Introduction

Renal cell carcinoma (RCC) is a prevalent malignant tumor of the urinary system, representing 2–3% of all malignancies in adults1. In 2018, more than 350,000 individuals worldwide were diagnosed with RCC, resulting in over 170,000 deaths2,3. Clear cell renal cell carcinoma (ccRCC), comprising 70–80% of all kidney cancers, is the most prevalent subtype of RCC4. Patients with ccRCC have higher tumor recurrence and metastasis rates than those with other renal cancer subtypes and generally exhibit poor prognoses5,6. Due to the high metastatic potential of renal cancer cells, patients are typically diagnosed at intermediate or advanced stages7. For newly diagnosed early RCC, the primary modality is surgical treatment, ranging from open surgery to laparoscopic minimally invasive surgery and robotic surgery, which brings significant advantages to patients. However, the prognosis of advanced metastatic RCC is poor and often lacks surgical indications; therefore, systematic therapy has become the main treatment option. Currently, targeted drugs used for the treatment of advanced RCC mainly belong to two categories: tyrosine kinase inhibitors (TKI), such as sunitinib, pazopanib, and acitinib8. The other group includes mammalian target of rapamycin (mTOR) inhibitors, such as everolimus and tisirolimus9. The advent of immunotherapy (alone or in combination) has revolutionized the treatment of advanced RCC. Immune checkpoint inhibitors (ICI) are effective therapeutic agents for various hematological and solid tumors10, including programmed cell death 1 (PD-1), programmed cell death ligand 1 (PD-L1), and cytotoxic T lymphocyte-associated antigen-4 (CTLA-4)10,11,12. However, prognostic and predictive biomarkers for immunotherapy in clinical practice are lacking. Additionally, ICI monotherapy and combination therapy were shown to be associated with higher levels of transaminemia in patients11. Patients who exhibit a poor response or resistance to immunotherapy face a risk of death, underscoring the need for effective solutions. Therefore, there is an urgent need to identify effective biomarkers to tailor treatment based on the patient and tumor characteristics, avoid unnecessary toxicity, and improve the patient’s quality of life13.

Previous research has demonstrated that tumor progression is not only related to the growth of the tumor itself but also to the tumor microenvironment (TME)14,15, which is a complex, highly heterogeneous, and dynamic system16,17. Immune cells infiltrating the TME contribute greatly to tumor development and assist tumor cells in their immune escape abilities18. Tumor cells can also interfere with the immune system by inducing and/or attracting immunosuppressive cells (mechanisms that prevent excessive inflammation or autoimmune disease) that impede antitumor activity of the immune system. Tumor-associated macrophages (TAM), Tregs, Myeloid-derived Suppressor Cells (MDSCs), Mast Cells, Neutrophils, B lymphocytes, and other immunosuppressive cells secrete several signaling molecules, including growth factors, chemokines, and cytokines that promote tumor progression. Secretion of angiogenic and/or invasive stroma-degrading enzymes, including MMP9 and other MMP molecules, induces and sustains tumor growth19. Attracted by chemokine ligands secreted by tumors and macrophages, Tregs infiltrate tumors and inhibit a variety of immunoactive cells, including CD8+ T cells, NK cells, B cells, and antigen-presenting cells20. In the TME, Tregs are activated by tumor-specific antigens or autoantigens released by tumor death, selectively inhibiting the activation of tumor-associated antigen (TAA)-specific effector T cells through a variety of mechanisms. The primary mechanism of tumor immune escape appears to be immunosuppression in the TME mediated by CD4, CD25–FoxP3–Tregs or other types of suppressor cells21. Phenotypic transformation of NK cells can inhibit immune cell function and promote tumor proliferation by expressing high levels of vascular endothelial growth factor (VEGF)22. High levels of TGF-β and anoxia in the TME may also induce NK cell transformation. CD8+ T cells play a major role in immunotherapy, mainly by interfering with their activity in treating patients with tumors23. In many tumors, the number of CD8+ T cells is reduced or inhibited by different mechanisms to induce immune escape. For example, TIRIB3 can reduce the infiltration of CD8+ T cells in colorectal cancer23, inhibit the response of CD8+ T cells to N6-methyladenosine-modified circIGF2BP3 in non-small cell lung carcinoma24, and promote CD8+ T cell rejection in Progranulin in breast cancer. These factors all induce immune escape25. TAM and dendritic cells (DC) in the TME can also inhibit tumor immunity26,27. M1-type macrophages promote antitumor immunity by presenting antigens, producing pro-inflammatory factors, and phagocytizing tumor cells, whereas M2-type macrophages promote tumor progression by upregulating cytokine secretion and protein expression. M0 macrophages have traditionally been thought to be precursors of polarized macrophages and have no specific functions. M2 macrophages enhance proliferation, migration, invasion, and epithelial–mesenchymal transition (EMT) of ccRCC cell line28. However, it has been found that macrophages infiltrating glioma tissue maintain a continuum between the M1 and M2 phenotypes, similar to M0 macrophages29. Moreover, the authors found that M0 macrophages in gliomas are associated with high-level tumors and poor prognosis, suggesting that M0 macrophages may play a tumorigenic role in tumor development30. Few studies have focused on the role of M0 macrophages in ccRCC tumorigenesis and development. ccRCC is sensitive to radiotherapy and chemotherapy and can benefit from immunotherapy. Therefore, this study analyzed the relationship between infiltrating M0 cells in the TME of ccRCC, patient prognosis, and immunotherapy.

This study focused on comprehensively characterizing immune cell infiltration in ccRCC and identifying immune cells and models relevant for prognosis and treatment. We investigated the abundance of immune-associated cells in ccRCC and evaluated their association with prognosis. We analyzed and filtered the genes in immune cells that had the most favorable prognosis between the groups with high and low infiltration. Additionally, risk models associated with immune cells were developed using key differentially expressed genes (DEGs). The sample was then divided into the high-risk group (HRG) and low-risk group (LRG) based on the risk score, and the survival status of the two groups was observed separately. To investigate the relationship between the risk score and immune characteristics, expression of ICs, IC50 values of targeted drugs, and immune efficacy, we performed a correlation analysis. Finally, immunohistochemistry was performed to verify the expression and functional roles of SAA1. This study investigated the potential role of M0 cells in the TME of ccRCC and the value of risk models for guiding patient medication and prognosis.

Methods

All methods were carried out in accordance with relevant guidelines and regulations.

Access to data

ccRCC sample data were obtained from The Cancer Genome Atlas (TCGA) (https://portal.gdc.cancer.gov/analysis_page?app=Downloads), Gene Expression Omnibus (GEO) database (cohort number: GSE40435) (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE40435), and ArrayExpress database (E-MTAB-1980) (https://www.ebi.ac.uk/biostudies/arrayexpress/studies?query=E-MTAB-1980). This comprehensive dataset encompasses RNA sequencing data, along with information on age, sex, and pathological stage, sourced from 541 ccRCC samples from TCGA, additional 101 samples from GEO, and 98 samples from ArrayExpress. Paraffin-embedded specimens of 66 patients with RCC (41 males and 25 females) and 18 normal renal tissue specimens and their corresponding clinical information were collected from the First Affiliated Hospital of Henan University of Science and Technology from January 1, 2020, to December 31, 2022. The 2009 TNM staging system was used for tumor staging. The tumor grade was determined according to the Fuhrman grading system (well-differentiated, grades 1 and 2; moderately differentiated, grade 3; and poorly differentiated, grade 4).

Analysis of immune cells

Firstly, we collected marker genes of 22 immune cells from the literature31. The abundance of immune cell infiltration in each sample was calculated according to the gene expression matrix (FPKM) through the R package CIBERSORT, setting perm = 1000. According to the infiltration abundance of each type of immune cells, the samples were divided into high and low infiltration groups, and then the survival differences between the two groups were compared by Kaplan–Meier (KM) survival analysis (Log-rank). p < 0.05 and |HR-1| > 0.5 were used to screen immune cells highly related to ccRCC.

DEGs analysis

The ccRCC cohort was categorized into high- and low-infiltration groups based on the extent of immune cell infiltration. The DEGs were analyzed using the Limma software package (version 4.22). We set the threshold of significance at p < 0.05 and |log fold change (FC)| > = 1.

Functional analysis

The DEGs identified in the previous step were entered into the STRING database, and a confidence threshold > 0.4 was set to construct the network. Subsequently, non-interacting genes were eliminated, fine-tuned, and embellished using Cytoscape (version 3.9.1). To further investigate the biological functions of these DEGs, we used the DAVID database for the enrichment analysis of Kyoto Encyclopedia of Genes and Genomes (KEGG)32,33,34 pathways, biological processes (BP), cellular components (CC), and molecular functions (MF). We considered terms with a p-value < 0.05 to be significant.

Development and validation of the immune-risk score

Univariate Cox regression was conducted to screen for prognostically relevant genes in the training cohort of ccRCC patients using the R survival package. To resolve issues related to collinearity and overfitting, the “glmnet” package (version 4.22) was employed to perform LASSO analysis on survival-related genes (p < 0.05) to filter predictors. Immunity-related risk models (IRRM) were constructed using multivariate Cox regression analysis based on the LASSO analysis results using the R package survival and survminer. Risk scores for all samples were calculated based on the weighted estimates of the expression of each gene. The ccRCC samples in both the training and validation cohorts were segregated into the LRG and HRG, respectively, based on the median risk score. KM survival analysis was used to investigate the prognostic value of survival in both the training and validation cohorts, with p < 0.05 as significance threshold. Additionally, the prognostic ability of the risk score was assessed using receiver operating characteristic (ROC) curves. We set the threshold of significance at p < 0.05 and AUC ≥ 0.65.

Creating a nomogram

Initially, we conducted Cox regression analyses of the selected genes to screen for independent prognostic factors for ccRCC. Therefore, we developed a nomogram to predict the likelihood of survival in patients with ccRCC. The method visually displays a set of scoring criteria derived from the magnitude of the regression coefficient for each variable and assigns a score to the level of each value for the respective variable. After calculating the total score for each patient using the method described above, we can approximate the prognostic probability for each patient using the function between the total score and prognostic probability35. Finally, to assess the efficacy of the nomogram, we generated corresponding calibration curves.

Calculation of enrichment scores for different genetic traits

A group of gene signatures was used to predict reactions to immunotherapy36. Furthermore, by tracking the tumor immunophenotype (TIP) (http://biocc.hrbmu.edu.cn/TIP/index.jsp), we obtained some gene features associated with the cancer immune cycle37. Pearson correlations were used to analyze the association between risk scores and gene features associated with cancer immune cycles using the R packages ggplot2 and ggcor.

Analysis of tumor immune dysfunction and exclusion (TIDE)

TIDE is primarily used to predict a tumor’s ability to survive an immune system attack based on its gene expression profile38. To calculate relevant immune parameters, such as microsatellite instability (MSI), TIDE, and CD8 and CD274 scores, and to predict ICIs response rates, we uploaded the gene expression files of the ccRCC samples to the TIDE website (http://tide.dfci). Unpaired t-tests were performed for tumor exclusion and TIDE scores between the HRG and LRG, while the association between CD8, CD274, and the risk score was analyzed using Spearman's correlation. A statistical significance level of p < 0.05 was set for all analyses.

Correlation between risk model and immunotherapy biomarkers

This study aimed to investigate the association between the immune risk score and IC. In addition, Pearson’s correlation was used to analyze the relationship between the risk score and five critical ICs (TIGIT, PDCD1, CTLA4, LAG3, and CD274) using the R packages circlize and corrplot. Additionally, we analyzed the differences in TIGIT, PDCD1, CTLA4, and LAG3 expression between the HRG and LRG.

Drug sensitivity analyses

The Oncoprecdict R package was used to predict drug responses among cancer patients in vivo39. It works by matching the expression data of the sample genes with the IC50 of the tumor cell line to the drug in the Genomics of Drug Sensitivity in Cancer. In this study, 283 drug sensitivities were compared between the HRG and LRG using an unpaired t-test set at p < 0.05.

Cell treatment

The cell lines (HEK-293, 786-O, 769-P, and THP-1) were obtained from HyCyteTM (Hai Xing, China). SAA1 siRNA was designed and synthesized by GenePharma and transfected according to the manufacturer's instructions. THP-1 cells in the logarithmic growth phase were treated with 100 ng/mL PMA and cultured for 48 h. THP-1 cells differentiated into wall-adherent macrophages.

Immunohistochemistry (IHC)

Paraffin sections were deparaffinized by heating. The sections were hydrated using a gradient of alcohol concentration. Sections were heated and boiled in a pot of antigen recovery solution for 15 min and then treated with 3% hydrogen peroxide and PBS. After drying, the primary antibody (1:400, Bioss bs-19359R) was added and incubated for 12 h. The next day, the primary antibodies were washed three times with PBS. Secondary antibody was added and incubated for 30 min before washing with PBS. Drops of DAB were added and observed. Staining was stopped if necessary, and the sections were rinsed with clean water. Finally, the dye was counterstained with hematoxylin.

Western blot

Cells were treated with RIPA lysis buffer (Solarbio) and centrifuged at 12,000 rpm for 30 min. The supernatants were collected for western blot analysis. Proteins in the supernatant were separated by SDS-PAGE (12%, Beyotime). The separated proteins were transferred to a membrane and sealed with nonfat dry milk. Primary antibody (abcam ab200485) was added and kept overnight at 4 °C. The following day, the primary antibody was washed off and incubated with a secondary antibody.

Proliferation experiment

The treated cells were seeded in 6-well plates at a density of 3 × 106 cells/well, and the cell state was stabilized. Experimental treatments were performed using the BeyoClickTM EDU-555 Cell Proliferation Assay Kit (Beyotime).

The prepared cells were transferred to 6-well plates at a density of 1000 cells per well. After 2 weeks of incubation, photographs were taken for observation.

Transwell

Matrigel gel, diluted 1:8 with serum-free cell culture medium, and 60 μL was taken and added evenly to the upper chamber, which was incubated for 3 h at 37 °C to polymerize the matrix gel into a thin film. Cells (5 × 104) were suspended in 150 μL serum-free medium and added to the upper chamber. 600 μL of medium was added to the lower chamber. After 48 h of incubation, the cells migrating across the membrane were fixed with formaldehyde and stained with 0.1% crystal violet. The migration assay did not require the Matrigel gel and the rest of the procedure was the same as that used for invasion.

Immunostaining evaluation

Staining results were evaluated by two professional investigators. The positive signal for SAA1 protein was localized in the nucleus and cytoplasm. Based on the proportion of positively stained cells in the total number of cells, positive cells were categorized and classified into four groups: 0 (0%), + 1 (1–5%), + 2 (5–10%), and + 3 (> 10%). SAA1 expression scores were 0 (negative), + 1 (weak), + 2 (moderate), and + 3 (strong). The two scores were multiplied to obtain the final score.

Statistical analysis

All data were analyzed using the R software (v4.1.3), and a t-test was used to analyze normally distributed data. Student’s t-test was used to compare two independent samples. Pearson’s correlation was used to analyze the correlation between the risk score, immune checkpoints, and MSI. The relationship between SAA1 expression and the clinicopathological features of ccRCC was analyzed using the Mann–Whitney U test. Statistical significance was set at p < 0.05.

Ethics approval and consent to participate

The ethics involved in this study was approved by the Ethics Committee of the First Affiliated Hospital of Henan University of Science and Technology. Confirmation that informed consent was obtained from all subjects and/or their legal guardian(s).

Results

Screening for immune cells associated with ccRCC prognosis

Initially, the ccRCC gene expression profile was translated into the degree of infiltration of 22 immune cells using CIBERSORT (Fig. 1A). Subsequently, we conducted KM survival analysis by the log-rank test and discovered that the prognosis of ccRCC patients was associated with five immune cells: mast cells (HR = 0.52, 95% [CI] = 0.39–0.7), dendritic cells (HR = 0.73, 95% [CI] = 0.54–0.99), M2 macrophages (HR = 0.65, 95% [CI] = 0.49–0.88), M0 macrophages (HR = 1.61, 95% [CI] = 1.2–2.17), and plasma cells (HR = 1.38, 95% [CI] = 1.02–1.85) (Fig. 1B). We then plotted survival curves to analyze the survival differences between the high- and low-infiltration groups of the five immune cell types. Our results demonstrated that patients with a high infiltration of M0 and plasma cells had a significantly poorer prognosis. In contrast, the other three cell types (resting mast cells, activated dendritic cells activated and macrophages M2) had significantly better prognoses in the high infiltration group than in the low infiltration group (Fig. 1C–G). In contrast, the other cell types did not show any significant differences, except for CD4+ memory-activated T cells, follicular helper T cells, and gamma delta T cells (Fig. S1A–N).

Figure 1
figure 1

Highly infiltrative M0 cells are associated with poor prognosis in ccRCC. (A) Gene expression matrices of ccRCC tissue in TCGA were converted to 22 immune cell infiltration abundances by CIBERSORT. (B) The effect of 77 significant immune cells on survival in ccRCC was analysed using the KM survival analysis. (C–G) Results of KM survival analysis of Mast cells, Dendritic cells, Macrophages M2, Macrophages M0, and Plasma cells in ccRCC.

Full size image

Analysis of DEGs between highly infiltrated and lowly infiltrated M0 cells

Expression data from ccRCC samples in the training cohort were translated to infiltration levels of 22 immune cells using CIBERSORT. Subsequently, based on the median level of M0 cell infiltration, the samples were categorized into high- and low-infiltration groups, resulting in 102 DEGs, including 11 upregulated and 91 downregulated genes (Fig. 2A,B) (Supplementary Dataset File 1). Most DEGs were highly expressed in highly infiltrated M0 cells as shown in Fig. 2B. Furthermore, the DEGs of the other four cell types were analyzed and found to be present only in resting mast cells, albeit in small numbers (Fig. S2A,B). Finally, a PPI network was constructed via PPI analysis, which indicated that 39 DEGs were associated with other genes (Fig. 2C).

Figure 2
figure 2

Identification of DEGs between the low and high infiltration groups of M0 cells in ccRCC. (A) The volcano plot was used visualize the DEGs (the red colour indicates upregulated genes, while blue colour indicates downregulated genes). (B) The heatmap plot was used visualize the DEGs. (C) The PPI network revealed associations between 39 genes associated with M0 macrophage infiltration.

Full size image

Prediction of DEG function from M0 cells

In total, 102 DEGs related to M0 cells were subjected to GO and KEGG enrichment analyses to identify the pathways in which they were enriched. The findings demonstrated that these DEGs were predominantly clustered in pathways associated with “neutrophil chemotaxis,” “glucose transmembrane transport,” “insulin-like growth factor I binding,” and “metabolic pathways” (Fig. 3A–D). In addition, we present the results of GO analyses (BP, MF, and CC) using chord and line plots (Fig. S3A–C). Our findings suggest that M0-related DEGs are predominantly involved in metabolic pathways.

Figure 3
figure 3

Functional analysis of DEGs in M0 cells. (A–D) KEGG and GO analysis of 39 genes associated with M0 macrophages.

Full size image

Risk model based on M0-related DEGs

First, we conducted univariate COX analysis on the 102 M0-related DEGs to determine their prognostic significance (Supplementary Dataset File 2). Among these, 11 genes were identified and 10 were randomly selected for visualization (Fig. 4A). Our findings revealed that SAA1 and PPP1R1A were associated with a poor overall survival, whereas the other eight genes were associated with a better overall survival (Fig. 4A). Subsequently, we performed a LASSO analysis to identify the three potential candidates with the smallest lambda values and used them to construct a model of M0-related risk (Fig. 4B,C). Third, we performed multivariate COX analysis on these three M0-associated DEGs and eventually constructed a risk model based on these genes (TMEM174, LRRC19, and SAA1) (Fig. 4D) (Supplementary Dataset File 3). Finally, we categorized the samples into two subgroups based on the expression level of the three risk genes, after which we conducted a KM prognostic analysis and generated a survivorship curve. The high and low expression subgroups of the three model genes showed significant differences in survival. Specifically, populations with high expression of TMEM174 and LRRC19 had a better prognosis than those with low expression (Fig. 4E,F), the SAA1 high expression group had worse survival than those with low expression (Fig. 4G).

Figure 4
figure 4

Develop an immunization riskscore. (A) Univariate Cox regression analysis was conducted on 102 genes associated with M0 macrophage infiltration. (B,C) Lasso analysis of other key genes linked to ccRCC survival. (D) A multivariate Cox regression analysis was performed on six genes that are linked to M0 macrophage infiltration. (E–G) Survival analyses showed survival differences between the low and high expression groups of TMEM174, LRRC19 and SAA1 in the training cohort.

Full size image

Constructing a nomogram from risk models and clinicopathological features

To further analyze whether risk scores have an impact on the prognosis of ccRCC patients, we used univariate and multivariate regression analyses to determine whether risk scores are independent prognostic factors (Supplementary Dataset File 4 and 5). The findings revealed that risk score, age, and pathological stage were independent prognostic factors in ccRCC patients (Fig. 5A). Immediately following this, we integrated multiple metrics, including risk scores, age, and pathological staging and plotted nomograms to describe the intervariable correlations (Fig. 5B). The calibration curves showed that the nomogram with risk scores had high predictive power for patient survival at 1, 3, and 5 years (Fig. 5C–E).

Figure 5
figure 5

Constructing the nomogram. (A) A multivariate Cox regression analysis was conducted on the riskscore, age, sex, pathological stage, and previous malignancy. (B) A nomogram was constructed using risk score, age, sex, pathological stage and previous history of malignancy. (C–E) Calibration plot for nomogram in the test group.

Full size image

IRRM showed high prognostic value

To further validate the predictive value of the risk model, we divided the ccRCC samples in the training cohort into the HRG and LRG based on the median risk score and conducted a prognostic analysis (Fig. 6A). Our findings revealed that the HRG had a worse overall survival rate (Fig. 6B) and a correspondingly higher mortality rate (Fig. 6C) than the LRG. Furthermore, we assessed the prognostic ability of IRRM using ROC curves, which showed that the sensitivity of IRRM in predicting overall survival at 1, 3, and 5 years was 0.766, 0.722, and 0.708, respectively (Fig. 6D). Furthermore, we observed different trends for the three model genes, with downregulated TMEM174 and LRRC19 expression downregulated and SAA1 expression upregulated in the HRG (Fig. 6E). These results demonstrate the excellent prognostic predictive value of risk models based on M0 in RCC.

Figure 6
figure 6

Risk models demonstrate high predictive value in training set. (A) Based on the results of the scoring, we divided the training cohort into a HRG and a LRG. (B) KM survival analysis results indicated differences between HRG and LRG. (C) The scatter plot results show surviving and dying cases in the HRG and LRG. (D) Time-dependent ROC analysis of immune risk models at 1, 3 and 5 years in the training cohort. (E) Expression of SAA1, TMEM174, LRRC19 in the HRG and LRG in the training cohort.

Full size image

Risk score correlated with tumor stage and grade in the validation cohort

We rechecked the stability of the IRRM using another dataset (GSE40435). A Sankey diagram was created to examine the relationship between the risk score and various stages of tumor development. Our results indicated that the majority of stages 3/4 and T3/4 corresponded to the HRG (Fig. 7A). Similarly, grades III/IV were markedly more common in the HRG than in the LRG, whereas the reverse was true for grades I/II (Fig. 7B). However, there were no apparent differences in age or sex between the HRG and LRG (Fig. 7B). These findings suggest that the HRG had a poorer prognosis, which is the same as the results of the IRRM analyzed in the previous section. Furthermore, the expression trends of the three genes in the validation set were consistent with those in the training set. Specifically, TMEM174 and LRRC19 expression were downregulated in the HRG, whereas the expression of SAA1 was upregulated (Fig. 7B). In addition, another validation cohort was used to verify the robustness of the constructed model, and the samples of the validation cohort were divided into the HRG and LRG (Fig. S4A). The results showed that the HRG was associated with a worse prognosis (Fig. S4B), and higher mortality (Fig. S4C) compared with the LRG in the validation cohort. The ROC curve showed that the sensitivities of IRRM in predicting 1, 3, and 5 years overall survival were 0.611, 0.601, and 0.591, respectively (Fig. S4D). TMEM174 and LRRC19 were downregulated and SAA1 was upregulated in the HRG (Fig. S4E). In the validation cohort, KM survival results showed that the low TMEM174 and LRRC19 expression group had a lower survival probability than the high expression group (Fig. S4F,G), while the SAA1 low expression group had a significantly higher survival probability than the high expression group (Fig. S4H).

Figure 7
figure 7

Riskscore correlated with ccRCC tumour stage and grade in the validation cohort. (A) Sankey demonstrates the relationship between risk score and tumour staging and grading in the validation cohort. (B) Clinical relevance of heat maps in relation to risk scores and tumour grading in a training cohort.

Full size image

Risk models related to immunity and the cancer-immunity cycle

We then demonstrated the potential link between the risk score and the cancer immunity cycle. Our findings revealed that the risk score exhibited a positive correlation with the GSVA enrichment score for all positive signals related to immunotherapy, except for the APM signal, suggesting that ccRCC may be closely associated with immunity and respond to immunotherapy (Fig. 8A).

Figure 8
figure 8

The immune risk model predicts the effect of immunotherapy and the cancer immune cycle. (A) The immune risk score correlates with the enrichment score of the immunotherapy prediction pathway. (B) Immunological risk scores are associated with the cancer-immunity cycle process.

Full size image

Furthermore, we observed that the risk score was inversely linked with some functions of the tumor immunity cycle, such as th17 cell recruiting, th22 cell recruiting, Treg cell recruitment, and infiltration of T-cells into the tumor, while it was positively correlated with most of the steps of the cancer-immune cycle, including the release of cancer cell antigens, macrophages, MDSC, neutrophils, B cells, basophil cells, CD8+ T, NK, Th1, and Th2 cell recruitment (Fig. 8B).

Risk models related to immunotherapy

To investigate the association between the risk score and immunotherapy, we examined the immunological features of the two groups using TIDE. Our results indicated that the risk score increased with an increase in the expression of CD8 (Fig. 9A) and CD274 (Fig. 9B); however, the trend was opposite to that of MSI (Fig. 9C). Additionally, we performed TIDE analyses and found that exclusion, TIDE scores, and dysfunction differed between the two groups, with higher scores observed in the HRG (Fig. 9E–G). These findings suggest that the HRG of patients with ccRCC may have a strong immune escape, leading to poorer ICIs treatment outcomes. Furthermore, consistent with these findings, the percentage of individuals who did not respond to ICIs treatment was higher in the HRG (75.4%) than in the LRG (67.7%) (Fig. 9D).

Figure 9
figure 9

Samples from the HRG of ccRCC were more sensitive to ICIs treatment. (A–C) Correlation between risk scores and CD8, CD274 and MSI scores. (D) Prediction of sample sizes of non-responders and responders to ICIs treatment in HRG and LRG. (E–G) Differences in TIDE scores, Exclusion scores and Dysfunction scores between the HRG and LRG.

Full size image

Relevance of risk models to IC

Next, we investigated the relationship between IRRM and ICs. Our findings revealed that the risk score was positively correlated with most ICs, including TIGIT, CTLA4, LAG3, CD274, and PDCD1 (Fig. 10A,B). Moreover, we compared the expression levels of LAG3, CTLA4, PDCD1, and TIGIT between the LRG and HRG. The expression of these ICs was remarkably higher in HRG than in LRG, as observed (Fig. 10C–F).

Figure 10
figure 10

The relationship between riskscore and ICs. (A) Correlation between immune risk scores and ICs. (B) The correlations between immune-risk score and common ICs. (C–F) Differences in the expression of LAG3, CTLA4, PDCD1 and TIGIT genes in the HRG and LRG.

Full size image

Risk models relevant to targeted therapy

To explore the sensitivity of HRG and LRG patients to targeted drugs, we used the Oncopredict algorithm to convert the sensitivity scores of 349 drugs based on the gene expression matrix in ccRCC. Supplementary Dataset File 6 shows the scores for each sample. Our analyses revealed a substantial difference in sensitivity to 283 drugs between the HRG and LRG (Fig. 11A), with the HRG being more resistant to 255 drugs (Fig. 11A). Moreover, we identified 28 drugs, including Axitinib, Pazopanib, Temsirolimus, and Sunitinib, with lower IC50 values in the HRG than in the LRG (Fig. 11B–E).

Figure 11
figure 11

Targeted therapy for LRG and HRG. (A) Heat map plots show the differences in drug sensitivity between the HRG and LRG. (B–E) The box plot shows the difference in IC 50 between Axitinib, Pazopanib, Temsirolimus and Sunitinib in the HRG and LRG.

Full size image

Immunohistochemical detection of SAA1 expression

Immunohistochemical staining demonstrated that SAA1 was absent in normal tissues but moderately or severely expressed in the immune cells of grade I, II, III, and IV tumor tissues and was located in the nucleus and cytoplasm (Fig. 12A). SAA1 was specifically expressed in RCC tissues and was significantly correlated with metastasis (p = 0.007), tumor stage (p = 0.022), and tumor grade (p = 0.015) (Table 1).

Figure 12
figure 12

Functional validation of model genes. (A) Immunohistochemical results of SAA1 in tissues. (B,C) Differential expression of SAA1 in normal and tumour cells. (D,E) Changes in SAA1 protein and RNA levels in cells after SAA1 knockdown. (F,G) Knockdown of SAA1 reduced the proliferative capacity of the cells. (H,I) After knocking down SAA1, cell migration and invasion abilities were decreased. Due to performing multiple sets of experiments on a single blot, the full-length blots had to be cropped. However, we provide the results of three replicate experiments in Supplementary Figs. 5, 6 and 7. SAA1 and GAPDH were probed on the same gel after cutting into half.

Full size image
Table 1 Correlation between clinicopathological and SAA1 expression.
Full size table

SAA1 promotes cell proliferation

To verify whether macrophages can induce SAA1 expression in renal tumor cells. Co-incubation of renal tumor cells with macrophage medium promoted SAA1 expression in renal tumor cells (Fig. 12B,C). First knock down SAA1 expression in macrophages (Fig. 12D,E). Tumor cells were treated with control and knockdown SAA1 macrophage medium, and it was found that the knockdown of SAA1 macrophage medium significantly reduced the ability of tumor cells to proliferate, migrate, and invade (Fig. 12F–I), suggesting that SAA1 plays a role in promoting tumor growth.

Discussion

The formation of an immunosuppressive TME is an indicator of tumor progression40. In this environment, tumor cells evade the immune system, which is a necessary precondition for tumor progression. Additionally, tumor cells must interconnect with other cells in the TME to provide the necessary signaling molecules and pathways. Macrophages are found in almost all body tissues, are highly flexible and differentiate into M1 and M2 under different conditions41. M0 macrophages, which are derived from the bone marrow, are generally believed to be precursors for M1 and M2 cells and are only in a quiescent state42. M1-like macrophages primarily eliminate pathogens and tumor cells, whereas M2-like macrophages are tumor-associated macrophages that play crucial roles in tumor development and immunosuppression43. Studies have indicated that Malignant gliomas are rich in M0 macrophages in addition to M1 and M2 cells that exhibit polarization44. It has even been suggested that M0 macrophages can be used to characterize glioblastoma malignancy44. Additionally, Liu et al. found that M0 macrophage infiltration was significantly higher in endometrial cancer45. In patients with high-grade tumors, M0 macrophages are associated with poor prognosis, indicating a role for M0 macrophages in promoting tumor growth42. Additionally, relevant studies have shown that elevated M0 cell levels in patients with early stage lung adenocarcinoma are correlated with reduced survival and poor patient prognosis46. Our results are consistent with these findings, as we showed that M0 macrophages are abundant in ccRCC tissues and are strongly associated with patient prognosis. Our analysis based on M0 macrophages showed that patients with a low abundance had a better prognosis than those with a high abundance.

SAA1 is an acute-phase reactive protein produced mainly in the liver that increases in response to injury and infection47. Elevated SAA1 levels are associated with many pathological conditions, including atherosclerosis and amyloidosis48,49. As an apolipoprotein, SAA regulates cholesterol metabolism by modulating lecithin/cholesterol acyltransferase activity or by altering the affinity pathway of high-density lipoprotein (HDL) for macrophages50,51. SAA1 also plays an important role in the immune system and serves as a diagnostic and prognostic marker for several diseases52. In glioblastoma and ovarian cancer, SAA1 shows promise as a potential therapeutic target and prognostic marker linked to immune infiltration53,54. Additionally, recent research on SAA1 has found that hypermethylation of a part of its promoter region is responsible for the significant upregulation of both transcription and protein expression in advanced RCC, indicating its diagnostic and prognostic value55. LRRC19 is a transmembrane receptor of the leucine-rich repeat (LRR) family that regulates the recruitment of immune cells to intestinal inflammation via associated chemokines56. Chai et al. proposed that LRRC19 induces pro-inflammatory cytokine expression through NF-κB activation and is involved in the inflammatory response57. In some cancers, such as ovarian, gastric, breast, lung, and colorectal cancers, a significant association between the downregulation of LRRC19 expression and poor patient prognosis has been observed58. According to the relevant literature, LRRC19 serves as both a prognostic indicator in RCC and a potentially useful prognostic and adjuvant therapeutic marker in ccRCC59. TMEM174, a recently discovered protein, consists of 243 amino acids and is highly expressed in human kidney tissues. It promotes cell proliferation by activating AP-1 and regulates NaPi2a, which is considered a regulator that prevents hyperphosphatemia60,61. Additionally, TMEM174 is differentially expressed in various RCC subtypes, with high expression in squamous cell carcinoma and migratory cell carcinoma and low expression in ccRCC, retroperitoneal metastatic ccRCC, and adrenal metastatic ccRCC62. In our investigation, we used differential gene analysis, multivariate regression analysis, and LASSO analysis to construct risk models for SAA1, TMEM174, and LRRC19 in M0 cells with high and low infiltration. Survival analysis of the three genes revealed that high expression of TMEM174 and LRRC19 indicated good prognosis, indicating their role as protective genes, whereas high expression of SAA1 indicated poor prognosis, implying that they function as risk genes. These results are consistent with those of previous studies, further emphasizing the significant roles of these genes in ccRCC. Furthermore, we conducted multivariate regression and prognostic analyses on the model based on SAA1, TMEM174, and LRRC19 and discovered that the risk score was an independent factor for evaluating the prognosis of patients with ccRCC, with those in the HRG having a poorer prognosis. ROC analysis indicated an AUC greater than 0.7, implying that the risk model had good predictive efficacy for 1-, 3-, and 5-year overall survival.

Immunotherapy and targeted therapies have made significant progress in recent years and show great potential for treating a wide range of cancers. Owing to the low efficacy of radiotherapy and chemotherapy, these treatments have become important options for patients with RCC63. The VEGF signaling pathway is a key driver of tumor vascular development and its inhibition is one of the primary therapeutic approaches for advanced RCC64. For instance, sunitinib is known to inhibit VEGF and PDGF receptors and has shown considerable improvement in the progression-free survival of patients65. Furthermore, pazopanib inhibits tyrosine kinase, which targets VEGF receptors 1, 2, and 3, PDGF receptors, and other signaling pathways. In a Phase III trial of first-line treatment, pazopanib demonstrated similar effectiveness to sunitinib in improving progression-free survival66. Moreover, inhibition of the mTOR pathway can have a profound impact on the functions of renal cancer cells such as cell growth, proliferation, metabolism, and angiogenesis67. Sirolimus is a well-known inhibitor of the mTOR pathway, which is downstream of the PI3K and Akt pathways. A Phase III trial involving 626 previously untreated low-risk patients confirmed the effectiveness and safety of tisirolimus, with a significantly improved overall survival benefit in the tisirolimus group as compared to IFN-α68. Tumor cells evade the immune system by interacting with PD-1 and PD-L1, leading to the development of PD-1 or PD-L1 checkpoint inhibitors with promising efficacy69. Tremendous advances in ICIs therapy have opened new therapeutic prospects for patients with cancer, and ICIs are now extensively used in the treatment of various cancer types. The main immune checkpoints are PD-L1, PD-1, and CTLA-4. RCC is highly immunogenic, with approximately 30% of RCC cases showing PD-L170. Opdivo is an immune checkpoint inhibitor antibody that is a humanized monoclonal immunoglobulin with a selective affinity for PD-1 expressing immune cells. This innovative therapy has shown promise and has been approved for the treatment of metastatic melanoma and non-small cell lung cancer71. Studies using Opdivo in treatment (CheckMate-025) have demonstrated significantly superior overall survival, hazard ratio, and objective response rate (ORR)71. However, it is important to note that although these treatments have improved the outcome for patients with RCC, relapse or progression of the disease still occurs occasionally, and targeted therapies and immunotherapies are less effective in a subset of patients. Consequently, identifying potential targets or immune-associated genes with significant therapeutic value is critical to improve the prognosis of patients with advanced RCC.

TIDE is a recently developed computational approach designed to predict a sample's response to ICIs treatment by combining two parameters: the T-cell dysfunction score and T-cell rejection score. However, in ccRCC, immune escape mainly occurs through T-cell dysfunction, resulting in higher T-cell levels and dysfunction scores38. Additionally, researchers have observed a negative correlation between T cell dysfunction and exclusion in five types of cancer38. Another important factor related to the efficacy of ICIs therapy is the MSI score72. In our study, we evaluated the correlation between risk scores and ICs, including TIGIT, CTLA-4, LAG3, PD-L1, and PDCD1, and found a positive correlation between them. In addition, we observed a negative correlation between MSI and risk scores.

Similar to previous findings, our study confirmed the presence of significant immune escape in the HRG, as evidenced by higher TIDE scores, exclusion, and T-cell dysfunction. Interestingly, we also found that the proportion of patients responding to ICIs treatment in the HRG was lower than that in the LRG, indicating that patients in the LRG may be more responsive to ICIs treatment and potentially benefit more from it. Finally, we performed sensitivity analyses of the targeted drugs in HRG and LRG patients using the OncoPredict algorithm. We calculated the sensitivity scores for 349 drugs and found that for 283 drugs, there was a difference in sensitivity between the HRG and LRG. The HRG exhibited the greatest resistance to 255 drugs. Meanwhile, the HRG was more sensitive to treatment with 28 drugs (axitinib, pazopanib, temsirolimus, and sunitinib) and had lower IC50 values than the LRG. In conclusion, our study suggests that although there is significant immune evasion in ccRCCs with HRG, their response to immunotherapy is not as favorable as that in ccRCCs with LRG. Moreover, their sensitivity to commonly targeted drugs is poor, leading to poor prognosis in patients with LRG.

As a risk gene, SAA1 may be closely associated with tumor formation and progression. Therefore, we verified the expression of SAA1 at the protein level using immunohistochemistry of the collected paraffin sections. Our results indicated that SAA1 was expressed only in tumor tissues and was associated with tumor metastasis, tumor T stage, and tumor grade, which is also consistent with our analysis. Furthermore, studies that interfere with SAA1 have suggested that SAA1 promotes the proliferation and invasion of renal tumor cells.

Unlike Fu et al.73, who developed a risk model for ccRCC based on genes related to necroptosis and pyroptosis, we developed a risk model based on genes related to tumor infiltration M0 macrophages. Fu et al.73 screened 10 genes related to necroptosis and pyroptosis to construct a risk model based on TCGA and GEO ccRCC datasets. Lin et al.74 used TGCA, ArrayExpress and checkmat-025 (CM-025) cohort datasets to analyze the characteristics of fatty acid metabolism-related mRNA, and screened 6 fatty acid metabolisM-related genes for the construction of risk models. Bao et al.75 screened 6 necroptosis related miRNAs from TCGA database to construct the risk model. In our study, we comprehensively used ccRCC datasets from TCGA, GEO and ArrayExpress databases to construct a risk model for related research. The AUC values of the risk model constructed by Fu et al.73 for predicting the 1-, 3-, and 5-year survival probability of ccRCC patients were 0.763, 0.707, and 0.728, respectively. The AUC values of the model constructed by Lin et al.74 for predicting the 3-, 5-, and 7-year survival probability of ccRCC patients were 0.745, 0.765, and 0.742, respectively. The AUC values of Bao et al.75 model in predicting the 1-, 3-, and 5-year survival probability of ccRCC patients were 0.721, 0.711, and 0.703, respectively. The AUC values of our model for predicting the 1-, 3-, and 5-year survival probability of ccRCC patients were 0.766, 0.722, and 0.708, respectively, which was similar to the prediction ability of Fu et al.'s model. Fu et al.73 showed that infiltrating M0 macrophages were positively correlated with the risk score, and the expression levels of model genes AIM2, CASP4, GSDMB, NOD2 and RBCK1 in tumor tissues were higher than those in adjacent normal tissues, among which CASP4 and GSDMB promoted the proliferation, migration and invasion of ccRCC cells. Lin et al.74 found that the expression levels of model genes ABCD1, ALOX12B, ALOX15B, CPT1B, and IL4I1 were up-regulated in tumor tissues, whereas HACD1 expression was lower in ccRCC tissues and the proportion of infiltrating M0 in high-risk group was higher than that in low-risk group. Furthermore, knockdown of IL4I1 in tumor cells inhibited the differentiation of M0 into M2 and inhibited the proliferation, migration and invasion of ccRCC cells. Bao et al.75 showed that the expression of risk model gene has-miR-193a-3p was significantly increased in ccRCC tissues, while the expression of miR-214-3p was significantly decreased. Our results showed that high M0 infiltration was associated with poor prognosis of ccRCC patients, and the expression of model gene SAA1 in tumor tissues was higher than that in normal tissues. Knockdown of SAA1 could inhibit the proliferation and migration of ccRCC tumor cells. Fu and Lin et al.73,74 also found that the infiltrating abundance of M0 was correlated with the survival prognosis of ccRCC patients. Fu et al.73 showed that ccRCC patients in high score group were sensitive to ICI, targeted drugs as well as rapamycin and temsirolimus, suggesting that these patients may achieve better results by receiving targeted therapy combined with immunotherapy. Lin et al.74 found that the risk score can predict the clinical efficacy of ICI-based immunotherapy, and ccRCC patients in the low-risk group are more sensitive to ICI. These findings are helpful for the screening of ccRCC therapeutic drugs and the precision treatment of ccRCC patients. Bao et al.75 showed that the target genes of miRNA TGFBR3, SIX4, DUSP1 and ARHGAP42 were positively correlated with drug sensitivity, while TAL1 and BTG2 were negatively correlated. The above results suggested that we can choose different therapeutic methods according to the expression level of related model genes in ccRCC patients in the future, and may achieve better results. For example, we can assess the prognosis of ccRCC patients by examining the abundance of M0 macrophage infiltration and the expression of model related genes in ccRCC tissues. For patients with high M0 infiltration or high expression of model genes, the prognosis may be poor, and intensive management combined with multiple combination therapies may improve the prognosis. In addition, more sensitive to targeted therapy in ccRCC patients with LRG, HSG patients may have a better response to immunotherapy. However, these need to be further confirmed in future studies.

In our study, the risk model built based on DEGs between the high-infiltration and low-infiltration M0 groups showed high prognostic value in both the training and validation datasets and was closely related to targeted therapy and immunotherapy of RCC. Therefore, in clinical practice, we can first assess the prognosis of RCC patients by detecting the infiltration abundance of M0 macrophages and the expression of related genes in the tumor tissues of patients. For the group with high infiltration of M0, the prognosis may be poor; therefore, we can improve the prognosis by strengthening management combined with multiple treatment combinations. According to our findings, patients in the LSG were more sensitive to targeted therapies, including Axitinib, Pazopanib, Temsirolimus and Sunitinib. Patients with HSG have a poorer prognosis and may need to respond better to immunotherapy. However, relevant large-sample multi-center verification and more basic research is needed to clarify the specific mechanism by which M0 affects ccRCC for our study. Moreover, M0 key genes may be transformed into new RCC tumor biomarkers for the diagnosis and prognosis of RCC in the future. For example, M0 macrophage-associated SAA1 is a potential new molecular target for RCC therapy. In the future, the relationship between infiltrating immune cells in the tumor microenvironment and tumor progression will gradually attract attention. M0 macrophages can differentiate into M1 or M2 cells when stimulated by different tumor microenvironments and signals. Further analysis of the relationship between M0, M1, and M2 macrophages in ccRCC is required. Owing to the differences in infiltrating immune cells in the tumor microenvironment between individuals, it is of great significance to guide clinical treatment and prognosis according to the differences in immune cell infiltration while grasping the common characteristics of the tumor microenvironment between individuals.

In conclusion, the risk model constructed using M0 macrophage-associated genes in the ccRCC tumor microenvironment can predict the prognosis of patients with CCRCC and the effect of immunotherapy. In addition, we found that the model gene SAA1 possessed the ability promoted clear renal cell proliferation and migration. This discovery may offer novel insights and avenues for prognostic prediction and therapeutic approaches for RCC. However, we must also clearly recognize the shortcomings in our work. First, the data used in this study were obtained from public databases, which may suffer from data bias and insufficient sample size. Second, the results of this study need to be validated in different cohorts and large samples. Finally, this study only verified the functional role of the genes at the cellular level, and the results need to be further verified by in vivo experiments.