Abstract: Background and purpose: There are various chemotherapy regimens for breast cancer, and multiple chemotherapy drugs are often used in combination. The chemotherapy cycle is long, and chemotherapy drugs cause many adverse reactions. In clinical diagnosis and treatment of tumor patients, liver injury caused by chemotherapy is also a common adverse reaction. Of the drug-induced liver injuries, liver injury caused by antitumor drugs accounts for 15%. We aimed to provide clinical strategies for chemotherapy patients to reduce adverse reactions, increase chemotherapy compliance and improve the quality of life of patients. Methods: We retrospectively analyzed the risk factors of liver injury caused by chemotherapy in patients with breast cancer who were treated in Fudan University Shanghai Cancer Center from January 2017 to December 2017. We enrolled breast cancer patients from Fudan University Shanghai Cancer Center and collected basic information including patient’s information, medical history, chemotherapy, and laboratory indicators related to liver function. According to the presence or absence of liver injury, the patients were divided into case group and control group. Univariate analysis of each factor level and multivariate analysis were used to explore the risk factors of liver injury caused by chemotherapy in patients with breast cancer. The prediction model of liver injury was established by binary logistic analysis, and the predictive ability of the model was tested by receiver operating characteristic (ROC) curve analysis. Results: A total of 724 patients with breast cancer in Fudan University Shanghai Cancer Center were eligible for inclusion in this study. The proportion of patients with liver injury during chemotherapy was 40.74%. The results of logistics analysis based on the results of univariate analysis showed that age, TNM stage, intensive chemotherapy regimen, paclitaxel combined with platinum chemotherapy regimen, anthracycline chemotherapy regimen and chemotherapy cycle became the independent risk factors for liver injury caused by chemotherapy in patients. The accuracy of the binary logistic model: P=1/1+Exp∑(0.901- AX ₁ +1.01X ₂ +TX ₃ -1.82X ₄ +5.225X ₅ +1.256X ₆ + 0.874X ₇ -0.764X ₈ ), with a specificity of 91.61% and a sensitivity of 81.69%, was 87.60%. The negative predictive value (NPV) was 87.92%, and the positive predictive value (PPV) was 87.00%. At the same time, ROC curve analysis showed that the area under the ROC curve was 0.923 (95% CI: 0.901-0.944, P<0.001). Conclusion: It is necessary to establish an effective prediction model to take certain intervention measures for patients with high-risk liver injury. The binary logistic prediction model established in this study has high accuracy, sensitivity and specificity, and can satisfy the prediction requirements of chemotherapy-induced liver injury in breast cancer patients. It lays the theoretical foundation for the prediction and clinical intervention of adverse reactions in breast cancer patients with liver injury at later stage.

Key words:  Breast cancer, Chemotherapy, Liver injury, Risk factors, Logistic regression analysis