Integration of a clinical survival model into the KM-plotter to identify high-risk populations in breast cancer

Abstract We aimed to establish clinically meaningful survival models in breast adenocarcinoma to support pharmacological research and guide therapy development. Using data from over 1.1 million patients in the SEER database, we performed univariate and multivariate Cox survival analyses on infiltrating ductal carcinoma cases stratified by pathological and demographic variables. By quantifying hazard ratios and Kaplan–Meier survival estimates across clinical subgroups, we identified patient cohorts with significantly reduced breast-cancer-specific survival. We further approximated the life-years lost in each subgroup to assess the cumulative survival burden and to prioritize cohorts for therapeutic intervention. Significant survival differences were observed across molecular subtypes and pathological categories, with ER-negative, triple-negative, node-positive, high-grade, advanced-stage, and larger tumors showing markedly reduced 60-month survival. For instance, survival declined from 96% in Stage I to 33% in Stage IV disease, from 98% in Grade 1 to 80% in Grade 4 tumors, and from 95% in tumors 10–19.9 mm to 51% in those ≥ 100 mm. Receptor-negative, high-grade tumors, despite their lower prevalence, account for a disproportionately large fraction of total life-years lost. To facilitate reproducibility and exploratory analyses, we integrated these clinical survival models into an upgraded version of the Kaplan–Meier plotter, which now supports user-defined cohort analysis based on clinicopathological variables. In summary, here we provide a large-scale, clinically grounded survival framework to identify and optimize target cohorts for pharmacology developments in breast cancer.