Validation of uPath HER2 dual-colour dual in-situ hybridisation image analysis tool for HER2/neu testing in breast...

HER2/neu gene is amplified in 15%-20% of invasive breast cancers (IBCs), serving as critical prognostic and predictive marker. HER2-targeted therapies have improved outcomes for HER2-positive patients, highlighting the importance of accurate assessment. Immunohistochemistry is commonly used for screening HER2 overexpression, with equivocal cases reflex tested using in situ hybridisation (ISH) methods like fluorescence (FISH) or dual-colour dual ISH (D-DISH). While FISH displays quantitative accuracy, it is expensive, time-consuming and technically demanding. D-DISH offers a faster, automated alternative using bright-field microscopy for easier interpretation and better archiving. Advances in digital pathology, such as whole slide imaging and automated image analysis (IA), promise to improve HER2 evaluation. The CE-IVD marked uPath HER2 Dual ISH IA algorithm by Ventana Medical Systems (Tucson, Arizona, USA) is designed to assist in this process, providing computer-assisted evaluation of HER2/neu. Thus, we undertook this study to standardise and validate uPath Dual ISH IA algorithm and assess interobserver reproducibility in interpreting D-DISH assay.

This study retrospectively analysed 106 IBC cases, evaluating the concordance between manual and algorithm-assisted D-DISH evaluations.

A consensus concordance rate of 91.5% and a Cohen's kappa value of 0.83 was observed between the manual and on-site IA evaluations, indicating near-perfect agreement. Remote IA evaluations also demonstrated substantial concordance, with a concordance rate of 88.89% and kappa value of 0.70.

We successfully validated the uPath IA algorithm as a time-efficient, screening modality as well as viable alternative to manual interpretation for both on-site and remote interpretation of HER2 D-DISH in a high-volume centre.