Prediction of the 70-Gene Signature (mammaprint) High Versus Low Risk by Nomograms among Axillary Lymph Node Positive (LN+) and Negative (LN-) Chinese Breast Cancer Patients, a Retrospective Study
Jie Lian,Ru Yao,Ying Xu,Linjuan Tan,Fangyuan Chen,Jiahui Zhang,Yang Qu,Lu Gao,Yanna Zhang,Songjie Shen,Qingli Zhu,Xinyu Ren,Lingyan Kong,Bo Pan,Qiang Sun,Yidong Zhou
DOI: https://doi.org/10.21203/rs.3.rs-4977540/v1
2024-01-01
Abstract:Background: Luminal-type breast cancer (BC) was characterized as hormonal receptor positive human epidermal receptor 2 negative (HR+/HER2-), which comprises the majority of breast cancer (BC). The 70-gene signature (70-GS, MammaPrint) test is recommended for assessing recurrence risk and guiding adjuvant chemotherapy decisions in Luminal-type BC. Based on our previously established nomogram models for predicting binary categorized risk of 70-GS, this study aims to update nomogram models to predict binary 70-GS risk for lymph node positive (LN+) and lymph node negative (LN-) luminal-type BC patients. Methods: This retrospective study included 301 consecutive female patients with HR+/HER2- BC treated at Peking Union Medical College Hospital from November 2019 to December 2023. Patients' medical history, imaging reports, and clinicopathological features were reviewed. Forty risk parameters were compared between 70-GS high vs. low-risk patients among LN+ and LN- groups. High risk stratification criterion in MonarchE and Natalee were compared between low and high 70-GS risk for the first time. Logistic regression was utilized to establish nomogram models predicting binary 70-GS risk for LN+ and LN- patients. The models' prediction performance was evaluated using accuracy, AUC of ROC curves, C-index, calibration curves, and decision curve analysis. Results: Significant differences were found in several risk parameters between 70-GS high vs. low-risk patients in both LN+ and LN- groups. Among LN+ patients, parameters including childbirth number (p=0.024), cardiovascular diseases (p=0.037), US min. diameter of tumor (p=0.034), Ki67 index (p<0.001) and PR positivity (p=0.007) were significant predictors. Among LN- patients, micro-calcifications (p=0.011), PR positivity (p=0.021), and Ki67 index (p<0.001) were significant. The nomogram models showed high predictive accuracy, with AUC of 0.948 in the training set (C-index 0.948, 0.914-0.982, accuracy 0.907) and 0.923 in the testing set (C-index 0.923, 0.919-0.927, accuracy 0.828) for LN+ patients and 0.917 in the training set (C-index 0917, 0.861-0.972, accuracy 0.870) and 0.917 in the testing sets (C-index 0917, 0.912-0.922, accuracy 0.808) among LN- patients. Calibration plots and decision curve analysis demonstrated the models' reliability and clinical utility. Conclusions: Our updated nomogram models for predicting 70-GS risk in LN+ and LN- luminal-type BC patients demonstrated improved prediction performance. The models facilitate individualized risk assessment and treatment decision-making, highlighting the distinct risk factor distributions between LN+ and LN- patients. These findings support the use of tailored approaches in managing luminal-type BC based on lymph node status.