CORRELATION OF ACTUAL OVERALL SURVIVAL AND SURVIVAL DETERMINED BY ONLINE PROGNOSTIC TOOLS FOR PATIENTS WITH BREAST CANCER: WHAT IS THEIR APPLICABILITY IN CLINICAL PRACTICE?

Introduction: Breast cancer is a public health problem, given its prevalence and the high costs of treatment. In patients with luminal tumors, the selection of the most effective treatment with fewer side effects is sometimes troublesome, and the evaluation is based on genomic tests which are not largely available due to their high cost. Alternatively, there are online prognostic tools that integrate clinical and anatomo-pathological characteristics to help select the bet adjuvant treatment. Objective: To correlate the survival results defined by the Nottingham Prognostic Index (NPI) and PREDICT, prognostic tools, with the actual survival of patients with early breast cancer, exploring their possible applicability in clinical practice. Methods: A retrospective study involving 100 patients with early stage breast cancer diagnosed in 2006 at a referral cancer center in São Paulo – Brazil. The NPI and PREDICT tools were used to calculate a 5-year and 10-year survival probability score, respectively, and these results were compared to the patients actual survival. The information was analyzed through descriptive statistics. Survival analysis was performed using the Kaplan-Meier method and Log-Rank test. A p-value < 0.005 was considered statistically significant. Results: In the sample, overall survival (OS) was 91% at 5 years and 82% at 10 years, recurrence free survival (RFS) was 92% at 5 years and 89% at 10 years. Most patients were aged between 45-65 years, invasive ductal histological type, histological and nuclear grade 2, absence of vascular-lymphatic and perineural invasions, free margins. Chemotherapy was used in 56% of patients and hormone therapy in 97%. Pathological features such as greater tumor size, greater number of affected lymph nodes, negative progesterone receptor, vascular-lymphatic invasion and undifferentiated histological grade correlated with lower OS and RFS. The NPI and PREDICT tools presented a satisfactory correlation with the actual survival. In the comparison of tools, PREDICT presented better performance in the sample studied. Conclusion: Online tools had a satisfactory correlation with the actual survival of the patients studied in the present cohort. Some pathological variables were associated with worse OS and RFS.