A spatial index of immune barriers in the tumour microenvironment

Abstract Background The presence of immune cells within a tumour does not guarantee an effective anti-tumour response, because suppressive barriers can hold these cells at the tumour margin. Abundance-based summaries of the immune infiltrate cannot distinguish immune cells that engage the tumour from those excluded at its periphery. Our objective was to distinguish immune-excluded tumours, in which immune cells are present but held back by a suppressive barrier, from the infiltrated and immune-desert tumours that abundance metrics conflate, and to provide a reproducible, interpretable measurement of that barrier from spatial transcriptomic data. Methods We introduce the Spatial Immune Barrier Index (SIBI), a spatial-transcriptomic measure that combines immune-cell presence at the tumour periphery with the strength of the suppressive barrier such cells must cross. SIBI is defined as the product of an immune-presence gate and a minimax barrier term computed over paths from the tumour core to its exterior. We applied SIBI to two independent cancers profiled by spatial transcriptomics, high-grade serous ovarian carcinoma and head and neck squamous cell carcinoma, and to an independent bulk ovarian-cancer cohort with survival data. Results Across the two spatial cohorts, SIBI separated favourable from unfavourable samples more effectively than immune abundance alone (area under the curve 0.90 versus 0.71), and its ranking was robust to the choices made in its construction. In an independent ovarian cohort, the directional suppression programme underlying SIBI retained significant prognostic value for overall survival. Conclusions SIBI turns immune exclusion into a reproducible, interpretable spatial measurement whose distribution reproduces across seven independent external cohorts spanning two platforms and four cancers, outperforms ten established and classical spatial methods in separating outcome groups in the discovery cohorts, and captures a functional immunosuppressive barrier distinct from a physical extracellular-matrix barrier. It is presented as an exploratory research index rather than a validated clinical classifier; the external cohorts establish reproducibility of the index rather than independent outcome validation, and prognostic claims are anchored to the discovery and survival cohorts, with prospective multi-centre validation as the clear next step.