Understanding exactly how tumors co-opt the immune system requires seeing protein expression and gene activity simultaneously, in the same tissue slice, at single-cell resolution — a technical barrier that has stymied cancer immunology for years. A new platform called IN-DEPTH now clears that hurdle, and its first application to a common blood cancer reveals a previously uncharacterized immunosuppressive circuit potentially driven by a latent herpesvirus.
IN-DEPTH (IN-situ DEtailed Phenotyping To High-resolution transcriptomics) is a streamlined laboratory workflow that performs single-cell spatial proteomics first — preserving protein epitopes — then uses the resulting imaging data to guide transcriptomic capture on the same slide without sacrificing RNA signal integrity. Critically, the team paired IN-DEPTH with a new computational framework, Spectral Graph Cross-Correlation (SGCC), which maps how cell populations functionally coordinate across space rather than simply cataloging cell types by location. Applied to diffuse large B-cell lymphoma (DLBCL) tissue, the combined approach distinguished Epstein-Barr virus (EBV)-positive from EBV-negative tumors and identified a coordinated remodeling involving tumor cells, macrophages, and CD4 T-cells. EBV-positive tumors showed enrichment of immunosuppressive C1Q-expressing macrophages, CD4 T-cell dysfunction, and a candidate IL27-STAT3 signaling axis as a putative driver of this suppressive niche.
This work sits at the intersection of spatial biology and cancer immunotherapy — two of the most rapidly evolving domains in biomedical research. The IL27-STAT3 axis is particularly compelling: IL-27 has a paradoxical dual role, promoting antitumor immunity in some contexts yet driving T-cell exhaustion in others, making it a plausible but underexplored therapeutic target in virus-associated lymphomas. The EBV-positive DLBCL subtype is notably more common in immunocompromised and elderly populations, so mechanistic clarity here has real clinical stakes. Key limitations include the early-stage, discovery-phase nature of the findings — validation in larger, prospectively collected cohorts is essential before any clinical translation. Nonetheless, the platform's compatibility with commercial spatial transcriptomics instruments positions IN-DEPTH as potentially scalable infrastructure for building the spatial multimodal AI models the authors envision.