1092.68 and m/z 1508.26, show high expression in CRF group. Six distinct mass spectrum peaks, such as m/z 13302.5 and m/z 14330.7, show low expression in CRF group. Fuzzy grouping algorithm was used in the fuzzy grouping analysis and principal component analysis （PCA） between CRF group and normal control group. The result showed discrimination, but partly overlapping. The predictive model of CRF is analyzed and established by bioinformatics with biological markers which are constituted with 7 distinct mass spectrum peaks, such as m/z 1049.61, m/z 1076.94, m/z 15295.7, and etc. The predictive model can be used in the sample classification between CRF group and normal control group. （The sensitivity of predictive model is 61.4%. The specificity is 57.3%. And predictive exactitude rate is 64.4%.） It is concluded that using technique of SELDI-TOF-MS, the proteinum markers of tongue coating of CRF have been preliminarily screened. The established predictive model provides objective evidence for the study on CRF diagnosis.