Development and validation of prognostic models based on cell cycle-related signatures for predicting the prognosis of patients with lung adenocarcinoma

Introduction

Lung cancer is a malignant neoplasm associated with a high incidence of morbidity and mortality, which threatens human life safety seriously (1). Among patients diagnosed with lung cancer, lung adenocarcinoma (LUAD) is without a doubt the most prevailing histological subtype. Currently, surgical treatment remains the primary approach for patients with LUAD. However, the risk of postoperative metastasis and recurrence stays a significant concern (2), which is also a major factor affecting patients’ long-term postoperative survival. In spite of the improved treatment of LUAD to a large extent, its therapeutic effect is still unsatisfactory (3). Patients with LUAD often present with mild symptoms that may not elicit concern, thereby delaying the diagnosis (4). Therefore, the exploration of characteristic biomolecular associated with LUAD plays a crucial role in improving patients’ long-term survival (5,6). Over the past few years, researchers have discovered a variety of biomarkers related to the prognosis of LUAD, such as carcinoembryonic antigen (CEA) and CYFRA21-1 (7,8). However, existing prognostic models for LUAD rely on conventional biomarkers typically and clinical features that have limitations in terms of predictive accuracy and biological interpretation (9).

Cell cycle is a crucial process for cell growth and division, the cell cycle regulatory mechanisms in LUAD are often disrupted (10,11). Furthermore, the cell cycle-related competing endogenous RNA (ceRNA) network is capable of taking into account the interactions among multiple RNA molecules (12). Many non-coding RNAs (ncRNAs) have a vital impact on regulating biological processes (BPs) and on the development of diseases (13). Long non-coding RNA (LncRNA) has been demonstrated to inhibit docetaxel resistance in LUAD through the regulation of autophagy, which is induced by autophagy-related 5 (ATG5) (14), and it has been shown to promote the proliferation, migration, and angiogenesis of malignant lung cells via the miR-942/TNS1 pathway (15). Recently, researchers prove that the microRNA (miRNA) and the regulated messenger RNAs (mRNAs) are closely related to the progression of LUAD, which might also be the prognostic markers in LUAD (16-19). Despite some known LUAD-related biomarkers, the mechanism of LUAD remains unclear. Hence, the study of ncRNAs based on ceRNA network will contribute to finding new prognostic biomarkers of LUAD.

In this research, in accordance with the symbol level characteristics of LUAD-related sample transcriptome, the Kyoto Encyclopedia of Genes and Genomes (KEGG) signaling pathway was quantified by gene set variation analysis (GSVA) algorithm to screen out LUAD-related genes, and then the cell cycle-related ceRNA network was built. Finally, we screened the optimal gene combinations in the regulatory network to assist with diagnosing and monitoring of LUAD. Our model provides a reliable prognostic assessment of the survival status of 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-24-1479/rc).

Methods

Data preprocessing

We obtained the gene profile data of the LUAD samples from The Cancer Genome Atlas (TCGA) database (https://www.cancer.gov/ccg/research/genome-sequencing/tcga), and the datasets were analyzed applying Illumina HiSeq 2000 RNA Sequencing (Illumina, California, US). We acquired 450 LUAD cases totally, including 18 normal samples and 432 LUAD cases, which were acted as the training dataset.

Meanwhile, we searched the keywords “Lung Adenocarcinoma and homo sapiens” to acquire the gene expression profile data from the Gene Expression Omnibus (GEO) repository (http://www.ncbi.nlm.nih.gov/geo/) (20). Searching criteria were as follows: (I) solid tissues of LUAD; (II) sample size was not less than 100; (III) the sample has histology type classification, such as LUAD or lung squamous cell carcinoma; (IV) the samples have clinical prognostic information. Two datasets were obtained from the GEO database and functioned as validation datasets: the GSE50081 dataset (21) included 181 LUAD samples, of which 127 had clinical prognostic information; the GSE37745 data set (22,23) consisted of 196 LUAD samples, with 106 of them having clinical prognostic information. The validation datasets were analyzed using the Affymetrix Human Genome U133 Plus 2.0 Array Assay Platform GPL570.

Screening of differentially expressed RNAs (DERs)

Limma package (Version 3.34.7, https://bioconductor.org/packages/release/bioc/html/limma.html) (24) was done in R3.6.1 to get the differences between normal and LUAD samples. The significance analysis of microarrays met the criteria of P<0.05 and |log2 fold change (FC)| >1. Moreover, we utilized the pheatmap package to generate the heatmap (Version 1.0.8, https://cran.r-project.org/web/packages/pheatmap/index.html) (25).

GSVA

We sourced all of the KEGG pathway information from the GSEA database and made it available for download (http://www.gsea-msigdb.org/gsea/downloads.jsp) (26). Then, based on the LUAD whole genome and significantly differentially expressed genes (DEGs) expression values, the GSVA in the R3.6.1 language (27) was used to quantify each KEGG signal pathway through gene expression values. The t-test was utilized to evaluate the discrepancies between the quantified KEGG signaling pathways in the LUAD tumor and the control samples across two distinct groups. The P value of less than 0.05 was selected as the criterion for identifying significant outcomes. The intersection of the two methods to obtain the KEGG pathway was retained as an important quantitative KEGG signaling pathway, and the significant DEGs involved in these KEGG signaling pathways were extracted for further analysis.

Construction of ceRNA network

First, the DIANA-LncBasev2 database (http://carolina.imis.athena-innovation.gr/diana_tools/web/index.php?r=lncbasev2%2Findex-experimental) (28) was carried out to find the lncRNA-miRNA connection pairs, and only pairs with opposite directions were kept. Moreover, the starBase database (including targetScan, picTar, RNA22, PITA and miRanda databases) (https://rnasysu.com/encori/) (29) was carried out to search for the target gene regulated miRNA. Next, we chose regulatory relationships that were included in a database shows the miRNA regulates the mRNA relationship pair, and the previously obtained important quantitative mRNAs involved in the KEGG signaling pathway were aligned with the gene targets, the miRNA-mRNA connection network was built by retaining the pairs of negatively correlated miRNA and mRNA expression levels. Finally, integrating the lncRNA-miRNA and miRNA-mRNA connected networks, we developed a ceRNA network. The Gene Ontology (GO) and KEGG pathway on the regulated mRNAs of the ceRNA regulatory network were analyzed through the DAVID 6.8 (https://david.ncifcrf.gov/) (30) with the P<0.05.

Generation of prognostic risk prediction model

Least absolute shrinkage and selection operator (LASSO) analysis was utilized to acquire the optimized gene combination from the ceRNA network. We used the R package to “penalize” and run 1,000-fold likelihood cross-validation to determine the optical lambda (https://cran.r-project.org/web/packages/penalized/index.html) (31), the following risk score (RS) model was built according to the prognostic coefficient of LASSO and the gene expression levels and RS values for each sample in the dataset were calculated using the following formula: RS = ∑βgenes × Exp genes (32). In this context, the term “∑βgenes” refers to the LASSO prognosis coefficient of the target gene. Meanwhile, “Exp genes” represents the gene expression levels as documented in the TCGA dataset.

Effectiveness evaluation of the RS model

With the median RS as the critical value, delineating the samples within the datasets into two distinct groups: a high-risk group and a low-risk group, then separated them into the training dataset and the validation dataset, respectively. The association between the prognosis and the risk model was subsequently evaluated through the application of the Kaplan-Meier (KM) curve in the R3.6.1 package (version 2.41-1, http://bioconductor.org/packages/survivalr/) (33).

Screening of independent prognostic clinical factors

We conducted an examination of the univariate and multivariate Cox regression analysis of the R3.6.1 language survival package (version 2.41-1) to identify the clinically relevant factors that are independent in the training dataset (P<0.05).

Generation of prognostic risk prediction model of nomogram

We built 3- and 5-year prognostic risk prediction models of nomogram by using R3.6.1 “rms” for further investigation of the association between individual factors and survival outcomes (version 5.1-2, https://cran.r-project.org/web/packages/rms/index.html) (34). Harrell’s concordance index (C-index) was counted utilizing the “survcomp” package (http://www.bioconductor.org/packages/release/bioc/html/survcomp.html) (35) in R, which was assessed for its predictive capacity. A C-index higher than 0.70 indicates a good model, while a score of around 0.50 implies a random background (36).

Evaluation of the multiple prognostic risk prediction model

We analyzed the predictive capacity of RS model using receiver operating characteristic (ROC) curves alone and other clinical characteristics which containing pathologic stage, tumor recurrence, clinical and RS prognostic risk prediction models.

Quantitative real-time polymerase chain reaction (qRT-PCR) and survival curve analysis

Tumor tissues and paracancerous tissues were collected from 10 patients diagnosed with LUAD in the First Hospital of Jilin University and stored at −80 °C. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the ethics review board of the First Hospital of Jilin University (No. 2025-028) and informed consent was obtained from all individual participants. Specifically, total RNA was extracted using Trizol reagent (Solarbio, Beijing, China) in accordance with the manufacturer’s instructions. Subsequently, complementary DNA (cDNA) synthesis was carried out with the HiScript II 1st Strand cDNA Synthesis Kit (R211-01, Vazyme). Then, the SYBR qPCR Master Mix (Q311-02, Vazyme) along with specific primers (Table 1) was employed for target gene detection on the LightCycler480 system (Roche, Basel, Switzerland). Each sample was assayed in triplicate, with GAPDH serving as the reference gene, the 2^−∆∆Ct method was utilized for expression quantification. After that, we divided the patients into two groups based on the gene expression level, then we drew the KM curves to assess the accuracy of the prognostic model from a clinical perspective in the R3.6.1 package (version 2.41-1).

Statistical analysis

All data were analyzed and plotted using R software (version 4.3.1). For the comparison between two groups, Student’s t-test was performed; for the comparison between multiple groups, Wilcoxon rank-sum test was carried out. All data were expressed as the mean ± standard deviation. The R package “pROC” was employed to construct ROC curves for evaluating the diagnostic value of hub genes. A P value <0.05 was considered statistically significant.

Results

DEGs screening

A total of 1,431 differentially expressed mRNAs (DEmRNAs) (498 down- and 933 up-regulated), 529 differentially expressed long non-coding RNAs (DElncRNAs) (110 down- and 419 up-regulated) and 162 differentially expressed microRNAs (DEmiRNAs) (18 down- and 144 up-regulated) were screened with the P<0.05 and |log2 FC| >1. The distribution of the DERs was illustrated on the volcano map (Figure 1A), and the distribution of each type of DERs is shown in Figure 1B, all DERs are shown in table available at https://cdn.amegroups.cn/static/public/tcr-24-1479-1.xlsx. According to the value of |log2 FC| we displayed a heatmap of DERs (Figure 1C). A two-way hierarchically clustered showed the expression levels of DERs.

Figure 1 The identification of DERs. (A) The volcano map of the DERs. P value <0.05 was the cutoff value. Red and blue dots suggest markedly up-regulated and down-regulated RNAs respectively, and gray dots indicated non-significant DERs. (B) Proportional bar chart of up and down DERs. Blue and pink represented the significant down-regulation and up-regulation of the proportion of DERs separately. (C) Two-way hierarchical clustering chart based on significantly DERs expression values. Red and blue meant up- and down-regulated DERs apart; black and white bar represent LUAD cases and normal samples independently. DERs, differentially expressed RNAs; FC, fold change; FDR, false discovery rate; lncRNA, long non-coding RNA; LUAD, lung adenocarcinoma; mRNA, messenger RNA; miRNA, microRNA.

GSVA analysis and the expression values of DEGs

We acquired the KEGG pathway information from GSEA database (table available at https://cdn.amegroups.cn/static/public/tcr-24-1479-2.xlsx). In total, 186 and 157 KEGG pathways were acquired according to LUAD whole genome and the expression values of DEGs, respectively. Then, a two-sample t-test was employed to ascertain the disparities between the quantified KEGG signaling pathways in the LUAD tumor and the normal samples, 116 and 136 quantified KEGG pathways were screened separately for significant differences in the allocation of the KEGG signaling pathways between the groups totally (table available at https://cdn.amegroups.cn/static/public/tcr-24-1479-3.xlsx). The terms of top 20 KEGG pathways are shown in Figures S1,S2, respectively (ranked according to P value). We found that the activity scores of DNA damage repair related pathways in LUAD were up-regulated, such as mismatch repair, DNA replication, and base excision repair, etc. Meanwhile, the activity scores of energy metabolism-related pathways in LUAD were also up-regulated, such as metabolism of fatty acid, propagate, and glycerophospholipid. Furthermore, we found that the activity scores of cell cycle related pathways in LUAD were also up-regulated, such as cell cycle, the mitogen-activated protein kinase (MAPK) signaling pathway, and the Wnt signaling pathway. These results demonstrate that biological pathways, including those involved in DNA damage repair, energy metabolism, and the cell cycle, play a pivotal role in the progression of LUAD. After comparing two sets of quantified KEGG signaling pathways, a total of 60 overlapping quantified KEGG pathways was screened, involving 240 significantly DEGs (tables available at https://cdn.amegroups.cn/static/public/tcr-24-1479-4.xlsx, https://cdn.amegroups.cn/static/public/tcr-24-1479-5.xlsx).

Construction of ceRNA network

We acquired 232 lncRNA-miRNA and 52 miRNA-mRNA connection pairs totally (table available at https://cdn.amegroups.cn/static/public/tcr-24-1479-6.xlsx). In total, 176 relationships between lncRNA, miRNA, and mRNA were acquired (129 DElncRNA, 21 DEmiRNA and 26 DEmRNA), the ceRNA network was built as shown in Figure S3. We found that 38 GO BPs and 8 KEGG pathways were yielded on DAVID version 6.8 (table available at https://cdn.amegroups.cn/static/public/tcr-24-1479-7.xlsx). Here, we exhibited the terms of TOP10 only (ranked according to P value). Among them, we found that the DEmRNAs were significantly associated with GO BP, such as cell surface receptor signaling pathway, aging and DNA replication (Figure 2A). The DEmRNAs were significantly linked to KEGG pathway database including pathways entitled “Neuroactive ligand-receptor interaction” and “progesterone-mediated oocyte maturation” (Figure 2B). Interestingly, we found that DEmRNA in the ceRNA network was mainly involved in cell division, DNA replication, and G1/S transition of mitotic cell cycle. These results suggest that the encoded genes in the ceRNA network play an essential function in the propagation of LUAD cells.

Figure 2 Enrichment analysis of (A) GO biological process and (B) KEGG pathway associated with significant genes of the ceRNA regulatory network. The horizontal axis suggests genes count, the vertical axis represents the GO and KEGG entry name; and the color suggests significance, a direct correlation can be observed with increasing red intensity. ceRNA, cycle-related competing endogenous RNA; GO, Gene Ontology; KEGG, Kyoto Encyclopedia of Genes and Genomes.

Generation and evaluation of RS survival prognostic risk prediction models

To identify precisely the ideal gene combination, we used the LASSO algorithm based on the 26 significantly DEGs contained in the ceRNA regulatory network. Finally, six optimal genes (ADRB2, IL1A, PIK3R2, CKD1, CCNB1 and CHRNA5) related to prognostic were obtained. Among them, CKD1, CCNB1 and PIK3R2 are all cell cycle related genes. Then, the following RS calculation formula was established: RS = −0.0421373556 × Exp_ADRB2 + 0.0333745460 × Exp_IL1A − 0.0094064148 × Exp_PIK3R2 + 0.0059610631 × Exp_CKD1 + 0.0334625200 × Exp_CCNB1 + 0.0006369962 × Exp_CHRNA5. RS values were calculated respectively in TCGA training dataset, GSE50081 and GSE37745 validation datasets, respectively. The boundary was defined by the median value of RS. The cases of the datasets were separated into two groups according to the medium score of RS values (high- and low-risk group); the distribution of RS value and survival prognosis time of each group were shown in Figure 3. We used the three datasets’ KM curves to assess the relationship between these two groups and actual prognostic information for LUAD (Figure 4). It indicated that the low-risk samples at the TCGA dataset yielded more favorable survival projections (P<0.001), and the GSE50081 and GSE37745 datasets also had the same trend (P value is 0.04 and 0.05, respectively). The results suggested that there was an obvious relationship between the different risk groups predicted based on the RS model and the actual prognosis, the centralized RS scores and grouping situations are shown in table available at https://cdn.amegroups.cn/static/public/tcr-24-1479-8.xlsx.

Figure 3 Effectiveness assessment of the RS model. TCGA training dataset (A), GSE50081 (B) and GSE37745 (C) validation datasets. RS distribution (above), survival time status (middle) and ROC curves of 1-, 3- and 5-year based on prognostic characteristics of important genes (bottom). AUC, area under the curve; ROC, receiver operating characteristic; RS, risk score; TCGA, The Cancer Genome Atlas.

Figure 4 KM curve of the RS prediction model and prognosis. TCGA training dataset (A), GSE50081 (B) and GSE37745 (C) validation datasets samples are based on the RS prediction model and prognosis. The blue and yellow curves suggest the low- and high-risk samples, respectively. KM, Kaplan-Meier; RS, risk score; TCGA, The Cancer Genome Atlas.

Screening of independent prognostic clinical factors

Three independent prognostic factors were identified: pathologic stage, tumor recurrence, and RS model status, which showed a significant correlation by univariate and multivariate Cox regression analyses (Table 2). We discovered that the pathologic stage, tumor recurrence and RS model status were independently prognostic related to other clinical factors. The KM survival analysis of pathologic stage and tumor recurrence in the TCGA dataset are displayed in Figure 5.

Figure 5 KM survival plots. Pathologic stage (A) and tumor recurrence (B) KM curves related to survival prognosis in the TCGA training dataset samples. KM, Kaplan-Meier; TCGA, The Cancer Genome Atlas.

Generation of prognostic risk prediction model of nomogram

To generate 3- and 5-year prognostic models of the nomogram and to ascertain the potential survival times of patients diagnosed with LUAD, the ROC curve of survival prediction based on RS were shown in Figure 6A. The C-index for 3- and 5-year prognostic models of nomogram were 0.7665 and 0.7104 respectively, which exerted accurate predictive ability. The nomogram calibration curve between the 3- and 5-year predicted probability of OS and the actual survival rate showed good consistency (Figure 6B).

Figure 6 Generation and validation of the nomogram consisted of the independent prognostic factors. (A) A quantitative nomogram is composed of risk score and independent prognostic factors for predicting the survival rate of the patients with LUAD. (B) Consistency plot of nomogram between the predicted survival rate and the actual survival rate in 3-year (top) and 5-year (bottom). The horizontal and vertical axis represents the predicted and actual survival rate, respectively. (C) ROC curves of different factors in TCGA training dataset and (D) GSE50081 validation dataset. AUC, area under the curve; LUAD, lung adenocarcinoma; OS, overall survival; ROC, receiver operating characteristic; RS, risk score; TCGA, The Cancer Genome Atlas.

Evaluation of multiple prognostic risk prediction model

The ROC curves were carried out to assess the effectiveness of the RS and clinical factors prognostic risk prediction models and the combinatorial model of the two models in TCGA dataset and GSE50081 dataset. The areas under the curve (AUCs) of pathologic stage, tumor recurrence, clinical, RS prognostic risk prediction models were 0.657, 0.682, 0.685 and 0.757 in the TCGA dataset, respectively. The AUCs of pathologic stage, tumor recurrence, clinical, RS prognostic risk prediction models were 0.597, 0.556, 0.605 and 0.730 in GSE50081 dataset, separately. The AUCs of the combination of RS and clinical factors prognostic risk prediction model were 0.869 and 0.770 in TCGA dataset and GSE50081 dataset, independently. The results indicated that the combination of RS and clinical factors prognostic risk prediction model has the highest AUC, which suggested the combination of RS and clinical factors prognostic risk prediction model had the better predictive ability (Figure 6C,6D).

Evaluation of the prognostic genes

To confirm the clinical utility of the identified biomarkers and prognostic models, we collected tumor tissues from ten patients with LUAD, and conducted the qRT-PCR assay. Compared with the adjacent non-tumor tissues, the expression levels of ADRB2, IL1A, PIK3R2, CKD1, CCNB1, and CHRNA5 in tumor tissues were significantly elevated (P<0.01, Figure 7A). Similarly, bioinformatics techniques were employed to analyze the prognosis of these 10 patients. We found that patients with high expression levels of the ADRB2, IL1A, CKD1, and CCNB1 had a relatively poor prognosis (with P values of 0.002, 0.03, 0.05, and 0.02, respectively, Figure 7B).

Figure 7 qRT-PCR assay (A) and KM curve (B) of the prognosis in ADRB2, IL1A, PIK3R2, CKD1, CCNB1, and CHRNA5. **, P<0.01 denotes statistical significance when compared to the control group. ADRB2, adrenoceptor beta 2; CCNB1, cyclin B1; CKD1, cyclin-dependent kinase 1; CHRNA5, cholinergic receptor nicotinic alpha 5; IL1A, interleukin 1 alpha; KM, Kaplan-Meier; LAUD, lung adenocarcinoma; PIK3R2, phosphoinositide-3-kinase regulatory subunit 2; qRT-PCR, quantitative real-time polymerase chain reaction.

Discussion

LUAD has always been regarded as one of the most deadly cancers, resulting in a massive financial burden every year, especially for advanced LUAD (37). While some progresses have been made to diagnose and treat LUAD, such as gene therapy and immunotherapy, due to the facts that some of the patients may develop drug resistance and others are not sensitive to the therapy, they are not suitable for all patients (2). Consequently, it is of the utmost importance to identify the specific biomarkers of LUAD in order to facilitate prompt diagnosis and evaluation of patient prognosis.

ceRNA is a novel way for studying the interactions of RNAs (37). Researches have proved that the ceRNA is linked to the progress of cancers, including breast cancer (38), glioblastoma (39), and lung cancer (37), etc. Therefore, the construction on ceRNA network of LUAD and the identification of important biomarkers are very meaningful for understanding the mechanism in the LUAD.

Six hub genes have been identified and found to be associated with the prognosis of LUAD as follows: RAD54L, CHEK1, RAD51, KIF18B, KIFC1 and FEN1 (40). In our study, the bioinformatics analysis was performed to acquire the differential LUAD-related genes firstly. Then, we built the ceRNA regulatory network to investigate the correlation between RNA levels and disease progression. GO and KEGG pathways analyses of mRNAs in the ceRNA network were carried out to explore the mRNAs in the ceRNA network, which demonstrated that the significant genes exhibited a strong correlation with cell surface receptor signaling pathways in this BP (five genes, including EDNRB, P2RY1, LEPR, CD36 and ADRB2), aging (four genes, including IL6, EDNRB, CNR1 and P2RY1), DNA replication (four genes, including CKD1, DNA2, CDC6 and CDC25A); and the KEGG pathway analysis revealed that the substantial genes were linked to neuroactive ligand-receptor interaction (seven genes, including EDNRB, CHRNA5, CNR1, P2RY1, LEPR, ADRB1 and ADRB2) and progesterone-mediated oocyte maturation (four genes, including CCNB1, CKD1, PIK3R2 and CDC25A). LASSO algorithm was utilized to identify the optimal gene combination within the ceRNA network. Finally, six optimal genes (ADRB2, IL1A, PIK3R2, CKD1, CCNB1 and CHRNA5) related to prognostic were obtained. Adrenoceptor beta 2 (ADRB2) has been proved to be the potential prognostic biomarker of LUAD (41). It has been verified that interleukin 1 alpha (IL1A) might be linked to the survival of the prostate cancer and breast cancer (42). Research has been demonstrated that phosphoinositide-3-kinase regulatory subunit 2 (PIK3R2) was associated with the prognostic of melanoma (43). Cyclin-dependent kinase 1 (CKD1) has been identified that the overexpression of it is related in most tumor tissues, such as kidney renal papillary cell carcinoma and LUAD, which may become a potential prognostic biomarker in tumor therapy (44). Cyclin B1 (CCNB1) has been testified to play a pivotal role in the prediction of numerous diseases including colorectal cancer (45), liver cancer (46) and hepatocellular carcinoma (47), etc. Cholinergic receptor nicotinic alpha 5 (CHRNA5) subunit might be a potential prognostic biomarker of breast cancer (48) and lung cancer (49).

Moreover, the RS prediction model was built dramatically related to prognostic of LUAD patients. The KM curves and the ROC curve were performed to evaluate the effectiveness of the predictive model, and we found that the prognostic risk prediction model for LUAD had satisfactory predictive ability. The prognostic ability of RS prediction model revealed a correlation between lower RS and higher survival rates among patients. Recent research has demonstrated that prognostic potential of pyroptosis-related genes (PRRS) is an independent prognostic factor for LUAD (50). A nomogram was subsequently constructed, incorporating PRRS and clinical characters specific to LUAD (51). To further explore the prognostic model with better performance, we constructed multiple prognostic models, including clinical factor prognostic model, RS prognostic model, the combination of clinical factor and RS prognostic model, and we found that the combination of clinical factor and RS prognostic risk prediction model had a better performance in predictive ability.

However, there are several limitations in our study. While the C-index values imply a relatively high level of predictive precision, it is essential to conduct further validation within larger and independent cohorts in order to ascertain the clinical practicality of the identified biomarkers and prognostic models.

Conclusions

Overall, a ceRNA network of LUAD was built and six optimal genes (ADRB2, IL1A, PIK3R2, CKD1, CCNB1 and CHRNA5) related to prognostic were obtained. Furthermore, the RS, nomogram and the combination of RS and clinical factors prognostic risk prediction models with good predictive ability were built, which may provide a powerful means for prognostic assessment 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-24-1479/rc

Data Sharing Statement: Available at https://tcr.amegroups.com/article/view/10.21037/tcr-24-1479/dss

Peer Review File: Available at https://tcr.amegroups.com/article/view/10.21037/tcr-24-1479/prf

Funding: None.

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://tcr.amegroups.com/article/view/10.21037/tcr-24-1479/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. The study was approved by the ethics review board of the First Hospital of Jilin University (No. 2025-028) and informed consent was obtained from all individual participants.

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. Chen Z, Huang Y, Hu Z, et al. Landscape and dynamics of single tumor and immune cells in early and advanced-stage lung adenocarcinoma. Clin Transl Med 2021;11:e350. [Crossref] [PubMed]
2. Boumahdi S, de Sauvage FJ. The great escape: tumour cell plasticity in resistance to targeted therapy. Nat Rev Drug Discov 2020;19:39-56. [Crossref] [PubMed]
3. Zhang L, Zhang K, Liu S, et al. Identification of a ceRNA Network in Lung Adenocarcinoma Based on Integration Analysis of Tumor-Associated Macrophage Signature Genes. Front Cell Dev Biol 2021;9:629941. [Crossref] [PubMed]
4. Zhang L, Luo Y, Cheng T, et al. Development and Validation of a Prognostic N6-Methyladenosine-Related Immune Gene Signature for Lung Adenocarcinoma. Pharmgenomics Pers Med 2021;14:1549-63. [Crossref] [PubMed]
5. Xu S, Xu Y, Chai W, et al. KLRB1 expression is associated with lung adenocarcinoma prognosis and immune infiltration and regulates lung adenocarcinoma cell proliferation and metastasis through the MAPK/ERK pathway. J Thorac Dis 2024;16:3764-81. [Crossref] [PubMed]
6. Xi Y, Zeng J, Zhou Y, et al. Prognostic value and potential biological function of PMSD11 in lung adenocarcinoma. J Thorac Dis 2024;16:7819-35. [Crossref] [PubMed]
7. Xu H, Zhao G, Lin J, et al. A combined preoperative red cell distribution width and carcinoembryonic antigen score contribute to prognosis prediction in stage I lung adenocarcinoma. World J Surg Oncol 2023;21:56. [Crossref] [PubMed]
8. Li Y, Shi D, Jiang Y, et al. Establishment of potential lncRNA-related hub genes involved competitive endogenous RNA in lung adenocarcinoma. BMC Cancer 2024;24:1371. [Crossref] [PubMed]
9. Zhang Z, Zhang P, Xie J, et al. Five-gene prognostic model based on autophagy-dependent cell death for predicting prognosis in lung adenocarcinoma. Sci Rep 2024;14:26449. [Crossref] [PubMed]
10. Xu Y, Jiao Y, Liu C, et al. R-loop and diseases: the cell cycle matters. Mol Cancer 2024;23:84. [Crossref] [PubMed]
11. Wu J, Fu G, Luo C, et al. Cuproptosis-related ceRNA axis triggers cell proliferation and cell cycle through CBX2 in lung adenocarcinoma. BMC Pulm Med 2024;24:85. [Crossref] [PubMed]
12. Ding D, Xu C, Zhang J, et al. Revealing underlying regulatory mechanisms of LINC00313 in Osimertinib-resistant LUAD cells by ceRNA network analysis. Transl Oncol 2024;43:101895. [Crossref] [PubMed]
13. Li J, Fang J, Jiang X, et al. RNAkines are secreted messengers shaping health and disease. Trends Endocrinol Metab 2024;35:201-18. [Crossref] [PubMed]
14. Yu Z, Tang H, Chen S, et al. Exosomal LOC85009 inhibits docetaxel resistance in lung adenocarcinoma through regulating ATG5-induced autophagy. Drug Resist Updat 2023;67:100915. [Crossref] [PubMed]
15. Wang J, Tan L, Yu X, et al. lncRNA ZNRD1-AS1 promotes malignant lung cell proliferation, migration, and angiogenesis via the miR-942/TNS1 axis and is positively regulated by the m(6)A reader YTHDC2. Mol Cancer 2022;21:229. [Crossref] [PubMed]
16. Chen H, Shen W, Ni S, et al. Construction of an immune-related lncRNA signature as a novel prognosis biomarker for LUAD. Aging (Albany NY) 2021;13:20684-97. [Crossref] [PubMed]
17. Ren Z, Hu M, Wang Z, et al. Ferroptosis-Related Genes in Lung Adenocarcinoma: Prognostic Signature and Immune, Drug Resistance, Mutation Analysis. Front Genet 2021;12:672904. [Crossref] [PubMed]
18. Ma Q, Xiao F, Hao Y, et al. The prognostic role of the Transducin-like Enhancer of split protein family in lung adenocarcinoma. Transl Lung Cancer Res 2021;10:3251-63. [Crossref] [PubMed]
19. Jia E, Shi X, Xue J. CCND2 is a prognostic biomarker and correlates with immune infiltration in lung adenocarcinoma. Transl Cancer Res 2024;13:1241-51. [Crossref] [PubMed]
20. Clough E, Barrett T, Wilhite SE, et al. NCBI GEO: archive for gene expression and epigenomics data sets: 23-year update. Nucleic Acids Res 2024;52:D138-44. [Crossref] [PubMed]
21. Tian L, Wang Y, Tian J, et al. Prognostic Value and Genome Signature of m6A/m5C Regulated Genes in Early-Stage Lung Adenocarcinoma. Int J Mol Sci 2023;24:6520. [Crossref] [PubMed]
22. Ding D, Wang L, Zhang Y, et al. Machine learning developed a programmed cell death signature for predicting prognosis and immunotherapy benefits in lung adenocarcinoma. Transl Oncol 2023;38:101784. [Crossref] [PubMed]
23. Huang J, Zhang J, Zhang F, et al. Identification of a disulfidptosis-related genes signature for prognostic implication in lung adenocarcinoma. Comput Biol Med 2023;165:107402. [Crossref] [PubMed]
24. Li J, Bai Y, Zhang H, et al. Single-cell RNA sequencing reveals the communications between tumor microenvironment components and tumor metastasis in osteosarcoma. Front Immunol 2024;15:1445555. [Crossref] [PubMed]
25. Liu W, Luo X, Zhang Z, et al. Construction of an immune predictive model and identification of TRIP6 as a prognostic marker and therapeutic target of CRC by integration of single-cell and bulk RNA-seq data. Cancer Immunol Immunother 2024;73:69. [Crossref] [PubMed]
26. Candia J, Ferrucci L. Assessment of Gene Set Enrichment Analysis using curated RNA-seq-based benchmarks. PLoS One 2024;19:e0302696. [Crossref] [PubMed]
27. Li BL, Wan XP. Prognostic significance of immune landscape in tumour microenvironment of endometrial cancer. J Cell Mol Med 2020;24:7767-77. [Crossref] [PubMed]
28. Paraskevopoulou MD, Vlachos IS, Karagkouni D, et al. DIANA-LncBase v2: indexing microRNA targets on non-coding transcripts. Nucleic Acids Res 2016;44:D231-8. [Crossref] [PubMed]
29. Liu J, Fu N, Yang Z, et al. The genetic and epigenetic regulation of CD55 and its pathway analysis in colon cancer. Front Immunol 2022;13:947136. [Crossref] [PubMed]
30. Sherman BT, Hao M, Qiu J, et al. DAVID: a web server for functional enrichment analysis and functional annotation of gene lists (2021 update). Nucleic Acids Res 2022;50:W216-21. [Crossref] [PubMed]
31. Zheng X, Amos CI, Frost HR. Comparison of pathway and gene-level models for cancer prognosis prediction. BMC Bioinformatics 2020;21:76. [Crossref] [PubMed]
32. Yang X, Wei Q. Identification of a novel Immune-Related prognostic model for patients with colorectal cancer based on 3 subtypes. Immunobiology 2023;228:152352. [Crossref] [PubMed]
33. Chen Y, Jia K, Sun Y, et al. Predicting response to immunotherapy in gastric cancer via multi-dimensional analyses of the tumour immune microenvironment. Nat Commun 2022;13:4851. [Crossref] [PubMed]
34. Guo H, Feng S, Li Z, et al. Prognostic Value of Body Composition and Systemic Inflammatory Markers in Patients with Locally Advanced Cervical Cancer Following Chemoradiotherapy. J Inflamm Res 2023;16:5145-56. [Crossref] [PubMed]
35. Shan S, Chen W, Jia JD. Transcriptome Analysis Revealed a Highly Connected Gene Module Associated With Cirrhosis to Hepatocellular Carcinoma Development. Front Genet 2019;10:305. [Crossref] [PubMed]
36. Huang B, Sollee J, Luo YH, et al. Prediction of lung malignancy progression and survival with machine learning based on pre-treatment FDG-PET/CT. EBioMedicine 2022;82:104127. [Crossref] [PubMed]
37. Wei D, Sun L, Feng W. hsa_circ_0058357 acts as a ceRNA to promote non small cell lung cancer progression via the hsa miR 24 3p/AVL9 axis. Mol Med Rep 2021;23:470. [Crossref] [PubMed]
38. Li Z, Yang HY, Dai XY, et al. CircMETTL3, upregulated in a m6A-dependent manner, promotes breast cancer progression. Int J Biol Sci 2021;17:1178-90. [Crossref] [PubMed]
39. Rahnama S, Bakhshinejad B, Farzam F, et al. Identification of dysregulated competing endogenous RNA networks in glioblastoma: A way toward improved therapeutic opportunities. Life Sci 2021;277:119488. [Crossref] [PubMed]
40. Liao Y, Wang Y, Cheng M, et al. Weighted Gene Coexpression Network Analysis of Features That Control Cancer Stem Cells Reveals Prognostic Biomarkers in Lung Adenocarcinoma. Front Genet 2020;11:311. [Crossref] [PubMed]
41. Zheng Q, Min S, Zhou Q. Identification of potential diagnostic and prognostic biomarkers for LUAD based on TCGA and GEO databases. Biosci Rep 2021;41:BSR20204370. [Crossref] [PubMed]
42. Zayed H. The identification of highly upregulated genes in claudin-low breast cancer through an integrative bioinformatics approach. Comput Biol Med 2020;127:103806. [Crossref] [PubMed]
43. Tian M, Yang J, Han J, et al. A novel immune checkpoint-related seven-gene signature for predicting prognosis and immunotherapy response in melanoma. Int Immunopharmacol 2020;87:106821. [Crossref] [PubMed]
44. Shen J, Gong X, Tan S, et al. CDK1 Acts as a Prognostic Biomarker Associated with Immune Infiltration in Pan-Cancer, Especially in Gastrointestinal Tumors. Curr Med Chem 2024; Epub ahead of print. [Crossref]
45. Zhang Z, Zhang X. Identification of m6A-Related Biomarkers Associated with Prognosis of Colorectal Cancer. Med Sci Monit 2021;27:e932370. [Crossref] [PubMed]
46. Hao L, Li S, Peng Q, et al. Anti-malarial drug dihydroartemisinin downregulates the expression levels of CDK1 and CCNB1 in liver cancer. Oncol Lett 2021;22:653. [Crossref] [PubMed]
47. Liu Z, Pu Y, Bao Y, et al. Investigation of Potential Molecular Biomarkers for Diagnosis and Prognosis of AFP-Negative HCC. Int J Gen Med 2021;14:4369-80. [Crossref] [PubMed]
48. Shehwana H, Keskus AG, Ozdemir SE, et al. CHRNA5 belongs to the secondary estrogen signaling network exhibiting prognostic significance in breast cancer. Cell Oncol (Dordr) 2021;44:453-72. [Crossref] [PubMed]
49. Pal K, Hussain T, Xie H, et al. Expression, correlation, and prognostic significance of different nicotinic acetylcholine receptors, programed death ligand 1, and dopamine receptor D2 in lung adenocarcinoma. Front Oncol 2022;12:959500. [Crossref] [PubMed]
50. Lin X, Zhou T, Hu S, et al. Prognostic significance of pyroptosis-related factors in lung adenocarcinoma. J Thorac Dis 2022;14:654-67. [Crossref] [PubMed]
51. Jiang Z, Wang X, Huang J, et al. Pyroptosis-based risk score predicts prognosis and drug sensitivity in lung adenocarcinoma. Open Med (Wars) 2023;18:20230663. [Crossref] [PubMed]