Nomograms for predicting the risk of arm lymphedema after axillary dissection in breast cancer

Abstract

Background: Lymphedema (LE) after axillary lymph node dissection (ALND) is a multifactorial, chronic, and disabling condition that currently affects an estimated 4 million people worldwide. Although several risk factors have been described, it is difficult to estimate the risk in individual patients. We therefore developed nomograms based on a large data set.

Methods: Clinicopathologic features were collected from a prospective cohort comprising 1,054 women with unilateral breast cancer undergoing ALND as part of their surgical treatment from August 2001 to November 2002. LE was defined as a volume difference of at least 200 ml between arms at 6 months or more after surgery. The cumulative incidence of LE was ascertained by the Kaplan-Meier method, and Cox proportional hazard models were used to predict the risk of developing LE on the basis of the available data at each time point: model 1, preoperatively; model 2, within 6 months from surgery; and model 3, at 6 months or later after surgery.

Results: The 5 year cumulative incidence of LE was 30.3%. Independent risk factors for LE were age, body mass index, ipsilateral arm chemotherapy infusions, level of ALND, location of radiotherapy field, development of postoperative seroma, infection, and early edema. When applied to the validation set, the concordance indices were 0.706, 0.729, and 0.736 for models 1, 2, and 3, respectively.

Conclusions: The proposed nomograms can help physicians and patients predict the 5 year probability of LE after ALND for breast cancer. Free online versions of the nomograms are available at http://www.lymphedemarisk.com/ .