Constructing an immune-related prognostic model and exploring the function of HMGB3, TNFSF4, and CORO2A in breast cancer

Introduction

As the leading oncological burden in women’s health, breast cancer (BC) continues to exert substantial socioeconomic impacts across healthcare systems (1,2). Epidemiological surveillance data from 2022 showed that female BC accounted for over 2.3 million newly diagnosed cases and 665,684 deaths worldwide, positioning it as the second most prevalent neoplasm and fourth leading cause of cancer-associated death (3). Conventional treatment paradigms encompassing surgical resection, cytotoxic agents, and ionizing radiation are frequently associated with clinically significant toxicities, highlighting the imperative for precision therapeutics with improved safety profiles (4,5).

The advent of immune checkpoint blockade has revolutionized therapeutic paradigms in cancer management, such as malignant melanoma (6), non-small cell lung cancer (7), and hepatocellular carcinoma (8). Initially regarded as a poorly immunogenic tumor, BC has increasingly demonstrated potential for benefiting from immunotherapeutic strategies. Immune checkpoint inhibitors (ICIs) targeting programmed death receptor-1/programmed death receptor ligand-1 (PD-1/PD-L1) and cytotoxic T lymphocyte-associated antigen-4 (CTLA-4) have demonstrated significant therapeutic potential in specific BC populations, particularly in triple-negative breast cancer (TNBC) and advanced-stage disease (9,10). Despite advancements, the effectiveness of immunotherapy in BC is still limited, emphasizing the critical need for dependable predictive biomarkers to identify patients who are most likely to benefit from these therapeutic approaches.

Emerging evidence highlights the critical involvement of immune-related genes (IRGs) in remodeling tumor microenvironment (TME) architecture and determining treatment effectiveness (11). Multigene prognostic models, which integrate transcriptomic data and clinical parameters, have shown potential in stratifying patients and guiding personalized treatment strategies (12). This investigation developed an immunogenomic signature for BC prognosis. Furthermore, we validated the model’s prognostic value using an independent dataset and explored the correlations between IRGs and clinical features, as well as immune infiltration. Our findings demonstrated the model’s potential as a prognostic tool and offered important guidance for refining immunotherapeutic regimens in BC management. We present this article in accordance with the TRIPOD reporting checklist (available at https://tcr.amegroups.com/article/view/10.21037/tcr-2025-1049/rc).

Methods

Data gathering

In the present study, clinical data and RNA sequencing profiles of the BC patient cohort were obtained from The Cancer Genome Atlas (TCGA) database (https://www.cancer.gov/), with a total of 1,050 cases ultimately included in the analysis. To identify differentially expressed genes (DEGs), two independent datasets—GSE45827 and GSE29431—were retrieved from the Gene Expression Omnibus (GEO; https://www.ncbi.nlm.nih.gov/geo/), which comprised 155 and 66 BC patients, respectively. Subsequently, these two GEO datasets were intersected with datasets derived from the TCGA and InnateDB (https://www.innatedb.com/) databases, leading to the identification of 62 DEGs. For the acquisition of tumor progression-associated genes (TPGs), datasets from TCGA and InnateDB were further intersected, resulting in 118 TPGs. Finally, the overlap between TPGs and DEGs yielded nine IRGs. Additionally, to validate the constructed model, the GSE20685 dataset was downloaded from the GEO database; this dataset provides transcriptomic profiles and associated clinicopathological features of 327 BC patients. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments.

Construction and validation of the prognostic model for BC

Least absolute shrinkage and selection operator (LASSO) regression was employed to screen the IRGs of BC. Subsequently, a composite risk score was mathematically derived by integrating expression levels of these genes with corresponding regression coefficients, stratifying the individuals into high and low-risk cohorts. Multivariate distribution patterns were visualized through uniform manifold approximation and projection (UMAP) and principal component analysis (PCA) methods. Survival probability was performed utilizing the Kaplan-Meier estimator, while temporal prognostic accuracy was evaluated via the receiver operating characteristic curve (ROC) tool at standardized clinical intervals (1-, 3-, and 5-year).

Establishment and evaluation of the prognostic nomogram

Associations between expression levels of IRGs and multidimensional clinical parameters (including age, gender, tumor stage, and vital status) were graphically represented through a clustered heatmap. Stepwise regression protocols (univariate and multivariate Cox proportional hazards modeling) were implemented to establish prognostic independence of these clinicopathological indicators. Next, the nomogram was established to forecast the long-term survival of BC via integrating multiple metrics, comprising age, tumor stage, and risk score. Validation analyses of the prognostic nomogram incorporated (I) longitudinal discriminative capacity evaluation via the ROC method; and (II) prognostic precision assessment using calibration curves.

Expression profiles of IRGs

For subsequent single-gene analysis, we only focused on “risk genes” (HMGB3, TNFSF4, CORO2A), which is a research focus-driven choice. The core objective of this study is to construct the BC risk model and explore the association between these risk genes and BC. This direction is in line with the clinical need to identify potential therapeutic targets (oncogenes are often more likely to be targeted for intervention than protective genes). Therefore, we give priority to the functional and clinical relevance analysis of risk genes. Firstly, expression analyses of the three IRGs were conducted using the “stats” and “car” packages for statistical testing, and “ggplot2” for visualization. Tumor and paired normal tissue RNA-sequencing data from the BC cohort were retrieved from the TCGA database. Expression values were log2-transformed [log2(value + 1)] prior to analysis, and paired two-tailed t-tests were applied to compare expression levels of IRGs between tumor and normal tissues. UALCAN (http://ualcan.path.uab.edu/analysis-prot.html) (13), and Human Protein Atlas (HPA, https://www.proteinatlas.org/) database were used to validate the protein expression level of IRGs, while the subcellular localizations of these genes were demonstrated through indirect immunofluorescence microscopy (14).

Clinical associations and drug sensitivity analyses of IRGs

Subsequently, correlation analyses between the expression profiles of IRGs and clinical characteristics of BC patients (tumor pathological stage and overall survival) were conducted. The Gene Set Cancer Analysis (GSCA, https://guolab.wchscu.cn/GSCA/#/) and Gene Expression Profiling Interactive Analysis (GEPIA, http://gepia2.cancer-pku.cn/#analysis) databases were employed to examine the relationship between the expression levels of the three genes and tumor stage, and patient survival outcomes, respectively (15,16). Associations between IRGs expression and drug sensitivity (IC50) were also conducted by the GSCA platform in the context of BC.

Immune landscape of IRGs

Relationship between microenvironmental scoring metrics (Stromal/Immune/ESTIMATE Scores) and IRGs expression levels was performed by R package “ESTIMATE”. The CIBERSORT algorithm was used to demonstrate the correlations between gene expression patterns and the infiltration level of various immune cells. The clinical significance of tumor immune subsets was further explored using the “outcome module” in the Tumor Immune Estimation Resource (TIMER) database (17). Single-cell data of BC were derived from the Tumor Immune Single-cell Hub (TISCH), and processed via R packages “MAESTRO” and “Seurat” (18). The TIDE (Tumor Immune Dysfunction and Exclusion) algorithm was employed to predict the potential immunotherapeutic effect for BC patients (19).

Statistical analysis

We employed the Shapiro-Wilk normality test to compare the normality of variables between groups. When the normality test is satisfied, we use the paired sample t-test; otherwise, we use the Wilcoxon signed rank test. Spearman correlation analysis was conducted to calculate correlation coefficients. We completed all statistical analyses using the software R and SPSS 25. The value of P<0.05 was considered significant (*, P<0.05; **, P<0.01; ***, P<0.001; ****, P<0.0001). The “ns” represented no statistically significant.

Results

Screening of IRGs

The screening process for DEGs, TPGs, and IRGs was visualized using a Venn diagram, as shown in Figure 1A. We identified 62 DEGs and 118 TPGs through cross-platform harmonization of three cancer genomics repositories: GEO, TCGA, and Innate-DB. Subsequent intersectional analysis revealed 9 overlapping candidate genes with two types of expression patterns in BC. In which, high mobility group box 3 (HMGB3), Coronin 2A (CORO2A), and tumor necrosis factor ligand superfamily member 4 (TNFSF4, also known as OX-40L) exhibited significant upregulation in BC tissues, while suppressor of cytokine signaling 3 (SOCS3), C-X-C motif chemokine ligand 2 (CXCL2), tachykinin receptor 1 (TACR1), protein S (PROS1), FRAS1 related extracellular matrix 1 (FREM1), and interleukin 33 (IL33) showed marked downregulation across tumor samples (Figure 1B,1C). Univariate Cox proportional hazards regression confirmed significant prognostic associations for all 9 IRGs (P<0.05), establishing their predictive capacity for clinical outcomes (Figure 1D).

Figure 1 Screening of IRGs. (A) Identifying DEGs and TPGs with overall survival (Venn diagram). (B) Heatmap of the expression of 9 IRGs in BC samples. (C) Expression levels of the 9 IRGs in BC and paired normal tissues. (D) Forest plot exhibiting univariate Cox regression analysis of 9 IRGs in BC patients. ***, P<0.001. BC, breast cancer; CI, confidence interval; DEGs, differentially expressed genes; HR, hazard ratio; IRGs, immune-related genes; TCGA, The Cancer Genome Atlas; TPGs, tumor progression-associated genes; TPM, transcripts per million.

Development of the IRGs prognostic model

Penalized regression analysis via LASSO methodology with ten-fold cross-validation identified 6 prognostic-related genes: HMGB3 (20), TNFSF4 (21,22), CORO2A (23,24), SOCS3 (25), TACR1 (26), and FREM1 (27), selected from the 9 candidate IRGs (Figure 2A). The important research findings of the six IRGs in BC were summarized in Box 1. All BC cases were stratified into low- and high-risk cohorts using median risk index thresholding, which was calculated through integration of these molecular predictors weighted by their respective LASSO regression coefficients (β). The specific calculation formula was as follows: Risk Index = Σ (β_i × X_i), where X represents gene expression levels, and the corresponding β values of HMGB3, TNFSF4, CORO2A, SOCS3, TACR1, and FREM1 were 0.14625006, 0.10697875, 0.07342066, −0.0208845, −0.0723798, and −0.1190182 (Figure 2B). The results of PCA and UMAP indicate that our model can well distinguish BC cohorts with different risks (Figure 2C). The prognostic stratification capacity was further validated via Kaplan-Meier analysis, demonstrating significant survival disparity (P=0.003) (Figure 2D). Temporal predictive accuracy assessment using the ROC tool yielded area under the curve (AUC) values of 0.676 (1-year), 0.646 (3-year), and 0.621 (5-year), showing robust prognostic discrimination capacity (Figure 2E).

Figure 2 Development of the IRGs prognostic model. (A) LASSO regression analysis and coefficient profiles of IRGs. (B) Risk score distribution and the survival status of patients with BC. (C) PCA and UMAP analyses of the BC patients displaying the distribution in high and low-risk populations. In the PCA plot, the horizontal axis represents PCA1, reflecting the largest direction of variation in the expression data of these prognostic genes, accounting for 33.99% of the total variation. The vertical axis represents PCA2, which indicates the second largest direction of variation in the data and explains 20.94% of the sample differences. In the UMAP graph, UMAP1 (horizontal axis) and UMAP2 (vertical axis) are two dimensions obtained after dimensionality reduction. They map the high-dimensional prognostic gene expression data into a two-dimensional space, allowing for a more intuitive observation of the sample distribution. The red and blue dots represent the high-risk and low-risk groups, respectively. In both the PCA and UMAP graphs, there is a certain degree of separation between the high-risk group and the low-risk group, indicating that the prognostic genes screened out by the model can reflect the differences in gene expression characteristics of breast cancer patients in different risk groups. (D) Survival analysis of patients in high and low-risk groups. (E) AUC of the time-dependent ROC curves. AUC, area under the curve; BC, breast cancer; CI, confidence interval; FPR, false positive rate; HR, hazard ratio; IRGs, immune-related genes; LASSO, least absolute shrinkage and selection operator; PCA, principal component; TPR, true positive rate; UMAP, Uniform Manifold Approximation and Projection.

Constructing the prognostic nomogram

Associations between risk index and various clinical traits (age, sex, tumor stage, and survival outcomes), and expression patterns of the six biomarkers were pictured in the heatmap (Figure 3A). Univariate Cox proportional hazards regression revealed significant prognostic associations for advanced age (HR =2.111, 95% CI: 1.507–2.958), late-stage disease (HR =2.669, 95% CI: 1.907–3.735), and elevated risk index (HR =4.307, 95% CI: 2.266–8.185) with clinical outcomes (Figure 3B). Multivariate adjustment confirmed that the age, tumor stage, and risk index retained independent prognostic capacity (Figure 3C). The following prognostic nomogram was established through integration of these three independent predictors, computing the 1-, 3-, and 5-year survival probability rates (Figure 3D). Temporal discrimination analysis demonstrated superior predictive accuracy with 1-year (AUC =0.877), 3-year (AUC =0.775), and 5-year (AUC =0.716) survival predictions (Figure 3E). Calibration curves exhibited excellent concordance between nomogram nomogram-predicted and observed survival probabilities across diverse timepoints (Figure 3F).

Figure 3 Constructing the prognostic nomogram. (A) Heatmap showing the association between IRGs expression and clinical variables and risk groups. Univariate (B) and multivariate (C) Cox regression analysis of BC patients (forest plots). (D) Nomogram-based gene signature for predicting 1-, 3-, and 5-year survival probability. ROC curves (E) and calibration curves (F) for predicting the 1-, 3-, and 5-year performance of the nomogram. AUC, area under the curve; BC, breast cancer; CI, confidence interval; FPR, false positive rate; HR, hazard ratio; IRGs, immune-related genes; ROC, receiver operating characteristic; TPR, true positive rate.

Validation of the IRGs-based prognostic model

Specimens of the independent gene cohort (n=327) were prospectively stratified into two prognostic subgroups through application of the median risk threshold (Figure 4A). Principal component analysis demonstrated effective risk-based segregation within the validation cohort (Figure 4B), in line with survival patterns observed through K-M estimation (P<0.001, HR =2.25) (Figure 4C). Figure 4D showed the time-dependent AUC values of 0.709, 0.612, and 0.637 at 1-, 3-, 5-year intervals, respectively, validating the generalizability of the model in heterogeneous individuals. As shown in Figure 4E, heatmap visualization elucidated significant relationships between the six immune regulators, clinical parameters, and risk stratification. The univariate and multivariate Cox regression analysis indicated tumor stage (P<0.001, HR =3.984, 95% CI: 2.552–6.222) and risk score (P<0.001, HR =4.112, 95% CI: 1.857–9.106) as independent prognostic metrics (Figure 4C). This external validation substantiated the model’s clinical robustness, exhibiting consistent prognostic distinction ability across different biological contexts.

Figure 4 Validating the IRGs-based prognostic model. (A) Risk score distribution of the survival status of BC patients. (B) UMAP plot and PCA analyses of BC patients exhibiting the distribution in high and low-risk groups. (C) K-M curves of patients in high and low-risk groups. (D) AUC of the time-dependent ROC curves. (E) Heatmap demonstrating the association between IRGs expression and clinical factors and risk groups. Univariate Cox regression analysis (F) and multivariate Cox regression analysis (G) of BC patients. AUC, area under the curve; BC, breast cancer; CI, confidence interval; FPR, false positive rate; HR, hazard ratio; IRGs, immune-related genes; PCA, principal component analysis; ROC, receiver operating characteristic; TPR, true positive rate; UMAP, Uniform Manifold Approximation and Projection.

Expression patterns and clinical correlations of the specific three IRGs

Cox regression analysis revealed three risk genes for BC: HMGB3, CORO2A, and TNFSF4. We then focused on analyzing relationships between the three genes’ expression, drug sensitivity, and immune infiltration. As graphed in Figure S1A, the protein expression level of HMGB3 was elevated in BC. Complementary immunohistochemistry (IHC) staining analysis demonstrated consistent upregulation of all three biomarkers in BC samples, which was in line with the expression landscapes at the gene level (Figure 5A). Subcellular localization analysis via immunofluorescence microscopy described distinct localization profiles: HMGB3 predominantly nuclear, CORO2A cytoskeletal-associated, and TNFSF4 exhibiting membrano-cytoplasmic distribution (Figure S1B). In addition, single-cell cluster analysis demonstrated that HMGB3 and TNFSF4 were highly expressed in Tprolif, while CORO2A was the most abundant expression in Tregs (Figure S2).

Figure 5 Expression patterns and clinical correlations of three specific upregulated IRGs. (A) Immunohistochemistry staining for HMGB3, CORO2A, and TNFSF4 in BC and para-carcinoma tissue. Images from the HPA—HMGB3 (breast tissue): https://images.proteinatlas.org/62583/141412_B_3_4.jpg; HMGB3 (breast cancer): https://images.proteinatlas.org/62583/141410_A_5_1.jpg; CORO2A (breast tissue): https://images.proteinatlas.org/41161/92983_B_1_4.jpg; CORO2A (breast cancer): https://images.proteinatlas.org/41161/92664_A_5_1.jpg; TNFSF4 (breast tissue): https://images.proteinatlas.org/59579/134109_B_3_4.jpg; TNFSF4 (breast cancer): https://images.proteinatlas.org/59579/134111_A_4_4.jpg.(B) Correlations between expression levels of IRGs and pathological stages of BC. (C) Correlations between expression levels of IRGs and clinical outcomes (overall survival). BC, breast cancer; HR, hazard ratio; IRGs, immune-related genes; RSEM, RNA-Seq by Expectation-Maximization; TPM, transcripts per million.

Figure 5B revealed a time-dependent upregulation of HMGB3 expression, with significant differential expression observed between pathologic stage III and stage II (P=0.01). In contrast, CORO2A and TNFSF4 demonstrated stable expression patterns among tumor stages, with all P values being above 0.05. Survival analysis revealed significant correlations between overexpression of the three genes and reduced OS: HMGB3 (P=0.02), CORO2A (P=0.02), and TNFSF4 (P=0.01) (Figure 5C). It is worth noting that throughout the survival period of BC, increased HMGB3 is not always associated with a poor prognosis. In all, the results of survival analysis further elucidated the significant impact of HMGB3, CORO2A, and TNFSF4 on the prognosis of BC patients.

Drug sensitivity analysis of three IRGs

Therapeutic resistance to cytotoxic and molecularly-targeted agents remains a persistent clinical obstacle in contemporary oncology practice (28). Pharmacogenomic landscape analysis integrating Genomics of Drug Sensitivity in Cancer (GDSC) and Cancer Therapeutics Response Portal (CTRP) datasets revealed distinct therapeutic sensitivity patterns. For GDSC results (Figure 6A), CORO2A expression was positively related to most drugs among the thirty drugs, excluding afatinib, cetuximab, and gefitinib. TNFSF4 demonstrated inverse associations with EGFR-targeted therapies (afatinib, cetuximab, and gefitinib). HMGB3 was positively associated with multiple drugs, especially the drug 5Z-7-Oxozeaenol. For CTRP results (Figure 6B), CORO2A, TNFSF4, and HMGB3 displayed robust correlations with various drugs. Drug sensitivity analysis landscape displayed 30 most relevant drugs of these three genes in the context of pan-cancer, providing valuable therapeutic strategies for clinical decisions.

Figure 6 Drug sensitivity analysis of three upregulated IRGs. (A) Correlation between GDSC drug sensitivity and three IRGs mRNA expression. (B) Correlation between CTRP drug sensitivity and three IRGs mRNA expression. Positive correlation suggests that a higher gene expression may result in drug resistance. Negative correlation indicates that a higher gene expression may lead drug sensitive. CTRP, Cancer Therapeutics Response Portal; FDR, false discovery rate; GDSC, Genomics of Drug Sensitivity in Cancer; IRGs, immune-related genes.

Immune landscape of three IRGs

Understanding the correlation between gene expression and immune characteristics is crucial for precision therapy of cancers (29). Our investigation thoroughly described the relationship between IRGs expression profiles and immune features in BC. As illustrated in Figure 7A-7C, HMGB3 expression demonstrated significantly negative relationships with the stromal score and ESTIMATE score. Additionally, CORO2A expression exhibited negative correlations with both the immune score and the ESTIMATE score. In contrast, TNFSF4 expression showed positive associations with these scores.

Figure 7 The association analysis between StromalScore, ImmuneScore, and ESTIMATEScore and expression levels of (A) HMGB3; (B) CORO2A; and (C) TNFSF4. TCGA, The Cancer Genome Atlas.

Immune cells exert a crucial influence on the tumorigenesis and progression, thereby impacting the clinical outcomes of cancer patients (30). Figure 8A reveals strong relationships between gene expressions and various immune cells. HMGB3 exhibited the strongest positive association with M0 macrophages, while showing the most inverse correlation with resting mast cells. CORO2A was most positively related to resting CD4+ memory T cells and resting mast cells, while most negatively correlated to follicular helper T cells. Notably, TNFSF4 displayed marked positive correlations with three immune subtypes (resting CD4+ memory T cells, neutrophil, and activated CD4+ memory T cells, but exhibited a robust negative association with activated NK cells.

Figure 8 Immune cell infiltration and clinical survival analysis of three IRGs expression in BC. (A) The lollipop chart shows the correlation between the three IRGs expression levels and 22 immune cells. (B) Kaplan-Meier plots of immune subsets and three IRGs expression levels based on TIMER 2.0. ns, P≥0.05; *, P<0.05; **, P<0.01; ***, P<0.001. BC, breast cancer; HR, hazard ratio; IRGs, immune-related genes; TIMER, Tumor Immune Estimation Resource.

We further analyzed the clinical correlation between expression of the three IRGs and immune subsets through the TIMER website (Figure 8B), revealing distinct correlation patterns. Patients with dual HMGB3 downregulation and CD8+ T cell deficiency exhibited poorer clinical outcomes compared to those with HMGB3-low/CD8+-high profiles (HR=0.578, P=0.008)—a pattern replicated in the CORO2A-low/CD8+-low cohort (HR=0.649, P=0.04). In the group with low HMGB3 expression, the prognosis of patients with high M2 macrophage infiltration was significantly worse compared to those with low M2 macrophage infiltration. Similarly, within both the low and high CORO2A/TNFSF4 expression groups, patients exhibiting high M2 macrophage infiltration demonstrated poorer prognostic outcomes than those with low infiltration. In a low-Tregs infiltration microenvironment, declined HMGB3/CORO2A/TNFSF4 expression levels were associated with unfavorable clinical outcomes. These findings highlight the influence of varying levels of immune cell infiltration on the prognosis of cancer patients and may suggest potential avenues for therapeutic intervention.

To investigate the immunomodulatory potential of the target genes, systematic correlations were performed between HMGB3/CORO2A/TNFSF4 expression profiles and 10 clinically relevant immune checkpoint markers (31,32), as displayed in Figure 9. High HMGB3 expression showed positive correlations with multiple checkpoint molecules, particularly the CTLA-4 and LAG-3, while demonstrating an inverse association with SIGLEC15 and ITPRIPL1 (Figure 9A). Upregulated CORO2A was negatively related to various checkpoint members, particularly CTLA4, LAG3, PDCD1, TIGIT, and ITPRIPL1, while positively related to SIGLEC15 and IGSF8 (Figure 9B). In the 10 checkpoint molecules, elevated TNFSF4 expression was associated with most genes, except for IGSF8 (Figure 9C). These results suggest a strong relationship between various immune inhibitors and gene expression.

Figure 9 Correlation analysis between three upregulated IRGs expression levels and immune checkpoints, and the benefits of immunotherapy in BC. (A-C) Grouped box plots. The x-axis represents different sample groups, and the y-axis represents the distribution of gene expression. Different colors represent different groups. (D-F) Violin plots. The y-axis represents the level of the TIDE score. Red represents high expression of three IRGs; Blue represents low expression of three IRGs. ns, P≥0.05; *, P<0.05; **, P<0.01; ***, P<0.001; ****, P<0.0001. BC, breast cancer; IRGs, immune-related genes; TIDE, Tumor Immune Dysfunctional and Exclusion.

The TIDE score was computed to further assess the immunotherapy effects of three IRGs in BC. Downregulated HMGB3 and CORO2A gained higher TIDE scores compared with upregulated HMGB3 and CORO2A groups (Figure 9D,9E). However, reduced TNFSF4 was correlated with a low score, as indicated in Figure 9F, suggesting a perhaps favorable immune response. In summary, TIDE evaluation may reveal better immunotherapeutic responsiveness in cases exhibiting elevated HMGB3/CORO2A expression compared to those with upregulated TNFSF4 levels.

Discussion

BC continues as a principal contributor to cancer-associated mortality among women (33). The advent of immunotherapy has brought hope to BC patients, leading to changes in treatment patterns for some BC subtypes (9,34). For example, in one clinical trial (NCT03197935), atezolizumab-based combination chemotherapy in early-stage TNBC demonstrated notably elevated pathological complete response (pCR) rates compared with placebo and chemotherapy subgroups (pCR from 41% to 59%) (35). This therapeutic breakthrough extends beyond TNBC, with accumulating trials investigating checkpoint inhibitors in hormone receptor (HR)-positive and HER2-positive BC (36). Despite advancements in immunotherapy, the great heterogeneity of this disease and the lack of robust biomarkers to predict response cause a critical challenge for therapeutic management (37). Our investigation developed an immune-related six-gene prognostic model for BC, stratifying patients into distinct prognostic cohorts and predicting immunotherapy sensitivity.

The prognostic model in our research integrated both risk (HMGB3, CORO2A, TNFSF4) and protective (TACR1, SOCS3, FREM1) genes, each with distinct roles in tumor biology. In our research, we focused on the three risk genes, aiming to explore their influence on BC. HMGB3, a chromatin-binding protein, plays an important role in multiple cellular processes, including DNA-repair, replication, recombination, and transcription (38). Accumulating evidence has proved that HMGB3 is overexpressed and related to poor survival in various types of cancer, including ovarian cancer (39), neuroblastoma (40), colorectal cancer (41), and BC. Our data consistently emphasized the potential importance of HMGB3 in BC. The expression of HMGB3 was significantly upregulated in the staging progression of BC, and the total protein level was also significantly increased in BC tissues. Interestingly, our results suggest that the correlation between HMGB3 expression and the prognosis of BC patients may be time-dependent. Specifically, 150 months ago, high expression of HMGB3 was associated with a shortened survival, but after that, this relationship was not absolute. However, this correlation still needs to be verified among long-term survivors. The silencing of HMGB3 can inhibit cell proliferation in vitro and tumor growth in vivo, which suggests its potential as a therapeutic target (20). Besides, elevated HMGB3 was related to low TIDE score, indicating a potential predictive value of immunotherapy. These results may also offer a new perspective for the future development of drugs for BC patients. CORO2A encodes a member of the WD repeat protein family, involved in a variety of cellular processes, including cell cycle progression, signal transduction, apoptosis, and gene regulation. CORO2A was statistically up-regulated in oral squamous cell carcinoma (OSCC) tissues, predicting 5-year survival in patients with OSCC (P=0.0203). In addition, knocking down CORO2A induced BC cell cycle stagnation in the G0/G1 phase, thereby inhibiting cell migration (23). TNFSF4, a transmembrane type II glycoprotein, is predominantly expressed on antigen-presenting cells (APCs) (42). It has been confirmed as a contributor to the progression of hepatocellular Carcinoma (21). Notably, diminished TNFSF4 expression correlated with adverse clinical outcomes in melanoma cohorts (22), contrasting with its beneficial prognostic association observed in BC cases, highlighting its context-dependent regulatory functions across malignancies.

Deciphering the interplay between gene expression and immune infiltration is essential for developing immunotherapy strategies. Tumor-associated macrophages (TAMs), important immune cell types of TME, are divided into M1 macrophages and M2 macrophages (43). M1 macrophages, often known for antitumor functions, are currently being considered as a potential strategy to treat cancers (44), like colorectal cancer (45) and BC (46). Increasing studies have revealed that M2-polarized phenotype can be induced by various inflammatory factors, promoting the progression and metastasis of multiple tumors, including gastric cancer (47), non-small cell lung cancer (48), and colorectal cancer. Among the numerous immune cells in the TME of BC, CD8+ T cells are key cytotoxic factors targeting cancer cells, and their presence is associated with improved clinical outcomes (49). Immune profiling revealed marked positive correlations of CORO2A/TNFSF4 with M2 macrophages infiltration, inversely associated with CD8+ T cells accumulation. Besides, BC patients with upregulated CORO2A/TNFSF4 combined with an elevated M2-polarized phenotype showed poor outcomes. However, reduced CORO2A/TNFSF4 combined with high CD8+ T cells displayed favorable outcomes. In conclusion, the results of the correlation analysis between immune subsets and BC survival potentially guide treatment. HMGB3/CORO2A/TNFSF4 showed close correlations with multiple immune checkpoints. Moreover, the three genes were also strongly associated with the TIDE score. These results indicate their potential as therapeutic targets, which may bring new hope to BC patients. More rigorous experiments will be needed in the future to verify their therapeutic value.

However, our study has certain limitations. The relatively small sample size in the validation cohort may limit the generalizability of our findings. In addition, to gain a more comprehensive understanding of the mechanisms through which the three identified “risk genes” contribute to BC development, further basic experimental research is warranted.

Conclusions

First, we developed a prognostic model based on IGRs to predict the survival outcomes of BC patients, aiming to provide potential guidance for clinical treatment strategies. Second, we conducted a preliminary investigation into the impact of the three identified “risk genes” on BC progression. Nevertheless, our findings are currently based solely on public database analyses, and more rigorous experimental validations will be required in future studies.

Acknowledgments

None.

Footnote

Reporting Checklist: The authors have completed the TRIPOD reporting checklist. Available at https://tcr.amegroups.com/article/view/10.21037/tcr-2025-1049/rc

Peer Review File: Available at https://tcr.amegroups.com/article/view/10.21037/tcr-2025-1049/prf

Funding: This work was supported by Project of Natural Science Foundation of Sichuan Province (No. 2023NSFSC0700).

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://tcr.amegroups.com/article/view/10.21037/tcr-2025-1049/coif). The authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Loibl S, Poortmans P, Morrow M, et al. Breast cancer. Lancet 2021;397:1750-69. [Crossref] [PubMed]
2. Siegel RL, Kratzer TB, Giaquinto AN, et al. Cancer statistics, 2025. CA Cancer J Clin 2025;75:10-45. [Crossref] [PubMed]
3. Bray F, Laversanne M, Sung H, et al. Global cancer statistics 2022: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin 2024;74:229-63. [Crossref] [PubMed]
4. Ramin C, Veiga LHS, Vo JB, et al. Risk of second primary cancer among women in the Kaiser Permanente Breast Cancer Survivors Cohort. Breast Cancer Res 2023;25:50. [Crossref] [PubMed]
5. Kerr AJ, Dodwell D, McGale P, et al. Adjuvant and neoadjuvant breast cancer treatments: A systematic review of their effects on mortality. Cancer Treat Rev 2022;105:102375. [Crossref] [PubMed]
6. Geels SN, Moshensky A, Sousa RS, et al. Interruption of the intratumor CD8(+) T cell:Treg crosstalk improves the efficacy of PD-1 immunotherapy. Cancer Cell 2024;42:1051-1066.e7. [Crossref] [PubMed]
7. Kang J, Zhang C, Zhong WZ. Neoadjuvant immunotherapy for non-small cell lung cancer: State of the art. Cancer Commun (Lond) 2021;41:287-302. [Crossref] [PubMed]
8. Liu Y, Xun Z, Ma K, et al. Identification of a tumour immune barrier in the HCC microenvironment that determines the efficacy of immunotherapy. J Hepatol 2023;78:770-82. [Crossref] [PubMed]
9. Liu Y, Hu Y, Xue J, et al. Advances in immunotherapy for triple-negative breast cancer. Mol Cancer 2023;22:145. [Crossref] [PubMed]
10. Emens LA, Cruz C, Eder JP, et al. Long-term Clinical Outcomes and Biomarker Analyses of Atezolizumab Therapy for Patients With Metastatic Triple-Negative Breast Cancer: A Phase 1 Study. JAMA Oncol 2019;5:74-82. [Crossref] [PubMed]
11. Bianchini G, De Angelis C, Licata L, et al. Treatment landscape of triple-negative breast cancer - expanded options, evolving needs. Nat Rev Clin Oncol 2022;19:91-113. [Crossref] [PubMed]
12. Yao Y, Kong X, Liu R, et al. Development of a Novel Immune-Related Gene Prognostic Index for Breast Cancer. Front Immunol 2022;13:845093. [Crossref] [PubMed]
13. Chen F, Zhang Y, Chandrashekar DS, et al. Global impact of somatic structural variation on the cancer proteome. Nat Commun 2023;14:5637. [Crossref] [PubMed]
14. Digre A, Lindskog C. The human protein atlas-Integrated omics for single cell mapping of the human proteome. Protein Sci 2023;32:e4562. [Crossref] [PubMed]
15. Liu CJ, Hu FF, Xie GY, et al. GSCA: an integrated platform for gene set cancer analysis at genomic, pharmacogenomic and immunogenomic levels. Brief Bioinform 2023;24:bbac558. [Crossref] [PubMed]
16. Li C, Tang Z, Zhang W, et al. GEPIA2021: integrating multiple deconvolution-based analysis into GEPIA. Nucleic Acids Res 2021;49:W242-6. [Crossref] [PubMed]
17. Li T, Fu J, Zeng Z, et al. TIMER2.0 for analysis of tumor-infiltrating immune cells. Nucleic Acids Res 2020;48:W509-14. [Crossref] [PubMed]
18. Han Y, Wang Y, Dong X, et al. TISCH2: expanded datasets and new tools for single-cell transcriptome analyses of the tumor microenvironment. Nucleic Acids Res 2023;51:D1425-31. [Crossref] [PubMed]
19. Jiang P, Gu S, Pan D, et al. Signatures of T cell dysfunction and exclusion predict cancer immunotherapy response. Nat Med 2018;24:1550-8. [Crossref] [PubMed]
20. Gu J, Xu T, Huang QH, et al. HMGB3 silence inhibits breast cancer cell proliferation and tumor growth by interacting with hypoxia-inducible factor 1α. Cancer Manag Res 2019;11:5075-89. [Crossref] [PubMed]
21. Deng Z, Li L, Meng Z, et al. Pan-cancer analysis shows that TNFSF4 is a potential prognostic and immunotherapeutic biomarker for multiple cancer types including liver cancer. BMC Cancer 2025;25:100. [Crossref] [PubMed]
22. Roszik J, Markovits E, Dobosz P, et al. TNFSF4 (OX40L) expression and survival in locally advanced and metastatic melanoma. Cancer Immunol Immunother 2019;68:1493-500. [Crossref] [PubMed]
23. Kase-Kato I, Asai S, Minemura C, et al. Molecular Pathogenesis of the Coronin Family: CORO2A Facilitates Migration and Invasion Abilities in Oral Squamous Cell Carcinoma. Int J Mol Sci 2021;22:12684. [Crossref] [PubMed]
24. Deng JL, Zhang HB, Zeng Y, et al. Effects of CORO2A on Cell Migration and Proliferation and Its Potential Regulatory Network in Breast Cancer. Front Oncol 2020;10:916. [Crossref] [PubMed]
25. Sasi W, Jiang WG, Sharma A, et al. Higher expression levels of SOCS 1,3,4,7 are associated with earlier tumour stage and better clinical outcome in human breast cancer. BMC Cancer 2010;10:178. [Crossref] [PubMed]
26. Padmanaban V, Keller I, Seltzer ES, et al. Neuronal substance P drives metastasis through an extracellular RNA-TLR7 axis. Nature 2024;633:207-15. [Crossref] [PubMed]
27. Li HN, Li XR, Lv ZT, et al. Elevated expression of FREM1 in breast cancer indicates favorable prognosis and high-level immune infiltration status. Cancer Med 2020;9:9554-70. [Crossref] [PubMed]
28. Gao L, Wu ZX, Assaraf YG, et al. Overcoming anti-cancer drug resistance via restoration of tumor suppressor gene function. Drug Resist Updat 2021;57:100770. [Crossref] [PubMed]
29. Thorsson V, Gibbs DL, Brown SD, et al. The Immune Landscape of Cancer. Immunity 2018;48:812-830.e14. [Crossref] [PubMed]
30. Zhang Y, Zhang Z. The history and advances in cancer immunotherapy: understanding the characteristics of tumor-infiltrating immune cells and their therapeutic implications. Cell Mol Immunol 2020;17:807-21. [Crossref] [PubMed]
31. Deng S, Zhang Y, Wang H, et al. ITPRIPL1 binds CD3ε to impede T cell activation and enable tumor immune evasion. Cell 2024;187:2305-2323.e33. [Crossref] [PubMed]
32. Yi L, Wu G, Guo L, et al. Comprehensive Analysis of the PD-L1 and Immune Infiltrates of m(6)A RNA Methylation Regulators in Head and Neck Squamous Cell Carcinoma. Mol Ther Nucleic Acids 2020;21:299-314. [Crossref] [PubMed]
33. Xiong X, Zheng LW, Ding Y, et al. Breast cancer: pathogenesis and treatments. Signal Transduct Target Ther 2025;10:49. [Crossref] [PubMed]
34. Zhu S, Wu Y, Song B, et al. Recent advances in targeted strategies for triple-negative breast cancer. J Hematol Oncol 2023;16:100. [Crossref] [PubMed]
35. Jin M, Fang J, Peng J, et al. PD-1/PD-L1 immune checkpoint blockade in breast cancer: research insights and sensitization strategies. Mol Cancer 2024;23:266. [Crossref] [PubMed]
36. Heater NK, Warrior S, Lu J. Current and future immunotherapy for breast cancer. J Hematol Oncol 2024;17:131. [Crossref] [PubMed]
37. Ye F, Dewanjee S, Li Y, et al. Advancements in clinical aspects of targeted therapy and immunotherapy in breast cancer. Mol Cancer 2023;22:105. [Crossref] [PubMed]
38. Chikhirzhina E, Tsimokha A, Tomilin AN, et al. Structure and Functions of HMGB3 Protein. Int J Mol Sci 2024;25:7656. [Crossref] [PubMed]
39. Ma H, Qi G, Han F, et al. HMGB3 promotes the malignant phenotypes and stemness of epithelial ovarian cancer through the MAPK/ERK signaling pathway. Cell Commun Signal 2023;21:144. [Crossref] [PubMed]
40. Zhong X, Zhang S, Zhang Y, et al. HMGB3 is Associated With an Unfavorable Prognosis of Neuroblastoma and Promotes Tumor Progression by Mediating TPX2. Front Cell Dev Biol 2021;9:769547. [Crossref] [PubMed]
41. Gong W, Guo Y, Yuan H, et al. HMGB3 is a Potential Therapeutic Target by Affecting the Migration and Proliferation of Colorectal Cancer. Front Cell Dev Biol 2022;10:891482. [Crossref] [PubMed]
42. Dannull J, Nair S, Su Z, et al. Enhancing the immunostimulatory function of dendritic cells by transfection with mRNA encoding OX40 ligand. Blood 2005;105:3206-13. [Crossref] [PubMed]
43. Xiang X, Wang J, Lu D, et al. Targeting tumor-associated macrophages to synergize tumor immunotherapy. Signal Transduct Target Ther 2021;6:75. [Crossref] [PubMed]
44. Guo L, Cheng X, Chen H, et al. Induction of breast cancer stem cells by M1 macrophages through Lin-28B-let-7-HMGA2 axis. Cancer Lett 2019;452:213-25. [Crossref] [PubMed]
45. Han Y, Sun J, Yang Y, et al. TMP195 Exerts Antitumor Effects on Colorectal Cancer by Promoting M1 Macrophages Polarization. Int J Biol Sci 2022;18:5653-66. [Crossref] [PubMed]
46. Nilkhet S, Mongkolpobsin K, Sillapachaiyaporn C, et al. M1 macrophages polarized by crude polysaccharides isolated from Auricularia polytricha exhibit anti-tumor effect on human breast cancer cells. Sci Rep 2024;14:8179. [Crossref] [PubMed]
47. Deng C, Huo M, Chu H, et al. Exosome circATP8A1 induces macrophage M2 polarization by regulating the miR-1-3p/STAT6 axis to promote gastric cancer progression. Mol Cancer 2024;23:49. [Crossref] [PubMed]
48. Yao Y, Chen C, Wang J, et al. Circular RNA circATP9A promotes non-small cell lung cancer progression by interacting with HuR and by promoting extracellular vesicles-mediated macrophage M2 polarization. J Exp Clin Cancer Res 2023;42:330. [Crossref] [PubMed]
49. Liang Y, Li J, Yuan Y, et al. Exosomal miR-106a-5p from highly metastatic colorectal cancer cells drives liver metastasis by inducing macrophage M2 polarization in the tumor microenvironment. J Exp Clin Cancer Res 2024;43:281. [Crossref] [PubMed]