Prediction model for distant metastasis of breast cancer based on magnetic resonance imaging
10.3969/j.issn.1000-8179.2019.07.130
- VernacularTitle:基于磁共振成像乳腺癌远处转移预测模型的研究*
- Author:
Jia TANG
1
;
Wenjuan MA
;
Junjun LIU
;
Zhenzhen SHAO
;
Peifang LIU
Author Information
1. 天津医科大学肿瘤医院乳腺影像诊断科
- Keywords:
breast cancer;
neoplasm metastasis,magnetic resonance imaging (MRI),prediction,molecular subtype
- From:
Chinese Journal of Clinical Oncology
2019;46(7):337-341
- CountryChina
- Language:Chinese
-
Abstract:
Objective: To establish a prediction model for the distant metastasis of breast cancer based on qualitative magnetic reso-nance imaging (MRI) parameters. Methods: A retrospective analysis of 3,032 patients with breast MRI from January 2011 to Decem-ber 2016 in Tianjin Medical University Cancer Institute and Hospital was conducted. After the confirmation of invasive breast cancer, the subjects were divided in 2 groups: metastasis and metastasis-free. A total of 93 patients were included in the metastasis group, and 186 patients without the presence of distant metastasis in the metastasis-free group. We analyzed the correlation between breast cancer molecular subtypes and distant metastasis in the metastasis group. Univariate and Logistic regression analyses of qualitative MRI features were performed for the groups. Subsequently, we used the results to establish prediction models. Results: The results showed that hormone receptor-positive tumors (Luminal type) had a greater tendency to develop bone metastasis in the metastasis group. Triple-negative tumors showed a greater tendency to develop lung metastasis. Human epidermal growth factor receptor 2 gene overexpression cases were more likely to develop liver metastasis. The results of the univariate analysis showed that the type of le-sion, multifocality or multicentricity of the cancer, T1-weighted signal uniformity, T2-weighted signal uniformity, and tumor size were statistically different between the groups (P<0.05). The results of the logistic regression analysis showed that the type of lesion, multi-focality or multicentricity of the cancer, T2-weighted signal uniformity, and tumor size were independent predictors of distant metasta-sis. Based on select independent predictors, we established a prediction model for the distant visceral metastasis of breast cancer. The accuracy, area under the curve, sensitivity, and specificity of the model were 82.8%, 0.801, 85.7%, and 75.0%, respectively. Conclu-sions: The prediction model based on the clinical pathology and MRI features established in this study can predict the distant metasta-sis of breast cancer.
