Abstract

   Systemic lupus erythematosus (SLE) patients exhibit a heightened risk
   of developing lung cancer, yet the underlying molecular mechanisms
   remain poorly understood. This study aimed to identify shared genetic
   factors linking SLE and LC using publicly available transcriptomic data
   from the Gene Expression Omnibus (GEO). Through integrated
   differentially expressed gene (DEG) analysis and weighted gene
   co-expression network analysis (WGCNA), we identified five genes
   consistently upregulated in both SLE and lung cancer. Gene set
   enrichment analysis (GSEA) revealed that these shared genes were
   enriched in inflammatory pathways, particularly those involving
   interferon-alpha, interferon-gamma, and general inflammatory responses.
   We applied least absolute shrinkage and selection operator (LASSO)
   regression to pinpoint potential diagnostic biomarkers and identified
   two key candidates: AIM2 and SLC26A8. These biomarkers demonstrated
   robust diagnostic performance with area under the ROC curve (AUC)
   values exceeding 0.75 in both training and validation cohorts. Immune
   infiltration and survival analyses using The Cancer Genome Atlas (TCGA)
   further supported their clinical relevance. Notably, high AIM2
   expression was significantly associated with poorer overall survival in
   female lung adenocarcinoma patients (P = 0.03), and SLC26A8 expression
   was significantly linked to survival outcomes only in patients with a
   history of smoking (P = 0.01). These findings are particularly
   meaningful in SLE, where most patients are female and smoking is a
   known risk factor. Our study enhances the understanding of
   autoimmune-driven carcinogenesis and opens new avenues for precision
   medicine strategies in managing patients with SLE at risk for lung
   cancer.

   Keywords: Systemic lupus erythematosus, Lung cancer, Tumorigenesis,
   Tumor immune microenvironment, AIM2, SLC26A8, Bioinformatics

1. Background

   Systemic lupus erythematosus (SLE) is one of the leading causes of
   death in young females [[27]1,[28]2], with an increasing prevalence
   ranging from 13 to 7713.5 per 100,000 persons worldwide [[29]3]. This
   chronic autoimmune disease features not only multi-organ involvement
   due to the disease itself but also susceptibility to various
   comorbidities, including an increased risk of certain types of cancer
   [[30][4], [31][5], [32][6], [33][7]]. A meta-analysis comprising 57,890
   SLE patients suggested a 1.6-fold increased risk of developing lung
   cancer in the SLE population compared with healthy controls (95 % CI:
   1.44–1.77; P < 0.00001) [[34]8]. Research has also indicated that the
   standardized mortality ratio for lung cancer in individuals with SLE is
   2.3 (95 % CI 1.6–3.0) in contrast to that in the general population
   [[35]9]. The distribution of lung cancer subtypes in SLE patients
   resembled that in lung cancer patients from the general population,
   with adenocarcinoma being the most prevalent subtype, followed by small
   cell carcinoma and squamous cell carcinoma [[36]10]. While most SLE
   patients who develop lung cancer are ever smokers, indicating that
   tobacco smoking is a shared risk factor for both SLE and lung cancer, a
   proportion of nonsmoking SLE individuals still develop lung
   malignancies with histological types not associated with smoking
   [[37]11].

   Although the exact mechanism of tumorigenesis in the SLE population
   remains elusive and complicated, emerging evidence suggests that
   persistent immune activation and pulmonary fibrosis may create a
   microenvironment conducive to the carcinogenesis of lung tissues
   [[38][12], [39][13], [40][14]]. Additionally, an increasing number of
   studies indicate that genetic susceptibility contributes to SLE and the
   occurrence of lung malignancy [[41][15], [42][16], [43][17], [44][18],
   [45][19], [46][20]], suggesting a potential genetic link that may
   predispose individuals with SLE to the development of lung cancer. The
   identification of shared biomarkers that promote tumorigenesis and
   autoimmune genesis is crucial for elucidating the commonalities and
   molecular mechanisms of SLE and lung cancer and further developing
   targeted therapeutic strategies.

   In this study, we explored the connections between SLE and lung cancer
   by analyzing the biological mechanisms involving DEGs shared by SLE and
   lung cancer patients. This investigation was conducted through
   comprehensive bioinformatics analysis and machine learning applied to
   gene expression profiles from whole blood samples from patients with
   these disorders.

2. Materials and methods

2.1. Data collection and processing

   The workflow chart of this study is shown in [47]Fig. 1. We searched
   for gene expression profiles of SLE and lung cancer patients in the
   Gene Expression Omnibus (GEO) database
   ([48]https://www.ncbi.nlm.nih.gov/geo/) [[49]21] via the terms
   "systemic lupus erythematosus" and "lung cancer". The criteria included
   (1) Homo sapiens; (2) Array and sequencing data should include both
   case and control groups; (3) samples obtained from whole blood cells or
   peripheral mononuclear cells; and (4) the number of samples in each
   dataset must be greater than 30. Eventually, [50]GSE72509
   ([51]GPL16791; containing 99 SLE and 18 control whole blood cell
   samples) and [52]GSE42830 ([53]GPL10558; containing 8 lung cancer and
   38 healthy whole blood samples) were obtained and used as training
   datasets. We further chose [54]GSE61635 ([55]GPL570; containing 99 SLE
   and 30 control whole blood cell samples) and [56]GSE42834
   ([57]GPL10558; containing 16 lung cancer and 143 healthy whole blood
   samples) for subsequent validation([58]Table 1). All expression data
   were downloaded into R software (version4.41) using R “GEOquery” for
   downstream analysis.

Fig. 1.

   [59]Fig. 1
   [60]Open in a new tab

   Flow chart.

Table 1.

   Information of GEO datasets containing the SLE/lung cancer patients.
   GSE number Platform Samples Disease Groups Sample type
   [61]GSE72509 [62]GPL16791 99 patients/18 controls SLE Training set
   Whole Blood
   [63]GSE42830 [64]GPL10558 8 patients/38 controls Lung cancer Training
   set Whole Blood
   [65]GSE61635 [66]GPL570 99 patients/30 controls SLE Validating set
   Whole Blood
   [67]GSE42834 [68]GPL10558 16 patients/143 controls Lung cancer
   Validating set Whole Blood
   [69]Open in a new tab

2.2. Identification of shared DEGs

   GEO2R ([70]http://www.ncbi.nlm.nih.gov/geo/geo2r) is a web tool
   allowing users to identify DEGs across self-defined groups, using
   Bioconductor R packages for analysis. We divided [71]GSE72509 and
   [72]GSE42830 into disease and control groups, respectively, based on
   the disease state of the sample. For the microarray data, the force
   normalization function was applied, and R “GEOquery” and “limma”
   packages were utilized. Significant DEGs were identified as those whose
   adjusted P value was ≤0.05 and whose | log[2] fold change(FC) | was ≥1.
   Moreover, probe IDs without a corresponding gene symbol are removed.
   The shared DEGs from the SLE dataset and lung cancer dataset were
   obtained and visualized via the R “ggvenn” package.

2.3. Weighted gene coexpression network (WGCNA) construction

   WGCNA is a bioinformatics method that clusters genes into modules on
   the basis of their coexpression patterns across all samples and
   identifies genes strongly correlated with the phenotype and the
   biological processes represented by these modules [[73]22]. We
   performed WGCNA via the WGCNA R package and initiated the process by
   excluding poor-quality genes and outlier samples through sample
   clustering and principal component analysis (PCA). The optimal soft
   power was then selected to balance the highest scale-free topology fit
   index (R^2) with adequate mean connectivity. We then constructed a
   topological overlap matrix (TOM) and computed the correlations between
   gene modules and phenotypes to identify modules associated with
   clinical traits. Heatmaps with Spearman's correlation coefficients
   between modules with SLE and those with lung cancer were drawn for
   module selection. The scripts used in this process are available in
   Additional file 1.

2.4. Functional enrichment analysis

   To gain insight into the potential shared pathological mechanism of SLE
   and lung cancer, we performed Gene Ontology (GO) enrichment analysis
   and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis on
   the shared DEGs and module genes identified via WGCNA via the
   Annotation, Visualization and Integrated Discovery (DAVID) database
   ([74]https://david.ncifcrf.gov). A P value < 0.05 was considered a
   significant term, and the results were visualized via the R “ggplot2”
   package.

2.5. Gene set enrichment analysis (GSEA)

   GSEA was conducted via the R “clusterProfiler” package to explore the
   biological functions of the hub genes. GSEA is a computational method
   that calculates gene set enrichment scores and helps understand the
   biological significance of gene expression changes in the context of
   the whole transcriptome [[75]23]. Gene expression data from SLE
   patients and lung cancer patients were first imported, and Spearman's
   correlation coefficient for each hub gene against all protein-coding
   genes in the transcriptome was computed. The resulting correlation
   values are then ranked against known gene sets. The hallmark gene sets
   were downloaded from the Molecular Signatures Database (MSigDB)
   ([76]https://www.gsea-msigdb.org/gsea/msigdb/index.jsp)
   [[77]24,[78]25]. Enriched gene sets with P values < 0.05, normalized
   enrichment scores >1, and FDR q values < 0.25 were considered
   significant. The scripts used in this process are available in
   Additional file 3.

2.6. Validation of core shared genes

   The LASSO regression algorithm, a machine learning approach that
   applies a penalty to the absolute values of the regression coefficients
   and shrinks some to zero, was employed via the R “glmnet” package to
   filter the most relevant cored shared genes with non-zero coefficients
   through 10-fold cross-validation. We further conducted receiver
   operating characteristic (ROC) curve analysis with the R package “pROC”
   to evaluate the diagnostic value of the selected genes for SLE and lung
   cancer. The area under the curve (AUC) was determined to assess the
   accuracy and sensitivity of these hub genes in disease diagnosis.
   Additionally, ROC analysis was carried out on the validation datasets
   to further confirm the diagnostic relevance of the identified shared
   genes.

2.7. Single cell RNA sequencing analysis

   To validate our findings at the single-cell level, we performed an
   integrative single-cell RNA sequencing (scRNA-seq) analysis using the
   publicly available dataset [79]GSE136035, which includes peripheral
   blood mononuclear cells (PBMCs) from four SLE patients and two healthy
   controls. Raw gene expression matrices were downloaded and processed
   using the Seurat (v5.0) R package. Each sample was independently loaded
   and filtered based on standard quality control criteria (cells with
   >200 and < 6000 detected features and <10 % mitochondrial gene
   content). After normalization and identification of highly variable
   features, samples were integrated using Seurat's anchor-based
   integration workflow. Dimensionality reduction was conducted via PCA,
   followed by clustering and Uniform Manifold Approximation and
   Projection (UMAP) visualization. For cell-type annotation, we employed
   the SingleR package using the BlueprintEncodeData reference. Predicted
   cell-type identities were added to the metadata of the integrated
   object, enabling cluster-level analysis of gene expression across
   immune cell subsets. To quantify transcript-level differences across
   conditions, we implemented a pseudobulk differential expression
   approach. For each sample, raw UMI counts were aggregated across all
   cells to generate sample-level gene expression matrices, which were
   then analyzed using DESeq2. This strategy accounts for sample-level
   biological variability while preserving the benefits of single-cell
   resolution. This single-cell level validation complements our bulk
   transcriptomic analysis and helps clarify the cell-type-specific
   expression patterns of candidate genes.

2.8. Immune cell infiltration

   The Tumor IMmune Estimation Resource (TIMER)
   ([80]https://cistrome.shinyapps.io/timer/) is a dedicated online tool
   designed to analyze the infiltration of immune cells in various cancer
   types via gene expression data [[81]26]. The gene module of TIMER
   provides insights into the tumor microenvironment by estimating the
   correlation of specific gene expression and the abundance of six
   different immune cell types, including B cells, T cells, macrophages,
   neutrophils, dendritic cells, and CD4^+ T cells, in selected cancer
   samples.

2.9. Survival analysis

   The Kaplan–Meier plotter ([82]https://kmplot.com/analysis/) is a
   popular web server that evaluates the correlation between a gene of
   interest and patient survival in various cancer types [[83]27]. The
   patient samples are split into two groups on the basis of the median
   expression of our gene of interest. Biased arrays were excluded, and a
   P value < 0.05 was considered significant.

3. Results

3.1. Identification and analysis of shared DEGs associated with SLE and lung
cancer

   We collected gene expression datasets of SLE and lung cancer patients
   and filtered the significant DEGs between the control and disease
   groups in each dataset via GEO2R with cutoff thresholds of P
   value < 0.05 and |log2FC| >1. There were 878 DEGs between SLE patients
   and healthy controls, with 117 genes downregulated and 761 genes
   upregulated, whereas 326 DEGs were identified between the lung cancer
   samples and the normal samples, including 160 downregulated genes and
   166 upregulated genes ([84]Fig. 2A and B; see also Additional file 2:
   [85]Table S1 and Table S2). A Venn diagram, consisting of 24
   upregulated genes and five downregulated genes, was used to display the
   DEGs shared between SLE patients and lung cancer patients ([86]Fig.
   2C).

Fig. 2.

   [87]Fig. 2
   [88]Open in a new tab

   Identification and analyses of DEGs from the SLE and lung cancer
   datasets. A The volcano plot presenting DEGs in [89]GSE72509. B The
   volcano plot presenting DEGs in [90]GSE42830. C Venn diagram showing
   the intersected DEGs of SLE and lung cancer. D GO analysis for shared
   DEGs.

   To gain an initial understanding of the molecular mechanisms involved
   in SLE and lung cancer, we imported these 29 shared DEGs into the DAVID
   online database for functional analysis. GO analysis revealed that the
   innate immune response and inflammatory response were significantly
   enriched ([91]Fig. 2D).

3.2. Construction of a weighted gene coexpression network and determination
of key modules in SLE and lung cancer

   Weighted gene coexpression network analysis (WGCNA) was performed to
   identify key modules associated with SLE and lung cancer. First, we
   excluded three outlier samples, including two SLE samples and one
   healthy control sample. A soft-thresholding power of 20 was selected on
   the basis of the average connectivity and scale independence ([92]Fig.
   3A). A total of 32 modules associated with SLE were generated, and a
   heatmap was used to visualize the correlation between the modules and
   SLE ([93]Fig. 3B). The MEmagenta module (r = 0.47, p = <0.05) and the
   MEdark turquoise module (r = 0.42, p < 0.05) were identified as the
   modules most positively correlated with SLE and were chosen as the key
   modules for further analysis of the [94]GSE72509 dataset. For
   [95]GSE42830, no outliers were detected, and a soft-thresholding power
   of 6 was selected to generate the network ([96]Fig. 3C). The
   MEturquoise module was most positively correlated with lung cancer
   (r = 0.66, p < 0.05) and was selected for subsequent exploration
   ([97]Fig. 3D). The intersection of the MEmagenta and MEdarkturquoise
   modules from [98]GSE72509 and the MEturquoise module from [99]GSE42830
   resulted in 125 shared module genes ([100]Fig. 4A).

Fig. 3.

   [101]Fig. 3
   [102]Open in a new tab

   WGCNA for identification of genes and modules positively correlated
   with SLE and lung cancer. A The scale-free topology model for
   identifying optimal soft power threshold in SLE dataset. B The heatmap
   illustrating the module-traits relationship of SLE. C The scale-free
   topology model for identifying optimal soft power threshold in lung
   cancer dataset. D The heatmap illustrating the module-traits
   relationship of lung cancer.

Fig. 4.

   [103]Fig. 4
   [104]Open in a new tab

   Identification and functional enrichment analysis of key modules genes
   in SLE and lung cancer. A The Venn diagram presenting the intersection
   of most positively correlated genes of SLE and lung cancer. B GO and
   KEGG enrichment analysis of shared modules genes.

   The gene ontology and KEGG pathway enrichment analysis of these shared
   module genes revealed significant involvement in immune-related
   biological processes, such as defense response to virus, positive
   regulation of NF-kappaB signaling, and cytokine-mediated signaling
   pathways ([105]Fig. 4B). Enriched cellular components include the
   cytosol, NLRP1 inflammasome complex, and plasma membrane components.
   Molecular function terms were enriched for RNA binding, receptor
   binding, and protein domain-specific binding. KEGG pathway analysis
   further highlighted enrichment in immune signaling pathways, including
   NOD-like receptor signaling, NF-kappaB signaling, and RIG-I-like
   receptor signaling pathways, indicating critical roles in inflammation
   and cellular signaling mechanisms.

3.3. Hub gene identification and gene set enrichment analysis

   To identify the shared DEGs primarily linked to both SLE and lung
   cancer, we intersected the genes from the MEmagenta and MEdarkturquoise
   modules of [106]GSE72509, the MEturquoise module of [107]GSE42830, and
   the shared DEGs. As shown in [108]Fig. 5, five shared DEGs associated
   with SLE and lung cancer were identified: AIM2 (absent in melanoma 2),
   ANKRD22(ankyrin repeat domain 22), CLEC4D (C-type lectin domain family
   4 member D), SLC26A8(solute carrier family 26-member 8), and TNFAIP6
   (TNF alpha induced protein 6). To elucidate the functional relevance of
   the five hub genes, we conducted GSEA based on transcriptome-wide
   Spearman's correlation profiles in SLE and lung cancer datasets. In
   SLE, all five genes consistently showed strong enrichment in
   immune-related hallmark pathways, including interferon alpha and gamma
   responses, inflammatory response, TNF-α signaling, IL6-JAK-STAT3
   signaling, and complement activation, highlighting their potential
   roles in the regulation of innate and adaptive immunity ([109]Fig. 6A).
   In the lung cancer dataset, while some heterogeneity was observed, most
   notably for SLC26A8 and ANKRD22, which were more prominently associated
   with cancer-specific pathways, such as PI3K-Akt-mTOR signaling and
   mTORC1 signaling, key immune pathways such as interferon responses,
   inflammatory signaling, and TNF-α signaling remained recurrent among
   AIM2, TNFAIP6, and CLEC4D([110]Fig. 6B). These findings suggest that
   these genes may contribute to a shared immunological axis in SLE and
   lung cancer, potentially bridging chronic inflammation and tumor immune
   microenvironment modulation.

Fig. 5.

   [111]Fig. 5
   [112]Open in a new tab

   Five hub genes were obtained by intersecting the shared DEGs and key
   module genes of SLE and lung cancer.

Fig. 6.

   [113]Fig. 6
   [114]Open in a new tab

   GSEA of core shared genes in SLE (A) and lung cancer(B).

3.4. Performance of the LASSO regression algorithm and evaluation of core
shared genes

   To identify the hub genes with the strongest predictive value, a LASSO
   regression model was established. AIM2, SLC26A8, and TNFAIP6 retained
   nonzero coefficients under the regularization parameter (λ)
   corresponding to the minimum mean squared error (MSE) and the λ within
   one standard error (SE) of the minimum MSE in the SLE dataset
   ([115]Table 2). In contrast, in the lung cancer dataset, only AIM2 and
   SLC26A8 showed nonzero coefficients through cross-validation.

Table 2.

   Coefficients of hub genes calculated through LASSO regression model.
   Gene Symbol (NCBI Gene ID) Coefficients
     __________________________________________________________________

   [116]GSE72509 (SLE)
     __________________________________________________________________

   [117]GSE42830 (lung cancer)
     __________________________________________________________________

   Minimum MSE λ λ within 1-SE Minimum MSE λ λ within 1-SE
   AIM2 (9447) 0.817 0.318 1.28 0.54
   ANKRD22(118932) 0 0 0.037 0
   CLEC4D (338339) 0 0 0 0
   SLC26A8(116369) 0.474 0.127 1.09 0.32
   TNFAIP6(7130) 0.038 0.026 0 0
   [118]Open in a new tab

   To assess the diagnostic potential of AIM2 and SLC26A8 for SLE and lung
   cancer, we utilized receiver operating characteristic (ROC) curve
   analysis on the [119]GSE72509 and [120]GSE42830 datasets. The AUC
   values for both AIM2 and SLC26A8 indicated substantial diagnostic
   significance for SLE and lung cancer in the training data. To verify
   these findings, we further tested their diagnostic accuracy via the
   [121]GSE61635 dataset for SLE patients and the [122]GSE42834 dataset
   for lung cancer patients. The AUC values for AIM2 and SLC26A8 were 0.93
   and 0.85, respectively, in [123]GSE61635, whereas both biomarkers
   performed well in [124]GSE42834, highlighting AIM2 and SLC26A8 as
   promising diagnostic markers for both SLE and lung cancer ([125]Fig.
   7).

Fig. 7.

   [126]Fig. 7
   [127]Open in a new tab

   ROC curves of AIM2 and SLC26A8 on training and validating datasets to
   evaluate and validate the diagnostic values for SLE and lung cancer.

   To validate our findings at single-cell resolution, we further
   performed an integrated analysis of the [128]GSE136035 scRNA-seq
   dataset, which includes PBMCs from four SLE patients and two healthy
   controls. Analysis of the UMAP plot, incorporating SingleR-derived cell
   type annotations, demonstrates that AIM2 expression is particularly
   elevated in B cell subsets ([129]Fig. 8). To quantify transcript-level
   differences across conditions, we implemented a pseudobulk differential
   expression approach. For each sample, raw UMI counts were aggregated
   across all cells to generate sample-level gene expression matrices,
   which were then analyzed using DESeq2. This strategy accounts for
   sample-level biological variability while preserving the benefits of
   single-cell resolution. AIM2 was significantly upregulated in SLE
   samples compared to controls (log[2] fold change = 1.025, p = 0.0356),
   consistent with findings from bulk RNA-seq analysis. However, SLC26A8
   expression was not detected in the scRNA-seq dataset, likely due to its
   low or absent expression in circulating PBMCs. Additionally,
   cell-type-specific pseudobulk analysis was not feasible due to limited
   overlap in annotated immune cell subsets between the SLE and control
   groups after quality filtering.

Fig. 8.

   [130]Fig. 8
   [131]Open in a new tab

   The UMAP visualization, integrating SingleR-predicted cell types with
   AIM2 gene expression, reveals a distinct enrichment of AIM2 expression
   within B cell populations.

3.5. Immune infiltration analysis of core shared genes in lung cancer

   Given that GSEA of the hub genes is closely linked to immune
   regulation, we investigated the correlation between the expression
   levels of AIM2 and SLC26A8 and immune infiltration in lung cancer via
   TIMER. AIM2 demonstrated a significant negative correlation with tumor
   purity in both LUAD and LUSC (p < 0.05), suggesting its predominant
   expression in the tumor immune microenvironment rather than within
   tumor cells ([132]Fig. 9A). Additionally, AIM2 showed moderate positive
   partial correlations with multiple immune cell populations, especially
   neutrophils and dendritic cells. These results indicate that AIM2 may
   play an active role in shaping or responding to immune infiltration
   across lung cancer subtypes.

Fig. 9.

   [133]Fig. 9
   [134]Open in a new tab

   Immune infiltration analysis of core shared genes in LUAD and LUSC. A
   Correlation between the expression of AIM2 and tumor immune
   microenvironment. B Correlation between the expression of SLC26A8 and
   tumor immune microenvironment.

   In contrast, SLC26A8 did not show a statistically significant
   correlation with tumor purity in either LUAD or LUSC([135]Fig. 9B).
   However, partial correlation analysis revealed modest but significant
   associations with several immune cell types in both cancer types,
   including B cells, CD8^+ T cells, CD4^+ T cells, and macrophages. While
   these correlations were lower than those observed for AIM2, the data
   suggest a potential, though weaker, immunological role for SLC26A8,
   particularly in the context of innate immune cell infiltration. These
   findings highlight the differential involvement of the two genes in the
   lung cancer immune landscape, with SLC26A8 more strongly linked to
   immune cell enrichment.

3.6. Assessing core shared gene expression and clinical outcomes

   Finally, we investigated the prognostic significance of AIM2 and
   SLC26A8 in lung cancer patients via Kaplan–Meier survival analysis. The
   results revealed that higher AIM2 expression was associated with poorer
   overall survival in the LUAD subtype ([136]Fig. 10A; see also
   Additional file 4:[137]Figure S3 A). Given the female predominance of
   SLE, we further explored sex-specific differences in LUAD and found
   that elevated AIM2 expression was significantly associated with worse
   outcomes exclusively in female patients. ([138]Fig. 10B). In contrast,
   SLC26A8 expression levels were not significantly associated with
   survival outcomes when stratified by lung cancer subtypes or patient
   sex (Additional file 4:[139]Figure S3 B and C). However, higher SLC26A8
   levels were linked to poorer overall survival specifically in patients
   with a smoking history, but not in those who never smoked, suggesting
   its potential relevance in smoking-related lung cancer, where tobacco
   exposure is also a shared risk factor with SLE ([140]Fig. 10C).

Fig. 10.

   [141]Fig. 10
   [142]Open in a new tab

   Kaplan‒Meier analysis of overall survival based on AIM2 and SLC26A8
   expression. A Higher AIM2 expression was associated with poorer overall
   survival in the LUAD subtype. B Elevated AIM2 expression is linked to
   poorer overall survival exclusively in female lung cancer patients. C
   Data stratified by smoking history indicated that lower SLC26A8
   expression is significantly associated with better survival outcomes
   only in patients with tobacco exposure.

4. Discussion

   Systemic lupus erythematosus (SLE) is a chronic autoimmune disease
   characterized by systemic inflammation and immune dysregulation that
   affects multiple organs and leads to 2.87-fold greater mortality than
   the general population does [[143]28]. Moreover, epidemiological data
   revealed that SLE patients have an approximately 1.62-fold increased
   risk of developing lung cancer compared with the general population
   [[144]29], and cancer is the second leading cause of death among SLE
   patients, following cardiovascular disease [[145]30]. Over the past
   decade, emerging evidence has suggested a potential link between
   chronic inflammation and the development of various cancers [[146][31],
   [147][32], [148][33], [149][34]]. Persistent immune activation and a
   dysregulated inflammatory response in SLE are thought to create a
   microenvironment conducive to tumorigenesis [[150][35], [151][36],
   [152][37]]. A previous study further demonstrated that genetically
   predisposed SLE directly contributes to an increased risk of lung
   cancer development through Mendelian randomization analysis [[153]38].
   This elevated risk underscores the importance of investigating the
   molecular and genetic mechanisms linking SLE and lung cancer. While the
   precise mechanisms connecting SLE and lung cancer are not fully
   understood, our study aims to investigate the genetic signatures
   correlated with SLE and lung cancer, with the goal of identifying more
   targeted therapies, improving cancer surveillance in SLE patients, and
   providing insights into the shared pathways of immune dysregulation and
   carcinogenesis.

   In the present research, we used a comprehensive approach to identify
   shared gene signatures between SLE patients and lung cancer patients by
   analyzing data from the GEO database. Through the integrated DEG and
   WGCNA analyses, we identified five genes shared between the two
   diseases. These genes are enriched primarily in immune and inflammatory
   pathways, such as the interferon-gamma and interferon-alpha pathways,
   as determined through GSEA. Additionally, via the LASSO regression
   algorithm, we found AIM2 and SLC26A to be the genes most significantly
   associated with both SLE and lung cancer. ROC curves were used to
   assess the diagnostic value of AIM2 and SLC26A8 for SLE and lung
   cancer. By calculating the AUC in the training and validation datasets,
   the values above 0.7 indicated that AIM2 and SLC26A8 are potential
   biomarkers for SLE and lung cancer.

   AIM2 is one of the key pathogen recognition receptors and plays a major
   role in the innate immune response [[154]39]. Located in the cytoplasm,
   AIM2 identifies double-stranded DNA and initiates the formation of the
   inflammasome [[155]40]. This results in the activation of caspase-1 and
   the production of proinflammatory interleukin-1β (IL-1β) and IL-18
   [[156]41]. This pathway is essential for protection against microbial
   infections. However, irregularities in AIM2 expression have been linked
   to various autoimmune and inflammatory conditions, including SLE
   [[157][42], [158][43], [159][44], [160][45]]. In line with our
   findings, AIM2 levels are significantly increased in SLE patients and
   enhance B-cell differentiation by regulating the Bcl-6-Blimp-1 pathway
   [[161]46]. SLE disease severity has also been shown to be correlated
   with AIM2 expression, as inhibiting AIM2 expression significantly
   alleviated SLE symptoms by reducing macrophage activation and
   suppressing the inflammatory response in lymphocyte-derived apoptotic
   DNA lupus mice [[162]47]. In addition, the overexpression of AIM2 is
   closely related to promote ng lung tumorigenesis [[163]48]. AIM2
   inflammasome-driven IL-1β promotes HIF-1α expression via the
   NF-κB/COX-2 pathway, while its interaction with mitochondria activates
   MAPK/ERK signaling, contributing to tumor development in NSCLC
   [[164]49,[165]50].

   SLC26A8 is a chloride ion transporter primarily known for its role in
   spermatogenesis and sperm motility [[166]51,[167]52]. Although its
   function has been well documented in reproductive biology, emerging
   evidence suggests that SLC26A8 may also play a role in cancer biology
   [[168][53], [169][54], [170][55]]. In a study by Han Z et al., SLC26A8
   was found to be one of the top upregulated genes in PBMCs from patients
   with hepatocarcinoma, and SLC26A8, along with five other dysregulated
   genes in liver cancer, was significantly enriched in several pathways,
   comprising the immune response, granulocyte activation, T-cell
   activation, Toll-like receptor binding, and GTPase regulator activity
   [[171]53]. SLC26A8 is also an eosinophil-related gene correlated with
   the clinical outcome of bladder urothelial carcinoma and a potential
   susceptibility gene for hereditary nonpolyposis colorectal cancer
   [[172]54,[173]55]. Additionally, increasing research has suggested that
   SLC26A8 is associated with autoimmune diseases and inflammatory
   regulation [[174][56], [175][57], [176][58], [177][59]]. SLC26A8 was
   identified as a matrix metalloproteinase (MMPS)-related disease marker
   and plays a role in the pathogenesis of inflammatory bowel disease
   [[178]56]. A study of moderate-to-severe asthma demonstrated that
   SLC26A8 is associated with mucosal immunity, cell metabolism, and
   airway remodeling [[179]57]. Moreover, SLC26A8 shows notable diagnostic
   performance in pediatric septic shock patients and is involved in the
   infiltration of immune cells [[180]58]. An in vitro study further
   confirmed that SLC26A8 mutation triggers vasoactive neuropeptide
   production and contributes to neurogenic inflammation during the
   development of rosacea, a chronic inflammatory skin disorder [[181]59].

   With respect to lung cancer treatment, the immune microenvironment has
   become crucial in tumor progression and therapeutic resistance
   [[182]60]. In this study, we assessed the correlation between core
   shared genes, AIM2 and SLC26A8, and the immune microenvironment in lung
   cancer. Our finding that elevated expression of AIM2 is positively
   associated with an active tumor microenvironment in LUAD and LUSC is
   consistent with the findings of previous studies [[183]61,[184]62]. In
   addition, we discovered that SLC26A8 has a mild but significant
   association with the activation of B cells and CD4^+ T cells in the
   lung cancer microenvironment. These results underscore the potential of
   targeting the tumor microenvironment, particularly AIM2 and SLC26A8, as
   a promising therapeutic approach for lung cancer patients.

   We aimed to further highlight the correlation between AIM2 and SLC26A8
   expression and the clinical outcome of patients with lung cancer.
   Specifically, in female patients with lung cancer, AIM2 expression
   significantly correlated with overall survival, whereas no such
   association was observed in male patients. Our findings also revealed
   that in female lung adenocarcinoma patients, higher AIM2 expression is
   linked to worse overall outcomes. Given that the majority of SLE
   patients are female and that lung adenocarcinoma is the most common
   lung cancer subtype in SLE patients, these results suggest a potential
   sex-specific role in which AIM2 may contribute to lung cancer
   progression through mechanisms such as hormonal responses. AIM2 may
   serve as both a valuable biomarker for assessing lung cancer prognosis
   and a potential therapeutic target, particularly in female patients
   with lung adenocarcinoma. On the other hand, the association of lower
   SLC26A8 expression with better overall survival in lung cancer patients
   with a smoking history, but the lack of association in nonsmokers,
   implies a potential role of SLC26A8 in smoking-related pathways or
   inflammatory responses in lung cancer.

   There are several limitations to our study. First, the current study is
   completely based on in silico analyses. Second, the precise mechanisms
   underlying the immune responses triggered by SLC26A8 require further
   investigation. Therefore, in vivo and in vitro experiments to verify
   the role of AIM2 and SLC26A8 in the tumorigenesis of SLE patients are
   necessary. In conclusion, this study provides novel insight into the
   molecular pathway connections between SLE and lung cancer. By utilizing
   multiple analytical techniques with validation via external datasets,
   we are committed to enhancing the robustness of the results and hope to
   offer immediate translational benefits, such as biomarkers for
   diagnosis and prognosis, as well as a more comprehensive understanding
   of the molecular pathways linking SLE and lung cancer.

5. Conclusion

   The present study revealed that the chronic inflammatory state may be
   linked to the mechanisms of SLE and lung cancer and that AIM2 and
   SLC26A8 are key shared genes. While the precise mechanisms connecting
   SLE and lung cancer are not fully understood, the use of integrated
   bioinformatics to detect and validate genes shared by SLE and lung
   cancer may provide potential therapeutic targets and pave the way for
   understanding common pathogenic mechanisms.

   List of abbreviations
   AIM2      absent in melanoma 2
   ANKRD22   ankyrin repeat domain 22
   AUC       area under the curve
   CLEC4D    C-type lectin domain family 4 member D
   DEGs      differentially expressed genes
   GEO       Gene Expression Omnibus
   GO        Gene Ontology
   GSEA      Gene Set Enrichment Analysis
   IL        interleukin
   KEGG      Kyoto Encyclopedia of Genes and Genomes
   LASSO     least absolute shrinkage and selection operator
   LUAD      lung adenocarcinoma
   LUSC      lung squamous cell carcinoma
   MMPS      matrix metalloproteinases
   MSE       minimum mean squared error
   PBMCs     peripheral blood mononuclear cells
   PCA       principal component analysis
   PPI       protein‒protein interaction
   ROC       receiver operating characteristic
   scRNA-seq single-cell RNA sequencing
   SE        standard error
   SLC26A8   solute carrier family 26-member 8
   SLE       systemic lupus erythematosus
   TNFAIP6   TNF alpha induced protein 6
   WGCNA     weighted gene coexpression network analysis
   [185]Open in a new tab

CRediT authorship contribution statement

   Chueh-Hsuan Hsu: Writing – original draft, Software, Formal analysis,
   Data curation. Shuo-Chueh Chen: Resources, Methodology. Yung-Luen Yu:
   Writing – review & editing, Supervision, Project administration,
   Conceptualization.

Ethics approval and consent to participate

   Not applicable.

Consent for publication

   All the authors have read and approved the content and agree to submit
   it for consideration for publication in the journal.

Availability of data and materials

   The datasets generated during and/or analyzed during the current study
   are available in the GEO database repository
   ([186]https://www.ncbi.nlm.nih.gov/geo/).

Funding

   This research was funded by grants from the National Science and
   Technology Council, Taiwan (NSTC 110-2314-B-039-034-MY3; NSTC
   112-2320-B-039-020-; NSTC 113-2314-B-039-019-MY3; NSTC
   114-2314-B-039-067-), China Medical University, Tai-wan (CMU108-MF-01;
   CMU109-MF-03), and China Medical University Hospital, Taiwan
   (DMR-112-198; DMR-113-020).

Declaration of competing interest

   The authors declare that they have no competing interests.

Acknowledgments