Immunohistochemical biomarker scoring in gastroesophageal cancers: Can computers help us?

The increasing complexity of cancer diagnostics and treatment selection has placed a growing burden on pathologists, particularly in the evaluation of immunohistochemical (IHC) biomarkers. In gastroesophageal cancers (GEC), both adenocarcinoma and squamous cell carcinoma subtypes, multiple prognostic and predictive biomarkers must be assessed to guide therapy. These evaluations require meticulous scoring, are time-consuming, and suffer from inter- and intra-observer variability. Given the worldwide shortage of pathologists, artificial intelligence (AI)-based tools have emerged as a potential solution to enhance efficiency and accuracy in biomarker scoring. This review aims to answer the question captured in its title: can AI help us in IHC biomarker scoring in GEC, and if so, how? A search of PubMed and Google Scholar was conducted to identify relevant studies. The analysis reveals that AI has demonstrated promise in improving reproducibility and reducing pathologist workload for biomarkers such as PD-L1 and HER2, although its applications in GEC remain limited compared to other cancer types. In parallel, predictive computational approaches are emerging that could revolutionize biomarker scoring altogether. By alleviating the burdens of complex scoring systems and costly additional assays, AI could have the potential to significantly enhance pathology practice in GEC biomarker evaluation.