MOSBY enables multi-omic inference and spatial biomarker discovery from whole slide images.

The utility of deep neural nets has been demonstrated for mapping hematoxylin-and-eosin (H&E) stained image features to expression of individual genes. However, these models have not been employed to discover clinically relevant spatial biomarkers. Here we develop MOSBY (Multi-Omic translation of whole slide images for Spatial Biomarker discoverY) that leverages contrastive self-supervised pretraining to extract improved H&E whole slide images features, learns a mapping between image and bulk omic profiles (RNA, DNA, and protein), and utilizes tile-level information to discover spatial biomarkers. We validate MOSBY gene and gene set predictions with spatial transcriptomic and serially-sectioned CD8 IHC image data. We demonstrate that MOSBY-inferred colocalization features have survival-predictive power orthogonal to gene expression, and enable concordance indices highly competitive with survival-trained multimodal networks. We identify and validate (1) an ER stress-associated colocalization feature as a chemotherapy-specific risk factor in lung adenocarcinoma, and (2) the colocalization of T effector cell vs cysteine signatures as a negative prognostic factor in multiple cancer indications. The discovery of clinically relevant biologically interpretable spatial biomarkers showcases the utility of the model in unraveling novel insights in cancer biology as well as informing clinical decision-making.

XIAP deletion sensitizes mice to TNF-induced and RIP1-mediated death.

XIAP is a caspase-inhibitory protein that blocks several cell death pathways, and mediates proper activation of inflammatory NOD2-RIP2 signaling. XIAP deficiency in patients with inflammatory diseases such as Crohn's disease, or those needing allogeneic hematopoietic cell transplantation, is associated with a worse prognosis. In this study, we show that XIAP absence sensitizes cells and mice to LPS- and TNF-mediated cell death without affecting LPS- or TNF-induced NF-κB and MAPK signaling. In XIAP deficient mice, RIP1 inhibition effectively blocks TNF-stimulated cell death, hypothermia, lethality, cytokine/chemokine release, intestinal tissue damage and granulocyte migration. By contrast, inhibition of the related kinase RIP2 does not affect TNF-stimulated events, suggesting a lack of involvement for the RIP2-NOD2 signaling pathway. Overall, our data indicate that in XIAP's absence RIP1 is a critical component of TNF-mediated inflammation, suggesting that RIP1 inhibition could be an attractive option for patients with XIAP deficiency.

Automated Dissection Protocol for Tumor Enrichment in Low Tumor Content Tissues.

Tumor enrichment in low tumor content tissues, those below 20% tumor content depending on the method, is required to generate quality data reproducibly with many downstream assays such as next generation sequencing. Automated tissue dissection is a new methodology that automates and improves tumor enrichment in these common, low tumor content tissues by decreasing the user-dependent imprecision of traditional macro-dissection and time, cost, and expertise limitations of laser capture microdissection by using digital image annotation overlay onto unstained slides. Here, digital hematoxylin and eosin (H&E) annotations are used to target small tumor areas using a blade that is 250 µm2 in diameter in unstained formalin fixed paraffin embedded (FFPE) or fresh frozen sections up to 20 µm in thickness for automated tumor enrichment prior to nucleic acid extraction and whole exome sequencing (WES). Automated dissection can harvest annotated regions in low tumor content tissues from single or multiple sections for nucleic acid extraction. It also allows for capture of extensive pre- and post-harvest collection metrics while improving accuracy, reproducibility, and increasing throughput with utilization of fewer slides. The described protocol enables digital annotation with automated dissection on animal and/or human FFPE or fresh frozen tissues with low tumor content and could also be used for any region of interest enrichment to boost adequacy for downstream sequencing applications in clinical or research workflows.