Abstract
Introduction: Breast cancer (BC) is the most common malignancy among women worldwide. Thus, this retrospective cohort study aimed to identify prognostic factors for BC-specific survival using parametric accelerated failure time models.
Methods: Data were collected from 820 female patients diagnosed between 2014 and 2020 at Shahid Mostafa Khomeini Hospital in Ilam, Iran, with follow-ups through 2023. Survival time was defined from diagnosis to BC death or censoring. The Kaplan-Meier method, along with log-rank tests, was applied in the model.
Results: Among 798 patients, 72 BC deaths occurred (9.0%). The goodness of fit in the log-normal model was evaluated using the Akaike Information Criterion (AIC=574.2) and the Bayesian Information Criterion (BIC=607.0). Moreover, histological type was the only significant independent predictor of survival. Compared with infiltrating ductal carcinoma, lobular carcinoma was associated with shorter survival (time ratio=0.613; 95% confidence interval: 0.401–0.939; P=0.024). Nonetheless, age, tumor topography, diagnostic method, and tumor grade were not statistically significant predictors.
Conclusion: Histological type is a key prognostic factor for breast cancer survival, with non-ductal tumors indicating poorer outcomes. The log-normal model provides the best statistical fit for this analysis.