pubmed.ncbi.nlm.nih.gov

Circadian Regulation Patterns With Distinct Immune Landscapes in Gliomas Aid in the Development of a Risk Model to Predict Prognosis and Therapeutic Response - PubMed

  • ️Sat Jan 01 2022

Circadian Regulation Patterns With Distinct Immune Landscapes in Gliomas Aid in the Development of a Risk Model to Predict Prognosis and Therapeutic Response

Ruotong Tian et al. Front Immunol. 2022.

Abstract

Circadian disruption in tumorigenesis has been extensively studied, but how circadian rhythm (CR) affects the formation of tumor microenvironment (TME) and the crosstalk between TME and cancer cells is largely unknown, especially in gliomas. Herein, we retrospectively analyzed transcriptome data and clinical parameters of glioma patients from public databases to explore circadian rhythm-controlled tumor heterogeneity and characteristics of TME in gliomas. Firstly, we pioneered the construction of a CR gene set collated from five datasets and review literatures. Unsupervised clustering was used to identify two CR clusters with different CR patterns on the basis of the expression of CR genes. Remarkably, the CR cluster-B was characterized by enriched myeloid cells and activated immune-related pathways. Next, we applied principal component analysis to construct a CRscore to quantify CR patterns of individual tumors, and the function of the CRscore in prognostic prediction was further verified by univariate and multivariate regression analyses in combination with a nomogram. The CRscore could not only be an independent factor to predict prognosis of glioma patients but also guide patients to choose suitable treatment strategies: immunotherapy or chemotherapy. A glioma patient with a high CRscore might respond to immune checkpoint blockade, whereas one with a low CRscore could benefit from chemotherapy. In this study, we revealed that circadian rhythms modulated tumor heterogeneity, TME diversity, and complexity in gliomas. Evaluating the CRscore of an individual tumor would contribute to gaining a greater understanding of the tumor immune status of each patient, enhancing the accuracy of prognostic prediction, and suggesting more effective treatment options.

Keywords: circadian rhythm; glioma; prognostic model; therapy; tumor microenvironment.

Copyright © 2022 Tian, Li and Shu.

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

Landscape of genomic variations and transcription profile of circadian rhythm genes in gliomas. (A) Volcano plot of differentially expressed CR genes between normal and glioma (TCGA-LGGGBM: TCGA-LGG and TCGA-GBM) samples (Wilcoxon test: adjust p < 0.05, and |log2FC| > 1). The overexpressed genes in gliomas were highlighted in red. (B) The interaction between CR genes in gliomas. The CR genes with different features were depicted by circles in different colors. The lines connecting CR genes represented their interaction with each other. The size of each circle represented the differential expression of CR genes in gliomas compared to normal. (C) The mutation frequency of CR genes in 584 patients from the TCGA cohort. Each column represented an individual patient. The upper bar plot indicated mutation-accumulation. The bar plot on the right indicated the proportion of each variant type with the number above representing mutation frequency. Only top 10 mutations were included. (D) The CNV frequency of CR genes in glioma patients from the TCGA cohort. The height of each column represented the alteration frequency. The amplification frequency, pink dot; The deletion frequency, blue dot. (E) The location of CR genes with CNV alteration on chromosomes. (F) The mRNA expression level of core CR genes among normal, TCGA-LGG, and TCGA-GBM samples. Normal, blue; LGG, pink; GBM, red. The asterisks represented the statistical p-value (Kruskal–Wallis test: *p < 0.05; ***p < 0.001).

Figure 2
Figure 2

Identification of two circadian patterns mediated by CR genes in TCGA cohort. (A) Unsupervised clustering of 91 CR genes for 596 glioma patients in the TCGA cohort resulted in two CR clusters. Age, gender, tissue, WHO grade, histopathology, IDH status, ATRX status, MGMT promoter status, TERT promoter status, and survival status are shown as patient annotations. Pink represented the relatively high expression of CR genes and blue represented the relatively low expression. (B) Principal component analyses for the transcriptome profiles of CR patterns in TCGA (left) and CGGA (right) cohorts, respectively, showing a remarkable difference on transcriptome between two CR patterns. (C) Survival analyses for CR patterns in the TCGA cohort (left) including 379 cases in CR cluster-A and 217 cases in CR cluster-B (Log-Rank test: p < 0.0001), and in the CGGA cohort (right) including 614 cases in CR cluster-A and 356 cases in CR cluster-B using Kaplan–Meier curves (Log-Rank test: p < 0.0001). (D) The transcriptome of core CR genes with remarkable differences in two CR patterns corresponded with the previous clustering. Pink represented the relatively high expression of core CR genes and blue represented the relatively low expression. (E, F) GSVA enrichment analyses in two CR clusters showing the activation states of Hallmark pathways (MSigDB) in TCGA (E) and CGGA (F) cohorts. Activated pathways, pink; Inhibited pathways, blue.

Figure 3
Figure 3

Characterization of the immune cell infiltration in distinct circadian patterns. (A) Heatmap of TME cell infiltration characteristics in two CR patterns assessed by four different methods. (B) Heatmap showing the differences of immune-related gene expression in two CR patterns. Upregulation, pink; Downregulation, blue. (C) The correlation between TME infiltration cell type and each core CR genes using spearman analyses. Negative correlation, blue; Positive correlation, pink. (D) CR clusters were distinguished by different TME relevant signatures and TME score (Wilcoxon test: ***p < 0.001).

Figure 4
Figure 4

Generation of circadian gene clusters and functional annotations for CR-related genes. (A) Volcano plot of CR pattern-related DEGs between CR cluster-A and -B (Wilcoxon test: adjust p < 0.05, and |log2FC| > 1.5). (B) Unsupervised clustering of CR pattern-related DEGs to classify patients of the TCGA cohort into different subtypes, termed CR gene cluster-A and -B, respectively. The gene clusters, CR clusters, and other parameters were used as patient annotations. (C) Survival analysis for CR gene clusters in the TCGA cohort including 368 cases in gene cluster-A and 228 cases in gene cluster-B using Kaplan–Meier curves (Log-Rank test, p < 0.0001). (D) Functional annotation for CR-related genes upregulated in gene cluster-B (Gene Set 1 in B) using GO enrichment analysis. The length of the bar plots represents the number of genes in that category. The color depth represented q-value. (E) Functional annotation for CR-related genes upregulated in gene cluster-A (Gene Set 2 in B) using GO enrichment analysis. The length of the bar plots represents the number of genes in that category. The color depth represented q-value.

Figure 5
Figure 5

Construction of the CRscore and exploration of its immunological relevance. (A) An overview of the association between CRscores and other patient annotations in TCGA cohort. (B) Comparison of CRscores between gene cluster A and B in the TCGA cohort, p < 0.001 (left); Comparison of CRscores between CR cluster A and B in the TCGA cohort, p < 0.001 (right). (C) The correlations between CRscore and TME relevant signatures in LGG as well as GBM of TCGA cohort, respectively (p > 0.05). Pink represented positive correlation and blue represented negative correlation in LGG; purple represented positive correlation and green represented negative correlation in GBM. (D) The correlations between CRscore and ssGSEA scores of TME cells in LGG as well as GBM of the TCGA cohort, respectively (p > 0.05). Pink represented positive correlation and blue represented negative correlation in LGG; purple represented positive correlation and green represented negative correlation in GBM. (E) The correlations between CRscore and steps of the cancer immunity cycle in LGG (right) as well as GBM (left) of TCGA cohort, respectively.

Figure 6
Figure 6

Evaluation of the prognostic potentiality of the CRscore. (A) The overlap among tissue types, CR clusters, gene clusters, and CR groups. (B) Survival analyses for CR groups in the TCGA cohort (left) including 357 cases in the CR-high group and 239 cases in the CR-low group (Log-Rank test, p < 0.0001), and in the CGGA (right) cohort including 507 cases in the CR-high group and 463 cases in the CR-low group using Kaplan–Meier curves (Log-Rank test, p < 0.0001). (C) Predictive accuracy at 5-year of CRscore compared with other histological or molecular indicators in TCGA (left) and CGGA (right) cohort, respectively. The accuracy was equal to the area under the ROC curves (AUC). (D) Time-dependent AUC of CRscore compared with other histological or molecular indicators in the TCGA (left) and CGGA (right) cohort, respectively. (E, F) Univariate and multivariate analyses of clinicopathological characteristics and CRscore with overall survival in the TCGA (E) and CGGA (F) cohort, respectively. (G) Nomograms for predicting the probability of patient mortality based on WHO Grade, CRscore, and age. (H) Plots depicted the calibration of nomograms. (I) Net decision curve analyses demonstrating the benefit for predicting overall survival on nomogram. (J) The net reduction analyses demonstrated in how many patients the nomogram could be avoided without prognosis of miscalculation.

Figure 7
Figure 7

Landscape of tumor somatic mutation in two CR groups. (A) Differences of TMB in CR-high and -low groups in TCGA cohort (Wilcoxon test: p < 0.001). (B) Top 20 most frequent mutations in patients from TCGA cohort and the distribution of mutations in CR high and low groups. (C) Differences in CR scores between wild-type and mutant groups of each of the top 20 genes in the TCGA cohort (Wilcoxon test: ns, P>0.05; *p < 0.05; **p < 0.01; ***p < 0.001). (D) Differences in the frequency of top 20 mutations between CR-high and -low groups in the TCGA cohort (Fisher’ exact test: p > 0.05; *p < 0.05; **p < 0.01; ***p < 0.001).

Figure 8
Figure 8

The role of the CRscore in the prediction of therapeutic benefits. (A, B) The relative distribution of TIDE (A) in TCGA (p > 0.001) and CGGA (p > 0.001) cohorts and ImmuCellAI (B) in TCGA (p = 0.17) and CGGA (p > 0.001) cohorts were compared between CR score high versus low groups, respectively. (C) The proportion of patients’ response to PD-1 blockade immunotherapy in CR-high or -low groups. PD, progressive disease; CR, complete response; PR, partial response (Fisher’ exact test: p = 0.013). (D) Survival analyses for low (8 cases) and high (19 cases) CR score patient groups in the anti-PD-1 immunotherapy cohort (GSE78220) using Kaplan–Meier curves (Log-Rank test: p = 0.0066). (E) The correlation of CRscore with clinical response to anti-PD-1 immunotherapy. (F) The correlation between CRscore and the estimated IC50 for drugs evaluated by the Spearman analysis. Each point represents a drug. (G) Pathways targeted by candidate drugs. (H) The box plots of the estimated IC50 for candidate drugs in CR-low and -high groups (Wilcoxon test: ***p < 0.001).

Similar articles

Cited by

References

    1. Ostrom QT, Patil N, Cioffi G, Waite K, Kruchko C, Barnholtz-Sloan JS. CBTRUS Statistical Report: Primary Brain and Other Central Nervous System Tumors Diagnosed in the United States in 2013-2017. Neuro Oncol (2020) 22(12 Suppl 2):iv1–96. doi: 10.1093/neuonc/noaa200 - DOI - PMC - PubMed
    1. Louis DN, Perry A, Reifenberger G, von Deimling A, Figarella-Branger D, Cavenee WK, et al. . The 2016 World Health Organization Classification of Tumors of the Central Nervous System: A Summary. Acta Neuropathol (2016) 131(6):803–20. doi: 10.1007/s00401-016-1545-1 - DOI - PubMed
    1. Bolcaen J, Kleynhans J, Nair S, Verhoeven J, Goethals I, Sathekge M, et al. . A Perspective on the Radiopharmaceutical Requirements for Imaging and Therapy of Glioblastoma. Theranostics (2021) 11(16):7911–47. doi: 10.7150/thno.56639 - DOI - PMC - PubMed
    1. Touat M, Idbaih A, Sanson M, Ligon KL. Glioblastoma Targeted Therapy: Updated Approaches From Recent Biological Insights. Ann Oncol (2017) 28(7):1457–72. doi: 10.1093/annonc/mdx106 - DOI - PMC - PubMed
    1. Nicholson JG, Fine HA. Diffuse Glioma Heterogeneity and Its Therapeutic Implications. Cancer Discov (2021) 11(3):575–90. doi: 10.1158/2159-8290.CD-20-1474 - DOI - PubMed

Publication types

MeSH terms