Integrative multi-omics and mendelian randomization reveal the critical role of pyroptosis in prognosis and therapy of lung squamous cell carcinomas

Abstract

Background Most types of programmed cell death (PCD) have been demonstrated to play critical roles in the pathogenesis and prognosis of lung squamous cell carcinoma (LUSC). However, the specific type of PCD with the most prominent driving effect and regulatory value in LUSC remains unclear. Methods Multi-omics data were integrated, and a multimodal autoencoder was employed to identify prognosis-related PCD gene modules and conduct weight ranking, leading to the construction of a PCD-associated neural network prognostic model. Summary-data-based Mendelian Randomization (SMR) analysis was applied to validate the causal relationship between key genes and LUSC. Combined with single-cell and spatial transcriptomics analyses, tumor–immune cell interactions were characterized, and the regulatory mechanism by which tumor cells modulate PCD in key immune cells through specific pathways was verified via cellular experiments. Results Twelve prognosis-related PCD gene modules were identified in LUSC, among which the Pyroptosis_4 module emerged as the core risk signature. The prognostic model based on Pyroptosis_4 enabled effective risk stratification of early-stage patients. SMR analysis confirmed that the key pyroptosis gene NOD1 was directly associated with LUSC susceptibility, and NOD1 expression in M2 macrophages regulated the tumor immune microenvironment. Tumor cells formed a close spatial network with NOD1- M2 macrophages and inhibited pyroptosis in M2 macrophages through the MDK/NCL signaling pathway. Conclusion Pyroptosis plays a crucial role in the prognosis of LUSC, and this effect is associated with differential NOD1 expression in M2 macrophages. Tumor cells and NOD1- M2 macrophages establish spatial interactions via the MDK/NCL pathway, emphasizing a potential candidate target for tumor immunoregulation-based strategies.