ABSTRACT Background The rising incidence and mortality of lung adenocarcinoma (LUAD) present a significant public health challenge. Histidine, an essential amino acid, plays a pivotal role in metabolic processes, yet its specific contribution to LUAD pathogenesis remains to be elucidated. Methods This study obtained bulk and single-cell RNA sequencing (scRNA-seq) data for LUAD from UCSC Xena and Code Ocean platforms, respectively. By integrating differential expression analysis, univariate/multivariate Cox analysis, and LASSO regression analysis, prognostic genes for LUAD were identified, and a prognostic risk model was constructed. Algorithms including ESTIMATE, ssGSEA, and CIBERSORT were employed to investigate immune heterogeneity across different groups. Furthermore, molecular subtypes of LUAD were identified through consensus clustering. Results This study, through the integration of bulk and scRNA-seq data, identified epithelial cells as the key effector cell population in LUAD, which can be further subdivided into four functionally heterogeneous subpopulations. Seven histidine metabolism-related epithelial cell-specific genes with prognostic significance in LUAD were identified (WIF1, GATA2, CD69, ID1, C4BPA, WFDC2, and CCL20), enabling the construction of a robust prognostic risk model. Immune infiltration analysis revealed that low-risk patients exhibited more robust immune infiltration and activity. Furthermore, cross-cancer exploratory evidence suggested potential sensitivity to CTLA-4 and PD-L1 inhibitors in this group. Furthermore, consensus clustering analysis successfully partitioned LUAD into two molecular subtypes exhibiting immune heterogeneity. Conclusion The prognostic model constructed based on epithelial cell-specific genes associated with histidine metabolism effectively distinguishes LUAD patients and their immune characteristics, revealing epithelial cells as a key cell population regulating LUAD histidine metabolism.

Comprehensive immunological profiling across diverse clinical and pathological subtypes of lung adenocarcinoma

Spatial and multi-omics transcriptomic dissects platinum resistance in lung adenocarcinoma: a five-gene predictive model with tumor microenvironment dynamics.

Harnessing retinoic acid metabolism-related genes to identify lung adenocarcinoma subtype and establish a risk model for predicting prognosis and drug therapy response.

Oxidative stress correlates with the development and prognosis of lung adenocarcinoma (LUAD). This study, on the basis of oxidative stress-related genes (OSRGs), commences to identify molecular subtypes and develop prognostic model for LUAD. LUAD samples were derived from the public database. OSRGs were acquired from GeneCards database. Molecular subtypes were classified by "ConsensusClusterPlus" package. Overall survival (OS) rate, clinical and immune infiltration features in different subtypes were compared. Differentially expressed genes (DEGs) were screened employing "limma" package. Thereafter, prognostic OSRGs signatures were identified via LASSO regression analysis. Further, we developed a RiskScore model and validated its predictive performance. Pathway enrichment analysis was carried out in different risk groups. Two molecular subtypes (Cluster1, Cluster2) of LUAD were classified with different survival outcomes, clinical features, and immune cell infiltration. Subsequently, 7-OSRGs prognostic signatures in LUAD were identified to establish RiskScore model, comprising TPSB2, CENPH, HIST1H1E, SULT2B1, CCL20, SERPINE1, and DKK1. High-risk group exhibited lower OS rate than low-risk group. The model exhibited robustness and was an independent indicator in predicting LUAD prognosis. Additionally, high-risk group was chiefly involved in cell function-relevant pathways, while low-risk group was chiefly implicated in immune-relevant pathways. This study distinguished two molecular subtypes and developed a prognostic RiskScore model linked to OSRGs in LUAD. However, these findings still require further verification through multiple prospective experiments. Taken together, our current research could offer some guidance for the precise stratification and treatment of LUAD.

Lung adenocarcinoma is classified into subtypes based on pathological features; however, the clinical and biological characteristics of each subtype, including driver mutations and programmed death-ligand 1 (PD-L1) expression, remain unclear. We aimed to clarify these characteristics. We retrospectively analyzed 1412 cases of stage I–III lung adenocarcinoma that underwent complete resection between 2004 and 2023. Clinical and biological characteristics were compared by predominant subtypes. Among the 1412 cases, the predominant subtypes were papillary (n = 686, 48.6%), lepidic (n = 317, 22.5%), acinar (n = 234, 16.6%), solid (n = 166, 11.8%), and micropapillary (n = 9, 0.6%). The lepidic subtype was more common in females (58.0% vs. 43.5%; p < 0.001), had a lower smoking rate (47.3% vs. 60.9%; p < 0.001), smaller invasive size (8 mm vs. 18 mm; p < 0.001), higher frequency of epidermal growth factor receptor (EGFR) mutation (63.4% vs. 45.2%; p < 0.001), and lower PD-L1 expression (15.5% vs. 45.2%; p < 0.001). The solid subtype was more prevalent in males (76.5% vs. 50.2%) and smokers (83.1% vs. 54.5%), with a larger invasive size (20.1 mm vs. 15.0 mm), fewer EGFR mutations (12.6% vs. 54.2%; p < 0.001), and higher PD-L1 expression (66.6% vs. 35.2%; p < 0.001). The 5-year Recurrence-free survival rates for the lepidic, acinar, papillary, and solid subtypes were 88.2, 72.9, 65.1, and 58.2%, respectively (p < 0.001). Lung adenocarcinoma subtypes present distinct clinical and pathological profiles, which may affect treatment strategies and prognostic evaluation.

The exon 19 deletion (19 Del) and the exon 21 L858R point mutation (21 L858R) are two main subtypes of EGFR-mutant lung adenocarcinoma (LUAD) with distinct response to targeted treatment and immunotherapy. Understanding the intratumor heterogeneity (ITH) of EGFR-mutant LUAD may explain the reason. 157 multi-region tumor samples and matched distant normal lung tissues from 29 treatment-naïve operable EGFR-mutant LUAD patients were collected to perform whole genome sequencing, panel sequencing and whole transcriptome sequencing. We aimed to comprehensively assess genomic and transcriptomic ITH between 19 Del and 21 L858R. The 21 L858R LUAD exhibited significantly higher copy number variation (CNV) ITH index (ITHi) compared to the 19 Del LUAD, but there was no significant difference in somatic single-nucleotide variant (SNV) ITHi between them. Meanwhile, 19 Del LUAD owned more clonal genetic alterations, while 21 L858R LUAD had more subclonal events. Both linear and branch evolution models existed in 19 Del and 21 L858R LUAD. Besides, 19 Del seemed to be more dominant for driving tumor development, while other driver mutations participated jointly with 21 L858R in tumor evolution. Moreover, 19 Del LUAD exhibited significantly higher immune score and checkpoint inhibition signature than 21 L858R. Additionally, it indicated that high-level TMB might be a favorable prognostic factor for EGFR-mutant LUAD. Our study demonstrated diverse genomic heterogeneity and tumor immune microenvironment in EGFR-mutant LUAD, which might elaborate on potential explanations for different efficacy between 19 Del and 21 L858R and provide valuable hints to treatment strategy.

The molecular landscape of lung adenocarcinoma (LUAD) is often illustrated as a driver-oncogene "pie chart," but identical mutations exhibit heterogeneous signaling shaped by co-mutations, transcriptional programs, and lineage context. We propose a lineage-integrated signaling framework using an EGFR mutation signature (mSig). We defined EGFR mSig using differentially expressed genes in EGFR-mutant (mt) LUADs. Semi-supervised clustering and machine learning models were used to test reproducibility in different combinations of datasets. We analyzed molecular subtypes, lineage markers, co-occurring mutations and EGFR copy number alterations in EGFR mSig-defined subtypes of LUAD. EGFR mSig showed robust classification performance (AUROC = 0.83-0.95; mean NPV = 96.3%). Validated gene expression subtypes and lung lineage markers were closely aligned with EGFR mSig status. Most EGFR mSig(+) tumors, including many without EGFR mutations belonged to Bronchioid subtype. A subset of canonical RAS mutations were mSig(+) and mirrored the EGFR mutation pattern. EGFR wild-type (WT)/mSig(-) tumors were enriched for non-Bronchioid subtypes and had co-mutations in TP53 or RAS/RAF/RTKs. We highlighted a parsimonious collection of coordinated mutations identified including RAS, KEAP1, STK11, TP53, and CDKN2A, supportive of prior reports. A novel EGFR mSig that captures the transcriptional footprint of EGFR activation revealed a subset of EGFR WT LUADs with "mt-like" features. mSig refines LUAD taxonomy beyond mutation-only pie-chart models by incorporating lineage and co-mutation context. Lineage-directed stratification with co-alteration identifies clinically relevant groups across EGFR and RAS states and highlights new treatment opportunities for patients currently considered "oncogene-negative." NCI U01CA272541, R01CA262296, U24CA264021, UG1CA233333, R01CA211939.

Effective diagnosis and treatment of lung adenocarcinoma depends on accurate typing, subtyping, and grading. Herein, we present the CLWD dataset, a valuable resource for the lung cancer pathology community, comprising 408 whole-slide images (WSIs) from 210 patients specifically curated for the study of lung adenocarcinoma subtypes. Scanned at 80 × magnification, it is one of the largest datasets in Asia, with a particular emphasis on Chinese patient demographics. Notably, the dataset includes comprehensive clinical information, such as age, sex, and diagnosis, providing a robust foundation for diverse research needs. Publicly accessible, it supports a range of applications, including machine learning model development and validation. An initial evaluation of lung adenocarcinoma subtype classification using a multi-instance learning framework demonstrated that this dataset can substantially advance global research and improve the accuracy of subtype diagnosis.

Lung adenocarcinoma (LUAD) is a malignancy with a high global incidence and cancer-related mortality rate. Over decades of development, the treatment of lung cancer has evolved from empirical approaches such as traditional chemotherapy and radiotherapy to a precision model that integrates targeted therapy, immunotherapy, and combination treatments. Through "molecular profiling and individualized treatment planning", targeted therapies focusing on biomarkers like EGFR and ALK, along with immunotherapy using programmed cell death protein 1 (PD-1)/programmed death ligand 1 (PD-L1) inhibitors, have become landmark achievements in precision medicine for lung cancer. Although various clinical trials have improved the prognosis of LUAD patients, their 5-year survival rate remains low, and precision therapy for lung cancer still faces multiple challenges. This study aims to improve the prognosis of LUAD patients through molecular subtype-based precision treatment. LUAD RNA-sequencing data sourced from an online database were used to screen for differentially expressed genes (DEGs). Weighted gene coexpression network analysis combined with univariate and multifactorial Cox analysis was used to identify hub prognostic genes. Based on these genes, partitioning around medoids clustering was applied to classify LUAD into two subtypes. The estimation of stromal and immune cells in malignant tumor via using expression data, immunophenoscore, and microenvironment cell populations counter algorithm was used to determine the microenvironmental purity and immune response of the two subtypes. Gene set enrichment analysis was performed to analyze the biological function. The correlation between hub gene and EGFR mutations was detected in clinical samples via immunohistochemical (IHC) staining. This study delineated two distinct subtypes of LUAD, and the survival rate for patients in cluster 2 was found to be significantly superior to that of cluster 1. Additionally, patients in cluster 2 had greater immune cell infiltration, a greater microenvironmental component, and a higher rate of EGFR mutation. In contrast, patients in cluster 1 exhibited a higher degree of fibroblast infiltration and a notable prevalence of NTRK3 mutations, as observed in the study of the tumor microenvironment. In addition, functional analysis suggested cluster 1 was associated with nucleotide sequence repair, while cluster 2 was mainly related to lipid metabolism and angiogenic pathways. IHC staining revealed that the expression level of BIRC5 was notably downregulated in early-stage patients with EGFR-mutant LUAD. Furthermore, in the advanced stage, the expression level of BIRC5 in the tissues of these patients was significantly higher compared to those in patients with the EGFR wild type. Patients in cluster 1 may benefit from anti-nucleotide repair therapies such as platinum therapy, radiotherapy, targeting of fibroblasts, and targeting of NTRK3, while patients in cluster 2 may benefit from immunotherapy, antiangiogenic therapy, targeting of lipid metabolism, and targeting of EGFR. This study may offer novel insights into improving the overall prognosis of patients with LUAD by leveraging molecular subtype-based precision therapy, as demonstrated by recent advancements in the identification of prognostic biomarkers and therapeutic targets.

Primary pulmonary enteric adenocarcinoma: A single-institute retrospective study on its imaging classification and survival outcomes.

Molecular subtypes of lung adenocarcinoma (LUAD) with varying prognosis and characteristics have been proposed based on one or two-dimensional studies but are not yet implemented into clinical routine. Epigenetic modifications in cancer cells are independent of sequence variants, directly linked to gene and genome regulation, and thus provide important information to guide subclassification efforts. We performed in-depth epigenomic profiling of 95 primary LUAD samples from a Swedish discovery cohort with comprehensive clinicopathological, epigenomic, genomic, transcriptomic, proteomic, and metabolomic data. Additionally, we estimated pure tumor cell methylomes using a computational approach. We subdivided the discovery cohort into four epigenetic subtypes, the epitypes, reflecting distinct tumor cell methylation states. Resulting epitypes were contrasted based on clinicopathological and molecular features, and our main findings were validated in two additional primary tumor cohorts totaling over 700 samples. Of the four DNA methylation epitypes, M1-M4, M1 and M4 were associated with the previously proposed mRNA subtypes Terminal Respiratory Unit and Proximal Proliferative, respectively. Epitypes M2 and M3 showed similar mRNA/protein subtype composition but differed with respect to e.g., higher expression of the LUAD histology-associated NAPSA/surfactant metabolism expression metagene in M3. Genes included in this metagene showed lower DNA methylation in M3, counter to a global tendency towards promoter hypermethylation in this epitype. To further delineate tumor intrinsic links between the epigenomic and expression phenotypes, 62 LUAD cell lines classified into the four epitypes were investigated and recapitulated several characteristics from the tumor epitypes, such as methylation and expression pattens of NAPSA/surfactant genes, highlighting epigenetic states as likely drivers or maintainers of broad tumor phenotypes and differentiation states. Dissecting LUAD based on combined biological characteristics using multiomics data has deepened our understanding of the heterogeneity in this complex disease and the mechanisms underlying phenotype formation and maintenance. There remains a critical need for large, publicly accessible, well-annotated multiomic LUAD cohorts to support rigorous subtype discovery and validation, particularly those linked to targeted therapy trial outcomes.

The impact of KRAS mutation subtypes on treatment response to first-line immunochemotherapy in advanced lung adenocarcinoma (LUAD) remains uncertain. This study evaluated treatment efficacy across KRAS subtypes and examined the role of programmed death-ligand 1 (PD-L1) expression and co-mutations. We retrospectively analyzed 335 patients with advanced KRAS-mutant LUAD treated with first-line immunochemotherapy between 2018 and 2022 at two centers. Patients were categorized into G12A (n = 36), G12C (n = 116), G12D (n = 62), G12V (n = 56), and other subtypes (n = 65). PD-L1 tumor proportion score (TPS) was stratified as <1%, 1-49%, or ≥50%. Endpoints included progression-free survival (PFS), objective response rate (ORR), and disease control rate (DCR). Median PFS in the overall cohort was 8.6 months, with an ORR of 34.0% and a DCR of 87.8%. Median PFS did not differ significantly among KRAS subtypes (p = .617), nor within PD-L1 TPS groups: <1% (p = .740), 1-49% (p = .652), and ≥50% (p = .481). In the major subtypes (G12A, G12C, G12D, and G12V), PD-L1 expression showed no significant association with PFS. STK11 co-mutations were enriched in G12C, G12V, and other subtypes (p = .004) and correlated with shorter PFS (p = .006). In conclusion, first-line immunochemotherapy yields comparable efficacy across KRAS subtypes, independent of PD-L1 expression. Within the major subgroups (G12A, G12C, G12D, and G12V), PD-L1 levels were not predictive of PFS. STK11 co-mutations were enriched in G12C, G12V, and other subtypes and were associated with shorter PFS.

Preoperative differentiation of precursor glandular lesions (PGL), minimally invasive (MIA), and invasive adenocarcinoma (IAC) in stage IA lung adenocarcinoma (LUAD) is critical for surgical planning but remains challenging due to overlapping CT features and interobserver variability. While existing artificial intelligence (AI) models focus predominantly on binary classification with limited multicenter validation, this study developed and validated a ternary classification framework using pretrained TabPFN and traditional machine learning (ML) algorithms based on AI-derived histogram features, benchmarking against intraoperative frozen section analysis. This multicenter retrospective study utilized preoperative CT scans from three institutions between September 2014 and October 2023. Data were divided into training, internal validation, and external test sets. Histogram features (n =26) were automatically extracted using a commercial AI system (InferRead CT Lung). TabPFN and five ML algorithms were trained with selected clinical and histogram features. Performance was evaluated by accuracy, macro-AUC, sensitivity, specificity, and Cohen's Kappa. Statistical comparisons included DeLong tests for AUC and chi-square for categorical variables. The cohort comprised 584 stage IA LUAD patients (mean age 57.9±11.0years; 386 female), divided into training/validation sets (n =412, center 1) and external test sets (n =114, center 2; n =58, center 3). TabPFN achieved macro-AUC of 0.781-0.911 and accuracy of 67.2-78.9% across external test sets, outperforming other ML algorithms. Of note, TabPFN achieved an overall better prediction accuracy compared to frozen section analysis on all test sets (internal: 92.3% vs 84.6%, P =0.503; external 1: 87.5% vs 75%, P =1.000; external 2: 67.2% vs 43.1%, P <0.001). Subgroup analysis revealed superior performance for mGGN lesions (85%) on both external test sets. TabPFN enables robust, generalizable ternary classification of LUAD subtypes, surpassing conventional ML and frozen section analysis. Its integration with automated histogram analysis offers a scalable solution for preoperative stratification of early-stage lung cancer.

Deep learning-based classification of lung adenocarcinoma subtypes in histopathological images using DS-EffNet.

BackgroundLung cancer, characterized by its high global incidence and mortality rates, necessitates comprehensive and precise stratification strategies to guide the diverse diagnostic approaches and therapeutic agents in clinical decision-making.ObjectivesShox2 and Rassf1a promoter methylation are established biomarkers for the early screening of lung cancer. This study comprehensively investigated the clinical utility of Shox2 and Rassf1a promoter methylation alongside driver mutations for molecular subtyping and stratification in 1027 lung adenocarcinoma (LUAD) patients.MethodsThis study included a cohort of 1027 LUAD patients who received treatment at the First Affiliated Hospital of Chongqing Medical University between January 2020 and August 2024. Comprehensive demographic and clinicopathological data were collected. Shox2 and Rassf1a methylation was quantified using the Lungme kit, while 10 driver mutations were detected by PCR assay. Chi-square tests were used to assess correlations between methylation status and clinicopathological characteristics; ROC analysis evaluated diagnostic performance for distinguishing LUAD subtypes. Multiple regression identified stage-associated hazardous factors.ResultsIn our cohort, Shox2 and Rassf1a methylation were correlated with more aggressive clinicopathological characteristics (age, sex, smoking, drinking, TNM stage and histological progression) and exhibited significant diagnostic potential for distinguishing early-stage lesions (adenocarcinoma in situ from LUAD across stages I-IV) and histological progression (minimally invasive adenocarcinoma versus invasive adenocarcinoma). Shox2 methylation exhibited significant co-occurrence with mutations of KRAS (p < 0.001) and MET (p = 0.02) and mutual exclusivity with mutations of EGFR (p < 0.001), RET (p = 0.013) and HER2 (p = 0.03). Rassf1a methylation showed no significant associations with these driver mutations. Combining Shox2 and Rassf1a methylation with EGFR mutations demonstrated enhanced discriminative capacity for early-stage lesions.ConclusionsOur study demonstrated that comprehensive analysis of methylation and gene mutations could provide a novel clinical strategy for molecular subtyping and precision medicine in LUAD.

Tumor histology reflects disease aggressiveness and clinical outcomes in cancer patients. Lung adenocarcinomas (LUADs) are classified based on predominant histologic patterns, including high-grade micropapillary and solid subtypes which portend unfavorable clinical features and prognosis. However, the cellular and molecular characteristics underlying these histologic subtypes remain largely unknown. We used scRNA-seq to profile 117,266 cells from 18 treatment-naïve LUADs with heterogeneous histologic patterns and also performed spatial transcriptomic analysis (10x Visium) for representative cases. By integrating single-cell transcriptomics with spatial information, we aimed to characterize the cellular identity and spatial organization driving LUAD heterogeneity. We demonstrated that histologic subtypes can be distinguished by subtype-specific cancer cell subpopulations and immunosuppressive phenotypes in the tumor microenvironment (TME). Our data reveal how intercellular interactions among cancer cells, macrophages, and CD8+ T cells in the prognostically unfavorable solid subtype are associated with cancer cell plasticity and promote an immunosuppressive TME. Additionally, we identify HMGA1 as a potential clinically relevant biomarker and therapeutic target for the solid subtype LUAD. These findings deepen our understanding of the histologic heterogeneity of LUAD and may facilitate the development of subtype-specific biomarkers and targeted therapeutic strategies.

BackgroundThis study aims to define a novel molecular subtype of LUAD by integrating multiple omics data. Additionally, we develop and validate an Artificial Intelligence Derived Prognostic Index (AIDPI) that predicts the prognosis of LUAD patients, identifies potential therapeutic targets.MethodsThis study employed ten clustering algorithms from the R package “MOVICS” to integrate multi-omics data of LUAD sourced from TCGA database for molecular typing. Subsequently, an Artificial Intelligence Derived Prognostic Index (AIDPI) was constructed as the most effective indicator for predicting the overall survival rate of LUAD patients. The biological functions and mechanisms of NPC2 in lung adenocarcinoma were elucidated through both in vitro and in vivo experiments, which included CCK-8 assays, colony formation assays, flow cytometry, Transwell assays, and xenograft tumor models. Additionally, the impact of NPC2 on Ribociclib sensitivity was investigated through drug correlation analysis and molecular docking, while the predictive value of NPC2 regarding immunotherapy benefits was validated using the immune cell infiltration analysis.ResultsThrough multi-omics clustering, we identified two subtypes of lung adenocarcinoma associated with prognosis, with the CS1 subtype exhibiting the most favorable prognostic outcomes. The low AIDPI group exhibited a more positive prognosis, accompanied by increased immune cell infiltration and activation of immune pathways. Meanwhile, NPC2 was recognized as a standalone risk factor for LUAD, with its high expression significantly improving the overall survival of LUAD patients. Functionally, the overexpression of NPC2 promotes tumorigenesis in LUAD both in vitro and in vivo. Mechanistically, the upregulation of NPC2 expression inhibits the progression of LUAD by suppressing the PI3K/AKT signaling pathway. Our study also demonstrated that high NPC2 expression is positively correlated with Ribociclib sensitivity, as confirmed by in vitro experiments. Furthermore, NPC2 expression is positively correlated with ImmuneScore, and may serve as a predictive indicator for the efficacy of immune checkpoint inhibitor (ICI) therapy.ConclusionThe comprehensive analysis of multiple omics data significantly enhances the molecular classification of lung adenocarcinoma. Furthermore, AIDPI is a potential biomarker that predicts the prognosis of LUAD patients. NPC2 inhibits the progression of LUAD by suppressing the PI3K/AKT signaling pathway and enhancing the chemotherapy sensitivity to Ribociclib.

BackgroundTo investigate the molecular genetic features of multiple primary lung cancer (MPLC) to provide a basis and new methods for its identification, diagnosis, and treatment.MethodsTranscriptome sequencing (RNA-seq) was performed on 16 tissue samples from eight patients with synchronous multiple primary lung adenocarcinoma (sMP-LUAD) and eight tissue samples from eight patients with single primary lung adenocarcinoma (SP-LUAD). Differentially expressed genes selected by bioinformatic methods were validated in 24 sets of sMP-LUAD and SP-LUAD samples using quantitative reverse transcription polymerase chain reaction (qRT-PCR). Based on The Cancer Genome Atlas (TCGA) database, the differentially expressed genes responsible for the biological behavior of lung adenocarcinoma and their clinical significance were analyzed.ResultsOverall, 194 differentially expressed genes were identified (P < 0.05), including 22 up-regulated and 172 down-regulated genes. Two up-regulated (DUOX1 and CACNA2D2) and three down-regulated (GPX8, COL1A2, and COL1A1) genes were selected for validation by qRT-PCR analysis. The qRT-PCR results showed that the expression of DUOX1 mRNA in the sMP-LUAD group was significantly higher (P < 0.05) than that in the SP-LUAD group; mRNA CACNA2D2, GPX8, COL1A2, and COL1A1 expression in the sMP-LUAD group was not statistically different from that in the SP-LUAD group (P > 0.05). Gene ontology (GO) enrichment analysis showed that DUOX1 mRNA was mainly enriched in the entries of positive regulation of wound healing and oxidation-reduction processes. Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis showed that DUOX1 can promote reactive oxygen species (ROS) production and be related to thyroid hormone production. Furthermore, based on the TCGA database, we analyzed the biological behavior and clinical significance of DUOX1 in lung adenocarcinoma using bioinformatics technology. DUOX1 mRNA expression was decreased in all stages and pathological subtypes of lung adenocarcinoma (P < 0.05). Immune infiltration analysis showed that DUOX1 with mast cells and eosinophils was positively correlated (P < 0.05), and Th2 cells were negatively correlated (P < 0.05). Logistic regression analysis showed that the expression of DUOX1 mRNA was significantly correlated with the patient’s age, lymph node metastasis, and pathologic stage (P < 0.05). Kaplan–Meier survival plots showed that low DUOX1 expression was not significantly correlated with OS, DSS, and PFI (P > 0.05). Univariate and multivariate COX regression analysis revealed that DUOX1 mRNA could not be used as an independent factor for predicting prognosis (P > 0.05). Therefore, we developed a predictive nomogram model combining clinicopathological variables and DUOX1 mRNA to predict the survival of patients with lung adenocarcinoma.

Lung adenocarcinoma (LUAD), the most common subtype of nonsmall cell lung cancer, exhibits substantial molecular heterogeneity, complicating subtype classification, progression assessment, and treatment decision-making. Advances in high-throughput sequencing enable multi-omics analysis to reveal cancer mechanisms and biomarkers, yet the high dimensionality, heterogeneity, and interrelationships of omics layers such as transcriptome, microRNA expression, methylome, and copy number variation remain challenging to integrate through conventional methods. Most existing graph-based approaches represent patients as nodes, obscuring gene-level regulatory dynamics and limiting biological interpretability. To address this, we propose the Multi-omics Hierarchical Graph Neural Network (MoAGNN), a novel architecture that represents genes as nodes, integrates four omics, and leverages graph convolution with self-attention–based graph pooling to identify informative molecular nodes, thereby enhancing predictive performance and interpretability for LUAD subtype classification, tumor staging, and prognosis prediction. Multi-omics datasets from The Cancer Genome Atlas (TCGA) were used and results showed that MoAGNN achieved a test accuracy of 0.89 for LUAD subtype classification, outperforming conventional models (Random Forest, Support Vector Machine and Multi-Layer Perceptron) as well as state-of-the-art graph-based models MoGCN, a multi-omics integration model based on graph convolutional network, and MOGLAM, an end-to-end interpretable multi-omics integration method. Furthermore, we validated the generalizability of this framework on the GSE81089 dataset, demonstrating its potential applicability to clinically relevant risk assessment. Subsequent functional enrichment and survival analyses validated the biological relevance of the key genes identified by MoAGNN, supporting their potential roles in LUAD progression, and suggesting the broader applicability of this framework in multi-omics cancer research.