Prognostic factors for hormone receptor-positive breast cancer with liver metastasis and establishment of novel nomograms for prediction: a SEER-based study

Abstract

Background: The prognosis of patients with hormone receptor (HR)-positive breast cancer with liver metastasis (BCLM) remains dismal and varies widely from person to person. Thus, we sought to construct nomograms to predict overall survival (OS) and breast cancer-specific survival (BCSS) in patients with HR-positive BCLM using data from the Surveillance, Epidemiology and End Results (SEER) database.

Methods: The data of patients with BCLM, who had received HR-positive diagnoses between 2010 and 2016, were collected from the SEER database. A Cox proportional hazards model was used to evaluate and identify the independent risk factors for OS and BCSS. Subsequently, two new nomograms were developed. Finally, the receiver operating characteristic (ROC) curves, calibration curves, and decision curve analysis (DCA) results were evaluated.

Results: The data of 1,780 patients diagnosed between 2010 and 2015 were used to build the nomogram models. Using both univariate and multivariate Cox regression analyses, nine variables, including age, marital status, grade, human epidermal growth factor receptor 2 (HER2) status, chemotherapy, surgery, bone metastasis, lung metastasis, and brain metastasis, were found to be significantly associated with OS. Conversely, 10 variables, including age, marital status, T stage, grade, HER2 status, chemotherapy, surgery, bone metastasis, lung metastasis, and brain metastasis, were identified as independent risk factors for BCSS. Using the risk factors listed above, we created 1-, 2-, and 3-year survival nomograms for OS and BCSS, respectively. Subsequently, the data of 312 patients, who had been diagnosed in 2016, were used for the external validation. These results, including the ROC curve, calibration curve, and DCA results, showed that our nomogram had strong predictive power.

Conclusions: Nomograms can effectively and reliably predict a patient's prognosis and could be useful in clinical decision making. The nomograms had strong discrimination, calibration, and clinical values. More aggressive treatment and closer monitoring should be considered when treating high-risk individuals.

Keywords: Breast cancer (BC); hormone receptor-positive (HR-positive); liver metastasis; nomogram; prognosis factors.

Figure 1

Flowchart of the study. SEER, Surveillance, Epidemiology and End Results; HR, hormone receptor; TNM, tumor-node-metastasis.

Figure 2

The nomogram to predict 1-, 2-, and 3-year OS of HR-positive BC patients with liver metastasis. **, P<0.01; ***, P<0.001. HER2, human epidermal growth factor receptor 2; OS, overall survival; HR, hormone receptor; BC, breast cancer.

Figure 3

The nomogram to predict 1-, 2-, and 3-year BCSS of HR-positive BC patients with liver metastasis. *, P<0.05; **, P<0.01; ***, P<0.001. HER2, human epidermal growth factor receptor 2; BCSS, breast cancer-specific survival; HR, hormone receptor; BC, breast cancer.

Figure 4

The ROC curves for the OS model for the training (A), internal validation (B), and external (C) cohorts. ROC, receiver operating characteristic; AUC, area under the curve; OS, overall survival.

Figure 5

The ROC curves for the BCSS model for the training (A), internal validation (B) and external (C) cohorts. ROC, receiver operating characteristic; AUC, area under the curve; BCSS, breast cancer-specific survival.

Figure 6

Calibration curves for predicting patients’ OS at 1-, 2-, and 3-year for the training (A-C), validation (D-F), and external (G-I) cohorts. OS, overall survival.

Figure 7

Calibration curves for predicting patients’ BCSS at 1-, 2-, and 3-year for the training (A-C), validation (D-F), and external (G-I) cohorts. BCSS, breast cancer-specific survival.

Figure 8

The DCA of the nomogram for predicting 1-, 2-, and 3-year OS for the training (A-C), validation (D-F), and external (G-I) cohorts. DCA, decision curve analysis; OS, overall survival.

Figure 9

The DCA of the nomogram for predicting 1-, 2-, and 3-year BCSS for the training (A-C), validation (D-F), and external (G-I) cohorts. DCA, decision curve analysis; BCSS, breast cancer-specific survival.

Figure 10

The K-M survival curves of the risk group stratification for OS in the training (A) and validation (B) cohorts and for BCSS in the cohort (C) and validation (D) cohort. K-M, Kaplan-Meier; OS, overall survival; BCSS, breast cancer-specific survival.

Figure 11

The K-M survival curves of the risk group stratification for OS (A) and BCSS (B) in the external cohort. K-M, Kaplan-Meier; OS, overall survival; BCSS, breast cancer-specific survival.