基于全病变的ADC直方图分析结合RESOLVE DWI预测子宫内膜癌病理组织学特征的价值

The Value of Total Pathology-Based ADC Histogram Analysis Combined with RESOLVE DWI in Predicting the Histopathological Features of Endometrial Carcinoma

目的应用基于整个病灶的表观扩散系数(ADC)直方图分析方法结合高分辨率扩散加权成像(RESOLVE DWI)以评估其在预测子宫内膜癌的病理分化、分期特征以及肌层浸润程度的价值。方法前瞻性搜集103例术前病理证实为子宫内膜癌并接受手术治疗的患者(排除26例)。所有患者术前一周内均行盆腔MRI检查(Siemens MAGNETOM Skyra 3.0 T超导磁共振扫描仪)且均行T_(1)WI、T_(2)WI、ssDWI和RESOLVE DWI序列扫描。对所有患者的子宫内膜癌病灶进行基于整个病变的ADC直方图分析,并对各亚组间(包括分化级别,分期级别)ADC直方图参数进行对比。采用受试者工作特征曲线(ROC)分析确定ADC直方图参数在预测子宫内膜癌的病理分化特征、浸润程度的诊断效能,并计算各参数所对应的ROC曲线下面积(AUC),采用Z检验比较两两之间差异。结果高分化组的中位数ADC(ADCmedian)、平均ADC(ADCmean)、P5、P10、P25、P75、P90和P95均显著高于中分化组(P均<0.05)。在评估高分化EC患者时,ADCmedian、ADCmean、P5、P10、P25、P75、P90和P95的AUC值分别为0.745,0.755,0.735,0.736,0.737,0.752,0.752,0.759。对各ADC参数的AUC使用Z检验进行两两组间比较,差异均无统计学意义,ADC95th预测高分化子宫内膜癌的最佳临界值为1059×10^(-6) mm^(2)/s,敏感度,特异度分别为60.98%,89.47%。中分化组与低分化组各ADC参数无统计学差异,ADC参数值鉴别低分化子宫内膜癌能力弱。其他非子宫内膜样腺癌组的最小ADC(ADCmin)值显著低于子宫内膜样腺癌组,差异有统计学意义(P<0.05)。深肌层浸润组的P10、P25显著低于浅表浸润组(P<0.05,P<0.05),P10、P25的AUC相对不高(分别为0.590和0.588),相比其他参数可区分子宫内膜癌的深/浅肌层浸润。其中P10的AUC值最大,与ADC25th在区分子宫内膜癌的深/浅肌层浸润能力相比,差异无统计学意义(0.590 vs.0.588,P=0.886)。结论ADCmedian和ADC25th在鉴别高/中分化和低分化子宫内膜癌中具有最大效能。RESOLVE DWI结合ADC直方图参数可作为辅助手段鉴别深/浅肌层浸润。

Objective To evaluate the value of whole-focus-based ADC histogram analysis combined with high-resolution Resolve DWI in predicting pathological differentiation,staging characteristics,and myometrium invasion of endometrial carcinoma.Methods 103 patients with endometrial carcinoma confirmed by preoperative pathology and treated with surgery were prospectively included(26 patients were excluded).All patients underwent pelvic MRI(Siemens MAGNETOM Skyra 3.0 T superconducting magnetic resonance scanner)within one week before surgery and homogeneous scanning techniques include T_(1)-weighted imaging,T_(2)-weighted imaging,ssDWI and RESOLVE DWI sequence scanning,etc.Whole-lesion ADC Histogram analysis was performed on endometrial cancer lesions of all patients,and the ADC histogram parameters of different subgroups(including differentiation level and stage level)were compared.Receiver operating Characteristic curve(ROC)analysis was used to determine the diagnostic efficacy of ADC histogram parameters in predicting pathological differentiation characteristics and invasion degree of endometrial cancer,and the areas under ROC curve corresponding to each parameter were calculated.The Z test was used to compare the differences between the two groups.Results ADCmedian,ADCmean,P5,P10,P25,P75,P90 and P95 in the highly differentiated group were significantly higher than those in the moderately differentiated group(P<0.05).In the evaluation of patients with highly differentiated EC,the area under curve(AUC)values of ADCmedian,ADCmean,P5,P10,P25,P75,P90 and P95 were 0.745,0.755,0.735,0.736,0.737,0.752,0.752,and 0.759,respectively.The AUC of each ADC parameter was compared between the two groups by Z test,and the difference was not statistically significant.The optimal critical value of ADC95th for predicting well-differentiated endometrial cancer was 1059×10^(-6)mm^(2)/s,and the sensitivity and specificity were 60.98%and 89.47%,respectively.There was no statistical difference in ADC parameters between the moderately differentiated group and the poorly differentiated group,and ADC parameter values were weak in differentiating poorly differentiated endometrial carcinoma.The ADCmin values of other non-endometrioid adenocarcinoma groups were significantly lower than those of endometrioid adenocarcinoma group,the difference was statistically significant(P<0.05).P10 and P25 in the deep muscle invasion group were significantly lower than those in the superficial invasion group(P<0.05,P<0.05),and the AUC of P10 and P25 was relatively low(0.590 and 0.588,respectively),which could distinguish the deep/superficial muscle invasion of endometrial carcinoma compared with other parameters.The AUC value of P10 was the largest,and the difference was not statistically significant compared with ADC25th in differentiating deep/superficial muscle invasion capacity of endometrial cancer(0.590 vs.0.588,P=0.886).Conclusion Research has showed that ADCmedian and ADC25th were most effective in differentiating between highly/moderately differentiated and poorly differentiated endometrial cancers.Resolve DWI combined with ADC histogram parameters could be used as an auxiliary means to identify the deep/superficial invasion of uterine muscle layer.