Cohort-scale Spatial Host-Microbiome Predicts Post-Resection Recurrence in Colorectal Cancer

The tumor microenvironment in colorectal cancer (CRC) is a heterogeneous ecosystem where host cells and microbial communities interact dynamically and influence disease progression. Clinical utility has been limited by the lack of a scalable spatial host-microbiome technique and by insufficient integration of artificial intelligence for interpreting high-dimensional multi-omics data.

To address these barriers, the authors present AlphaFISH, a platform technology that combines technical and computational innovations for multi-omics spatial analysis of clinical biopsies at subcellular resolution. It uses a sequencing-free, high-throughput spatial profiling method to construct what they describe as the largest clinical spatial transcriptomics and spatial microbiome datasets to date: 149 colorectal biopsies from 68 human subjects. These data are supported by a comprehensive scRNA-seq atlas covering 4.27 million cells across 650 patients for robust cell annotation.

Deep learning of the cohort-scale dual-omics data, consisting of more than 10 million subcellular sampling vectors, enables a transformer model with joint embeddings of gene expression, spatial architecture, and the microbial microenvironment in colon tissues. The model achieves nearly 90% accuracy in predicting CRC-associated pathological features from unseen spatial omics inputs. The AI interrogation further predicts tumour recurrence at 81% accuracy in 28 patients followed within 1 year after tumor resection.

AlphaFISH reveals that spatial interactions between Fusobacterium and a cellular niche consisting of tumor and T cells serve as key markers of CRC malignancy, progression, and recurrence. This indicates a critical role for spatial bacterial-immune crosstalk in CRC.