T cell receptor signaling pathway subgroups and construction of a novel prognostic model in osteosarcoma

Abstract

Background: T cell receptor (TCR) signaling pathway is closely related to tumor progress and immunotherapy. This study aimed to explore the clinical significance, prognosis, immune infiltration and chemotherapy sensitivity of TCR in osteosarcoma (OS).

Material and methods: OS data were obtained from TARGET and GEO database. TCR signaling pathway-related genes (TCRGs) were extracted from Molecular Signatures Database. Unsupervised non-negative matrix factorization clustering analysis was used to identify OS molecular subtypes. Differential expressed TCRGs between molecular subtypes were screened with univariate Cox regression, LASSO regression and multivariate Cox regression. Subsequently, an OS-associated prognostic model was constructed and validated. Nomogram was established and verified. Immune landscape analysis including immune infiltration analysis, ESTIMATE algorithm and immune checkpoints expression levels of molecular subtypes and different risk groups were analyzed. Finally, chemotherapy sensitivity and potential therapeutic agents between different risk groups was identified.

Results: Two TCRGs related subclusters were identified. Two hundred and seventy-two Differential expressed TCRGs were screened between two subclusters. A robust prognostic model were constructed. High and low risk groups were stratified. Low risk group showed higher ESTIMATE, immune and stromal scores, while high risk group exhibited higher tumor purity and the lower expression levels of immune checkpoints. A nomogram comprising metastasis and risk score was successfully built. The sensitivity to chemotherapy agents were different across high and low risk groups.

Conclusions: Our study proposed TCR related molecular subtypes and provided a prognostic model for OS. Our findings may bring a new insight into the immunotherapy for OS patients.

Keywords: Osteosarcoma; Prognosis; Signal Transduction.

Fig. 1

TCRGs subclusters of OS. (A) Cophenetic coefficient for r = 2–10 is shown. (B) Consensus matrix heatmap for r = 2 is displayed. (C) Silhouette coefficient for r = 2–10 is shown. (D–E) Both UMAP and t-SNE analyses supported the stratification into two OS subclusters. (F–H) Comparisons of gender, age and metastasis between two clusters.

Fig. 2

(A) K-M plot of overall survival. (B) K-M plot of progression free survival.

Fig. 3

(A–B) DEGs and TCRGs between two subtypes. (C–F) Enrichment plots of top five enrichments for biological process, molecular function, cellular component and KEGG.

Fig. 4

(A) Forest plot of eight TCRGs closely linked to OS prognosis mined by univariate Cox regression analysis. (B) LASSO coefficient plot of the eight TCRGs. (C) Penalty plot of the LASSO regression for the eight TCRGs. (D) Forest plot of two independent prognostic TCRGs screened by multivariate Cox regression. (E) Cut point plot of the risk score.

Fig. 5

(A–B) Comparisons of gender between high and low risk groups in TARGET and GSE21257 cohorts. (C–D) Comparisons of age between high and low risk groups in TARGET and GSE21257 cohorts. (E–F) Comparisons of metastasis between high and low risk groups in TARGET and GSE21257 cohorts. (G–H) Expression level of CD8A between high and low risk groups in TARGET and GSE21257 cohorts. (I–J) Expression level of WAS between high and low risk groups in TARGET and GSE21257 cohorts.

Fig. 6

(A–B) K-M plot of overall survival in TARGET and GSE21257 cohorts. (C–D) Time-dependent ROC curve of OS prognostic model at 1, 3, and 5 years in TARGET and GSE21257 cohorts. (E–F) Survival distribution map and risk heatmap of TARGET and GSE21257 cohorts. (G–H) Expression levels of CD8A and WAS between osteosarcoma cell lines and normal osteoblast and bone samples in GSE36001 cohort.

Fig. 7

(A–B) Forest plots of significant covariates screened by univariate and multivariate Cox regressions. (C) Nomogram predicting 1- year, 3- year and 5-year survival of OS patients. (D–F) Calibration curves of the nomogram at 1- year, 3- year and 5-year in Target cohort. (G–I) Calibration curves of the nomogram at 1- year, 3- year and 5-year in GSE21257 cohort.

Fig. 8

(A–B) Immune infiltration analysis in Target and GSE21257 cohorts. (C) Infiltrating levels of immunocyte lines in two OS subclusters. (D–E) Infiltrating levels of immunocyte lines in high and low risk groups at TARGET and GSE21257 cohorts. (∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001).

Fig. 9

(A–D) ESTIMATE, immune and stromal scores and tumor purity between two OS clusters. (E–H) ESTIMATE, immune and stromal scores and tumor purity between high and low risk groups at TARGET cohorts. (I–L) ESTIMATE, immune and stromal scores and tumor purity between high and low risk groups at GSE21257 cohorts.

Fig. 10

(A) Expression levels of immune check points in cluster 1/2. (B–C) Expression levels of immune check points in high and low risk groups at TARGET and GSE21257 cohorts. (∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001).

Fig. 11

(A–C) Correlation analyses between the expression levels of CD8A and WAS and ESTIMATE analysis, proportion of macrophage types and immune checkpoints in TARGET cohort. (D–F) Correlation analyses between the expression levels of CD8A and WAS and ESTIMATE analysis, proportion of macrophage types and immune checkpoints in GSE21257 cohort.

Fig. 12

(A–B) IC50 values of chemotherapy agents in high and low risk groups at TARGET and GSE21257 cohorts. (∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001).

Fig. 13

(A–B) IC50 values of chemotherapy agents intersected between TARGET and GSE21257 cohorts in high and low risk groups at TARGET and GSE21257 cohorts. (∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001).