Objective: development of a predictive model based on binary regression to determine the likelihood of progression of I stage breast cancer.Materials and methods. A retrospective analysis of data of 385 patients with T1N0M0 stage breast cancer was performed. The minimum follow-up period was 120 months and the maximum made 256 months, with an average follow-up of 191 ± 36 months (16 ± 3 years). Using a forward stepwise selection (binary regression), the most important prognostic factors were selected, on the basis of which the predictive model “Risk Assessment Algorithm for Recurrence of Breast Carcinoma” was constructed.Results. During the study period, recurrence of stage I breast cancer was reported in 67 patients, representing 17.4 % of the total cohort. Five prognostic factors were selected by binary regression: grade, histological type, estrogen receptor expression, HER2 / neu overexpression and Ki-67 amplification. Kaplan–Meier analysis and Cox proportional hazards method demonstrated the influence of each of the selected factors on disease-free survival. Comparative analysis with other existing models showed that our prognostic model is inferior to Adjuvant! Online in terms of sensitivity (85 % ver- sus 95 %). However, it is superior in specificity (58 % versus 38 %), PPV (69 % versus 63 %) and AUC (84 % versus 70 %).Conclusions. In I stage breast cancer, factors such as grade, histological type, estrogen receptor expression, HER2 / neu overexpression and Ki-67 amplification are the most significant predictive factors influencing recurrence rates. The algorithm for assessing the risk of recurrence of stage I breast cancer can predict the risk of tumour progression with a sensitivity of 84 % and a specificity of 58 % (p <0.05).
Prognostic factors in patients with stage I, estrogen receptor-negative carcinoma of the breast. A clinicopathologic study