Transcriptomic and experimental validation identify biomarkers and regulatory mechanisms associated with the aryl hydrocarbon receptor in nasopharyngeal carcinoma

Introduction

Nasopharyngeal carcinoma (NPC), a malignancy arising from nasopharyngeal epithelial cells, has a pathogenesis attributed to multifactorial interactions and is marked by high invasiveness (1). In 2022, global estimates recorded 120,434 new cases and 73,482 deaths from NPC, posing a significant threat to human health (2). Consistently higher incidence rate, however, is a long-standing feature of southern China’s Cantonese population, while intermediate rates characterize indigenous groups in Southeast Asia, the Arctic, North Africa, and the Middle East (3). The cause of NPC remains unknown at present. Studies have shown that the possible causes of NPC include Epstein-Barr virus (EBV) infection, genetic factors and environmental factors. Some unhealthy habits such as heavy smoking and consumption of pickled foods may also trigger the occurrence of this disease (3,4). The primary treatment modalities for NPC consist of radiotherapy, chemotherapy, and targeted therapy, with radiotherapy serving as the preferred approach. Chemotherapy is often combined with radiotherapy and can use drugs such as cisplatin injection, which helps to reduce the tumor size and control distant metastasis. Targeted therapy, such as nimotuzumab, can specifically act on the cancer cell growth signaling pathways and is effective for some advanced patients (5-7). However, the prognosis of patients shows significant heterogeneity (such as a wide range of survival periods), and the existing classification criteria are insufficient for predicting treatment responses. Therefore, in-depth exploration of the pathogenic mechanism of NPC is of crucial clinical significance for improving the long-term prognosis of patients.

The aryl hydrocarbon receptor (AhR) gene encodes a ligand-activated transcription factor that is ubiquitously expressed in various tissues (8). The mechanism of action involves binding to aromatic hydrocarbon compounds such as dioxins and polycyclic aromatic hydrocarbons. After activation, it moves from the cytoplasm to the nucleus, forms a heterodimer with the aryl hydrocarbon receptor nuclear translocator (ARNT), and binds to xenobiotic response elements to regulate the expression of downstream genes (9,10). The AhR has multiple functions in the tumor microenvironment (TME) that affect both tumor cells and immune cells (11). The activation of AhR has been proven to be carcinogenic through its effects such as promoting cell proliferation, migration, and inhibiting cell apoptosis (12). Studies have shown that AhR/aryl hydrocarbon receptor nuclear translocator protein 2 (ARNT2) is overexpressed in NPC and regulates prostaglandin-endoperoxide synthase 2 (PTGS2), which is involved in the cancer process (13). Cisplatin inhibits indoleamine 2,3-dioxygenase 1 (IDO1) activity, preventing tryptophan-to-kynurenine (Kyn) conversion via the kynurenine pathway. This action suppresses the IDO1-Kyn-AhR axis and reduces NPC cell migration and invasion (14). This study integrated public transcriptome datasets, employing differential expression analysis, univariate Cox regression, and machine learning methods to identify key AhR-linked genes in NPC. Furthermore, by combining immune infiltration, molecular regulatory networks and drug sensitivity analysis, it systematically clarified the role and potential mechanisms of candidate biomarkers in the NPC microenvironment, and conducted experimental verification using reverse transcription quantitative polymerase chain reaction (RT-qPCR). These findings offer novel molecular insights and potential targets for intervention, thereby contributing to a deeper understanding of NPC pathogenesis, improved prognosis assessment, and the advancement of targeted therapies. We present this article in accordance with the REMARK reporting checklist (available at https://tcr.amegroups.com/article/view/10.21037/tcr-2026-0879/rc).

Methods

Data correlation

GSE53819 (GPL6480) from Gene Expression Omnibus (GEO, https://www.ncbi.nlm.nih.gov/geo/) was used as the training set, with 18 NPC and 18 control tissues. The validation dataset GSE102349 (GPL11154) was obtained from GEO and contained 113 NPC samples, of which 88 had prognostic data, while 25 were excluded due to missing survival data. The clinical information table was presented in https://cdn.amegroups.cn/static/public/tcr-2026-0879-1.xlsx. A gene set consisting of 215 aryl hydrocarbon receptor-related genes (AhRGs) was obtained from the GenomeNet (https://www.genome.jp/) (15). The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments.

Differential expression analysis

Using the “limma” package (v 3.54.0) (16), differentially expressed genes [DEGs; |log₂fold change (FC)| >0.5, P adjust <0.05] were screened from the GSE53819 training set. The results were visualized with the “ggplot2” package (v 3.4.1) (17) in a volcano plot depicting the top 10 |log₂FC| genes per direction, and with the “pheatmap” package (v 1.0.12) (18) to show expression of the top 20 DEGs. The “VennDiagram” package (v 1.7.3) (19) then identified candidate genes from the intersection of DEGs and AhR-related sets. Finally, correlations among candidates (|correlation coefficient (cor)| >0.3, P<0.05) were analyzed using the “Hmisc” package (v 5.1.3) (20) to assess their role in tumor progression.

Function enrichment analysis and construction of protein-protein interaction (PPI) network

To elucidate the functional mechanisms of candidate genes in the disease, Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analyses were conducted using the “clusterProfiler” package (v 4.2.2) (21). GO analysis annotated functions across biological processes (BP), cellular components (CC), and molecular functions (MF), with Benjamini-Hochberg correction (P adjust <0.05). For visualization, the top 5 GO terms per category were extracted. KEGG enrichment analysis (P adjust <0.05) identified the top 5 pathways, and protein interactions were modeled in STRING (http://string-db.org, score >0.4) with the network rendered in Cytoscape (v 3.8.2) (22).

Identification of biomarkers

Candidate genes were assessed for prognosis in GSE102349 NPC samples. Survival-associated genes [hazard ratio (HR) ≠1, P<0.05] were identified by univariate Cox regression (“survival“ v 3.7.0) (23) visualized using “forestplot“ (v 3.1.5) (24), with the proportional hazards (PH) assumption confirmed via the cox.zph function (P>0.05). To address gene collinearity, the genes that satisfied the PH assumption were subjected to least absolute shrinkage and selection operator (LASSO) regression analysis using the “glmnet“ package (v 4.1.8) (25) to reduce the number of genes. Candidate biomarkers were screened through ten-fold cross-validation. Based on median expression, GSE102349 samples were categorized, and Kaplan-Meier (K-M) curves plotted with “survminer” (v 0.4.9) (26) assessed progression-free survival (PFS) differences by log-rank test (P<0.05), establishing significant genes as validated biomarkers.

Gene multiple association network integration algorithm (GeneMANIA)

To analyze the interactions and functional associations between biomarkers and their functionally similar genes in greater detail, first, the list of biomarkers was imported into the GeneMANIA online database (http://genemania.org/), with the analyzed species explicitly specified as “Homo sapiens” (human). Subsequently, this tool was used to construct a detailed interaction network, which intuitively revealed the complex relationships between genes.

Functional similarity analysis and subcellular localization

Using the “GOSemSim” package (v 2.33.0) (27), functional similarity scores were calculated for NPC biomarkers (threshold >0.5), and FASTA sequences were obtained from the Gene database (https://www.ncbi.nlm.nih.gov/gene/). Subsequently, the mRNA Locater database (http://bio-bigdata.cn/mRNALocater/) was utilized, and data of these biomarkers were inputted for subcellular localization prediction. Finally, this information was presented in the form of a subcellular localization map, thereby enabling an intuitive understanding of the specific distribution locations of the biomarkers within cells.

Immune and matrix scoring

Estimation of STromal and Immune cells in MAlignant Tumor tissues using Expression data (ESTIMATE)-derived immune and stromal scores were employed to quantify immune/stromal cell abundance and immune infiltration levels, providing an integrated assessment of tumor purity in the tissue microenvironment. First, the “estimate“ (v 1.0.13) (28) and “limma“ packages (v 3.54.0) were loaded: the former was used for score calculation, and the latter for data normalization. Co-expressed genes were screened using the filterCommonGenes() function, and a .gct file conforming to the ESTIMATE input format was generated. To compare score distributions between biomarker expression groups (P<0.05), Immune, Stromal, ESTIMATE scores, and Tumor Purity were derived with estimateScore(), and the “ggplot2” package (v 3.4.1) was used for box or violin plot generation.

Immune checkpoints and immunotherapy response analysis

To compare the expression of 26 immune checkpoint genes and assess immunotherapy response in GSE102349 NPC samples (29), Wilcoxon test (P<0.05) and the tumor immune dysfunction and exclusion (TIDE) tool (http://tide.dfci.harvard.edu/) were employed, respectively, with the rank sum test used for TIDE score comparisons (P<0.05).

Molecular regulatory network

The miRDB database (https://mirdb.org/) was employed to predict microRNA (miRNA) targets for the biomarkers. Based on prediction scores, the top 20 miRNA-mRNA regulatory relationships were screened and extracted. Subsequently, the Encyclopedia of RNA Interactomes database (ENCORI, https://rnasysu.com/encori/) was utilized to perform association analysis between miRNAs and long non-coding RNAs (lncRNAs), and the top 10 lncRNAs significantly associated with the target miRNAs were selected. Based on the competing endogenous RNA (ceRNA) theory, it was hypothesized that lncRNAs indirectly regulate the mRNA expression of biomarkers by competitively binding to miRNAs. Finally, the two types of relationships (miRNA-mRNA and miRNA-lncRNA) were integrated and organized into a unified format. Upon ensuring uniform node identifiers, a multi-level mRNA-miRNA-lncRNA network was assembled and its structure rendered using Cytoscape (v 3.8.2).

Immune infiltration analysis

To compare immune cell infiltration, single-sample gene set enrichment analysis (ssGSEA) (30) was performed on GSE102349 NPC samples, evaluating 28 immune subsets between biomarker expression groups. Differential immune cells determined by Wilcoxon test (P<0.05) were plotted with “ggplot2” (v 3.4.1). Biomarker-immune infiltration correlations were analyzed via Spearman method using the “psych” package (v 2.1.6) (31), and a heatmap presented significant associations (|cor| >0.3, P<0.05).

Drug susceptibility analysis

Chemotherapy sensitivity profiles were compared between biomarker expression groups using drug response data from the Genomics of Drug Sensitivity in Cancer (GDSC) database (https://www.cancerrxgene.org), covering eight NPC-relevant agents such as cisplatin and gemcitabine. Half-maximal inhibitory concentration (IC₅₀) values were predicted in NPC samples using the “pRRophetic” package (v 0.5) (32). Wilcoxon test (P<0.05) then compared these values between expression groups in GSE102349 to assess sensitivity variations.

Expression levels of biomarkers

To compare biomarker levels in tumor versus adjacent normal tissues from GSE53819, the Wilcoxon rank-sum test (P<0.05) was employed.

RT-qPCR

This study used the immortalized human nasopharyngeal epithelial cell line NP69 and the human NPC cell lines NPC/HK1, HONE-1, and 5-8F (AoRuiCell, Shanghai, China). All cells were cultured in RPMI-1640 medium supplemented with 10% fetal bovine serum and 1% penicillin/streptomycin and maintained in a humidified incubator at 37 ℃ with 5% CO₂. When cell confluence reached 80–90%, cells were digested with 0.25% trypsin and passaged (Figure S1). Total RNA was extracted using the TRIzol method. After RNA concentration and purity were measured with a NanoPhotometer N50, 2 µg of RNA was reverse-transcribed into cDNA using the HP All-in-one qRT Master Mix II kit according to the manufacturer’s instructions. qPCR amplification was performed on a CFX96 real-time quantitative PCR instrument using 2× Universal Blue SYBR Green qPCR Master Mix, with glyceraldehyde-3-phosphate dehydrogenase (GAPDH) serving as the internal reference gene. Three technical replicates were set for each sample. The PCR conditions were as follows: pre-denaturation at 95 ℃ for 1 min, followed by 40 cycles of 95 ℃ for 20 s, 55 ℃ for 20 s, and 72 ℃ for 30 s. Primer sequences are listed in Table 1. Primer specificity was confirmed by a single peak in the melting curve (Table S1). The relative expression levels of target genes were calculated using the 2^-∆∆Ct method, and the results were expressed as mean ± standard deviation. P value <0.05 was considered statistically significant. As this study involved only established human cell lines, public datasets, and commercially available reagents, ethical approval was not required.

Statistical analysis

In the R environment (v 4.2.2), the Wilcoxon test or t-test was applied for between-group comparisons, adopting P<0.05 as the threshold for statistical significance.

Results

Identification of candidate genes

DEG screening identified 4,611 genes (1,967 upregulated, 2,644 downregulated; |log₂FC| >0.5, P adjust <0.05) (Figure 1A,1B). Overlap of these DEGs with 215 AhRGs produced 54 candidates (Figure 1C). Among these, GSTA1, GSTA2, GSTA3, and GSTA5 showed the strongest inter-correlation (|cor| >0.3, P<0.05) (Figure 1D). A total of 584 signaling pathways were enriched through GO analysis, including 507 BPs, 11 CCs, and 66 MFs. Among these, the top 5 BP pathways ranked by enrichment degree were “epithelial cell proliferation”, “response to xenobiotic stimulus”, “cellular response to peptide hormone stimulus”, “cellular response to peptide”, and “regulation of epithelial cell proliferation”. The top 5 CC pathways were “transferase complex (transferring phosphorus-containing groups)”, “extrinsic component of membrane”, “intercellular bridge”, “RNA polymerase II transcription regulator complex”, and “membrane raft”. Among MF pathways, the top 5 (P adjust <0.05) included glutathione transferase activity, transferase activity (alkyl/aryl groups), DNA-binding transcription factor binding, steroid binding, and nuclear receptor activity. In KEGG enrichment, 202 pathways were identified, and the five most significant (P adjust <0.05) were chemical carcinogenesis-receptor activation, hepatocellular carcinoma, chemical carcinogenesis-reactive oxygen species, human cytomegalovirus infection, and Breast cancer (Figure 1E). In the PPI network, MAPK3 and SRC emerged as hub nodes, both with a degree of 29. A total of 344 interaction relationships were identified among 53 proteins, among which one discrete gene was detected (interaction score >0.4) (Figure 1F).

Figure 1 Candidate genes related to NPC. (A) Volcano plot of DEGs. The red dots represent the up-regulated genes and the blue dots represent the downregulated genes, while gray dots indicate genes with no significant expression correlation (P adjust <0.05, |log₂FC| >0.5). (B) Heatmap of top 10 upregulated and downregulated DEGs. (C) Venn diagram of candidate genes. The red circles indicate DEGs, blue-green circles denote AhRGs, and the overlapping genes in the center appear in both categories. Numbers within circles represent gene counts, while percentages in parentheses indicate each gene’s proportion among all genes. (D) Correlation heatmap of candidate genes. The colour in the top-left corner of each square represents the correlation coefficient (cor), while the colour in the bottom-right corner denotes the P value [|correlation coefficient (cor)| >0.3, P<0.05]. (E) GO enrichment analysis and main KEGG pathways of candidate genes (P adjust <0.05). The 10 most significant GO terms of BPs, MFs, and CCs are shown. (F) PPI network of candidate genes (interaction score >0.4). The size and colour intensity of each circle represent the degree value of the gene; the greater the degree value, the larger the circle and the deeper the colour. AhR, aryl hydrocarbon receptor; AhRG, aryl hydrocarbon receptor-related gene; BP, biological process; CC, cellular component; DEG, differentially expressed gene; FC, fold change; GO, Gene Ontology; KEGG, Kyoto Encyclopedia of Genes and Genomes; MF, molecular function; NPC, nasopharyngeal carcinoma; PPI, protein-protein interaction.

Cholinergic receptor nicotinic beta 2 subunit (CHRNB2), adenylate cyclase 4 (ADCY4), and cell division cycle 6 (CDC6) were identified as biomarkers

Eleven survival-related genes were identified through univariate Cox analysis, namely PRKCA, ADCY4, PIK3CB, FGF7, CDC6, PIK3R3, BIRC5, CDC25A, ADCY2, CHRNB2, and STAT3 (HR ≠1, P<0.05) (Figure 2A). The HR values and their 95% confidence intervals for each gene were presented in Table 2. The PHs assumption test excluded FGF7 due to a P value <0.05, while the other 10 genes (P>0.05) were retained for further analysis (Figure 2B-2L). The results of the PH assumption test were presented in Table S2. Ten genes assumed based on PH were subjected to LASSO analysis, and five candidate biomarkers were obtained, namely CHRNB2, ADCY2, ADCY4, CDC6, and PRKCA (lambda.min =−4.143) (Figure 2M,2N). Log-rank P values from K-M curves were 0.014 (CHRNB2), 0.041 (ADCY4), and 0.003 (CDC6), supporting their identification as biomarkers (P<0.05) (Figure 2O-2Q).

Figure 2 CHRNB2, ADCY4, and CDC6 were identified as biomarkers. (A) Univariate Cox analysis. Each red dot corresponds to a gene; dots positioned to the right indicate HR >1, while those to the left denote HR <1 (HR ≠1, P<0.05). (B-L) The PH assumption. (M,N) LASSO regression analysis. (M) The left figure depicts the LASSO logical coefficient penalty plot, with the horizontal axis representing log(Lambda) and the vertical axis indicating cross-validation error. Lambda 1se denotes the optimal lambda value. Each curve in the figure traces the variation path of an individual independent variable coefficient, with the y-axis representing the coefficient value and the x-axis denoting the number of non-zero coefficients in the model at that point. (N) The right-hand figure displays the lambda selection plot for the model, where red dots indicate the MSE and its upper and lower one-standard-deviation bounds. A lower MSE signifies a superior model; the number above each dot indicates the remaining independent variables in the model. The first dashed line marks the point of minimum MSE. (O-Q) Survival differences analysis between up- and down-regulated groups of genes (P<0.05). ADCY4, adenylate cyclase 4; CDC6, cell division cycle 6; CHRNB2, cholinergic receptor nicotinic beta 2 subunit; CI, confidence interval; HR, hazard ratio; LASSO, least absolute shrinkage and selection operator; MSE, mean squared error; NPC, nasopharyngeal carcinoma; PH, proportional hazard.

Functional similarity analysis and subcellular localization

Based on the prediction results from the GeneMANIA database, CHRNB2, ADCY4, and CDC6 were predicted to have physical interactions with multiple genes. For example, CHRNB2 was predicted to interact with genes such as CHRNA5, CHRFAM7A, and CHRNA3. These results suggested that potential interactions and functional associations might exist between the biomarkers and their functionally similar genes (Figure 3A). Functional similarity analysis showed that ADCY4 and CDC6 had a relatively high similarity score (0.59, threshold >0.5) (Figure 3B). Subcellular localization predictions indicated that CDC6 was mainly localized in the nucleus, whereas ADCY4 and CHRNB2 were mainly localized in the cytoplasm (Figure 3C).

Figure 3 Functional similarity analysis and subcellular localization. (A) GeneMANIA. The inner circle comprises three biomarkers, with colours within it representing distinct functional categories. The outer circle displays genes associated with the biomarkers’ functions. Connections between the inner and outer circles indicate functional relationships or interactions among genes, with differing line colours signifying varying degrees of interaction. A greater number of connections denotes a stronger relationship between genes. (B) Heatmap of functional similarity among biomarkers. (C) Subcellular localization. ADCY4, adenylate cyclase 4; CDC6, cell division cycle 6; CHRNB2, cholinergic receptor nicotinic beta 2 subunit; GeneMANIA, gene multiple association network integration algorithm.

Immune and matrix scoring

Elevated immune, stromal, and ESTIMATE scores were associated with high ADCY4 expression, in contrast to CDC6, which showed a reverse correlation with these scores. In CHRNB2, high expression was associated with lower immune and ESTIMATE scores (P<0.05) (Figure 4). Thus, high expression of ADCY4 might have indicated immune-enriched TME features, whereas high expression of CDC6 and CHRNB2 was associated with insufficient immune cell infiltration, and this association might have been linked to tumor immune escape and poor prognosis.

Figure 4 Immune and matrix scoring. Red indicates the biomarker high-expression group; blue denotes the low-expression group. ns, P≥0.05; **, P<0.01; ***, P<0.001. ADCY4, adenylate cyclase 4; CDC6, cell division cycle 6; CHRNB2, cholinergic receptor nicotinic beta 2 subunit; ns, not significant.

Immune checkpoints and immunotherapy response analysis

BTN2A1, CD274, CD276, CD47, LGALS9, PVR, and SIRPA levels were higher in the CHRNB2 high-expression group, while BTL1 expression was greater in the low-expression group (P<0.05) (Figure 5A). CDC6 expression groups showed differential immune checkpoint gene profiles: eight genes (e.g., BTLA, CTLA4, PDCD1) were enriched in the low-expression group, and seven genes (e.g., BTN2A1, CD276, LGALS9) were enriched in the high-expression group (Figure 5B). High ADCY4 expression correlated with increased expression of ten immune checkpoint genes (including BTN2A2, CD160, TNFRSF14), while CD276 was elevated in the low ADCY4 group (P<0.05) (Figure 5C). Patients with high CDC6 (P<0.001) and CHRNB2 (P<0.001) expression also displayed higher TIDE scores (Figure 5D), and this suggested that the likelihood of benefiting from immunotherapy was lower. (P<0.05).

Figure 5 Immune checkpoints and immunotherapy response analysis. (A-C) Differential analysis of immune checkpoint gene expression in high- and low-expression biomarker groups. The horizontal axis represents immune checkpoints, while the vertical axis denotes expression levels. (D) Differential analysis of TIDE scores in high- and low-expression groups of biomarkers. Red indicates high-expression groups, and blue denotes low-expression groups. *, P<0.05; **, P<0.01; ***, P<0.001. ADCY4, adenylate cyclase 4; CDC6, cell division cycle 6; CHRNB2, cholinergic receptor nicotinic beta 2 subunit; TIDE, tumor immune dysfunction and exclusion.

Molecular regulatory network

The mRNA-miRNA-lncRNA molecular regulatory network constructed based on CDC6 and CHRNB2 included 9 miRNAs (hsa-miR-26a-5p, hsa-miR-26b-5p, hsa-miR-491-5p) and 31 lncRNAs (GAS5, SNHG6, MALAT1) (Figure 6). The network suggested that the biomarkers might have been involved in the pathogenesis of NPC through related lncRNAs and miRNAs, but the specific regulatory relationships still required experimental verification.

Figure 6 Molecular regulatory network. Rectangles denote biomarkers, rhombuses represent miRNAs, and ellipses denote lncRNAs. CDC6, cell division cycle 6; CHRNB2, cholinergic receptor nicotinic beta 2 subunit; lncRNAs, long non-coding RNAs; miRNAs, microRNAs.

Immune infiltration and drug susceptibility analyses

ADCY4, CDC6, and CHRNB2 expression groups differed significantly in immune cell infiltration, with 13, 25, and 18 differential cell types, respectively (P<0.05; Figure 7A). The following significant associations were observed: ADCY4 with plasmacytoid dendritic cells (cor =0.48, P<0.001), CDC6 with central memory CD4 T cells (cor =−0.75, P<0.001), and CHRNB2 with activated B cells (cor =−0.50, P<0.001) (|cor| >0.3, P<0.05; Figure 7B). Immune infiltration analysis revealed that high CDC6 expression correlated inversely with ssGSEA scores, suggesting that it might have been associated with the suppression of immune activity. In the in silico drug sensitivity analysis based on the pRRophetic/GDSC databases, high ADCY4 was associated with elevated IC₅₀ for gemcitabine, mitomycin C, docetaxel, and cisplatin; high CDC6 with reduced IC₅₀ for gemcitabine, methotrexate, docetaxel, and vinorelbine; and high CHRNB2 with lower IC₅₀ for gemcitabine, mitomycin C, docetaxel, 5-fluorouracil, and cisplatin (P<0.05) (Figure 7C). It should be emphasized that these drug sensitivity analyses were in silico hypothesis-generating results, and the predicted IC₅₀ values could not be considered as evidence for clinical therapeutic benefit; their accuracy still required experimental validation through studies with modulation of the candidate genes

Figure 7 Immune infiltration analysis drug susceptibility analysis. (A) The top, middle and bottom represent three biomarkers. The horizontal axis denotes immune cells, while the vertical axis shows numerical values (0–1). Red indicates the high-expression group, and blue denotes the low-expression group. (B) Heatmap of correlation analysis between biomarker expression and ssGSEA score of immune cell infiltration. The vertical axis represents immune cells, while the horizontal axis denotes biomarkers. A deeper shade of red indicates a stronger correlation, whereas a deeper shade of blue signifies a weaker correlation (|cor| >0.3, P<0.05). (C) Differences in IC₅₀ values for various anticancer drugs across biomarker high- and low-expression groups. The horizontal axis represents drugs, the vertical axis represents IC₅₀ values, blue denotes the high-expression group, and orange denotes the low-expression group. *, P<0.05; **, P<0.01; ***, P<0.001; ****, P<0.0001. ADCY4, adenylate cyclase 4; CDC6, cell division cycle 6; CHRNB2, cholinergic receptor nicotinic beta 2 subunit; IC₅₀, half-maximal inhibitory concentration; ssGSEA, single-sample gene set enrichment analysis.

Expression levels of biomarkers

The results of the biomarkers expression analysis indicated that ADCY4 was underexpressed in NPC samples compared with normal samples, while CDC6 and CHRNB2 were overexpressed in NPC samples relative to normal samples (P<0.05) (Figure 8A). Compared to nasopharyngeal epithelial cells, ADCY4 was downregulated in NPC cell lines (NPC/HK1, HONE-1, and 5-8F), while CDC6 and CHRNB2 were upregulated, as shown by RT-qPCR (Figure 8B-8D, Table 3).

Figure 8 The expression differences of ADCY4, CDC6, and CHRNB2 between normal and NPC samples were identified and verified in cell lines using RT-qPCR. (A) Expression levels of biomarkers. Analysis of biomarker expression levels in training set GSE53819. The horizontal axis represents biomarkers, while the vertical axis denotes expression levels. Red indicates the NPC disease group, and blue denotes the control group. (B-D) RT-qPCR results for biomarkers. The x-axis represents the NPC control strain NP69 and NPC strains: NPC/HK1, HONE-1, and 5-8F. The y-axis represents gene expression levels. ns, P≥0.05; *, P<0.05; **, P<0.01; ***, P<0.001. ADCY4, adenylate cyclase 4; CDC6, cell division cycle 6; CHRNB2, cholinergic receptor nicotinic beta 2 subunit; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; NPC, nasopharyngeal carcinoma; ns, not significant; ns, not significant; RT-qPCR, reverse transcription quantitative polymerase chain reaction.

Discussion

NPC is a respiratory epithelial malignancy with distinctive epidemiological patterns, showing pronounced regional, ethnic, and gender-based variations in disease distribution (33). Although the mortality of NPC has been decreasing in recent years, approximately 10–20% of patients still experience recurrence following initial treatment. Most targeted therapies remain in the preclinical stage, and their precise mechanisms require further investigation (6). As a critical nuclear transcription factor, the AhR signaling pathway plays an essential role in tumor progression and development. Research has demonstrated that AhR and its transcriptional co-activator ARNT2 are aberrantly overexpressed in NPC, where they promote disease progression through the regulation of downstream targets such as PTGS2. This suggests that the AhR signaling axis is a potential therapeutic target for NPC (13). This study successfully screened and verified three NPC biomarkers closely related to the AhR signaling pathway through the integrated analysis of public transcriptome data, combined with univariate Cox regression, LASSO machine learning, and multi-dimensional functional validation. These biomarkers are CHRNB2, CDC6, and ADCY4. Among them, ADCY4 is expressed at a lower level in NPC, while CDC6 and CHRNB2 are expressed at a higher level in NPC. These biomarkers are potential correlated with patient prognosis and immune microcellular infiltration characteristics, and are also associated with specific non-coding RNA regulatory networks and chemotherapy drug sensitivity. This finding provides novel insights into AhR function in NPC progression and offers potential molecular targets for AhR signaling‑directed therapeutic strategies.

ADCY4 is located at the position 14q12 on the human chromosome. ADCY4 can be regulated by various extracellular signals through its interaction with membrane-bound G protein-coupled proteins. These signals include hormones, neurotransmitters, and glucose, etc. The ADCY4 protein, by embedding its adenylate cyclase active domain in the cell membrane, converts adenosine triphosphate (ATP) into cyclic adenosine monophosphate (cAMP), thereby regulating various cellular functions (34). ADCY4 catalyzes the increase of cAMP levels in immune cells by enhancing the ATP concentration. This induces broad anti-inflammatory responses and stimulates regulatory mediators like IDO1 in T cells. The IDO1/tryptophan 2,3-dioxygenase 2 (TDO2)-derived kynurenine (Kyn) activates AhR, driving regulatory T cell differentiation and shaping an immunosuppressive TME (35,36). Studies indicate that ADCY4 exerts tumor-suppressive effects in breast cancer by inhibiting the focal adhesion kinase (FAK)/protein kinase B (AKT) and extracellular signal-regulated kinase (ERK) signaling pathways. However, its expression is often silenced by DNA methylation, suggesting that its downregulation in NPC may also involve epigenetic regulation, resulting in weakened cAMP/protein kinase A (PKA) signaling and subsequently affecting cell proliferation and survival (35). It is worth noting that there is a potential bidirectional regulatory relationship between ADCY4 and the AhR signaling pathway. As the key sensor for environmental and metabolic signals, the activation of AhR can be regulated by metabolites such as canadine. The ADCY4-cAMP-PKA axis may affect the transcriptional activity of AhR or coordinate with it to regulate downstream target genes. On the other hand, AhR has also been confirmed to regulate the expression of immune-related genes by influencing the binding duration of transcription factors such as signal transducer and activator of transcription 6 (STAT6), suggesting that in the NPC microenvironment, the downregulation of ADCY4 may be associated with the disorder of immune regulation mediated by AhR (37,38).

CDC6, located at the position of human chromosome 17q21.2. It is one of the main proteins that constitute the pre-replication complex (Pre-RC), regulating the transition of cells from the G1 phase to the S phase. It also participates in the activation and maintenance of the mitotic S-M checkpoint mechanism. Recent studies have also identified its oncogenic properties, with elevated expression observed across various tumor cells. It plays a significant role in tumorigenesis and progression, potentially involving the INK4/ARF signaling pathway (39). In NPC, the upregulation of CDC6 likely promotes tumor proliferation by accelerating cell cycle progression and inducing genomic instability. Additionally, the abnormal expression of CDC6 is closely related to tumor treatment resistance. For instance, in radiotherapy resistance, the enhanced stability of CDC6 protein can promote tumor tolerance by regulating the processes of cell senescence and epithelial-mesenchymal transition. It is worth noting that there may be functional interactions between CDC6 and the AhR signaling pathway (40). AhR can activate E2F transcription factor 1 (E2F1), which induces cell proliferation through transcription. E2F1 upregulates the expression of CDC6, thereby mediating cell proliferation and migration (41). AhR indirectly regulates CDC6 expression via E2F1. By forming a complex with p53, AhR facilitates p53-mediated upregulation of the E3 ubiquitin ligase Mdm2, which then binds to CDC2, promotes its ubiquitination, and ultimately triggers apoptosis in cancer cells (42). Furthermore, insufficient phosphorylation of CDC6 impairs nuclear-to-cytoplasmic protein translocation, resulting in the accumulation of subcellular components when cells are arrested at the G0/G1 phase. This mechanism modulates senescence and epithelial-mesenchymal transition, enhances radioresistance, thereby establishing CDC6 as a promising novel target for cancer radiosensitization therapy (40). The CHRNB2 is located on human chromosome 1q21.3. Its main function is to encode the beta 2 subunit of the nicotinic cholinergic receptor. It binds to acetylcholine on the postsynaptic membrane and triggers the opening of cation channels, leading to neuronal excitation (43). Research indicates that nicotine promotes NPC cell proliferation via the alpha 7 nicotinic acetylcholine receptor (α7AChR) and multiple signaling pathways including ERK, hypoxia-inducible factor-1α, and vascular endothelial growth factor (VEGF)/pigment epithelium-derived factor (PEDF) (44). However, CHRNB2 can suppress tumor migration and invasion in pancreatic cancer by inhibiting the β-catenin pathway, indicating its potential for tissue-specific regulatory functions across different tumor types. It is notable that the susceptibility locus of smoking-related NPC is located at 15q25.1, and the AhR signal plays a crucial role in the carcinogenesis caused by smoking. It can promote carcinogenesis by inducing mediators such as adrenal medullary hormone, indicating that the upregulation of CHRNB2 may be associated with the AhR-mediated smoking-related carcinogenic pathway (43,45,46). From a mechanistic perspective, AhR and STAT3 form a synergistic regulatory network in chemical carcinogenesis. The absence of CHRNB2 can affect cell adhesion and the expression of genes related to neurodegenerative responses, indicating that CHRNB2 may be involved in regulating intercellular interactions and stress responses in the TME (47,48). Additionally, AhR regulates the transcriptional response of monocytes to interleukin-4 (IL-4) by prolonging STAT6 binding within promoter regions, suggesting its potential indirect influence on the expression patterns of neuroimmune-related genes such as CHRNB2 (38). In summary, the upregulation of CHRNB2 in NPC may not only be related to cell proliferation mediated by the nicotine receptor signaling pathway, but also may form a functional dialogue with the AhR-STAT3/STAT6 signaling network, jointly regulating tumor growth, microenvironment remodeling, and immune responses.

Due to the significant immunogenicity of malignant tumors, a large number of immune-related molecular markers [such as programmed death-ligand 1 (PD-L1)] exist in their TME. These markers can stimulate immune-mediated recognition and elimination of tumor cells. Furthermore, immune infiltration analysis showed that CDC6 expression was negatively correlated with ssGSEA scores for multiple immune cell subsets, indicating its association with suppressed immune infiltration. Consistent with this, elevated CDC6 levels in glioma patients inversely correlated with CD8⁺ T cells, underscoring its potential regulatory role in tumor immunity (49). In pancreatic cancer, CDC6 enhances CD8⁺ T cell levels by modulating matrix metalloproteinase 28 (MMP28), which in turn promotes cancer cell proliferation, migration, and invasion (50). In this study, effector memory CD8⁺ T cells demonstrated significant positive correlations with all three biomarkers. This suggests that these biomarkers may modulate immune cell functions—such as those of CD8⁺ T cells—thereby promoting cancer cell proliferation, migration, and invasion, and influencing NPC progression. Drug sensitivity analysis revealed lower IC₅₀ values for gemcitabine and docetaxel in patients with high expression of CDC6 and CHRNB2, suggesting greater treatment sensitivity and potential clinical benefit. Previous studies reported a 3‑year recurrence‑free survival rate of 85.3% with gemcitabine-cisplatin induction therapy, compared to 76.5% for standard treatment. Moreover, the addition of induction chemotherapy to chemoradiotherapy significantly improved both recurrence-free and overall survival over chemoradiotherapy alone in locoregionally advanced NPC (51). These biomarkers may thus represent novel therapeutic targets for NPC treatment.

This study initially explored the potential molecular functions of ADCY4, CDC6 and CHRNB2 in NPC, emphasized the possible role of the AhR signal in the pathogenesis, and proposed potential therapeutic targets that need to be verified. However, there are still several limitations. Firstly, the sample size of the public database used in this study is relatively small, and it has not been validated in an independent external cohort. Therefore, there may be a certain risk of overfitting in the results; the existing findings are exploratory and hypothesis-generating in nature, and their robustness and generalization ability need to be confirmed in multi-center, larger sample-size independent datasets. Secondly, the specific molecular regulatory mechanisms of these biomarkers have not been fully elucidated. The associations obtained based on GeneMANIA, miRNA-lncRNA prediction, subcellular localization prediction, immune infiltration score, and TIDE analysis can only provide clues and cannot serve as direct evidence of causality or regulatory mechanisms. Moreover, the prognostic analysis is limited by incomplete clinical annotations in the public datasets, and can only perform univariate Cox regression and LASSO screening. Due to the lack of key covariates such as age, gender, treatment regimen, and EBV-DNA, the independent prognostic value of CHRNB2, ADCY4, and CDC6 cannot be evaluated through the multivariate Cox model. In future work, we plan to collect multi-center, larger sample-size independent datasets, design strict validation procedures to evaluate the robustness and clinical translational potential of these biomarkers, and verify their independent prognostic significance through multivariate correction analysis; at the same time, we will conduct in-depth exploration of the related regulatory mechanisms through cell experiments and animal models.

Conclusions

In summary, by integrating public transcriptomic data with multiple bioinformatics approaches, this study screened and preliminarily characterized three potential biomarkers of NPC associated with the AhR signaling pathway—namely, CHRNB2, CDC6, and ADCY4. These biomarkers exhibited differential expression in NPC and showed certain correlations with immune microenvironment features and computational drug sensitivity. All these findings are exploratory results derived from data mining. They provide clues and candidate molecules for future in‑depth investigation of the functional role of the AhR pathway in NPC progression and the development of related targeted therapeutic strategies. Nevertheless, the precise mechanisms and clinical translational potential warrant further validation through experimental studies and independent cohorts.

Acknowledgments

We would like to express our sincere gratitude to all individuals and organizations who supported and assisted us throughout this research. Special thanks to the funding support from the Talent Fund of Guizhou Provincial People’s Hospital. In conclusion, we extend our thanks to everyone who has supported and assisted us along the way. Without your support, this research would not have been possible.

Footnote

Reporting Checklist: The authors have completed the REMARK reporting checklist. Available at https://tcr.amegroups.com/article/view/10.21037/tcr-2026-0879/rc

Data Sharing Statement: Available at https://tcr.amegroups.com/article/view/10.21037/tcr-2026-0879/dss

Peer Review File: Available at https://tcr.amegroups.com/article/view/10.21037/tcr-2026-0879/prf

Funding: This work was supported by Guizhou Provincial People’s Hospital Talent Fund (grant number 2024-62).

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://tcr.amegroups.com/article/view/10.21037/tcr-2026-0879/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. All experiments utilized only established human cell lines (NP69, NPC/HK1, HONE-1, and 5-8F), publicly available transcriptomic data, and commercially available reagents. No human subjects, animal experiments, or primary patient samples were involved.

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. Rueda Domínguez A, Cirauqui B, García Castaño A, et al. SEOM-TTCC clinical guideline for nasopharyngeal carcinoma (update 2025). Clin Transl Oncol 2026;28:1151-64. [Crossref] [PubMed]
2. Jang W, Kim S, Son Y, et al. Global, Regional, and National Burden of Pharyngeal Cancer and Projections to 2050 in 185 Countries: A Population-Based Systematic Analysis of GLOBOCAN 2022. J Korean Med Sci 2025;40:e177. [Crossref] [PubMed]
3. Chang ET, Ye W, Zeng YX, et al. The Evolving Epidemiology of Nasopharyngeal Carcinoma. Cancer Epidemiol Biomarkers Prev 2021;30:1035-47. [Crossref] [PubMed]
4. Chan SM, Paterson IC, Yap LF. Nasopharyngeal Carcinoma in Southeast Asia: Current Landscape and Future Priorities. Br J Biomed Sci 2025;82:15902. [Crossref] [PubMed]
5. Luo Y, Xiang X, Ma X. Clinical observational study on the efficacy of induction chemotherapy sequential concurrent radiotherapy combined with targeted therapy in patients with locally advanced EGFR-positive nasopharyngeal carcinoma: prediction model construction and efficacy testing. Eur Arch Otorhinolaryngol 2023;280:5409-16. [Crossref] [PubMed]
6. Kang Y, He W, Ren C, et al. Advances in targeted therapy mainly based on signal pathways for nasopharyngeal carcinoma. Signal Transduct Target Ther 2020;5:245. [Crossref] [PubMed]
7. Wong KCW, Hui EP, Lo KW, et al. Nasopharyngeal carcinoma: an evolving paradigm. Nat Rev Clin Oncol 2021;18:679-95. [Crossref] [PubMed]
8. Polonio CM, McHale KA, Sherr DH, et al. The aryl hydrocarbon receptor: a rehabilitated target for therapeutic immune modulation. Nat Rev Drug Discov 2025;24:610-30. [Crossref] [PubMed]
9. Sondermann NC, Faßbender S, Hartung F, et al. Functions of the aryl hydrocarbon receptor (AHR) beyond the canonical AHR/ARNT signaling pathway. Biochem Pharmacol 2023;208:115371. [Crossref] [PubMed]
10. Coumoul X, Barouki R, Esser C, et al. The aryl hydrocarbon receptor: structure, signaling, physiology and pathology. Signal Transduct Target Ther 2026;11:20. [Crossref] [PubMed]
11. Griffith BD, Frankel TL. The Aryl Hydrocarbon Receptor: Impact on the Tumor Immune Microenvironment and Modulation as a Potential Therapy. Cancers (Basel) 2024;16:472. [Crossref] [PubMed]
12. Kober C, Roewe J, Schmees N, et al. Targeting the aryl hydrocarbon receptor (AhR) with BAY 2416964: a selective small molecule inhibitor for cancer immunotherapy. J Immunother Cancer 2023;11:e007495. [Crossref] [PubMed]
13. Huang SW, Chen G, Li JD, et al. Potential Molecular Mechanism of Upregulated Aryl Hydrocarbon Receptor Nuclear Translocator 2 in Nasopharyngeal Carcinoma. Comput Math Methods Med 2022;2022:9137282. [Crossref] [PubMed]
14. Zou X, Zhao D, Wen X, et al. NLG-919 combined with cisplatin to enhance inhibitory effect on cell migration and invasion via IDO1-Kyn-AhR pathway in human nasopharyngeal carcinoma cell. Can J Physiol Pharmacol 2023;101:599-609. [Crossref] [PubMed]
15. Li Q, Li H. Integrating bioinformatics and machine learning to identify AhR-related gene signatures for prognosis and tumor microenvironment modulation in melanoma. Front Immunol 2024;15:1519345. [Crossref] [PubMed]
16. 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]
17. Gustavsson EK, Zhang D, Reynolds RH, et al. ggtranscript: an R package for the visualization and interpretation of transcript isoforms using ggplot2. Bioinformatics 2022;38:3844-6. [Crossref] [PubMed]
18. Gu Z, Eils R, Schlesner M. Complex heatmaps reveal patterns and correlations in multidimensional genomic data. Bioinformatics 2016;32:2847-9. [Crossref] [PubMed]
19. Chen H, Boutros PC. VennDiagram: a package for the generation of highly-customizable Venn and Euler diagrams in R. BMC Bioinformatics 2011;12:35. [Crossref] [PubMed]
20. Kiș AM, Buzatu R, Chisavu L, et al. Time-to-Treatment Delays and Their Prognostic Implications in Pharyngeal Cancer-An Exploratory Analysis in Western Romania. Clin Pract 2024;14:1270-84. [Crossref] [PubMed]
21. Wu T, Hu E, Xu S, et al. clusterProfiler 4.0: A universal enrichment tool for interpreting omics data. Innovation (Camb) 2021;2:100141. [Crossref] [PubMed]
22. Liu P, Xu H, Shi Y, et al. Potential Molecular Mechanisms of Plantain in the Treatment of Gout and Hyperuricemia Based on Network Pharmacology. Evid Based Complement Alternat Med 2020;2020:3023127. [Crossref] [PubMed]
23. Lei J, Qu T, Cha L, et al. Clinicopathological characteristics of pheochromocytoma/paraganglioma and screening of prognostic markers. J Surg Oncol 2023;128:510-8. [Crossref] [PubMed]
24. Li Y, Lu F, Yin Y. Applying logistic LASSO regression for the diagnosis of atypical Crohn's disease. Sci Rep 2022;12:11340. [Crossref] [PubMed]
25. Friedman J, Hastie T, Tibshirani R. Regularization Paths for Generalized Linear Models via Coordinate Descent. J Stat Softw 2010;33:1-22.
26. Liu TT, Li R, Huo C, et al. Identification of CDK2-Related Immune Forecast Model and ceRNA in Lung Adenocarcinoma, a Pan-Cancer Analysis. Front Cell Dev Biol 2021;9:682002. [Crossref] [PubMed]
27. Yu G. Gene Ontology Semantic Similarity Analysis Using GOSemSim. Methods Mol Biol 2020;2117:207-15. [Crossref] [PubMed]
28. Becht E, Giraldo NA, Lacroix L, et al. Estimating the population abundance of tissue-infiltrating immune and stromal cell populations using gene expression. Genome Biol 2016;17:218. [Crossref] [PubMed]
29. Ru B, Wong CN, Tong Y, et al. TISIDB: an integrated repository portal for tumor-immune system interactions. Bioinformatics 2019;35:4200-2. [Crossref] [PubMed]
30. Gour N, Dong X. The MRGPR family of receptors in immunity. Immunity 2024;57:28-39. [Crossref] [PubMed]
31. Orifjon S, Jammatov J, Sousa C, et al. Translation and Adaptation of the Adult Developmental Coordination Disorder/Dyspraxia Checklist (ADC) into Asian Uzbekistan. Sports (Basel) 2023;11:135. [Crossref] [PubMed]
32. Geeleher P, Cox N, Huang RS. pRRophetic: an R package for prediction of clinical chemotherapeutic response from tumor gene expression levels. PLoS One 2014;9:e107468. [Crossref] [PubMed]
33. Huang SH, Ng WT, Bakst RL, et al. Precision in Practice: A Critical Overview of Recent Advances in Nasopharyngeal Cancer Management. Int J Radiat Oncol Biol Phys 2026;125:783-94. [Crossref] [PubMed]
34. Watts VJ. Molecular mechanisms for heterologous sensitization of adenylate cyclase. J Pharmacol Exp Ther 2002;302:1-7. [Crossref] [PubMed]
35. Pan G, Huang M, Fu S, et al. ADCY4 inhibits cAMP-induced growth of breast cancer by inactivating FAK/AKT and ERK signaling but is frequently silenced by DNA methylation. Sci Rep 2025;15:20426. [Crossref] [PubMed]
36. Nguyen HO, Tiberio L, Facchinetti F, et al. Modulation of Human Dendritic Cell Functions by Phosphodiesterase-4 Inhibitors: Potential Relevance for the Treatment of Respiratory Diseases. Pharmaceutics 2023;15:2254. [Crossref] [PubMed]
37. Grishanova AY, Perepechaeva ML. Kynurenic Acid/AhR Signaling at the Junction of Inflammation and Cardiovascular Diseases. Int J Mol Sci 2024;25:6933. [Crossref] [PubMed]
38. de Juan A, Tabtim-On D, Coillard A, et al. The aryl hydrocarbon receptor shapes monocyte transcriptional responses to interleukin-4 by prolonging STAT6 binding to promoters. Sci Signal 2024;17:eadn6324. [Crossref] [PubMed]
39. Kumar R, Rao GN. Novel Role of Prereplication Complex Component Cell Division Cycle 6 in Retinal Neovascularization. Arterioscler Thromb Vasc Biol 2022;42:407-27. [Crossref] [PubMed]
40. Yu X, Liu Y, Yin L, et al. Radiation-promoted CDC6 protein stability contributes to radioresistance by regulating senescence and epithelial to mesenchymal transition. Oncogene 2019;38:549-63. [Crossref] [PubMed]
41. Yang K, Yin J, Sheng B, et al. AhR‑E2F1‑KGFR signaling is involved in KGF‑induced intestinal epithelial cell proliferation. Mol Med Rep 2017;15:3019-26. [Crossref] [PubMed]
42. Zhao Y, Cai J, Shi K, et al. Germacrone induces lung cancer cell apoptosis and cell cycle arrest via the Akt/MDM2/p53 signaling pathway. Mol Med Rep 2021;23:452. [Crossref] [PubMed]
43. Qin C, Li T, Wang Y, et al. CHRNB2 represses pancreatic cancer migration and invasion via inhibiting β-catenin pathway. Cancer Cell Int 2022;22:340. [Crossref] [PubMed]
44. Shi D, Guo W, Chen W, et al. Nicotine promotes proliferation of human nasopharyngeal carcinoma cells by regulating α7AChR, ERK, HIF-1α and VEGF/PEDF signaling. PLoS One 2012;7:e43898. [Crossref] [PubMed]
45. Ji X, Zhang W, Gui J, et al. Role of a genetic variant on the 15q25.1 lung cancer susceptibility locus in smoking-associated nasopharyngeal carcinoma. PLoS One 2014;9:e109036. [Crossref] [PubMed]
46. Portal-Nuñez S, Shankavaram UT, Rao M, et al. Aryl hydrocarbon receptor-induced adrenomedullin mediates cigarette smoke carcinogenicity in humans and mice. Cancer Res 2012;72:5790-800. [Crossref] [PubMed]
47. Minacori M, Fiorini S, Perugini M, et al. AhR and STAT3: A Dangerous Duo in Chemical Carcinogenesis. Int J Mol Sci 2025;26:2744. [Crossref] [PubMed]
48. Rubin CM, van der List DA, Ballesteros JM, et al. Mouse mutants for the nicotinic acetylcholine receptor ß2 subunit display changes in cell adhesion and neurodegeneration response genes. PLoS One 2011;6:e18626. [Crossref] [PubMed]
49. Wang F, Zhao F, Zhang L, et al. CDC6 is a prognostic biomarker and correlated with immune infiltrates in glioma. Mol Cancer 2022;21:153. [Crossref] [PubMed]
50. Liu N, Zhong L, Ni G, et al. High Matrix Metalloproteinase 28 Expression is Associated with Poor Prognosis in Pancreatic Adenocarcinoma. Onco Targets Ther 2021;14:4391-406. [Crossref] [PubMed]
51. Zhang Y, Chen L, Hu GQ, et al. Gemcitabine and Cisplatin Induction Chemotherapy in Nasopharyngeal Carcinoma. N Engl J Med 2019;381:1124-35. [Crossref] [PubMed]