Association of Treatment Arm with Quantitative Cosmetic Outcome in a Randomized Trial Comparing Whole Breast Irradiation Dosing Schedules

Abstract

<h3>Purpose/Objective(s)</h3> We have previously validated that two measures of breast symmetry derived from an anteroposterior photograph, termed M1 and M2, correlate with patient- and physician-reported cosmetic outcome and offer an opportunity to quantitate breast cosmesis without relying on subjective human assessment. Within the context of a prospective randomized trial comparing two dosing schedules for whole breast irradiation (WBI) (50 Gy/25fx+10-14 Gy/5-7fx boost [CF-WBI] versus 42.56 Gy/16fx+10-12.5 Gy/4-5fx boost [HF-WBI]), we evaluated the impact of treatment arm on M1 and M2 through five years follow up. <h3>Materials/Methods</h3> 2D AP photographs framed from the mid-neck to umbilicus were obtained at baseline, 6 months, and yearly through 5 years on a prospective randomized trial of 287 patients with early breast cancer who were randomly assigned to CF-WBI+boost or HF-WBI+boost. For each photograph, we marked the coordinates of each nipple and inframammary fold and used these values to calculate M1 and M2 as per prior methods (PMID 28333010). M1 and M2 range from 0 to 1 with higher values indicating better nipple symmetry (M1) or vertical breast symmetry (M2). Multivariable logistic regression longitudinal models were created with M1 and M2 as outcomes, dichotomized at the median, to evaluate associations of treatment arm with M1 and M2 adjusting for relevant clinical, demographic, and treatment variables. <h3>Results</h3> A total of 287 patients were enrolled on this trial (n=149 allocated to CF-WBI; n=138 to HF-WBI) with median follow up 48 months. Median age was 60 years with 11% Black and 13% Hispanic and 44% obese. A total of 1600 photographs were acquired and marked to calculate M1 and M2. At five-years follow up, M1 was 0.92 for CF-WBI and 0.91 for HF-WBI (P=0.55) and M2 was 0.63 for CF-WBI and 0.66 for HF-WBI (P=0.40). In multivariable longitudinal analysis, treatment arm was not correlated with either M1 (OR 0.76; P=0.16) or M2 (OR 1.19; P=0.39). In contrast, obesity was associated with worse M1 (OR 0.50; P=0.003) and worse M2 (OR 0.58; P=0.03) and age≥70 years was associated with worse M2 (OR 0.39; P=0.02). <h3>Conclusion</h3> This is one of the first randomized trials to use an objective, quantitative approach to measure breast cosmesis and compare outcomes by treatment arm. We found that cosmesis as assessed via M1 and M2 was similar by treatment arm, providing high quality evidence supporting cosmetic equivalence of CF-WBI and HF-WBI. Future breast cancer radiation trials could adopt this approach to compare different radiation regimens.