Analysis of Clinical Prognostic Variables for Triple Negative Breast Cancer Histological Grading and Lymph Node Metastasis

Background: Triple Negative Breast Cancer (TNBC) is a type of breast cancer with very bad prognosis. Predicting the histological grade (HG) and the lymph nodes metastasis is crucial for developing more suitable treatment strategies. Methods: We present the main clinical and pathological variables to predict the histological grade and lymph nodes metastasis via novel machine learning techniques. These variables are currently being used for prognosis and treatment in medical practice. This analysis was performed using a database of 102 Caucasian women diagnosed with TNBC. The results were cross-validated using random simulations of this dataset. Results: HG was predicted with an accuracy of 93.8% using a list of 6 prognostic variables with significant implications: Ki67 expression, use of Oral contraceptives, Col11A1 expression, Col11A1 score, E-cad truncated and Tumor size. The lymph nodes metastasis was predicted with an accuracy of almost 85% using only 6 prognostic variables: Vascular invasion, Tumor size, Perineural invasion, Age at diagnosis, Ki67 expression, and Col11A1 score. This analysis also served to establish the median signatures of the groups with and without lymph node metastasis, and proved the existence of a kind of small-size tumors (around 2.15 cm) with lymph node metastasis but not showing vascular and perineural invasions and higher protein Col11A1 score. Besides, these signatures proved to be very stable. Conclusions: The additional information conveyed by the prognostic variables found in these two classification problems provides new insight about the genesis and progression of this disease and can be used in medical practice to improve decisions in patient diagnosis and further treatment.