Letter to the Editor: “Risk of Lymphedema After Sentinel Node Biopsy in Patients With Breast Cancer”

Dear editor,

We read the article “Risk of Lymphedema After Sentinel Node Biopsy in Patients with Breast Cancer” published by Byeon et al. [1], with great interest because some years ago we have previously assessed lymphedema risk after breast cancer treatment in a Brazilian cohort with long follow-up [2] and we are aware the challenge and interest this subject represents.

In this current study [1], the authors developed a lymphedema prediction model after sentinel lymph node biopsy and used variables which showed significant impact on outcome: radiotherapy, chemotherapy, body mass index ≥ 30 kg/m² and more than three harvested lymph nodes. However, we think their results need to be interpreted cautiously and we suggest some considerations. We are afraid that some aspects of the study compromise the internal validity of this study and, thus, its external validity.

Radiotherapy has been reported to increase lymphedema risk if lymphatic drainage pathways are targeted, whereas the same has not been observed when radiotherapy is restricted to the breast [3]. In the study [1], it will be important to know whether groups are similar in their radiotherapy approach.

Also, Byeon et al.’s mention [1] of the impact of chemotherapy without additional information may be misleading. Taxane and ipsilateral limb infusion are related to increased lymphedema risk [2, 4, 5], thus, a simple “yes” or “no,” although appealing to construct an easily understandable risk model may not be accurate enough to reach its objective.

As for lymphedema definition itself, the authors considered it a limitation of the study for it depended on patient’s symptoms previously to referral. Additionally, employment of three different methods to evaluate the same population can produce important classification bias. 5% of the study population were diagnosed using circumference difference between the upper limbs because these patients presented restricted range of motion due to postoperative adhesive capsulitis. The reason for using this method may be related to the increased risk of lymphedema.

Finally, we think that the author demonstrated that the clinical stage (tumor, node, metastasis; TNM) is more important than the size of the tumor (T stage) for it showed high collinearity with the variables included in the lymphedema predictor model. In other words, patients with advanced disease are those who undergo chemotherapy and radiotherapy in drainage chains. Therefore, construction of a multiple model requires testing of variable collinearity to support the final model. Furthermore, to evaluate T stage, a larger sample size is necessary for five categories. This caused the large confidence interval observed in the univariate analysis and in the erroneous OR of the T4 stage (odds ratio, 0.00; 95% confidence interval, 0.00–65.8). Maybe, the authors could consider this variable into two broader categories: Initial (Tis − T2) × Advanced (T3 − T4).