NSCLC represents a critical diagnostic challenge. Non-small cell lung carcinoma accounts for roughly 85% of all lung malignancies and carries a poor prognosis largely because most patients present with advanced-stage disease. Despite advances in targeted therapies and immunotherapy, early diagnostic markers and precise treatment targets remain urgently needed.
LPS metabolism drives inflammation and tumor progression. Lipopolysaccharide (LPS) metabolism regulates immune responses and the tumor microenvironment, influencing tumor immune escape and inflammatory signaling. LPS enhances tumor cell invasiveness and metastasis by upregulating oncogenes and triggering local inflammation, particularly in NSCLC, partly through activation of hypoxia-inducible factor 1-alpha and the TLR4-NF-kB pathway.
Circadian disruption removes a critical tumor suppression mechanism. Circadian rhythms normally govern cell cycle checkpoints, DNA damage response, and metabolic homeostasis. Disruption of these rhythms eliminates the temporal oversight that keeps inflammatory responses transient and localized. In NSCLC patients, loss of circadian control shifts the tumor microenvironment toward a constitutive pro-inflammatory, pro-tumorigenic state characterized by persistent cytokine secretion and facilitated immune evasion.
A two-hit model links LPS signaling and circadian disruption. The functional synergy between LPS-mediated TLR4 inflammatory signaling and circadian clock disruption represents a critical two-hit model in NSCLC development. Circadian gating normally constrains LPS-TLR4 inflammatory responses to brief, localized flares. When circadian rhythm is disrupted in NSCLC, this control is lost, creating persistent cytokine secretion that drives malignant progression.
Eight GEO datasets formed the analysis foundation. Gene expression data were extracted from eight datasets covering NSCLC, lung adenocarcinoma, and lung squamous cell carcinoma in the Gene Expression Omnibus (GEO) repository. Six datasets (GSE103512, GSE116959, GSE140797, GSE101929, GSE40791, GSE49644) served as the training set, while GSE54495 and GSE18842 were reserved as independent testing datasets. Batch effects were corrected using the ComBat algorithm.
LPS and circadian gene libraries guided the search. From the published literature, 6,571 LPS-related genes (LRGs) and 2,091 circadian rhythm-related genes (CRGs) were compiled as the candidate gene pools. Weighted Gene Co-expression Network Analysis (WGCNA) was then used to identify disease-relevant co-expression modules from the 889 significant differentially expressed genes identified in the training data.
Ten machine learning algorithms were benchmarked. To establish the optimal diagnostic model, ten algorithms including Lasso regression, Random Forest, XGBoost, Elastic Net, Partial Least Squares Regression (plsRglm), Naive Bayes, Support Vector Machine, and others were systematically evaluated under ten-fold cross-validation. The plsRglm algorithm was selected as the best diagnostic model because it extracts orthogonal latent components that maximize covariance between gene expression and the NSCLC phenotype, handling the high multicollinearity of transcriptomic data better than RF or SVM.
Hub genes were identified through Lasso-RF intersection. From the 59 genes at the intersection of the WGCNA disease-associated module with LRGs and CRGs, Lasso regression identified 20 key candidate genes, and Random Forest ranked the top 20 by importance. The intersection of both lists yielded nine hub genes: CACNA2D2, ASPA, LRRN3, ABCA6, TNFSF12, AHNAK, TACC1, ID4, and TSLP. Cellular experiments in A549 lung cancer cells validated the functional role of the lead hub gene CACNA2D2.
DEG analysis identified 889 significantly dysregulated genes. Differential expression analysis between NSCLC and healthy control samples identified 889 significant DEGs, including 271 upregulated and 618 downregulated genes. Functional enrichment revealed strong enrichment in inflammatory pathways including IL-17, Wnt, and PI3K-Akt signaling, as well as in fatty acid metabolism and cell cycle regulation, consistent with the LPS-circadian disruption hypothesis.
The magenta WGCNA module was most strongly linked to NSCLC. Module-trait correlation analysis identified the magenta module (MEmagenta) as the most strongly associated with the NSCLC phenotype (R = 0.71, P = 1e-65). The 542 genes in this module were enriched in angiogenesis, vascular development, extracellular matrix organization, and TGF-beta signaling, implicating them in tumor metastasis and proliferation. Fifty-nine of these genes intersected with LRGs and CRGs.
All nine hub genes showed excellent diagnostic AUC values. ROC analysis in the training dataset confirmed that all nine hub genes achieved AUC values ranging from 0.832 to 0.906. ASPA and LRRN3 showed the highest individual AUC of 0.906. In independent validation datasets, CACNA2D2, TNFSF12, ID4, LRRN3, AHNAK, and ABCA6 achieved AUC values close to or reaching 1.000, confirming highly consistent discriminative performance across external cohorts.
XGBoost model achieved near-perfect training performance. Consensus clustering stratified NSCLC samples into two distinct molecular subtypes (k=2). An XGBoost classification model trained on the nine hub genes achieved AUC of 1.000 on the training set and 0.914 on the test set. SHAP interpretability analysis identified TSLP, ID4, and TACC1 as the three most influential features driving model predictions, and decision curve analysis confirmed the model's clinical utility across threshold probabilities.
NSCLC displays a characteristic immunosuppressive microenvironment. CIBERSORT immune cell quantification revealed significantly elevated infiltration of M0 and M1 macrophages and activated dendritic cells in NSCLC compared to controls, while resting CD4+ memory T cells and resting mast cells were substantially depleted. This pattern reflects a dominant immunosuppressive state consistent with tumor-associated macrophage-mediated immune evasion.
Each hub gene has a distinct immune correlation signature. All nine hub genes showed statistically significant correlations with immune cell infiltration patterns (P less than 0.001). ABCA6 positively correlated with resting CD4+ memory T cells and monocytes while inversely correlating with M0/M1 macrophages. CACNA2D2 positively associated with monocytes and resting mast cells but inversely with M1 macrophages and activated CD4+ T cells. These distinct immunomodulatory profiles suggest the genes collectively shape NSCLC's immune landscape.
Key genes connect T-cell regulation to tumor immune evasion. LRRN3 and AHNAK were identified as involved in T-cell receptor signaling and lymphocyte activation, with their downregulation consistent with T-cell exhaustion and reduced CD4+ memory T cell infiltration observed in NSCLC. TNFSF12 (TWEAK) modulates macrophage recruitment and pro-inflammatory cytokine production, while TSLP bridges innate and adaptive immunity and is dysregulated in the lung to promote pro-tumorigenic inflammation.
A nomogram integrated eight validated hub genes. A nomogram predicting NSCLC risk was built from eight genes consistently validated across external datasets (CACNA2D2, ASPA, LRRN3, ABCA6, TNFSF12, AHNAK, TACC1, ID4). Calibration plots confirmed that predicted probabilities aligned reasonably well with actual outcomes. CACNA2D2, TNFSF12, and ID4 contributed the most to risk prediction, consistent with their biological roles at key regulatory nodes.
CACNA2D2 sits at the intersection of LPS metabolism and circadian control. CACNA2D2 encodes an auxiliary subunit of voltage-gated calcium channels and was uniquely categorized as both an LPS-related gene and a circadian rhythm-related gene, suggesting a dual regulatory role. As a master regulator of Ca2+ homeostasis, CACNA2D2 provides a mechanistic link between calcium signaling, circadian oscillation maintenance, and LPS-induced inflammatory buffering, making it the most biologically compelling hub gene in this study.
Knockdown and overexpression experiments confirmed tumor suppressor function. In A549 NSCLC cells, siRNA-mediated knockdown of CACNA2D2 increased cell viability, clonogenicity, migration, and invasion. Conversely, CACNA2D2 overexpression reduced all these oncogenic phenotypes. Live/dead staining, CCK-8 viability assay, colony formation assay, Transwell invasion assay, and scratch migration assay collectively confirmed that CACNA2D2 constrains fundamental hallmarks of cancer progression across multiple independent readouts.
Low CACNA2D2 expression predicts worse overall survival in LUAD. Survival analysis using the GEPIA2 database in the TCGA-LUAD cohort showed that patients with low CACNA2D2 expression had significantly shorter overall survival compared to those with high expression (Log-rank P = 0.022). This prognostic association reinforces that CACNA2D2 loss has real consequences for patient longevity beyond its diagnostic utility.
Downregulation creates a dual regulatory failure. The authors proposed that marked downregulation of CACNA2D2 in NSCLC represents a dual regulatory failure: collapse of circadian-mediated calcium rhythmicity and a compromised cellular capacity to buffer LPS-induced inflammatory stress. This synergistic disruption culminates in enhanced tumor aggressiveness, providing a mechanistic explanation for why CACNA2D2 loss is such a consistent feature of NSCLC and such a strong predictor of poor outcomes.
First systematic linkage of LPS-circadian axis to NSCLC immune landscape. This study was among the first to systematically link LPS metabolism and circadian rhythm gene disruption to the tumor immune microenvironment in NSCLC, providing a new mechanistic framework for understanding immune escape in this cancer. Prior studies had addressed LPS or circadian rhythms separately, but the integrated two-hit model is a novel contribution.
The nine-gene signature offers actionable diagnostic value. The identified hub genes (CACNA2D2, ASPA, LRRN3, ABCA6, TNFSF12, AHNAK, TACC1, ID4, TSLP) demonstrated consistent diagnostic performance across independent validation datasets with AUC values up to 0.906 and near-1.000 for several genes. The eight-gene nomogram provides a practical clinical tool for quantifying NSCLC risk from gene expression data.
Limitations point toward future research directions. The analysis relied on public datasets, which carry inherent batch effects and sample heterogeneity limitations despite ComBat correction. The study focused on gene expression without addressing epigenetic regulation or protein modification mechanisms. In vivo validation and larger prospective clinical studies are needed to fully establish the therapeutic relevance of CACNA2D2 and the other hub genes before clinical translation.