Spatiotemporal loco-regional recurrence patterns after breast-conserving surgery.

Abstract

e12580 Background: Loco-regional recurrence (LRR) following breast-conserving surgery (BCS) represents a heterogeneous class of disease that has significant variation in its biological behavior and prognosis. Herein we report for the first time the spatiotemporal patterns of LRR post-lumpectomy in a Chinese population-based cohort. Methods: This retrospective study analyzed the single institutional patient-level data of breast cancer patients who underwent BCS between January 2006 and December 2016 at Shanghai Cancer Hospital of Fudan University. The enrolled patients were followed up regularly after initial surgery. Multivariate Cox regression models were performed to identify independent risk factors for LRR events. Recurrence patterns were scrutinized based on recurrence type and recurrence-free interval (RFI). Annual recurrence rates (ARR) were compared according to recurrence type and molecular subtype. Results: 4,325 patients were included, with a median follow-up of 66 months. A total of 120 (2.8%) LRRs were recorded as the first site of failure. Age at onset, pathologic stage, and molecular subtype were identified as predictors of loco-regional relapse. The major type of LRR was ipsilateral breast tumour recurrence (IBTR), which mainly (83.6%) occurred ≤5y post surgery. ARR curves showed that relapse peaked in the first 2.5 years in the overall population. Patients with regional nodal recurrence (RNR), shorter RFI, and synchronous distant metastasis were associated with a poorer prognosis. HER2+ positive disease had a higher rate of LRR events, more likely to have in-breast recurrence, and had an earlier relapse peak in the first 2 years after surgery. Conclusions: Currently, the LRR rate is generally low. Different recurrence patterns after BCS were related to distinct clinical outcomes. Management of LRR should be largely individualized and tailored to the extent of disease, the molecular profile of the recurrence, and to baseline clinical variables.