Novel AI Model Predicts TP53 Mutations and Survival Across 32 Cancer Types from Routine Pathology Slides
Key Insights
Researchers developed a single Vision Transformer AI model that predicts TP53 (search) mutation status, RNA expression, tumor type, and survival outcomes from routine H&E whole slide images.
The model achieved an AUROC of 0.766 for TP53 (search) mutation detection across 32 solid tumor types in an independent validation set of 1,729 slides.
Trained on over 11,000 primary tumor cases from the Pan-Cancer (search) Atlas, the model uses weakly supervised learning to avoid costly pixel-level annotations.
Researchers have developed a novel artificial intelligence model that can analyze a routine whole histopathology image and simultaneously predict cancer (search) subtype, specific genetic mutations, and survival outcomes across 32 different solid cancers rather than focusing on a single cancer type. The findings, published in The American Journal of Pathology (published by Elsevier (search)), highlight the potential of computational pathology to connect routine diagnostic imaging with molecular oncology.
Histopathology, the microscopic study of tissue, remains the gold standard for diagnosing cancer (search) and identifying prognostic features across most solid tumors. However, current clinical workflows still depend on additional molecular and genomic assays to identify key alterations, such as those in TP53 (search), which is one of the most frequently altered tumor suppressor genes across human cancers. Because mutations in TP53 influence tumor growth and treatment resistance, accurate diagnosis and molecular profiling are essential for guiding treatment and improving patient outcomes.
"Standard molecular profiling for TP53 (search) mutations is often costly and inaccessible in underprivileged or remote clinical settings," explains co-lead investigator Alex W. Hewitt, PhD, Menzies Institute for Medical Research and School of Medicine, University of Tasmania. "We wanted to develop a more practical tool for pathologists. Currently, most deep learning-based models are used for single-model concepts; one model for one task. We developed a single model that can generate seven outputs simultaneously from the whole histopathology image, including TP53 mutation status, TP53 RNA expression, tumor type, and survival-related outcomes at the slide level."
A Multi-Task Model Built on Weakly Supervised Learning
The AI-based Vision Transformer model analyzed routine hematoxylin and eosin (H&E) stained whole slide images of human solid tumors. The model was trained on a dataset that included more than 11,000 primary tumor cases retrieved from the Pan-Cancer (search) Atlas, with corresponding somatic mutation, RNA-sequencing, and clinical outcome data.
Because whole slide images are extremely large and complex, and obtaining detailed expert annotations for every relevant tumor region is difficult, costly, and often subjective, the researchers deployed a weakly supervised learning strategy. "In this study, molecular labels such as TP53 (search) mutation status were available at the patch level, but whole slide images containing TP53-associated morphological information were not manually labeled. Weak supervision enabled the model to learn from slide-level labels and identify relevant patterns across image patches without requiring exhaustive pixel- or region-level annotations," notes Dr. Hewitt.
Validation Performance and Clinical Implications
The most significant result was that the model achieved a strong predictive accuracy score (AUROC of 0.766) for TP53 (search) mutation detection across 32 solid tumor types in an independent validation set of 1,729 slides. The model also demonstrated the ability to infer TP53 RNA expression levels and tumor taxonomy directly from whole slide images.
Co-lead investigator Abadh K. Chaurasia, PhD, Menzies Institute for Medical Research, University of Tasmania, and Pandani Solutions Pty Ltd (search), notes, "This approach could help identify patients who may benefit from confirmatory molecular testing, support triage in settings with limited genomic testing, and provide additional decision support to clinicians. Importantly, this method should be viewed as complementary to molecular testing, not a replacement. Its potential impact is strongest as a screening, prioritization, or decision-support tool within broader diagnostic pathways."
Toward Clinically Integrated Computational Pathology
The study aligns with a broader push to move computational pathology from technical innovation toward clinical integration. A related Perspective in Science Bulletin, "Toward clinically integrated computational pathology: advancing diagnostic insight and practice transformation," argues that accuracy alone is not enough for AI to truly help pathologists. Useful AI tools must be interpretable, reportable, and compatible with existing pathology workflows, showing what was measured, where the evidence came from, how reliable the result is, and whether further testing is needed.
The Perspective also emphasizes the need for real-world validation, assessing computational pathology systems not only by accuracy but also by their effects on turnaround time, diagnostic consistency, case prioritization, unnecessary testing, clinical decision-making, and implementation costs. Near-term opportunities include biomarker prescreening, confirmatory test prioritization, quantitative scoring, treatment response assessment, and intraoperative support.
Dr. Hewitt concludes, "Accurate molecular profiling from routine histopathology slides, already widely used in cancer (search) care, could transform clinical oncology. This new AI-based model integrates diagnostic, molecular, and prognostic tasks, and could help clinicians obtain more information from existing pathology workflows, ultimately supporting more accessible precision cancer care and early intervention."
The study was supported by an Australian National Health and Medical Research Council Leadership Award (grant number GNT2009079).