45P A three-gene signature marks the time to loco-regional recurrence in luminal-like breast cancer

Gene expression profiling (GEP)-based prognostic signatures are rapidly integrated into clinical decision-making for systemic management of breast cancer patients. However, GEP remains relatively underdeveloped loco-regional risk assessment. Yet, local-regional recurrence (LRR), especially early after surgery, is associated with poor survival. was performed on two independent luminal-like cohorts patients developing (≤ 5 years surgery) or late (>5 years) LRR and used, by training testing approach, to build a gene signature able intercept women at LRR. The data in silico datasets third cohort were used explore its value. Analysis the first lead identification three genes, CSTB, CCDC91 ITGB1, whose expression, derived principal component analysis, generated 3-gene significantly both (p-value: <0.001 0.005, respectively), overcoming discriminatory capability age, hormone receptor status therapy. Remarkably, integration these variables leads an AUC 0.878 (95% CI: 0.810-0.945). In we found that classifier retained association, showing higher values relapsed Moreover, additional cohort, relapse-free survival (HR 2.88; 95% CI 1.11-7.49, p-value: 0.030). Our represents new exploitable tool aid treatment choice recurrence.