Personalized treatment strategies for breast adenoid cystic carcinoma: A machine learning approach

Background: Breast adenoid cystic carcinoma (BACC) is a rare subtype of breast cancer that accounts for less than 0.1 % of all cases. This study was designed to assess the efficacy of various treatment approaches for BACC and to create the first web-based tool to facilitate personalized treatment decisions.

Methods: The Surveillance, Epidemiology, and End Results (SEER) database was used for this study's analysis. To identify the prognostic variables, we conducted Cox regression analysis and constructed prognostic models using five Machine Learning (ML) algorithms to predict the 5-year survival. A validation method incorporating the area under the curve (AUC) of the receiver operating characteristic (ROC) curve was used to validate the accuracy and reliability of ML models. We also performed a Kaplan-Meier (K-M) survival analysis.

Results: This study included 1212 patients. The median age was 60 years, with most tumors being localized and less than 2 cm in size. The 5-year overall survival (OS) rates were highest for surgery + radiotherapy (RT) (94.9 %) and lowest for surgery + chemotherapy (CTX) + RT (80.1 %). Positive estrogen receptor (ER) status and younger age were associated with better survival outcomes. ML models identified key predictive features for survival, including age, nodal status, and ER status.

Conclusion: Age, lymph node metastasis, and ER status are crucial prognostic indicators for BACC. Although postoperative RT enhances survival, the advantages of adjuvant CTX are uncertain, implying that it may be eschewed to avert adverse effects. Our online tool offers essential resources for prognostication and treatment optimization.