Graduate student University of Ulsan College of Medicine, Asan Medical Center, Brain Korea 21 project Songpa-gu, Republic of Korea
Disclosure(s):
Yongjae Kim, MD: No financial relationships to disclose
Introduction/Rationale: Understanding primary tumor features predicting post-recurrence immunotherapy response in resectable stage III non-small cell lung cancer (NSCLC) could guide adjuvant strategies. We hypothesized that spatial profiling of surgical specimens would reveal microenvironmental determinants of programmed death ligand-1 (PD-L1)-discordant responses to subsequent immunotherapy.
Methods: Surgical specimens from eight stage III NSCLC patients with post-recurrence progression were analyzed by Xenium 5K spatial transcriptomics. Patients were stratified into four PD-L1-high (≥50%) non-responders and four PD-L1-negative ( < 50%) responders based on primary tumor PD-L1 expression and post-recurrence immune checkpoint inhibitor (ICI) response. Analyses included pathway enrichment, T cell exhaustion mapping, fibroblast subtyping, spatial neighborhoods, and NicheNet inference.
Results: Non-responders displayed elevated glycolysis and hypoxia pathways indicating metabolic reprogramming, while responders did not. Spatial analysis revealed T cell exhaustion proximal to tumors in non-responders with marked attenuation in responders. Fibroblast subtyping identified metabolically active CAFs (meCAFs) enriched adjacent to glycolytic tumors in non-responders, with tighter tumor-meCAF spatial coupling. NicheNet identified meCAF-derived VEGFA as a key signaling mediator targeting tumor cells, with predicted targets including ACKR3, PTGS2, and PLAU, implicating angiogenic-metabolic crosstalk driving immune exclusion.
Conclusion: Spatial profiling reveals that PD-L1-discordant ICI responses are determined by coordinated interactions of metabolically reprogrammed tumors, localized T cell exhaustion, and meCAF-driven signaling. These findings establish meCAF-VEGFA-tumor crosstalk as a therapeutic vulnerability and support spatially-resolved tumor-stroma profiling as a prognostic tool beyond PD-L1 for predicting ICI responses.
