Establishing a prognostic model based on immune-related genes and identification of BIRC5 as a potential biomarker for lung adenocarcinoma patients

As an extremely malignant tumor, lung cancer is among the most commonly diagnosed cancers (11.6% of the total cases) and is the leading cause of cancer death (18.4% of all cancer deaths) [1]. All lung cancers consist of two main subtypes: small-cell lung carcinoma (SCLC) and non-small cell lung carcinoma (NSCLC) [2]. Accounting for over 40% of non-small cell lung cancers, lung adenocarcinoma (LUAD) is overwhelmingly the most common histologic type of lung cancer [3]. The most advantageous approach for managing patients with locally advanced non-small cell lung cancer (NSCLC) that is amenable to surgical resection involves administering chemoradiation as a minimum [4]. The utilization of trimodality treatment, which encompasses surgical resection, has been a contentious topic for numerous decades. Furthermore, for patients with inoperable or unresectable locally advanced disease, the adoption of immunotherapy consolidation following chemoradiation has established a novel benchmark for care. Despite improvements in surgery in recent years, the prognosis of lung cancer remains unfavorable. Thus, more comprehensive therapies are urgently needed. The development of cancer genomics in recent decades has permitted the identification of several gene alterations as driver gene mutations for LUAD, including anaplastic lymphoma kinase (ALK), epidermal growth factor receptor (EGFR) and KRAS [5‐7]. Several therapies that target these gene alterations have been employed. EGFR is found to be mutated in as much as 59.7% of NSCLC tumors in Asian patients and approximately 16.7% of those in Caucasian patients. Novel therapeutic agents known as tyrosine kinase inhibitors (TKIs) have been developed to specifically target these mutations, including erlotinib, gefitinib, and afatinib, which have demonstrated response rates of up to 75% [8]. The mechanism of targeted therapy focused on these mutation sites involves the use of drugs that specifically inhibit the activity of the altered protein. For example, EGFR inhibitors, such as erlotinib and gefitinib, block the activity of the EGFR protein, preventing the activation of downstream signaling pathways that promote cell growth and survival. Similarly, ALK inhibitors, such as crizotinib and ceritinib, block the activity of the ALK protein, which is often altered in lung cancer. In summary, the mechanism of targeted therapy focused on ALK, EGFR, KRAS, and other mutation sites involves the use of drugs that specifically inhibit the activity of the altered protein. These drugs can effectively target cancer cells while sparing healthy cells, reducing side effects, and improving patient outcomes. Despite improvements in the prognoses of some patients after receiving targeted treatments, a large number of patients eventually become resistant to targeted therapy [9]. For example, all patients possessing activating mutations in EGFR eventually encounter resistance to TKIs after a median duration of 12 months. The most prominent resistance mechanism observed is a secondary point mutation located in exon 20 of EGFR (T790M), wherein methionine is substituted by threonine at amino acid position 790 [10]. Under these circumstances, the advent of immunotherapy provides novel insight into lung cancer therapy.

The rapid development of cancer immunology in recent years has provided a novel perspective for cancer therapy [11]. A complex network has been well established to regulate interactions between the immune system and cancers. The human immune system can recognize and extinguish abnormal tumor cells. Immune checkpoint inhibitors (ICIs) have recently gained increasing attention as an essential part of immunotherapy [12]. Furthermore, tremendous advances in immune checkpoint blockade have introduced a paradigm shift in treatment for patients with lung cancer. In addition, immune checkpoint inhibitor (ICI) treatment has functioned as the standard of care for patients with extensive-stage small cell lung cancer or locally advanced/metastatic non-small cell lung cancer without EGFR/ALK alterations [13]. Thus, ICIs have been widely used in LUAD therapy. For example, immune checkpoint inhibitors (ICIs) that target programmed cell death 1 (PD-1) and programmed cell death-ligand 1 (PD-L1) play a significant role in the immune check-point pathway, exhibiting excellent and durable antitumor activity in LUAD patients [14]. The signaling pathway of programmed cell death 1 (PD-1) is often co-opted by malignant cells as a means of evading immunological scrutiny. Consequently, the PD-1 pathway serves to stifle T cell activities, such as their activation, proliferation, and production of cytokines. As it stands, antibodies that obstruct either PD-1 itself or its ligand, PD-L1, have gained regulatory approval for employment in treating an array of solid and hematologic neoplasms [15]. In addition to PD-1 and PD-L1, cytotoxic T-lymphocyte-associated antigen 4 (CTLA-4) also presents promising results in the treatment of advanced-stage lung cancer patients [16]. Despite the great impact of immunotherapy on the treatment of LUAD patients, many patients still experience disease progression during treatment or after treatment discontinuation due to immune resistance [17]. In immunocompetent individuals, neoplastic cells can undergo three outcomes: eradication, stasis, or evasion. Tumor immune evasion (TIE) refers to a mechanism by which the immune milieu of molded neoplasms can proliferate via an unbridled route [18]. The continual interactions amidst neoplastic cells and the neoplastic microenvironment are pivotal in neoplasm inception, advancement, metastasis, and reaction to therapeutic interventions [19]. The tumor microenvironment plays an important role in the immunotherapy response. Exploring the tumor microenvironment (TME) can improve the effect of cancer immunotherapy. Under these circumstances, constructing an immune gene signature is crucial to predict the prognosis and efficacy of LUAD immunotherapy.

In our study, a novel immune gene signature marker that can predict the response to immunotherapy was developed. After being validated by the GEO database, its prediction value in the prognosis of LUAD patients was proven to be excellent. Then, several clinicopathological characteristics were analyzed to explore the correlations between them and the prognostic model. To elucidate the TME of LUAD, the tumor mutation burden (TMB) and immune infiltration were further analyzed. In addition, the prediction of immunotherapy response and prognostic ability of various models were compared. Furthermore, BIRC5, an immune gene in the prognostic model, was identified to be significantly enriched in T cells by single-cell sequencing analysis. Finally, cell experiments were further performed to confirm the effects of BIRC5 on LUAD cells.

Recently, great progress has been made in immunotherapy, especially for non-small cell lung cancer (NSCLC), shedding novel light on the therapeutic strategy of patients diagnosed with NSCLC [30]. Nevertheless, many LUAD patients still suffer from this malignant tumor due to the low response rate [31]. The lack of precise therapeutic targets or limited knowledge of the TME might account for this dilemma [32]. The tumor microenvironment, consisting of not only diverse immune and stromal cells but also the factors they secrete, has been deemed to correlate with treatment efficacy and patient outcomes [33]. Under these circumstances, a prognostic model based on immune-related genes was constructed to help select patients for immunotherapy and discover potential biomarkers.

In our study, 675 DEIRGs were obtained between tumor and normal tissues based on TCGA and IMMPORT databases. Then, 56 immune-related genes were identified using univariate Cox regression analysis. Multivariate Cox regression analysis was applied to identify 13 key immune-related genes, calculate coefficients and construct the risk model. As expected, we found that patients in the high-risk group had shorter survival than those in the low-risk group. Subsequently, forest plots and a nomogram were constructed to evaluate the clinical applicability of the model. Plotting the ROC curve and survival curve established that our model had an excellent predictive effect. Furthermore, our model still performed well after external validation with two GEO datasets (GSE72094 and GSE26939). In addition, our model was remarkably correlated with prognostic malignant clinicopathologic characteristics (such as clinical stage, T stage, and N stage), further revealing its outstanding prognostic efficacy.

For the GSEA based on the DEIRGs, adaptive immune response, B-cell activation, and B-cell mediated immunity were significantly enriched, which could exert enormous influence on the tumor microenvironment, limiting tumor invasion to some extent [34, 35]. Then, all samples were subjected to GSEA based on the high- and low-risk groups. Interestingly, we found that mitotic nuclear division, which could lead to chromosomal instability and promote the migration of NSCLC [36], was remarkably enriched in the high-risk group. In addition, response to interleukin-1 and cellular response to interleukin-1 were also observed to be enriched. Besides, antigen receptor mediated signaling pathway was found enriched in high-risk group, which has been confirmed to mediate superior antitumor effects [37]. Recently, a paper revealed that the tumor response to cetuximab could be enhanced by increasing the levels of IL-1α [38]. In conclusion, patients in the high-risk group might have a worse prognosis but respond better to immunotherapy.

Tumors consist not only of cancer cells but also the tumor microenvironment, which consists of stromal cells (tumor-infiltrating immune cells and cancer-associated fibroblasts), the extracellular matrix, and various cytokines and metabolites. Representing most of the tumor mass, the TME actively participates in tumorigenesis [39]. The development and progression of cancer are accompanied by modifications in the adjacent stroma. Cancerous cells are capable of manipulating their microenvironment in a functional manner, by means of excreting diverse cytokines, chemokines, and other factors. As a consequence, a reprogramming of the neighboring cells is induced, allowing them to assume a decisive function in the sustenance and advancement of the neoplasm [40]. The TME can communicate with tumor cells, permitting them to proliferate and protecting them from apoptosis. Thus, the TME might play an essential role in therapeutic efficacy [41]. Under these circumstances, we comprehensively analyzed the TME with the ESTIMATE algorithm based on transcriptomic data. The immune score and stromal score represent the status of immune and stromal cell infiltration within the TME in LUAD. The results revealed that patients in the low-risk group shared remarkably higher immune scores than those in the high-risk group. In addition, the immune score was negatively correlated with clinical stage and positively associated with survival time, indicating that immune cells might function as protective factors, providing a favorable prognosis for patients diagnosed with LUAD. However, the stromal score was not found to be significantly linked to the risk score and clinical characteristics, which implied that stromal cells might not play a significant role in the tumorigenesis of our samples. Moreover, six immune subtypes of cancer could influence the prognosis by determining immune response patterns [42], which consist of C1 (wound healing), C2 (IFN-γ dominant), C3 (inflammatory), C4 (lymphocyte depleted), C5 (immunologically quiet) and C6 (TGF-γ dominant). The distribution of various immune subtypes between the high- and low-risk groups was analyzed by the chi-square test (Fig. S4). The results showed that the C1 and C2 subgroups accounted for more patients in the high-risk group (28% and 37%), while the low-risk group mainly correlated with the C3 subgroup. As reported, CD4⁺ T cells can function as tumor growth suppressors and induce cytolysis by secreting interferon-γ (IFN-γ) [43]. However, chronic inflammation can induce tumor progression, triggering treatment resistance [2]. To summarize, we can infer that patients in the high-risk group might respond well to immunotherapy based on immune subtype analysis.

Furthermore, to elucidate the TME immune landscape, we explored the infiltration status of 22 immune cells in LUAD. Consistent with previous results, most of the immune cells were enriched in the low-risk group, including resting memory CD4 T cells, monocytes, resting dendritic cells, and resting mast cells, which were related to a longer survival time. Correlated with worse prognosis, activated memory CD4 T cells and M0 macrophages were significantly enriched in the high-risk group.

Immune cell infiltration has been accepted to play an essential role in tumor progression and the response to immunotherapy in LUAD [44]. By eradicating tumor cells directly through cytolytic mechanisms or modulating the TME indirectly, CD4 + T cells can target tumor cells in various ways [45]. By helping to induce a gene expression program in CD8 + T cells that promotes cytotoxic T lymphocyte (CTL) function through various molecular mechanisms, CD4 + T cells assist CTLs in overcoming the barriers that sharply hinder antitumor immunity [46]. In addition, Probst, H C. et al. showed that peripheral CD8 + T-cell tolerance could result from antigen presentation by resting dendritic cells [47], revealing the vital role that resting dendritic cells play in immunotherapy resistance. On the other hand, numerous types of immune cells, comprising regulatory T cells, macrophages (M2), and terminally exhausted CD8 + T cells, have the potential to result in adverse clinical consequences due to their immune dysregulation [48]. These results indicated that patients in the high-risk group might have a better response to immunotherapy.

As a crucial part of immunotherapy, immune checkpoints can regulate T-cell effector function, bringing about breakthroughs and even constituting a paradigm shift in cancer therapy [49]. Recently, great efforts have been made to develop immune checkpoint blockade treatments, mainly targeting PD-1, PD-L1, and CTLA-4 [50]. However, in contrast to LUSC patients, LUAD patients benefit little from CTLA4 and anti-PD-1 or anti-PD-L1 therapy. Thus, improved ICI-based treatment approaches beyond those targeting the CTLA4 and PD-1/PD-L1 pathways are urgently needed. In this study, the mRNA expression levels of diverse immune checkpoints other than PD-1, PD-L1, and CTLA-4 were analyzed with the TCGA database. The results showed that CD27, IDO2, CD200R1, TNFRSF25, CD40LG, ADORA2A, and BTLA were significantly enriched in the low-risk group and remarkably correlated with vital clinicopathologic features. Wang, Qinchuan et al. revealed that several immune points (including BTLA, IDO, and CD27) were optimal biomarkers for tumor recurrence and survival in renal cell carcinoma patients, and a high expression level of BTLA was also found to be related to decreased survival [51]. Acting as gatekeepers of the immune response, several inhibitory immunoreceptors have been identified and exploited in past decades, including PD-1, CTLA-4, LAG3, TIM3, TIGIT and BTLA [52]. As surface molecules, their activity can be easily restrained by blocking antibodies that inhibit ligand‒receptor engagement [53]. In addition, activating costimulatory T-cell receptors is deemed a promising therapeutic strategy in clinical practice [54]. Furthermore, the six immune checkpoints were identified as significantly connected with genes in our prognostic model as well as the risk score, implying that patients in the low-risk group are more suitable for immunotherapy based on costimulatory receptor targeting therapy.

Then, the tumor immune dysfunction and exclusion (TIDE) algorithm, which simulates two main immune escape mechanisms of tumors to predict the ICI response, was used to predict the response to immunotherapy [55]. The TIDE score has excellent performance in tumor immune escape prediction [56], which illustrates that patients with lower scores are more likely to share favorable responses to immunotherapy. In our study, the high-risk group had a lower TIDE score, representing more benefits from immunotherapy. However, as discussed before, patients in the high-risk group presented a lower immune score, suggesting that high immune cell proportions do not necessarily predict high immunogenicity.

Tumor mutation burden (TMB) has been deemed an efficient biomarker not only for measuring the number of mutations in a cancer but also for immunotherapy response [57, 58], where higher TMB tends to correlate with more promising outcomes from immunotherapy. To further elucidate the immune characteristics of LUAD patients, gene mutations were analyzed based on high- and low-risk subgroups, where missense mutations were most common. Various genomic alterations, including alterations in EGFR, KRAS, ALK, and TP53, have been proven to be related to ICI efficacy [13]. In particular, cooccurring mutations in KRAS and TP53 have been determined to have predictive value in immune checkpoint inhibitors [59]. The top 20 genes with the highest mutation rates are displayed, among which TP53 shared the highest mutation frequency, with a higher level in the high-risk group. As reported before, TP53 was significantly correlated with oncogenic pathways, such as DNA replication, mismatch repair, and the cell cycle, contributing to undesirable clinical outcomes in LUAD patients [60]. In addition, mutations in MET, KRAS, and TP53 have been revealed to sharply correlate with high PD-L1 expression and a favorable ICI response [61]. These results corresponded with our observation that patients in the high-risk group had a worse prognosis but better immunotherapy responses. In addition, TMB was found to be positively associated with the risk score, further illustrating that patients in the high-risk group could benefit more from immunotherapy, and the prognostic model possesses excellent prediction value in measuring the TMB and immunotherapy response.

In contrast to bulk data that measure the averaged attributes of whole tissues, single-cell RNA sequencing (scRNA-seq) facilitates the identification of cell types and lineages of various cell subpopulations based on heterogeneous tissue ecosystems [62]. As discussed above, our prognostic model exhibited great efficacy in TMB measurement as well as prognosis and immunotherapy prediction. To explore the expression level of prognostic genes in different cell subpopulations, scRNA-seq was performed. The results showed that only BIRC5 was significantly enriched in T cells, which play a vital role in antitumor immunogenicity. Wang Y et al. revealed that baculoviral inhibitor of apoptosis protein (IAP) repeat containing 5 (BIRC5) expression can be regulated by the circCAMSAP1/miR-1182/BIRC5 axis, promoting NSCLC progression [63]. Besides, it has been revealed that the attenuation of the long non-coding RNA LINC00857 significantly augments the susceptibility of lung adenocarcinoma cells to radiotherapy, contingent upon BIRC5 expression, by inducing the recruitment of NF-κB1 [64]. As a well-known cancer therapeutic target, BIRC5 has been extensively researched, providing new insight into immunotherapy [65]. Based on these findings, we infer that BIRC5 could function as a biomarker and even therapeutic target in LUAD. Finally, the functional phenotype of BIRC5 was further explored by preliminary experiments. The significant BIRC5 mRNA levels in LUAD tissues and cell lines were confirmed by cell experiments, which was consistent with several findings obtained previously [66‐68]. In addition, BIRC5 gene knockdown in LUAD cell lines was proven to significantly inhibit the activity and proliferation of cancer cells. The elevated expression of BIRC5 has been notably demonstrated to significantly facilitate tumorigenesis and migration, exerting a profound influence on the early detection and accurate prediction of the immunotherapeutic response in patients with LUAD.

In summary, an immune-related prognostic model was constructed based on the TCGA database to predict the OS of LUAD patients, which was validated by the GEO database. The risk score and clinical stage were found to be independent prognostic factors. The immunotherapy response was further analyzed, reflecting our model’s robust and capacious perspective in utilization. Unavoidably, deficiencies remain in our study because this is still a retrospective analysis. Thus, a prospective study or clinical samples and methods of animal models in vivo are needed to further confirm our results.

We performed comprehensive bioinformatics analysis and identified a predictive model for LUAD prognosis based on thirteen immune-related genes. We assessed the prognosis and immunological microenvironment of LUAD patients via this model. By scRNA analysis, the expression of BIRC5 was identified significantly high in T cell. We also verified the role of BIRC5 in LUAD by cell assay, which may provide novel insight into LUAD patients management.