Establishing a Predictive Model to Predict Survival of de Novo Metastatic HER2-Low Breast Cancer: A National Cancer Database Analysis.

Background: Breast cancer with low human epidermal growth factor receptor (HER2) expression is increasingly considered as a distinct subtype which consists of types of HER2 immunohistochemistry (IHC) 1+ and HER2 IHC 2+/in-situ hybridization (ISH)-negative. We aim to assess the survival difference between HER2 IHC 1+ and HER2 IHC 2+/ISH-negative breast cancer patients with metastasis at presentation and construct a prognostic nomogram for HER2-low patients.

Method: Patients diagnosed with de novo metastatic HER2-low breast cancer from 2010 to 2015 were included and analyzed using the National Cancer Database (NCDB). Cox proportional hazards regression model and Kaplan-Meier (KM) method were used for survival analysis. Nomograms were built to predict survival.

Result: A total of 7897 patients were included in the final analysis, among which 5458 (69.1%) patients were HER2 IHC 1+ and 2439 (30.9%) were HER2 IHC 2+/ISH-negative. Although the Kaplan-Meier survival analysis showed difference in survival, this survival difference was lost in the multivariate Cox analysis (multivariate: HR (hazard ratio) = 0.97; 95% CI (confidence interval) [0.92-1.03]). A prognostic nomogram was successfully constructed for individually predicting the long-term survival rate of HER2-low patients, which exhibited an acceptable predictive capability in training (C index: 0.719) and validation cohort (C index: 0.706). This nomogram could easily divide patients into high and low-risk subgroups with distinct prognoses.

Conclusions: Our data suggest no statistical survival differences between HER2 1+ and HER2 2+ breast cancer. Additionally, a nomogram was constructed with an acceptable capacity to individually predict the long-term outcome of HER2-low metastatic breast cancer patients.