In this presentation, Rudy Hovelinck reflected on his extensive experience in diagnostics, drug development, and digital pathology. Hovelinck began by noting the inspiration for his presentation’s movie theme, which he used to introduce insights from his career, spanning the creation of diagnostic antibodies and immunohistochemistry kits, to his work at AstraZeneca on launching drug therapies with robust diagnostic biomarkers for precision medicine. In recent years, he has been active in digital pathology, offering a broad perspective on the field.
Hovelinck addressed common misconceptions about immunohistochemistry, such as its perceived lack of reproducibility and semi-quantitative nature. He argued that, despite these criticisms, immunohistochemistry remains a valuable and accessible biomarker technique. He traced its evolution, highlighting the pivotal role of Dora Richardson, who, while working at ICI Chemicals in Manchester, contributed to the development of oestrogen inhibitors and the first oestrogen receptor tests. This work laid the foundation for modern immunohistochemistry, which is now routine in pathology labs worldwide, with rapid turnaround times and broad accessibility.
Hovelinck emphasised that the majority of precision medicine drug therapies – estimated at 80% - are prescribed based on immunohistochemistry results. He discussed the challenges of training pathologists in complex algorithms, particularly for tests like PD-L1, and described how digital pathology platforms, such as those developed by Pathomation, facilitate global training and assessment. These platforms enable both self-learning and classroom settings, supporting reproducibility and certification.
The presentation explored the integration of AI in pathology, demonstrating how AI-assisted analysis can enhance training and diagnostic accuracy. Hovelinck highlighted the need for retraining with the advent of new therapies, such as those targeting HER2 ultra-low expression, and discussed the limitations of current assays. He advocated for an “App Store” model for AI tools, allowing laboratories to select from a menu of algorithms via SaaS platforms, thus improving flexibility and collaboration.
In conclusion, Hovelinck asserted the enduring importance and reliability of immunohistochemistry, the transformative potential of AI image analysis, and the necessity of ongoing training to ensure reproducibility and quality in pathology.
