| 摘要: |
| 目的 探讨乳腺导管原位癌（ductal carcinoma in situ，DCIS）及其伴微浸润（ductal carcinoma in situ with microinvasion，DCIS-Mi）的超声特征，基于临床及超声特征构建列线图模型，评估其在术前鉴别DCIS与DCIS-Mi中的临床应用价值。方法 回顾性分析浙江省肿瘤医院及衢州市人民医院的316例DCIS（203例）与DCIS-Mi（113例）患者资料，按7:3比例随机分为训练集（n = 221）和验证集（n = 95）。采用单因素及多因素logistic回归分析筛选DCIS-Mi的独立预测因子，构建临床预测模型并绘制列线图。通过受试者工作特征曲线（receiver operating characteristic curve，ROC）、校准曲线及决策曲线（decision curve analysis，DCA）评估模型的诊断效能与临床适用性。结果 多因素分析显示，中级别核分级（OR = 4.78，P = 0.043）、高级别核分级（OR = 6.51，P = 0.011）、边缘模糊（OR = 5.43，P = 0.027）、成角/毛刺（OR = 6.00，P = 0.012）、血流信号（OR = 2.54，P = 0.011）及周边回声增强（OR = 4.53，P = 0.043）是DCIS-Mi的独立预测因子。基于上述特征构建的列线图在训练集中的曲线下面积（area under the curve，AUC）、准确度、灵敏度、特异度分别为0.82、0.75、0.74、0.77，验证集中分别为0.77、0.73、0.78、0.61，校准曲线显示模型预测值与实际值一致性良好（P均>0.05）。决策曲线分析表明，模型在较宽阈值范围内（训练集0.05-0.83;验证集0.10-0.82）均具有临床净收益。结论 中/高级别核分级、边缘模糊或成角/毛刺、血流信号及周边回声增强对DCIS-Mi具有鉴别价值，基于列线图的预测模型可为术前评估提供影像学依据，优化临床决策。 |
| 关键词: 乳腺 导管原位癌 微浸润 超声检查 列线图 |
| DOI： |
| 分类号: |
| 基金项目: |
|
| The Application of a Clinical and Ultrasound Features Nomogram Model in Predicting Microinvasion in Ductal Carcinoma In Situ of the Breast |
|
Liu Dan1,2,3,4, Ou Di5,6, Zhou Jinhong2,3,4, Zhou Lingyan5, Yan Yuqi7,5,6, Xu Dong5,6
|
|
1.Graduate School of Zhejiang Chinese Medical University;2.Department of Ultrasound Medicine,The Quzhou Affiliated Hospital of Wenzhou Medical University, Quzhou People'3.'4.s Hospital;5.Department of Ultrasound Medicine, Zhejiang Cancer Hospital;6.Wenling Institute of Big Data and Artificial Intelligence Institute in Medicine;7.Postgraduate Training Base, Wenzhou Medical University
|
| Abstract: |
| Objective To investigate the sonographic characteristics of ductal carcinoma in situ (DCIS) and DCIS with microinvasion (DCIS-Mi), and to develop a clinical-ultrasound nomogram for preoperative differentiation between these entities while assessing its clinical diagnostic value. Methods A retrospective analysis was conducted on 316 patients (203 DCIS and 113 DCIS-Mi) from Zhejiang Cancer Hospital and Quzhou People’s Hospital. Patients were randomly divided into a training set (n = 221) and a validation set (n = 95) in a 7:3 ratio. Independent predictors of DCIS-Mi were identified using univariate and multivariate logistic regression, and a clinical prediction model was constructed with a nomogram. The model’s diagnostic performance and clinical utility were evaluated using receiver operating characteristic (ROC) curves, calibration curves, and decision curve analysis (DCA). Results Multivariate analysis revealed that intermediate nuclear grade (OR = 4.78, P = 0.043), high nuclear grade (OR = 6.51, P = 0.011), indistinct margins (OR = 5.43, P = 0.027), angular/spiculated margins (OR = 6.00, P = 0.012), presence of color doppler flow signals (OR = 2.54, P = 0.011), and peripheral echo enhancement (OR = 4.53, P = 0.043) were independent predictors of DCIS-Mi. The nomogram demonstrated an area under the curve (AUC), accuracy, sensitivity, and specificity of 0.82, 0.75, 0.74, and 0.77 in the training set, and 0.77, 0.73, 0.78, and 0.61 in the validation set, respectively. Calibration curves indicated good agreement between predicted and observed probabilities (all P > 0.05). DCA confirmed clinical net benefits across a wide threshold probability range (training set: 0.05–0.83; validation set: 0.10–0.82). Conclusion Intermediate/high nuclear grade, indistinct or angular/spiculated margins, color doppler flow signals, and peripheral echo enhancement have significant discriminatory value for DCIS-Mi. The nomogram-based model provides a noninvasive preoperative assessment tool to optimize clinical decision-making. |
| Key words: Breast Ductal carcinoma in situ Microinvasion Ultrasonography Nomogram |