SS 12 - Breast 1 - Toxicity
83 - A Nomogram for Predicting Lymphedema Risk Following Axillary Surgery and Regional Nodal Irradiation Developed From the NCIC MA.20 Trial
Monday, October 22
10:45 AM - 10:55 AM
Location: Room 214 C/D
Jeffrey Gross, MD, MS
McGaw Medical Center of Northwestern University: Resident physician: Employee
A Nomogram for Predicting Lymphedema Risk Following Axillary Surgery and Regional Nodal Irradiation Developed From the NCIC MA.20 Trial
J. Gross, E. D. Donnelly, and J. B. Strauss; Department of Radiation Oncology, Northwestern University Feinberg School of Medicine, Chicago, IL
Purpose/Objective(s): Both axillary surgery and regional nodal irradiation (RNI) utilized in the treatment of breast cancer can lead to upper extremity lymphedema. The impact of radiotherapy field design on lymphedema risk, and interaction with extent of axillary surgery has not been well characterized. Using data from the NCIC MA.20 randomized trial, this study sought to correlate variations in radiotherapy field design and other risk factors with lymphedema outcomes, and build a nomogram to predict lymphedema risk.
Materials/Methods: Data from the NCIC MA.20 trial was obtained from the National Cancer Institute’s National Clinical Trial Network Data Archive. The MA.20 trial randomized 1,832 women to RNI versus whole breast irradiation (WBI) alone. Eligible patients had node-positive or high-risk node negative breast cancer treated with breast conserving and axillary surgery. Per protocol, women received RNI to a limited field defined as axillary level III, supraclavicular fossa (SCF) and internal mammary (IM) nodes. However, women with fewer than 10 nodes removed, or greater than 3 nodes positive received RNI to the full axilla, including axillary levels I and II. On the trial, Grade 2 lymphedema events, defined as moderate lymphedema requiring compression, or higher were reported as per National Cancer Institute Common Toxicity Criteria version 2.0. Multivariable logistic regression of the MA.20 cohort was performed for nomogram derivation. A retrospective cohort of 763 women from our institution, with matching trial eligibility criteria, was used for nomogram external validation.
Results: The nomogram was constructed from multivariable logistic regression by purposeful covariate selection on a randomly-selected training subset of 1,374 MA.20 patients. The model identified body mass index (OR 1.06, 95% confidence interval [CI] 1.03-1.09), number of nodes removed (0-7 nodes, reference; 8-11 nodes, OR 3.6, 95% CI 1.7-8.9; 12-15 nodes, OR 5.6, 95% CI 2.3-15.5; >16 nodes, OR 7.1, 95% CI 2.9-19.4), and radiotherapy field group (no RNI, reference; limited axillary RNI; OR 1.6, 95% CI 1.6-6.5; full axillary RNI, OR 3.2, 95% 2.3-8.9) as significant predictors of lymphedema. The accuracy and internal validity of the nomogram, measured by the area under the receiver operating characteristic (ROC) curve, was 0.74 when applied to the randomly-selected test subset of 458 MA.20 patients. When applied to our cohort of 763 similar patients as external validation, the area under the ROC curve was 0.72.
Conclusion: The nomogram developed from the NCIC.MA20 trial and externally validated can be used to predict expected lymphedema rates following axillary surgery and RNI based on patient body mass index, number of nodes removed, and radiotherapy field group. This information can be readily used by breast surgeons and radiation oncologists to counsel patients about expected lymphedema risk following potential treatments. We thank the Canadian Cancer Trials Group for their contributions to this work.
Author Disclosure: J. Gross: None. E.D. Donnelly: None. J.B. Strauss: Advisory Board; American Imaging Management. President; Chicago Radiological Society.