Exploration of the Prognostic and Immunotherapeutic Value of B and T Lymphocyte Attenuator in Skin Cutaneous Melanoma

Abstract

B and T lymphocyte attenuator (BTLA) is a newly identified immune checkpoint molecular belonging to the CD28 immunoglobulin superfamily. However, the expression and clinical value of BTLA in skin cutaneous melanoma (SKCM) has not been widely characterized. We found that BTLA levels were upregulated in metastatic melanoma compared to normal skin tissues and primary melanoma. Higher BTLA was also correlated with improved prognosis in SKCM based on several datasets. The multivariate Cox regression model revealed that BTLA was an independent survival indicator in metastatic melanoma. Tumor microenvironment analysis indicated BTLA was positively associated with the infiltrating levels of different immune cells and the activity of the anti-cancer immunity cycle. Importantly, BTLA accurately predicted the outcome of melanoma patients treated with MAGE-A3 blocker or first-line anti-PD-1. The present findings disclose that BTLA is a reliable biomarker for prognosis and immunotherapeutic response and might contribute to developing novel SKCM immunological treatment strategies.

Keywords: B and T lymphocyte attenuator (BTLA); immunotherapy; prognosis; skin cutaneous melanoma; tumor microenvironment; tumor-infiltration.

Figure 1

The expression of BTLA in SKCM. (A) BTLA mRNA levels in normal tissues, primary and metastatic SKCM tissues by TCGA and GTEx database. (B) BTLA mRNA expression levels in primary and metastatic SKCM tissues by GSE65904. (C) BTLA mRNA expression levels in HEMa and melanoma cell lines (SK-MEL-28, LOXIMVI, and M21). (D) BTLA protein expression levels in normal tissues, primary and metastatic SKCM tissues by the human protein atlas database. HEMa, normal human epidermal melanocytes; n.s., no significance. Statistical methods were as follows: (A, B) Wilcoxon rank-sum test.

Figure 2

Kaplan-Meier survival curves comparing the BTLA^high and BTLA^low groups in SKCM in TCGA cohorts and GEO cohorts. (A–C) Survival curves of OS, DSS, DFI in TCGA SKCM cohorts. (D, E) Low BTLA expression was correlated with poor DSS in the GSE65904 and GSE53118 cohorts. (F) Low BTLA expression was correlated with poor DFI in the GSE98394 cohort. OS, overall survival; DSS, disease-specific survival; DFI, disease-free interval.

Figure 3

The survival curves and multivariate Cox analysis model of BTLA. (A, B) The OS and DSS curves using combinations of BTLA expression level and tumor subtypes. (C, D) The age, T, N, radiation therapy, and BTLA expression showed significance in multivariate Cox analysis forest graph of OS and DSS prognosis. OS, overall survival; DSS, disease-specific survival; T, tumor stage; N, regional lymph nodes metastasis stage.

Figure 4

Function enrichment analysis of the co-expressed genes of BTLA. (A) The enrichment results of the GO. (B) The enrichment results of the KEGG signal pathway. Clusters are based on the gene count to measure the similarity between terms. The size of the gray circle represents the number of gene counts. Different color circles represent the correlation with BTLA. GO, the gene ontology; KEGG, Kyoto Encyclopedia of Genes and Genomes.

Figure 5

Comparison of tumor microenvironment scores with BTLA expression profiles and the relative proportion and correlation of tumor-infiltrating immune cells in SKCM. (A) Spearman correlations between BTLA and microenvironment scores calculated based on the ESTIMATE algorithm in primary and metastatic SKCM. (B, C) Heatmap illustrating the proportion of the immune cells in primary and metastatic SKCM microenvironment using xCell and CIBERSORT. (D, E) Correlation clustering of immune cells based on expression data to classify them into four groups by xCell and CIBERSORT.

Figure 6

BTLA expression and tumor-infiltrating immune cells. Compare the immune infiltrated cell components quantified in (A, B) primary and (C, D) metastatic SKCM by algorithms xCell and CIBERSORT in the BTLA^high and BTLA^low groups. The scattered dots represent all score values, and the thick line represents the median value within each group. The bottom and top of the boxes are the 25th and 75th percentiles (interquartile range). The difference between BTLA^high and BTLA^low groups was compared through the two-sided Wilcoxon rank-sum test. ns: not significant, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

Figure 7

Compare the levels of cancer-immunity cycle, MSI and TMB between the different BTLA expressional groups. (A, B) Compare the levels of cancer-immunity cycle in primary and metastatic SKCM. (C, D) Compare the levels of MSI and TMB in primary and metastatic SKCM. TMB, tumor mutational burden; MSI, microsatellite instability. ns: not significant, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

Figure 8

BTLA is predictive in immunotherapy response in advanced melanoma. (A) The comparison of the BTLA and PD-L1 expression between MAGE-A3 antigen vaccine therapy response groups. (B) The MAGE-A3 antigen vaccine therapy response in the different BTLA and PD-L1 expressional groups. (C) The predictive value of the BTLA and PD-L1 was measured by ROC curves in GSE35640 cohort. (D) The comparison of the BTLA and PD-L1 expression between anti-PD-1 immunotherapy response (Nivolumab) groups. (E) The Nivolumab immunotherapy response in the different BTLA and PD-L1 expressional groups. (F) The predictive value of the BTLA and PD-L1 was measured by ROC curves in GSE91061 cohort. (G) The comparison of the BTLA and PD-L1 expression between anti-PD-1 immunotherapy response (Pembrolizumab) groups. (H) The Pembrolizumab immunotherapy response in the different BTLA and PD-L1 expressional groups. (I) The predictive value of the BTLA and PD-L1 was measured by ROC curves in Liu et al. cohort. NR/R, non-response or response; PR/CR, partial or complete response; PD/SD, progressive or stable disease. ROC, receiver operating characteristic. Statistical methods were as follows: (A, D, G) Wilcoxon rank-sum test; (B, E, H) Two-sided Fisher exact test.