Abstract
Background
Breast cancer stage at diagnosis varies across populations and healthcare settings. We aimed to identify factors independently associated with stage III disease across different healthcare contexts.Methods
We conducted a multicentre retrospective cohort study of patients with stage I-III invasive breast cancer diagnosed between January 1 and December 31, 2025, at four tertiary institutions in Spain, Costa Rica, Guatemala, and Panama. The primary outcome was stage III versus stage I-II disease. Multivariable logistic regression assessed age, country, molecular subtype, and diagnostic method. A sensitivity analysis replaced diagnostic method with routine screening. Model discrimination and calibration were assessed.Results
Among 918 patients initially identified, 843 were included; 632 (75.0%) had stage I-II and 211 (25.0%) stage III disease. Stage III prevalence ranged from 11.1% in Panama to 51.6% in Guatemala. Compared with Luminal A, Luminal B (adjusted OR [aOR], 2.28; 95% CI, 1.45-3.58), HER2-positive (non-luminal) (aOR, 4.01; 95% CI, 2.37-6.79), and triple-negative tumours (aOR, 1.91; 95% CI, 1.06-3.43) were associated with higher odds of stage III disease. Guatemala was also strongly associated with stage III presentation (aOR, 4.08; 95% CI, 2.68-6.20). Diagnostic method was not independently associated (aOR, 1.35; 95% CI, 0.94-1.94). In the sensitivity analysis (n=789), routine screening was not independently associated with stage III disease (aOR, 0.93; 95% CI, 0.64-1.35). The model showed an AUROC of 0.764 (95% CI, 0.733-0.806).Conclusions
Stage III presentation was independently associated with molecular subtype and participating country/centre. Reducing advanced-stage disease may require integrated strategies addressing tumour biology, screening, timely diagnosis, and healthcare-system factors.