A Novel DNA Repair-Gene Model to Predict Responses to Immunotherapy and Prognosis in Patients With EGFR-Mutant Non-Small Cell Lung Cancer

Abstract

Background: The epidermal growth factor receptor mutant (EGFRm) non-small cell lung cancer (NSCLC) has a unique "cold" immune profile. DNA damage repair (DDR) genes are closely related to tumorigenesis and the effectiveness of immunotherapy in many tumors. However, the role and mechanism of DDR in the genesis and progression of EGFRm NSCLC remain unclear.

Methods: This study included 101 EGFRm NSCLC samples from The Cancer Genome Atlas (TCGA) dataset and a GSE31210 dataset (external set) from the GEO database. Cluster analysis was used to identify different subtypes of EGFRm NSCLC based on the expression of DDR genes. Univariate and LASSO regression analysis was used to develop a DDR-based predictive model. The prognostic significance of this model was assessed using Cox regression, Kaplan-Meier, and receiver operating characteristic (ROC) curve analyses. Bioinformatics analysis was performed to investigate the clinicopathological characteristics and immune profiles associated with this model. In vitro experiment was performed to testify the role of DDR genes in EGFRm NSCLC.

Results: We identified two subtypes of EGFRm NSCLC: DDR-activated and DDR-suppressed. The DDR-activated subtype showed more aggressive clinical behavior and poorer prognosis and was more responsive to immunotherapy. A prognostic model for EGFRm NSCLC was constructed using four DDR genes: CAPS, FAM83A, IGLV8-61, and SLC7A5. The derived risk score could serve as an independent prognostic indicator. High- and low-risk patients exhibited distinct clinicopathological characteristics, immune profiles, and responses to immunotherapy. The T-cell inflammation and Tumor Immune Dysfunction and Exclusion (TIDE) scores differed between the high- and low-risk subgroups, with both showing enhanced effectiveness of immunotherapy in the low-risk subgroup. Targeted therapy such as BI.2536, an inhibitor of polo-like kinase 1, could be effective for patients with high-risk EGFRm NSCLC. Meanwhile, in vitro detection approved the role of DDR genes in EGFRm NSCLC response.

Conclusion: This study demonstrated a diversity of DDR genes in EGFRm NSCLC and developed a predictive model using these genes. This model could assist in identifying potential candidates for immunotherapy and in assessing personalized treatment and prognosis of patients with EGFRm NSCLC.

Keywords: DNA‐damage repair; epidermal growth factor receptor; non‐small cell lung cancer; prognostic model.

FIGURE 1

Study flowchart.

FIGURE 2

Classification based on differences in DNA damage repair (DDR) genes and clinical characteristics between clusters. (A) Consensus clustering analysis based on DDR gene expression. (B) Evaluation of differences in clinical characteristics between DDR‐activated/suppressed groups. (C) Kaplan–Meier plot showing the overall survival of the two clusters. *p < 0.05, ns, not statistically significant.

FIGURE 3

Association between DNA damage repair (DDR) genes and the tumor immune microenvironment (TME). (A) Relationship between the TME score and selected alterations in the DDR genes. (B) Comparison of the immune cell composition between the two clusters. (C) Heatmap showing the correlation between immune cells in the two clusters. (D) Comparison of tumor mutation burden (TMB) in the two clusters. *p < 0.05, **p < 0.01, ***p < 0.001.

FIGURE 4

Identification of hub DDR genes associated with the prognosis. (A) Volcano plot showing differentially expressed genes (DEGs) between the two clusters. (B) Heatmap showing the top 20 DEGs in the two clusters. (C) Coefficient profiles of the LASSO regression model. (D) Cross‐validation for tuning parameter screening in the LASSO regression model. (E) Results of LASSO‐Cox regression analysis of hub DDR genes.

FIGURE 5

Construction and validation of the prognostic model. (A) Risk score plot. (B) The relationship between risk scores and survival status of each patient. (C) Kaplan–Meier survival curves for high‐risk and low‐risk patients in the training set, testing set, and whole set. (D) ROC curves of the prognostic model in the training set, testing set, and whole set. (E) ROC curve of the prognostic model in the external set. (F) Clustering heatmap of hub DDR genes.

FIGURE 6

Analysis of correlations between clinical features and risk scores. (A) Difference in risk scores based on TNM stages. (B) Difference in risk scores based on sex. (C) Correlation between risk scores and tumor purity. (D) Correlation between risk scores and age. (E) Univariate cox regression analysis for prognosis. (F) Multivariable cox regression analysis for prognosis. **p < 0.01.

FIGURE 7

GSEA of specific pathways enriched by the high‐risk and low‐risk groups.

FIGURE 8

Drug sensitivity and tumor immune microenvironment (TME) analysis of the prognostic model. (A) Drug sensitivity analysis. (B) Comparison of the T‐cell inflammation score (left) and Tumor Immune Dysfunction and Exclusion (TIDE) score (right) in the low‐ and high‐risk groups. (C) Comparison of the immune cell composition in the low‐ and high‐risk groups. *p < 0.05, **p < 0.01, ***p < 0.001.

FIGURE 9

The influence of DDR‐related gene on NSCLC cells in vitro. (A) The protein content in five groups with western blotting analysis. (B) The quantitative analysis results. (C) Then the quantitative real‐time PCR assay results showed that the expression trend of Ki‐67 in each group. (D) Then the quantitative real‐time PCR assay results showed that the expression trend of Bcl‐2 in each group. (E) Then the quantitative real‐time PCR assay results showed that the expression trend of caspase in each group. (F) The transwell assay showed that compared with the control group, more cells could be found in the OE‐FAM83A group and OE‐SLC7A5 group, with lower cell number in the OE‐CAPS group and OE‐IGLV8‐61 group. *p < 0.05, **p < 0.01, ***p < 0.001. Bar = 200 μm.