摘要
目的:在免疫组化分型的基础上,探讨弥漫性大B细胞淋巴瘤患者石蜡标本基因表达与免疫组化分型之间的关系。方法:收集弥漫性大B细胞淋巴瘤患者石蜡切片,常规免疫组化检测CD10、Bcl-6和MUM1,按Hans分类标准进行亚型分类。石蜡组织中提取RNA,real-time PCR法检测Bcl-2、CCND2、LMO2、FOXP1、Bcl-6、HGAL、FN1、CCL3、MME、MUM1和REL基因的表达。利用Logistic回归建立预测模型,并用ROC曲线对其应用价值进行评价。结果:免疫组化结果显示,CD10阳性率为17.07%,Bcl-6阳性率为78.05%,MUM1阳性率为60.98%,Hans标准分型GCB型24.39%,non-GCB型75.61%。Logistic回归模型单因素分析结果表明,MME、LMO2和Bcl-2基因与DLBCL的免疫分型有关,P<0.05;多因素逐步回归分析显示,MME、LMO2和Bcl-2有显著回归效果,建立预测模型公式P=e-3.946-0.687×Bcl-2+0.199×MME+0.421×LMO2/(1+e-3.946-0.687×Bcl-2+0.199×MME+0.421×LMO2)。ROC曲线显示,该模型的曲线下面积为0.970,灵敏度为0.900,特异度为0.968。该模型判断DLBCL分型与免疫组化的总符合率为95.1%。结论:从石蜡中提取RNA,real-time PCR检测基因表达,建立Logistic回归模型,利用该模型对DLBCL分型是可行的。
OBJECTIVE： To detect the expression of mRNA and protein in diffuse large-E-cell lymphoma patients samples,and study the relationship between the expression of mRNA and subgroups base on the immunohistochemisty. METHODS： Collect the paraffin samples of diffuse large-B-cell lymphoma patients. Divide the patients into two subgroups according to the immunohischemisty of CD10,Bcl-6 and MUM1. Extract RNA from the paraffin-embedded tissues,Meas- ure the expression of Bcl-2, CCND2, LMO2, FOXP1, Bcl-6, HGAL, FN1, CCL3, MME, MUM1 and REL by real-time RCR. Construct a predictive model through Logistic regression and evaluate the application value by ROC curve. RE- SULTS：According to the results of immunohistochemisty, the positive rate of CD10 was 17.07%, the positive rate of Bcl-6 is 78.05% ,and the positive rate of MUM1 was 60.98%. The percent of GCB was 24.39%o and non-GCB was 75.61% ac- cording to the algorithm described by Hans. The results of univariate Logistic regression indicated that the expressions of MME, LMO2 and Bcl-2 were associated with DLBCL immunohistochemical classification （P〈0.05）, and the results of multivariate stepwise regression showed that the regression effect of MME,LM02 and Bcl-2 were significance. Construc- ted the predictive model formula ： P = e-3-3.946-0.587×Bcl-2＋0.199×MME＋0.421×LMO2 / （ 1 ＋ e-3.946-0.587×Bcl-2＋0.199×MME＋0.421×LMO2 ）. Ac- cording to ROC curve AUC of the model was 0. 970, sensitivity was 0. 900 and specificity was 0. 968. The model predicted DLBCL classification of 95.1% in accordance with the result of immunohistochemisty. CONCLUSION： The constructed Logistic regression model is valuable for DLBCL classification.
出处
《中华肿瘤防治杂志》
CAS
北大核心
2012年第12期908-912,共5页
Chinese Journal of Cancer Prevention and Treatment
