Single-Cell and Bulk Transcriptomic Integration Reveals an Adverse SPP1 Inflammatory TAM State and INHBA as a
Sheng Zhang1, Guojun Zhang2, Wenxuan Wang1
1Department of Radiology, Beijing Anzhen Nanchong Hospital of Medical University & Nanchong Central Hospital, Nanchong, China.
Background:
Biomarkers that capture the microenvironmental basis of response to neoadjuvant immunotherapy in non-small cell lung cancer (NSCLC) remain limited. We integrated single-cell and bulk transcriptomic data to identify response-associated immune programs with patient-level relevance.
Methods:
Single-cell RNA-seq and clinical annotations from GSE207422 were used to construct an NSCLC tumor microenvironment atlas and refine myeloid and T cell states. UCell scoring, sample-level summaries, pseudobulk differential analysis, Monocle2 trajectory inference, and targeted macrophage-T cell interaction analysis were applied. A posttreatment SPP1 inflammatory TAM-derived NR-TAM signature was projected to pretreatment bulk RNA-seq from the same cohort. INHBA was evaluated in pretreatment bulk data, TCGA-LUAD/TCGA-LUSC, and lung cancer cell line assays.
Results:
SPP1 inflammatory TAM was enriched in nonmajor pathologic response (NMPR) samples and co-occurred with exhausted CD8 T cell abundance, malignant cell hypoxia, and EMT programs. The NR-TAM signature was partially recapitulated in pretreatment bulk RNA-seq, with higher scores in NMPR but substantial group overlap. Exploratory ROC analysis showed modest discrimination for NR-TAM score (AUC = 0.60), INHBA (AUC = 0.64), and the combined model (AUC = 0.64; LOOCV AUC = 0.37). INHBA was higher in NMPR cases, correlated with residual tumor burden and selected TAM-related genes, showed less favorable survival trends particularly in LUSC, and its knockdown reduced proliferation and migration in lung cancer cells.
Conclusions:
An SPP1 inflammatory TAM program is associated with unfavorable pathologic response in NSCLC neoadjuvant immunotherapy. INHBA is a prioritized exploratory candidate within this program. Because the evidence is associative and predictive performance is modest, these markers require prospective validation before clinical use.