Introduction

Oncology Letters

1792-1074 1792-1082

D.A. Spandidos

10.3892/ol.2025.14902

OL-29-3-14902

Articles

Construction of a lung adenocarcinoma prognostic model based on KEAP1/NRF2/HO‑1 mutation‑mediated upregulated genes and bioinformatic analysis

Zhu

Wei

1 * Zhang

1 * Yang

Lingyun

2 * Chen

1 Chen

Chaobo

3 4 Shi

Qifeng

1 Xu

Zipeng

1Department of Pathology, Xishan People's Hospital of Wuxi City, Wuxi, Jiangsu 214105, P.R. China 2Department of Renal and Rheumatology, Affiliated Children's Hospital of Jiangnan University (Wuxi Children's Hospital), Wuxi, Jiangsu 214000, P.R. China 3Department of General Surgery, Xishan People's Hospital of Wuxi City, Wuxi, Jiangsu 214105, P.R. China 4Department of Hepatobiliary and Transplantation Surgery, The Affiliated Drum Tower Hospital of Nanjing University Medical School, Nanjing, Jiangsu 210000, P.R. China

Correspondence to: Dr Zipeng Xu, Department of General Surgery, Xishan People's Hospital of Wuxi City, 1128 Dacheng Road, Wuxi, Jiangsu 214105, P.R. China, E-mail: xuzipeng1989@126.com yuehongshen@hotmail.com Dr Qifeng Shi, Department of Pathology, Xishan People's Hospital of Wuxi City, 1128 Dacheng Road, Wuxi, Jiangsu 214105, P.R. China, E-mail: shiqifeng1969@sina.com yuehongshen@hotmail.com *

Contributed equally

03 2025

23 01 2025

29 3

155

27082024 06012025

2025

This is an open access article distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivs License, which permits use and distribution in any medium, provided the original work is properly cited, the use is non-commercial and no modifications or adaptations are made.

Lung adenocarcinoma (LUAD) is a prevalent malignant tumor of the respiratory tract. The Kelch like ECH associated protein 1 (KEAP1)/nuclear factor erythroid 2-related factor 2 (NRF2)/heme oxygenase 1 (HO-1) axis serves a pivotal role in the occurrence and progression of LUAD. The present study aimed to identify specific genes regulated by mutations of the KEAP1/NRF2/HO-1 axis and to investigate their prognostic potential in LUAD, as well as their association with the tumor microenvironment. Immunohistochemistry was performed to assess the expression levels of KEAP1, NRF2 and HO-1 in LUAD tissues and to evaluate their association with clinical pathology. Sequencing data and clinical information were obtained from The Cancer Genome Atlas (TCGA)-LUAD and Gene Expression Omnibus (GSE68465) databases, whilst mutation information was sourced from the cBio Cancer Genomics Portal website. The R package ‘limma’ and Venn diagram were utilized to identify upregulated differentially expressed genes. Subsequently, a prognostic model was constructed using univariate Cox regression analysis and 101 machine learning methods. A nomogram of the prognostic model was generated to assess its efficacy in evaluating survival among patients with LUAD. The ‘ImmuCellAI’ and ‘oncoPredict’ R packages were used to compare and evaluate differences in immune cell infiltration and immunotherapy between high- and low-risk groups, as well as the sensitivity of LUAD to chemotherapy drugs. Compared with the group with negative expression, the results revealed that the group with positive expression of NRF2 and HO-1 exhibited advanced tumor, lymph node and clinical stages and a worse prognosis. A predictive model incorporating four genes (kynureninase, serpin family B member 5, insulin like 4 and γ-aminobutyric acid type A receptor subunit α3) was constructed based on KEAP1/NRF2/HO-1 mutation-mediated upregulated genes (KNHMUGs). Risk score was an independent prognostic factor for patients with LUAD (hazard ratio, 1.038; 95% confidence interval, 1.034–1.043; P<0.001). A nomogram was developed to predict the prognosis of patients with LUAD, which was validated as a reliable prognostic tool. The low-risk group exhibited higher immune cell infiltration, including CD4⁺ T, CD8⁺ T, natural killer (NK) and NKT cells, compared with the high-risk group. In addition, it demonstrated increased expression levels of immune checkpoint inhibitory genes such as cytotoxic T-lymphocyte associated protein 4, T cell immunoreceptor with Ig and ITIM domains, hepatitis A virus cellular receptor 2 and B and T lymphocyte associated protein. Moreover, it displayed enhanced sensitivity to immunotherapy. Drug sensitivity analysis revealed that the high-risk group exhibited increased sensitivity towards vinblastine, docetaxel and cisplatin, whereas the low-risk group showed increased sensitivity to BMS_754807, SB505124_1194 and JQ1_2172. In conclusion, a KNHMUGs-based gene signature was constructed in the present study, which holds promise as a biomarker for evaluating patient prognosis and guiding treatment by effectively assessing immunotherapy response and chemotherapy sensitivity in patients with LUAD.

lung adenocarcinoma Kelch like ECH associated protein 1/nuclear factor erythroid 2-related factor 2/heme oxygenase 1 mutation prognostic model tumor immune microenvironment immunotherapy drug prediction

Top Talent Support Program for Young and Middle-Aged People of Wuxi Health Committee

HB2023116

Innovation Cultivation Fund of Xishan People's Hospital

202103

The present research was supported by the scientific research projects of the Top Talent Support Program for Young and Middle-Aged People of Wuxi Health Committee (grant no. HB2023116) and the Innovation Cultivation Fund of Xishan People's Hospital (grant no. 202103).

Introduction

Lung cancer is the most common malignant tumor, ranking as the second most prevalent and lethal malignant tumor globally (1). Lung cancer can be broadly classified into two main subtypes: i) Non-small cell lung cancer (NSCLC) and ii) small cell lung cancer, from both pathological and therapeutic perspectives. Lung adenocarcinoma (LUAD) is a pathological subtype of NSCLC, accounting for ~45% of all cases (2). Despite important advancements in molecular targeting, chemotherapy, radiation therapy and immunotherapy, the prognosis of LUAD remains unfavorable, with a 5-year overall survival rate of ~15% (3,4). Previous studies have demonstrated the pivotal role of genetic analysis in the diagnosis and treatment of lung cancer (5–7). For instance, the polymorphisms of the epidermal growth factor receptor are associated with side effects and outcome of tyrosine kinase inhibitor therapy administered to patients with NSCLC (5–7). Therefore, there is an urgent need for novel and effective screening methods based on genetic analysis to improve diagnostic accuracy and treatment efficiency, in order to improve the prognosis of patients with LUAD.

The Kelch like ECH associated protein 1 (KEAP1)/nuclear factor erythroid 2-related factor 2 (NRF2)/heme oxygenase 1 (HO-1) pathway serves a pivotal role in metabolic reprogramming, immune remodeling and treatment resistance in LUAD (8). In total, 20–30% of LUADs exhibit KEAP1 loss-of-function mutations, leading to aberrant activation of NRF2 and subsequently resulting in poor prognosis (9,10). KEAP1 mutations facilitate the progression of LUAD by augmenting glutamine metabolism and upregulating solute carrier family 33 member 1 expression (11,12). Furthermore, KEAP1-mutant tumors diminish dendritic cell and T cell responses, thus driving immunotherapy resistance in LUAD (13). NRF2, a transcription factor, serves a crucial role in the antioxidant pathway, promoting tumor cell growth, proliferation and drug resistance (14). NRF2 is the target of KEAP1 and abnormal activation of NRF2 is not only associated with KEAP1 mutations but also with gain-of-function mutations in NRF2 (15). Previous studies have found multiple gain-of-function mutation sites affecting NRF2 in LUAD (16). NRF2 mutations promote tumor cell immune escape by inhibiting stimulator of interferon response CGAMP interactor 1 activation (17). HO-1, which is regulated by NRF2, serves a crucial role as an antioxidant factor. When HO-1 expression is dysregulated, it can promote the proliferation of lung cancer cells and make them resistant to radiotherapy (18,19). However, several crucial and unresolved issues remain, encompassing the potential reclassification of LUAD based on KEAP1/NRF2/HO-1 mutations, the intricate interplay between KEAP1/NRF2/HO-1 mutations and the tumor microenvironment, as well as the strategic implementation of therapeutic interventions to effectively control tumor progression contingent upon KEAP1/NRF2/HO-1 mutations.

Therefore, the present study aimed to identify specific genes regulated by gene mutations of the KEAP1/NRF2/HO-1 axis and construct a prognostic model using upregulated genes. All patients were stratified into low- or high-risk groups based on the risk score, and the diagnostic efficacy was validated in both the training and test sets. Furthermore, the association between risk score and immune cell infiltration in the tumor microenvironment was explored and the individualized sensitivity to chemotherapy and immunotherapy were simultaneously predicted for each patient with LUAD.

Materials and methods <sec> <title>Validation of the expression of KEAP1, NRF2, HO-1 and Ki-67

Postoperative paraffin-embedded tissue blocks of patients with LUAD who underwent surgical procedures at Xishan People's Hospital of Wuxi City (Wuxi, China) between January 2020 and June 2023 were included in the present study. The inclusion criterion was a histologically confirmed diagnosis of LUAD, whilst the exclusion criteria were prior chemotherapy or radiation therapy. After selection, a total of 104 tissue wax blocks from patients with LUAD were included in the study. The patients consisted of 42 men and 62 women, with an mean age of 66.6 years (range, 32–92 years). The present study was approved by the Ethics Committee of Xishan People's Hospital (approval no. xs2023ky021).

Tissue sections (5 µm) were prepared and the expression of KEAP1, NRF2 and HO-1 was assessed using immunohistochemistry as previously described (19), using anti-KEAP1 (cat. no. ab226997; 1:100 dilution; Abcam), anti-NRF2 (cat. no. ab313825; 1:100; Abcam), anti-HO-1 (cat. no. ab137749; 1:100; Abcam) and anti-Ki-67 (cat. no. ab230460; 1:100; Abcam) antibodies as the primary antibodies. The Elivision™ plus Polyer HRP (mouse/rabbit) IHC Kit (cat. no. KIT-9902; Maixin Biotech. Co., Ltd.) served as the secondary antibody reagent in this study.

Data acquisition

High-throughput sequence-fragments per kilobase of transcript per million mapped reads data and clinicopathological information of patients with LUAD were acquired from The Cancer Genome Atlas (TCGA)-LUAD database (https://portal.gdc.cancer.gov/). Gene expression and survival data associated with the GSE68465 (20) dataset were obtained from the Gene Expression Omnibus (GEO) database (http://www.ncbi.nlm.nih.gov/geo/). Gene mutation information for KEAP1, NRF2 and HO-1 was retrieved from the cBio Cancer Genomics Portal (cBioPortal) website (http://www.cbioportal.org/). The association between KEAP1, NRF2 and HO-1 expression levels and patient survival was analyzed using the Kaplan-Meier Plotter website (https://kmplot.com/analysis/index.php?p=service) (21). All patients diagnosed with lung adenocarcinoma across the 17 datasets were included in this study. All possible cut-off values between the lower and upper quartiles were computed and the best performing threshold was used as a cut-off.

Identification of mutation-associated differentially expressed genes

Based on the mutation information obtained from the cBioPortal website, 112 tumor samples were classified into the mutation group, 403 tumor samples into the wild-type group and 59 normal pulmonary tissues into the normal group. The ‘limma’ package in R (version 4.1.3; http://www.r-project.org) was utilized to identify differentially expressed genes (log₂ fold-change >1; P<0.05) by comparing gene expression between the mutant group and both the wild-type and normal groups (22).

Construction of a prognostic model

Candidate genes in the TCGA-LUAD and GSE68465 datasets were initially screened using univariate Cox analysis. Subsequently, 101 combinations of algorithms were used to construct a prognostic model based on the leave-one-out cross-validation framework (23). The TCGA-LUAD dataset was utilized as the training group, whilst the GSE68465 dataset served as the validation group. Furthermore, the consistency index (C-index) of each model was calculated in the validation group (23).

Survival analysis and nomogram construction based on clinical characteristics

Univariate and multivariate Cox regression analyses were used to identify independent prognostic factors in patients with LUAD. The ‘rms’ package in R (version 4.1.3) was utilized for constructing a nomogram based on risk score and two characteristic factors. Calibration curves were employed to evaluate the accuracy of these predictions. Additionally, decision curves generated using the ‘rmda’ package in R (version 4.1.3) were used to assess the predictive utility of the nomogram and to compare different prediction models.

Immune infiltration assessment and immunotherapy response prediction

The ‘ESTIMATE’ package in R (version 4.1.3) was used to evaluate the levels of immune cells and stromal cells in tumor tissues. The ImmuCellAI website (https://guolab.wchscu.cn/ImmuCellAI/#!/) was utilized for the analysis of immune cell types and levels within tumor tissues, facilitating a comparative assessment of immune cell levels between low- and high-risk groups (24). The effectiveness of immunotherapy in the aforementioned groups was evaluated through the prediction of individual response to immunotherapy among patients with LUAD.

Assessment of chemotherapy effectiveness

The ‘oncoPredict’ package in R (version 4.1.3) was used to predict chemotherapy drug sensitivity in patients with LUAD, aiming to compare the sensitivity to chemotherapy drugs between the high- and low-risk groups (25).

Expression of hub genes at the single-cell level

The Tumor Immune Single-cell Hub website (http://tisch1.comp-genomics.org/) was utilized to evaluate the expression of these hub genes in several cell types.

Statistical analysis

Statistical analyses were performed using R software (version 4.1.3). The Wilcoxon rank-sum test was used to determine the significance of differences between two groups. Frequency counts were used for statistical description of categorical (qualitative) data, and comparisons between groups were performed using either the χ² test or the Fisher's exact test. All P-values were two-tailed. P<0.05 was considered to indicate a statistically significant difference.

Results <sec> <title>Association between KEAP, NRF2 and HO-1 expression and clinicopathology

The expression of KEAP1, NRF2 and HO-1 was assessed using immunohistochemistry based on 104 cases of LUAD and their association with pathological characteristics was evaluated. The immunohistochemical analysis revealed that KEAP1 and HO-1 were predominantly expressed in the cytoplasm, whereas NRF2 and Ki-67 were primarily localized in the nucleus (Fig. 1). No significant association was demonstrated between the expression of KEAP1 and clinical features (Table I). Compared to the NRF2-negative group, the NRF2-positive group exhibited enhanced tumor aggressiveness, higher proportions of lymph node and distant metastases, more advanced TNM stage (26) and increased proliferative activity (Table I). Compared to the HO-1-negative group, the HO-1-positive group exhibited significantly higher grades in T stage, N stage, M stage and the overall TNM staging system. Additionally, the HO-1 positive group demonstrated a larger tumor diameter and a markedly increased expression level of Ki-67 (Table I).

Association between KEAP1, NRF2 and HO-1 expression and prognosis of patients with LUAD

The association between the expression of KEAP1, NRF2 and HO-1 and the prognosis of patients with LUAD was analyzed using the Kaplan-Meier Plotter website. The results indicated that, compared with low KEAP1 expression, high KEAP1 expression was not associated with overall survival (OS; Fig. 2A), but was significantly associated with improved first progression survival (FPS; Fig. 2B). Conversely, high expression of NRF2 and HO-1 was significantly associated with worse OS and FPS, in comparison with low NRF2 and HO-1 expression (Fig. 2C-F). The association between KEAP1, NRF2 and HO-1 expression levels and disease-specific survival (DSS), disease-free interval and progression-free interval (PFI) were further assessed based on data from the TCGA-LUAD dataset. The results revealed that only the expression of HO-1 was significantly associated with poor DSS and PFI (Fig. S1).

Identification of KEAP1/NRF2/HO-1 mutation-mediated upregulated genes (KNHMUGs)

From the cBioPortal database, KEAP1, NRF2 and HO-1 mutations in LUAD from TCGA pan-cancer atlas, which included 566 patients, were analyzed. Genetic alterations, such as in-frame, missense and truncation mutations of KEAP1 (19%), NRF2 (4%) and HO-1 (1.8%), were identified in patients with LUAD (Fig. 3). Based on the status of KEAP1/NRF2/HO-1 mutations, patients were classified into two groups: i) Mutant and ii) wild-type. Subsequently, the differential genes between the mutant, wild-type and normal groups were compared. A total of 2,346 upregulated genes specific to the mutant group were identified compared with the normal group (Fig. 4A), with 253 upregulated genes in the mutant group compared with the wild-type group (Fig. 4B). Furthermore, 172 common differentially upregulated genes in the mutant group were identified using a Venn diagram (Fig. 4C).

Prognostic model based on KNHMUGs

Univariate Cox regression analysis was then performed on 172 candidate genes in the TCGA and GSE68465 datasets. The TCGA dataset identified a total of 53 genes associated with prognosis, whereas the GSE68465 dataset revealed 29 genes that were prognostically relevant. Ultimately, an overlap of 21 genes with prognostic significance was observed in both datasets (Fig. 4D). Subsequently, these 21 genes were subjected to analysis using a combination of machine algorithms (101 combinations) with TCGA as the training set and GSE68465 as the validation set. By evaluating the C-index of both sets and considering the number of genes in optimal combinations, StepCox (backward) + random survival forest (RSF) algorithms emerged as the most effective prediction model (Fig. 5A), comprising four characterized genes, kynureninase (KYNU), serpin family B member 5 (SERPINB5), insulin like 4 (INSL4) and γ-aminobutyric acid type A receptor subunit α3 (GABRA3). Based on the median risk score, patients were stratified into high- and low-risk groups. Notably, OS was significantly lower for patients classified as high-risk compared with that for patients categorized as low-risk in both the TCGA [hazard ratio (HR), 12.21; 95% confidence interval (CI), 8.00–18.64; P<0.001] and GSE68465 (HR, 1.36; 95% CI, 1.05–1.75; P=0.012) datasets (Fig. 5B and D). It was also observed that the survival rate of patients significantly decreased as the risk score increased. Notably, the KYNU, SERPINB5, INSL4 and GABRA3 genes were identified as risk factors with an increasing trend in expression as the risk score rose (Fig. 5C and F).

Association between risk score and clinicopathological features

The association between risk score and clinicopathological characteristics was assessed. Patients with advanced-stage tumors [tumor stage (T) 3–4; lymph node stage (N)1-3; and stage III–IV] demonstrated a significantly higher risk score compared with those with early-stage tumors (T1-2; N0; and stage I–II), suggesting that this could be one of the factors contributing to the unfavorable prognosis observed in the high-risk group (Fig. 6A-C).

Construction of the nomogram

Univariate and multivariate Cox regression analyses were performed to assess the clinical characteristics and risk score of 490 patients (Fig. 6D and E). The results of the multivariate regression analysis revealed that risk score, N stage and metastasis (M) stage independently served as prognostic factors for patients with LUAD. Consequently, a nomogram was constructed based on these three independent risk factors (Fig. 6G) and calibration curves demonstrated a notable agreement between the predicted and actual values of the prediction model at 1, 2 and 3 years (Fig. 6F).

Association of risk score with immune infiltration and immunotherapy

The ESTIMATE results revealed that the low-risk group had significantly increased stromal and immune scores compared with the high-risk group, indicating a higher level of immune cell infiltration within this group (Fig. 7A). Subsequently, further analysis on immune cell infiltration in tumor tissues was performed. It was revealed that the low-risk group exhibited significantly enhanced infiltration of CD4⁺ naive, CD8⁺ naive, CD4⁺ T, CD8⁺ T, natural killer (NK), NKT cells and other immune cells compared with the high-risk group (Fig. 7C). Additionally, the expression levels of immune checkpoint-related genes were assessed. Notably, the low-risk group demonstrated a significantly increased expression of immune checkpoint-stimulated genes such as CD27, CD80 and C-X3-C motif chemokine ligand 1, as well as immune checkpoint-inhibited genes such as cytotoxic T-lymphocyte associated protein 4 (CTLA4) and hepatitis A virus cellular receptor 2 (HAVCR2), compared with the high-risk group (Fig. 7D and E). Furthermore, the efficacy of immunotherapy in different subgroups was predicted, and it was revealed that immune-checkpoint blockade therapy response prediction indicated a higher rate of response in the low-risk group compared with that in the high-risk group (Fig. 7B).

Association of risk score with drug therapy

The ‘oncoPredict’ R package was used to compare the differences in chemotherapy sensitivity between the high- and low-risk groups. The results demonstrated that the high-risk group had a significantly increased sensitivity to vincristine, docetaxel and cisplatin, whilst the low-risk group exhibited a significantly increased sensitivity to BMS_754807 (insulin receptor inhibitor), SB505124_1194 (TGF-β receptor inhibitor) and JQ1_2172 [bromodomain and extraterminal (BET) inhibitor] (Fig. 8).

Expression of hub genes in different risk groups and their localization in LUAD

To analyze the expression of hub genes in different risk groups, the mRNA expression levels of hub genes were assessed. The findings revealed that KYNU, SERPINB5, INSL4 and GABRA3 were significantly upregulated in the high-risk group compared with that of the low-risk group (P<0.05; Fig. 9A). Additionally, the Tumor Immune Single-cell Hub website was utilized to evaluate the expression of these hub genes in several cell types. Annotation analysis of the GSE148071 single-cell dataset demonstrated that tumor tissue consists of diverse cells, including malignant cells, epithelial cells and fibroblasts. Notably, KYNU, SERPINB5, INSL4 and GABRA3 exhibited a predominant expression in malignant cells (Fig. 9B). These results suggest that these hub genes primarily contribute to LUAD progression by modulating biological functions, specifically within malignant cells.

Discussion

A total of 20–30% LUAD tumors carry mutations in KEAP1 and NRF2. Both loss-of-function mutations in KEAP1 and gain-of-function mutations in NFE2-like BZIP transcription factor 2 can activate the KEAP1/NRF2 pathway in LUAD (15). Previous studies have demonstrated that the KEAP1/NRF2/HO-1 axis inhibits cisplatin-induced ferroptosis by upregulating the iron death-related genes solute carrier family 7 member 11 and NAD(P)H quinone dehydrogenase 1 (27,28). Furthermore, the NRF2/HO-1 axis could attenuate resistance to programmed cell death 1 immunotherapy by inhibiting the transformation of M2 into M1 macrophages (29). Notably, KEAP1/NRF2 mutations are considered a pivotal driver of immune-suppressive tumor microenvironment metabolic reprogramming in LUAD, including potent induction of T-regulatory cells and diminishing the number of dendritic cells and tissue-resident memory CD8+ T cells, thereby compromising the efficacy of immunotherapy (13,30,31). However, combination therapy involving glutaminase inhibition and immune checkpoint blockade is reported to hold promise for reversing the immune suppression mediated by KEAP1/NRF2 mutations (13,30). Mutations in the KEAP1/NRF2/HO-1 genes serve a crucial role in the development of LUAD. However, there is a lack of established clinical models to assess the prognosis, tumor microenvironment and response to chemotherapy and immunotherapy in patients with LUAD based on genes regulated by KEAP1/NRF2/HO-1 mutations.

The prognostic model in the present study consisted of four genes: KYNU, SERPINB5, INSL4 and GABRA3. KYNU acts as a key enzyme in tryptophan metabolism, and its metabolite 3-hydroxyanthranilic acid is notably associated with poor prognosis in patients with NSCLC (32,33). Further research reported that NRF2 upregulated KYNU, which was positively associated with an inhibitory tumor microenvironment (34). SERPINB2 belongs to the family of protease inhibitors, which is associated with tumorigenesis and regulation of several vital biological processes (35,36). The expression of SERPINB5 has been reported to be upregulated in LUAD, and its high expression levels were notably associated with poor OS. Furthermore, overexpression of SERPINB5 promoted cell proliferation, migration and invasion and epithelial-mesenchymal transition (EMT) in LUAD cells (37). INSL4 is a member of the insulin superfamily, which includes genes for insulin, insulin-like growth factors and relaxin. In LUAD, high levels of INSL4 expression have been reported to be associated with poor prognosis (38). Increased expression of INSL4 can promote tumor cell proliferation and invasion (38,39). GABRA3 acts as a key chloride ion transporter and also functions as a subunit of γ-aminobutyric acid A receptors (40). In LUAD, increased expression of GABRA3 has been closely associated with poor prognosis. By upregulating the expression of MMP-2 and MMP-9 via activation of the JNK/activator protein-1 signaling pathway, GABRA3 facilitates lymphatic metastasis in LUAD (41). Therefore, the aforementioned four genes are associated with the pathogenesis and progression of LUAD and may be considered potential therapeutic targets. Currently, there are studies investigating the regulatory association between KEAP1, NRF2 and HO-1 and KYNU (34). Subsequent research should further explore the interaction between KEAP1, NRF2 and HO-1, and SERPINB5, along with INSL4 and GABRA3.

The tumor microenvironment consists of diverse immune cells, stromal cells, extracellular matrix and several cytokines (42). These different components not only impact patient prognosis but also have a close association with immunotherapy (43). The present study revealed that the low-risk group exhibited a higher percentage of infiltration of CD4⁺ T, CD8⁺ T, NK and NKT cells compared with the high-risk group. Previous research has demonstrated that an increased presence of CD4⁺ T and CD8⁺ T cells is associated with improved prognosis and enhanced immunotherapeutic efficacy (44). NK cells are a subset of innate immune cells capable of rapidly recognizing and eliminating tumor cells whilst also promoting adaptive T-cell immune responses to suppress tumor aggressiveness (45). NKT cells represent a class of lymphocytes expressing both NK and T cell surface markers. They serve as a link between intrinsic and adaptive immunity and serve a crucial role in the tumor immune microenvironment (46). Therefore, increased infiltration of these cells may explain why the low-risk group exhibited improved prognosis compared with the high-risk group.

At present, research on immune checkpoint inhibitors is a hot topic. The current study assessed the expression of immune checkpoint suppressor genes in both the high- and low-risk groups. The results revealed high expression of immune checkpoint inhibitors, including CTLA4, T cell immunoreceptor with Ig and ITIM domains (TIGIT), HAVCR2 and B and T lymphocyte associated protein (BTLA) in the low-risk group. CTLA4 is a suppressor molecule located on the surface of T cells that can promote tumor tolerance and T-cell exhaustion when abnormally expressed (47). TIGIT is an immune checkpoint gene expressed on several types of T and NK cells. Its binding to CD155 on immune-suppressive cells or tumor cell surfaces inhibits immune cell function (48). HAVCR2, as a negatively regulated immune checkpoint, not only regulates the exhaustion of T and NK cells, but also regulates the intrinsic immune escape of tumors mediated by macrophages and dendritic cells (49). BTLA is a suppressor expressed on the surface of immune cells such as T and B cells. Upon binding to herpesvirus entry mediator, it sends inhibitory signals to T cells, leading to a decrease in T cell proliferation and activation (50). These findings indicate that there are more immunotherapeutic targets within the low-risk group with higher potential benefits from immunotherapy compared with the high-risk group, aligning with the predicted response rate for immunotherapy.

Notably, chemotherapy is a crucial component of LUAD treatment. Half-maximal inhibitory concentration analysis was performed to screen potential compounds and the findings revealed that the high-risk group exhibited greater sensitivity to vincristine, docetaxel and cisplatin, whilst the low-risk group showed increased sensitivity to BMS_754807, SB505124_1194 and JQ1_2172. Clinically, vincristine, docetaxel and cisplatin are commonly utilized chemotherapeutic agents for LUAD treatment, which are often administered in combination to enhance patient prognosis (51). BMS-754807 is a potent inhibitor of the insulin-like growth factor 1 receptor/insulin receptor family of kinases that synergistically interacts with platinum-based chemotherapeutics to induce apoptosis in lung cancer cells (52). Furthermore, dasatinib combined with BMS-754807 could inhibit lung cancer cell proliferation by inducing autophagy and arresting the cell cycle at G1 phase (53). SB505124_1194 is a selective TGF-β1 receptor inhibitor that notably impedes TGF-β1-mediated EMT in lung cancer cells, thereby reducing recurrence and metastasis (54,55). In the present study, higher expression of TGF-β1 in the low-risk group was associated with drug prediction. JQ1_2172 is a small molecule inhibitor targeting the BET protein that binds reversibly to its bromodomain, thus disrupting its interaction with acetylated lysines on histones or transcription factors. This inhibition leads to the suppression of oncogene expression, ultimately resulting in cancer cell proliferation arrest (56). Additionally, JQ1_2172 inhibits the upregulation of programmed death-ligand 1 during concurrent radiotherapy for lung cancer, thus exerting antitumor immune escape effects (57). Despite their effective eradication of tumor cells, anti-neoplastic drugs exert certain effects on both patients and hospital workers exposed to them, including increased DNA damage, elevated oxidative stress parameters and impairment on complete blood counts (58). Therefore, the implementation of personalized treatment strategies based on risk score could facilitate the selection of appropriate drug treatment plans, thereby enhancing treatment efficacy and minimizing the adverse effects of anti-neoplastic drugs on patients and healthcare workers.

Although a robust prognostic model based on KNHMUGs was constructed in the present study using public data, it is relied on public databases and further clinical validation is therefore required to confirm the model efficacy in prognostic prediction and assessment of immunotherapy and chemotherapy sensitivity. Secondly, whilst the hub genes screened in the present study have been investigated in a small number of LUAD cases, their mechanisms remain unclear and require further exploration. Subsequent studies should be conducted to address the aforementioned questions.

In summary, KEAP1, NRF2 and HO-1 expression is closely associated with clinicopathological features and prognosis in LUAD. The prognostic model constructed based on KNHMUGs may serve as an independent prognostic factor for predicting the prognosis of patients with LUAD. It integrates important clinical parameters and risk scores of prognoses, demonstrating reliability and effectiveness. Furthermore, the association between risk score and tumor immune microenvironment, immunotherapy and chemotherapy response may offer a potentially powerful new tool for clinical decision-making, patient prognosis determination and personalized treatment.

Supplementary Material

Supporting Data

Acknowledgements

Not applicable.

Availability of data and materials

The data generated in the present study may be requested from the corresponding author.

Authors' contributions

ZX designed the research and drafted the manuscript. CC and QS utilized software code to perform an in-depth analysis and visualization of the data and reviewed the present paper. WZ and YZ analyzed the paraffin sections. LY designed the mathematical methods, collected and analyzed the data. LC performed the immunohistochemistry. WZ and ZX confirm the authenticity of all the raw data. All authors read and approved the final manuscript.

Ethics approval and consent to participate

The present study complied with the Declaration of Helsinki and was approved by the Xishan People's Hospital Ethics Committee (approval no. xs2024ky037). The samples used in the present study were sourced from Xishan People's Hospital in Wuxi City, collected from patients with LUAD undergoing surgery between January 2020 and June 2023. Written informed consent forms were signed by all patients.

Patient consent for publication

Not applicable.

Competing interests

The authors declare that they have no competing interests.

References 1

Sung

Ferlay

Siegel

Laversanne

Soerjomataram

Jemal

Bray

Global cancer statistics 2020: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries

CA Cancer J Clin712092492021

10.3322/caac.21660

33538338

Denisenko

Budkevich

Zhivotovsky

Cell death-based treatment of lung adenocarcinoma

Cell Death Dis91172018

10.1038/s41419-017-0063-y

29371589

Reck

Rabe

Precision diagnosis and treatment for advanced non-small-cell lung cancer

N Engl J Med3778498612017

10.1056/NEJMra1703413

28854088

Yan

Advances and challenges in the treatment of lung cancer

Biomed Pharmacother1691158912023

10.1016/j.biopha.2023.115891

37979378

Obradovic

Todosijevic

Jurisic

Application of the conventional and novel methods in testing EGFR variants for NSCLC patients in the last 10 years through different regions: A systematic review

Mol Biol Rep48359336042021

10.1007/s11033-021-06379-w

33973139

Jurisic

Vukovic

Obradovic

Gulyaeva

Kushlinskii

Djordjevic

EGFR polymorphism and survival of NSCLC patients treated with TKIs: A systematic review and meta-analysis

J Oncol202019732412020

10.1155/2020/1973241

32256580

Obradovic

Todosijevic

Jurisic

Side effects of tyrosine kinase inhibitors therapy in patients with non-small cell lung cancer and associations with EGFR polymorphisms: A systematic review and meta-analysis

Oncol Lett25622023

10.3892/ol.2022.13649

36644136

Hall

SRR

Pengs

Yang

Yao

Battles against aberrant KEAP1-NRF2 signaling in lung cancer: Intertwined metabolic and immune networks

Theranostics137047232023

10.7150/thno.80184

36632216

Ghareghomi

Moosavi-Movahedi

Saso

Habibi-Rezaei

Khatibi

Hong

Moosavi-Movahedi

Modulation of Nrf2/HO-1 by natural compounds in lung cancer

Antioxidants (Basel)127352023

10.3390/antiox12030735

36978983

Imielinski

Berger

Hammerman

Hernandez

Pugh

Hodis

Cho

Suh

Capelletti

Sivachenko

Mapping the hallmarks of lung adenocarcinoma with massively parallel sequencing

Cell150110711202012

10.1016/j.cell.2012.08.029

22980975

Romero

Sanchez-Rivera

Westcott

PMK

Mercer

Bhutkar

Muir

Robles

Rodriguez

Liao

Keap1 mutation renders lung adenocarcinomas dependent on Slc33a1

Nat Cancer15896022020

10.1038/s43018-020-00118-z

34414377

Romero

Sayin

Davidson

Bauer

Singh

LeBoeuf

Karakousi

Ellis

Bhutkar

Sanchez-Rivera

Keap1 loss promotes Kras-driven lung cancer and results in dependence on glutaminolysis

Nat Med23136213682017

10.1038/nm.4407

28967920

Zavitsanou

Pillai

Hao

Bartnicki

Karakousi

Rajalingam

Herrera

Karatza

Rashidfarrokhi

KEAP1 mutation in lung adenocarcinoma promotes immune evasion and immunotherapy resistance

Cell Rep421132952023

10.1016/j.celrep.2023.113295

37889752

Ghareghomi

Habibi-Rezaei

Arese

Saso

Moosavi-Movahedi

Nrf2 modulation in breast cancer

Biomedicines1026682022

10.3390/biomedicines10102668

36289931

Menegon

Columbano

Giordano

The dual roles of NRF2 in cancer

Trends Mol Med225785932016

10.1016/j.molmed.2016.05.002

27263465

Kerins

Ooi

A catalogue of somatic NRF2 gain-of-function mutations in cancer

Sci Rep8128462018

10.1038/s41598-018-31281-0

30150714

Liang

Yan

Wang

NRF2 mutation enhances the immune escape of hepatocellular carcinoma by reducing STING activation

Biochem Biophys Res Commun6981495362024

10.1016/j.bbrc.2024.149536

38271834

Tang

Liu

Yan

Jin

Huang

Liu

Inhibition of LNC EBLN3P enhances radiation-induced mitochondrial damage in lung cancer cells by targeting the Keap1/Nrf2/HO-1 axis

Biology (Basel)1212082023

37759607

Spampinato

Sferrazzo

Pittala

Di Rosa

Vanella

Salerno

Sorrenti

Carota

Parrinello

Raffaele

Non-competitive heme oxygenase-1 activity inhibitor reduces non-small cell lung cancer glutathione content and regulates cell proliferation

Mol Biol Rep47194919642020

10.1007/s11033-020-05292-y

32056044

Director's Challenge Consortium for the Molecular Classification of Lung

Adenocarcinoma

Shedden

Taylor

Enkemann

Tsao

Yeatman

Gerald

Eschrich

Jurisica

Giordano

Gene expression-based survival prediction in lung adenocarcinoma: A multi-site, blinded validation study

Nat Med148228272008

10.1038/nm.1790

Lanczky

Gyorffy

Web-based survival analysis tool tailored for medical research (KMplot): Development and implementation

J Med Internet Res23e276332021

10.2196/27633

34309564

Ritchie

Phipson

Law

Shi

Smyth

limma powers differential expression analyses for RNA-sequencing and microarray studies

Nucleic Acids Res43e472015

10.1093/nar/gkv007

25605792

Liu

Weng

Guo

Dang

Wang

Zhang

Sun

Han

Machine learning-based integration develops an immune-derived lncRNA signature for improving outcomes in colorectal cancer

Nat Commun138162022

10.1038/s41586-022-04952-2

35145098

Miao

Zhang

Lei

Luo

Xie

Wang

Guo

ImmuCellAI: A unique method for comprehensive T-cell subsets abundance prediction and its application in cancer immunotherapy

Adv Sci (Weinh)719028802020

10.1002/advs.201902880

32274301

Maeser

Gruener

Huang

oncoPredict: An R package for predicting in vivo or cancer patient drug response and biomarkers from cell line screening data

Brief Bioinform22bbab2602021

10.1093/bib/bbab260

34260682

Detterbeck

Woodard

Bader

Dacic

Grant

Park

Tanoue

The proposed ninth edition TNM classification of lung cancer

Chest1668828952024

10.1016/j.chest.2024.05.026

38885896

Lou

Zhao

Huang

Chen

Tai

Tsim

KWK

Chen

Xie

Ginkgetin derived from Ginkgo biloba leaves enhances the therapeutic effect of cisplatin via ferroptosis-mediated disruption of the Nrf2/HO-1 axis in EGFR wild-type non-small-cell lung cancer

Phytomedicine801533702021

10.1016/j.phymed.2020.153370

33113504

Liang

Zhao

Zhang

Jin

Zheng

Bian

Chen

Huang

miR-6077 promotes cisplatin/pemetrexed resistance in lung adenocarcinoma via CDKN1A/cell cycle arrest and KEAP1/ferroptosis pathways

Mol Ther Nucleic Acids283663862022

10.1016/j.omtn.2022.03.020

35505963

Mei

Long

Mei

Qiu

APOC1 reduced anti-PD-1 immunotherapy of nonsmall cell lung cancer via the transformation of M2 into M1 macrophages by ferroptosis by NRF2/HO-1

Anticancer Drugs353333432024

10.1097/CAD.0000000000001573

38241194

Wei

Zhang

Fan

Chen

Wang

Yang

Deng

Yang

Redox(high) phenotype mediated by KEAP1/STK11/SMARCA4/NRF2 mutations diminishes tissue-resident memory CD8+ T cells and attenuates the efficacy of immunotherapy in lung adenocarcinoma

Oncoimmunology1323401542024

10.1080/2162402X.2024.2340154

38601319

Fahrmann

Tanaka

Irajizad

Mao

Dennison

Murage

Casabar

Mayo

Peng

Celiktas

Mutational activation of the NRF2 pathway upregulates kynureninase resulting in tumor immunosuppression and poor outcome in lung adenocarcinoma

Cancers (Basel)1425432022

10.3390/cancers14102543

35626147

Phillips

Structure and mechanism of kynureninase

Arch Biochem Biophys54469742014

10.1016/j.abb.2013.10.020

24200862

Karayama

Masuda

Mori

Yasui

Hozumi

Suzuki

Furuhashi

Fujisawa

Enomoto

Nakamura

Comprehensive assessment of multiple tryptophan metabolites as potential biomarkers for immune checkpoint inhibitors in patients with non-small cell lung cancer

Clin Transl Oncol234184232021

10.1007/s12094-020-02421-8

32533317

Leon-Letelier

Sater

Chen

Srejbhjv

Kynureninase upregulation is a prominent feature of NFR2-activated cancers and is associated with tumor immunosuppression and poor prognosis

Cancers (Basel)158342023

10.3390/cancers15030834

36765792

Liu

Xie

Zhang

Zeng

Tao

Yang

Chen

Zeng

SERPINB5 promotes colorectal cancer invasion and migration by promoting EMT and angiogenesis via the TNF-α/NF-κB pathway

Int Immunopharmacol1311117592024

10.1016/j.intimp.2024.111759

38460302

Rathod

Franz

Beyersdorfer

Wanuske

Leal-Fischer

Hanns

Stüdle

Zimmermann

Buczak

Schinner

Spindler

DPM1 modulates desmosomal adhesion and epidermal differentiation through SERPINB5

J Cell Biol223e2023050062024

10.1083/jcb.202305006

38477878

Huang

Wang

Zhang

Guo

Zhang

Wen

Zhang

SERPINB5 is a prognostic biomarker and promotes proliferation, metastasis and epithelial-mesenchymal transition (EMT) in lung adenocarcinoma

Thorac Cancer14227522872023

10.1111/1759-7714.15013

37424293

Scopetti

Piobbico

Brunacci

Pieroni

Bellezza

Castelli

Ludovini

Tofanetti

Cagini

Sidoni

INSL4 as prognostic marker for proliferation and invasiveness in non-small-cell lung cancer

J Cancer12378137952021

10.7150/jca.51332

34093787

Yang

Chen

Zhou

Kaye

Role of INSL4 signaling in sustaining the growth and viability of LKB1-inactivated lung cancer

J Natl Cancer Inst1116646742019

10.1093/jnci/djy166

30423141

Bhattacharya

Gawali

Kallay

Toukam

Koehler

Stambrook

Krummel

Sengupta

Therapeutically leveraging GABA(A) receptors in cancer

Exp Biol Med (Maywood)246212821352021

10.1177/15353702211032549

34649481

Liu

Yang

Shen

Huang

Lin

Tang

Liang

Shao

Zhang

GABRA3 promotes lymphatic metastasis in lung adenocarcinoma by mediating upregulation of matrix metalloproteinases

Oncotarget732341323502016

10.18632/oncotarget.8700

27081042

Hanahan

Coussens

Accessories to the crime: Functions of cells recruited to the tumor microenvironment

Cancer Cell213093222012

10.1016/j.ccr.2012.02.022

22439926

Binnewies

Roberts

Kersten

Chan

Fearon

Merad

Coussens

Gabrilovich

Ostrand-Rosenberg

Hedrick

Understanding the tumor immune microenvironment (TIME) for effective therapy

Nat Med245415502018

10.1038/s41591-018-0014-x

29686425

Genova

Dellepiane

Carrega

Sommariva

Ferlazzo

Pronzato

Gangemi

Filaci

Coco

Croce

Therapeutic implications of tumor microenvironment in lung cancer: Focus on immune checkpoint blockade

Front Immunol127994552021

10.3389/fimmu.2021.799455

35069581

Song

Guo

Ying

Gao

Identification and validation of a novel signature based on NK cell marker genes to predict prognosis and immunotherapy response in lung adenocarcinoma by integrated analysis of single-cell and bulk RNA-sequencing

Front Immunol138507452022

10.3389/fimmu.2022.850745

35757748

Terabe

Berzofsky

Tissue-specific roles of NKT cells in tumor immunity

Front Immunol918382018

10.3389/fimmu.2018.01838

30158927

Rad

Monkman

Warkiani

Ladwa

O'Byrne

Rezaei

Kulasinghe

Understanding the tumor microenvironment for effective immunotherapy

Med Res Rev41147414982021

10.1002/med.21765

33277742

Peppelenbosch

Sprengers

Kwekkeboom

TIGIT, the next step towards successful combination immune checkpoint therapy in cancer

Front Immunol126998952021

10.3389/fimmu.2021.699895

34367161

Acharya

Sabatos-Peyton

Anderson

Tim-3 finds its place in the cancer immunotherapy landscape

J Immunother Cancer8e0009112020

10.1136/jitc-2020-000911

32601081

Wojciechowicz

Spodzieja

Wardowska

The BTLA-HVEM complex-The future of cancer immunotherapy

Eur J Med Chem2681162312024

10.1016/j.ejmech.2024.116231

38387336

Gebbia

Lorusso

Galetta

Caruso

Palomba

Riccardi

Borsellino

Carrozza

Leo

Ferraù

First-line cisplatin with docetaxel or vinorelbine in patients with advanced non-small-cell lung cancer: A quality of life directed phase II randomized trial of Gruppo Oncologico Italia Meridionale

Lung Cancer692182242010

10.1016/j.lungcan.2009.10.008

19910076

Fuentes-Baile

Ventero

Encinar

García-Morales

Poveda-Deltell

Pérez-Valenciano

Barberá

Gallego-Plazas

Rodríguez-Lescure

Martín-Nieto

Saceda

Differential effects of IGF-1r small molecule tyrosine kinase inhibitors BMS-754807 and OSI-906 on human cancer cell lines

Cancers (Basel)1237172020

10.3390/cancers12123717

33322337

Zhang

Zhao

Wang

Gong

Zhao

Zhang

Niu

Zhao

Dasatinib in combination with BMS-754807 induce synergistic cytotoxicity in lung cancer cells through inhibiting lung cancer cell growth, and inducing autophagy as well as cell cycle arrest at the G1 phase

Invest New Drugs414384522023

10.1007/s10637-023-01360-9

37097369

DaCosta Byfield

Major

Laping

Roberts

SB-505124 is a selective inhibitor of transforming growth factor-beta type I receptors ALK4, ALK5, and ALK7

Mol Pharmacol657447522004

10.1124/mol.65.3.744

14978253

Chou

Chiu

Tang

Yeh

Validation of the effects of TGF-β1 on tumor recurrence and prognosis through tumor retrieval and cell mechanical properties

Cancer Cell Int14202014

10.1186/1475-2867-14-20

24581230

Stathis

Bertoni

BET proteins as targets for anticancer treatment

Cancer Discov824362018

10.1158/2159-8290.CD-17-0605

29263030

Wang

Rao

Zhang

Tang

Zhang

Jie

Zhu

Wang

BRD4-IRF1 axis regulates chemoradiotherapy-induced PD-L1 expression and immune evasion in non-small cell lung cancer

Clin Transl Med12e7182022

10.1002/ctm2.718

35083874

Mrdjanovic

Solajic

Srdenovic-Conic

Bogdanović

Dea

Kladar

Jurišić

The oxidative stress parameters as useful tools in evaluating the DNA damage and changes in the complete blood count in hospital workers exposed to low doses of antineoplastic drugs and ionizing radiation

Int J Environ Res Public Health1884452021

10.3390/ijerph18168445

34444191

Figure 1.

Immunohistochemical detection of KEAP1, NRF2, HO-1 and Ki-67 expression in lung adenocarcinoma tissues. Negative expression of (A) KEAP1, (B) NRF2, (C) HO-1 and (D) Ki-67. Positive expression of (E) KEAP1, (F) NRF2, (G) HO-1 and (H) Ki-67 (magnification, ×100). KEAP1, Kelch like ECH associated protein 1; NRF2, nuclear factor erythroid 2-related factor 2; HO-1, heme oxygenase 1.

Figure 2.

Association between KEAP1, NRF2 and HO-1 expression and prognosis of patients with lung adenocarcinoma. Kaplan-Meier survival curves of (A) overall survival and (B) first progression survival for KEAP1; (C) overall survival and (D) first progression survival for NRF2; and (E) overall survival and (F) first progression survival for HO-1 expression. KEAP1, Kelch like ECH associated protein 1; NRF2, nuclear factor erythroid 2-related factor 2; HO-1, heme oxygenase 1; HR, hazard ratio.

Figure 3.

Overview of genetic changes in KEAP1, NRF2 and HO-1 mutations in patients with lung adenocarcinoma in The Cancer Genome Atlas database. (A) Types and frequencies of KEAP1, NRF2 and HO-1 genetic alterations. Lollipop charts of the genetic localization of (B) KEAP1, (C) NRF2 and (D) HO-1 mutations. KEAP1, Kelch like ECH associated protein 1; NRF2, nuclear factor erythroid 2-related factor 2; HO-1/HMOX1, heme oxygenase 1; GISTIC, genomic identification of significant targets in cancer.

Figure 4.

Prognostic-related genes upregulated by mutations in lung adenocarcinoma. Volcano plot of differentially expressed genes in the (A) t mutant vs. normal groups and (B) mutant vs. wild-type groups. (C) Common differential genes of mutant group vs. normal group and mutant group vs. wild-type group. (D) Common prognosis-related genes between The Cancer Genome Atlas and GSE68465 datasets.

Figure 5.

Construction of a prognostic signature based on KEAP1, NRF2 and HO-1 mutation-mediated upregulated genes. (A) A total of 101 types of prediction models were constructed via a leave-one-out cross-validation framework and the C-index of each model was further calculated. Kaplan-Meier survival curves of OS for high- and low-risk groups of patients in the (B) TCGA and (D) GSE68465 databases. Risk curves of the prognostic risk model in the (C) TCGA and (E) GSE68465 databases. TCGA, The Cancer Genome Atlas; OS, overall survival; KYNU, kynureninase; SERPINB5, serpin family B member 5; INSL4, insulin like 4; GABRA3, γ-aminobutyric acid type A receptor subunit α3; HR, hazard ratio; CI, confidence interval.

Figure 6.

Clinical pathological features and prognosis analysis. Discrepancies in risk score in (A) T, (B) N and (C) TNM stages. Independent prognostic predictors obtained by (D) univariate and (E) multivariate Cox analyses. (G) Nomogram of risk score, N stage and M stage. (F) Calibration curve of the nomogram. *P<0.05, ****P<0.0001. T, tumor; N, lymph node; TNM, tumor-lymph node-metastasis; HR, hazard ratio; CI, confidence interval; OS, overall survival.

Figure 7.

Immune infiltration and immune association analyses. Differences in (A) immune, stromal and ESTIMATE scores, (B) immune-checkpoint blockade therapy response prediction and (C) immune cells between the high- and low-risk groups. Differences in the expression of immune-checkpoint (D) immune stimulated and (E) immune inhibited genes between the high- and low-risk groups. *P<0.05; **P<0.01; ***P<0.001; ***P<0.0001. -, not significant; FPKM, fragments per kilobase of transcript per million mapped reads.

Figure 8.

Differences in drug sensitivity between high- and low-risk groups. IC₅₀ values for (A) vinblastine, (B) docetaxel and (C) cisplatin, which were more sensitive in the high-risk group. IC₅₀ values for (D) BMS-754804, (E) SB05124 and (F) JQ1, which were more sensitive in the low-risk group. **P<0.01; ***P<0.001; ****P<0.0001.

Figure 9.

Expression of hub genes in different risk groups and their localization in lung adenocarcinoma. Expression of KYNU, SERPINB5, INSL4 and GABRA3 in (A) high- and low-risk groups, and (B) different cell types in the GSE148071 dataset. ***P<0.001, ****P<0.0001. KYNU, kynureninase; SERPINB5, serpin family B member 5; INSL4, insulin like 4; GABRA3, γ-aminobutyric acid type A receptor subunit α3; FPKM, fragments per kilobase of transcript per million mapped reads.

Table I.

Association between KEAP1/NRF2/HO-1 expression and clinicopathological features in lung adenocarcinoma.

A, KEAP1 expression

Clinicopathological feature	Positive (n=72)	Negative (n=32)	P-value
Age, years	66.04±11.58	67.78±11.15	0.47
Sex			0.80
Female	44	18
Male	28	14
T stage			0.06
T1-2	62	32
T3-4	10	0
N stage			0.50
N0	50	25
N1-3	22	7
M stage			0.09
M0	63	32
M1	9	0
TNM stage			0.12
I–II	54	29
III–IV	18	3
Tumor diameter, cm	2.01±1.44	2.10±1.03	0.72
Ki-67 expression ratio, %	41±15	39±13	0.37

B, NRF2 expression

Clinicopathological feature	Positive (n=58)	Negative (n=46)	P-value

Age, years	68.38±11.57	64.30±10.93	0.07
Sex			0.44
Female	37	25
Male	21	21
T stage			<0.01
T1-2	48	46
T3-4	10	0
N stage			<0.01
N0	34	41
N1-3	24	5
M stage			0.01
M0	49	46
M1	9	0
TNM stage			<0.001
I–II	39	44
III–IV	19	2
Tumor diameter, cm	2.52±1.50	1.43±0.68	<0.001
Ki-67 expression ratio, %	46±15	34±10	<0.001

C, HO-1 expression

Clinicopathological feature	Positive (n=63)	Negative (n=41)	P-value

Age, years	68.00±11.24	64.39±11.49	0.12
Sex			0.21
Female	34	28
Male	29	13
T stage			0.02
T1-2	53	41
T3-4	10	0
N stage			<0.001
N0	36	39
N1-3	27	2
M stage			0.03
M0	54	41
M1	9	0
TNM stage			<0.001
I–II	42	41
III–IV	21	0
Tumor diameter, cm	2.43±1.50	1.42±0.64	<0.001
Ki-67 expression ratio, %	45±15	33±9	<0.001

Data are presented as n or the mean ± standard deviation. KEAP1, Kelch like ECH associated protein 1; NRF2, nuclear factor erythroid 2-related factor 2; HO-1, heme oxygenase 1; T, tumor; N, lymph node; M, metastasis.