Predicting Nonsentinel Lymph Node Metastasis Using Lymphoscintigraphy in Patients with Breast Cancer

Journal of Nuclear Medicine

Eun Sook Lee¹  Seok Won Kim¹  Seok-ki Kim¹  Hyo Sang Lee¹  Seeyoun Lee¹  So-Youn Jung¹  Han-Sung Kang¹  Tae-sung Kim¹  Byoung-Hee Kim¹  Youngmi Kwon¹ 

关键词: lymphoscintigraphy;    nonsentinel lymph node;    breast cancer;    metastasis prediction;   

DOI  :  10.2967/jnumed.112.106260

学科分类：医学（综合）

来源: Society of Nuclear Medicine

【 摘 要 】

Several models for predicting the likelihood of nonsentinel lymph node (NSLN) metastasis using histopathologic parameters in sentinel-positive breast cancer patients have been proposed. In this study, we established a new model that uses sentinel lymphoscintigraphic findings and histopathologic parameters as covariates and assessed its predictive performance. Methods: The analysis included breast cancer patients (n = 301 women) who underwent sentinel lymphoscintigraphy (SLS) using 99mTc-labeled human serum albumin, had sentinel lymph node biopsy results positive for metastasis, and subsequently underwent complete axillary lymph node dissection. First, we devised a grading system relating SLS patterns to the risk of NSLN metastasis positivity. Second, we developed a multivariate logistic regression model for the prediction of NSLN metastasis using the SLS pattern and histopathologic parameters as covariates and compared its performance with that of the extensively validated Memorial Sloan-Kettering Cancer Center model using receiver-operating-characteristic curve analysis. Results: The SLS visual grade was strongly correlated with the presence of NSLN metastases. A well-calibrated prediction model for NSLN metastasis was constructed using SLS grade and histopathologic findings. The mean area under the curve of our model was 0.812, which is significantly greater than that of the Memorial Sloan-Kettering Cancer Center model (P < 0.001). A nomogram was drawn to facilitate the application of our model. Conclusion: SLS can aid in predicting NSLN metastasis in patients with breast cancer. Our model performed better than did established prediction models.

【 授权许可】

Unknown