Propensity Score Matched Analysis of Post-mastectomy Reconstruction: What are the potential prognostic factors after twenty years?
Wooram F. Jung, MSc1, Meridith P. Pollie, BS1, Kaylee K. Ho, MS2, Elizabeth A. Mauer, MS2, Lisa A. Newman, MD, MPH3, David M. Otterburn, MD1
1Plastic and Reconstructive Surgery, NewYork-Presbyterian - Weill Cornell, New York, NY, 2Department of Population Sciences, Weill Cornell Medicine, New York, NY, 3Division of Breast Surgery, NewYork-Presbyterian - Weill Cornell, New York, NY
Background: Immediate breast reconstruction after mastectomy has been demonstrated to be safe over longitudinal periods for breast cancer patients. However, there are no randomized controlled trials directly comparing implant and flap-based reconstruction. Recently, a small Korean cohort to overcome this limitation using propensity score matching (PSM), who showed worse outcomes for high tumor grade flap patients compared implants. We wished to validate their results in a more diverse patient population. Over a twenty-year period, we evaluated survival and recurrence outcomes of implant and flap reconstruction using PSM in a heterogenous population.
Methods: We performed a retrospective study of mastectomy patients with immediate reconstruction using the Weill Cornell Breast Cancer Registry between January 1998 to January 2020. Patients were matched using propensity scores (1:1 nearest neighbor with 0.1 calipers) based on age, marital status, smoking, insurance, pathological stage, tumor grade, ER, PR, reconstruction year, chemotherapy timing, and radiation timing. Kaplan-Meier estimates for time-to-event endpoints and hazard ratios (HR) from Cox-regression models were used.
Results: A total of 1933 patients met our inclusion criteria. After propensity score matching, 538 implant and 538 flaps patients were used for analysis. Median patient age was 49 (IQR: 43, 57) and 50 (IQR: 42, 57) years for implants and flaps, respectively. There was no difference in the matching criteria variables between the two treatment groups. Survival probability did not differ between groups based on Kaplan-Meier analysis (p=0.72). Reconstruction type was not associated with survival (Flap—HR: 1.39, 95% CI: 0.87, 2.23; p=0.2), however, Medicaid/Medicare (HR: 3.28, 95% CI: 1.42, 7.55; p=0.005), pathological stage II (HR: 7.18, 95% CI: 2.19, 23.5; p=0.001) and III (HR: 12.9, 95% CI: 3.28, 50.6; p<0.001) were associated with worse survival. Adjuvant chemotherapy was associated with better survival (HR: 0.38, 95% CI: 0.20, 0.70; p=0.002). No difference in overall recurrence, locoregional, and distant recurrence was observed. Further subset analysis of high tumor grade patients failed to demonstrate worse survival with flap-based reconstruction (HR: 1.37, 95% CI: 0.68, 2.76; p=0.4).
Conclusions: To our knowledge, this is the first study using PSM to directly compare outcomes in implant and flap-based reconstruction using a large heterogenous patient population. After propensity matching, there was no statistical difference in survival or recurrence between different types of immediate breast reconstruction, even for high tumor grade patients. Instead, worse survival was associated with Medicaid/Medicare status, pathological stage II and III patients.
Back to 2021 Abstracts