Use of technology to facilitate a prospective surveillance program for breast cancer-related lymphedema at the Massachusetts General Hospital

Breast Cancer-Related Lymphedema (BCRL) and Bioimpedance Spectroscopy: Long-Term Follow-Up, Surveillance Recommendations, and Multidisciplinary Risk Factors

BACKGROUND

Early detection and intervention for breast cancer-related lymphedema (BCRL) significantly decreases progression to persistent BCRL (pBCRL). We aimed to provide long-term follow-up on our early detection with bioimpedance spectroscopy (BIS) and early home intervention demonstrating reduced pBCRL to guide surveillance recommendations.

PATIENTS AND METHODS

In total, 148 female patients with breast cancer who had axillary lymph node dissection (ALND) from November 2014 to December 2017 were analyzed. Baseline BIS measurements and postoperative follow-up occurred every 3 months for 1 year, biannual for 1 year, and then annually. An elevated BIS triggered evaluation and initiation of at-home interventions with reassessment for resolution versus persistent BCRL (pBCRL). High-risk factors and timing were analyzed.

RESULTS

Mean follow-up was 55 months, and 65 (44%) patients had an abnormal BIS. Of these, 54 (82%) resolved with home intervention. The overall pBCRL rate was 8%. Average time to first abnormal BIS was 11.7 months. None of the stage 0 patients (0/34) and only 5/25 (20%) of stage 1 patients had pBCRL. All of stage 2 and stage 3 patients (7/7) had pBCRL. pBCRL correlated with number of positive nodes, percentage of positive nodes, stage of lymphedema at diagnosis, and recurring abnormal BIS measurements (p < 0.05).

CONCLUSIONS

We have shown that patients undergoing ALND with early BCRL identified by BIS who performed home interventions had an 8% pBCRL rate. Patients at high risk for pBCRL should have routine surveillance starting at 9 months postoperatively to identify an opportunity for early intervention.

Lymphedema in Inflammatory Breast Cancer Patients Following Trimodal Treatment

BACKGROUND

Breast cancer-related lymphedema (BCRL) is a debilitating sequela of breast cancer treatment and is becoming a greater concern in light of improved long-term survival. Inflammatory breast cancer (IBC) is a rare and aggressive malignancy for which systemic therapy, surgery, and radiotherapy remain the standard of care, thereby making IBC patients highly susceptible to developing BCRL. This study evaluated BCRL in IBC following trimodal therapy.

METHODS

IBC patients treated from 2016 to 2019 were identified from an institutional database. Patients were excluded if they presented with recurrent disease, underwent bilateral axillary surgery, did not complete trimodal therapy, or were lost to follow-up. Demographic, clinicopathologic factors, oncologic outcomes, and perometer measurements were recorded. BCRL was defined by clinician diagnosis and/or objective perometer measurements when available. Time to development of BCRL and treatment received were captured.

RESULTS

Eighty-three patients were included. Median follow-up was 33 months. The incidence of BCRL was 50.6% (n = 42). Mean time to BCRL from surgery was 13 (range 2-24) months. Demographic and clinicopathologic features were similar between patients with and without BCRL with exception of higher proportion receiving delayed reconstruction in the BCRL group (38.1% vs. 14.6%, p = 0.03). Forty patients (95.2%) underwent BCRL treatment, which included physical therapy (n = 39), compression (n = 38), therapeutic lymphovenous bypass (n = 13), and/or vascularized lymph node transfer (n = 12).

CONCLUSIONS

IBC patients are at high-risk for BCRL after treatment, impacting 51% of patients in this cohort. Strategies to reduce or prevent BCRL and improve real-time diagnosis should be implemented to better direct early management in this patient population.

Comparison of machine learning models for predicting the risk of breast cancer-related lymphedema in Chinese women

Objective

Predictive models for the occurrence of cancer symptoms by using machine learning (ML) algorithms could be used to aid clinical decision-making in order to enhance the quality of cancer care. This study aimed to develop and validate a selection of classification models that used ML algorithms to predict the occurrence of breast cancer-related lymphedema (BCRL) among Chinese women.

Methods

This was a retrospective cohort study of consecutive cases that had been diagnosed with breast cancer, stages I-IV. Forty-eight variables were grouped into five feature sets. Five classification models with ML algorithms were developed, and the models' performance and the variables’ relative importance were assessed accordingly.

Results

Of 370 eligible female participants, 91 had BCRL (24.6%). The mean age of this study sample was 49.89 (SD ​= ​7.45). All participants had had breast cancer surgery, and more than half of them had had a modified radical mastectomy (n ​= ​206, 55.5%). The mean follow-up time after breast cancer surgery was 28.73 months (SD ​= ​11.71). Most of the tumors were either stage I (n ​= ​49, 31.2%) or stage II (n ​= ​252, 68.1%). More than half of the sample had had postoperative chemotherapy (n ​= ​227, 61.4%). Overall, the logistic regression model achieved the best performance in terms of accuracy (91.6%), precision (82.1%), and recall (91.4%) for BCRL. Although this study included 48 predicting variables, we found that the five models required only 22 variables to achieve predictive performance. The most important variable was the number of positive lymph nodes, followed in descending order by the BCRL occurring on the same side as the surgery, a history of sentinel lymph node biopsy, a dietary preference for meat and fried food, and an exercise frequency of less than three times per week. These factors were the most influential predictors for enhancing the ML models’ performance.

Conclusions

This study found that in the ML training dataset, the multilayer perceptron model and the logistic regression model were the best discrimination models for predicting the outcome of BCRL, and the k-nearest neighbors and support vector machine models demonstrated good calibration performance in the ML validation dataset. Future research will need to use large-sample datasets to establish a more robust ML model for predicting BCRL deeply and reliably.