Abstract Background: Tertiary lymphoid structures (TLSs) are associated with improved survival and enhanced response to anti-cancer immunotherapy. This study aimed to develop a computed tomography (CT) imaging-based approach for non-invasive assessment of TLSs and immunotherapy response. Methods: This study involved 3,155 patients with gastric cancer (GC). TLSs were classified into four stages (absence, Agg, FL-I, FL-II) according to their maturation based on immunohistochemistry staining. A CT imaging-based TLSs scoring model (ctTLSs) was developed to assess TLSs status, subsequently classified into four classes (ctTLSs-0/1/2/3). We next evaluated the model's associations with prognosis and immunotherapy response. To enhance the model's interpretability, we analyzed multi-omics data and employed the Shapley value strategy. Results: The ctTLSs model achieved high accuracies in predicting TLSs status in the internal validation (AUC range: 0.727-0.809) and external validation (0.704-0.807) cohorts. In retrospective and prospective validation cohorts, ctTLSs exhibited significant associations with both disease-free and overall survival (HR range: 0.206-0.634, all P0.01). Shapley value analysis highlighted ctTLSs as the strongest predictor of TLSs status. Upon analyzing multi-omics data, we found that higher ctTLSs levels positively correlated with tumor immune activation and apoptosis signaling, while displaying a negative correlation with tumor proliferation and metabolism signaling. Intriguingly, patients with high ctTLSs (but not low ctTLSs) exhibited substantial benefits from immunotherapy (P0.0001). The objective response rate of four ctTLSs classes was 16.7% in ctTLSs-0, 35.5% in ctTLSs-1, 45.8% in ctTLSs-2, and 53.8% in ctTLSs-3. Conclusion: The ctTLSs model could noninvasively assess TLSs status, enabling improved prognosis evaluation and informed decisions regarding immunotherapy.
Sun et al. (2026) studied this question.