Integrative bulk, single-cell and spatial transcriptomics characterizes the immune context of proliferative recurrence risk in early lung adenocarcinoma

<title>Abstract</title> <p>Background Stage IA lung adenocarcinoma (LUAD) is usually treated with curative-intent resection, yet a clinically relevant subset of patients recurs. Proliferation-related transcriptomic programs have prognostic associations in resected LUAD, but the immune context in which proliferative risk is expressed and the cellular sources of accompanying immunoregulatory signals remain incompletely resolved. We therefore tested whether a locked cell-cycle program associated with recurrence-related outcomes coexists with immune engagement, immunoregulation, or immune depletion, and then localized these signals using independent cell-resolved datasets. Results A cell-cycle program score (CCPS) was developed in 131 TCGA stage IA tumors (36 progression-free-interval events) and frozen before transport to GSE31210 (109 stage IA tumors; 15 recurrence-free-survival events); GSE8894 (63 postoperative LUADs; 33 events) was supportive because stage was unavailable. Across the two confirmed stage IA cohorts, CCPS correlated with immune engagement (pooled Spearman ρ = 0.223, 95% CI 0.098–0.341) and more strongly with the mapped immunoregulatory core (ρ = 0.474, 95% CI 0.369–0.567). Age- and sex-adjusted CCPS hazard ratios were 1.435 (95% CI 1.037–1.984) in TCGA and 1.877 (1.093–3.225) in GSE31210. Adding CCPS improved likelihood-based model fit, whereas bootstrap ΔC-index intervals approached or crossed zero. Post hoc analyses suggested attenuation from stage I to stage II, including within 0–36 months, despite substantial score overlap. Single-cell pseudobulks localized CCPS to malignant epithelium and checkpoint transcripts to T/NK, myeloid, and selected epithelial sources; the epithelial–T/NK immunoregulation correlation was nominal only (ρ = 0.738, P = 0.0458, FDR q = 0.210). Nine-lesion AIS–MIA–IAC analyses showed exploratory transcriptional remodeling, whereas unadjusted spatial proximity disappeared after capture-depth adjustment. Conclusions Higher CCPS was associated with recurrence-related outcomes and an immune-engaged, immunoregulatory transcriptional context. Frozen external transport, cell-resolved localization, exploratory progression biology, annotation robustness, and spatial sensitivity define a coherent but bounded evidence hierarchy; they do not establish clinical predictive utility, causal immune regulation, or depth-independent spatial confirmation.</p>