磁共振DWI及其ADC对乳腺导管原位癌伴微浸润腋窝淋巴结转移的诊断价值

Diagnostic value of magnetic resonance DWI and ADC in breast ductal carcinoma in situ with axillary lymph node metastasis

目的:探讨磁共振扩散加权成像(diffusion weighted imaging,DWI)及其表观扩散系数(apparent diffusion coefficient,ADC)对乳腺导管原位癌(ductal carcinoma in situ,DCIS)伴微浸润(micro-invasio,Mi)腋窝淋巴结(axillary lymph nodes,ALN)转移的诊断价值。方法:选取2018年2月至2021年3月于河南省安阳市肿瘤医院行磁共振成像(magnetic resonance imaging,MRI)常规平扫、DWI且经术后病理结果均为DCIS-Mi的女性患者120例纳入训练集,另将2018年11月至2019年8月收治的女性DCIS-Mi患者51例纳入验证集。根据术后病理诊断结果将训练集患者分别分为阳性组(ALN转移)与阴性组(无ALN转移)。训练集中,阳性组17例,阴性组103例。收集患者的一般临床资料及ADC图中ALN纹理特征,比较训练集中阳性组和阴性组纹理特征及一般临床资料;采用LASSO回归及Logistic回归分析筛选影响因素。基于筛选的影响因素构建列线图模型,并对模型进行评价。各独立因素与DCIS-Mi患者发生ALN转移风险的相关性采用Spearman相关性分析。结果:变异度、偏度、峰度、1%ADC值、10%ADC值、能量、和方差、和熵、熵是影响DCIS-Mi患者发生ALN转移的独立影响因素(P<0.05)。1%ADC值与DCIS-Mi患者发生ALN转移风险呈明显负相关(P<0.05),变异度、偏度、峰度、10%ADC值、能量、和方差、和熵、熵与DCIS-Mi患者发生ALN转移风险呈明显正相关(P<0.05)。基于独立影响因素构建的Nomogram模型具有较高的特异性和灵敏度,临床适用性较高。结论:基于ADC图纹理参数构建的列线图模型预测DCIS-Mi患者ALN转移具有较好的预测效能。

Objective To investigate the diagnostic value of DWI and ADC in axillary lymph node(ALN)metastasis of breast ductal carcinoma in situ(DCIS)with micro-invasion(Mi).Method A total of 120 female patients who underwent conventional magnetic resonance imaging(MRI)scan,DWI and postoperative pathological results of DCIS-Mi in Anyang Cancer Hospital of Henan Province from February 2018 to March 2021 were included in the training set.In addition,51 female DCIS-Mi patients admitted from November 2018 to August 2019 were included in the validation set.According to the results of postoperative pathological diagnosis,the patients in the training set were divided into the positive group(ALN metastasis)and the negative group(no ALN metastasis).In the training set,there were 17 cases in the positive group and 103 cases in the negative group.The patient's general clinical data and the ADC ALN texture feature in the figure were collected and compared between positive group and negative group.LASSO regression and Logistic regression analysis were used to screen the influencing factors.A nomogram model was constructed and evaluated based on the selected influencing factors.The correlation between independent factors and the risk of ALN metastasis in DCIS-Mi patients was analyzed using Spearman correlation analysis.Result Variation,skewness,kurtosis,1%ADC value,10%ADC value,energy,variance of sum,entropy of sum and entropy were independent factors affecting ALN metastasis in patients with DCIS-Mi(P<0.05).The 1%ADC value was negatively correlated with the risk of ALN metastasis in DCIS-Mi patients(P<0.05),while variation,skewness,kurtosis,10%ADC value,energy,variance of sum,entropy of sum and entropy were positively correlated with the risk of ALN metastasis in DCIS-Mi patients(P<0.05).The Nomogram model based on independent factors has high specificity,sensitivity and clinical applicability.Conclusion The Nomogram model based on ADC texture parameters is effective in predicting ALN metastasis in DCIS-Mi patients.