目的筛选新的大肠癌候选肿瘤标志物,建立分期诊断模型,并探讨其临床意义。方法采用表面加强激光解吸电离-飞行时间质谱(SELDI-TOF-MS)技术和CM10蛋白质芯片,检测76例大肠癌患者术前血清蛋白质指纹图谱;运用支持向量机分析判别处理数据、筛选标志物,建立并验证分期模型;应用时间序列分析的方法,将DukesA、B、C、D各期以二维散点图的形式表示。结果诊断模型Ⅰ由6个蛋白质峰组合构建,其质荷比分别为2759．6、2964．7、2048．0、4795．9、4139．8和37 761．6,鉴别局限性大肠癌(Dukes A、B期)和区域性(Dukes C期)大肠癌的总准确率为86．7％。诊断模型Ⅱ由3个蛋白质峰组合构建,其质荷比分别为6885．3、2058．3和8567．8,鉴别局限区域性(Dukes A、B、C期)和系统性(Dukes D期)大肠癌的总准确率为75．0％。诊断模型Ⅲ鉴别Dukes A期和B期大肠癌的总准确率为86．2％;诊断模型Ⅳ鉴别Dukes A期和C期大肠癌的总准确率为84．6％;诊断模型V鉴别Dukes B期和C期大肠癌的总准确率为85．7％;诊断模型Ⅵ鉴别Dukes B期和D期大肠癌的总准确率为80．0％;诊断模型Ⅶ鉴别Dukes C期和D期大肠癌的总准确率为78．7％。通过二维散点图,可以明显看出Dukes A、B、C、D各期之间的区别。结论通过SELDI-TOF- MS技术和CM10蛋白质芯片所筛选的候选肿瘤标志物可以指导大肠癌的综合治疗,所建立的诊断模型可以辅助临床明确大肠癌的术前分期。 Objective To screen new tumor markers of colorectal cancer and establish a staging diagnostic model and to explore its clinical significance. Methods Serum protein fingerprints of 76 patients with colorectal cancer were detected by surface enhanced laser desorption / ionization-time of flight mass spectrometry (SELDI-TOF-MS) and CM10 protein chip. The support vector machine was used to analyze the discriminant data, Establish and verify the staging model; apply the method of time series analysis to represent DukesA, B, C and D in the form of two-dimensional scatter plot. Results The diagnostic model Ⅰ was constructed by the combination of six protein peaks with mass-to-charge ratios of 2759.6, 2964.7, 2048.0, 4795.9, 4193.8 and 37761.6, respectively. The differential diagnosis of localized colorectal cancer (Dukes A, B) and regional (Dukes C) colorectal cancer with a total accuracy of 86.7%. The diagnostic model II was constructed by the combination of three protein peaks with mass-to-charge ratios of 6885.3, 2058.3 and 8567.8, respectively, to identify localized regions (Dukes A, B, C) and systemic (Dukes D) The overall accuracy of colorectal cancer was 75.0%. The diagnostic accuracy of diagnostic model Ⅲ in differentiating colorectal cancer from Dukes A and B was 86.2%. The diagnostic accuracy of diagnostic model Ⅳ in distinguishing colorectal cancer from Dukes A and C was 84.6%. The diagnostic model V identified Dukes B The overall accuracy rate of colorectal cancer in stage and C was 85.7%. The diagnostic accuracy of diagnostic model Ⅵ in distinguishing colorectal cancer from Dukes B and D was 80.0%. Diagnostic model Ⅶ identified Dukes C stage and D stage colorectal cancer The total accuracy of 78.7%. The difference between Dukes A, B, C, and D stages is evident through the two-dimensional scatter plot. Conclusions Candidate tumor markers screened by SELDI-TOF-MS and CM10 protein chip can guide the comprehensive treatment of colorectal cancer. The established diagnostic model can help clinical clear preoperative staging of colorectal cancer.