3.1 Expression and prognosis of DcR3 in CRC

DcR3 expression was significantly higher in tumor tissue compared to normal tissue (Fig. 1a). To gain further insights, we conducted survival analyses, which yielded intriguing results. High DcR3 expression was associated with a favorable prognosis in terms of both Overall Survival (OS) and Disease-Free Survival (DFS) (Fig. 1b–d). To better understand the clinical implications of DcR3, we examined its correlation with various clinical features. Interestingly, patients with high DcR3 expression were more likely to be in the M0 stage, indicating a potential association with less metastatic spread (Fig. 1e). Conversely, low DcR3 expression seemed to be linked with cancer progression to stage 3/4. Furthermore, immunohistochemical (IHC) analysis has revealed elevated levels of the DcR3 protein within tumor tissue (Figure f,g). These intriguing findings raised questions about the role of DcR3 in CRC. Despite its high expression in tumor tissue, it appears to be associated with improved survival outcomes in CRC patients. To shed light on the underlying mechanisms and potential reasons for this observation, we aim to delve deeper into the molecular and cellular factors influenced by DcR3 in CRC. Further investigation into DcR3's functional roles in CRC could provide valuable insights and contribute to our understanding of this complex disease.

Fig. 1

Expression and Prognosis of DcR3 in CRC. a DcR2 mRNA expression in READ and COAD. b OS of DcR3 from TCGA, KM plotter c and DFS of DcR3 (TCGA). d Expression of DcR3 in different clinical feature stage. f IHC straning of DcR3 in normal and tumor tissue. e Expression of DcR3 protein in normal and tumor tissue

3.2 Difference in immune, TMB, and pathway of DcR3 high and low group

In our efforts to understand the reasons behind the favorable prognosis observed in high DcR3 CRC patients, we examined the abundance of 22 immune cell types in the high and low DcR3 expression groups. Notably, Macrophages M0 (p = 0.03), Macrophages M2 (p = 0.09), Macrophages M1 (p = 0.06), and T cells CD4 memory resting (p = 0.05) were found to be more abundant in the high DcR3 group compared to other cells (Fig. 2a). Specifically, Macrophages M1 and T cells CD4 memory resting primarily accumulated in the DcR3 high group, while Macrophages M0 and Macrophages M2 were more prevalent in the DcR3 low group. Additionally, we evaluated the scores of 29 immune feature sets, encompassing 16 immune cells and 13 immune functions. We observed higher levels of immune-related features, such as APC_co_inhibition, APC_co_stimulation, T_cell_co_inhibition, T_cell_co_stimulation, TIL, and CD8_T_cells in the DcR3 high group (Fig. 2b). Furthermore, the DcR3 high group displayed a higher Tumor Mutational Burden (TMB) score, suggesting potential benefits for targeted cancer treatment (Fig. 2c). And GSEA revealed that pathways like Antigen processing and presentation, Cytokine-cytokine receptor interaction, and IL-17 signaling were enriched in the DcR3 high group (Fig. 2d). In contrast, the DcR3 low group exhibited enrichment in pathways such as Bile secretion, Drug metabolism-cytochrome P450, Drug metabolism-other enzymes, and Metabolism of xenobiotics by cytochrome P450 (Fig. 2e). These findings suggest that the immune system may play a crucial role in the favorable prognosis observed in the DcR3 high group, while metabolism pathways may be disturbed in the DcR3 low group. These results offer valuable insights into the potential mechanisms underlying the favorable outcomes observed in high DcR3 CRC patients and provide a basis for further exploration of immune-related pathways and therapeutic opportunities in colorectal cancer management.

Fig. 2

Difference in Immune, TMB, and Pathway of DcR3 High/Low Group. a Immune cell abundance in DcR3 High/Low Group. b Difference of immune feature sets in DcR3 High/Low Group. c TMB score in DcR3 High/Low Group. d Enrichment pathway in DcR3 High group and Low group (e)

3.3 DcR3-associated risk score (DARS) model

To delve deeper into the potential functions of DcR3, we sought to construct a riskscore model that would be associated with DcR3. In the initial step, we employed the Pearson correlation method to identify 769 genes that were correlated with DcR3 (r > 0.3 or r < 0.3, p < 0.05). Next, a univariate Cox regression analysis was performed on the training set, leading us to select 7 genes (ABCA7, DPP7, HDAC10, HMHA1, KDM3A, NFKB2, and TMEM86B) for further analysis through the Lasso method (Fig. 3a, c). In the training cohort, we calculated the riskscore for each sample and categorized them into either high or low-risk groups based on the median riskscore. The results demonstrated that the high-risk group had a higher mortality rate, whereas the low-risk group had a higher survival rate (Fig. 3b). We further validated these findings in the test group, where we observed a similar pattern with the high-risk group exhibiting increased mortality (Figure S1). Subsequently, utilizing Lasso and Multivariate Cox regression analysis, we successfully identified a refined riskscore model comprising 3 genes (DPP7, KDM3A, and TMEM86B) (Fig. 3d). This model, named DARS, allowed us to calculate the riskscore using the formula: riskscore = (0.52 × expression of DPP7) + (0.69 × expression of KDM3A) + (0.34 × expression of TMEM86B). The DARS model provides a promising tool to assess risk levels and predict patient outcomes based on the expression of these 3 genes. By incorporating the DARS model, we aim to gain better insights into the potential impact of DcR3 on CRC prognosis, as well as identify potential therapeutic targets for personalized cancer management.

Fig. 3

DcR3-Associated Risk Score (DARS) Model. a Lasso coefficient profiles of the 7 DcR3-Associated genes. b Survival time and status of High/Low Risk patients and expression of three genes in High/Low Risk group. c LASSO regression analysis was used to prevent the overfitting effects of the model. d The screened genes were brought into multivariate Cox regression analysis and constructed the DARS Model

3.4 Estimation of risk model and establishment of nomograms

To thoroughly evaluate the performance of the riskscore model, we conducted survival analyses and ROC curve analysis. Our results consistently showed that the high-risk group exhibited poorer prognosis in both the train and test sets (Fig. 4a, b). External validation with the GSE17536 dataset also confirmed the association between the high-risk group and unfavorable prognosis (Fig. 4c). ROC curve analysis further demonstrated the accuracy of the riskscore model, with AUC values of 0.73 and 0.753 for 3-year and 5-year survival in the train set, respectively. The AUC values for the 3-year and 5-year survival in the test set were 0.722 and 0.707, respectively (Fig. 4d, e). Next, we performed univariate and multivariate Cox regression analyses using TCGA and GSE17536 clinical data to explore the prognostic significance of the risk model alongside other clinical factors. Univariate Cox regression revealed that age, stage, M_STAGE, N_STAGE, T_STAGE, and riskScore were significantly associated with prognosis in TCGA data, while stage, grade, and riskScore were associated with prognosis in GSE17536 data (Fig. 4f and Figure S2a). Multivariate Cox regression analysis in TCGA data identified age, sex, M_STAGE, T_STAGE, and riskScore as independent prognostic factors, while in GSE17536 data, stage and riskScore were the independent prognostic factors (Fig. 4g and Figure S2a). To provide clinicians with a practical tool for predicting survival risk, we developed nomograms for both TCGA and GEO cohorts. These nomograms integrated riskScore, age, sex, stage, M_STAGE, N_STAGE, and T_STAGE for the TCGA cohort and riskScore, age, gender, stage, and grade for the GEO cohort (Fig. 4h and Figure S2b). Furthermore, we validated the accuracy and net benefit of the DARS model using Calibration curve and Decision Curve Analysis (DCA) in both TCGA and GEO cohorts (Fig. 4i, j and Figure S2c). The results confirmed the stability and clinical utility of our riskscore model. It can serve as an independent prognostic indicator, aiding clinicians in more accurately predicting patient outcomes and potentially guiding treatment decisions for CRC patients.

Fig. 4

Estimation of Risk Model and Establishment of Nomograms. a Prognosis of DARS in Train set, Test set (b) and GEO set (c). d, e ROC curve about 3-year and 5-year survival. f, g Univariate and Multivariate Cox regression analyses. h Nomograms including riskScore, age, sex, stage, M_STAGE, N_STAGE, and T_STAGE. i Calibration curve of DARS in 3-year and 5-year. j Evaluation of the clinical usefulness of the DARS

3.5 Assessment of correlation between DARS model and clinical features

Clinical features serve as important indicators for the development of cancers [26]. To gain deeper insights into the association between the DARS model and CRC, we conducted a Chisq-test to analyze the relationship between the DARS risk group and various clinical features in CRC. The distribution of high and low risk groups was profiled for each clinical feature (Fig. 5a, b). Our findings indicated that the high DARS group was positively correlated with age, suggesting that higher DARS may be associated with older age in CRC patients. Additionally, high DARS was found to promote cancer development from stage I/II to stage III/IV, implying that elevated DARS might be linked to disease progression (Fig. 5c). Moreover, the DARS was associated with M_STAGE and N_STAGE in the TCGA cohort, indicating that high DARS could be indicative of tumor metastasis and increased lymph node involvement (Fig. 5c). These results were further validated in the GEO cohort, providing additional support for the impact of DARS on CRC development (Fig. 5d). More detailed statistics on the correlation between DARS and clinical features can be found in Table 1. This table presents comprehensive data on the relationship between clinical features and the DARS group across different cohorts, including train, test, TCGA, and GSE17536. Furthermore, we analyzed the difference in riskScore across different stages of clinical features. The results aligned with our previous findings, demonstrating that patients over 60 years of age and those with clinical stages at III/IV had higher risks (Fig. 5e, f). These findings are in line with well-established clinical facts, further strengthening the significance of our DARS model as a reliable prognostic tool for CRC patients. In conclusion, the correlation analysis between the DARS model and clinical features sheds light on the potential impact of DARS on CRC development and progression. By identifying these associations, our model proves to be a valuable asset for predicting patient outcomes and guiding treatment decisions in CRC management.

Fig. 5

Assessment of Correlation between DARS Model and Clinical Features. Proportion of high and low risk groups was profiled for each clinical feature in TCGA cohort (a, c) and GEO cohort (b, d). e, f Correlation between DARS and clinical features

Table 1 Correlation between DARS and clinical features3.6 Mutation and enrichment analysis between high and low riskscore groups

To explore the differences in mutation patterns, we obtained CRC mutation data from TCGA. We identified the top 20 genes with the highest mutation frequency in both the high and low riskscore groups (Fig. 6a). The mutation types included Nonsense Mutation, Missense Mutation, Frame Shift Del, Frame Shift Ins, Splice Site, In Frame Del, In Frame Ins, and Multi Hit. Nonsense Mutation emerged as the predominant cause of gene mutation. In the low riskscore group, the top 10 mutated genes were TTN (34%), APC (63%), MUC16 (19%), SYNE1 (19%), TP53 (48%), FAT4 (17%), KRAS (33%), RYR2 (16%), OBSCN (14%), and PIK3CA (19%). Conversely, in the high riskscore group, the top 10 mutated genes were TTN (42%), APC (59%), MUC16 (22%), SYNE1 (24%), TP53 (46%), FAT4 (18%), KRAS (32%), LRP1B (15%), ZFHX4 (15%), and PIK3CA (21%) (Fig. 6b). We conducted survival analysis to investigate the correlation between mutations and prognosis, revealing that mutations in SYNE1 were associated with poor prognosis (Fig. 6c). These results indicated that the high riskscore group had a higher mutation frequency, and specifically, mutations in SYNE1 were linked to decreased survival time. Additionally, we performed enrichment analysis to uncover the different pathways between the high and low riskscore groups. We focused on both biological processes (BP) and the Kyoto Encyclopedia of Genes and Genomes (KEGG). The pathways enriched in the high riskscore group included organic hydroxy compound transport, regulation of system process, embryonic organ morphogenesis, skeletal system development, and calcium ion transport, among others (Fig. 7a). In contrast, the pathways enriched in the low riskscore group involved ncRNA processing, epidermis development, ribonucleoprotein complex biogenesis, ribosome biogenesis, and rRNA processing (Fig. 7a). KEGG analysis indicated that the high riskscore group showed enrichment in pathways such as the Synaptic vesicle cycle, Hematopoietic cell lineage, Nicotine addiction, GnRH secretion, and Neuroactive ligand-receptor interaction, while the low riskscore group displayed enrichment in pathways like Oxidative phosphorylation, Chemical carcinogenesis—reactive oxygen species, Protein export, IL-17 signaling pathway, and Basal transcription factors (Fig. 7b). Notably, the low riskscore group exhibited activation of the IL-17 signaling pathway. This suggests that the low riskscore group may promote immune reactions through the IL-17 signaling pathway, potentially contributing to their more favorable prognosis. The combination of mutation and enrichment analysis provides valuable insights into the underlying biological mechanisms and potential therapeutic targets associated with the DARS model in CRC.

Fig. 6

Mutation and Enrichment Analysis between High and Low Riskscore Groups. Landscape of mutation in High/Low risk groups (a, b). c Survival analysis of SYNE1

Fig. 7

Enrichments of pathway in High/Low risk groups of DARS. a Biology process. b KEGG

3.7 Correlation between DcR3 and riskScore group

The observed differences in survival outcomes between high DcR3 and low DcR3 patients, along with the correlation analysis with the IL-17 signaling pathway, provide valuable insights into the potential reasons behind the contrasting prognosis in CRC patients. Firstly, the analysis of IL-17 related genes revealed that MMP13, CXCL3, MMP3, CCL11, CXCL1, CXCL6, MMP1, S100A9, S100A8, DEFB4A, S100A7, and TNFRSF6B exhibited higher expression levels in the low riskScore group (Fig. 8a). Subsequently, the correlation analysis demonstrated that DcR3 expression was positively correlated with the expression of CXCL3 (r = 0.464, p < 0.001), CXCL1 (r = 0.51, p < 0.001), and S100A9 (r = 0.354, p < 0.001) (Fig. 8b). This suggests that DcR3 may play a role in activating the immune system by modulating the expression of these IL-17 related genes. The upregulation of these genes in the low riskScore group might contribute to the better prognosis observed in these patients. Additionally, the analysis of DcR3 expression in the high riskScore group showed that DcR3 levels were relatively low (p = 0.1), and conversely, in the high DcR3 expression group, the riskScore was also relatively low (p = 0.08) (Fig. 8c). This further supports the notion that high DcR3 expression is associated with a more favorable prognosis, as evidenced by the lower riskScore in these patients. Collectively, these findings suggest that the interaction between DcR3 and the IL-17 signaling pathway may be a key factor in determining the survival outcomes of CRC patients. DcR3 might act as an activator of the immune system through its regulation of IL-17 related genes, leading to improved prognosis in CRC patients with higher DcR3 expression. Understanding the interplay between DcR3 and the IL-17 signaling pathway can provide important insights into the molecular mechanisms underlying CRC development and progression, and it highlights DcR3 as a potential target for further investigation and therapeutic interventions.

Fig. 8

Correlation between DcR3 and riskScore group of DARS. a Heatmap of DcR3 and genes in IL-17 signaling pathway. b Correlation between DcR3 and genes. c Correlation between DcR3 and riskScore