表观弥散系数值在原发性中枢神经系统淋巴瘤诊断与鉴别中的应用价值

Value of apparent diffusion coefficient in diagnosis and differential diagnosis of primary central nervous system lymphoma

目的探讨表观弥散系数(ADC)值对原发性中枢神经系统淋巴瘤(PCNSL)与高级别胶质瘤(HGG)鉴别诊断的价值。方法选择经临床及病理证实的PCNSL患者34例为PCNSL组,另选同期确诊的HGG患者36例为HGG组,行MRI平扫、增强及弥散加权成像检查,得到平均ADC值、最小ADC值、最大ADC值及对侧正常脑白质平均(对侧)ADC值,进行logistic回归分析,用ROC曲线分析各指标的诊断效能。结果 PCNSL组平均ADC值、最小ADC值、最大ADC值及对侧ADC值两两比较均有统计学差异(P<0.05);HGG组最小ADC值与对侧ADC值比较无统计学差异(P>0.05),其余各项两两比较均有统计学差异(P<0.05);PCNSL组平均ADC值、最小ADC值、最大ADC值均低于HGG组(P<0.05)。logistic回归分析显示,最大ADC值是独立危险因素,而平均ADC值、最小ADC值是保护因素(P<0.05,P<0.01)。ROC曲线分析显示,最小ADC值单独鉴别诊断的最佳临界值为0.727×10^(3)mm^(2)/s时,曲线下面积(AUC)为0.922;各指标联合鉴别诊断的AUC为0.978,准确率92.9%、敏感性91.2%、特异性94.4%(P=0.000)。结论肿瘤实质不同ADC值参数为无创性鉴别PCNSL与HGG提供分子影像学依据,运用logistic回归模型下三者联合诊断可有效提高诊断效能。

Objective To study the value of ADC in differential diagnosis of primary central nervous system lymphoma(PCNSL) and high grade glioma(HGG).Methods Thirty-four PCNSL patients served as a PCNSL group and 36 HGG patients served as a control group.The patients underwent MRI scanning, enhanced and diffusion-weighted imaging.The mean, minimal, maximal ADC values and the mean contralateral normal white matter ADC value in tumor parenchyma were tested with post-processing software and analyzed by logistic regression analysis.The efficacy of different ADC values in differential diagnosis of PCNSL and HGG was analyzed by ROC curve analysis.Results Significant difference was detected in mean, minimal, maximal ADC values and contralateral normal white matter ADC value between the two groups(P<0.05).Significant difference was detected in minimal contralateral ADC values and other parameters in HGG group(P<0.05).The mean, minimal, maximal ADC values were significantly lower in PCNSL group than in HGG group(P<0.05).Logistic regression analysis showed that the maximal ADC value was an independent risk factor while the mean and minimal ADC values were a protective factor for PCNSL and HGG(P<0.05,P<0.01).ROC curve analysis displayed that the AUC of minimal ADC value for the differential diagnosis of PCNSL and HGG was 0.922 when its optimal cut-off value was 0.727×10^(3)mm^(2)/s while that of mean, minimal, maximal ADC values in combination for the differential diagnosis of PCNSL and HGG was 0.978 with an accuracy of 92.9%,a sensitivity of 91.2%,a specificity of 94.4%,a positive predictiion value of 93.9% and a negative prediction value of 91.9%(P=0.000).Conclusion Different ADC values in tumor parenchyma can provide the molecular imaging evidence for noninvasive differential diagnosis of PCNSL and HGG.Logistic regression model can effectively improve the diagnostic efficiency of PCNSL and HGG when the mean, minimal, maximal ADC values are applied in combination.