Establishment And Evaluation Of Prediction Models For Predicting Breast Cancer-Related Lymphedema

Purpose:The aim of our study was to evaluate the predictive values of the Scoring System for predicting the risk of breast cancer-related lymphedema (including Logistic scoring system and Additive scoring system) and also to develop a new nomogram for predicting the risk of lymphedema.Methods:Part I:Validation the Scoring System for predicting the risk of breast cancer related lymphedema. The research was started from October 2014 to November 2015 in two hospital of Anhui Province. The social demographic and clinical information of breast cancer patients that underwent breast surgery with unilateral axillary lymph node dissection (ALND) were collected. Lymphedema was defined by circumferential measurement. We calculate the risk score of lymphedema by the prediction model of the Scoring System. The receiver operating characteristic (ROC) curve (AUC) was used to evaluate the predictive efficiency of the Scoring System for predicting the risk of breast cancer-related lymphedema. Part II:Establishment a nomogram for predicting the risk of breast cancer related lymphedema. The research was started from October 2014 to November 2015 in two hospital of Anhui Province.Four categories of demographic-related, disease-related,treatment-related and behavior-related information of breast cancer patients were collected. Lymphedema was defined by circumferential measurement. Risk factors of lymphedema was determined by logistic regression analysis. A nomogram for predicting the risk of breast cancer related lymphedema was built by R soft. Receiver operating characteristic (ROC) curve (AUC) and Hosmer-Lemeshow goodness of fit test were used to evaluate the predictive efficiency of the nomogram for predicting the risk of breast cancer related lymphedema.Results:The total objective of this study was 221. The average age of the patients was 49 years,74 patients with lymphedema,147 patients without Lymphedema. Part Ⅰ:The AUC of Logistic score system was 0.779 (P<0.001,95% CI:0.717-0.842). The AUC of Additive score system was 0.804(P<0.001,95%CI:0.745-0.864). Part Ⅱ:By comparing the groups between lymphedema and non-lymphedema, the statistical difference was found in incision type, the level of ALND, chemotherapy, radiotherapy, complications, functional exercise and Lymphedema Risk-Reduction Behavior (avoiding strenuous exercise, avoiding lift heavy objects) (P<0.05).The multiple stepwise logistic regression showed that the level of ALND, radiation therapy, functional exercise andLymphedema Risk-Reduction Behavior (avoiding strenuous exercise, avoiding lift heavy objects) were significant correlated with breast cancer-related lymphedema. Five variables were the predictive factors of lymphedema. The AUC of nomogram was 0.815 (P<0.001, 95%CI:0.748～0.883), Hosmer-Lemeshow goodness of fit test (x2=3.499, P>0.05).Conclusions:It is the first time design to evaluate the Scoring System for predicting the risk of breast cancer-related lymphedema in demestic and also the first time to develop a nomogram for predicting breast cancer-related lymphedema.The Scoring System have good clinical application value which can predict the risk of Lymphedema effectively. The level of ALND, radiation therapy, functional exercise and Lymphedema Risk-Reduction Behavior (avoiding strenuous exercise, avoiding lift heavy objects) were the risk factor of lymphedema in this nomogram.Comparing with the Scoring System, this new nomogram has better AUC with fewer variables. Physicians and patients can predict the probability of lymphedema by individual nomogram accurately. In order to predict the risk of lymphedema, the Physicians should consider the characteristics of prediction models to improve the patients’ quality of life.