Neutrophil-epithelial niches arrest B-cell maturation in dysfunctional tertiary lymphoid structures of gastric cancer neoadjuvant immunochemotherapy.

BACKGROUND

Tertiary lymphoid structures (TLS) are widely viewed as favorable in cancer immunotherapy, yet their clinical impact varies dramatically, and the local determinants of TLS quality remain poorly defined.

METHODS

We integrated spatial transcriptomics, matched single-cell RNA sequencing, multiplex tissue imaging, in vitro perturbation assays, orthogonal spatial-proteomic validation, and external bulk transcriptomic cohorts encompassing more than 1100 patients to define TLS functional states and their associated immune-ecological determinants in gastric cancer.

RESULTS

We identify a single fate-determining TLS subtype in which response-divergent germinal-center and plasma-cell trajectories unfold. In non-responders, B cells arrest at a naive/antigen-presenting state through collapse of a mid-trajectory NF-κB licensing module. This arrest is imposed by a coupled peri-TLS niche of FCN1_neutrophils and an ATF4-driven, immunologically silent tumor-epithelial state. Neutrophils deliver two convergent suppressive arms: contact-dependent TRAIL-caspase-8-RIP1 cleavage and an EGR1-DGUOK-uric-acid metabolic-stress axis-both targeting B-cell NF-κB licensing, while adjacent ATF4-driven epithelium compounds the block through serine and purine competition. Translating this logic into a bulk-transcriptomics-compatible two-axis framework, three TLS Immuno-Ecological Archetypes stratify immunotherapy response (functional 47.1%, dysfunctional 16.7%, excluded 10%), with the TLS Response Index outperforming abundance alone and consistently stratifying survival across >1100 patients.

CONCLUSIONS

In gastric cancer neoadjuvant immunochemotherapy, TLS function is governed by B-cell NF-κB-dependent maturation, and the surrounding neutrophil–epithelial niche. This explains why TLS abundance alone underpredicts benefit and nominates the neutrophil–epithelial peri-TLS niche as a candidate therapeutic target and biomarker layer.