Integrated multi-omic analysis unravels the characteristics of the metabolism-related intratumoral microbes and establishes a novel signature for predicting prognosis and therapeutic response in lung adenocarcinoma

Introduction

Lung adenocarcinoma (LUAD), a prevalent and lethal malignancy, accounts for over 10% of cancer-related mortality worldwide. Most patients are diagnosed with LUAD at an intermediate or advanced stage, and the 5-year survival rate for LUAD patients in advanced stages is less than 20% (1). Conventional treatment modalities for LUAD, including surgery, radiotherapy, and chemotherapy, pose a higher risk of recurrence, metastasis, and drug resistance development. In recent years, therapeutic strategies for tumors have shifted towards recognizing the pivotal role of aberrant molecular regulation in tumor cell growth. Despite demonstrating significant therapeutic efficacy for a minority proportion of LUAD patients, the application of emerging immunotherapy remains limited due to tumor heterogeneity (2,3). Therefore, revealing the molecular characteristics and exploring novel biomarkers hold immense significance in LUAD treatment.

The microbial community plays a crucial role in maintaining the homeostasis of human life. In the past decade, advancements in high-throughput sequencing technology have enhanced our understanding of the intricate relationship between microbiota and hosts, specifically focusing on intestinal microbes (4). However, there is a significant knowledge gap regarding the interaction between other microbial populations and tumors. The intratumoral microbiota refers to the microbial community present within tumors, which actively shape the tumor microenvironment. Accumulating evidence suggests that intratumoral microbes exert diverse effects on tumor biological behaviors, including host metabolism, immune regulation, mutagenesis, and drug resistance (5-7). Although an increasing number of studies have reported the connection between microbial dysbiosis and lung cancer development, the underlying mechanism remains elusive.

Metabolic intermediates are recognized as crucial regulators of cellular signaling and epigenetic modifications. Tumor-associated metabolites can function as signal transduction molecules, influencing the activity of tumor signaling pathways and modulating tumorigenesis through direct intervention, transcriptional regulation, protein binding, and other mechanisms (8,9). Dysregulation of metabolism represents a crucial hallmark of tumor cells. Tumor cells undergo metabolic reprogramming to meet the demands for energy, materials, and redox power required for their rapid proliferation. The metabolic pathways involving glucose, glutamine, lipids, and one-carbon play essential roles in tumor metabolism. Metabolic disorders in these pathways can modulate the cellular microenvironment exerting a pivotal influence on tumorigenesis and progression (10-13). Collectively, the relationship between intratumoral microbes and cancerous metabolism is mutually interdependent and reciprocally influential. Elucidating this association is crucial for gaining profound insights into the mechanisms underlying tumor development and offers novel perspectives for LUAD prevention and treatment.

As the significant association between metabolism-related intratumoral microbes (MRIMs) and tumor occurrence and development continues to be elucidated, it highlights the considerable potential of MRIMs as a novel biomarker and therapeutic target. However, research in this field remains limited and is largely at an early stage, with reliance on conventional methodologies. Traditional 16S rRNA sequencing straggles in poor specificity in identification of microbial taxa, while metagenomic approaches are susceptible to host DNA contamination, which compromises detection accuracy. Molecular detection techniques such as fluorescence in situ hybridization (FISH) suffer from limited sensitivity due to the low abundance of intratumoral microbiota (14,15). While conventional methods often exhibit shortcomings in sensitivity, resolution, and functional characterization, emerging integrated multi-omics approaches, including single-cell omics, RNA transcriptomics, and combined microbiome analyses—offer potential resolution in deciphering the functional, spatial, and dynamic roles of MRIMs. This paradigm shift is critical for the discovery of prognostic biomarkers and the development of microbiome-targeted cancer therapies. However, relevant research in this area remains scarce, particularly in LUAD. In this study, we integrated single-cell and bulk transcriptomic data with microbiome profiles to investigate the association between MRIMs and the LUAD tumor microenvironment. In this study, we identified MRIMs in LUAD, analyzed their distinctive subtype characteristics, revealed the regulatory network of MRIMs-featured cells, and developed a prognostic signature for LUAD. Furthermore, we assessed the impact of microbial abundance of MRIMs on the immune microenvironment and evaluated the potential of microbial-based signatures in predicting immunotherapeutic benefit as well as drug response to chemotherapy and targeted therapy treatments. These findings provide a systematic overview of metabolically relevant intratumoral microbes in LUAD that may offer valuable insights for therapeutic approaches in patients with LUAD. We present this article in accordance with the TRIPOD reporting checklist (available at https://tcr.amegroups.com/article/view/10.21037/tcr-2025-357/rc).

Methods

The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments.

Data acquisition

The transcriptomic data of 561 LUAD samples were obtained from the Genomic Data Commons (GDC) data portal (https://gdc.cancer.gov/access-data/gdc-data-portal), along with comprehensive clinical information and somatic mutation data, including age, sex, survival status, follow-up time, smoking history, tissue origin, pathological stage (16). The corresponding intratumoral microbial abundance profile was retrieved from the online archive (http://ftp.microbio.me/pub/cancer_microbiome_analysis) (17). Furthermore, the single-cell RNA sequencing data were acquired from the Gene Expression Omnibus (GEO, https://www.ncbi.nlm.nih.gov/gds) repository dataset GSE149655 comprising two paired LUAD and normal tissues (18). A total of 11,699 single cells were selected for subsequent analysis. The gene sets relevant to metabolism in the Kyoto Encyclopedia of Genes and Genomes (KEGG, https://www.kegg.jp/kegg/pathway.html) database served as a reference (19). The overview of this study is illustrated in Figure 1.

Figure 1 Overview of this study. LUAD, lung adenocarcinoma; MRIMs, metabolism-related intratumoral microbes; NES, normalized enrichment score; UMAP, Uniform Manifold Approximation and Projection.

Reverse transcription-polymerase chain reaction (RT-PCR) assessment

The expression levels of BCL3, KLF3, and NFKB2 in LUAD cells were assessed using the RT-PCR method. The primer sequences used are listed in Table S1. Total RNA was extracted from BEAS-2B, A549, and PC-9 cell lines using TRizol reagent (TIANGEN, Beijing, China). The isolated RNA was then reverse transcribed into cDNA following the manufacturer’s instructions (Thermo, Waltham, USA). For PCR amplification, a 20 µL reaction mixture was prepared containing 10 µL of 2× PerfectStart Green qPCR SuperMix (TransGen, AQ601-04, Beijing, China), 2 µL of diluted cDNA in nuclease-free water, and 1 µL of 10 µM primer solution, with the remaining volume made up of nuclease-free water. Amplification and detection were performed using an ABI Q1 PCR system. GAPDH was employed as the internal reference gene. The relative expression levels of the target genes were calculated using the 2^-ΔΔCt method.

Identification of MRIMs and MRIMs-derived subtypes in LUAD

The metabolic activity of LUAD samples was assessed using gene set variation analysis (GSVA) (20). The correlation between intratumoral microbiota and metabolic activity was determined using Spearman’s coefficient, with a threshold of |Corr| >0.3 and P<0.05 set as statistically significant (21). Subsequently, unsupervised clustering analysis was conducted to classify subtypes based on the microbial abundance of MRIMs (22). A heatmap was generated to demonstrate variations in microbial abundance, metabolic activity, and clinical characteristics among LUAD subtypes associated with MRIMs (23). The prognostic difference between the MRIMs-derived subtypes was compared using Kaplan-Meier survival analysis. The “maftools” package was employed to visualize the landscape of genomic alterations in populations of these subtypes (24). The correlation between MRIM subtypes and the immune microenvironment was analyzed using the ESTIMATE and CIBERSORT algorithms implemented in the R package Immunedeconv (25).

Investigation of MRIMs features at the single cell level

The Scissor algorithm identifies cell subpopulations in single-cell data that have the highest correlation with phenotypes collected from bulk samples (26). The microbial abundance profile of MRIMs was used to calculate the phenotypical feature for LUAD samples, and a correlation matrix was constructed to evaluate concordance between bulk transcriptome and single-cell data of LUAD. Consequently, cells were classified as MRIMs-positive (MRIMs⁺) cells, MRIMs-negative (MRIMs⁻) cells, or background cells based on their correlations with the phenotypical activities. The main cell types were annotated by referring to the marker genes in the CellMarker 2.0 database (27). These MRIMs-featured cells were further analyzed to investigate their association with specific cell types. Furthermore, a quantitative assessment was performed to evaluate the metabolic characteristics of MRIMs-featured cells (28).

Regulatory network of MRIMS-featured cells

The utilization of “Single-Cell rEgulatory Network Inference and Clustering” (SCENIC) analysis on scRNA-seq data enabled the identification and reconstruction of gene regulatory networks in MRIMs-featured cells (29). Regulons with an activity difference >0.3 between MRIMs^+/− cells were recognized as specific regulatory factors, which were employed for the construction of regulatory network. This regulatory association in MRIMs-featured cells was visualized through Cystoscope (30).

Construction and validation of MRIMs signature

The prognostic value of MRIMs was evaluated using univariate cox proportional hazard models, with a significance level set at P value <0.05 (31). Differentially abundant microbes (DAMs) were identified using the “limma” package in R software, applying a threshold of absolute fold change (FC) >1.5 and P<0.01 (32). The intersection of prognostic MRIMs and DAMs was selected to develop the signature for predicting LUAD prognosis. The microbial risk score (MRS) was calculated based on the coefficients from Cox regression analysis and microbial abundance, following the formula as follows:

MRS=∑i=1ncoefficient(i)×microbial_abundance(i)[1]

Consequently, each sample was assigned an MRS label. Subsequently, the samples were divided into low- and high-MRS groups based on the median MRS. To assess the predictive capability of the MRIMs signature, LUAD samples were randomly allocated to training and test cohorts in a 7:3 ratio for Kaplan-Meier survival analysis. The receiver operating characteristic (ROC) analysis implemented with the R package ’survivalROC’ was used to evaluate the predictive accuracy of overall survival (OS) at 1, 3, and 5 years. A nomogram incorporating MRS and clinical characteristics was developed to quantitatively predict OS. Finally, a calibration curve was employed to assess the robustness of the nomogram.

Functional enrichment analysis

The functional assessment of MRIMs-featured cells in the scRNA-seq dataset was conducted employing gene set enrichment analysis (GSEA) (33). The differentially expressed genes (DEGs) between the low- and high-MRS samples were subjected to Gene Ontology (GO) and KEGG pathway analyses utilizing Fisher’s exact test (19,34,35). A P value below 0.05 was considered statistically significant.

Correlation between MRIMs and immune microenvironment

The Mantel test was employed to assess the correlation between microbial abundance at the phylum class and infiltrating immune cells within the tumor microenvironment (36). The “EaSIeR” algorithm facilitates the estimation of patients’ likelihood to respond to immune checkpoint blockade (ICB) therapies, and repressed immune resistance scores (resF, resF_up, resF_down) were calculated for individuals (37). A comparison of the expression levels of immune checkpoint marker genes between low- and high-MRS groups was conducted to evaluate their association with MRIMs in LUAD.

Prediction of drug sensitivity

The “oncoPredict” R package was employed for the computation of the 50% inhibitory concentration (IC₅₀) of therapeutic drugs in patients (38). To evaluate the predictive potential of MRIMs signature in LUAD treatment, a comparison was made between the drug sensitivity to these drugs in patients with low- and high-MRS group.

Statistical analysis

The statistical analyses were conducted using R software (version 4.3.3) for data processing and visualization purposes. The correlation between LUAD subtypes and clinical features was assessed using a Chi-squared test. The Wilcoxon rank-sum test was employed to compare quantitative variables among different MRS subgroups. Kaplan-Meier survival analysis was performed to evaluate the prognostic impact of MRIMs signature on LUAD, followed by calculating the log-rank P value. Single cell data processing utilized the “Seurat” package (version 5.0.1) (39). Statistical significance was considered when the P value fell below 0.05.

Results

Identification of MRIMs and construction of featured subtypes in LUAD

The metabolic activities of LUAD samples were assessed using GSVA scores, and their correlation with intratumoral microbiota was determined through Spearman’s correlation coefficient analysis. A total of 10 genera of microbes, including Marichromatium, Collimonas, Campylobacter, Succinimonas, Sutterella, Synechococcus, Chitinivibrio, Luteibacter, Microvirga and Listeria were found to be significantly associated with metabolic regulation pathways and identified as MRIMs (Figure 2A). The LUAD samples were classified into two subtypes (C1 and C2) using an unsupervised clustering method based on the microbial abundance of MRIMs (Figure 2B), and the robustness of the subtype classification was demonstrated through a 3D-pca plot (Figure 2C).

Figure 2 Identification of metabolism-related intratumoral microbes and distinctive subtypes in LUAD. (A) Spearman’s coefficient analysis between microbial abundance and metabolic activities. (B) Unsupervised clustering of LUAD samples based on MRIMs. (C) 3D-PCA plot illustrating the microbe-derived subtypes of LUAD. 3D-PCA, 3-dimensional principal component analysis; LUAD, lung adenocarcinoma; MRIMs, metabolism-related intratumoral microbes.

Correlation between MRIMs-derived subtypes and LUAD

The associations between MRIMs-derived subtypes and prognosis, clinical characteristics, as well as genomic alteration features were investigated to elucidate their relationship with LUAD. The heatmap displayed the distinct microbial abundance of MRIMs and illustrated the differences in metabolic activities between subtypes C1 and C2 (Figure 3A). The correlation assessment findings revealed microbe-associated subtypes were clearly linked to patients’ survival status, tissue locations, pathologic N, and stage in LUAD. This relationship was depicted using an alluvial plot (Figure 3B). Kaplan-Meier survival curve indicated a poorer prognosis of LUAD patients in C1 group (Figure 3C). The landscape of genomic alterations revealed distinct mutations between the C1 and C2 groups. Glutamate rich protein 3 (ERICH3) was found to be the most distinctive gene in two subgroups (Figure 3D), suggesting its potential as a therapeutic target associated with microbe-host crosstalk for LUAD patients. Furthermore, the mutation frequencies were compared between C1 and C2 groups across eight genes, namely EGFR, BRAF, ALK, ROS1, RET, PIK3CA, MET, and HER2. These genes are commonly detected markers for targeted therapy in the clinical management of LUAD. However, no significant difference was observed in the mutation frequencies of these target genes within the two subtypes (Figure 3E). The ESTIMATE analysis revealed higher immune scores in the C2 group compared to the C1 subtype (Figure S1A). A significant difference was observed between the two MRIM-derived subgroups in terms of the infiltration levels of naïve B cells, follicular helper T cells, regulatory T cells, gamma delta T cells, activated NK cells, M2 macrophages, eosinophils, and neutrophils (Figure S1B). Furthermore, comparisons of immune checkpoint gene expression levels were performed between the C1 and C2 cohorts (Figure S1C). Within the C2 group, the expression levels of twenty-one genes —including TNFSF15, CD40LG, CD96, CD47, TNFSF9, CD48, HLA-DRB1, BTLA, BTN2A1, BTN2A2, HLA-DRA, HLA-DMA, HLA-DMB, HLA-DPA1, HLA-DPB1, CD160, BTNL9, CD200R1, CD226, CD244, and IDO2—were significantly elevated compared to those in the C1 group. In contrast, the expression levels of TDO2, TNFSF4, TNFSF18, and CD276 were higher in the C1 cohort. These findings indicated a strong association between MRIMs and immune cell infiltration in LUAD.

Figure 3 Correlations between microbe-associated subtype and clinical features in LUAD. (A) Heatmap of MRIMs abundance and metabolic activities in C1 and C2 groups. (B) Sankey diagram between subclusters and associated clinical characteristics. (C) Kaplan-Meier survival curve between C1 and C2 groups. (D) Genomic alteration landscape of different mutant genes between two subgroups. (E) Comparison of the mutant frequency of eight commonly detected target genes in C1 and C2 groups. *, 0.01≤P<0.05; **, 0.001≤P<0.01; ***, P<0.001. LUAD, lung adenocarcinoma; M, metastasis; MRIMs, metabolism-related intratumoral microbes; N, node; T, tumor.

Investigation of MRIMs feature at single-cell resolution

By implementing the Scissor algorithm, integration of scRNA-seq and bulk sequencing data enables a meticulous exploration of MRIMs features within the LUAD microenvironment at single-cell resolution. Following Uniform Manifold Approximation and Projection (UMAP) dimensional reduction, a total of 11,699 single cells were classified into 14 distinct cell clusters (Figure 4A), with seven distinct cell types identified based on the expression of specific marker genes (Figure 4B,4C and Figure S2): epithelial cells [4,391], endothelial cells [1,366], fibroblasts [735], macrophages [1,890], T cells [1,740], mast cells [1,198] and ciliated cells [379]. According to the microbial abundance level of MRIMs in The Cancer Genome Atlas-Lung Adenocarcinoma (TCGA-LUAD), the matched bulk samples were assigned as phenotypes to identify the most relevant cell subpopulations from LUAD single-cell data. Consequently, 3,124 MRIMs⁺ cells and 3,688 MRIMs⁻ cells were categorized using this microbial phenotype (Figure 4D). A significant difference was observed between MRIMs^+/− cell subclusters in the proportions of T cells, endothelial cells, macrophages, fibroblasts, ciliated cells, and mast cells (Figure 4E,4F). To elucidate the functional status of immune and stromal cells within the MRIMs^+/− subpopulations, a comparative analysis of these cell types across distinct MRIM-characterized subpopulations was conducted using pathway analyses (Figure 4G). Notably, mast cells exhibited higher folate biosynthesis activity in the MRIM⁺ subpopulation, as did macrophages, whereas stromal MRIM⁺-featured fibroblasts and ciliated cells displayed significantly reduced levels of this activity. Macrophages also demonstrated elevated activity in ubiquinone and other terpenoid-quinone biosynthesis within MRIM⁺ cells. With regard to fibroblasts, a functionally distinct profile was observed in glycosphingolipid biosynthesis between MRIMs^+/− subpopulations. Moreover, MRIM⁺-endothelial cells exhibited enhanced functional activity in both the pentose phosphate pathway and pyrimidine metabolism compared to endothelial cells in the MRIM⁻ subcluster. Additionally, epithelial and ciliated cells exhibiting MRIM⁺ features showed increased functional activities in oxidative phosphorylation and pyrimidine metabolism. These findings implicated the potential impact of MRIMs on the cellular functions in LUAD.

Figure 4 Analysis of MRIMs-featured cells in LUAD. (A) UMAP plot of cell clusters of LUAD scRNA-seq dataset. (B) Cell type annotations assigned to specific clusters. (C) The expression levels of marker genes for each cell type. (D) UMAP visualization of MRIMs-featured cells. (E) Cell proportions of identified cell types among MRIMs^+/− and background subclusters. (F) UMAP plots of identified cells within MRIMs^+/− subclusters. (G) Metabolic activities of different MRIMs-featured cells in LUAD. LUAD, lung adenocarcinoma; MRIMs, metabolism-related intratumoral microbes; UMAP, Uniform Manifold Approximation and Projection.

Regulatory network in MRIMs-featured cells

The results of GSEA revealed that MRIMs+ cells primarily exhibited biological functions related to immunity and metabolism, such as interleukin signaling, immune response, inflammation, regulation of carbohydrate metabolic processes, and peptidase activity. Conversely, MRIMs⁻ cells were found to be enriched in the cellular response to inorganic substances including zinc and cadmium ions and may affect the functions of lung basophil mast 1/2 cells (Figure 5A). These findings suggested a certain role of MRIMs in the microenvironment of LUAD. Additionally, SCENIC analysis was performed to further elucidate the underlying regulatory network of MRIMs-featured cells. The Venn diagram displayed the top 10 ranked regulons for each featured cell type (Figure 5B). Activity scores indicated specifically higher levels of NFKB2, KLF3, and BCL3 in MRIMs⁺ cells compared to MRIMs⁻ cells (Figure 5C). The regulatory network of MRIMs⁺ cells may play a crucial role in mediating microbe-host interactions, offering valuable insights for the identification of promising biomarkers in the treatment of LUAD (Figure 5D). Experimentally, RT-PCR analysis revealed a significant upregulation of NFKB2 expression in the A549 and PC-9 LUAD cell lines compared to human bronchial epithelial cells (BEAS-2B). BCL3 was also highly expressed in PC-9 cells, while KLF3 showed markedly reduced expression levels in these cells (Figure S3). These results further suggest their potential utility as novel biomarkers for LUAD.

Figure 5 Uncovering the underlying regulatory network of MRIM-featured cells. (A) The potential involvement of MRIMs^+/− cells in biological processes. (B) Top 10 regulons in MRIMs^+/− and background cells. (C) The activity scores of exclusive regulons in MRIMs^+/− cells. (D) Regulatory network in MRIMs⁺ cells. MRIMs, metabolism-related intratumoral microbes; NES, normalized enrichment score.

Construction and validation of prognostic MRIMs signature

The prognostic risk factors were identified through univariate Cox regression analysis (Figure 6A) and DAMs screening (Figure 6B). Succinimonas, Collimonas, and Marichromatium exhibited a significant predictive effect for poor prognosis [hazard ratio (HR) value >1, P<0.05], and their microbial abundance was enriched in LUAD tissue. The scoring system for this microbe-associated prognostic signature was calculated as follows: MRS = Succinimonas abundance × 0.163 + Collimonas abundance × 0.121 + Marichromatium abundance × 0.06. TCGA-LUAD samples were randomly divided into train and test cohorts in a 7:3 ratio, and patients in each cohort were classified into high- and low-MRS groups using the median MRS as cutoffs. The prognostic value of the signature was analyzed in the training, test, and full sets, which demonstrated a significant association between high-MRS and a worse survival outcome in LUAD (Figure 6C-6E). The AUC values for predicting survival at intervals of 1, 3, and 5 years highlighted the accuracy and robustness of the prognostic MRIMs signature (Figure 6F). In comparison to clinical characteristics, the nomogram indicated that MRS may serve as a promising indicator of prognosis for patients with LUAD (Figure 6G), while the calibration curve validated the reliability of this signature (Figure 6H).

Figure 6 Establishment and validation of MRIMs signature for LUAD prognosis. (A) Univariate Cox proportional hazards regression analysis of MRIMs. (B) Analysis of differently abundant microbes between LUAD and adjacent normal tissues. (C) Kaplan-Meier survival curve of high- and low-MRS groups within the training cohort. (D) Kaplan-Meier survival curve of high- and low-MRS groups in the test cohort. (E) Kaplan-Meier survival curve of high- and low-MRS groups in the full dataset. (F) ROC analysis for overall survival at 1-, 3-, and 5-year periods in LUAD. (G) Construction of a nomogram using MRS and clinical factors. (H) The calibration curve of the nomogram. *, 0.01≤P<0.05; **, 0.001≤P<0.01. AUC, area under the curve; CI, confidence interval; FDR, false discovery rate; HR, hazard ratio; LUAD, lung adenocarcinoma; MRIMs, metabolism-related intratumoral microbes; M, metastasis; MRS, microbial risk score; N, node; OS, overall survival; T, tumor.

Enriched functional evaluations

The DEGs between the high- and low-MRS groups were subjected to GO and KEGG pathway analyses. The GO analysis revealed that these DEGs were predominantly enriched in immune regulation and metabolism-related biological processes, including the regulation of T cell proliferation, inflammatory response, collagen catabolic process, triglyceride catabolic process, xenobiotic metabolism, one-carbon metabolism, reactive oxygen species metabolism, and other small molecule metabolic processes. Additionally, they showed responses to hypoxia, organic substances, glucose stimuli, etc. In terms of cellular components, the enriched GO terms included extracellular region and space, proteinaceous extracellular matrix, collagen trimer, and alveolar lamellar body. Regarding molecular functions, the DEGs primarily exhibited binding activities with fibronectin, calcium, wnt-protein, collagen, and metal ions. Additionally, the dehydratase activity of carbonate was observed (Figure 7A). Furthermore, the pathway analysis uncovered that dysregulation of PI3K-AKT and PPAR signaling pathways may impact the tumor progression, and the aberrant protein and nitrogen metabolism may also contribute to oncogenesis and development in LUAD (Figure 7B).

Figure 7 Functional analysis of differentially expressed genes between high- and low-MRS groups in LUAD. (A) Enriched GO terms of DEGs. (B) Enriched KEGG pathways of DEGs. The color gradient, ranging from shades of grey to deep red, represents different genes. DEGs, differentially expressed genes; ECM, extracellular matrix; GO, Gene Ontology; LUAD, lung adenocarcinoma; MRS, microbial risk score; KEGG, Kyoto Encyclopedia of Genes and Genomes.

Connection of MRIMs with immune microenvironment of LUAD

The correlation between metabolism-associated microbes and various types of immune cells was examined to evaluate their relationship with the LUAD immune microenvironment. Mantel test analysis revealed a significant correlation at the phylum level among MRIMs and activated CD4 memory T cells, gamma delta T cells, follicular helper T cells, M1 macrophages, resting natural killer (NK) cells, resting and activated dendritic cells, activated mast cells, and neutrophils (Figure 8A). The expression levels of immune checkpoint genes were compared across the MRS subgroups. Higher expression of BTN2A1/2, BTNL9, CD160, CD40LG, CD47, HLA-DMA/B, HLA-DOA HLA-DPA1 and HLA-DPB1 were observed in the low-MRS group as opposed to the high-MRS group. On the other hand, lower expressions of CD276, TDO2, TNFRSF18, TNFRSF9, and TNFSF4 were found in the former group (Figure 8B). Furthermore, EasieR was used to assess whether MRIMs could predict response to immune checkpoint inhibitors (ICIs) by estimating their ability to suppress immune resistance. A certain positive association was observed between resF_up and both Marichromutium abundance as well as Succinimonas (Figure 8C), indicating potential involvement of microbial abundance in immunotherapeutic resistance of LUAD treatment.

Figure 8 The correlation of microbial abundance and immune microenvironment in LUAD. (A) Mantel test on the immune cell fractions and MRIMs at phylum level in matched samples. (B) Comparison between the expression level of the immune checkpoint marker genes in high- and low-MRS groups. (C) The relationship between microbial abundance and suppressed immune resistance activity. *, 0.01 ≤P<0.05; **, 0.001≤P<0.01; ***, P<0.001. LUAD, lung adenocarcinoma; MRIMs, metabolism-related intratumoral microbes; MRS, microbial risk score.

The predictive value of MRS in drug sensitivity

The OncoPredict algorithm was used to assess the potential of microbial abundance in predicting drug sensitivity for LUAD therapy, allowing for the evaluation of IC50 values for therapeutic drugs within low- and high-MRS cohorts. Notably, a significant inverse correlation (|r|>0.3, P<0.05) was observed between the IC50 values of cisplatin, cytarabine, pyrimethamine, olaparib, bicalutamide and vorinostat and MRS levels (Figure 9A-9F). This indicated that patients with elevated MRS in the LUAD group exhibited increased susceptibility to these drugs compared to individuals with lower MRS levels. These findings underscored the promising utility of MRS as a predictive biomarker for guiding clinical decision-making regarding these medications in LUAD treatment.

Figure 9 Correlation between MRS and the drug sensitivity of (A) cisplatin, (B) cytarabine, (C) pyrimethamine, (D) olaparib, (E) bicalutamide, and (F) vorinostat. ***, P<0.001. IC₅₀, 50% inhibitory concentration; MRS, microbial risk score.

Discussion

The global mortality rate of LUAD remains high, posing a significant threat to human health. Despite the rapid advancements in treatment strategies for malignant tumors in recent years, the efficacy of these treatments has been unsatisfactory. Accumulative research on the regulatory function of microbial communities in cancer occurrence and development has revealed their significant impacts on tumor initiation, progression, metastasis, and treatment response (5-7). On the other hand, metabolic dysfunction represents a pivotal hallmark of cancer cells, and emerging evidence suggests the involvement of intratumoral microbes in this intricate process (9,40). Thus, elucidating the connection between metabolism-related intratumoral microbiota and tumor occurrence and development can favor identifying novel biomarkers and therapeutic targets for LUAD treatment.

The high degree of heterogeneity in malignant tumors is a significant factor resulting in diverse treatment outcomes and prognoses for cancer patients. Precise tumor classification has the potential to provide valuable insights and decision-making support for clinical management. Previous studies have identified distinct subtypes of hepatocellular carcinoma, breast cancer, colorectal cancer, and pancreatic cancer based on intratumoral microbiota (21,41-43), but no similar classification has been reported in LUAD research. In this study, we screened ten MRIMs and identified two distinct subtypes based on their microbial abundance. These microbe-derived subtypes demonstrate strong correlations with clinical features including survival status, tissue locations, pathologic N, and pathologic stage. Furthermore, significant differences in prognosis were observed between the two subtypes. To identify the potential functional pathways influenced by MRIM subtypes, KEGG pathway analysis was conducted based on DEGs between the C1 and C2 clusters. The results indicated that several metabolic, immune-related, and cancer-associated signaling pathways may play roles in the progression and development of LUAD. These pathways include protein digestion and absorption, proteoglycans in cancer, pyrimidine metabolism, galactose metabolism, the IL-17 signaling pathway, the PI3K-Akt signaling pathway, and the p53 signaling pathway (Figure S4). These findings further highlight the significant role of MRIMs in the initiation and progression of LUAD. In the genetic alteration profile, ERICH3 emerged as the gene exhibiting the most pronounced variation in mutant frequency between MRIMs-derived subtypes. It plays a role within retrograde intraflagellar transport (IFT)-associated pathways by modulating cilium length and ciliary levels of Shh signaling molecules (44). The antisense RNA of ERICH3 has already been identified as a diagnostic biomarker and an independent prognostic indicator for gastric cancer (45). However, the role of ERICH3 in LUAD remains unexplored warranting further investigations.

Single-cell sequencing has emerged as a great tool for studying tumor microenvironment and cell heterogeneity in the past several years. The generation of large-scale scRNA-seq data and advancements in computational methods have made it possible to analyze the characteristics and functions of MRIMs in LUAD from a single-cell perspective. As a novel algorithm, Scissor is capable of mapping the phenotypic features represented by bulk transcriptome data onto single cell sequencing data for multidimensional analysis. In this study, considering the significant association between MRIMs and LUAD, single cells from LUAD were labeled with MRIMs⁺ or MRIMs⁻ based on the phenotypic activity of TCGA-LUAD RNA-seq data using Scissor. Sun et al. applied Scissor to identify subsets of cells associated with TP53 mutations and worse survival in lung cancer (26). Zhang’s research developed an iron, copper, and sulfur-metabolism (ICSM)-associated signature to predict LUAD prognosis and therapeutic response, as well as identified ICSM-selected cells using the Scissor algorithm (46). In another study on LUAD, two groups of hypoxia-featured epithelial cells were identified through Scissor, which are involved in tumor invasion, metastasis, and immune therapy non-response (47). Recently, Li et al. proposed LP_SGL, a novel approach that integrates scRNA-seq, bulk expression, and bulk phenotype data by incorporating cell group structure (48). Definitely, their study demonstrated the potential of this method in predicting cancer, immunotherapy response, and survival outcomes through the identification of signaling genes within specific cell subsets in LUAD.

Reprogramming of metabolism is recognized as a critical indicator of cancer progression (49). This study assessed the correlation between intracellular microbes and host cell metabolic reprogramming at the single-cell level by evaluating the heterogeneity of MRIMs⁺ and MRIMs⁻ cells within aberrant metabolic pathways. Generally, these findings from both bulk data and single-cell data suggested that Intercellular microbes were linked to cellular metabolism in LUAD. Additionally, the enriched functional analysis of MRIMs^+/− cells demonstrated the dominant involvement of these cells with MRIMs features in biological processes related to immune mediation. It is currently hypothesized that bacteria in lung cancer primarily impact tumor progression by modulating the immune microenvironment and orchestrating local immune responses (50). Jin et al. found that increased local bacterial load and changes in microbiota composition stimulate the production of IL-1β and IL-23 by myeloid immune cells through the MyD88-dependent pathway in LUAD. These cytokines stimulate the activation and proliferation of Vγ6⁺ Vδ1⁺ γδ T cells, which in turn produce IL-17 and other effector molecules that promote inflammation and tumorigenesis (51). Dong et al. uncovered the association of Prevotella with inflammatory phenotype, including an enhanced Th17 lymphocyte and neutrophil response (52). In another study, following antibiotic aerosol inhalation, the bacterial load in tumoral murine lungs decreased, thereby activating tumor-infiltrating T cells and NK cells, reducing the number of immunosuppressive regulatory T cells, and enhancing local anti-tumor immune responses (53). In the present study, we further investigated the functional disparities of T cells between MRIMs^+/− cell clusters. Consequently, we observed elevated expression levels of LAG3, KLRG1, HAVCR2, IFNG and TGFB1 in MRIMS⁺ cells (Figure S5), which serve as biomarkers indicative of suppressed and exhausted T cell activity. This finding may elucidate the associations between intratumoral microbes and unfavorable survival outcomes in LUAD. In addition to the vital impacts of MIRMs on immunity in lung cancer, this work identified some cancerous signaling pathways in which MRIMs-associated DEGs participate, such as PI3K-Akt and PPAR. Previous study indicated that the intratumoral microbiome may contribute to the progression and metastasis of lung cancer by upregulating the activity of the PI3K pathway (54). Tsay et al. discovered a positive correlation between the upregulation of PI3K signaling pathways and an elevated abundance of Veillonella and Streptococcus in the lower respiratory tract among individuals diagnosed with lung cancer (55). The PPAR signaling pathway is intricately linked to the regulation of fatty acid metabolism. Xu et al. delineated a possible SIRT1/PGC-1α/PPAR-γ signaling-related molecular metabolic mechanism underlying hypoxia-induced chemotherapy resistance in the microenvironment of non-small cell lung cancer (NSCLC) (56). Chen et al. found that the inhibition of HLF facilitates distant metastases in multiple organs in NSCLC via the PPAR/NF-κb signaling pathway (57). Besides, a distinctive signature comprising of eight genes related to PPAR was developed for the prognosis prediction of LUAD. The crosstalk between MRIMs and the host may occur via these signaling pathways in LUAD (58). Importantly, this study explored the regulatory network in MRIM-featured cells, BCL3, KLF3 and NFKB2 were identified as the specific regulons in the MRIMs⁺ cells of LUAD. Among these regulators, the protein encoded by BCL3 functions as a transcriptional co-activator that activates through its association with NF-kappa B homodimers (59). Salameh’s study has reported on the role of BCL3 dysregulation in the pathogenesis of lung cancer (60). A genetic variation of BCL3 (rs8100239) may also be considered as a novel prognostic indicator in NSCLC (61). The transcription factor KLF3 enables sequence-specific double-stranded DNA binding activity. Previous study demonstrated the involvement of the miR-326/SP1/KLF3 regulatory axis in lung cancer progression (62), while Wei et al. reported that mir-130a downregulates KLF3 and inhibits lung cancer growth (63). Regarding NFKB2, its complex is expressed in numerous cell types and functions as a central activator of genes involved in inflammation and immune regulation. The dysregulation of the alternative NF-kB pathway in NSCLC may play a pivotal role in its pathogenesis and progression (64,65). Overall, elucidating the roles of these regulons in LUAD may facilitate the exploration of underlying mechanisms of MRIMs within the tumor microenvironment.

The symbiotic intratumoral microbiota are an important component in the microenvironment of solid tissue tumors and exert significant influence on tumor occurrence and progression (66). Numerous studies have identified distinct microbial signatures within various tumor types, including colorectal cancer, hepatocellular carcinoma, pancreatic cancer, and breast cancer, which can be utilized for prognostic prediction (41,43,67,68). In this study, we initially developed a metabolic-associated intratumoral microbiota-based prognostic signature in LUAD and validated its predictive accuracy and robustness. Within this signature, Succinimonas has been identified as a bacterial biomarker with predictive value for lung cancer recurrence or metastasis (69). The succinate production of Succinimonas activates tumor angiogenesis through SUCNR1-mediated ERK1/2 and STAT3 signaling pathways (70). Collimonas are well adapted to nutrient-poor environments, its production of N-acyl homoserine lactones were reported that play a role in microbial interactions and metabolism readjustment (71). Marichromatium is a bacterium associated with sulfur metabolism and exhibits the ability to utilize hydrogen sulfide (H₂S) as a metabolic substrate, potentially compromising immune signaling mediated by H₂S and thereby contributing to tumor progression (72,73). However, to date, no investigations have been reported on the role of these microbes in LUAD, underscoring the imperative for further exploration.

The mutation rate of microbial DNA is substantially lower than that of human genes, rendering it less vulnerable to the interference of tumor heterogeneity. In addition, distinct tumor types exhibit unique characteristics of intratumoral microbial communities. This indicates that intratumoral microorganisms possess excellent specificity as biomolecular markers. Notably, the metabolic products of intratumoral microorganisms can be involved in the regulation of tumor and associated cellular immune signals, which undoubtedly provides novel perspectives and a theoretical foundation for uncovering new molecular targets for tumor treatment. However, it is essential to acknowledge that certain inherent limitations may impose restrictions on the application of intratumoral microorganisms as biomolecular markers. For one thing, their low abundance in tumor tissues can readily result in false-positive detection outcomes. For another, tissue samples are prone to contamination during the collection process, thereby influencing the accuracy of detection. Moreover, factors such as antibiotic usage and diet may potentially alter the levels of microbial markers. The interference from the host-microbiota interaction can also decrease the detection efficiency of intratumoral microorganisms (74). Looking ahead, the employment of dual UMI-labeled amplicon sequencing (e.g., the Primer-Scheme method) can be considered to mitigate amplification bias, thus enhancing the precision of microbiota detection. Additionally, the use of microbial flow sorting to directly isolate viable bacteria can effectively circumvent the impact of DNA contamination. Promisingly, spatial metagenomics technology holds new promise for unveiling the dynamic regulatory mechanisms of intratumoral microorganisms within the tumor microenvironment (75,76).

The essential role of the immune microenvironment in LUAD progression has been demonstrated. As our understanding deepens regarding the intricate relationship between immune responses and the tumor microenvironment, novel prognostic and therapeutic biomarkers are being discovered. The present study unveiled the correlation between microbial abundance of MRIMs and immune cell infiltration at the phylum level, including T lymphocytes, macrophages, NK cells, dendritic cells, mast cells, and neutrophils. These findings implicate the impact of MRIMs on the immune-related functions in LUAD. Nowadays, immunotherapy has emerged as a promising approach for cancer treatment by disrupting interactions between immune checkpoints to achieve anti-tumor effects (77). For patients with relapsed or metastatic LUAD, immunotherapy may offer a potential therapeutic option; however, heterogeneous responses to ICIs primarily result from variable expression levels of genes associated with immune checkpoints. Individual response to ICIs also varies significantly (78). Therefore, there is significant interest in exploring predictive signatures that can guide immunotherapy decisions. In this study, differential expressions were observed in several immune checkpoint genes between high- and low-MRS subgroups of LUAD patients while significant links were found between microbial abundance of MRIMs and suppressed immune resistance activity suggesting potential utility in predicting immunotherapy efficacy for treating LUAD. Moreover, the MRIMs signature also exhibited potential to inform clinical decision-making regarding drug use in chemotherapy and targeted therapy thereby favoring management strategies for treating LUAD patients.

As the increasing recognition of the role of intracellular microbiota in tumor tissue, elucidating their relationship with the tumor microenvironment may offer a novel approach to cancer treatment. Our study first identified intratumoral microbes associated with metabolism and revealed distinct subtypes linked to MRIMs in LUAD. Additionally, we investigated the cellular characteristics of MRIMs and their regulatory network at a single-cell resolution. The developed signature of MRIMs demonstrated potential for prognostic prediction, indicating immunotherapeutic efficacy, and estimating drug sensitivity in the treatment of chemo- and targeted therapy. However, there are several limitations in this study. The retrospective nature of the predictive signature limits its robustness, and the causal relationship between MRIMS and LUAD is yet to be confirmed. Future studies involving vitro/vivo experiments are needed to explore how MRIMs interact with the host immune and metabolism system.

Conclusions

In general, this study identified intratumoral microbes related to metabolism in LUAD and established a prognostic signature that can also inform the immunotherapeutic benefit and drug response in treatment. Furthermore, by integrating multi-omics analyses, this work unveiled the characteristics and potential mechanisms of MRIMs in LUAD, thereby offering novel insights into the pathogenesis of LUAD.

Acknowledgments

None.

Footnote

Reporting Checklist: The authors have completed the TRIPOD reporting checklist. Available at https://tcr.amegroups.com/article/view/10.21037/tcr-2025-357/rc

Peer Review File: Available at https://tcr.amegroups.com/article/view/10.21037/tcr-2025-357/prf

Funding: This work was supported by the Science and Technology Program Fund of Huaihua (grant No. 2021R3111) and the Fund of Huaihua Technology Innovation Platform (grant No. 2021R2206).

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://tcr.amegroups.com/article/view/10.21037/tcr-2025-357/coif). The authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Bray F, Laversanne M, Sung H, et al. Global cancer statistics 2022: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin 2024;74:229-63. [Crossref] [PubMed]
2. Lahiri A, Maji A, Potdar PD, et al. Lung cancer immunotherapy: progress, pitfalls, and promises. Mol Cancer 2023;22:40. [Crossref] [PubMed]
3. Galetta D. Advances in Lung Cancer Therapy. Cancers (Basel) 2023;15:2671. [Crossref] [PubMed]
4. DiMaio D, Emu B, Goodman AL, et al. Cancer Microbiology. J Natl Cancer Inst 2022;114:651-63. [Crossref] [PubMed]
5. Bullman S. The intratumoral microbiota: From microniches to single cells. Cell 2023;186:1532-4. [Crossref] [PubMed]
6. Gao F, Yu B, Rao B, et al. The effect of the intratumoral microbiome on tumor occurrence, progression, prognosis and treatment. Front Immunol 2022;13:1051987. [Crossref] [PubMed]
7. Galeano Niño JL, Wu H, LaCourse KD, et al. Effect of the intratumoral microbiota on spatial and cellular heterogeneity in cancer. Nature 2022;611:810-7. [Crossref] [PubMed]
8. Wang W, Rong Z, Wang G, et al. Cancer metabolites: promising biomarkers for cancer liquid biopsy. Biomark Res 2023;11:66. [Crossref] [PubMed]
9. Li M, Yang Y, Xiong L, et al. Metabolism, metabolites, and macrophages in cancer. J Hematol Oncol 2023;16:80. [Crossref] [PubMed]
10. Zhou D, Li Y. Gut microbiota and tumor-associated macrophages: potential in tumor diagnosis and treatment. Gut Microbes 2023;15:2276314. [Crossref] [PubMed]
11. Jin J, Byun JK, Choi YK, et al. Targeting glutamine metabolism as a therapeutic strategy for cancer. Exp Mol Med 2023;55:706-15. [Crossref] [PubMed]
12. Pavlova NN, Zhu J, Thompson CB. The hallmarks of cancer metabolism: Still emerging. Cell Metab 2022;34:355-77. [Crossref] [PubMed]
13. Pavlova NN, Thompson CB. The Emerging Hallmarks of Cancer Metabolism. Cell Metab 2016;23:27-47. [Crossref] [PubMed]
14. Tomczak K, Czerwińska P, Wiznerowicz M. The Cancer Genome Atlas (TCGA): an immeasurable source of knowledge. Contemp Oncol (Pozn) 2015;19:A68-77. [Crossref] [PubMed]
15. Gao Z, Jiang A, Li Z, et al. Heterogeneity of Intratumoral Microbiota Within the Tumor Microenvironment and Relationship to Tumor Development. Med Research 2025;1:32-61.
16. Wu Y, Wang P, Zhao X, et al. Dynamic interplay of intratumoral microbiota within the colorectal cancer microenvironment. View 2024;5:20230100.
17. Sepich-Poore GD, Zitvogel L, Straussman R, et al. The microbiome and human cancer. Science 2021;371:eabc4552. [Crossref] [PubMed]
18. O'Leary NA, Wright MW, Brister JR, et al. Reference sequence (RefSeq) database at NCBI: current status, taxonomic expansion, and functional annotation. Nucleic Acids Res 2016;44:D733-45. [Crossref] [PubMed]
19. Kanehisa M, Furumichi M, Tanabe M, et al. KEGG: new perspectives on genomes, pathways, diseases and drugs. Nucleic Acids Res 2017;45:D353-61. [Crossref] [PubMed]
20. Hänzelmann S, Castelo R, Guinney J. GSVA: gene set variation analysis for microarray and RNA-seq data. BMC Bioinformatics 2013;14:7. [Crossref] [PubMed]
21. Chen F, Yang J, Guo Y, et al. Integrating bulk and single-cell RNA sequencing data reveals the relationship between intratumor microbiome signature and host metabolic heterogeneity in breast cancer. Front Immunol 2023;14:1140995. [Crossref] [PubMed]
22. Wilkerson MD, Hayes DN. ConsensusClusterPlus: a class discovery tool with confidence assessments and item tracking. Bioinformatics 2010;26:1572-3. [Crossref] [PubMed]
23. Gu Z, Eils R, Schlesner M. Complex heatmaps reveal patterns and correlations in multidimensional genomic data. Bioinformatics 2016;32:2847-9. [Crossref] [PubMed]
24. Mayakonda A, Lin DC, Assenov Y, et al. Maftools: efficient and comprehensive analysis of somatic variants in cancer. Genome Res 2018;28:1747-56. [Crossref] [PubMed]
25. Sturm G, Finotello F, List M. Immunedeconv: An R Package for Unified Access to Computational Methods for Estimating Immune Cell Fractions from Bulk RNA-Sequencing Data. Methods Mol Biol 2020;2120:223-32. [Crossref] [PubMed]
26. Sun D, Guan X, Moran AE, et al. Identifying phenotype-associated subpopulations by integrating bulk and single-cell sequencing data. Nat Biotechnol 2022;40:527-38. [Crossref] [PubMed]
27. Hu C, Li T, Xu Y, et al. CellMarker 2.0: an updated database of manually curated cell markers in human/mouse and web tools based on scRNA-seq data. Nucleic Acids Res 2023;51:D870-6. [Crossref] [PubMed]
28. Wu Y, Yang S, Ma J, et al. Spatiotemporal Immune Landscape of Colorectal Cancer Liver Metastasis at Single-Cell Level. Cancer Discov 2022;12:134-53. [Crossref] [PubMed]
29. Aibar S, González-Blas CB, Moerman T, et al. SCENIC: single-cell regulatory network inference and clustering. Nat Methods 2017;14:1083-6. [Crossref] [PubMed]
30. Demchak B, Hull T, Reich M, et al. Cytoscape: the network visualization tool for GenomeSpace workflows. F1000Res 2014;3:151. [Crossref] [PubMed]
31. Ata N, Sozer TM. Cox Regression Models with Nonproportional Hazards Applied to Lung Cancer Survival Data. Hacettepe Journal of Mathematics Statistics 2007;36:157-67.
32. Ritchie ME, Phipson B, Wu D, et al. limma powers differential expression analyses for RNA-sequencing and microarray studies. Nucleic Acids Res 2015;43:e47. [Crossref] [PubMed]
33. Subramanian A, Tamayo P, Mootha VK, et al. Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles. Proc Natl Acad Sci U S A 2005;102:15545-50. [Crossref] [PubMed]
34. Upton GJG. Fisher's Exact Test. Journal of the Royal Statistical Society Series A: Statistics in Society 1992;155:395-402.
35. Ashburner M, Ball CA, Blake JA, et al. Gene ontology: tool for the unification of biology. The Gene Ontology Consortium. Nat Genet 2000;25:25-9. [Crossref] [PubMed]
36. Anderson MJ, Walsh DCI. PERMANOVA, ANOSIM, and the Mantel test in the face of heterogeneous dispersions: What null hypothesis are you testing? Ecological Monographs 2013;83:557-74.
37. Lapuente-Santana Ó, van Genderen M, Hilbers PAJ, et al. Interpretable systems biomarkers predict response to immune-checkpoint inhibitors. Patterns (N Y) 2021;2:100293. [Crossref] [PubMed]
38. Maeser D, Gruener RF, Huang RS. oncoPredict: an R package for predicting in vivo or cancer patient drug response and biomarkers from cell line screening data. Brief Bioinform 2021;22:bbab260. [Crossref] [PubMed]
39. Stuart T, Butler A, Hoffman P, et al. Comprehensive Integration of Single-Cell Data. Cell 2019;177:1888-1902.e21. [Crossref] [PubMed]
40. Dzutsev A, Badger JH, Perez-Chanona E, et al. Microbes and Cancer. Annu Rev Immunol 2017;35:199-228. [Crossref] [PubMed]
41. Zhang B, Liu J, Li H, et al. Integrated multi-omics identified the novel intratumor microbiome-derived subtypes and signature to predict the outcome, tumor microenvironment heterogeneity, and immunotherapy response for pancreatic cancer patients. Front Pharmacol 2023;14:1244752. [Crossref] [PubMed]
42. Jin M, Shang F, Wu J, et al. Tumor-Associated Microbiota in Proximal and Distal Colorectal Cancer and Their Relationships With Clinical Outcomes. Front Microbiol 2021;12:727937. [Crossref] [PubMed]
43. Huang JH, Wang J, Chai XQ, et al. The Intratumoral Bacterial Metataxonomic Signature of Hepatocellular Carcinoma. Microbiol Spectr 2022;10:e0098322. [Crossref] [PubMed]
44. Alsolami M, Kuhns S, Alsulami M, et al. ERICH3 in Primary Cilia Regulates Cilium Formation and the Localisations of Ciliary Transport and Sonic Hedgehog Signaling Proteins. Sci Rep 2019;9:16519. [Crossref] [PubMed]
45. Chen Q, Huang X, Dong X, et al. Long non-coding RNA ERICH3-AS1 is an unfavorable prognostic factor for gastric cancer. PeerJ 2020;8:e8050. [Crossref] [PubMed]
46. Zhang L, Zhang X, Guan M, et al. Machine-learning developed an iron, copper, and sulfur-metabolism associated signature predicts lung adenocarcinoma prognosis and therapy response. Respir Res 2024;25:206. [Crossref] [PubMed]
47. Li M, Zhou B, Zheng C. An integrated bioinformatic analysis of bulk and single-cell sequencing clarifies immune microenvironment and metabolic profiles of lung adenocarcinoma to predict immunotherapy efficacy. Front Cell Dev Biol 2023;11:1163314. [Crossref] [PubMed]
48. Li J, Zhang H, Mu B, et al. Identifying phenotype-associated subpopulations through LP_SGL. Brief Bioinform 2023;25:bbad424. [Crossref] [PubMed]
49. Hanahan D, Weinberg RA. Hallmarks of cancer: the next generation. Cell 2011;144:646-74. [Crossref] [PubMed]
50. Sommariva M, Le Noci V, Bianchi F, et al. The lung microbiota: role in maintaining pulmonary immune homeostasis and its implications in cancer development and therapy. Cell Mol Life Sci 2020;77:2739-49. [Crossref] [PubMed]
51. Jin C, Lagoudas GK, Zhao C, et al. Commensal Microbiota Promote Lung Cancer Development via γδ T Cells. Cell 2019;176:998-1013.e16. [Crossref] [PubMed]
52. Dong Q, Chen ES, Zhao C, et al. Host-Microbiome Interaction in Lung Cancer. Front Immunol 2021;12:679829. [Crossref] [PubMed]
53. Le Noci V, Guglielmetti S, Arioli S, et al. Modulation of Pulmonary Microbiota by Antibiotic or Probiotic Aerosol Therapy: A Strategy to Promote Immunosurveillance against Lung Metastases. Cell Rep 2018;24:3528-38. [Crossref] [PubMed]
54. Gustafson AM, Soldi R, Anderlind C, et al. Airway PI3K pathway activation is an early and reversible event in lung cancer development. Sci Transl Med 2010;2:26ra25. [Crossref] [PubMed]
55. Tsay JJ, Wu BG, Badri MH, et al. Airway Microbiota Is Associated with Upregulation of the PI3K Pathway in Lung Cancer. Am J Respir Crit Care Med 2018;198:1188-98. [Crossref] [PubMed]
56. Xu R, Luo X, Ye X, et al. SIRT1/PGC-1α/PPAR-γ Correlate With Hypoxia-Induced Chemoresistance in Non-Small Cell Lung Cancer. Front Oncol 2021;11:682762. [Crossref] [PubMed]
57. Chen J, Liu A, Lin Z, et al. Downregulation of the circadian rhythm regulator HLF promotes multiple-organ distant metastases in non-small cell lung cancer through PPAR/NF-κb signaling. Cancer Lett 2020;482:56-71. [Crossref] [PubMed]
58. Zhang W, Liu J, Ren X, et al. Identification of the novel markers of PPAR signalling affecting immune microenvironment and immunotherapy response of lung adenocarcinoma patients. J Cell Mol Med 2024;28:e17877. [Crossref] [PubMed]
59. Seaton G, Smith H, Brancale A, et al. Multifaceted roles for BCL3 in cancer: a proto-oncogene comes of age. Mol Cancer 2024;23:7. [Crossref] [PubMed]
60. Salameh L, Bhamidimarri P, Elemam N, et al. The Role of BCL3 Dysregulation in the Pathogenesis of Asthma and Lung Cancer. Am J Respir Crit Care Med. 2023;207:A5758.
61. Dimitrakopoulos FI, Antonacopoulou AG, Kottorou A, et al. Variant of BCL3 gene is strongly associated with five-year survival of non-small-cell lung cancer patients. Lung Cancer 2015;89:311-9. [Crossref] [PubMed]
62. Wang R, Xu J, Xu J, et al. MiR-326/Sp1/KLF3: A novel regulatory axis in lung cancer progression. Cell Prolif 2019;52:e12551. [Crossref] [PubMed]
63. Wei MC, Wang YM, Wang DW. miR-130a-Mediated KLF3 Can Inhibit the Growth of Lung Cancer Cells. Cancer Manag Res 2021;13:2995-3004. [Crossref] [PubMed]
64. Dimitrakopoulos FD, Antonacopoulou A, Kottorou AE, et al. Shedding light on the alternative pathway of NF-kB in non-small cell lung cancer. J Clin Oncol 2011;29:e21123.
65. Lombardi L, Ciana P, Cappellini C, et al. Structural and functional characterization of the promoter regions of the NFKB2 gene. Nucleic Acids Res 1995;23:2328-36. [Crossref] [PubMed]
66. Garrett WS. Cancer and the microbiota. Science 2015;348:80-6. [Crossref] [PubMed]
67. Parida S, Siddharth S, Xia Y, et al. Concomitant analyses of intratumoral microbiota and genomic features reveal distinct racial differences in breast cancer. NPJ Breast Cancer 2023;9:4. [Crossref] [PubMed]
68. Jin N, Hoyd R, Yilmaz AS, et al. Microbiome signature, global methylation and immune landscape in early onset colorectal cancer. J Clin Oncol 2021;39:abstr 3519.
69. Yuan X, Wang Z, Li C, et al. Bacterial biomarkers capable of identifying recurrence or metastasis carry disease severity information for lung cancer. Front Microbiol 2022;13:1007831. [Crossref] [PubMed]
70. Kuo CC, Wu JY, Wu KK. Cancer-derived extracellular succinate: a driver of cancer metastasis. J Biomed Sci 2022;29:93. [Crossref] [PubMed]
71. Uroz S, Geisler O, Fauchery L, et al. Genomic and transcriptomic characterization of the Collimonas quorum sensing genes and regulon. FEMS Microbiol Ecol 2022;98:fiac100. [Crossref] [PubMed]
72. Dahl C, Rákhely G, Pott-Sperling AS, et al. Genes involved in hydrogen and sulfur metabolism in phototrophic sulfur bacteria. FEMS Microbiol Lett 1999;180:317-24. [Crossref] [PubMed]
73. Joulia E, Metallo CM. Methionine and H(2)S alter cancer-immune dialogue. Nat Metab 2023;5:1456-8. [Crossref] [PubMed]
74. Asgharzadeh S, Pourhajibagher M, Bahador A. The microbial landscape of tumors: a deep dive into intratumoral microbiota. Front Microbiol 2025;16:1542142. [Crossref] [PubMed]
75. Tan Q, Cao X, Zou F, et al. Spatial Heterogeneity of Intratumoral Microbiota: A New Frontier in Cancer Immunotherapy Resistance. Biomedicines 2025;13:1261. [Crossref] [PubMed]
76. Ribas A, Wolchok JD. Cancer immunotherapy using checkpoint blockade. Science 2018;359:1350-5. [Crossref] [PubMed]
77. Mamdani H, Matosevic S, Khalid AB, et al. Immunotherapy in Lung Cancer: Current Landscape and Future Directions. Front Immunol 2022;13:823618. [Crossref] [PubMed]
78. Wu J, Zhang P, Mei W, et al. Intratumoral microbiota: implications for cancer onset, progression, and therapy. Front Immunol 2023;14:1301506. [Crossref] [PubMed]