Near-Surface Dosimetry as a Predictor for Unplanned Surgery or Implant Failure After Post-Mastectomy Radiotherapy

Abstract

<h3>Purpose/Objective(s)</h3> Post-mastectomy radiotherapy (PMRT) improves survival in appropriately selected patients but is associated with a risk of unplanned surgery (US) and implant loss (IL) due to the impact of radiation on skin integrity, infection risk, contracture, and cosmetic outcome. If reliable dosimetric correlates for these outcomes can be found, optimization during radiation treatment planning may be useful to reduce their occurrence. We hypothesize that near-surface dosimetry predicts for increased US and IL risk. <h3>Materials/Methods</h3> Patients who received PMRT at a single institution from 2016-2019 were retrospectively analyzed. Demographic, clinical, and treatment parameters were reviewed. Three structures were retrospectively generated, encompassing the volumes bound by 0-3 mm (SR3), 0-5 mm (SR5), and 5-10 mm (SR10) from the skin surface. Dose volume histograms (DVHs) were exported for analysis. Univariate (UVA) and multivariate analyses (MVA) were used to identify predictors of US and IL among demographic and treatment variables. Dosimetric analysis was performed to determine candidate dose constraints for SR3, SR5, and SR10. Clinical variables included age, smoking history, hypertension, diabetes, body mass index (BMI), menopause status, TNM stages with total and positive lymph nodes as continuous variables, receptor status, margin status, type of lymph node surgery, reconstruction type, mastectomy type, and receipt of chemotherapy, trastuzumab, or hormone therapy. <h3>Results</h3> Of 126 patients reviewed, 109 were analyzable. Median follow up was 2.3 years. Twenty-five patients (23%) underwent US, 10 (9.2%) of whom experienced IF. Among clinical variables, BMI significantly predicted for US on UVA and MVA. There were no significant predictors of IF per se. Dosimetric analysis in 2 Gy intervals identified V44 Gy < 81% to SR3 (AUC 0.711) or V44 Gy < 83% to SR5 (AUC 0.685) as potentially useful for predicting US. V44 Gy remained a significant predictor of US when included on MVA. There were no significant dosimetric parameters using SR10. <h3>Conclusion</h3> Near-surface dosimetry is useful to predict US following PMRT. V44 Gy < 81% to SR3 should be further explored as a constraint for treatment planning optimization. The absence of significant predictors for IF is likely due to a paucity of events in this cohort.