Software Predicts Lymphedema Risk After Axillary Dissection

Tool for Deciding About Surgery?

Nick Mulcahy

September 07, 2011

September 7, 2011 — Newly developed mathematical models are at least 70% accurate in predicting the risk for lymphedema in women undergoing axillary dissection as part of their breast cancer treatment, according to Brazilian researchers.

The software models, or nomograms, were developed in conjunction with the largest-ever prospective study of lymphedema occurrence, said José Bevilacqua, MD, PhD, a surgical oncologist at Hospital Sirio Libanes in Sao Paulo, Brazil. He spoke at a press conference held in advance of the 2011 Breast Cancer Symposium, being held September 8 to 10 in San Francisco, California, where the new study will be presented as a poster.

The ability to predict lymphedema is "an important step forward" that "allows us to identify patients who can be appropriately triaged for early intervention" of the condition, said Andrew Seidman, MD, from Memorial Sloan-Kettering Cancer Center in New York City, who moderated the press briefing.

Dr. Bevilacqua concurred with that assessment, but had an additional proposal: that the nomograms could be used in discussions with some patients about the "necessity" of surgery.

Dr. Seidman did not comment directly on that idea, but noted that it was "interesting" that the study "comes at a real time of transition in the management of the axilla."

"Fewer women need to undergo axillary dissection today than decades ago," Dr. Seidman said. But patients with "more extensive lymph node involvement" still need the procedure, he pointed out.

The need for axillary dissection might be influenced by the number of involved nodes and the type of treatment a patient receives, such as regional radiation, he suggested.

A number of recent trials in early breast cancer have evaluated axillary dissection. A landmark trial in early breast cancer published this year found that, in certain women with 0 to 2 positive nodes, sentinel lymph node dissection did not result in inferior survival, compared with axillary dissection.

In addition, a trial presented at the 2011 meeting of the American Society of Clinical Oncology raised questions about the need for axillary dissection in women with early breast cancer and 1 to 3 positive nodes who are treated with regional nodal irradiation.

In the prospective study by Dr. Bevilacqua and colleagues, 1054 women with breast cancer underwent axillary dissection in 2001 and 2002. The overall 5-year incidence of lymphedema in the cohort was 30.3%. This incidence was expected, reported Dr. Bevilacqua, who explained that the chronic condition affects about one third of patients worldwide who have axillary lymph node surgery.

The median follow-up was 41 months, and 66 patients were lost to follow-up, he said. The arm volume of patients was measured with a simple measuring tape immediately before and after surgery, and every 6 months thereafter.

The Models, Variables and Findings

Using mathematical software, the investigators developed 3 models to predict the risk of developing lymphedema, which was defined as a volume difference of at least 200 mL between arms at 6 months or more after surgery.

The 3 models differ in a couple of ways: the point in time that risk is assessed and the variables used.

Model 1 predicted lymphedema risk in the study population in advance of axillary dissection. The model uses variables such as age, body mass index, and the number of chemotherapy infusions in the ipsilateral arm prior to surgery.

Dr. Bevilacqua and his colleagues compared the predictions of model 1 with the actual occurrence of lymphedema in this group of women, and found a "concordance index" of 0.706.

Model 2 predicted risk up to 6 months after surgery. It uses the same predictors as the first model, along with the extent of axillary dissection and the location of the radiotherapy field. The model had a concordance index of 0.729.

Model 3 predicted risk 6 months or more after surgery. It includes the variables of the development of postoperative seroma, infection, and early edema. The model had a concordance index of 0.736.

In short, all 3 models correctly predicted that a patient would develop lymphedema roughly 7 in 10 times. Dr. Bevilacqua noted that the accuracy of the models was on par with the accuracy of mammography in detecting breast cancer.

Dr. Bevilacqua said that he and his team plan to refine the models to increase their accuracy. Currently, some of their tools are available online for free. A tool to calculate arm volume can be accessed at The models to estimate the risk for lymphedema are available during the 2011 Breast Cancer Symposium (until September 10) at

Early Intervention

The nomograms and their ability to assess risk for lymphedema raise a "bigger question," Dr. Seidman noted. "Can some form of early intervention be useful?"

A "world of lymphedema experts" believe "early intervention matters," he said. At Memorial Sloan-Kettering, a team of physical rehabilitation experts is attempting to "control the progression of lymphedema."

Methods for treating and controlling the condition include the use of compression garments, arm elevation, not carrying backpacks and purses on the affected side of the body, and moderate exercise.

The authors have disclosed no relevant financial relationships.

2011 Breast Cancer Symposium. Abstract 8. To be presented September 8, 2011.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.