摘要
目的:探讨全容积合成MRI联合3D-ASL直方图术前无创预测胶质瘤MGMT启动子甲基化状态的临床价值.方法:回顾性将2020年8月至2022年8月在本院行颅脑合成MRI和3D-ASL序列扫描且术后病理证实的50例胶质瘤患者纳入研究.由2名放射科医师使用3D-Slicer软件勾画肿瘤全容积后分别提取合成MRI和3D-ASL的定量参数T1、T2、PD及CBF的直方图特征(10th位数、90th位数、均值、熵、峰度、偏度).根据MGMT启动子甲基化表达状态将胶质瘤患者分为甲基化组与非甲基化组.比较2名观察者测量各参数值的一致性差异;采用独立样本t检验或Mann-Whitney U检验比较甲基化组与未甲基化组各定量参数的直方图特征差异;采用ROC和二分类logistic回归分析各参数及其联合的诊断效能,计算ROC曲线下面积(area under the curve,AUC),选择约登指数并确定最佳截断值及其鉴别诊断的灵敏度及特异度.采用Spearman相关分析研究各参数值与MGMT启动子甲基化的相关性.以P＜0.05为差异有统计学意义.结果:2名观察者测量各参数值的一致性均较好(ICC＞0.80).MGMT 启动子甲基化组的 T1-10th 位数(613.36±118.01)、T2-10th 位数(63.70±5.88)及CBF-熵值[0.77(0.72,0.90)]均低于MGMT启动子非甲基化组[分别为843.42±87.67、68.92±5.34、1.21(1.02,1.39)],差异具有统计学意义(分别为 t=7.77、t=3.27、Z=5.37,P 均＜0.05),多因素logistic 回归分析显示 T1-10th 位数[OR=1.013,95％CI(1.002～1.025),P=0.02]、T2-10th 位数[OR=1.081,95％CI(0.905～1.292),P=0.038]、CBF-熵值[OR=1.007,95％CI(2.795～3.631),P=0.027]值是预测胶质瘤MGMT甲基化的独立影响因素,且其与MGMT启动子甲基化表达状态均呈负相关(r 分别为-0.724、-0.409、-0.767,P＜0.01).ROC 显示 T1-10th 位数、T2-10th 位数、CBF-熵值鉴别MGMT启动子甲基化组与MGMT启动子非甲基化组的曲线下面积(AUC)分别为0.918、0.736、0.913.多参数分析中三者联合的诊断效能最高(AUC=0.973),灵敏度和特异度分别为95.8％和92.3％.结论:全容积合成MRI联合3D-ASL直方图为术前无创预测胶质瘤MGMT启动子甲基化表达状态提供了 一种有效的辅助诊断方法.联合诊断可提高诊断效能,优于部分单一参数.
Abstract
Objective:To explore the clinical value of preoperative noninvasive prediction of MG-MT promoter methylation status in glioma using whole-volume synthetic MRI combined with 3D-ASL histogram analysis.Methods:50 postoperative pathological identified glioma patients were retrospective enrolled and underwent MRI scanning in our hospital from August 2020 to August 2022 and were con-firmed by.Two radiologists used 3D-slicer software to delineate the entire tumor volume and extract histogram features of the quantitative parameters T1,T2,PD,and CBF from synthetic MRI and 3D-ASL,including the 10th,90th percentile,mean,entropy,kurtosis,and skewness.All patients were divid-ed into methylated and unmethylated groups based on MGMT promoter methylation status.The con-sistency of parameter values measured by the two observers was compared,and the histogram feature differences of each quantitative parameter in the methylated and unmethylated groups were analyzed using independent sample t-tests or Mann-Whitney U tests.ROC and dichotomous logistic regression analyses were used to evaluate the diagnostic performance of each parameter and their combination,calculate the area under the ROC curve(AUC),select the Jorden index,and determine the optimal cut-off value along with sensitivity and specificity for differential diagnosis.Spearman correlation anal-ysis was performed to study the correlation between parameter values and MGMT promoter methyla-tion status.A P-value of less than 0.05 was considered statistically significant.Results:The consistency of the parameter values measured by the two observers was good(ICC＞0.80).The T1-10th percentile(613.36±118.01),T2-10th percentile(63.70±5.88),and CBF-entropy[0.77(0.72,0.90)]of the MG-MT promoter methylated group were lower than those of the non-methylated group[843.42±87.67,68.92±5.34,1.21(1.02,1.39),respectively],and the differences were statistically significant(t=7.77,t=3.27,Z=5.37,P＜0.05).Multivariate logistic regression analysis showed that T1-10th per-centile[OR=1.013,95％CI(1.002～1.025),P=0.02],T2-10th percentile[OR=1.081,95％CI(0.905～1.292),P=0.038],CBF-entropy[OR=1.007,95％CI(2.795～3.631),P=0.027]value is an independent influencing factor in predicting glioma MGMT methylationand,and it was negatively cor-related with the methylation expression status of MGMT promoter(r=-0.724,-0.409,-0.767,P＜0.01).ROC showed that the area under the curve(AUC)of T1-10th percentile,T2-10th percen-tile,and CBF-entropy in distinguishing the MGMT promoter methylated group from the non-methyla-ted group were 0.918,0.736,and 0.913,respectively.In the multi-parameter analysis,the combined di-agnosis had the highest diagnostic performance(AUC=0.973),with a sensitivity of 95.8％and a spe-cificity of 92.3％.Conclusion:The study demonstrates that certain histogram features from synthetic MRI and 3D-ASL can be used to noninvasively predict the MGMT promoter methylation status in gli-oma patients,preoperatively.T1-10th,T2-10th percentile,and CBF-entropy showed significant differ-ences between methylated and unmethylated groups,indicating their potential as predictive biomarkers.
基金项目
宁夏回族自治区卫生健康系统科研课题(2023-NWKYP-046)
宁夏回族自治区2023年自然科学基金(2023AAC03557)
NETL
NSTL
万方数据