Enhancing Ki-67 Quantification in Canine Mammary Carcinoma: A Digital Pathology Approach Using QuPath

Davids Rebelo; Joana Ferreira Rodrigues; Ana Laura Vinagre; Adelina Gama; Pedro Faísca

doi:10.60543/rlcmv.v20i.11078

Davids Rebelo Department of Veterinary Sciences, School of Agrarian and Veterinary Sciences (ECAV), University of Trás-os-Montes e Alto Douro (UTAD), Vila Real, Portugal
Joana Ferreira Rodrigues Gulbenkian Institute for Molecular Medicine, Oeiras, Portugal
Ana Laura Vinagre Gulbenkian Institute for Molecular Medicine, Oeiras, Portugal
Adelina Gama Department of Veterinary Sciences, School of Agrarian and Veterinary Sciences (ECAV), University of Trás-os-Montes e Alto Douro (UTAD), Vila Real, Portugal
Pedro Faísca Gulbenkian Institute for Molecular Medicine, Oeiras, Portugal

DOI: https://doi.org/10.60543/rlcmv.v20i.11078

Resumo

Objectives: Accurate Ki-67 quantification is crucial for canine mammary carcinoma molecular (MAC) classification. Traditional manual quantification (MQ) suffers from variability and inconsistencies. This study evaluated QuPath, a free digital tool, to standardize Ki-67 assessment and improve reproducibility.

Materials and Methods: Ki-67 immunohistochemistry was performed on 25 MAC cases.Three different approaches were used and compared with MQ obtained from five observer-selected hotspots (~1000 cells) from Targeted High-Power Field (HPF) images: 1 HPF Analysis: Automated quantification (AQ) of the same fields as the MQ (n=25) 2 Area-Based Analysis: Manual outlining of whole-slide tumor areas (~1000 cells), excluding non-tumor regions (n=21) 3 Algorithmic Hotspot Detection (AHS): Whole-slide algorithmic detection of the most proliferative areas (n=21)

Results:

Targeted HPF Analysis

MQ and AQ showed a strong correlation (r=0,92, p<0,0001) but the results were statistically different (p=0,01). MQ yielded higher Ki-67 ratios in 72% of cases, with an average difference of 3,71%. AQ identified more negative cells (80% of cases), while detecting more positive cells in only 36% of cases. The higher quantification of negative cells is due to the inability of Qupath to distinguish neoplastic negative cells from non-neoplastic negative cells.

Area-Based Analysis

MQ still showed higher ratios in 57% of cases compared to area-based AQ but differences were not statistically different (p=0,7081). The ability to avoid non neoplastic regions with this method improved accuracy.

Algorithmic Hotspot Analysis (AHS)

AHS revealed higher Ki-67 ratios in 76% of cases with the results being statistically different compared to MQ (p= 0,0077). QuPath's systematic, unbiased whole-slide scanning likely contributed to detecting hotspots with higher ratios.

Conclusion: AHD seems to be the most effective method for identifying hotspots. While AQ enhances Ki-67 quantification, it remains susceptible to artifacts and requires optimal staining. Further standardization is crucial for improving accessibility and reliability in veterinary pathology.

Keywords: Ki67; QuPath; Mammary Anaplastic Carcinoma, dog

Downloads

Não há dados estatísticos.