Increased Complication Rates with Proton Therapy in Breast Cancer Patients with Immediate, Implant-Based Reconstruction: Single-Institution Comparative Effectiveness Analysis

Abstract

To compare the impact of protons vs. photons on breast reconstruction complications, for patients (pts) receiving postmastectomy radiation (PMRT) with either single-stage direct-to-implant (DTI) or two-staged expander/implant (TE/I). We reviewed the charts of 578 pts who underwent immediate reconstruction followed by radiation at our institution between 2010 and 2020. Pts with implant-based reconstruction using either TE/I or DTI and PMRT delivery in the presence of the prosthesis were included. Pts enrolled in active ongoing clinical trials were excluded from the analysis. The photon group received 3D conformal or IMRT/VMAT treatment with a median dose of 50-50.4 Gy in 25 to 28 fractions. For proton pts, treatment was delivered mainly with pencil beam scanning technique (PBS); few pts received passively scattered proton spread-out Bragg peak (SOBP). The complications were defined as infection/skin necrosis (I/N) requiring operative debridement, capsular contracture (CC) necessitating capsulotomy, and overall implant failure (ORF) as the removal of the permanent implant irrespective of replacement outcomes (i.e., with and without salvage reconstruction). We fit inverse-probability weighted cumulative incidence curves to adjust for confounding and non-random loss to follow-up. Various sensitivity analyses were conducted. Four hundred ninety-five pts were available for the final analysis with an overall median follow-up of 55 months. 66 (13%) received protons, of which14 were treated with SOBP protons. 137 (28%) and 256 (56%) received photons with and without chest wall boost (CWB), respectively. The 5-year inverse probability-weighted risk of CC post-PMRT was 31% for protons vs. 10% for photons (RR:3.09, 95% CI: 1.77, 5.40). The 5 years ORF risk was 35.6% in protons compared to 22.7% in photons pts (RR: 1.57; 95% CI 1.0, 2.48). Hazard ratios from the adjusted Cox models were 3.79 (p<0.001) for CC and 2.05 (p<0.01) for ORF. No difference in I/N was noted between protons and photons pts. Sensitivity analysis showed that protons significantly increased CC risk vs photons both with CWB (HR:3.56, P<0.001) and without CWB (HR:3.9, p<0.001). Similar outcomes were observed with ORF, where protons increased the rate of ORF compared to photons, irrespective of CWB (HR 1.8, p = 0.038 with CWB; HR 2.4, p = 0.004 without CWB). No differences between PBS and SOBP proton techniques were noted. Compared to photons, proton therapy increases the risk of capsular contracture requiring surgical intervention and hence overall reconstruction failure. This data should inform discussions about the risks and benefits of protons in patients with reconstruction, while awaiting mature data from ongoing clinical trials (RADCOMP) utilizing protons for breast cancer.