Spatial dose-distribution-based risk mapping to predict moist desquamation in breast radiotherapy

<i>Objective.</i>A relationship between the regional spatial distribution of skin dose and the development of moist desquamation (MD) was established for patients treated with breast radiotherapy.<i>Approach.</i>A 56-patient dataset was used to develop and validate a dose-distance based metric to predict MD. Dose distributions for the skin were extracted from AcurosXB treatment plans, and patient reported outcomes were used to classify the incidence of MD across the whole breast and then more specifically in the inferior breast. The sensitivity and specificity of the metric was compared against dose-area (A38 Gy ⩽ 50 cm²) and dose-volume (V105% ⩽ 2% of the breast volume) predictive metrics with the same dataset.<i>Main results.</i>With a sensitivity of 70% and a specificity of 72%, the dose-distance metric outperformed the dose-area (45%, 55%) and dose-volume (43%, 56%) predictive metrics. The test performance improves to a sensitivity and specificity of 81% when excluding the full coverage breast support devices that confounded the skin dose identification in the analysis.<i>Significance.</i>This metric offers regional MD prediction and risk mapping to highlight regions at high risk of developing severe skin toxicity and is suitable for implementation within the treatment planning process.This work is based on data acquired for the following clinical trials: ClinicalTrials.gov NCT04543851 and NCT04257396.