TLS Atlas: Turning Tumor Immune Architecture into a Clinical Biomarker: Science | May 2026
Tertiary lymphoid structures are local immune hubs within tumors, but this study shows that their mere presence is not enough; their maturation, location, and cellular organization determine biological and clinical relevance.
This pan-cancer study analyzed TLS biology across 12 cancer types, combining spatial transcriptomics with artificial intelligence applied to routine H&E pathology slides.
TLSs were classified along a maturation spectrum: early TLS, primary follicle–like TLS, and secondary follicle–like TLS, reflecting increasing immune organization.
Mature TLSs showed coordinated immune architecture, including B-cell and T-cell zoning, follicular dendritic cell networks, chemokine signaling, and interferon-related immune activation.
The spatial position of TLSs mattered. Intratumoral, peritumoral, and distal TLSs were associated with different immune and tumor signaling patterns.
Tumor regions close to intratumoral TLSs showed stronger antigen presentation and immune activation signatures, including interferon and MHC class II pathways.
Tumor regions farther from TLSs showed relatively higher proliferative and invasive programs, including MYC signaling, cell-cycle activity, and epithelial–mesenchymal transition.
The authors developed an AI-based model capable of detecting and phenotyping TLSs directly from standard H&E whole-slide images, making the approach potentially scalable for routine pathology.
A maturation-aware TLS composite score performed better than simple “TLS present or absent” assessment for patient stratification across cancer and treatment settings.
This work suggests that TLS profiling may become a future immuno-oncology biomarker, helping predict prognosis, treatment response, and possibly guide trial design.
The clinical implication is important: future cancer immunotherapy decisions may depend not only on tumor genomics or PD-L1 expression, but also on the spatial immune architecture of the tumor microenvironment.
Prospective validation is still needed before TLS scoring can be used routinely in clinical practice.
Bottom line: This study transforms TLSs from a descriptive pathology feature into a measurable, AI-enabled spatial biomarker with potential relevance for immunotherapy selection and cancer prognosis.
