Predictive Models for Colon Adenocarcinoma Diagnosis,Prognosis,and Immune Microenvironment Based on 2 Hypoxia-Related Genes: KDM3A and ENO3

Abstract

Background: Hypoxia is known to play a critical role in tumor occurrence, progression, prognosis, and therapy resistance. However, few studies have investigated hypoxia markers for diagnosing and predicting prognosis in colon adenocarcinoma (COAD). This study aims to identify a hypoxia genes-based biomarker for predicting COAD patients’ prognosis and response to immunotherapy on an individual basis. Methods: Hypoxia-related genes were extracted from the Molecular Signatures Database. Gene expression, clinical data, and mutation data of COAD were collected retrospectively from the Cancer Genome Atlas, the Gene Expression Omnibus, and the International Cancer Genome Consortium databases. Univariate and multivariate cox regression, and the least absolute shrinkage and selection operator method were used to select the genes most associated with the prognosis of COAD patients. Kaplan–Meier survival analysis, receiver operating characteristic curves, calibration curves, and decision curve analyses were performed to validate the efficacy of the signature in predicting the prognosis of COAD patients. EdU incorporation assays, cell survival assays, western blot assays, and trans-well invasion assays were performed to further confirm the function of the screened genes in tumorigenesis. Results: ENO3 and KDM3A were identified as key genes for constructing prognostic and diagnostic signatures, which were found to be independent risk factors for predicting the prognosis and diagnosis of COAD patients. Using these signatures, COAD patients could be stratified into high-risk and low-risk groups, with the latter exhibiting better overall survival outcomes. Moreover, the high-risk group displayed elevated levels of immune checkpoint genes and tumor mutation burden, indicating that these patients may benefit from immune checkpoint inhibitor therapy. Conclusion: The signature developed in this study demonstrates excellent efficacy in prognosticating the outcomes of COAD patients. Moreover, it can serve as a valuable tool for clinicians to identify COAD patients who are suitable for ICI therapy.

Keywords

hypoxia ENO3 KDM3A immune checkpoint inhibitor colon adenocarcinoma

Introduction

Colon cancer is one of the most common malignancies of the digestive system, with the third incidence and the second cause of cancer-related mortality worldwide.^1,2 Colon adenocarcinoma (COAD) is the most common type of colon cancer, accounting for the majority of cases. Despite recent advancements in early diagnosis and treatment strategies, the 5-year survival rate of COAD patients remains unsatisfactory.³ Current treatment plans and clinical outcomes are primarily evaluated based on the conventional clinicopathologic risks and prognostic factors, such as TNM stage, histological differentiation, vascular and nerve invasion.^4–6 However, COAD is a highly heterogeneous disease, and the prognosis outcome varies greatly even for patients with similar clinical features and treatment regimens.⁷ Some patients experience longer disease-free survival after receiving recommended therapy based on conventional clinicopathologic factors, while others suffer from recurrence.⁸ Thus, identifying reliable biomarkers for predicting the prognosis and response to specific therapy is crucial for personalized modification in clinical management.

Hypoxia, a condition characterized by low oxygen tension in tissues, is a critical feature of tumor pathophysiology.⁹ This microenvironmental condition is prevalent in most solid tumors due to the abnormally active metabolism of tumor cells. Under hypoxia, hypoxia-inducible factors are activated in the tumor microenvironment, promoting the transcription of genes involved in angiogenesis, tumor growth, metastasis, metabolic reprograming, radio-resistance, and chemoresistance.^10–12 Furthermore, hypoxia can induce immunosuppression by reducing the activity of various immune cells in the tumor microenvironment^13,14 or by promoting the release of suppressors and the expression of immune checkpoint inhibitors (ICIs) through the production of immune stimulators.^15,16 This makes hypoxia to be a potential target for improving immunotherapy.¹⁷ Therefore, further investigation into the relationship between hypoxia and tumor microenvironment is of great importance in enhancing the therapeutic efficacy of COAD patients.

In recent years, anticancer immunotherapies involving the use of ICIs or adoptive cellular transfer have emerged as new therapeutic pillars within oncology.¹⁸ However, despite patients having the same disease stage, different prognosis was observed due to the molecular heterogeneity of COAD.¹⁹ Tumor mutation burden (TMB), an emerging biomarker for reflecting response to ICIs in multiple tumor types, has been found to be positively correlated with the number of neoantigens in cancers.^20,21 Patients with high TMB, indicating more neoantigens, are more likely to benefit from immunotherapy,^22,23 suggesting that TMB is a biomarker for predicting the individual responsiveness to immunotherapy.

Hypoxia plays a crucial role in regulating tumor progression and treatment resistance, and ultimately affects the prognosis of tumors. However, the precise association between hypoxia-related genes and the prognosis of COAD patients, as well as their responsiveness to immunotherapy, remains unclear. Therefore, our objective is to develop a novel hypoxia gene-based molecular biomarker that can accurately predict the prognosis of COAD patients and their individual responsiveness to immunotherapy.

In this study, 2 hypoxia-related genes ENO3 and KDM3A were screened out and further used to construct the predictive model. ENO3, also known as beta-enolase (ENO-β), is a metalloenzyme that functions during glycolysis and has been found to be ectopic expression in different cancers.²⁴ Lysine demethylase 3A (KDM3A) is a member of the Jumonji C domain-containing histone demethylase family, which plays a critical role in promoting gene expression by facilitating the demethylation of histones at lysine-9 of di-methylated histone H3 (H3K9me2) or mono-methylated (H3K9me1).²⁵

In the present study, we constructed predictive models for the prognosis and diagnosis of COAD patients. Then we explored the relationship between hypoxia and immune microenvironment in COAD patients, and identify candidates for immunotherapy. Finally, we verified our conclusions through several online databases and in vitro experiments. Taken together, our findings can assist in developing individualized therapeutic strategies and improve the outcomes of COAD patients.

Methods

Acquisition of Gene Expression Profiles and Clinical Data

The mRNA sequence of COAD patients with available clinical data was collected retrospectively from the the Cancer Genome Atlas (TCGA) (http://www.tcga.org/) database, Gene Expression Omnibus (GEO) (https://www.ncbi.nlm.nih.gov/geo/) database and International Cancer Genome Consortium (ICGC) (https://dcc.icgc.org/) database. In total, 473 COAD tissue samples and 41 normal tissue samples, 443 colon cancer samples and 19 normal tissues, and 441 COAD tissue samples and 41 normal tissues in TCGA-COAD cohort, GSE39582 cohort, and ICGC cohort were enrolled in this study, respectively. The TCGA dataset was used as a training cohort, and the GSE39582 and ICGC datasets were applied as independent validation cohorts. The mutation data were downloaded from the TCGA datasets.

Hypoxia-Related Genes

A total of 197 hypoxia-related genes were extracted from the Molecular Signatures Database (https://www.gsea-msigdb.org/gsea/msigdb/index.jsp, MSigDB). The 197 hypoxia-related genes were matched with COAD-associated mRNA obtained from the RNA-sequencing platform in TCGA database, GEO database, and ICGC database.

Establishment of the Predictive Prognostic Gene Signature and Risk Stratification

Univariate Cox proportional hazard regression analysis was performed for each gene regarding survival status to identify the prognostic hypoxia-related genes, and genes with a P value < .05 were considered statistically significant. Subsequently, the genes that exhibited statistically significant results in the univariate cox proportional hazard regression analysis were included in the multivariate Cox regression analysis for further investigation. Least absolute shrinkage and selection operator (LASSO) regression was then performed to identify the genes that were most strongly associated with the prognosis of COAD patients. Finally, the genes with nonzero penalty coefficients in the LASSO regression were included in a step-wise multivariate cox regression to select highly correlated genes and construct a prognostic signature. For each patient, the prognostic signature was calculated by multiplying the expression value of a gene with its corresponding Cox proportion regression coefficient (prognostic signature = ∑ Cox regression of gene_i * expression level of gene_i). Using the X-tile software, the COAD patients were divided into high-risk and low-risk groups based on the prognostic signature. The predictive ability of the prognostic signature was evaluated using Kaplan–Meier survival analysis.

Establishment and Evaluation of a Predictive Nomogram

Univariate and multivariate Cox regression analyses were conducted to determine whether the prognostic gene signature was independent of other clinical characteristics, including age, gender, and TNM stage. Hazard ratio (HR) and 95% confidence intervals were calculated, and factors with P value < .05 were considered statistically significant. Subsequently, all the independent predictors were used to construct a predictive nomogram, which was then used to predict the 1-, 3-, and 5-year survival probability of COAD patients. Then the nomogram was used to predict the 1-,3-, and 5-year survival probability of the COAD patients. The predictive performance of the nomogram was validated using calibration curves and time-independent receiver operating characteristic (ROC) curves. Decision curve analysis (DCA) was performed to assess the clinical usefulness of the nomogram.

Internal and External Validations of the Expression Patterns and Prognostic Performance of the Hypoxia-Related Genes

Wilcoxon signed rank tests were utilized to validate the expression patterns of hypoxia-related genes in COAD and normal tissues from the ICGC and GSE81558 datasets. A P value < .05 was deemed statistically significant. Furthermore, regression analysis was conducted to investigate the potential of the prognostic genes in discriminating COAD from normal samples.

IHC Staining

Formalin-fixed paraffin-embedded tumor tissues and adjacent nontumor tissues of the same patient were obtained from the department of pathology in our institution. The tissues were preserved for 1 week in 10% formalin before being embedded in paraffin. Then 4-µm pieces of tissue specimens were prepared. The slides were deparaffinized in xylene, and then rehydrated in alcohol. The tissue slices were put in the repair box of TRIS-EDTA repair solution (PH8.0), and carry out antigen repair in the microwave oven. Then we used 3% of hydrogen peroxide (H₂O₂) to block endogenous peroxidase activity. Slides were blocked with 3% bovine serum albumin for 30 min and incubated in the anti-KDM3A antibody (diluted 1:300; ab24364; Abcam) and anti-ENO3 antibody (diluted 1:500; 55234-1-ap; Wuhan Sanying Biotechnology Co., Ltd) overnight at 4 °C. The following day, following 3 washes with PBS, the slides were incubated with a secondary antibody for 50 min. Visualization was achieved using the 3,3-diaminobenzidine kit, with hematoxylin utilized to stain nuclei. Postexperiment, the slides were observed under a microscope.

Cell Culture and Transfection

Human COAD cell lines DLD1 (accession number CCL-221) and HCT116 (accession number CCL-247EMT) were purchased from the American Type Culture Collection. The cells were cultured in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum (FBS), 100 μg/mL streptomycin (Sigma) and 100 μg/mL penicillin and maintained in a humidified cell incubator with an atmosphere of 5% CO₂ at 37 °C. To knockdown mRNA expression, 10 μM siRNA of ENO3 and KDM3A (Ribobio) was added to DLD1 and HCT116 cells for 48 h. Knockdown efficiency was tested using the western blot.

Western Blot Assay

Western blot assays were performed using the methodology as previously described.²⁶ Cells were lysed with RIPA lysate (Beyotime Biotechnology), and cell lysates were prepared. Proteins were separated by 10% SDS-PAGE. Proteins were separated by 10% SDS-PAGE and transferred to PVDF membranes. Antibodies used in this study were as follows: KDM3A (diluted 1:1000; Cat# ab191389; Abcam), ENO3 (diluted 1:1000; Cat# ab119327; Abcam), GAPDH (diluted 1:5000; Cat No.60004-1; Proteintech), anti-Rabbit-IgG-HRP (diluted 1:10000; Cat No.PR30012; Proteintech), and anti-Mouse-IgG-HRP (diluted 1:10000; Cat No.PR30012; Proteintech). ECL reagent was applied to detect the signals.

Cell Survival Assays

To analyze the survival rate of COAD cells, a cell counting kit-8 (CCK-8, Beyotime Biotechnology) was used. Briefly, 5 × 10³ COAD cells infected with or without si-ENO3 and si-KDM3A were seeded into 96-well plates for 48 h. Then, 20 μL CCK-8 solution was added to each well and incubated for 1 h. The fluorescence was detected using a microplate reader (BioTek) at 450 nm and the survival rates were calculated.

EdU Straining Assay

5 × 10⁵ cells infected with or without si-ENO3 and si-KDM3A were seeded into 6-well plates for 48 h. EdU regent was added to DLD1 and HCT116 cells for 4 h and then fixed in 4% paraformaldehyde for 15 min. The cells were permeabilized with 0.3% Triton X-100 for 10 min and strained with EdU dye solution for 20 min. The nucleus was stained with 2-(4-amidinophenyl)-6-indolecarbamidine dihydrochloride (DAPI). The fluorescence was observed using a microscope, and the number of EdU-positive cell was calculated.

Trans-Well Assay

Trans-well chamber filters were coated with Matrigel. After infection with lentivirus, DLD1 cells or HCT116 cells were suspended in serum-free DMEM media and then 2 × 10⁴ cells were seeded into the upper chamber in a volume of 500 μL. The chamber was then cultured in a well containing 500 μL of DMEM media with 10% FBS at 37 °C for 24 h. Cells on the upper side of the membrane were removed using cotton swabs, and those on the other side were stained and counted. Four high-powered fields were counted for each membrane.

Statistical Analysis

Statistical analysis was performed using SPSS and R software (version 4.5.1). Patients with incomplete clinical data were excluded from the study cohort. Differences between 2 independent groups were measured using Student's t test or Wilcoxon test. A P-value of less than .05 was considered statistically significant.

Results

Identification of Hypoxia-Related Genes Related to Prognosis of COAD Patients and Construction of the Prognostic Signature

A total of 197 hypoxia-related genes extracted from the Molecular Signatures Database and matched with the mRNA-sequencing data from the TCGA, GEO, and ICGC databases. Univariate Cox proportional hazard regression analysis was performed on the 197 hypoxia-related genes using 403 COAD samples with survival status and overall survival (OS) time, resulting in the identification of 23 genes with P < .05 (Figure 1A). Multivariate Cox regression was then performed to select highly correlated genes with the prognosis of COAD patients, resulting in the identification of 9 statistically significant genes (Figure 1B). Further implementation of LASSO regression (Figure 1C and D) and step-wise multivariate Cox regression analysis led to the identification of 2 genes that were statistically significant and used to construct the prognostic signature (Figure 1E). The prognostic signature is described as follows: PS = 1.32* ENO3 + 1.31* KDM3A, with both ENO3 and KDM3A associated with poor prognosis.

Figure 1.

Identification of key hypoxia-related genes for COAD patients. (A) Forest plot for univariate cox regression. (B) Forest plot for multivariate cox regression. (C and D) LASSO cox regression in selecting the key hypoxia-related genes. (E) Forest plot for step-wise multivariate cox regression in selecting the independent prognostic genes. Abbreviations: COAD, colon adenocarcinoma; LASSO, least absolute shrinkage and selection operator.

Performance Evaluation of the Prognostic Signature

The prognostic signature was used to calculate the risk score for each sample in the TCGA cohort. The patients were then divided into high- and low-risk groups based on their risk score using the X-tile software. Kaplan–Meier analyses were performed to compare the OS of the 2 groups, and the log-rank test was used for statistical analysis. The survival curves revealed that patients in the low-risk group had better OS (P < .001) (Figure 2A). Figure 2B depicted the expression levels of the 2 hypoxia-related prognostic genes and the corresponding risk scores between the high- and low-risk patients. To further validate the efficacy of the hypoxia-genes-based prognostic signature, we plotted time-dependent ROC curves. The area under the curve (AUC) for 1-, 3-, and 5-years in the TCGA dataset were 0.701, 0.740, and 0.770, respectively (Figure 2C), indicating that the predictive model performed well in predicting the prognosis of the COAD patients.

Figure 2.

Kaplan–Meier survival analysis, risk score distribution and time-dependent ROC curves of a prognostic model in the COAD cohort from TCGA and GSE29621. (A and D), Kaplan–Meier survival curves indicated that the OS in the high-risk group was markedly poorer than that in the low-risk group (P < .001) in both the training cohort (TCGA) and validation cohort (GSE29621). (B and E) Distribution of risk scores under different gene expression characteristics of COAD patients in TCGA and GSE29621 cohorts, respectively. (C and F) Time-dependent ROC curve analysis for measuring the predictive performance on OS of COAD patients in TCGA and GSE29621 cohorts, respectively. Abbreviations: COAD, colon adenocarcinoma; ROC, receiver operating characteristic; TCGA, the Cancer Genome Atlas; OS, overall survival.

To further verify the generalizability and the robustness of the model, we performed the same analysis on an external data set GSE29621. Consistent with the results in the training dataset, the OS of patients in the low-risk group was significantly better than that in the high-risk group (P < .001) (Figure 2D). The prognostic gene expression and corresponding riskScore of the validation cohort are shown in Figure 2E. The AUC for 1-, 3-, and 5-years were 0.812, 0.649, and 0.598, respectively (Figure 2F), indicating that the model had good generalizability and robustness.

Construction and Validation of a Predictive Nomogram for COAD Patients

Univariate and multivariate Cox regression analyses were performed to determine whether the prognostic signature (riskScore) was an independent predictive factor for OS in COAD patients, independent of other clinical factors such as age, gender, and TNM stage. The results revealed that the riskScore (P < .001, HR = 1.313), TNM stage (P = .046, HR = 2.058) and age (P = .003, HR = 1.051) were independent risk factors for OS in COAD patients (Figure 3A). These 3 independent risk factors were then integrated to construct a predictive nomogram that could accurately predict the clinical outcomes of COAD patients (Figure 3B). The calibration curves for predicting the OS at 1-,3-, and 5-years indicated that the nomogram-predicted OS outcomes were consistent with the actual outcomes (Figure 3C toE). Additionally, the ROC curve analysis confirmed the superior predictive performance of the nomogram, with AUCs of 0.685, 0.739, and 0.776 for 1, 3, and 5 years, respectively, compared to any single independent risk factor (Figure 3F to H). Furthermore, the DCA demonstrated that the nomogram achieved the optimal net benefit compared to any single independent risk factor (Figure 3I to K).

Figure 3.

Construction and validation of a predictive nomogram. (A) Univariate and multivariate Cox regression confirmed that the prognostic signature, age, and TNM stage were independent prognostic predictors for COAD patients. (B) Nomogram for predicting the OS of COAD patients at 1, 3, and 5 years. (C-E) Calibration curves of the nomogram for OS prediction at 1, 3, and 5 years. (F-H) ROC curves to evaluate the predictive ability of the nomogram. (I-K) DCA curves determined that the nomogram can provide optimal clinical decision-making benefits. Abbreviations: COAD, colon adenocarcinoma; ROC, receiver operating characteristic; DCA, decision curve analysis; OS, overall survival.

Establishment of a Diagnostic Model Based on the 2 Hypoxia-Related Genes in COAD

A diagnostic model based on 2 hypoxia-related genes was established for COAD using logistic regression. The model demonstrated a high specificity of 100.0% and a sensitivity of 71.4% in training cohort (GSE81558 dataset) (Figure 4A), with an AUC of 0.918 (Figure 4B). The consistency between the predicted and actual outcomes is depicted in Figure 4C. The ICGC dataset was utilized to validate the diagnostic model, which consists of 441 COAD samples and 38 normal samples. The model achieved a specificity of 85.4% and a sensitivity of 78.0% in the validation cohort (Figure 4D), with an AUC of 0.846 (Figure 4E). The consistency between the predicted and actual outcome in the validation dataset is shown in Figure 4F. These results indicate that the diagnostic model based on the 2 hypoxia-related genes can effectively distinguish COAD patients from normal individuals.

Figure 4.

A diagnostic model for distinguishing COAD patients from normal samples in the COAD cohort from GSE81558 and ICGC. (A and D) Confusion matrix for the binary classification results of the diagnostic model in the training cohort (GSE81558) and validation cohort (ICGC), respectively. (B and E) ROC curves for evaluating the predictive performance of the diagnostic model in GSE81558 and ICGC cohort, respectively. (C and F) The expression profiles of ENO3 and KDM3A of COAD patients and the consistency between the predicted disease status and real disease status in GSE81558 and ICGC cohort, respectively. Abbreviations: COAD, colon adenocarcinoma; ROC, receiver operating characteristic; ICGC, International Cancer Genome Consortium.

Comparison of the Immune Microenvironment Between the High-Risk and low-Risk Groups

Immunotherapy has become a powerful clinical strategy for treating colon cancers. The TMB and neoantigen load in tumors promote the infiltration of immune effector cells, rendering the tumor susceptible to ICIs.²⁰ Thus, TMB has been established as a biomarker of immunotherapy response in colon cancers. We downloaded the mutation data from the TCGA datasets and calculated the TMB of each COAD patients. The TMB in somatic cells of COAD patients in the high- and low-risk groups are presented in Figure 5A and B, respectively. We also analyzed the association between riskScore and TMB level and found that patients in the high-risk group had a higher TMB level than those in the low-risk group (P = .046) (Figure 5C). The Kaplan–Meier analysis showed that the patients with lower TMB levels had better OS rates than those with higher TMB levels (P = .011) (Figure 5D).

Figure 5.

Correlations between risk scores and TMB, and the predictive performance of TMB on OS. (A and B) The differences in TMB in somatic cells in patients with COAD between the high-risk (A) and low-risk (B) groups. (C) The high-risk group showed a higher level of TMB than the low-risk group. (D) OS rates in patients with low TMB were higher than those in patients with high TMB. Abbreviations: COAD, colon adenocarcinoma; TMB, tumor mutation burden; OS, overall survival.

To further explore the relationship between the riskScore and immune cells and functions, we employed the ssGSEA algorithm to calculate the level of tumor-infiltrating immune cells (Figure 6A). Figure 6B shows the relationship between the riskScore and the expression of the immune checkpoints. As depicted in Figure 6C to F, the expression levels of immune checkpoints CTLA4, IDO1, LAG3, and TIGIT were significantly higher in high-risk patients than in low-risk patients (P < .05).

Figure 6.

The landscape of immune infiltration and expression of immune checkpoints in colon adenocarcinoma (COAD) patients with different risk scores. (A) The correlations between risk score and immune infiltration of 22 immune cell types in COAD patients. (B) The relationship between the risk score and the expression of immune checkpoints. (C-F) The expression levels of immune checkpoints CTLA4 (C), IDO1 (D), LAG3 (E), TIGIT (F) in the high-risk group were higher than those in the low-risk group (P < .05).

Internal and External Validation of the Expression Patterns and Prognostic Predictive Performance of the 2 Hypoxia-Related Genes

The expression levels of ENO3 and KDM3A were found to be significantly higher in the COAD samples from the ICGC cohort compared to normal samples (Figure 7A and B), which is consistent with our analysis in screening for predictive prognostic genes. To further validate our findings, we analyzed the expression levels of these 2 genes in the GSE81558 cohort, which also showed significantly higher expression levels in COAD tissues than in adjacent tissues (Figure 7C and D). In order to confirm the effect of the single gene on the prognosis of COAD patients, we used 2 online databases (UALCAN and GEPIA) to analyze the relationship between survival time and the expression levels of ENO3 and KDM3A. The results revealed that high levels of ENO3 and KDM3A were associated with poor OS time (Figure 7E to H). Furthermore, we performed an immunochemistry experiment on 2 pairs of human COAD tissues and corresponding adjacent tissues. The results showed that ENO3 and KDM3A were expressed at higher levels in tumor tissue than in adjacent tissues, which is consistent with our expected analysis (Figure 7I to L).

Figure 7.

Validation of the expression patterns of the 2 hypoxia-related genes (ENO3 and KDM3A). (A-B) Expression levels of the ENO3 and KDM3A of COAD and normal samples from ICGC cohort. (C-D) Expression levels of the ENO3 and KDM3A of COAD and normal samples from GSE81558 cohort. (E-F) Effects of ENO3 and KDM3A expression levels on COAD patient survival in UALCAN databases. (G-H) Effects of ENO3 and KDM3A expression levels on COAD patient survival in GEPIA databases. (I-L) Immunochemistry results of expression patterns of the ENO3 and KDM3A in COAD tissues and normal tissues. Abbreviations: COAD, colon adenocarcinoma; ICGC, International Cancer Genome Consortium.

Determination of the Oncogenic Effect of ENO3 and KDM3A in COAD Cells

To confirm the malignant biofunctions of ENO3 and KDM3A in COAD cells, we employed gene interference strategy to knockdown the expression of ENO3 and KDM3A. Western blotting analysis confirmed that the application of ENO3 siRNA and KDM3A siRNA significantly inhibited the expression of ENO3 and KDM3A in COAD cells (Figure 8A and B). Furthermore, the results of EdU straining assay showed that the percentage of EdU-positive cells in si-ENO3 or si-KDM3A cells was lower than that in the negative control (Figure 8C and D), indicating that ENO3 and KDM3A promote cell proliferation in COAD cells. Trans-well assay was used to explore the effects of ENO3 and KDM3A on cell invasion in COAD cells. The results demonstrated that the number of invasion cells was decreased in si-ENO3-infected cells as well as si-KDM3A-infected cells (Figure 8E and F). In addition, the viability of DLD1 and HCT116 cells transfected with si-ENO3 or si-KDM3A was sharply decreased compared to the negative control (Figure 8G to J). These data indicated that ENO3 and KDM3A play an important role in cell proliferation and metastasis in COAD cells.

Figure 8.

The relationship between the expression level of hypoxia-related genes (ENO3 and KDM3A) and the progression of colon adenocarcinoma (COAD). (A-B) Western blot analysis confirmed that the expression of ENO3 and KDM3A were inhibited in the COAD cell lines with transfection of ENO3 and KDM3A siRNA. (C-D) DLD1 and HCT116 cells were transfected with ENO3 and KDM3A siRNA. The EdU incorporation assay was performed using a fluorescence method. The images represent 1 field under microscopy (200× ). (E-F) Trans-well invasion assays of DLD1 and HCT116 cells were performed following transfection with the indicated specific siRNA. The invaded cells were stained and counted. The images represent 1 field under microscopy (100×). (G-J) The quantitative statistical results of the viability of DLD1 and HCT116 cells transfected with si-ENO3 or si-KDM3A. All the experiments were repeated 3 times. The data are shown as the mean ± SD. Note: ns is an abbreviation of no significance and indicates P > .05, *P < .05, **P < .01, ***P < .001, ****P < .0001.

Discussion

Colorectal cancer is a prevalent malignant tumor that poses a significant health burden on society. Despite the availability of various staging systems to aid in decision-making and predict the prognosis of COAD patients, the prognosis of these patients remains highly variable. This is primarily due to the fact that these systems are based on clinicopathological criteria and do not account for the complex molecular pathogenic mechanisms involved. In the era of precision medicine, there is an urgent need to develop a robust classifier that can accurately predict the prognosis of COAD patients. Such a classifier would be of great importance in maximizing the benefits of personalized treatment.

The emergence of high-throughput array technology has provided a platform for discovering new genes involved in the onset and progression of COAD. Hypoxia, a typical microenvironmental feature in solid tumors, plays a crucial role in tumor growth, apoptosis, angiogenesis, invasion and metastasis, energy metabolism, and chemoradiotherapy resistance.^27,28 It has been demonstrated to be associated with poor prognosis in patients with various types of tumors. Thus, we developed and validated a predictive hypoxia-related gene-based prognostic signature. Our study confirmed that the signature based on the 2 hypoxia-related genes is an independent prognostic factor for COAD patients. Based on the signature, COAD patients can be classified into high-risk and low-risk groups, which has been proven to be superior to the conventional TNM stage for predicting the prognosis of COAD patients. Additionally, the diagnostic model constructed with the 2 genes could distinguish COAD patients from normal individuals with high sensitivity and specificity. Furthermore, the signature can be used to assess patients’ responsiveness to ICIs therapy, which will greatly benefit COAD patients. Bai et al²⁹ also explored the value of hypoxia-related genes in the prognosis of colorectal cancer patients. Different from our study, they employed an unsupervised consensus clustering algorithm as an initial step to delineate distinct hypoxia-related patterns in COAD patients, followed by the identification of genes associated with prognosis. In contrast, our approach was to directly screen for hypoxia-related genes associated with prognosis using univariate and multivariate Cox regression, as well as LASSO regression, based on survival information. Furthermore, our research has a distinct advantage over Bai's study as we conducted a range of experiments to validate the functionality of the identified genes in tumorigenesis. These included EdU incorporation assays, cell survival assays, western blot assays, and trans-well invasion assays. Our rigorous approach provides robust evidence for the role of these genes in cancer development, enhancing the significance and impact of our findings.

ENO3 has been confirmed to be overexpressed in metastatic colon cancer and is associated with poor prognosis.³⁰ Recent research has shown that knocking down ENO3 has an anticancer impact on STK11 mutant cells, indicating that ENO3-based targeted treatment may be beneficial for patients with STK11 mutations.³¹ KDM3A, a member of the KDM3 family of histone demethylases, plays a crucial role in the tumorigenic potential and survival of human colorectal cancer stem cells by epigenetically activating Wnt target gene transcription.³² A study has demonstrated that KDM3A is an important biomarker for predicting metastasis and prognosis in COAD patients and functions as an oncogene in controlling cell migration and invasion by modifying epithelial–mesenchymal transition (EMT) and matrix metalloproteinases.³³

Immunotherapy has revolutionized cancer treatment and revitalized the field of tumor immunology, achieving long-term durable responses for multiple previously difficult-to-treat solid cancers. However, the therapeutic outcome of immunotherapy varies greatly from patient to patient.³⁴ Researchers have conducted in-depth exploration of the mechanism of immunotherapy resistance and found that the immunosuppressive mechanism is activated in immune-resistant patients. This activation leads to an increase in immune-suppressive cells and molecules, ultimately establishing an immune escape microenvironment.³⁵ Therefore, the immune response may be stimulated by blocking the immune-suppressive mechanism, and ICI therapy works by using this principle. As more clinical trials of ICI therapy are conducted, TMB and immune checkpoints have been found to be reliable biomarkers for assessing whether ICI therapy is suitable for specific individuals.³⁶ Therefore, we also analyzed the TMB in 2 groups classified by signature and found that TMB in the high-risk group is statistically higher than that in the low-risk group. Moreover, the level of TMB was negatively correlated with survival time, consistent with previous research.³⁷ In addition, we detected the expression of immune checkpoints in the 2 groups. The high-risk group exhibited higher expression levels of CTLA4, IDO1, LAG3, and TIGIT compared to the low-risk group, suggesting that the poor prognosis of the high-risk group was partly due to an immunosuppressive microenvironment. CTLA4 is an immunoglobulin superfamily member expressed by activated T cells that transmits an inhibitory signal to T cells, functioning as an immune checkpoint and downregulating immune responses. Previous study has confirmed that anti-CTLA-4 antibodies can enhance antitumor activity by reducing intratumoral regulatory T cells.³⁸ IDO1 is an interferon-inducible protein associated with powerful immunosuppressive effects in cancer.³⁹ The LAG3 protein belongs to immunoglobulin superfamily. LAG3 protein, like CTLA-4 and PD-1, inhibits T cell proliferation, activation, and homeostasis and has been linked to Treg suppressive action.^40,41 It is also reported to be helpful in maintaining CD8+ T cells in a tolerogenic state.⁴² TIGIT is an inhibitory receptor expressed on lymphocytes that can inhibit T cell and NK cell activities. TIGIT has emerged as a crucial anti-tumor response inhibitor that can hinder several stages of the cancer immunity cycle. Blocking TIGIT has been shown in preclinical tests to protect against a variety of solid and hematological malignancies.⁴³

This study has some limitations that must be acknowledged. Firstly, while several independent external validations were included, it was not possible to account for all variances across patients from different geographic regions due to the retrospective acquisition of samples from publicly accessible sources. Secondly, the intricate interplay between tumor cells and immune cells in hypoxic environments requires further investigation. In the future, we intend to delve deeper into the role of ENO3 and KDM3A in tumorigenesis.

Conclusion

In this study, we identified 2 genes that are associated with hypoxia and have predictive value in determining the prognosis of patients with COAD. Our prognostic and diagnostic models, based on these genes, have demonstrated superior performance compared to the conventional TNM stage. Furthermore, our model can assist clinicians in identifying COAD patients who are likely to benefit from immunotherapy. These findings have significant implications for the clinical management of COAD patients and highlight the importance of incorporating molecular biomarkers into prognostic models.

Supplemental Material

sj-docx-1-tct-10.1177_15330338231195494 - Supplemental material for Predictive Models for Colon Adenocarcinoma Diagnosis, Prognosis, and Immune Microenvironment Based on 2 Hypoxia-Related Genes: KDM3A and ENO3

Supplemental material, sj-docx-1-tct-10.1177_15330338231195494 for Predictive Models for Colon Adenocarcinoma Diagnosis, Prognosis, and Immune Microenvironment Based on 2 Hypoxia-Related Genes: KDM3A and ENO3 by Chunli Kong, Liyun Zheng, Shiji Fang, Minjiang Chen, Guihan Lin, Rongfang Qiu, Zhongwei Zhao, Weiqian Chen, Jingjing Song, Yang Yang and Jiansong Ji in Technology in Cancer Research & Treatment

Footnotes

Author's Contribution

Jiansong Ji and Yang Yang performed the study concept and design; Chunli Kong and Minjiang Chen analyzed the data; Guihan Lin and Rongfang Qiu contributed analysis tools; Liyun Zheng and Shiji Fang performed the experiments; Zhongwei Zhao, Weiqian Chen, and Jingjing Song provided acquisition, analysis and interpretation of data, and statistical analysis; Yang Yang and Jiansong Ji edited the paper. All authors read and approved the final paper.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Ethics Statement

This study was approved by the Institutional Review Board and Human Ethics Committee of the Fifth Affiliated Hospital of Wenzhou Medical University (approval number: 2022-312, Lishui, China), and the requirement for informed consent was waived for this is a retrospective study.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study was supported by Zhejiang Medical and Health Science Project (2022RC087).

ORCID iD

Jiansong Ji

Supplemental Material

Supplemental material for this article is available online.

Abbreviations

References

Sung

Ferlay

Siegel

, et al. Global cancer statistics 2020: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin. 2021;71(3):209‐249. doi:10.3322/caac.21660

Siegel

Miller

Fuchs

Jemal

. Cancer statistics, 2021. CA Cancer J Clin. 2021;71(1):7‐33. doi:10.3322/caac.21654

Siegel

Miller

Goding Sauer

, et al. Colorectal cancer statistics, 2020. CA Cancer J Clin. 2020;70(3):145‐164. doi:10.3322/caac.21601

Edge

Compton

. The American joint committee on cancer: The 7th edition of the AJCC cancer staging manual and the future of TNM. Ann Surg Oncol. 2010;17(6):1471‐1474. doi:10.1245/s10434-010-0985-4

Knijn

van Exsel

UEM

de Noo

Nagtegaal

. The value of intramural vascular invasion in colorectal cancer - a systematic review and meta-analysis. Histopathology. 2018;72(5):721‐728. doi:10.1111/his.13404

Huang

Qin

Xiao

, et al. Association of tumor differentiation and prognosis in patients with rectal cancer undergoing neoadjuvant chemoradiation therapy. Gastroenterol Rep (Oxf). 2019;7(4):283‐290. doi:10.1093/gastro/goy045

Zhang

, et al. Gene expression signature to predict prognosis and adjuvant chemosensitivity of colorectal cancer patients. Cancer Manag Res. 2020;12:3301‐3310. doi:10.2147/cmar.S243490

Stelzner

Radulova-Mauersberger

Zschuppe

, et al. Prognosis in patients with synchronous colorectal cancer metastases after complete resection of the primary tumor and the metastases. J Surg Oncol. 2019;120(3):438‐445. doi:10.1002/jso.25578

Joseph

Harishankar

Pillai

Devi

. Hypoxia induced EMT: A review on the mechanism of tumor progression and metastasis in OSCC. Oral Oncol. 2018;80:23‐32. doi:10.1016/j.oraloncology.2018.03.004

10.

Kim

Lee

Park

. Quinacrine-Mediated Inhibition of Nrf2 Reverses Hypoxia-Induced 5-Fluorouracil Resistance in Colorectal Cancer. Int J Mol Sci. 2019;20(18):4366. 10.3390/ijms20184366

11.

Emami Nejad

Najafgholian

Rostami

, et al. The role of hypoxia in the tumor microenvironment and development of cancer stem cell: A novel approach to developing treatment. Cancer Cell Int. 2021;21(1):62. doi:10.1186/s12935-020-01719-5

12.

Tang

Chen

Bao

, et al. Hypoxic tumor microenvironment activates GLI2 via HIF-1α and TGF-β2 to promote chemoresistance in colorectal cancer. Proc Natl Acad Sci U S A. 2018;115(26):E5990‐e5999. doi:10.1073/pnas.1801348115

13.

Noman

Desantis

Janji

, et al. PD-L1 is a novel direct target of HIF-1α, and its blockade under hypoxia enhanced MDSC-mediated T cell activation. J Exp Med. 2014;211(5):781‐790. doi:10.1084/jem.20131916

14.

Chouaib

Noman

Kosmatopoulos

Curran

. Hypoxic stress: Obstacles and opportunities for innovative immunotherapy of cancer. Oncogene. 2017;36(4):439‐445. doi:10.1038/onc.2016.225

15.

Yuen

Wong

. Hypoxia-inducible factors and innate immunity in liver cancer. J Clin Invest. 2020;130(10):5052‐5062. doi:10.1172/jci137553

16.

Zhang

Wang

, et al. Effects of hypoxia in intestinal tumors on immune cell behavior in the tumor microenvironment. Front Immunol. 2021;12:645320. doi:10.3389/fimmu.2021.645320

17.

Wang

Zhao

Zhang

, et al. Targeting hypoxia in the tumor microenvironment: A potential strategy to improve cancer immunotherapy. J Exp Clin Cancer Res. 2021;40(1):24. doi:10.1186/s13046-020-01820-7

18.

O'Donnell

Teng

MWL

Smyth

. Cancer immunoediting and resistance to T cell-based immunotherapy. Nat Rev Clin Oncol. 2019;16(3):151‐167. doi:10.1038/s41571-018-0142-8

19.

Sinicrope

Shi

Allegra

, et al. Association of DNA mismatch repair and mutations in BRAF and KRAS with survival after recurrence in stage III colon cancers : A secondary analysis of 2 randomized clinical trials. JAMA Oncol. 2017;3(4):472‐480. doi:10.1001/jamaoncol.2016.5469

20.

Samstein

Lee

Shoushtari

, et al. Tumor mutational load predicts survival after immunotherapy across multiple cancer types. Nat Genet. 2019;51(2):202‐206. doi:10.1038/s41588-018-0312-8

21.

Cristescu

Mogg

Ayers

, et al. Pan-tumor genomic biomarkers for PD-1 checkpoint blockade-based immunotherapy. Science. 2018;362(6411):eaar3593. doi:10.1126/science.aar3593

22.

Schrock

Ouyang

Sandhu

, et al. Tumor mutational burden is predictive of response to immune checkpoint inhibitors in MSI-high metastatic colorectal cancer. Ann Oncol. 2019;30(7):1096‐1103. doi:10.1093/annonc/mdz134

23.

Picard

Verschoor

Pawelec

. Relationships between immune landscapes, genetic subtypes and responses to immunotherapy in colorectal cancer. Front Immunol. 2020;11:369. doi:10.3389/fimmu.2020.00369

24.

Cui

Guo

Zhang

, et al. ENO3 Inhibits growth and metastasis of hepatocellular carcinoma via wnt/β-catenin signaling pathway. Front Cell Dev Biol. 2021;9:797102. doi:10.3389/fcell.2021.797102

25.

Wang

Long

Tang

, et al. Histone demethylase KDM3A is required for enhancer activation of hippo target genes in colorectal cancer. Nucleic Acids Res. 2019;47(5):2349‐2364. doi:10.1093/nar/gky1317

26.

Fang

Zhang

Weng

, et al. CPSF7 Regulates liver cancer growth and metastasis by facilitating WWP2-FL and targeting the WWP2/PTEN/AKT signaling pathway. Biochim Biophys Acta Mol Cell Res. 2020;1867(2):118624. doi:10.1016/j.bbamcr.2019.118624

27.

Shao

Yang

Miao

, et al. Role of hypoxia-induced exosomes in tumor biology. Mol Cancer. 2018;17(1):120. doi:10.1186/s12943-018-0869-y

28.

Zhang

Yang

Peng

, et al. Hypoxia constructing the prognostic model of colorectal adenocarcinoma and related to the immune microenvironment. Front Cell Dev Biol. 2021;9:665364. doi:10.3389/fcell.2021.665364

29.

Bai

Chen

Yan

, et al. Exploration of different hypoxia patterns and construction of a hypoxia-related gene prognostic Index in colorectal cancer. Front Immunol. 2022;13:853352. doi:10.3389/fimmu.2022.853352

30.

Cui

Sun

Bhandari

, et al. Comprehensive analysis of glycolysis-related genes for prognosis, immune features, and candidate drug development in colon cancer. Front Cell Dev Biol. 2021;9:684322. doi:10.3389/fcell.2021.684322

31.

Harada

Matsumoto

Hirota

, et al. Chemically modified antisense oligonucleotide against ARL4C inhibits primary and metastatic liver tumor growth. Mol Cancer Ther. 2019;18(3):602‐612. doi:10.1158/1535-7163.Mct-18-0824

32.

Deng

, et al. KDM3 Epigenetically controls tumorigenic potentials of human colorectal cancer stem cells through wnt/β-catenin signalling. Nat Commun. 2017;8:15146. doi:10.1038/ncomms15146

33.

Liu

Liang

Zhangsun

. KDM3A Is associated with tumor metastasis and modulates colorectal cancer cell migration and invasion. Int J Biol Macromol. 2019;126:318‐325. doi:10.1016/j.ijbiomac.2018.12.105

34.

Bagchi

Yuan

Engleman

. Immune checkpoint inhibitors for the treatment of cancer: Clinical impact and mechanisms of response and resistance. Annu Rev Pathol. 2021;16:223‐249. doi:10.1146/annurev-pathol-042020-042741

35.

Jackson

Choi

Lim

. Mechanisms of immunotherapy resistance: Lessons from glioblastoma. Nat Immunol. 2019;20(9):1100‐1109. doi:10.1038/s41590-019-0433-y

36.

Chabanon

Pedrero

Lefebvre

Marabelle

Soria

Postel-Vinay

. Mutational landscape and sensitivity to immune checkpoint blockers. Clin Cancer Res. 2016;22(17):4309‐4321. doi:10.1158/1078-0432.Ccr-16-0903

37.

Zhou

Xie

Wang

, et al. Correlations between tumor mutation burden and immunocyte infiltration and their prognostic value in colon cancer. Front Genet. 2021;12:623424. doi:10.3389/fgene.2021.623424

38.

Selby

Engelhardt

Quigley

, et al. Anti-CTLA-4 antibodies of IgG2a isotype enhance antitumor activity through reduction of intratumoral regulatory T cells. Cancer Immunol Res. 2013;1(1):32‐42. doi:10.1158/2326-6066.Cir-13-0013

39.

Munn

Mellor

. IDO In the tumor microenvironment: Inflammation, counter-regulation, and tolerance. Trends Immunol. 2016;37(3):193‐207. doi:10.1016/j.it.2016.01.002

40.

Workman

Vignali

. The CD4-related molecule, LAG-3 (CD223), regulates the expansion of activated T cells. Eur J Immunol. 2003;33(4):970‐979. doi:10.1002/eji.200323382

41.

Huang

Workman

Flies

, et al. Role of LAG-3 in regulatory T cells. Immunity. 2004;21(4):503‐513. doi:10.1016/j.immuni.2004.08.010

42.

Grosso

Kelleher

Harris

, et al. LAG-3 regulates CD8+ T cell accumulation and effector function in murine self- and tumor-tolerance systems. J Clin Invest. 2007;117(11):3383‐3392. doi:10.1172/jci31184

43.

Harjunpää

Guillerey

. TIGIT As an emerging immune checkpoint. Clin Exp Immunol. 2020;200(2):108‐119. doi:10.1111/cei.13407

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.66 MB