A dynamic nomogram predicting persistent breast cancer-related lymphedema: a retrospective cohort study in China.

Abstract

Background: Once developed, persistent lymphedema (PLE) is irreversible and imposes multiple adverse challenges and a heavy economic burden on patients and the healthcare industry. This study aims to develop a risk nomogram model for PLE in breast cancer-related lymphedema (BCRL) patients and visualize it as a free online prediction website to guide individualized risk stratification and graded management.

Methods: 418 BCRL patients who underwent axillary lymph node dissection (ALND) among 2176 postoperative breast cancer patients from January 2020 to December 2022 were retrospectively enrolled as research subjects. Univariate and logistic regression models were performed to identify risk factors. A visual dynamic nomogram was constructed using R and Shinyapps software, followed by validation of its discrimination, calibration, and clinical validity.

Results: PLE incidence was 32.78%. Age, ALND level, severity of lymphedema, dominant side, and lymph node metastasis were significant risk factors for PLE (P < 0.05). The nomogram's C index was 0.827 (95%CI 0.774-0.880) in the training cohort and 0.849 (95%CI 0.782-0.916) in the validation cohort. Calibration curves showed good consistency across both cohorts. Decision curve analysis confirmed the good clinical validity, identifying 36% as the optimal threshold probability.

Conclusion: The dynamic nomogram, leveraging readily available clinical parameters, offers a clinically applicable web-based platform for dynamic risk quantification of PLE, facilitating early prediction, resource allocation, and prognosis management for high-risk PLE patients.