Are we using the right cut-offs to obtain maximum information from Ki-67 in patients with early breast cancer?

person Oana Gabriela Trifanescu Prof Dr. Al. Trestioreanu Bucharest Institute of Oncology, Bucharest, Romania info_outline Oana Gabriela Trifanescu, Liviu Bilteanu, Silvia Ilie, Diana Maria Bran, Inga Safta, Laurentia Minea Gales, Rodica Maricela Anghel

Authors person Oana Gabriela Trifanescu Prof Dr. Al. Trestioreanu Bucharest Institute of Oncology, Bucharest, Romania info_outline Oana Gabriela Trifanescu, Liviu Bilteanu, Silvia Ilie, Diana Maria Bran, Inga Safta, Laurentia Minea Gales, Rodica Maricela Anghel Organizations Prof Dr. Al. Trestioreanu Bucharest Institute of Oncology, Bucharest, Romania, University of Agronomic Sciences and Veterinary Medicine, Bucharest, Romania, Centre Georges-Francois Leclerc, Dijon, France, Prof. Dr. Al. Trestioreanu Bucharest Institute of Oncology, Bucharest, Romania, Department of Medical Oncology, Centre Antoine Lacassagne, Nice, France, Prof. Dr. Al. Trestioreanu Bucharest Institute of Oncology, Bucuresti, Romania Abstract Disclosures Research Funding No funding received None. Background: Ki-67 is one of the most critical prognostic proliferation markers in breast cancer, used on a large scale by oncologists to select treatment options. This study aimed to validate Ki-67 as a prognostic marker and establish the optimal Ki67 cut-offs for stratifying patient prognosis and treatment decisions. Methods: Medical files of 582 consecutive patients with early breast cancer (stage IB-IIIA) who underwent surgery as a first treatment in our institution between 2005-2019 were retrospectively reviewed. Values of Ki-67 were divided into 5 categories: 0-9%, 10-19%, 20-29%, 30-39%, and more than 40%. Results: Kolmogorov-Smirnov and Shapiro-Wilk tests showed that the patient’s age (yrs) at diagnostic, Ki-67 (in %), tumor size (in mm), estrogen and progesterone receptors abundance (%), exhibit normal distributions (p < 0.05). Median age at diagnosis was 54.8 (± 11.5) years and mean Ki-67 of 25.5 ± 16.5 and a median value of 20. Stage distribution was: stage IB 5.5%, IC 12%, IIA 46.6%, IIB 20.4%, IIIA 15.5%. The mean tumor dimension at diagnosis was 21.3 mm. After a median follow-up of 48 months (6-118), estimated disease-free survival rates (DFS) for all patients at 3 and 5 years were 92% and 79%. Disease-free survival rates at 5 years was 96% for patients with Ki-67 0-9%, 90% for patients with Ki-67 10-19%, 80% for patients with Ki-67 20-29%, 75% for patients with Ki-67 30-39%, and 72% for patients with Ki-67 more than 40%. The difference between the median DFS and each ki67 category was statistically significant (p < 0.0001). Using the fit of mixture method, we identify a new cut-off of 29% that negatively predicts DSF for patients (HR = 4.51 95% CI = 1.69-12.04, p = 0.001). Pearson correlation coefficients (p < 0.05) for Ki-67 and DFS, age, ER%, PR%, p53, and tumor size were -0.110, -0.096, -0.139, -0.136, 0.301, and 0.175, respectively. Moreover, DFS was also negatively correlated (p < 0.01) with the patient’s age (-0.106) and tumor size (-0.180). Though the correlation coefficients were small, they oriented the creation of linear models of DFS as a dependent variable and Ki-67, tumor size, and age. DFS has a statistically significant dependence only on the first two variables (standardized coefficients -0.103, p = 0.045 and -0.161, p = 0.02). Conclusions: Ki-67 is a strong prognostic factor in early breast cancer patients. New Ki-67 cut-off levels that may be used in clinical practice were identified. In addition, multivariate analysis showed a significant statistical influence of other factors, such as Ki67 and tumor size. Such correlations can be the basis of a scoring system for progression risk.