致密型腺体乳腺癌早期超声诊断假阳性特征及影响因素分析

To Analyze the Characteristics and Influencing Factors of False Positive Results in Early Ultrasound Diagnosis of Breast Cancer with Dense Glands

探讨致密型腺体乳腺癌早期超声诊断假阳性特征及影响因素。选取2021年1月至2023年1月于本院就诊经超声诊断为致密型腺体乳腺癌患者190例,依据术后病理结果将致密型腺体乳腺癌患者分为假阳性组(71例)和真阳性组(119例)。比较两组的临床特征及超声征象水平。采用多因素Logistic回归分析影响超声诊断为致密型腺体乳腺癌假阳性的危险因素及进一步逐步回归筛选,并建立预测超声诊断致密型腺体乳腺癌为假阳性的模型,收集同期37例患者作为外部验证集,用于验证模型的预测效果,使用X-tile软件根据Logistic风险得分将该模型分层,探讨该模型的临床应用价值。结果显示,假阳性组的年龄<50岁、BMI≥25 kg/m^(2)、病灶位置是外上象限、肿瘤分期T1期、绝经、高回声晕、后方回声增强、纵横比<1、微钙化占比均高于真阳性组(P<0.05)。年龄(<50岁)、病灶位置(外上象限)、肿瘤分期(T1期)、后方回声(增强)、微钙化、PI≥3.45、VI≥12.37、FI≥32.52、RI<1.02是影响超声诊断致密型腺体乳腺癌为假阳性的独立危险因素(P<0.05)。进一步逐步回归分析显示这9个临床因素与超声诊断致密型腺体乳腺癌为假阳性关联最紧密,超声诊断致密型腺体乳腺癌为假阳性的公式:P=1-1/(1+e^(1.166+0.373×年龄+0.890×病灶位置+0.949×肿瘤分期(T1期)+1.831×后方回声(增强)+0.870×微钙化+1.538×PI+1.383×VI+1.417×FI+1.731×RI)),利用回归方程计算超声诊断致密型腺体乳腺癌为假阳性的可能性。结果表明:P=0.65时,约登指数数值最大,为77.808。提示模型预测效果较好。验证模型结果提示训练集和验证集的C指数分别为0.852、0.828,两集的AUC分别为0.847、0.818,两集的校正曲线均与理想曲线拟合反应良好。低风险组患者的超声诊断为致密型腺体乳腺癌假阳性概率明显低于中、高风险组(P<0.05)。患者年龄(<50岁)、病灶位置(外上象限)、肿瘤分期(T1期)、后方回声(增强)、有微钙化、PI≥3.45、VI≥12.37、FI≥32.52及RI<1.02等因素可能增加假阳性风险,临床应引起重视。

To investigate the characteristics and influencing factors of false positive in early ultrasound diagnosis of dense glandular breast cancer.A total of 190 patients with dense glandular breast cancer diagnosed by ultrasound in our hospital from January 2021 to January 2023 were selected,which were divided into a false positive group(71 cases)and true positive group(119 cases)according to postoperative pathological results.The clinical features and ultrasound signs of the two groups were compared.Multivariate Logistic regression analysis was used to analyze the risk factors affecting false positives of dense glandular breast cancer diagnosed by ultrasound and further step-by-step regression screening,and a model was established to predict false positives of dense glandular breast cancer diagnosed by ultrasound.37 patients in the same period were collected as an external validation set to verify the prediction effect of the model.X-tile software was used to stratify the model according to the risk score of the Logistic,and the clinical application value of the model was further discussed.The results showed that age<50 years old,BMI≥25 kg/m^(2),lesion location in the upper outer quadrant,tumor stage T1,menopause,hyperechoic halo,posterior echo enhancement,aspect ratio<1,microcalcification proportion in false positive group were higher than those in true positive group(P<0.05).Age(<50 years),lesion location(upper outer quadrant),tumor stage(T1 stage),posterior echo(enhanced),microcalcification,PI≥3.45,VI≥12.37,FI≥32.52,RI<1.02 were independent risk factors for false positive ultrasound diagnosis of dense glandular breast cancer(P<0.05).Further stepwise regression analysis showed that these 9 clinical factors were most closely associated with false positive ultrasound diagnosis of dense glandular breast cancer,and the formula for false positive ultrasound diagnosis of dense glandular breast cancer was as follows:P=1-1/(1+e^(1.166+0.373×age+0.890×lesion location+0.949×tumor stage(T1 stage)+1.831×posterior echo(enhanced))+0.870×microcalcification+1.538×PI+1.383×VI+1.417×FI+1.731×RI)by regression equation.To calculate the possibility of false positivity in ultrasound diagnosis of dense glandular breast cancer.The results show that when P=0.65,the highest value of Youden index is 77.808.It is suggested that the prediction effect of the model is better.The results of the verification model indicate that the C-index of the training set and the verification set are 0.852 and 0.828,respectively,and the AUC of the two sets are 0.847 and 0.818,respectively.The calibration curves of the two sets fit well with the ideal curves.The false positive probability of ultrasound diagnosis of dense glandular breast cancer in low risk group was significantly lower than that in medium and high risk group(P<0.05).Factors such as patient age(<50 years old),lesion location(upper outer quadrant),tumor stage(T1 stage),posterior echo(enhanced),microcalcification,PI≥3.45,VI≥12.37,FI≥32.52 and RI<1.02 may increase the risk of false positives,which should be paid attention to clinically.