BACKGROUND: Despite recent discoveries concerning molecular alterations involved in astrocytoma tumorigenesis, the conventional histological analysis remains the best diagnostic method and a reasonable prognostic indicator for astrocytic tumors. However, the histopathological routine recommended by the World Health Organization (WHO) is laborious and marked by low reproducibility. OBJECTIVE: The present study aimed to develop auxiliary strategies for astrocytic tumor graduation according to histological criteria established by WHO. MATERIAL AND METHOD: A clinical and epidemiological analysis and a histopathological evaluation were performed in 55 astrocytomas of different graduations (13 grade I, 14 grade II, seven grade III, 21 grade IV) and five samples of non-tumor tissue (control group). RESULTS: The distribution by age, sex, and tumor localization of astrocytomas in patients reproduced, in a general way, worldwide trends. The presence of multinucleated cells and gemistocytes was associated with tumoral malignancy. The histopathological findings evaluated through semi-quantitative criteria and CART modeling confirmed the primary classification, reaffirming their applicability and reproducibility. CONCLUSION: These results indicate the CART decision tree and the totality of semi-quantitative scores as practical and auxiliary strategies for astrocytoma graduation as stated by WHO criteria.
Astrocytomas; Histopathology; Artificial intelligence; WHO
