Introduction

According to statistics, lung cancer remains the most common malignant tumor in the world, regardless of gender. In 2020, new cases of lung cancer accounted for 11.4% of all cancer cases globally, with 1.8 million deaths attributed to the disease. Lung cancer is currently the leading cause of cancer-related mortality1. Lung adenocarcinoma is the most common pathological subtype of lung cancer, accounting for approximately 85%. It is often diagnosed at an advanced stage of the disease due to its early asymptomatic status. Surgery, immunotherapy, chemotherapy, and targeted therapy are common treatments. However, the therapeutic effects of these treatment methods are limited. Therefore, there is an urgent need to discover new targets.

Impaired apoptosis is a significant feature of tumor cells and a fundamental cause of drug resistance and poor patient prognosis2,3,4. Cellular pyroptosis, apoptosis, and necrosis are clear pathways of programmed cell death (PCD) that can significantly regulate tumor related diseases5. Although these three pathways differ significantly due to their respective characteristics, an increasing number of studies indicate a strong correlation between them. The three types of PCD interact with each other, resulting in a dynamic equilibrium in the end6. Thus, researchers have classified forms of apoptosis that cannot be explained by a single pathway or specific feature as pan-apoptosis (PANoptosis). Up to the present, it has been found that infectious diseases and neurological diseases have PANoptotic characteristics. However, research on oncological diseases mainly focuses on colorectal cancer and prostate cancer, with less research on lung adenocarcinoma.

As in other tumorigenic diseases, the progression of lung adenocarcinoma is the result of numerous genetic variants and pathways influenced by oncogenic factors. More personalized and precise approaches to tumor treatment are currently required. Therefore, the main goal of modern medical research on lung adenocarcinoma patients is to discover inherent molecular regulatory mechanisms, search for potentially sensitive drugs, and determine personalized and precise treatment strategies.

Method

Data collection

This research primarily relied on data from the Gene Expression Omnibus (GEO) and The Cancer Genome Atlas (TCGA) databases. The TCGA-GDC program provides comprehensive expression profiles, tumor mutation burden (TMB) information, and related clinical data, which were subsequently integrated into TCGA-LUAD. Each expression file for each patient was formatted using STAR counts, while the clinical information was formatted using BCR-xml. The authors utilized the Perl and R programming codes to execute the data procession. In particular, GSE30219 was chosen using the GPL570 [HG-U133-Plus_2] Affymetrix human genome U133Plus 2.0 array annotation platform. The expression profile and clinical information of patients in GSE30219 were downloaded from the “Series Matrix File(s)” link. Before analysis, the expression profile data of GSE30219 were pre-processed using the limma package, including missing value completion, correction, normalization, and other necessary steps.

Collection of the molecules involved in the PANoptosis process

AIFM1, AKT3, APAF1, APPL1, BMF, BNIP3L, BOK, CASP10, CASP2, CASP6, CASP7, CASP8, CD14, CHUK, CRADD, DFFA, DFFB, E2F1, HMGB2, IGF1, LY96, PPP3R1, TFDP1, TICAM1, TNFRSF1A, TNFSF10, TP73, TRAF2, UACA, UNC5B, YWHAE, YWHAG, FADD, FAS, FASLG, MLKL, RIPK1, RIPK3, TLR3, TNF, BAK1, BAX, CASP1, CASP3, CASP4, CASP5, CHMP2A, CHMP2B, CHMP3, CHMP4A, CHMP4B, CHMP4C, CHMP6, CHMP7, CYCS, ELANE, GSDMD, GSDME, GZMB, HMGB1, IL18, IL1A, IL1B, IRF1, IRF2.

Protein–protein interaction (PPI) network

The protein–protein interaction (PPI) network was constructed using the STRING database (https://cn.string-db.org/), with moderate confidence as the minimum required interaction score, using Homo as organism. Cytoscape was used as the PPI network visualization tool.

Identification of genes in LUAD

The RNA sequencing data and related clinical details of LUAD patients from the TCGA database were retrieved, including 503 LUAD samples and 54 noncancerous adjacent samples (NAT). From the previous article, a total of 65 PANoptoses were downloaded7. The differential analysis of the PRGs between LUAD and NATs was conducted using the ‘limma’ R package after obtaining the intersection from the above two datasets. The false discovery rate (FDR) was used to adjust the p value. A statistically significant result was considered when FDR was less than 0.05 and |log2 (FC)| was greater than or equal to 18.

LUAD cluster analysis

The cluster was evaluated using the R package ‘ConsensusClusterPlus’. After clustering the TCGA data, the optimal number of categories is determined based on the change area under the CDF curve. Based on the clustering results, survival analysis of the different groups was performed using the R packages ‘survival’ and ‘survminer.’ PCA analysis was then performed with R.

Association of molecular patterns with clinical features and prognosis of LUAD

The difference in prognosis between the two groups was compared using K–M (Kaplan–Meier) survival curves. A p value of less than 0.05 was considered statistically significant. We used univariate and multivariate logistic regression to identify variables that may independently affect the survival of patients. The hazard ratio (HR) and 95% confidence interval (CI) were calculated to identify genes associated with overall survival.

Construction of prognostic gene scores associated with PANoptosis

The genes associated with prognosis were identified using univariate Cox regression as the input genes. LASSO regression was used to reduce data dimensions and determine variables that need to be optimized. The feature recognition of prognosis is performed using multivariate Cox regression, with the equation as follows: “Risk score = (Gene A × A) + (Gene B × B) + (Genetic C × C) + (⋯) + (Genetic N × N)”, where A–N is the coefficient. K–M and receiver operating characteristic (ROC) curves were applied to evaluate the effectiveness of the prognostication prediction. For the TCGA cohort, patients were randomly divided into a training cohort and a validation cohort in a 1:1 ratio. The GSE30219 cohort was selected for external validation. ROC and C-index are used to evaluate the predicted efficacy.

Creation of predictive nomogram

The nomogram plot is a valuable clinical predictor of the risk and clinical pathological characteristics of LUAD patients, especially the overall survival rate (OS) at 1, 3, and 5 years. The nomogram plot combining clinical features and risk scores was generated using R package rms and replot. A calibration curve analysis was assessed the consistency between actual survival and survival predicted by the nomogram.

Functional enrichment analysis

The R package 'GSVA' was used to conduct GSVA enrichment analysis9. This non-parametric, unsupervised method is mainly used to evaluate biological processes and path changes in the sample. In order to investigate the potential biological mechanisms and functions of the selected PRG in signatures, gene ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis were conducted on the PRGs8.

Immune cell infiltration and difference analysis

Based on gene expression profiles, various algorithms, including CIBERSORT, were used to measure the tumor immune microenvironment of LUAD patients.

Immunotherapy and drug sensitivity analysis

The Tumor Immune Dysfunction and Exclusion (TIDE) program assessed patient immunotherapy responses10. Each patient was assigned a TIDE score. A score greater than 0 indicates a poor response to immunotherapy (or responder). The Genomics of Drug Sensitivity in Cancer (GDSC) database was used to test drug sensitivity11. The study compared high-risk and low-risk groups by calculating differences in tumor mutational burden and tumor microenvironment.

Statistical analysis

The analysis was conducted using R and PERL software. The test level was set as that P < 0.05 was considered statistically significant.

Result

PANoptosis-related genes in lung cancer

Based on the published literature, we have collected the molecules involved in PANoptosis, as shown in Table 17. These genes play a role in pyroptosis, necroptosis, and apoptosis. PANoptotic gene alterations are shown in Fig. 1a. Among them, UNC5B has the highest mutation frequency. Among them, UNC5B has the highest mutation frequency. Missense mutations are the most common type. Figure 1b shows the PPI network of these PANoptosis-related genes. RNA transcriptome data and corresponding clinical data of 549 LUAD patients were extracted from the TCGA database. The differentially expressed PANoptosis genes between tumor and non-cancerous adjacent samples (NATs) were found to be significant, as shown in Fig. 1c (Supplementary Data 1). The survival results presented are based on PRGs including BAK1, CASP10, CYCS, FADD, HMGB2, IL1A, MLKL, TGDP1, TICAM1, TRAF2, UACA, YWHAG and YWHAE. As shown in the K-M plot, TRAF2, BAK1, CYCS, FADD, HMGB2, IL1A, MLKL, TFDP1, T1CAM1, TRAF2, UACA, and YWHAE had a greater prognostic value in patients with lung adenocarcinoma. In summary, these genes are PRGs that affect the prognosis of patients with LUAD (Fig. 1d, Supplementary Data 2).

Table 1 PANoptosis genes.
Full size table
Fig. 1
figure 1

The analysis of PRGs in PANoptosis. (a) Waterfall plot of PANoptosis genes. (b) PPI network. (c) Differently expressed genes in LUAD. (d) Prognostic genes in LUAD.

Full size image

Analysis cluster of genes

A cluster parameter of 2 was chosen from 1 to 9 based on the cluster analysis. The patients with LUAD were then divided into 2 clusters, A and B (Fig. 2a–d, Supplementary Data 3). Figure 2e presents the survival analysis based on the PANoptosis cluster, which shows a significant difference in prognosis between the different groups. PCA analysis showed that the gene expression of PANoptosis significantly distinguished patients from different identified risk groups, demonstrating the accuracy of the model (Fig. 2f).

Fig. 2
figure 2

The result of PRGs cluster analysis. (a–d) Cluster analysis. (e) Survival analysis of different groups. (f) PCA analysis.

Full size image

The result of Functional analysis and differences of immune cell enrichment

The infiltration of immune cells was then investigated. Among the 23 immune cells, activated B cells and eosinophils were significantly enriched in cluster A. The others showed no difference or just the opposite (Fig. 3a, Supplementary Data 4). This indicates that the two clusters are related to the tumor immune microenvironment. Next, GSVA was performed to explore various expressed pathways in different groups, showing differences in the top 20 pathways. GSVA indicates that the KEGG and GO analyses differ entirely in the two clusters. According to the GSVA analysis results, cluster B is mainly involved in small cell lung cancer, and the p53 signaling pathway. At the same time, cluster B is engaged in multiple functions, including the positive regulation of cell cycle checkpoints (Fig. 3b,c, Supplementary Data 5, 6). Cluster B is involved in organelle division (BP), collagen-containing extracellular matrix (CC), and ATP hydrolysis activity (MF). We set the corrected p value filter condition to 1 and found the enrichment of pertussis and micro RNAs by the software in the KEGG pathway (Fig. 3d,e, Supplementary Data 7, 8). To sum up, the above results indicate a correlation between PANoptosis and the development of lung adenocarcinoma.

Fig. 3
figure 3

Result of Functional Enrichment Analysis. (a) The infiltration of immune cells in different groups. (b–c) The differentially expressed pathways in different groups. (d–e) The results of GO and KEGG analysis.

Full size image

Development and validation of the PRGs signature

A total of 477 patients who met the inclusion and exclusion criteria were randomly divided into a training set and a validation set in a 1:1 ratio to establish features for predicting the survival status of LUAD patients. In the training group, we performed univariate Cox regression on the differentially expressed PRGs to identify genes which are positively associated with prognosis. 12 optimal prognostic PRGs were further selected by LASSO analysis. Finally, a risk score model including the selected genes was developed in the training set: Risk score = ADM × 0.20554 + FLNC × 0.14125 + PITX3 × 0.68369 + LINGO2 × 0.27907 + FSIP2 × 0.26154 + C11orf16 × (-0.23574) + MS4A1 × (− 0.15237) (Fig. 4a,b, Supplementary Data 9). According to the median risk score calculated using the above formula in the training group, testing group, and the entire group of patients were further divided into high-risk and low-risk groups. The risk scores of these two clusters are different (Fig. 4c). Figure 4d shows the distribution of the two clusters regarding risk groups and survival outcomes. The validation confirmed the association between the clinical characteristics and the risk scores. The K–M curve shows significant differences in survival rates between risk groups in terms of tumor staging, age, gender, and lymph node metastasis (Fig. 4e). According to the median risk score, the training group, test group, and all patients were divided into high-risk and low-risk groups (Fig. 4f). Figure 4g showed that an increasing risk score was positively correlated with the accumulation of patients with poor survival status. The expression levels of signature PRGs are shown in Fig. 4h.

Fig. 4
figure 4

Establishment of risk scores and study of differences. (a–b) LASSO regression. (c) Relationship between risk and different clusters. (d) Relationship Between Cluster, Risk Score, and Survival Outcomes for Patients. (e) Relationship between patient prognosis and clinical characteristics. (f) Survival curves between different risk groups for training patients’ samples in TCGA and test patients’ samples in GEO, entire patients in TCGA. (g) Survival statuses of patients in different risk groups grouped by the signature in the training, validation, and entire set, respectively. (h) Expressions for the selected PRG in training, testing, and the entire set, respectively.

Full size image

Construction of a nomogram model of risk scores and clinical factors to predict cancer progression

Correlation between patient clinical data and risk scores was investigated. The constructed model and graphical representation of stages confirmed their high correlation with patient prognosis (Fig. 5a,b, Supplementary Data 10–11). Figures 5c,d illustrate the accuracy of the AUC curve for clinical data and risk score in forecasting patient prognosis at 1, 3 and 5 years. We plotted the C-index curve and noticed that the accuracy of the risk score was very high (Fig. 5e). We created a column chart using clinical parameters to evaluate the 1-, 3-, and 5-years overall survival rates of Luad patients. The calibration curve of this column chart shows a high degree of accuracy between actual and predicted values (Fig. 5f,g).

Fig. 5
figure 5

Predicting efficacy and the creation of a Nomogram plot. (a–b) Forest plots for single-factor and multifactor analyses. (c–e) Predictive efficacy of risk scores and clinical traits. (f–g) Establishment of nomogram and calibration curve.

Full size image

Immune-related analysis

In order to better understand the changes in immune checkpoints between the high-risk and low-risk groups, we found that immune checkpoints CD276, CD20, CD27, and TNFSF9 were highly expressed in the high-risk group (Fig. 6a). Next, we evaluated the tumor mutation burden between the two groups and found a significant difference in mutation burden between the high-risk and low-risk groups (Fig. 6b). There is an impact of the level of tumor mutation load on patient prognosis (Fig. 6c). The immune microenvironment between the two groups was evaluated, and significant differences in immune cells were observed between the two groups (Fig. 6d). In the high-risk group, compared with the low-risk group, CD4 memory T cells and M0 macrophages were significantly increased (Fig. 6e,f). The evaluation of TIDE between the two groups showed that the TIDE score of the high-risk group was significantly higher than that of the low-risk group, and the likelihood of immune escape in the high-risk group was higher than that in the low-risk group. This indicates that patients in high-risk populations may have poorer responses to immunotherapy (Fig. 6g). Finally, we confirmed this risk model using a database, which showed that patients in the high-risk group had a worse overall prognosis (Fig. 6g). Finally, we plotted the K–M curve. By comparing the survival rates of the high-risk and low-risk groups, it was found that there was a significant difference in OS between the two risk groups (Fig. 6h).

Fig. 6
figure 6

Result of Immune analysis. (a) Differences in immune checkpoints between the two groups. (b) Relationship between high and low TMB and patient prognosis. (c) Differences in tumor mutational burden between high and low-risk groups. (d) Differences in tumor microenvironment between high and low-risk groups. (e–f) Differences in immune cell infiltration between high and low-risk groups. (g) Differences in TIDE score between high and low-risk groups. (h) Differences in patient prognostic outcomes between the two groups.

Full size image

The drug sensitivity in LUAD

Finally, we screened potential drugs between the two groups and found that axitinib, doramapimod and leflunomide were more effective for patients in the high-risk group (Fig. 7).

Fig. 7
figure 7

Result of drug sensitivity analysis. Sensitivity of chemical drugs to different groups of patients.

Full size image

Discussion

Lung cancer has become the leading malignancy threatening human health. Among its primary pathological subtypes is lung adenocarcinoma. Despite advances in modern medicine, early detection, diagnosis, and treatment of malignant tumors remain a challenge. As a result, more than half of individuals are often diagnosed with lung cancer at an advanced stage due to the insidious symptoms of the disease. Millions of people still die from lung cancer every year. With the development of precision medicine and the deepening of research on the development process of cancer, the focus of treatment has shifted towards personalized treatment, with a greater emphasis on driving gene research. Consequently, searching for effective tumor therapeutic targets is an essential part of the current task.

With the deepening of research on programmed cell death, more and more studies have confirmed that the three apoptosis modes of cell pyroptosis, necroptosis, and apoptosis generally exhibit a dynamic balance. Therefore, based on this phenomenon, the concept of pan apoptosis was established5. Rajendra et al. found that Interferon regulatory factor 1 (IRF1) could inhibit colon tumor development by activating pan-apoptosis12. TNF-α and IFN-γ could inhibit the growth of various tumors by inducing pan-apoptosis13. Therefore, more and more experts and scholars believe that pan-cell apoptosis has excellent potential in preventing and treating tumours.

At present, there is relatively little research on lung adenocarcinoma. Immunotherapy is an emerging method for treating various types of tumors. However, its benefits vary from person to person. In addition, the relationship between pan-apoptosis and immunotherapy, clinical features, and prognosis has yet to be reported.

This document compiles the PANoptosis gene collection from the literature that has been published before. We are the first to use TCGA data to study patient prognostic expression data. We identified a total of 50 genes exhibiting different expression levels. The proteins CASP1, CASP3, CASP8, CYCS, TNF, MLKL, FADD, and IL1B could be vital in the progression of tumors, as depicted in the PPI network diagram. Further investigation was conducted on the correlation between and patient survival rates. The study's findings unequivocally showed that BAK1, CASP10, CYCS, FADD, HMGB2, IL1A, MLKL, TGDP1, TICAM1, TRAF2, UACA, YWHAG, and YWHAE significantly influenced the prognosis of patients.

These genes have been found to be indispensable in tumor development in previous literature. The miRNA-125b affects the outcome of breast cancer treatment by targeting and regulating BAK114. Zihang et al. found that Gelsolin increase CASP10 expression to inhibit colon cancer cell proliferation15. In glioblastoma, upregulation of CYCS can be used to inhibit tumor cell proliferation16. FADD is the most interesting of these genes. It mainly participates in post-translation modifications and has different meanings in different organizations. In lung cancer and cervical cancer, this usually indicates poor prognosis, but in thyroid cancer, the opposite is true17. Variations in HMGB2 expression are observed across multiple cancer types, such as skin cancer and hepatocellular carcinoma, and hold significance for predicting patient outcomes. Therefore, some have proposed that targeting this gene may be a feasible treatment method. Research has shown that HMGB2 plays a role as an oncogene through pp38MAPK in renal cell carcinoma and is associated with poor prognosis18. Tianke et al. used comprehensive bioinformatics analysis to determine that IL1A may be a potential target for oral cancer and is associated with poor overall prognosis in patients19. Sofie et al. described that the oncogenic or pro-oncogenic effects of MLKL are dependent on the tumor type and the tumor microenvironment, for example, with completely opposite effects between melanoma and pancreatic cancer20. Zhiyi et al. compared the survival rates of renal blastoma patients with high and low expression of TICAM1 and found that the overall survival rate of patients in the high expression group was significantly higher than that in the low expression group, which is considered a potential prognostic marker21. Yawei et al. discovered a high expression of TRAF2 in individuals with clear cell nephroblastoma, potentially linked to unfavourable outcomes22. The underlying process could be attributed to the role of TRAF2 in advancing tumors through the control of regulating macrophage polarization, migration and angiogenesis. It was proposed that UACA could be a possible focus of miR-30a-3p, serving as an independent prognostic factor for pancreatic cancer23. Jian et al. revealed that LncRNACERS6-AS1 promotes pancreatic cancer through YWHAG proliferation and metastasis24. Yi-Fang et al. deduced that the heightened expression of YWJAE enhances the growth and movement capabilities of breast cancer cells, whereas knockdown the YWHAE levels improves the effectiveness of chemotherapy for breast cancer25.

Next, we used cluster analysis on these genes and divided them into two groups based on the results, with k set to 2. There was a significant difference in the prognosis of the patients between the two groups. By comparison, the prognosis of cluster B was found to be much worse than that of cluster A. For cluster B, activated B cells and eosinophils are more abundant in cluster A, while other immune cells are more abundant or undifferentiated in cluster B. This indicates that the infiltration of tumor immune cells (such as CD56 natural killer cells and regular T cells) is more remarkable, leading to a poorer prognosis in cluster B. Daniela et al. described the function of neutrophils in detail and mentioned that there were five neutrophil states in the tumor cells and that infiltration of neutrophils predicted a poorer prognosis26. RegTcells were found to be associated with poorer prognosis of patients, advanced staging and metastasis of lung cancer. Regulary T cells may inhibit anti-tumor function through a variety of mechanisms26. For example, they exhibit suppression of anti-tumor function by expressing soluble immunosuppressive cytokines27,28. In mouse models, it has been found that tumor immune suppression can also be achieved by inhibiting natural killer cells29,30.

Then perform functional enrichment analysis to clarify the differences between their potential functions and the pathways involved. KEGG enrichment analysis revealed the central involvement of microRNAs in cancer. Different miRNAs have different effects on the prognosis of lung adenocarcinoma patients. Hongshuang et al.31 found that miR-1260b, miR-21-3p and miR-92a-3p were associated with early recurrence and metastasis in 160 samples of lung adenocarcinoma patients. Differences between the two clusters in the pathways revealed that PANoptosis genes play an essential role in the transformation of lung adenocarcinoma into non-small cell lung cancer. Statistically, less than 10 percent of patients with EGFR-mutated non-small cell lung cancer, especially lung adenocarcinoma, will transform into small cell lung cancer. This process usually occurs within 18 months of diagnosis. The primary mechanism of pathogenesis is unclear and may be related to p53 and its mechanism pathways. A common precursor transformation theory exists to explain this phenomenon. The reason for this is that small-cell lung cancer and non-small-cell lung cancer may originate from the same alveolar type II cells32. PANoptotic genes mainly control the cell cycle and their role in transformation deserves further research to discover. GO enrichment analysis revealed that it is mainly involved in nuclear division, organelle fission, chromosome region, extracellular matrix or structural organization in biological processes (BP), mainly in collagen-containing extracellular matrix, chromosomal region, condensed chromosome, chromosome, centromeric region (CC), and ATP hydrolysis activity in molecular function (MF). The above results suggest that it plays a vital role in cell growth, proliferation, and differentiation, further indicating its precise role in tumor development. Next, we calculated the risk score. Further validation was performed based on the median risk score with TCGA data as the train group and GEO as the test group for external data. We found a very consistent survival relationship in the experimental group, test group, and all data.

After comparing the survival data between the high-risk and low-risk groups based on clinical data such as lymph node metastasis, age, and gender, significant differences were found between the high-risk and low-risk groups. Then, we visualized the survival rates of samples from the high-risk and low-risk groups, and it was evident that the survival rates of patients in the high-risk group were lower than those in the low-risk group. A heat map was plotted for the expression of the selected genes, where ADM, FLNC, PITX3, LING02, and FSIP2 were expressed in the high-risk group, while C11ofF16 and MS4A1 were more abundantly expressed in the low-risk group.

Next, forest plots were used to demonstrate that risk score and stage are factors that affect patient prognosis. Then, based on the clinical information of the patient, the ROC curve and C-index curve were plotted, and it was found that the risk score had high accuracy in predicting patient prognosis. Subsequently, we created a column chart based on the clinical information of the patient and found that the expected and actual outcomes were closer in predicting patient prognosis. These column charts provide a more practical tool to assist clinicians in developing appropriate personalized treatment plans for LUAD patients, which can help improve clinical outcomes.

Further research was conducted on the immune-related differences between the two groups of patients. In contrast, we found significant differences in immune checkpoints between the high-risk and low-risk groups. In the high-risk group, the expression levels of CD276, CD20, CD27, and TNFSF9 are higher. There were also significant differences in TMB between the two groups. Meanwhile, it is not difficult to observe from the K–M curve that the prognosis of patients in the high TMB group is significantly better than that of patients in the low TMB group.

Next, we analyzed the tumor microenvironment between the two groups and found that although there was no difference in estimated scores between the two groups, there was a difference in immune scores. Meanwhile, we compared the immune cell differences between the two groups and found that plasma cell, T cell cd4 memory resting, DC resting, and mast cell resting differed between the two groups. Simultaneously, the low-risk group was higher than the high-risk group. Lastly, we evaluated TIDE between the two groups and found that the high-risk group had a higher likelihood of immune escape. According to the Imvigor database, there was a significant difference in OS between the two groups.

We finally screened patients in the high-risk group for sensitive drugs. Many sensitivity drugs were detected to to be more sensitive for patients in the high-risk group like Axitinib, Doramapimod, and leflunomide. Benjamin et al.33 found that Axitinib may benefit some advanced-stage patients in terms of survival. Shitong et al.34 found that Doramapimod may play a better role in cervical cancer. However, there are fewer studies on lung cancer. Rui et al.35 revealed the cancer-inhibitory effect of leflunomide. They found that leflunomide inhibited the growth of tumor cells and organized the metastasis of lung cancer cells through multiple pathways and modalities.

Although this study remedies a gap area of previous research. The PRGs in the signature could refine the prediction performance of LUAD survival outcome, and evaluate the immune conditions and forecast the immune checkpoints of LUAD patients. The PRG identified in this study may provide potential therapeutic targets or prognostic factors for LUAD patients. Our research has several limitations. This is a retrospective study based on gene expression profiles and some clinical variables in the TCGA database. Some specific clinical information related to lung cancer may not be available.

Conclusion

PANoptosis genes play a pivotal role in predicting the prognosis of LUAD patients, notably BAK1, CASP10, CYCS, FADD, HMGB2, IL1A, MLKL, TGDP1, TICAM1, TRAF2, UACA, YWHAG, as key PANoptosis genes. In addition, the PRG score can serve as a predictive indicator for the prognosis of patients with LUAD.