基于MRI的纹理分析预测颅内原发性中枢神经系统淋巴瘤患者Bcl-2,C-myc表达状态的应用价值

The Value of Texture Analysis Based on MRI in Predicting the Expression of Bcl-2 and C-myc in Patients with Encephalic PCNSL

目的 探讨MRI图像纹理分析预测颅内原发性中枢神经系统淋巴瘤(PCNSL)患者Bcl-2,C-myc表达状态的应用价值。方法 回顾性分析2015年1月至2020年12月共45例颅内PCNSL患者资料,采用免疫组织化学染色法检测患者病理标本中Bcl-2和C-myc蛋白的表达,将患者分为Bcl-2、C-myc共表达组和非共表达组。对45例患者基线期头颅磁共振T_(1)WI增强图像进行纹理分析,比较两组中值、平均值、标准差、偏度、峰度、能量和熵值等定量参数的差异。结果 对MRI图像纹理分析所得的定量参数中,峰值和偏度2个参数在Bcl-2、C-myc共表达组和非共表达组间的差异无统计学意义;中值、平均值、标准差、能量和熵值5个参数在Bcl-2、C-myc共表达组和非共表达组间的差异有统计学意义(P<0.05);熵值参数的鉴别效能显著优于其他参数,曲线下面积(AUC)为0.889,敏感度为91.3%,特异度为90.9%;对中值、平均值、标准差、能量和熵值5个定量参数进行多参数联合分析,AUC为0.913,当阈值为7.472时,其诊断敏感度为95.6%、特异度为88.9%,较利用单个纹理分析定量参数鉴别效能高。结论 中值、平均值、标准差、能量和熵值5个纹理分析定量参数有助于预测颅内PCNSL相关基因Bcl-2和C-myc的表达,有利于术前无创性评估肿瘤的恶性程度并为预后和治疗提供新的依据。

Objective To investigate the value of MRI image texture analysis in predicting the expression of Bcl-2 and C-myc in patients with encephalic PCNSL.Methods 45 encephalic PCNSL patients from January 2015 to December 2020 were collected in this retrospective study.Immunohistochemical staining was used to detect the expression of Bcl-2 andC-myc protein in pathological specimens and the patients were divided into Bcl-2,C-myc co-expression group and non-co-expression group.Texture analysis was performed on the T1WI-enhanced cranial magnetic resonance images of 45 patients atbaseline,and the differences in quantitative parameters such as median,mean,standard deviation,skewness,kurtosis,energyand entropy were compared between the two groups.Results Among the quantitative parameters obtained from MRI imagetexture analysis,there was no significant difference in the kurtosis and skewness between the co-expression group and non-co-expression group,but significant difference in skewness,kurtosis,energy and entropy between Bcl-2 and C-myc proteinco-expression group and non-co-expression group existed(P< 0.05);The identification efficiency of entropy parameterswas obviously better than that of other parameters.Its AUC was 0.889,sensitivity was 91.3% and specificity was 90.0%;Multi-parameter joint analysis was performed among five quantitative parameters:median,mean,standard deviation,energy,and entropy.When the threshold was 7.472,the diagnostic sensitivity and the specificity was 95.6% and 88.9%,indicatingmulti-parameter joint analysis was more efficient.Conclusion Among the quantitative parameters,median,mean,standarddeviation,energy and entropy contribute to the prediction of PCNSL-related genes Bcl-2 and C-myc expression,which areconducive to preoperative non-invasive assessment of tumor malignancy and provided novel and valuable evidence for prog-nosis and treatment.