目的建立基于Kaiser评分(Kaiser score,KS)临床-多参数乳腺MRI影像诊断模型,并探讨其在乳腺良恶性病变诊断及鉴别中的价值.材料与方法回顾性分析2019年1月至2022年12月经病理证实乳腺肿瘤患者389名(共403例病灶)的术前MRI及临床病理资料,其中良性组100例及恶性组303例.记录基于KS中的MRI图像特征、表观扩散系数(apparent diffusion coefficient,ADC)值及相关临床指标,单因素分析比较乳腺良恶性病变组间各指标之间的差异,多因素logistic回归分析筛选乳腺恶性病变的独立危险因素,建立临床-多参数MRI影像诊断模型,绘制受试者工作特征(receiver operating characteristic,ROC)曲线评估其诊断效能,DeLong检验比较临床-多参数MRI影像诊断模型与单纯KS的诊断效能.结果乳腺良恶性病变组间根征、时间-信号强度曲线(time-signal intensity curve,TIC)类型、边缘、内部强化、水肿、ADC值、年龄、妇科肿瘤病史、绝经史差异有统计学意义(P<0.001),多因素logistic回归分析显示病灶根征阳性、TIC为Ⅲ型、边缘不光整、年龄大、存在妇科肿瘤史[比值比(odds ratio,OR)=7.889、7.707、4.398、1.122、0.239,P<0.05]是乳腺恶性病变的独立预测因子,基于KS相关特征、年龄、妇科肿瘤史建立临床-多参数MRI影像诊断模型.以乳腺良恶性为标准绘制KS及临床-多参数MRI影像诊断模型的ROC曲线,敏感度分别为97.4％、91.1％,特异度为69.3％、84.2％,曲线下面积(area under the curve,AUC)值为0.912、0.950;DeLong检验显示两者的AUC差异有统计学意义(P=0.006).腋窝淋巴结(axillary lymph node,ALN)转移阳性组与阴性组在乳腺癌根征(x2=6.477,P=0.011)、瘤周水肿(x2=12.241,P<0.001)、ADC值(Z=10.988,P=0.015)差异有统计学意义.多因素logistic回归分析显示瘤周水肿(OR=2.807,P=0.006)会增加ALN转移风险,存在瘤周水肿增加ALN转移的风险是无此特征患者的2.807倍.结论KS对乳腺病灶有较高的诊断价值,基于KS的临床-多参数MRI影像诊断模型有助于提高乳腺良恶性病变的诊断效能,且乳腺MRI原发灶存在瘤周水肿可作为乳腺癌ALN转移的独立预测因子.
Value of a clinical-multiparametric MRI diagnostic model based on Kaiser score in the differential diagnosis of benign and malignant breast lesions
Objective:To establish a clinical-multiparameter breast MRI diagnostic model based on Kaiser score (KS) and explore its value in the diagnosis and differentiation of benign and malignant breast lesions. Materials and Methods:Clinical and preoperative MRI data of 389 patients with 403 lesions confirmed by pathology were retrospectively analyzed between January 2019 and December 2022,collected MRI,clinical and pathological data of breast lesions,including 100 cases in benign group and 303 cases in malignant group. Based on MRI image features,apparent diffusion coefficient (ADC) value and related clinical indicators in KS,comparing the differences between the indicators of benign and malignant breast lesions by univariate analysis,multivariate logistic regression analysis established clinical-multiparameter MRI imaging diagnosis model. The receiver operating characteristic (ROC) cruve was plotted to evaluate the diagnostic performance. DeLong test was used to compare the diagnostic efficacy of clinical-multiparameter MRI imaging diagnosis model with the KS. Results:Root features,time-signal intensity curves (TIC) type,margin,internal enhancement,edema,ADC value,age,gynecological tumor history,menopausal status between benign and malignant breast lesions with a statistical difference (P<0.001). Multivariate logistic regression analysis showed positive root sign,TIC type Ⅲ,rough margins,old age,and history of gynecological tumors[odds ratio (OR)=7.889,7.707,4.398,1.122,0.239,P<0.05]was an independent predictor of malignant breast lesions. A clinical-multiparametric MRI imaging diagnostic model was established based on KS correlation characteristics,age,and gynecological tumor history. The ROC curves of KS and clinically-multi-parameter MRI diagnostic models were mapped using benign and malignant breast as criteria. Sensitivity was 97.4％ and 91.1％,specificity was 69.3％ and 84.2％,respectively. Area under the curve (AUC) values were 0.912 and 0.950. The AUC difference was statistically significant (P=0.006). There were significant differences between the positive and negative ALN metastasis groups in breast cancer root sign (x2=6.477,P=0.011),peritumoral edema (x2=12.241,P<0.001),and ADC value (Z=10.988,P=0.015). Multivariate logistic regression analysis showed that peritumoral brain edema (OR=2.807,P=0.006) increased the risk of axillary lymph node (ALN) metastasis,and the presence of peritumoral edema increased the risk of ALN metastasis 2.807 times higher than in patients without this feature. Conclusions:KS has high diagnostic value for breast lesions,the clinical-multiparametric MRI diagnostic model based on KS is subservient to improve the diagnostic efficacy of benign and malignant breast lesions,and the presence of peritumoral edema in the primary breast MRI can be used as an independent predictor of ALN metastasis in breast cancer.
breast cancermagnetic resonance imagingKaiser scorelymph node metastasisclinical factors
国家科技期刊平台
NSTL
万方数据
