目的

探索乳腺癌与乳腺良性疾病和健康人血清蛋白质谱表达差异,寻找具有鉴别诊断意义的血清蛋白标志物。

方法

实验分为两大组:(1)决策树模型组共293 例标本, 包括3 个亚组,分别为乳腺癌组110 例标本、乳腺良性疾病组113 例和健康组70 例,建立决策树(乳腺癌诊断)模型;(2)盲法验证组共34 例标本, 包括3 个亚组分别为乳腺癌组7 例标本、乳腺良性疾病组13 例及健康组14 例,进行盲筛验证决策树模型。 采用弱阳离子磁珠捕获乳腺癌患者血清中的蛋白,使用基质辅助激光解析电离飞行时间质谱(MALDI-TOF-MS)仪检测绘制蛋白峰。 应用Biomarker Wizard TM 3.1 软件和Biomarker Patterns TM 5.0 软件分析数据。 统计分析采用方差分析法和秩和检验法。计算决策树模型诊断的准确率以及盲法验证模型诊断乳腺癌的敏感性和特异性。

结果

在决策树模型组中检测到了47 个差异有统计学意义的蛋白峰(P＜0.050)。 应用BPS 5.0 软件,以相对损失最小的原则从这47 个蛋白峰中选取了4 个蛋白峰,分别为相对分子质量(M_r,本文中相当于质荷比m/z) 9292.5、M_r 11707.2、M_r 15504.5 和M_r 16107.9,用其建立决策树模型(乳腺癌诊断模型)。 该模型判断乳腺癌、乳腺良性疾病及健康人的准确率分别为99.09%、95.58%、92.86%。 盲法验证该模型诊断乳腺癌的敏感性为71.43%,特异性为88.89%。

结论

应用MALDI-TOF-MS 联合磁珠技术可以检测乳腺癌血清中差异蛋白峰并可以建立决策树(乳腺癌诊断)模型。 选择的4 个差异蛋白蜂建立的决策树模型诊断乳腺癌具有好的准确性和较好的敏感性及特异性。 决策树模型能将乳腺癌与乳腺良性疾病及健康人相鉴别。 寻找到的M_r 9292.54、M_r 11707.2、M_r 15504.5以及M_r16107.9 的蛋白峰有望成为鉴别乳腺癌与乳腺良性疾病和健康人的有效的肿瘤血清蛋白标记物。

Objective

To explore the different expressions of proteins in serum among patients with breast cancer, benign mammary disease and healthy people and find out potential serum biomarkers differentiating breast cancer from benign mammary disease and healthy people.

Methods

This study included two experiment groups, the decision tree model group with three subgroups including 110 cases of breast cancer,113 cases of benign mammary disease and 70 healthy controls to build breast cancer diagnosis model, and the blind test group with three subgroups including 7 cases of breast cancer,13 cases of benign mammary disease and 14 healthy controls to test the sensitivity and specificity of the decision tree model. The serum proteins were captured using the weak cation magnetic beads, and differently expressed proteins were identified by matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF-MS). Biomarker Wizard TM Software 3.1 and Biomarker Patterns TM Software (BPS) 5.0 were used to analyze the data. Variance analysis or rank sum test was applied for statistical analysis. The accuracy rate of the decision tree model and the sensitivity and specificity of the model tested by blind test were calculated.

Results

A total of 47 statistically different protein peaks (P＜0.05)were tested in the decision tree model group. Based on the principle of least relative loss, four protein peaks with the relative molecular mass (M_r, equal to m/z) of 9292.54,11707.2,15504.5 and 16107.9 were selected from the 47 protein peaks, and they used to construct the decision tree model for diagnosis of breast cancer using BPS 5.0. The accuracy rate of the decision tree identifying breast cancer, benign mammary disease and healthy people was 99.09%,95.58% and 92.86%,respectively. The blind test showed that the sensitivity and specificity of the decision tree diagnosing breast cancer was 71.43% and 88.89%,respectively.

Conclusions

Using the technique of MALDI-TOF MS combined with magnetic beads, different serum protein peaks in breast cancer can be detected. The decision tree model constructed with the four potential biomarkers has good accuracy and better sensitivity and specificity of diagnosing breast cancer. The decision tree model can identify breast cancer from not only benign breast disease but also healthy person. The four protein peaks of M_r 9292.54, M_r 11707.2, M_r 15504.5 and M_r 16107.9 selected are promising serum protein biomarkers for breast cancer.