Classification of Breast Lymphedema in a Racially Diverse Cohort.

Abstract

Breast lymphedema is a common sequela of breast conservation that delays healing and reduces quality of life. No rigorous classification system exists for this condition. We explored approaches for classifying breast lymph-edema based on breast ultrasound, physical exam, and patient-reported outcomes. We enrolled 80 patients from two institutions. Each site enrolled 30 invasive breast cancer patients treated with breast conservation and radiotherapy, and 10 control patients evaluated for benign breast complaints. All patients underwent bilateral breast ultrasound to measure dermal thickness and were assessed for physical signs of breast lymphedema. Patients reported quality of life impacts on standard questionnaires. We derived breast lymphedema classifiers using (1) a simple ultrasound-based metric of dermal thickness difference, and (2) a multiparameter machine learning classifier based on dermal thickness difference, physical exam, and patient-reported impacts. Ultrasound-defined breast lymphedema was present in 72% (95% CI: 59 to 82%) of invasive breast cancer patients. The multiparameter classifier identified three distinct patient groups: one with little evidence of breast lymph-edema, and two with increasingly severe breast lymphedema. A simple ultrasound-based measure and a novel multiparameter classifier both show promise for rigorous classification of breast lymphedema and warrant further development in larger patient cohorts.