Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2024 Feb 5:15:1325447.
doi: 10.3389/fphar.2024.1325447. eCollection 2024.

Identification of molecular subtypes and diagnostic model in clear cell renal cell carcinoma based on collagen-related genes may predict the response of immunotherapy

Affiliations

Identification of molecular subtypes and diagnostic model in clear cell renal cell carcinoma based on collagen-related genes may predict the response of immunotherapy

Yulong Hong et al. Front Pharmacol. .

Abstract

Background: Collagen represents a prominent constituent of the tumor's extracellular matrix (ECM). Nonetheless, its correlation with the molecular subtype attributes of clear cell renal cell carcinoma (ccRCC) remains elusive. Our objective is to delineate collagen-associated molecular subtypes and further construct diagnostic model, offering insights conducive to the precise selection of ccRCC patients for immunotherapeutic interventions. Methods: We performed unsupervised non-negative matrix factorization (NMF) analysis on TCGA-KIRC samples, utilizing a set of 33 collagen-related differentially expressed genes (33CRDs) for clustering. Our analysis encompassed evaluations of subtype-associated differences in pathways, immune profiles, and somatic mutations. Through weighted gene co-expression network analysis (WGCNA) and four machine learning algorithms, two core genes were found and a diagnostic model was constructed. This was subsequently validated in a clinical immunotherapy cohort. Single cell sequencing analysis and experiments demonstrated the role of core genes in ccRCC. Finally, we also analyzed the roles of MMP9 and SCGN in pan-cancer. Results: We described two novel collagen related molecular subtypes in ccRCC, designated subtype 1 and subtype 2. Compared with subtype 1, subtype 2 showed more infiltration of immune components, but had a higher TIDE (tumor immunedysfunctionandexclusion) score and increased levels of immune checkpoint molecules. Furthermore, reduced prognosis for subtype 2 was a consistent finding in both high and low mutation load subgroups. MMP9 and SCGN were identified as key genes for distinguishing subtype 1 and subtype 2. The diagnostic model based on them could better distinguish the subtype of patients, and the differentiated patients had different progression free survival (PFS) in the clinical immunotherapy cohort. MMP9 was predominantly expressed in macrophages and has been extensively documented in the literature. Meanwhile, SCGN, which was overexpressed in tumor cells, underwent experimental validation, emphasizing its role in ccRCC. In various cancers, MMP9 and SCGN were associated with immune-related molecules and immune cells. Conclusion: Our study identifies two collagen-related molecular subtypes of ccRCC and constructs a diagnostic model to help select appropriate patients for immunotherapy.

Keywords: clear cell renal cell carcinoma; collagen; diagnostic model; immunotherapy; machine learning; molecular subtypes.

PubMed Disclaimer

Conflict of interest statement

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Figures

FIGURE 1
FIGURE 1
Screening and analysis of 33 CRGs required for subtype identification. (A, B) CRGs Differentially expressed between tumor tissues and normal tissues. (C) 33CRDs correlated with OS in ccRCC. (D) PPI network, core network and core genes of 33CRDs. (E) Frequencies of CNV gain, loss, and non-CNV among 33CRDs. (F) Circus plots of chromosome distributions of 33CRDs. CRGs, collagen-related genes; 33CRDs, 33 collagen-related DEGs; OS, overall survival; CNV, copy number variation.
FIGURE 2
FIGURE 2
Identification of collagen subtypes in KIRC. (A) The cophenetic correlation coefficient is for optimal number of subtypes. (B) Consensus matrix of the molecular subtypes: subtype 1 and subtype 2. (C) t-SNE scatterplot supports ccRCC subtypes based on mRNA expression profiles. (D) Kaplan-Meier OS curves for subtype 1 and subtype 2. (E) Expression differences of 33CRDs between subtype 1 and subtype 2.
FIGURE 3
FIGURE 3
Difference analysis and GSEA between subtype 1 and subtype 2. (A) DeSeq2 differential analysis heatmap and corresponding clinical information between subtype 1 and subtype 2. (B) GSEA results on pathways in GO, including BP, MF and CC. (C) GSEA results on pathways in KEGG. GSEA, gene set enrichment analysis; GO, gene ontology; KEGG, Kyoto-encyclopedia of genes and genomes; BP, biological process; MF, molecular function; CC, cellular component; *p < 0.05; **p < 0.01; ***p < 0.001; not significant, p > 0.05.
FIGURE 4
FIGURE 4
Immune infiltration analysis of collagen-associated ccRCC subtypes. The immune score (A), stromal score (B), ESTIMATE score (C) and tumor purity (D) between subtype 1 and subtype 2. Comparisons of immune cells (E) and immune functions (F) between subtype 1 and subtype 2. (G) The differences in the TIDE score between subtype 1 and subtype 2. (H–M) Differences in expression of six molecules related to Immune evasion and T cell exhaustion compared between subtype 1 and subtype 2. TIDE, tumor immunedysfunctionandexclusion.
FIGURE 5
FIGURE 5
Analysis of TMB characteristics. (A) Comparison of TMB between subtype 1 and subtype 2. Waterfall maps of the somatic mutations in the subtype 1 (B) and the subtype 2 (C). (D) Difference in OS between high TMB and low TMB groups. (E) Difference in OS based on TMB and two subtypes. TMB, tumor mutation burden. OS, overall survival.
FIGURE 6
FIGURE 6
Identification of core genes that differentiate subtypes. (A) Scale independence and mean connectivity analyzes are used to determine the optimal soft threshold. (B) Gene dendrogram as a result of clustering, where colored rows below the dendrogram indicate different modules. (C) Heatmap of correlations between modules and two subtypes. (D) RDCP for RF, SVM, XGB, and GLM, each curve represents a model. (E) BPR for RF, SVM, XGB, and GLM, each boxplot represents a model. (F) ROC represents the discriminative performance of the four machine learning models for subtype 1 and subtype 2 in the validation set. (G) The most important top ten genes of the three models with significantly high and similar predictive performance are intersected. RDCP, reverse cumulative distribution plot; RF, random forest; SVM, support vector machine; XGB, extreme gradient boosting; GLM, generalized linear model; BPR, boxplot of Residuals; ROC, receiver operating characteristic.
FIGURE 7
FIGURE 7
Diagnostic model based on MMP9 and SCGN can predict immunotherapy efficacy. (A, B) The expression level of MMP9 and SCGN in subtype 1 and subtype 2. (C) Nomogram of diagnostic model. (D) ROC of nomogram. (E) Calibration curve of nomogram. (F, G) In the JAVELIN Renal 101 cohort, the expression level of MMP9 and SCGN in subtype 1 and subtype 2. (H) In the JAVELIN Renal 101 cohort, patients treated with avelumab + axitinib are classified as having low PFS in subtype 2. PFS, progression free survival. (I) Immune cell infiltration of subtype 1 and subtype 2 in the immunotherapy cohort.
FIGURE 8
FIGURE 8
Single-cell expression analysis of MMP9 and SCGN in ccRCC. (A) Cell clustering of GSE171306. (B) Distribution of MMP9 in different cell populations. (C) Distribution of SCGN in different cell populations. (D) Expression levels of MMP9 in different cell populations. (E) Expression levels of SCGN in different cell populations. (F) Molecular function enrichment analysis of SCGN in different subpopulations of malignant cells. (G) Enrichment analysis of metabolic pathways of SCGN in different subpopulations of malignant cells.
FIGURE 9
FIGURE 9
SCGN increases the proliferation and invasion ability of tumor cells. (A, B) SCGN protein and mRNA expression levels in normal and tumor tissues, from UALCAN (https://ualcan.path.uab.edu/). (C) Comparison of IHC staining of SCGN in tumor tissue and normal tissue, from HPA (https://www.proteinatlas.org/), Anti-body:CAB068232. (D) After knocking down SCGN in 786-O and ACHN cell lines, the relative expression of SCGN decreased. (E) Cell invasion was attenuated after knockdown of SCGN in 786-O and ACHN cell lines. (F) Knockdown of SCGN in 786-O and ACHN cell lines weakened cell proliferation. IHC, Immunohistochemistry.

Similar articles

Cited by

References

    1. Ai K., Yi L., Wang Y., Li Y. (2023). CircRNA_33702 promotes renal fibrosis by targeting the miR-29b-3p/WNT1-inducible signaling pathway protein 1 pathway. J. Pharmacol. Exp. Ther. 384 (1), 61–71. 10.1124/jpet.122.001280 - DOI - PubMed
    1. Bai D., Chen S., Feng H., Lu J., Ma Y. (2021). Integrated analysis of immune-related gene subtype and immune index for immunotherapy in clear cell renal cell carcinoma. Pathol. Res. Pract. 225, 153557. 10.1016/j.prp.2021.153557 - DOI - PubMed
    1. Barata P. C., Rini B. I. (2017). Treatment of renal cell carcinoma: current status and future directions. J. Clin. 67 (6), 507–524. 10.3322/caac.21411 - DOI - PubMed
    1. Braun D. A., Bakouny Z., Hirsch L., Flippot R., Van Allen E. M., Wu C. J., et al. (2021). Beyond conventional immune-checkpoint inhibition — novel immunotherapies for renal cell carcinoma. Nat. Rev. Clin. Oncol. 18 (4), 199–214. 10.1038/s41571-020-00455-z - DOI - PMC - PubMed
    1. Capitanio U., Bensalah K., Bex A., Boorjian S. A., Bray F., Coleman J., et al. (2019). Epidemiology of renal cell carcinoma. Eur. Urol. 75 (1), 74–84. 10.1016/j.eururo.2018.08.036 - DOI - PMC - PubMed