Prognostic significance and pathological correlation analysis of DKK4 in colorectal cancer

Introduction

Colorectal tumors primarily encompass colorectal cancer (CRC) and colorectal adenoma (CRA) (1). As one of the most prevalent and lethal malignancies of the digestive system worldwide, CRC typically develops through a multi-step progression from normal mucosa-adenoma to precancerous lesion and finally invasive carcinoma (2). Notably, CRAs account for over 85% of CRC precancerous lesions, with malignant transformation risk strongly associated with adenoma size, histological type, and degree of dysplasia (3,4). The hallmarks of colorectal intraepithelial neoplasia include cellular and architectural atypia, with approximately 80% of CRCs arising from adenoma malignant transformation (5). Current diagnosis of dysplastic adenomas relies heavily on colonoscopy and histopathological biopsy (6). However, early precancerous lesions often present asymptomatically and may evade endoscopic detection, particularly flat or diminutive lesions. Conventional endoscopic techniques (e.g., white-light endoscopy) exhibit limited sensitivity (60–85%) and specificity (70–90%) for early lesions, with diagnostic accuracy significantly influenced by operator experience (7). Pathological interpretation similarly suffers from inter-observer variability in assessing precancerous changes.

Early detection of CRC precursors could reduce CRC incidence by 40–60% and mortality by 29–52% (8). Therefore, the identification of reliable molecular biomarkers for detecting these lesions is of paramount importance for early diagnosis, prognostic evaluation, and personalized treatment strategies to improve the early detection rate of CRAs. Previous screening of molecular biomarkers has primarily focused on differentially expressed genes (DEGs) between CRC and adjacent tissues, while neglecting precancerous lesions. If specific biomarkers could be identified from precancerous lesions, would they help improve the detection rate of colorectal precancerous lesions and enhance the preventive effect against CRC? Furthermore, mechanistic studies of these biomarkers could assist in improving the effective clinical and prognostic management of CRC across multiple aspects, including disease prevention, molecular typing, and personalized therapy.

Based on this, our study utilized public database mining to identify DEGs between colorectal precancerous lesions and healthy populations, establishing Dickkopf-related protein 4 (DKK4) as a candidate target. Through single-gene functional analysis, we predicted DKK4’s biological functions, associated signaling pathways, and immunoregulatory roles. Using least absolute shrinkage and selection operator (LASSO) regression, we constructed a DKK4-based risk prediction model for CRC, validated this predictive model in online datasets with LASSO, and confirmed DKK4 expression in CRC and precancerous lesions through immunohistochemistry. This research explores the correlation between DKK4 and CRC, aiming to identify potential molecular markers for colorectal precancerous lesions and provide new insights into the development and progression of CRC. We present this article in accordance with the TRIPOD reporting checklist (available at https://tcr.amegroups.com/article/view/10.21037/tcr-2026-1-0021/rc).

Methods

Study subjects

Patients who underwent colonoscopy at General Hospital of Ningxia Medical University from May 2024 to December 2024 were enrolled in this study. All colonoscopic examinations were performed by attending physicians or higher-ranking endoscopists with more than 5 years of colonoscopy experience. This study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. This study was approved by the Ethics Committee of General Hospital of Ningxia Medical University (No. KYLL-2023-1429), and written informed consent was obtained from all participating patients prior to their procedures.

Inclusion and exclusion criteria

Pathological diagnosis was classified into adenoma and adenocarcinoma according to the Diagnostic Criteria for Digestive System Tumors (1). Adenomas were further categorized based on tumor cell atypia: low-grade intraepithelial neoplasia (LGIN) indicated mild or moderate dysplasia with relatively mild cellular and structural atypia, while high-grade intraepithelial neoplasia (HGIN) was equivalent to severe dysplasia or carcinoma in situ, showing significant cellular and structural atypia (5). Inclusion criteria: CRA group: patients diagnosed with CRA by colonoscopy and confirmed by pathological examination. CRC group: patients with CRC detected by colonoscopy and pathologically confirmed as advanced CRC. Normal colorectal epithelium group: Subjects with no abnormalities in colorectal mucosa by colonoscopy, and without other digestive system diseases such as inflammatory bowel disease, celiac disease, intestinal infection, chronic pancreatitis, or portal hypertension. All patients had complete clinical case data. Exclusion criteria: patients unable to cooperate due to poor cardiopulmonary function; patients who had received chemotherapy, radiotherapy, or immunotherapy before surgery; patients with significant abnormalities in the digestive tract by endoscopy; patients with a history of CRA resection or digestive system tumors; patients with a Boston Bowel Preparation Scale score <6. All tissue specimens underwent routine hematoxylin and eosin (H&E) staining for pathological examination. According to pathological classification, patients were divided into normal colorectal epithelium group, LGIN group, HGIN group, and CRC group (5).

Data mining of DKK4 in CRC-related public databases

Downloaded raw datasets from NCBI-GEO: (I) GSE41657: 12 normal intestinal tissues, 21 LGIN, 30 HGIN, and 25 invasive CRCs (9). (II) GSE37364: 38 normal colorectal epithelium samples, 16 LGIN, and 13 HGIN (10). All sequencing data that passed quality control were analyzed using edgeR to perform differential expression analysis. For the initial screening, the following criteria were set: a fold change ≥2 with statistical significance (P<0.05); the secondary screening was based on biological characteristics including immune infiltration, migration, proliferation and other features.

The study describes the inclusion of 603 samples from TCGA (https://portal.gdc.cancer.gov) (11). Cluster heatmaps, ROC curves and survival analysis plots were generated using software tools from the OmicStudio platform (https://www.omicstudio.cn). DKK4-related mRNAs were used as input for GO functional enrichment analysis of biological functions using the DAVID database (https://david.ncifcrf.gov/) (12). Pathway classification studies were performed using the KEGG database (13). Single gene enrichment analysis of target genes was conducted using the GSEA database (https://www.gsea-msigdb.org/), and immune pathway correlation analysis was performed using the Gene Set Variation Analysis (GSVA) method (14). Immune infiltration analysis of target genes was performed using the CIBERSORT package (15). Using TCGA clinical data as input, univariate and multivariate Cox regression analyses of target genes were performed using edgeR (16). After identifying independent risk genes, nomograms were constructed to predict patient survival. The GSE39582 dataset was selected as the validation set to analyze the predictive performance of the risk model.

Risk prediction model for DKK4 in CRC survival and prognosis based on LASSO model

The LASSO regression model was constructed with ten-fold cross-validation to determine the optimal λ value for selecting candidate risk-prediction genes. Risk scores were calculated using the identified candidate genes, followed by generation of risk-associated plots (ggrisk). Patients were stratified into high- and low-risk groups based on median risk score cutoff values. Time-dependent receiver operating characteristic (ROC) curves were plotted to evaluate 1-, 3-, and 5-year survival predictions in both training and validation cohorts. Univariate and multivariate Cox regression analyses were performed to identify independent risk factors. A prognostic nomogram incorporating these risk factors and patient age was subsequently developed to predict survival outcomes.

H&E staining

Paraffin-embedded tissue sections were sequentially dewaxed in xylene and rehydrated through graded ethanol series (100% to 75%). Sections were immersed in pre-treatment solution for 1 minute at constant temperature. Nuclear staining was performed using modified Harris hematoxylin (Sigma-Aldrich) for 3–5 minutes, followed by differentiation in acid alcohol, blueing in 1% ammonia water, and rinsing with deionized water. Cytoplasmic staining was achieved with eosin for 15 seconds after dehydration in 95% ethanol. Complete dehydration was accomplished through ethanol, n-butanol, and xylene series before mounting. Whole-slide imaging was performed using Pannoramic 250 Flash digital slide scanner (3DHISTECH, Hungary), with nuclear and cytoplasmic staining appearing blue and red, respectively.

Immunohistochemistry

Paraffin-embedded sections were sequentially dewaxed and rehydrated through graded ethanol series (100% to 75%). Antigen retrieval was performed using EDTA buffer with microwave treatment. Non-specific binding sites were blocked with 3% BSA. Sections were incubated overnight with primary antibody against DKK4 (Proteintech, Cat No. 27080-1-AP, rabbit polyclonal, 1:100 dilution), followed by secondary antibody (1:200) incubation. DAB substrate was applied for chromogenic development, with reaction termination upon observation of brown-yellow positive signals under microscopic monitoring. Counterstaining was performed using modified Harris hematoxylin (Sigma-Aldrich), followed by acid alcohol differentiation, bluing in 1% ammonia water, and deionized water rinsing. Sections were dehydrated through ethanol series, cleared in n-butanol and xylene, and scanned using Pannoramic 250 Flash whole slide scanner (3DHISTECH, Hungary), with nuclei appearing blue and DKK4-positive staining exhibiting brown-yellow coloration. Positive rate scoring: ≤1%: 0 points, 2–25%: 1 point, 26–50%: 2 points, 51–75%: 3 points, 75%: 4 points. Staining intensity scoring: colorless 0 points, pale yellow 1 point, yellow 2 points, brown-yellow 3 points. Total score = positive rate × intensity: 0 negative (−), 1–4 weakly positive (+), 5–8 positive (++), 9–12 strongly positive (+++).

Statistical analysis

Statistical analyses were performed using SPSS 25.0 and GraphPad Prism 9.0 software. Normally distributed data were analyzed using Student’s t-test for comparisons between two groups or one-way ANOVA for multiple group comparisons. Non-normally distributed data were analyzed using the Mann-Whitney U test. All data are presented as mean ± standard deviation. A P value <0.05 was considered statistically significant.

Differential expression and functional prediction of DKK4 in colorectal tumors

Using the GSE37364 dataset, we compared high-grade intraepithelial neoplasia (target group) against other specimens (control group). With FC >2 and P<0.05 as thresholds, we identified 1,752 DEGs related to tumorigenesis, proliferation, migration, differentiation, and immunity. Differential expression and functional prediction of DKK4 in colorectal tumors as shown in Figure 1. Top 100 DEGs were visualized in a cluster heatmap, with the most significant 10 DEGs highlighted in the volcano plot (Figure 1A,1B). Literature review identified DKK4 as our candidate target.

Figure 1 Differential distribution and functional prediction of DKK4 in colorectal adenoma and carcinoma tissues. (A) Cluster heatmap analysis of the top 100 differentially expressed mRNAs in colorectal adenoma with intraepithelial neoplasia from the GSE41657 dataset (selected differences). (B) Scatter plot analysis of differentially expressed mRNAs in colorectal adenoma with intraepithelial neoplasia; the top 5 upregulated genes are labeled in red, and the top 5 downregulated genes are labeled in blue. (C) Top 15 enriched upregulated and downregulated genes between high- and low-DKK4 expression groups. (D) TCGA data analysis of DKK4 expression differences between colon cancer/adjacent normal tissues and rectal cancer/adjacent normal tissues. (E) Single-gene GSEA enrichment analysis of DKK4. (F) Regulatory effect of DKK4 on key molecules of the Wnt signaling pathway. DEGs, differentially expressed genes; DKK4, Dickkopf-related protein 4; FDR, false discovery rate; GSEA, gene set enrichment analysis; TCGA, The Cancer Genome Atlas; TPM, transcripts per million.

Analysis of TCGA data (275 colon cancer cases vs. 349 controls; 92 rectal cancer cases vs. 318 controls) demonstrated significantly elevated DKK4 expression in both colon and rectal cancer patients compared to adjacent normal tissues (P<0.05, Figure 1D). Clinical data analysis revealed a significant association between DKK4 expression levels and tumor mutational burden (TMB) in CRC patients (P=0.008, Table 1). Differential gene enrichment analysis was performed after stratifying patients into high- and low-expression groups based on median DKK4 expression levels. Cluster and volcano plots of the top 15 upregulated/downregulated genes were generated (Figure 1C). Gene set enrichment analysis (GSEA) and KEGG pathway analysis of the top 30 DEGs showed significant enrichment of DKK4 in the Wnt signaling pathway (Figure 1E,1F). These findings suggest that DKK4 is strongly associated with abnormal tumor proliferation and invasion, and its differential expression may influence the efficacy of tumor immunotherapy.

DKK4 correlation with immune cells and function

As shown in Figure 2, DKK4 is correlation with immune cells and function. GSVA and CIBERSORT analyses of TCGA data revealed significant correlations between DKK4 expression and CD4+ T cells as well as macrophages (Figure 2A,2B). Integration of expression abundance and correlation coefficients demonstrated DKK4’s predominant involvement in immune cell interactions, particularly among macrophages, T cells, dendritic cells, NK cells, and B cells (Figure 2C). Stratification by DKK4 expression levels (high vs. low) showed differential DKK4 expression primarily enriched in CD4+ T cells, NK cells, and macrophages (Figure 2D). Further CIBERSORT analysis of tumor immune pathway gene expression profiles identified significant enrichment of DKK4 in macrophages, with associated genes including MMP9, TM4SF19, CD68, and CYBB (Figure 2E,2F). Immunological functional analysis revealed that DEGs between high- and low-DKK4 groups in CRC patients were primarily clustered in immune processes such as antigen presentation, cytolytic activity, and type I/II interferon responses (Figure 2G). Focusing on immune regulatory genes, DKK4 expression levels significantly influenced both immunostimulatory genes (CD80, ENTPD1, IFNA2, IFNG, PRF1; Figure 2H,2I) and immunosuppressive genes (ARG1, CD274, EDNRB, VEGFB; Figure 2J). Further quantitative analysis of immune functions showed that DKK4 positively regulated both immune stimulatory and immune inhibitory genes, with a significantly stronger correlation with immune inhibitory genes CD274 (r=0.35, P<0.001) and ARG1 (r=0.30, P<0.001) than with immune stimulatory genes IFNG (r=0.31, P<0.001) and CD80 (r=0.28, P<0.001). Additionally, CIBERSORT analysis confirmed that the infiltration proportion of M2-type macrophages was significantly increased (P<0.05), while the infiltration proportion of CD4+ effector T cells was significantly decreased (P<0.05) in tumor tissues of the high DKK4 expression group. These findings suggest that DKK4 can drive the transformation of the colorectal tumor microenvironment toward an immune-suppressive state through the biased regulation of immune-related genes and immune cell polarization.

Figure 2 Correlation between DKK4 and immune function. (A) Correlation analysis of DKK4 expression with immune cell infiltration using TCGA data. (B) Distribution of DKK4 expression across 22 immune cell types (CIBERSORT analysis). (C) Interaction network analysis of DKK4 with immune cell populations. (D) Boxplot showing differential DKK4 expression in 22 immune cell subtypes. (E,F) Differential expression of tumor immunity-related genes (7 immune cell types) between high- and low-DKK4 groups (boxplot). (G) Boxplots showing immune response-related gene expression (antigen presentation, cytolytic activity, IFN-I/II response) stratified by DKK4 expression. (H,I) Differential expression of tumor-promoting immune genes between DKK4 expression groups (boxplot). (J) Differential expression of immunosuppressive genes between DKK4 expression groups (boxplot). *, P<0.05; **, P<0.01; ***, P<0.001; ****, P<0.0001. CIBERSORT, Cell-type Identification By Estimating Relative Subsets Of RNA Transcripts; DKK4, Dickkopf-related protein 4; TCGA, The Cancer Genome Atlas.

DKK4-based prognostic model construction using LASSO regression

Transcriptomic data from 603 tumor samples with survival outcomes were extracted from TCGA database to establish a LASSO regression model (Figure 3). Ten-fold cross-validation was performed to determine the optimal λ value (Figure 3A,3B). The optimal λ value was identified as 0.00413, which selected 12 candidate genes (including DKK4). Risk scores were calculated using the regression coefficients of these 12 genes, and risk association plots (ggrisk) demonstrated that increasing risk scores correlated with decreased overall survival and increased mortality in CRC patients (Figure 3C,3D). Univariate and multivariate analyses of clinical parameters (age, sex, TNM stage, AJCC grade, venous/lymphatic invasion) revealed AJCC grade and risk score as key independent prognostic factors (Figure 3E,3F). The 12-gene signature (including DKK4) successfully constructed a prognostic risk model for CRC survival.

Figure 3 LASSO regression model of DKK4 in colorectal cancer. (A) Ten-fold cross-validation plot of 603 colorectal cancer transcriptome samples with survival outcomes from TCGA. (B) Variation in feature coefficients of differentially expressed genes across λ values for risk assessment gene selection. (C,D) Risk factor association plots (ggrisk) integrating risk scores and patient survival status to evaluate prognostic correlation. (E,F) Univariate and multivariate Cox regression analyses of clinical factors (age, sex, TNM stage, AJCC grade, venous/lymphatic invasion) and risk scores on patient survival. AJCC, American Joint Committee on Cancer; CI, confidence interval; DKK4, Dickkopf-related protein 4; HR, hazard ratio; LASSO, least absolute shrinkage and selection operator; TCGA, The Cancer Genome Atlas; TNM, Tumor, Node, Metastasis.

Validation of the DKK4-based prognostic model

Using the median risk score as cutoff, ROC curves were generated to evaluate 1-, 3-, and 5-year survival probabilities in 603 patients (Figure 4). The high-risk group showed significantly poorer survival compared to low-risk patients (P<0.001, Figure 4A). Time-dependent AUC values demonstrated good predictive performance for 1-year (AUC =0.676), 3-year (AUC =0.695), and 5-year (AUC =0.658) survival (Figure 4B-4D). Univariate and multivariate Cox regression analyses identified independent risk factors, which were incorporated with patient age to construct a prognostic nomogram. Calibration curves indicated better predictive accuracy for 1- and 3-year survival compared to 5-year outcomes (Figure 4E,4F). These results suggest the DKK4-based risk model is more suitable for short-term (1-3 year) than long-term (5-year) survival prediction in CRC patients.

Figure 4 DKK4-based prognostic risk prediction model for colorectal cancer patients. (A) Kaplan-Meier survival curves comparing high- and low-risk groups stratified by risk scores in 603 colorectal cancer patients. (B-D) Time-dependent ROC curves showing predictive accuracy of risk scores for 1-, 3-, and 5-year survival. (E) Prognostic nomogram integrating risk scores with clinical characteristics. (F) Calibration curves assessing nomogram performance at 1, 3, and 5 years. AJCC, American Joint Committee on Cancer; AUC, area under the curve; DKK4, Dickkopf-related protein 4; OS, overall survival; ROC, receiver operating characteristic.

External validation using GSE39582 dataset

A CRC transcriptomic dataset (GSE39582) was randomly selected from public databases for model validation (Figure 5). Risk association plots demonstrated an inverse correlation between risk scores and overall survival, with increasing mortality observed at higher risk levels (Figure 5A,5B). Survival curve analysis using the model’s risk cutoff revealed significantly poorer outcomes in high-risk versus low-risk patients (P=0.0045, Figure 5C). Time-dependent ROC analysis showed satisfactory 1-year predictive accuracy (AUC =0.634), but declining performance for 3-year (AUC =0.596) and 5-year (AUC =0.561) survival (Figure 5D-5F). Nomogram and calibration curve analyses confirmed better predictive capability for 1–3-year compared to 5-year outcomes (Figure 5G,5H). These findings further validate the DKK4-based model’s superior performance for short-term (1–3-year) versus long-term (5-year) survival prediction in CRC patients.

Figure 5 Validation of the risk prediction model in independent cohorts. (A,B) Risk factor association plots (ggrisk) validating the prediction model in the GSE39582 dataset. (C) Kaplan-Meier survival analysis of colorectal cancer patients stratified by risk scores in the GSE39582 validation cohort. (D-F) Time-dependent ROC curves demonstrating 1-, 3-, and 5-year survival prediction accuracy in the validation set. (G) Prognostic nomogram incorporating risk scores and clinical parameters from GSE39582 dataset. (H) Calibration curves evaluating nomogram performance at 1, 3, and 5 years in the validation cohort. AJCC, American Joint Committee on Cancer; AUC, area under the curve; OS, overall survival; ROC, receiver operating characteristic.

DKK4 expression in CRA-carcinoma sequence

Eighty patients meeting inclusion criteria were enrolled (19 normal colorectal epithelium, 31 LGIN, 9 HGIN, and 21 invasive carcinomas). Histopathological analysis revealed: Normal mucosa exhibited single-layer columnar epithelium with goblet cells and densely packed, straight tubular glands in the lamina propria (Figure 6). Immunohistochemistry showed minimal DKK4 expression in mucosal epithelium and glandular margins (Figure 6A,6C). LGIN specimens demonstrated hyperchromasia, increased glandular density, goblet cell proliferation, luminal basophilic material, edema, and lymphocyte infiltration. DKK4 expression was diffusely elevated, particularly within glandular structures (Figure 6B,6D). HGIN lesions displayed marked cellular basophilia, high nuclear-cytoplasmic ratios, elongated nuclei, glandular crowding, and intraluminal tumor cells. DKK4 expression was markedly increased and predominantly localized to glandular elements (Figure 6E,6G). Invasive carcinomas showed cellular pleomorphism, glandular architecture with necrotic debris, and mixed inflammatory infiltrates. DKK4 expression was significantly reduced (Figure 6F,6H). Statistical analysis of immunohistochemical scores revealed: Significant DKK4 differences between adenomas (LGIN/HGIN) and normal/carcinoma tissues (P<0.05). Marked DKK4 elevation in HGIN versus all other groups (P<0.05) (Table 2). These findings suggest that DKK4 is not continuously highly expressed throughout the initiation and progression of CRC, but instead plays a core regulatory role in the precancerous stage of adenoma-carcinoma progression. Its high expression in adenomas is closely associated with abnormal proliferation and structural atypia of glandular epithelium, whereas its downregulation in invasive carcinoma may be related to the destruction of acinar structures and phenotypic alterations of epithelial cells during malignant transformation. This unique expression profile renders DKK4 a potential specific biomarker for the screening of precancerous lesions of CRC, compensating for the deficiency of existing biomarkers that mostly target cancerous tissues and are insufficient for identifying precancerous lesions, and provides a novel molecular target for the early screening and intervention of CRC.

Figure 6 Expression of DKK4 in colorectal cancer tissues. (A,C) H&E staining and DKK4 IHC analysis of normal colorectal tissues (5×, 50×). (B,D) H&E staining and DKK4 IHC analysis of adenoma with low-grade intraepithelial neoplasia (5×, 50×). (E,G) H&E staining and DKK4 IHC analysis of adenoma with high-grade intraepithelial neoplasia (5×, 50×). (F,H) H&E staining and DKK4 IHC analysis of colorectal carcinoma (5×, 50×). DKK4, Dickkopf-related protein 4; H&E, hematoxylin and eosin; IHC, immunohistochemistry.

Discussion

CRAs are benign tumors that develop on the colorectal mucosa and may progress to malignant CRC over time (6). These lesions typically originate from glandular cells in the mucosa, where genetic and environmental factors drive uncontrolled cellular proliferation through mutations. Early identification and treatment of precancerous colorectal lesions are critical for preventing cancer development (17). Regular screening programs, such as colonoscopy, enable the detection and removal of these lesions, significantly reducing cancer risk. However, conventional diagnostic methods have limitations, including operator-dependent variability and relatively high miss rates (18). Therefore, identifying specific biomarkers from precancerous lesions could improve early detection of high-risk populations and enhance the diagnostic yield of CRAs (19). Previous biomarker research has predominantly focused on DEGs between CRC and adjacent normal tissues, often overlooking precancerous lesions. Screening for specific markers in precancerous stages may improve early detection rates and preventive efficacy against CRC. Recent advances in high-throughput sequencing have shifted attention toward identifying precancerous lesion-specific biomarkers. Compared to traditional approaches analyzing cancer versus normal tissue differences, biomarkers derived from precancerous lesions offer several advantages: (I) improved detection rates of CRA, enabling early CRC prevention (20); (II) elucidation of molecular mechanisms underlying CRC initiation and progression (21); (III) identification of novel targets for personalized therapy (22); and (IV) refinement of molecular subtyping in CRC (23). Thus, investigating and screening colorectal tumor biomarkers and their mechanisms holds significant clinical value.

Public data mining, an emerging bioinformatics approach, systematically integrates and analyzes open-access biological databases, offering new perspectives for cancer research (24). In this study, we leveraged GEO datasets containing high-grade intraepithelial neoplasia (HGIN) samples to identify DKK4 as a key regulator in the adenoma-carcinoma transition through differential expression and functional enrichment analyses. As a member of the Dickkopf family, DKK4 exhibits dual roles in tumorigenesis: (I) acting as a negative regulator of Wnt signaling by binding to LRP5/6 receptors and inhibiting β-catenin nuclear translocation (25); and (II) promoting epithelial-mesenchymal transition (EMT) in specific tumor microenvironments, enhancing cancer cell invasiveness (25). Notably, DKK4 expression peaks in HGIN, suggesting its involvement in early malignant transformation. Its significant correlation with TMB further supports a role in CRC pathogenesis. Functional enrichment analyses consistently implicated the Wnt pathway as DKK4’s primary oncogenic mechanism, aligning with its known Wnt-antagonistic function.

This study confirmed a significant negative correlation between DKK4 expression and TMB in patients with CRC (P=0.008), with the mean TMB in the low DKK4 expression group being significantly higher than that in the high DKK4 expression group. At high expression levels, DKK4 negatively regulates the Wnt signaling pathway by inhibiting β-catenin nuclear translocation, which maintains the genomic stability of tumor cells, reduces mutation accumulation and thereby lowers TMB levels. Meanwhile, it constructs an immunosuppressive tumor microenvironment by upregulating immunosuppressive genes and promoting M2 macrophage polarization, which facilitates the abnormal proliferation of cells at the adenoma stage. In contrast, decreased DKK4 expression relieves the inhibitory effect on the Wnt pathway, impairs the genomic stability of tumor cells and increases the probability of mutations, leading to elevated TMB. Concurrently, the imbalance of the immune regulatory network further exacerbates the immunosuppressive state of the tumor microenvironment, driving the malignant transformation of adenomas into invasive carcinomas. These findings confirm the stage-specific regulatory role of DKK4 during CRA-carcinoma transition, which is highly consistent with the expression pattern of DKK4 identified in this study—high expression in adenomas and low expression in invasive carcinomas.

Immune profiling revealed novel immunomodulatory properties of DKK4. Its strong associations with CD4+ T cells and macrophages, along with dual regulation of immunostimulatory (CD80, IFNG, PRF1) and immunosuppressive (CD274, ARG1, VEGFB) genes, suggest contributions to an immunosuppressive tumor microenvironment. These findings corroborate prior reports of DKK4’s multifaceted roles in tumor immunomodulation (26). Emerging evidence indicates that DKK4 mediates immune activation-suppression balance via Wnt/β-catenin signaling (27), through mechanisms including: (I) promoting M2 macrophage polarization while suppressing M1 activation (28); (II) impairing dendritic cell maturation and antigen presentation (29); and (III) enhancing Treg differentiation while inhibiting effector and cytotoxic T cells (30). In colorectal tumors, high DKK4 expression correlates strongly with immunosuppressive microenvironments (27), characterized by reduced tumor-infiltrating lymphocytes and elevated PD-1/PD-L1 levels (31). These observations provide mechanistic insights into DKK4-associated immunotherapy resistance in solid tumors.

LASSO regression offers distinct advantages for oncological applications: effectively addressing high-dimensional data challenges, enabling multi-omics integration, facilitating target prioritization, and providing quantitative tools for personalized prognosis (32,33). Here, we constructed a CRC risk prediction model centered on DKK4 and 12 strongly correlated (r>0.8) protein-protein interaction network genes. The resulting 12-gene signature demonstrated robust risk stratification, with superior short-term (1–3-year) versus long-term (5-year) predictive accuracy. These results align with previous findings that high DKK4 expression correlates with elevated risk scores (34) and deeper tumor invasion (35), suggesting predominant involvement in early rather than metastatic progression. The time-dependent predictive performance underscores clinical utility for early intervention while highlighting the need for complementary long-term prognostic markers. We speculate that the main reasons for the limitations of the model such as: the model was constructed solely on the basis of transcriptomic data, without integrating multi-dimensional data including clinicopathological characteristics and epigenetics; DKK4 mainly functions during the precancerous stage, and its regulatory effect on long-term survival following invasion and metastasis of CRC is attenuated, leading to insufficient predictive performance for long-term prognosis.

Integrated pathological and immunohistochemical analyses revealed dynamic DKK4 expression patterns during the adenoma-carcinoma sequence. DKK4 levels peaked in HGIN but declined in invasive cancer, indicating stage-specific roles. Its glandular localization and decreased expression accompanying glandular structure disruption suggest utility as an epithelial differentiation marker. These histopathological findings substantiate DKK4’s involvement in CRC development, consistent with reports of its differential expression in early-stage tumors and Wnt pathway modulation (34,36).

DKK4 differs significantly from classic CRC biomarkers such as CEA and CA19-9. It is highly expressed in precancerous lesions (high-grade intraepithelial neoplasia, HGIN) with prominent predictive value, but its expression decreases after malignant transformation—distinct from the application scenario of traditional biomarkers that target the invasive/metastatic stages post-carcinogenesis. CRC exhibits marked molecular subtype heterogeneity, and refined subtyping of enrolled patients has not yet been completed (relevant work is ongoing); thus, pan-cancer level comparisons fail to highlight its stage-specific advantages. DKK4 is detected via molecular pathology in tissue samples, whereas classic biomarkers mostly rely on serological assays or gene sequencing. Based on the results of this study, DKK4 is more suitable as a target for molecular pathological auxiliary diagnosis, facilitating the auxiliary assessment of the inflammation-cancer transition process and its role.

Study limitations warrant consideration: (I) potential selection bias in retrospective public data analyses; (II) moderate predictive efficacy of DKK4 for prognosis after tumor progression; and (III) requirement for multicenter prospective validation of the prognostic model. Future investigations should explore DKK4’s therapeutic potential and tumor-stroma interaction mechanisms. This study aims to clarify the stage-specific expression pattern of DKK4 during the CRA-carcinoma sequence. Rather than focusing on in-depth exploration of molecular mechanisms, the primary objective of this research is to dissect the unique characteristics and clinical translational value of DKK4 in CRA carcinogenesis. Investigations into the mechanisms underlying CRA-carcinoma transformation are more suitably conducted using a continuous mouse model of colitis-adenoma-CRC. Establishing this model requires a sufficient number of experimental animals, long-term modeling, anatomical dissection, and refined pathological diagnosis to accurately distinguish between low-grade and high-grade intraepithelial neoplasia, thereby enabling targeted analysis of the regulatory roles of DKK4 at distinct malignant transformation stages. In future work, our team plans to construct a mouse model of CRA and systematically elucidate the molecular mechanisms by which DKK4 regulates adenoma carcinogenesis.

Conclusions

In conclusion, our multi-omics approach establishes DKK4 as a promising biomarker for early CRC detection and risk prediction. The developed prognostic model demonstrates particular clinical value for short-term outcome stratification. DKK4, predominantly expressed in colorectal glandular cells, associates with aberrant proliferation, invasion, and pathological features during adenoma-carcinoma transition. These findings provide a foundation for elucidating DKK4’s molecular mechanisms and therapeutic implications in CRC progression.

Acknowledgments

None.

Footnote

Reporting Checklist: The authors have completed the TRIPOD reporting checklist. Available at https://tcr.amegroups.com/article/view/10.21037/tcr-2026-1-0021/rc

Data Sharing Statement: Available at https://tcr.amegroups.com/article/view/10.21037/tcr-2026-1-0021/dss

Peer Review File: Available at https://tcr.amegroups.com/article/view/10.21037/tcr-2026-1-0021/prf

Funding: This work was supported by the Ningxia Natural Science Foundation Project (No. 2024AAC03671, to X.W.), Scientific Research Projects of Higher Education Institutions in Ningxia (No. NYG2024127, to X.W.), Project of the Health Commission of the Ningxia Hui Autonomous Region (No. 2024-NWQP-B012, to X.W.), Central Guidance for Local Science and Technology Development Special Funds Project (No. 2024FRD05103, to X.Z.), Yinchuan Science and Technology Plan Project (No. 2024SF044, to X.Z.), and Special Talent Introduction Project of Ningxia Autonomous Region Key R&D Programs (No. 2023BSB03053, to X.Z.).

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://tcr.amegroups.com/article/view/10.21037/tcr-2026-1-0021/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. This study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the Ethics Committee of General Hospital of Ningxia Medical University (No. KYLL-2023-1429), and written informed consent was obtained from all participating patients prior to their procedures.

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. Sung H, Ferlay J, Siegel RL, et al. Global Cancer Statistics 2020: GLOBOCAN Estimates of Incidence and Mortality Worldwide for 36 Cancers in 185 Countries. CA Cancer J Clin 2021;71:209-49. [Crossref] [PubMed]
2. Siegel RL, Wagle NS, Cercek A, et al. Colorectal cancer statistics, 2023. CA Cancer J Clin 2023;73:233-54. [Crossref] [PubMed]
3. Yamada A, Kondo T. Hereditary Colorectal Cancer: Clinical Implications of Genomic Medicine and Precision Oncology. J Anus Rectum Colon 2025;9:167-78. [Crossref] [PubMed]
4. Wang F, Chen G, Zhang Z, et al. The Chinese Society of Clinical Oncology (CSCO): Clinical guidelines for the diagnosis and treatment of colorectal cancer, 2024 update. Cancer Commun (Lond) 2025;45:332-79. [Crossref] [PubMed]
5. Soluk-Tekkesin M, Wright JM. The World Health Organization Classification of Odontogenic Lesions: A Summary of the Changes of the 2022 (5th) Edition. Turk Patoloji Derg 2022;38:168-84. [Crossref] [PubMed]
6. Vogelstein B, Fearon ER, Hamilton SR, et al. Genetic alterations during colorectal-tumor development. N Engl J Med 1988;319:525-32. [Crossref] [PubMed]
7. Mitsala A, Tsalikidis C, Pitiakoudis M, et al. Artificial Intelligence in Colorectal Cancer Screening, Diagnosis and Treatment. A New Era. Curr Oncol 2021;28:1581-607.
8. Siegel RL, Miller KD, Fuchs HE, et al. Cancer Statistics, 2021. CA Cancer J Clin 2021;71:7-33. [Crossref] [PubMed]
9. Hou S, Heng S, Xie S, et al. Integrated multi-omics and machine learning reveal an immunogenic cell death-related signature for prognostic stratification and therapeutic optimization in colorectal cancer. Front Immunol 2025;16:1606874. [Crossref] [PubMed]
10. Wang B, Zhang J, Wang X, et al. Identification and clinical validation of key genes as the potential biomarkers in colorectal adenoma. BMC Cancer 2023;23:39. [Crossref] [PubMed]
11. Ellrott K, Wong CK, Yau C, et al. Classification of non-TCGA cancer samples to TCGA molecular subtypes using compact feature sets. Cancer Cell 2025;43:195-212.e11. [Crossref] [PubMed]
12. Gene Ontology Consortium. The Gene Ontology resource: enriching a GOld mine. Nucleic Acids Res 2021;49:D325-D334. [Crossref] [PubMed]
13. Kanehisa M, Furumichi M, Sato Y, et al. KEGG: biological systems database as a model of the real world. Nucleic Acids Res 2025;53:D672-7. [Crossref] [PubMed]
14. Tang N, Zhou Q, Liu S, et al. GSEA analysis identifies potential drug targets and their interaction networks in coronary microcirculation disorders. SLAS Technol 2024;29:100152. [Crossref] [PubMed]
15. Guan M, Jiao Y, Zhou L. Immune Infiltration Analysis with the CIBERSORT Method in Lung Cancer. Dis Markers 2022;2022:3186427. [Crossref] [PubMed]
16. Liu S, Wang Z, Zhu R, et al. Three Differential Expression Analysis Methods for RNA Sequencing: limma, EdgeR, DESeq2. J Vis Exp 2021;
17. Jacobsson M, Wagner V, Kanneganti S. Screening for Colorectal Cancer. Surg Clin North Am 2024;104:595-607. [Crossref] [PubMed]
18. Lund M, Trads M, Njor SH, et al. Quality indicators for screening colonoscopy and colonoscopist performance and the subsequent risk of interval colorectal cancer: a systematic review. JBI Database System Rev Implement Rep 2019;17:2265-300. [Crossref] [PubMed]
19. Mo S, Ye L, Wang D, et al. Early Detection of Molecular Residual Disease and Risk Stratification for Stage I to III Colorectal Cancer via Circulating Tumor DNA Methylation. JAMA Oncol 2023;9:770-8. [Crossref] [PubMed]
20. Jayasinghe M, Prathiraja O, Caldera D, et al. Colon Cancer Screening Methods: 2023 Update. Cureus 2023;15:e37509. [Crossref] [PubMed]
21. Bruhm DC, Vulpescu NA, Foda ZH, et al. Genomic and fragmentomic landscapes of cell-free DNA for early cancer detection. Nat Rev Cancer 2025;25:341-58. [Crossref] [PubMed]
22. Dunne PD, Arends MJ. Molecular pathological classification of colorectal cancer-an update. Virchows Arch 2024;484:273-85. [Crossref] [PubMed]
23. Guinney J, Dienstmann R, Wang X, et al. The consensus molecular subtypes of colorectal cancer. Nat Med 2015;21:1350-6. [Crossref] [PubMed]
24. Clough E, Barrett T, Wilhite SE, et al. NCBI GEO: archive for gene expression and epigenomics data sets: 23-year update. Nucleic Acids Res 2024;52:D138-44. [Crossref] [PubMed]
25. Liang J, Sun L, Li Y, et al. Wnt signaling modulator DKK4 inhibits colorectal cancer metastasis through an AKT/Wnt/β-catenin negative feedback pathway. J Biol Chem 2022;298:102545. [Crossref] [PubMed]
26. Desterke C, Fu Y, Bonifacio-Mundaca J, et al. Single-Cell RNA Sequencing Reveals LEF1-Driven Wnt Pathway Activation as a Shared Oncogenic Program in Hepatoblastoma and Medulloblastoma. Curr Oncol 2025;32:35. [Crossref] [PubMed]
27. Wang M, Ni B, Zhuang C, et al. Aberrant accumulation of Dickkopf 4 promotes tumor progression via forming the immune suppressive microenvironment in gastrointestinal stromal tumor. Cancer Med 2019;8:5352-66. [Crossref] [PubMed]
28. Kikuchi A, Matsumoto S, Sada R. Dickkopf signaling, beyond Wnt-mediated biology. Semin Cell Dev Biol 2022;125:55-65. [Crossref] [PubMed]
29. Ding Y, Bian TT, Li QY, et al. A new risk model for CSTA, FAM83A, and MYCT1 predicts poor prognosis and is related to immune infiltration in lung squamous cell carcinoma. Am J Transl Res 2022;14:7705-25.
30. Hiramitsu S, Terauchi M, Kubota T. The effects of Dickkopf-4 on the proliferation, differentiation, and apoptosis of osteoblasts. Endocrinology 2013;154:4618-26. [Crossref] [PubMed]
31. Mourtada J, Thibaudeau C, Wasylyk B, et al. The Multifaceted Role of Human Dickkopf-3 (DKK-3) in Development, Immune Modulation and Cancer. Cells 2023;13:75. [Crossref] [PubMed]
32. Kang J, Kim Y, Jeong Y, et al. Development and validation of a LASSO-Based FDG PET/CT model for predicting colorectal adenoma in asymptomatic individuals undergoing colonoscopy. Ann Nucl Med 2026;40:218-25. [Crossref] [PubMed]
33. Ali H, Shahzad M, Sarfraz S, et al. Application and impact of Lasso regression in gastroenterology: A systematic review. Indian J Gastroenterol 2023;42:780-90. [Crossref] [PubMed]
34. Park SJ, Cho JG, Han SH, et al. Dickkopf 4 Alone and in Combination with Leucyl-tRNA Synthetase as a Good Prognostic Biomarker for Human Colorectal Cancer. Biomed Res Int 2023;2023:9057735. [Crossref] [PubMed]
35. Zhai W, Hu GH, Zheng JH, et al. High expression of the secreted protein dickkopf homolog 4: Roles in invasion and metastasis of renal cell carcinoma and its association with Von Hippel-Lindau gene. Int J Mol Med 2023;51:47. [Crossref] [PubMed]
36. Lou X, Meng Y, Hou Y. A literature review on function and regulation mechanism of DKK4. J Cell Mol Med 2021;25:2786-94. [Crossref] [PubMed]