Work

Predictive model for assessing probability of breast cancer biological subtype change during treatment

Year: 2026
Type: article
Study design: Observational Study
Abstract: Background. Breast cancer demonstrates significant biological heterogeneity, which determines disease prognosis and the choice of systemic therapy. In clinical practice, the molecular subtype of the tumor is traditionally considered a relatively stable characteristic; however, recent data indicate the possibility of its change during the course of treatment. Investigation of patterns of tumor biological transformation represents an important area of personalized oncology, as such changes may influence the efficacy and appropriateness of the administered therapy. Aim. To develop a multiparametric prognostic model integrating histological, immunohistochemical, and clinical data for predicting changes in the molecular subtype of breast cancer. Materials and methods. The retrospective study included 261 patients with histologically verified invasive breast cancer treated between 2006 and 2025 at the "Medical City" Multidisciplinary Clinical Medical Center (Tyumen). Depending on the presence or absence of a change in the tumor molecular subtype, defined as a change in the immunohistochemical subtype between sequential examinations, patients were divided into two groups: group 1 – patients with confirmed subtype change; group 2 – patients without subtype change. Results. The developed prognostic model demonstrated AUC (area under curve, 0.779 (95 % confidence interval 0.71–0.82) with specificity of 85 % and sensitivity of 51 % at the standard cut-off threshold ( 0.5). With optimized threshold of p = 0.32, sensitivity of the model reached 72.4 %, specificity – 71.3 %. Internal bootstrap validation (1000 repetitions) confirmed stability of the model: AUC 0.79 (95 % confidence interval 0.72–0.86). To improve predictive accuracy, additional machine learning methods were applied (Random Forest: AUC 0.79, F1 0.77; LASSO: AUC 0.81, F1 0.78), which confirmed robustness of the identified predictors and consistency of prognostic estimates. Conclusion. Changes in the molecular subtype of breast cancer are a predictable event. The developed model enables identification of patients at high risk of tumor biological transformation, which may help optimize dynamic monitoring and ensure timely adjustment of therapeutic strategies.
Language: English
Subfield: Cancer Research
Open Access status: hybrid
OpenAlex ID: W7204990344
FWCI:
Cites: 17
Cited by:
Related to: 9