JMIR Publications

ABSTRACT

Background:

Breast cancer is one of the most common malignancies among women globally. Patients do not achieve a pathological complete response (pCR) post neoadjuvant chemotherapy (NAC) are associated with poorer survival outcomes.

Objective:

This study aimed to develop and validate a Random Survival Forest (RSF) model to predict survival risk in breast cancer patients who do not achieve pCR after NAC.

Methods:

We reviewed patients diagnosed with breast cancer at the First Affiliated Hospital of Chongqing Medical University from January 2019 to January 2023, focusing on those who underwent NAC. After treatment, patients who were evaluated for non-pCR were included in the study. We developed an RSF model and a Cox regression model. The models were validated using external datasets from Duke University and the SEER database.

Results:

A total of 306 patients were included in the internal dataset, with a median disease-free survival (DFS) of 25.9 months. The RSF model demonstrated superior predictive performance compared to the Cox regression model, with AUC values of 0.811, 0.834, and 0.810 at 1-year, 3-year, and 5-year, respectively. The C-index was 0.803 (95%CI, 0.747-0.859). Validation in the Duke dataset yielded AUC values of 0.912 , 0.803, and 0.776 at 1-year, 3-year, and 5-year, respectively. While in the SEER dataset showed AUC values of 0.823, 0.756, and 0.731 at 1-year, 3-year, and 5-year, respectively. The RSF model effectively stratified patients into high-risk and low-risk groups, with significant differences in survival outcomes.

Conclusions:

The RSF model, based solely on clinicopathological variables, provides a promising tool for identifying high-risk breast cancer patients post-NAC. This approach may facilitate personalized treatment strategies and improve patient management in clinical practice.