A preoperative prediction model for breast benign and malignant phyllodes tumors
10.3781/j.issn.1000-7431.2023.2206-0457
- VernacularTitle:乳腺良性叶状肿瘤与恶性叶状肿瘤的术前预测模型
- Author:
Jialin LIU
1
;
Xianyu ZHANG
;
Abiyasi NANDING
;
Siliang ZHANG
;
Wei MENG
;
Da PANG
Author Information
1. 哈尔滨医科大学附属肿瘤医院乳腺外科,黑龙江 哈尔滨 150000
- Keywords:
Phyllodes tumor of breast(PTB);
Growth rate of tumor;
Ultrasound BI-RADS category;
Differential diagnosis
- From:
Tumor
2023;43(2):106-113
- CountryChina
- Language:Chinese
-
Abstract:
Objective:To establish a predictive model for preoperative diagnosis of benign and malignant phyllodes tumor of the breast(PTB). Methods:The clinicopathological data of 69 patients with benign PTB and 41 patients with malignant PTB(24 borderline and 17 malignant)who underwent multiple(≥2)preoperative ultrasound follow-ups in the Cancer Hospital of Harbin Medical University from January 2011 to December 2018 were retrospectively analyzed.The preoperative prediction models of benign and malignant PTB were constructed by using the influencing factors determined by multivariate logistic regression analysis.The receiver operating characteristic(ROC)curve was used to evaluate the efficiency of the prediction model.In addition,the clinicopathological data of 22 patients of benign PTB and 19 patients of malignant PTB(12 borderline and 7 malignant)admitted to the hospital from January 2019 to April 2022 were selected for external verification. Results:Logistic regression analysis showed that growth rate of tumor>2 mm/month and ultrasound BI-RADS category≥4b were independent predictors for the diagnosis of malignant PTB(OR:4.476,95％CI:1.673～11.975;OR:9.448,95％CI:3.149～28.345;P<0.01).The logistic regression equation:Logit(P)=-1.868+1.499×growth rate of tumor+2.246×ultrasound BI-RADS category.The AUC for the training cohort was 0.795(95％CI:0.699～0.890),the best cut-off value was 0.421,the corresponding sensitivity was 0.732,the specificity was 0.826,and the Jorden index was 0.558,P<0.001.The AUC for the the validation cohort was 0.772(95％CI:0.624～0.919),with the sensitivity of 0.526 and the specificity of 0.773,positive predictive value was 0.667 and negative predictive value was 0.654,P = 0.003.The AUC of the training cohort and the validation cohort were both>0.75,indicating that the model has certain predictive ability. Conclusion:The predictive model constructed by clinicopathological parameters can be used for preoperative diagnosis of benign PTB and malignant PTB,and provide a certain reference value for clinicians to select the appropriate surgical resection scope.
