3.1 BID is highly expressed in ccRCC tissues

We used transcriptome information of 539 ccRCC tumor samples and 72 matched paracancerous samples to explore the differential expression of BID. Results as shown in Fig. 1A, compared with normal renal tissue, the expression levels of BID in ccRCC tissue was significantly up-regulated (P < 0.001). In addition, we validate the result by the GSE53757 dataset and GSE40435 dataset of the GEO database. Consistent with our expected result, the differential expression of BID was further verified in tumor tissues and normal renal tissues (Fig. 1B, C, P < 0.001). Moreover, the results of transcriptome sequencing of tumor samples also demonstrated that BID was highly expressed in ccRCC tissues and lowly expressed in normal kidney tissues (Fig. 1D, P < 0.001).

Fig. 1

BID is highly expressed in ccRCC tissues. A In the TCGA database, the expression levels of BID was significantly up-regulated in ccRCC tissues. B, C The differential expression of BID in normal kidney tissues and ccRCC tissues in the GEO dataset. D The transcriptome sequencing of tumor specimens confirmed the differential expression of BID in normal renal tissues and ccRCC tissues

3.2 The prognostic value of BID in ccRCC

Kaplan–Meier survival curves revealed a shorter OS and PFS for BID high expression group patients compare to patients in BID low expression group (Fig. 2A, B, P < 0.001). We introduced ROC curves to evaluate the predictive potential of BID on the prognosis of ccRCC patients. The results revealed that the AUC values of the ROC curves of BID predicting the 1-, 3-, 5- and 7-year OS of patients were 0.686, 0.651, 0.652 and 0.677, respectively (Fig. 2C). The AUC values of the ROC curves of BID predicting the 1-, 3-, 5- and 7-year PFS of patients were 0.670, 0.639, 0.677 and 0.693, respectively (Fig. 2D). This suggests that BID has good prognostic predictive ability for ccRCC patients. The results of univariate Cox independent prognostic analysis demonstrated that BID (HR = 1.216, P < 0.001), age (HR = 1.033, P < 0.001), T stage (HR = 1.941, P < 0.001), M stage (HR = 4.284, P < 0.001), histological grade (HR = 2.293, P < 0.001) and clinical stage (HR = 1.889, P < 0.001) were all correlated with the OS of patients (Fig. 2E). After excluding confounding factors by multivariate Cox prognostic analysis, BID was still an independent risk factor for the prognosis of ccRCC patients (Fig. 2F). This indicates BID may be a risk gene of ccRCC and participate in tumor progression.

Fig. 2

The prognostic value of BID in ccRCC. A The Kaplan–Meier survival curve of BID predicting the OS of patients. B The Kaplan–Meier survival curve of BID predicting the PFS of patients. C The ROC curve of BID predicting the OS of 1-year, 3-year, 5-year and 7-year for ccRCC patients. D The ROC curve of BID predicting the PFS of 1-year, 3-year, 5-year and 7-year for ccRCC patients. E The result of univariate Cox independent prognostic analysis. F The result of multivariate Cox independent prognostic analysis

3.3 Clinical correlation between the expression levels of BID and ccRCC

We found BID was clearly associated with common clinicopathological features of ccRCC patients except for age (Fig. 3A). Specifically, there was no difference in the expression levels of BID among different age groups (Fig. 3B), but it was higher in male patients (Fig. 3C). In addition, the high expression of BID may predict worse T stage, N stage, histological grade and clinicopathological stage, and lead to distant metastasis (Fig. 3D–H).

Fig. 3

The relationship between the BID and the clinicopathological features of ccRCC patients. A The heat map of the correlation. B The expression level of BID in different age groups of patients. C The expression level of BID in male and female. D–H The relationship between the expression levels of BID and the T stage, N stage, M stage, histological grade and clinical stage of patients (*: P < 0.05; **: P < 0.01; ***: P < 0.001)

3.4 Construction of BID-related prognostic model

In the above analysis, we identified the risk factors with independent prognostic value for ccRCC. We constructed a risk prognosis model by multivariate Cox regression analysis in the TCGA training cohort.

$$\mathrm=\left(0.023060099\times }\right)+\left(0.088244672\times \mathrm\right)+\left(0.306695441\times }\right)+\left(0.545605079\times }\right)$$

The Kaplan–Meier survival analysis demonstrated a distinctly decreased OS of patients with high riskscore (Fig. 4A, P < 0.001). The AUC values of the ROC curves for predicting the 1-, 3-, and 5-year OS via the model were 0.863, 0.840, and 0.793, respectively, demonstrating the good predictive performance of the model (Fig. 4B). In addition, the results of PCA showed that the model could accurately distinguish patients with different risks (Fig. 4C). As the risk score increased, the patient's survival time decreased and the number of deaths significantly increased, indicating the risk score was negatively correlated with the prognosis of patients (Fig. 4D). Moreover, the multi-index ROC curves revealed that the AUC value of the model was better than that of any single clinical trait, no matter at 1 year, 3 years, or 5 years, showing good prognostic predictive ability (Fig. 4E–G).

Fig. 4

Construction of a BID-related prognostic model. A There is a significant difference in OS between the high- and low-risk group. B The ROC curve of the model predicting the 1-year, 3-year, and 5-year OS of patients. C The result of PCA analysis. D The risk distribution curve and survival status plots of patients. E–G The multi-index ROC curves

3.5 Validation of BID-related prognostic model

We validated the model’s accuracy and effectiveness in the testing cohort, ICGC external cohort and GEO external cohort. Kaplan–Meier survival analysis showed a shorter OS for patients in the high-risk group in the testing cohort (Fig. 5A, P < 0.001). Interestingly, consistent with the result of the TCGA training cohort, the same difference in OS was observed between two groups in ICGC external cohort and GEO external cohort (Fig. 5B, C, P < 0.001). In the testing cohort, the AUC values of 1 -, 3-and 5-year OS of the prognostic model were 0.878, 0.771 and 0.703, respectively (Fig. 5D). In the ICGC external cohort, the AUC values of 1 -, 3-and 5-year OS of the prognostic model were 0.595, 0.700 and 0.767, respectively (Fig. 5E). In the GEO external cohort, the AUC values of 1 -, 3-and 5-year OS of the prognostic model were 0.783, 0.857 and 0.808, respectively (Fig. 5F). In addition, the results of the PCA of the three cohorts showed that the model could also accurately distinguish patients with different risks (Fig. 5G–I). The risk distribution curve and survival status map of the three cohorts also demonstrated that the risk score was negatively correlated with the prognosis of patients (Fig. 5J–L). Anyway, these results revealed the accuracy and effectiveness of the BID-related risk model in predicting the prognosis of ccRCC patients.

Fig. 5

Internal and external validation of BID-related prognostic models. A–C Kaplan–Meier survival curves of patients in the high- and low-risk groups in the testing cohort, ICGC external cohort and GEO external cohort. D–F The ROC curves of the prognostic model in the testing cohort, ICGC external cohort and GEO external cohort. G–I The results of the PCA analysis for the testing cohort, ICGC external cohort and GEO external cohort. J–L The risk distribution curve and survival status plots for the testing cohort, ICGC external cohort and GEO external cohort

3.6 A nomogram for predicting the prognosis of ccRCC

We constructed a nomogram that can predict the 1 -, 3 -, and 5-year OS for ccRCC. (Fig. 6A). It combines the expression levels of BID and three clinical characteristics with independent prognostic (age, histological grade, clinical stage), which can be better used for clinical assessment of individual survival probability. The calibration curve of the nomogram demonstrated good agreement between the actual survival rate of the patients and the predicted survival rate (Fig. 6B–D), revealing the superior predictive performance of the nomogram.

Fig. 6

The Nomogram for predicting prognosis of ccRCC. A Nomogram that can predict 1-, 3-, and 5-year survival rates of ccRCC. B–D The calibration curve of the nomogram demonstrated good agreement between the actual survival rate and predicted survival rate

3.7 The GSEA of BID

To research the potential mechanism of BID in the occurrence and development of ccRCC, we performed GSEA on the BID. The results manifested that BID overexpression was distinctly enriched in immune processes such as cytokine_cytokine_receptor_interaction, natural killer cell-mediated cytotoxicity, and antigen processing and presentation; while the down-regulation of BID was significantly enriched in insulin signaling pathway, WNT signaling pathway, renal cell carcinoma, pathways in cancer and MAPK signaling pathway (Fig. 7A).

Fig. 7

BID-related enrichment pathways and characteristics of immune infiltration. A The GSEA of BID. B Correlation between the expression of BID and the immune score of ccRCC samples. C Correlation between the expression of BID and the stromal score of ccRCC samples. D Differences in TME scores of ccRCC samples between BID low- and high-expression group. E The correlation between the expression levels of BID and the immune cells in TME. (F) Differences in infiltration of immune cells between BID high- and low-expression group (*: P < 0.05; **: P < 0.01; ***: P < 0.001)

3.8 BID-mediated tumor immune infiltration features

Accumulating evidence indicates that TME plays a central role in tumor initiation, progression and metastasis. The results of correlation analysis showed that the correlation coefficient between the expression of BID and immune score was 0.43 (Fig. 7B), and that between BID and stromal score was 0.21 (Fig. 7C), which indicates that the expression levels of BID was positively correlated with TME score in patients with ccRCC. Besides, the samples with high expression of BID had higher immune score, stromal score and ESTIMATE score (Fig. 7D , P < 0.001), indicating that the tumor tissue with high expression of BID may have higher content of immune cells and stromal cells, and higher tumor purity.

The infiltration of immune cells in TME is complex and diverse, which is intimately correlated to the efficacy of immunotherapy. The CIBERSORT analysis results demonstrated that the expression levels of BID was correlated with 10 kinds of immune cells in the TME (Fig. 7E). Specifically, the expression levels of BID was positively correlated with the infiltration levels of T cells regulatory (Tregs), T cells CD4 memory activated, Macrophages M0, T cells follicular helper, T cells gamma delta, plasma cells and T cells CD8; while that was negatively correlated with the infiltration levels of Mast cells resting, Dendritic cells resting and T cells CD4 memory resting. The radar map showed that the infiltration levels of Tregs, T cells gamma delta, T cells follicular helper, T cells CD8, T cells CD4 memory activated and plasma cells were higher in the BID high expression group, while the infiltration levels of Mast cells resting and Dendritic cells resting were decreased (Fig. 7F). Survival analysis of these differentially infiltrated immune cells revealed that high infiltration of Tregs and low infiltration of Mast cells resting and Dendritic cells resting were associated with poor prognosis, while the infiltration levels of T cells CD8, T cells gamma delta, T cells follicular helper, T cells CD4 memory activated and plasma cells had on correlation with prognosis of patients (Fig. 8).

Fig. 8

Survival analysis of of 8 kinds of immune infiltration cells

3.9 ssGSEA

The ssGSEA can evaluate the score and immune activity of immune infiltrating cells in tumor samples, so as to further explore the effect of BID on tumor immune microenvironment. The results of ssGSEA showed that the samples with high expression of BID usually had higher immune cell infiltration scores, and among the comparison of 16 types of immune cells, the infiltration scores of 14 types of immune cells were increased (Fig. 9A). In addition, the activity of immune-related pathways in samples with BID high expression was also higher. Except for the Type_II_IFN_Reponse, the activities of 12 immune pathways in samples with BID high expression were higher than those in samples with BID low expression (Fig. 9B).

Fig. 9

The results of ssGSEA for BID and the relationship between BID and common immune checkpoints. A The differences in enrichment fractions of 16 types of immune cells between BID high- and low-expression group. B The difference of the activity of 13 immune-related pathways between BID high- and low-expression group. C Correlation between BID and 9 immune checkpoints. D, E The expression levels of 10 immune checkpoints in BID high- and low-expression group (*: P < 0.05; ***: P < 0.001)

3.10 The relationship between BID and immune checkpoints

We evaluated the relativity between BID and common immune checkpoints. The results showed that 4 immunostimulatory molecules and 5 immunosuppressive molecules were significantly correlated with the expression of BID (Fig. 9C). Then, differential expression analysis was performed on these immune checkpoints between BID high- and low-expression group. We found that these 9 immune checkpoint genes were significantly up-regulated in samples with BID high expression, while down-regulated in the samples with BID low-expression (Fig. 9D, E, P  < 0.001).

3.11 Tumor mutation burden and somatic mutations

TMB is a key factor affecting tumor immune response and immunotherapy. We found that the expression levels of BID was positively correlated with TMB in ccRCC samples (Fig. 10A). The results of differential analysis showed that ccRCC samples with BID high expression had higher TMB compared with samples with BID low expression (Fig. 10B). Kaplan–Meier survival analysis demonstrated that patients in the TMB high group had a decreased OS compared with the TMB low group (Fig. 10C). Then, we used waterfall plot to show the mutation status of the 20 genes with the highest somatic mutation frequency between BID high- and low-expression groups (Fig. 10D, E). We found that missense mutations were the most common type in ccRCC samples, followed by frameshift deletions and nonsense mutations. The VHL mutation and PBRM1 mutation were the most common in both BID high- and low-expression groups. The biggest differences of mutations between groups were VHL, SETD2 and BAP1 mutations. Specifically, VHL mutation, SETD2 mutation and BAP1 mutation were more common in BID high expression group (52% vs. 42%, 17% vs. 7%, 15% vs. 5%).

Fig. 10

The relationship of BID to TMB and somatic mutations. A Correlation between BID and TMB of ccRCC samples. B The difference in TMB between BID high- and low-expression group. C The survival analysis of TMB. D Distribution of somatic mutations in samples with BID high expression. E Distribution of somatic mutations in samples with BID low expression

3.12 TIDE scores

The TIDE scores can predict the response of patients to ICIs. We found that the samples with BID high expression had higher TIDE scores (Fig. 11A), suggesting that such patients may benefit less from the treatment of ICIs. The T cell exclusion scores was lower in samples with BID high expression (Fig. 11B), while T cell dysfunction scores was higher (Fig. 11C). Finally, we also compared the differences of MSI between the two groups. The MSI was higher in samples with BID low expression (Fig. 11D).

Fig. 11

The TIDE scores of ccRCC. A Differences in TIDE scores of ccRCC samples between BID high- and low-expression group. B Differences in T cell exclusion scores of ccRCC samples between two groups. C Differences in T cell dysfunction scores of ccRCC samples between two groups. D Differences in MSI of ccRCC samples between two groups (*: P < 0.05; **: P < 0.01; ***: P < 0.001)

3.13 The results of immunohistochemical staining

We obtained the results of immunohistochemical staining of BID protein in ccRCC tissue and normal renal tissue from HPA database. From the Fig. 12A, B we can find that BID protein was highly expressed in ccRCC tissues, but low or not expressed in normal renal tissues. The results of immunohistochemical staining of tumor samples collected from the hospital also demonstrated that BID protein was highly expressed in the cytoplasm of ccRCC tissues, while low or no expression in normal renal tissues adjacent to cancer (Fig. 12C, D).

Fig. 12

The results of immunohistochemical staining of BID. A BID protein was highly expressed in ccRCC tissues in the HPA database. B BID protein was lowly or not expressed in normal kidney tissues in the HPA database. C BID protein was highly expressed in ccRCC specimens. D BID protein was lowly or not expressed in normal kidney specimens