Recent Developments in Molecular Pathology

Recent advances in machine learning inform precision medicine and translational research. We introduce a pioneering approach that integrates pathology imaging with transcriptomics and proteomics to identify predictive histology features associated with critical clinical outcomes in cancer. We utilize 2,755 H&E-stained histopathological slides from 657 patients across 6 cancer types from CPTAC. Our models effectively recapitulate distinctions readily made by human pathologists: tumor vs. normal (AUROC = 0.995) and tissue-of-origin (AUROC = 0.979). We further investigate predictive power on tasks not normally performed from H&E alone, including TP53 prediction and pathologic stage. Importantly, we describe predictive morphologies not previously utilized in a clinical setting. The incorporation of transcriptomics and proteomics identifies pathway-level signatures and cellular processes driving predictive histology features. Model generalizability and interpretability is confirmed using TCGA. We propose a classification system for these tasks, and suggest potential clinical applications for this integrated human and machine learning approach. A publicly available web-based platform implements these models. Interesting paper from the Clinical Proteomic Tumor Analysis Consortium: https://lnkd.in/eAArJwDv

Research from Harvard & MIT used AI to unlock molecular insights in cancer pathology. Foundation models are revolutionizing computational pathology. But, most struggle to analyze entire whole-slide images (WSIs) and incorporate molecular data. 𝗧𝗛𝗥𝗘𝗔𝗗𝗦 𝗶𝗻𝘁𝗿𝗼𝗱𝘂𝗰𝗲𝘀 𝗮 𝗺𝘂𝗹𝘁𝗶𝗺𝗼𝗱𝗮𝗹 𝗳𝗼𝘂𝗻𝗱𝗮𝘁𝗶𝗼𝗻 𝗺𝗼𝗱𝗲𝗹 𝘁𝗵𝗮𝘁 𝗹𝗲𝗮𝗿𝗻𝘀 𝗳𝗿𝗼𝗺 𝗯𝗼𝘁𝗵 𝗵𝗶𝘀𝘁𝗼𝗽𝗮𝘁𝗵𝗼𝗹𝗼𝗴𝘆 𝘀𝗹𝗶𝗱𝗲𝘀 𝗮𝗻𝗱 𝗺𝗼𝗹𝗲𝗰𝘂𝗹𝗮𝗿 𝗽𝗿𝗼𝗳𝗶𝗹𝗲𝘀. • 𝗣𝗿𝗲𝘁𝗿𝗮𝗶𝗻𝗲𝗱 𝗼𝗻 𝟰𝟳,𝟭𝟳𝟭 𝗛&𝗘-𝘀𝘁𝗮𝗶𝗻𝗲𝗱 𝗪𝗦𝗜𝘀 𝘄𝗶𝘁𝗵 𝗴𝗲𝗻𝗼𝗺𝗶𝗰 𝗮𝗻𝗱 𝘁𝗿𝗮𝗻𝘀𝗰𝗿𝗶𝗽𝘁𝗼𝗺𝗶𝗰 𝗽𝗿𝗼𝗳𝗶𝗹𝗲𝘀, the largest dataset of its kind. • Enabled state-of-the-art survival prediction, identifying high-risk patients with up to 8.9% higher accuracy than previous models. • 𝗘𝘅𝗰𝗲𝗹𝗹𝗲𝗱 𝗶𝗻 𝗹𝗼𝘄-𝗱𝗮𝘁𝗮 𝘀𝗰𝗲𝗻𝗮𝗿𝗶𝗼𝘀, achieving near-clinical accuracy with just 4 training samples per class. • Introduced “molecular prompting”, allowing AI to classify cancer types and mutations without task-specific training. I like that the architecture of THREADS is notably modular. It begins with an ROI encoder based on CONCHV1.5 (a ViT-L model fine-tuned with vision–language data) to extract patch features. The patch features are then aggregated into a slide-level embedding via an attention-based multiple instance learning (ABMIL) slide encoder. In parallel, distinct encoders for transcriptomic data (a modified scGPT) and genomic data (a multi-layer perceptron) create molecular embeddings. This design not only enables integration of heterogeneous data types but also achieves remarkable parameter efficiency. For instance, THREADS is reported to be 4× smaller than PRISM and 7.5× smaller than GIGAPATH, yet outperforms them on 54 oncology tasks. Here's the awesome work: https://lnkd.in/g5y5HFuV Congrats to Faisal Mahmood, Anurag Vaidya, Andrew Zhang, Guillaume Jaume, and co! I post my takes on the latest developments in health AI – 𝗰𝗼𝗻𝗻𝗲𝗰𝘁 𝘄𝗶𝘁𝗵 𝗺𝗲 𝘁𝗼 𝘀𝘁𝗮𝘆 𝘂𝗽𝗱𝗮𝘁𝗲𝗱! Also, check out my health AI blog here: https://lnkd.in/g3nrQFxW

Small-molecule drugs are effective and thus most widely used. However, their applications are limited by their reliance on active high-affinity binding sites, restricting their target options. A breakthrough approach involves molecular glues, a novel class of small-molecule compounds capable of inducing protein-protein interactions. This opens avenues to target conventionally undruggable proteins, overcoming limitations seen in conventional small-molecule drugs. Molecular glues play a key role in targeted protein degradation (TPD) techniques, including ubiquitin-proteasome system-based approaches such as Proteolysis Targeting Chimeras (PROTACs) and Molecular Glue Degraders and recently emergent lysosome system-based techniques like Molecular Degraders of Extracellular proteins through the Asialoglycoprotein receptors (MoDE-As) and Macroautophagy Degradation Targeting Chimeras (MADTACs). These techniques enable an innovative targeted degradation strategy for prolonged inhibition of pathology-associated proteins. This review provides an overview of them, emphasizing the clinical potential of molecular glues and guiding the development of molecular-glue-mediated TPD techniques.

🔬 Breaking New Ground in 3D Tissue Imaging - A Game-Changing Collaboration! 🚀 I'm incredibly excited to share our latest breakthrough published in Nature Communications, representing a powerful collaboration between Vanderbilt University Vanderbilt University Medical Center, KAIST TOMOCUBE, INC., and leading medical institutes. "Revealing 3D microanatomical structures of unlabeled thick cancer tissues using holotomography and virtual H&E staining" Led by the visionary Prof. YongKeun 'Paul' Park (KAIST and TOMOCUBE, INC.) and myself at Vanderbilt University Medical Center Vanderbilt University our team has achieved what was once thought impossible: 🌟 THE WORLD'S FIRST completely non-destructive, label-free method to visualize living tissue in stunning 3D detail - no staining, no sectioning, no damage! 🎯 Revolutionary Breakthroughs: 📸 See inside tissues 50 μm thick - that's 12.5x thicker than traditional methods! 🔍 Watch individual cells in their natural 3D environment 💎 Preserve 100% of precious tissue for additional testing ⚡ Get results in hours with no prep 🤖 AI-powered virtual staining matches traditional H&E quality 💡 Why This Changes Everything: Imagine being able to "fly through" a tumor in 3D, tracking every cancer cell without destroying the sample. That's now reality! 🚀 Game-Changing Applications: ✅ Precision Cancer Surgery - See exact tumor margins in 3D ✅ Rare Disease Diagnosis - Analyze precious biopsies without waste ✅ Drug Development - Watch drugs penetrate tissues in real-time ✅ Single-Cell Genomics - Preserve tissue for multi-omics analysis ✅ Digital Pathology 2.0 - Enable global 3D consultations ✅ Personalized Medicine - Tailor treatments to 3D tissue architecture This technology, powered by TOMOCUBE, INC.'s cutting-edge holotomography systems, represents the future of medical imaging and diagnostics. Special thanks to our incredible team across Vanderbilt University Medical Center Vanderbilt University Mayo Clinic, Yonsei University, and TOMOCUBE, INC. for making this vision a reality! 🔗 Read the full paper: https://rdcu.be/encJ1

🔬 AI is Revolutionizing Cancer Research & Precision Medicine See how MSK is leading the charge! By combining AI, Machine Learning, real-world oncology data, and molecular & imaging data, Memorial Sloan Kettering Cancer Center is redefining what’s possible in cancer research. Their latest study proves the power of unifying multimodal data. The MSK-CHORD Clinicogenomic dataset transforms precision oncology by analyzing nearly 25,000 patient records using natural language processing and machine learning. 🔎 Why This Is a Breakthrough: ✅ NLP achieves >90% accuracy, extracting insights from unstructured clinical notes, pathology reports, and radiology data, eliminating tedious manual review. ✅ Multimodal AI models outperform traditional staging, predicting survival and metastasis risks with greater accuracy. ✅ New biomarker discovery: SETD2 mutations in lung adenocarcinoma are linked to lower metastatic potential and better immunotherapy response, a game-changer for precision medicine. The power of AI + multimodal data is no longer just theory, it’s already improving patient stratification, accelerating biomarker discovery, and driving better clinical outcomes. 💡 Precision medicine isn’t just digital. It’s intelligent. For those working at the intersection of AI, Oncology, and Digital Pathology, what innovations excite you most? Let’s discuss.

⚡️🔬📣 Here are our two latest preprints on how AI for Pathology can advance pre-clinical drug safety and toxicity assessment. Work led by our superstar postdoc Guillaume Jaume, Deep Learning-based Modeling for Preclinical Drug Safety Assessment 📄 Preprint: https://lnkd.in/dDsQrkfJ 🔍 Demo: https://lnkd.in/dPbKC2Xb 🌟 Insights: We trained a Vision Transformer model (TRACE) on H&E-stained whole-slide images from 150+ preclinical toxicity studies. We showed that TRACE can assist and augment pathological assessment for lesion detection, quantification and automatic dose-response characterization. TRACE was also evaluated alongside ten expert pathologists and showed better agreement with the consensus. AI-driven Discovery of Morphomolecular Signatures in Toxicology 📄 Preprint: https://lnkd.in/d-uTy9dr 🔍 Demo: https://lnkd.in/dRB-n96v 🌟Insights: We developed GEESE, an AI model trained to predict gene expression of 1,500+ targets from histology. We showed that GEESE can reveal molecular signatures associated with distinct morphologies and toxicity mechanisms that are preserved across multiple compounds and species. Congrats to Thomas Peeters, Simone de Brot, Andrew Song and everyone involved! Stay tuned for more coming soon.