Microscopia digital x microscopia óptica na avaliação histológica do câncer de mama: estudo de concordância e análise dos fatores pré-analíticos

Abstract

INTRODUCTION: The use of whole slide imaging has proven to be a promising tool in the evaluation of invasive breast carcinoma, especially when integrated with advances such as artificial intelligence and automated analysis. However, challenges remain in ensuring reproducibility and reducing interobserver variability. Factors such as pre-analytical control, pathologist experience, and the evaluation platform itself must be assessed to improve breast cancer diagnosis and treatment. OBJECTIVES: To compare the analysis of prognostic and predictive factors in breast cancer using conventional optical microscopy (OM) versus digital microscopy (DM) by pathologists with different levels of experience; to compare the evaluation time for each observer using both methods; and to assess the influence of pre-analytical factors on these analyses. METHODS: This is a cross-sectional study was conducted with samples from 37 cases of invasive breast carcinomas with no prior treatment, collected with and without pre-analytical phase control. The samples were stained with hematoxylin and eosin and immunohistochemical markers and evaluated by three pathologists with varying experience levels using both OM and DM. RESULTS: The study demonstrated good intraobserver agreement between OM and DM for the main variables, especially for histological type (κ between 0.56 and 0.63; p < 0.001) and histological grade (κ between 0.57 and 0.78; p < 0.001), both showing moderate to substantial agreement. Interobserver analysis showed moderate agreement for histological type in OM (κ = 0.45; p < 0.001) and fair agreement in DM (κ = 0.35; p < 0.001). Mitotic count showed the lowest interobserver agreement in both modalities (κ = 0.27 in OM; κ = 0.26 in DM; p < 0.001). In some variables, the group without pre-analytical control showed higher kappa values. The mean evaluation time was longer with DM. MAIN CONCLUSION: The results indicated moderate to strong intraobserver agreement between the methodologies, with a tendency toward higher histological grading and increased HER2 expression in digital microscopy (DM). Interobserver analysis showed lower reproducibility, particularly for the mitotic index. A longer average evaluation time was also observed in DM, reflecting the need for greater familiarity with the platform. It is concluded that, although promising, digital microscopy requires specific training, technical standardization, and strict pre-analytical control to ensure diagnostic safety and reproducibility comparable to conventional optical microscopy.