Dong Yunyun, Huang Yuqing, Qiu Lanyan, Yang Yu, Feng Wei, Shi Xian-Quan
Department of Ultrasound, Beijing Friendship Hospital, Capital Medical University, Beijing, China.
Department of Ultrasound, Beijing Friendship Hospital, Capital Medical University, Beijing, China.
J Surg Res. 2025 Jan;305:275-285. doi: 10.1016/j.jss.2024.11.047. Epub 2024 Dec 27.
The objective of this study was to develop a model for predicting hematoma formation after ultrasound-guided vacuum-assisted excision (US-VAE) in patients with benign breast lesions.
This retrospective study included 302 benign breast lesions from 276 patients who had undergone US-VAE. The patients were divided into training (190 patients, 201 lesions) and validation (86 patients, 101 lesions) datasets. The risk factors for hematoma were analyzed, including the lesion depth, location, maximum diameter, and pathological results, distance from the nipple, number of lesions removed, tissue components surrounding the lesions, color Doppler blood flow image characteristics, and breast thickness. Binary logistic regression was used to construct the prediction model, and a nomogram was constructed. The performance of the prediction model was assessed by obtaining the area under the receiver operating characteristic curve (AUC) and calibration plots for both training and validation datasets.
Lesion depth ≥ 1.5 cm or < 0.7 cm, color Doppler blood flow image Adler grade 2 or 3, non-fibroadenoma pathological type, and breast thickness > 2 cm were important predictors of hematoma occurrence, with odds ratios of 2.303 (P = 0.037), 2.582 (P = 0.004), 2.133 (P = 0.017), and 2.133 (P = 0.024), respectively. The prediction model performed well in both the training (AUC = 0.701, 95% CI = 0.626-0.775) and validation (AUC = 0.740, 95% CI = 0.644-0.836) datasets.
This prediction model can be used to predict the probability of hematoma after US-VAE in patients with benign breast lesions.
本研究的目的是建立一个模型，用于预测超声引导下真空辅助切除（US-VAE）治疗良性乳腺病变患者后血肿形成的情况。
这项回顾性研究纳入了276例行US-VAE的患者的302个良性乳腺病变。患者被分为训练数据集（190例患者，201个病变）和验证数据集（86例患者，101个病变）。分析了血肿的危险因素，包括病变深度、位置、最大直径、病理结果、距乳头距离、切除病变数量、病变周围组织成分、彩色多普勒血流图像特征和乳腺厚度。采用二元逻辑回归构建预测模型，并绘制列线图。通过获得训练和验证数据集的受试者操作特征曲线（AUC）下面积和校准图来评估预测模型的性能。
病变深度≥1.5 cm或<0.7 cm、彩色多普勒血流图像Adler分级2级或3级、非纤维腺瘤病理类型以及乳腺厚度>2 cm是血肿发生的重要预测因素，比值比分别为2.303（P = 0.037）、2.582（P = 0.004）、2.133（P = 0.017）和2.133（P = 0.024）。预测模型在训练数据集（AUC = 0.701，95%CI = 0.626 - 0.775）和验证数据集（AUC = 0.740，95%CI = 0.644 - 0.836）中均表现良好。
该预测模型可用于预测良性乳腺病变患者US-VAE后发生血肿的概率。
