Predicting Cancer-Related Lymphedema: A Decade of Outcomes and Model Development from a Cohort of 15,666 Cases

PURPOSE: Lymphedema is a common, debilitating condition following lymph node dissection and affecting 1 in 3 breast cancer patients. Current interventions focus on symptom management rather than prevention. There are no models looking at which patients are at highest risk and the timeline for development of lymphdemea. This decade-long study aimed to identify lymphedema predictors and develop an algorithm estimating both risk and onset. METHODS: All patients undergoing axillary lymph node dissection (ALND) at Yale Cancer Center (YCC) between 2013–2024 were included. Demographic and clinical variables were collected, including age, BMI, race, ethnicity, chemotherapy, radiation, and comorbidities. Two multivariate regression models were developed to assess risk factors for lymphedema development and predict time to lymphedema diagnosis. RESULTS: Among 15,666 ALND cases, 2,345 patients (14.9%) developed lymphedema, with an average onset of 20.5 months post-surgery. Independent risk factors included BMI >30 (OR 1.385, 95% CI 1.255–1.528, p<0.0001), chemotherapy (OR 2.445, 95% CI 2.189–2.734, p<0.0001), diabetes (OR 1.168, 95% CI 1.040–1.312, p=0.0085), Black/African American race (OR 1.443, 95% CI 1.248–1.666, p<0.0001), and radiation (OR 1.960, 95% CI 1.777–2.162, p<0.0001). Radiation (β=-4.496, 95% CI -6.376 to -2.617, p<0.0001), Black/African American race (β=-4.159, 95% CI -6.876 to -1.442, p=0.027) and Asian race (β=-6.796, 95% CI -12.76 to -0.8300, p=0.0256) were associated with earlier onset, while preexisting diabetes was linked to delayed presentation (β=2.813, 95% CI 0.6189 to 5.008, p=0.0120). The model demonstrated adequate calibration (Hosmer-Lemeshow p=0.7985), with statistically significant discriminative ability (AUC = 0.6873, 95% CI: 0.6755–0.6990, p<0.0001). CONCLUSIONS: This is the largest study to date to develop a clinically applicable prediction model for cancer-related lymphedema after axillary lymph node dissection (ALND). This model incorporates patient-specific demographics, clinical information and comorbidities. The Yale Cancer Center (YCC) lymphedema prediction model can be integrated into lymphedema care to proactively identify high-risk patients, estimate the timing of onset, and implement novel preventative strategies such as immediate lymphatic reconstruction. This predictive approach shifts the focus from late-stage management to prevention of this incurable condition, improving patient outcomes and long-term quality of life.