Time-dependent prognostic impact of tumor grade in stage I-III breast cancers

Breast cancer (BC) mortality risk extends over decades; yet, the temporal prognostic patterns of histological grade across subtypes and stages remain unclear.

Women with stage I-III BC diagnosed between 2000 and 2018 were identified from Surveillance, Epidemiology, and End Results registries (N = 767 218). Breast cancer-specific mortality (BCSM), as the primary endpoint, was analyzed using cumulative incidence functions including competing risk and annual hazard rates stratified by grade, estrogen receptor (ER) status, stage, and adjuvant chemotherapy. Restricted mean survival time differences were calculated to quantify absolute survival differences across follow-up intervals. Landmark survival analyses were carried out for 0 to <60, 60 to <120, and 120 to <180 months from diagnosis. Time-varying effects of grade were assessed through interval-specific modeling to evaluate nonproportional hazards.

Overall, 23%, 44%, and 33% of cancers were grades 1, 2, and 3, respectively, accounting for 8%, 35%, and 57% of BC deaths. In each consecutive follow-up period, the contribution of grade 1 cancers to BCSM increased, and the contribution of grade 3 cancers decreased. Overall, 59% and 30% of deaths from grades 1 and 3 cancers, respectively, occurred after 5 years. In ER-positive disease, grade 3 tumors showed peak annual hazards between years 3 and 5, whereas grade 1 tumors had lower but sustained hazards extending beyond 10 years; the hazard curves converged between years 10 and 12. In ER-negative disease, early hazards were higher, and convergence also started earlier at years 5-8. Absolute risk was influenced by nodal status and tumor stage within each grade category and follow-up interval.

Grade 1 cancers show sustained risk beyond 10 years, whereas grade 3 cancers exhibit front-loaded risk. These temporal risk patterns, together with other clinical and genomic data, could inform extended endocrine therapy decisions and surveillance strategies.