Tumour morphology as a visible readout of cancer biology: Towards artificial intelligence‐enabled multimodal risk stratification in anal squamous cell carcinoma

Tumour morphology as a visible readout of cancer biology: Towards artificial intelligence-enabled multimodal risk stratification in anal squamous cell carcinomaTo the Editor, We read with great interest the study by Puthiyakunnel Saji et al.[1], which demonstrated that ulcerated morphology in anal squamous cell carcinoma (ASCC) was associated with more advanced disease, poorer complete response to chemoradiotherapy, higher local recurrence and inferior survival.This study is valuable not CORRESPONDENCE multimodal model incorporating visible morphology, MRI-defined tumour volume, nodal status, treatment intent and radiomic signatures may therefore provide a more robust risk framework than morphology or stage alone.Third, pretreatment biopsy specimens offer an opportunity to connect visible ulceration with microscopic biology.AI-assisted digital pathology could quantify invasion patterns, keratinization, necrosis, tumour-stroma ratio, lymphocyte density and spatial immune organization on whole-slide images.Given the rapid development of AI in digital pathology, such pathomic features may help determine whether ulcerated ASCC represents delayed diagnosis, aggressive local biology, or a distinct tumour-microenvironment phenotype.Future studies should therefore prospectively record morphology using standardized image acquisition, central review and interobserver agreement testing.AI models should be externally validated across institutions, scanners and imaging protocols, with attention to calibration, interpretability and clinical utility.Ultimately, the goal should not be to replace clinical judgement, but to convert a simple bedside observation into a reproducible, multimodal and biologically interpretable tool for risk stratification, surveillance design and translational sampling in ASCC.