Abstract
Background: Although low-grade glioma (LGG) has a good prognosis, it is
prone to malignant transformation into high-grade glioma. It has been
confirmed that the characteristics of inflammatory factors and immune
microenvironment are closely related to the occurrence and development
of tumors. It is necessary to clarify the role of inflammatory genes
and immune infiltration in LGG.
Methods: We downloaded the transcriptome gene expression data and
corresponding clinical data of LGG patients from the TCGA and GTEX
databases to screen prognosis-related differentially expressed
inflammatory genes with the difference analysis and single-factor Cox
regression analysis. The prognostic risk model was constructed by LASSO
Cox regression analysis, which enables us to compare the overall
survival rate of high- and low-risk groups in the model by Kaplan–Meier
analysis and subsequently draw the risk curve and survival status
diagram. We analyzed the accuracy of the prediction model via ROC
curves and performed GSEA enrichment analysis. The ssGSEA algorithm was
used to calculate the score of immune cell infiltration and the
activity of immune-related pathways. The CellMiner database was used to
study drug sensitivity.
Results: In this study, 3 genes (CALCRL, MMP14, and SELL) were selected
from 9 prognosis-related differential inflammation genes through LASSO
Cox regression analysis to construct a prognostic risk model. Further
analysis showed that the risk score was negatively correlated with the
prognosis, and the ROC curve showed that the accuracy of the model was
better. The age, grade, and risk score can be used as independent
prognostic factors (p < 0.001). GSEA analysis confirmed that 6
immune-related pathways were enriched in the high-risk group. We found
that the degree of infiltration of 12 immune cell subpopulations and
the scores of 13 immune functions and pathways in the high-risk group
were significantly increased by applying the ssGSEA method (p < 0.05).
Finally, we explored the relationship between the genes in the model
and the susceptibility of drugs.
Conclusion: This study analyzed the correlation between the
inflammation-related risk model and the immune microenvironment. It is
expected to provide a reference for the screening of LGG prognostic
markers and the evaluation of immune response.
Keywords: low-grade glioma, inflammatory response, prognosis, immune
infiltration, drug sensitivity
Introduction
Glioma is the most common primary malignant tumor of the central
nervous system, and its rate of fatality and disability are both high
([36]Lapointe et al., 2018; [37]Ferlay et al., 2019). Low-grade glioma
is classified into grade II and grade III with isocitrate dehydrogenase
(IDH) mutations according to the WHO histopathological grading system
([38]De Blank et al., 2020). Although its prognosis is relatively
better than that of high-grade gliomas, nearly 70% of LGG patients,
during the period of occurrence and development, are prone to
transition to high-grade gliomas (HGGs), which express the
characteristics such as higher malignant degree and stronger
invasiveness ([39]Appolloni et al., 2019; [40]Huang, 2019; [41]Nejo et
al., 2019; [42]Youssef and Miller, 2020). At present, the main clinical
treatments for LGG include surgical resection, radiotherapy, and
chemotherapy. However, the existing treatments still fail to
significantly improve the survival rate of patients. It is well known
that cancer is closely related to inflammation. Rudolf Virchow et al.
regarded inflammation as a possible driver of tumorigenesis through
“lymphatic reticular infiltration,” and the inflammatory cells and
cytokines in tumor contributed to the growth, progression, and
immunosuppression of cancer ([43]Bergamin et al., 2015; [44]Lepore et
al., 2018). Inflammation is one of the characteristics of tumors, and
uncontrollable inflammation is closely related to the occurrence,
development, invasion, and metastasis of tumors ([45]Singh et al.,
2017). The growth of tumors depends not only on the genetic changes of
malignant tumor cells but also on the changes in the tumor
microenvironment, such as matrix, blood vessels, and infiltrating
inflammatory cells, and it is immunity and inflammation that constitute
the two cores of the tumor microenvironment ([46]Chen and Mellman,
2017; [47]Suarez-Carmona et al., 2017). In recent years, increasing
evidence has shown that tumor-related inflammation can advance tumor
growth and progression by promoting angiogenesis and metastasis,
subverting antitumor immune responses, and changing the sensitivity of
tumor cells to chemotherapeutics ([48]Khandia and Munjal, 2020).
Cytokines, which are produced by chronic inflammation, induce gene
mutations, change the expression and transformation of oncogenes and
tumor suppressor genes, inhibit cell apoptosis, evoke angiogenesis, and
result in abnormal inflammatory signal transduction pathways. In
addition, chronic inflammation can recruit a variety of
immunosuppressive cells (such as M2-TAMs, MDSC, and Treg) to facilitate
the establishment of immunosuppressive tumor microenvironment and
accelerate the occurrence of the tumor malignant biological behavior
([49]Whiteside, 2018; [50]Zhang et al., 2019; [51]Weber et al., 2021).
Therefore, it is of great significance to effectively control chronic
inflammation in the process of inhibiting tumor occurrence and
enhancing antitumor immune response. It is this idea that is the
starting point of this article to carry out research, hoping to provide
a certain reference for related research on tumor inflammation and
immunity.
Some studies have shown that inflammatory media, which contain human
leukocyte antigen-G, CD8 T cells, IL-1beta, IL-6, TAM, the S100A family
and so on, are having high expression in the high-risk group ([52]Bloch
et al., 2013; [53]Kelly et al., 2020). It is seen that the
proliferation, deterioration, and prognosis of glioma are closely
linked with the inflammatory microenvironment. Currently, some
inflammatory response–related genes were reported to predict the
metastasis potential of LGG, but further utilization in the prognosis
of LGG remains to be studied. In addition, the integrity of the
blood–brain barrier of glioma is easily destroyed with the help of
pathological conditions, which provide opportunities for immune-related
cell infiltration and recognition ([54]Cserr et al., 1992; [55]Davies,
2002; [56]Goldmann et al., 2006). It has laid a theoretical foundation
for developing immunotherapy in LGG. Therefore, it is necessary to
establish an inflammatory factor–related model derived from LGG samples
to scientifically predict its prognosis.
In this study, we downloaded the transcriptome gene expression data and
corresponding clinical data of LGG patients from public databases.
Then, we constructed a prognostic model using differentially expressed
genes related to inflammation in the TCGA database, and verified the
accuracy of the model in predicting the survival of LGG patients
through the ROC curve. Then, we further conducted a functional
enrichment analysis to explore its potential immune mechanism. In
addition, we also analyzed the relationship between the prognostic gene
expression and the type of immune infiltration, and discussed the
feasibility of inflammation-related risk models in predicting the
immunotherapy response. Finally, we analyzed the inflammatory genes in
the prognosis model and the sensitivity of drugs. Our research has
discovered that some inflammation-related genes may act as early
warning markers and were considered to be related to the poor prognosis
of LGG, and meanwhile, we also clarified its relevance to the immune
microenvironment, which may provide an important basis for the
evaluation of the clinical efficacy of immunotherapy.
Methods
Data Acquisition Extract
We downloaded transcriptional group gene expression data from the tumor
samples of 529 patients with LGG through the Cancer Genome Atlas (TCGA,
[57]https://portal.gdc.cancer.gov/) database while obtaining the
corresponding clinical information, including age, gender, grade,
overall survival time, and survival status. We downloaded the gene
expression data of 1,152 normal brain tissues in the GTEX database
through the UCSC website ([58]http://xena.ucsc.edu/). We extracted and
verified the reaction-related genes: downloaded the human inflammatory
response gene set from the GSEA
([59]http://www.gsea-msigdb.org/gsea/index.jsp) database
([60]Supplementary Tables 6, 7, and 9).
Screening of Differential Genes and Prognosis-Related Inflammation Genes
In order to obtain differentially expressed inflammation-related genes,
we used the “Bioconductor Limma” R package to analyze 529 LGG tumor
tissues and 1,152 normal brain tissues. If a gene satisfies the
condition of | log2FC | > 2 and FDR <0.05, it was considered to be a
differential inflammatory gene. In order to clarify the relationship
between inflammation-related genes and the prognosis of LGG patients,
based on the expression data and clinical data of tumor samples in the
TCGA database, we used the “survival” R package to perform univariate
Cox regression analysis on inflammation genes. A gene was considered to
be a prognosis-related gene if p < 0.05.
Construction and Evaluation of Inflammatory Gene–Related Prognostic Models
In order to determine the value of inflammatory genes in evaluating the
prognosis of LGG, we adopted LASSO Cox regression analysis to construct
a prognostic risk model which employed the abovementioned nine
prognosis-related differential inflammatory genes. The risk score is
calculated using the following formula: Risk score =
[MATH:
∑i=1
nCoe
fi*xi :MATH]
, In this formula, Coefi is the risk coefficient and xi is the
expression of each gene. According to the median risk score, LGG
patients were divided into the high-risk group and low-risk group,
meanwhile drawing the risk curve and survival status chart. We employed
PCA analysis by using the “ggplot2”R package to explore the
distribution of genes in different groups based on the expression level
of genes in the model.
Analyzing the prognostic value of the model that enables us to exploit
the “survival” package and “survminer” R package to analyze the
relationship between patients with different risk groups and overall
survival, the survival curve was drawn. The “timeROC” R package was
used to construct the receiver operating characteristic (ROC) curve to
evaluate prediction efficiency. Besides, we eliminated samples with
incomplete clinical information in the TCGA database, and utilized
univariate and multivariate Cox regression analyses to explore the
feasibility of prognostic risk models as independent prognostic
markers. Finally, the relationship of LGG patients’ age, gender, and
classification between the risk score of the prognostic model was
clarified by means of clinical correlation analysis.
GSEA Enrichment Analysis
To illuminate the enrichment of high- and low-risk LGG sample groups in
terms of the immune function, the study used gene set enrichment
analysis (GSEA) and GSEA 4.1.0 software to carry out Genome
Encyclopedia (KEGG) pathway enrichment analysis. We believe that when p
< 0.05, this pathway is considered statistically significant.
Comprehensive Analysis of Tumor Microenvironment and Tumor Immune Correlation
With the purpose of clarifying the correlation between the inflammatory
response and immune infiltration, our team first adopted the tumor
immune cell infiltration score and immune-related function score for
each LGG sample in the TCGA database by applying the ssGSEA method and
sequentially used the “Bioconductor Limma” R package to operate the
differential analysis of immune scoring and immune typing, and drew the
box plot. We used the Spearman correlation test to evaluate the risk
score to explore the relationship between the expression of immune
checkpoints such as PD-1 and PD-L1 and the stem cell index (DNA-based
DNA methylation and RNA-based RNA expression). Next, we performed an
analysis of the immune score, matrix score, and comprehensive score on
LGG samples in the TCGA database, using the “estimate” R package and
the “Limma” R package to export a scatter chart.
Drug Sensitivity Analysis
With an aim to clarify the influence of inflammatory genes in the
prognostic model on drug sensitivity and tolerance, we downloaded
transcriptome data from the CellMiner database
([61]https://discover.nci.nih.gov/cellminer/) and FDA-certified drug
sensitivity–related data. The Pearson correlation test was utilized to
analyze the relationship between gene expression and drug sensitivity.
Next, we used the “pRRophetic” R package to analyze the relationship
between the high- and low-risk groups in the prognosis model and
LGG-related drugs, and draw a box plot.
Statistical Analysis
We adopt the Wilcoxon rank-sum test to analyze the gene differences
between tumor tissues and normal tissues. Our group had taken the
method of the LASSO Cox regression algorithm to establish a risk
prognosis model, wherein the relationship between its overall survival
rate in the high- and low-risk group gene expression group was used to
generate the Kaplan–Meier survival curve, and the accuracy of the model
was tested by the ROC curve. Univariate and multivariate Cox regression
analyses were used to evaluate the feasibility of the risk score as an
independent prognostic factor. The chi-square test was used to compare
the differences of clinical traits between different risk groups. We
used two correlation test means. One is Spearman's test which analyzed
the relationship between the sample risk score and the expression of
immune checkpoints, such as PD-1 and PD-L1, stem cell index, and the
tumor microenvironment score. The other is Pearson's test which
evaluated the correlation between the gene expression and drug
sensitivity in the model. All above statistical analyses were performed
using R (version 4.0.4) software. If p < 0.05, it was considered
statistically significant.
Results
Screening of Prognosis-Related Differential Inflammation Genes
We downloaded the human inflammatory response gene set using the GSEA
database, which contained 200 inflammatory response–related genes, such
as ABCA1, ABI1, and ACVR1 B. The expression levels of these
inflammatory genes in LGG tissues and normal brain tissues were
obtained from the TCGA database and GTEX database, and 13
differentially expressed inflammatory genes were screened. Compared
with normal tissues, ABCA1, APLNR, BTG2, C3AR1, CALCRL, CD14, CYBB,
HIF1A, MMP14, MYC, SELL, and SLC4A4 were upregulated in LGG tumor
tissues, and SCN1B was downregulated in tumor tissues ([62]Figure 1A).
Then, we used the expression data and clinical information of LGG
samples in the TCGA database for univariate COX analysis. A total of
140 inflammatory genes related to prognosis were obtained, including 21
low-risk genes and 119 high-risk genes ([63]Figure 1B). Finally, we
crossed the differentially expressed genes with prognosis-related genes
and obtained 9 inflammation-related genes that can mediate prognosis
and have differential expression in LGG patients ([64]Figures 1C–E).
FIGURE 1.
[65]FIGURE 1
[66]Open in a new tab
Screening of prognostic differential inflammatory genes. (A) Thermal
map of all inflammatory differential gene expression in LGG tumor
tissues and normal tissues. (B) Based on 200 inflammatory genes, the
forest map of prognosis-related genes was screened.(C) Wayne diagram of
intersection of differential genes and prognosis-related genes. (D)
Thermograms of nine prognosis-related inflammatory genes. (E) Forest
map of nine prognosis-related inflammatory genes.
Construction and Evaluation of a Prognostic Model of Inflammation-Related
Genes
We used LASSO Cox regression analysis to analyze the significant
prognostic differential inflammatory genes in the abovementioned nine
univariate results, and finally identified three genes (CALCRL, MMP14,
and SELL) for the construction of prognostic risk models ([67]Figures
2A,B). At the same time, the weight coefficient of each gene was
determined, and the risk score was calculated according to the
following formula: Risk score =
-0.195793×CALCRL+0.310011×MMP14-0.015233×SELL.
FIGURE 2.
[68]FIGURE 2
[69]Open in a new tab
Construction and evaluation of prognostic model of inflammation-related
genes. (A,B) LASSO Cox regression analysis screened three
differentially expressed inflammatory genes and established a
prognostic model. (C) Risk curve constructed according to the median of
the risk score. (D) Feasibility of PCA-based analysis and judgment
models. (E) Survival status of LGG patients with different risk scores.
(F) Survival analysis of LGG patients in different risk groups.(G) ROC
curve suggested that the model had good accuracy in predicting the
1-year, 3-year, and 5-year survival of LGG patients.
According to the median risk score, LGG patients were divided into
high- and low-risk groups ([70]Figure 2C). PCA analysis showed that
patients with different risk groups were distributed in two directions,
indicating that the expression of three genes in the model can
effectively classify LGG patients into high- and low-risk groups
([71]Figure 2D). The survival status scatterplot showed that the number
of patients who died gradually increased with the increase in the risk
value ([72]Figure 2E). Further survival analysis showed that the
overall survival time was significantly different between the high-risk
group and the low-risk group, and the overall survival rate of the
high-risk group was significantly lower than that of the low-risk group
(p < 0.001) ([73]Figure 2F). The ROC curve showed that the AUC values
of the model for 1 year, 3 years, and 5 years were 0.870, 0.833, and
0.787, respectively, indicating that the model had high accuracy in
predicting the survival of LGG patients ([74]Figure 2G).
Independent Prognostic Analysis and Clinical Correlation Analysis
We explored the independent prognostic value of the
inflammation-related gene prognostic risk model through single-factor
regression analysis and multifactor regression analysis. Univariate
analysis showed that age (p < 0.001), grading (p < 0.001), and risk
score (p < 0.001) were significantly correlated with the overall
survival rate of patients ([75]Figure 3A). Further multivariate
analysis showed that age (p < 0.001), grading (p < 0.001), and risk
score (p < 0.001) were still significantly correlated with the overall
survival rate of patients ([76]Figure 3B). Therefore, we believe that
the risk score of the model can be used as an independent prognostic
factor for LGG. By analyzing the relationship between the risk score
and clinical characteristics, we found that the risk score of LGG
patients aged ≥65 years was significantly higher than that of patients
aged <65 years (p < 0.01), and the risk score of patients with tumor
grade 3 was significantly higher. For grade 2 patients (p < 0.001),
there was no significant difference in risk scores between different
genders (p > 0.05) ([77]Figures 3C–E).
FIGURE 3.
[78]FIGURE 3
[79]Open in a new tab
Independent prognostic analysis and clinical correlation analysis. (A)
Single-factor Cox regression analysis of different clinical
characteristics and risk scores. (B) Multivariate Cox regression
analysis of different clinical characteristics and risk scores. (C–E)
Differences in age, gender, and tumor grade between high- and low-risk
groups.
Pathway Enrichment Analysis
We performed GSEA pathway enrichment analysis on the high- and low-risk
groups, and the results showed that 64 pathways were significantly
enriched in the high-risk group (false discovery rate <0.05). Among
them, the statistically prominent pathways include leukocyte
transendothelial migration, glutathione metabolism, regulation of actin
cytoskeleton, and apoptosis ([80]Figure 4G). Further, we conducted an
in-depth analysis of immune-related pathways, and the results indicate
that the immune-related pathways of this model include antigen
processing and presentation, primary immunodeficiency, natural killer
cell–mediated cytotoxicity, intestinal immune IGA production network,
and B-cell receptor body signaling pathway and T-cell receptor
signaling pathway, but failure to find statistically significant
pathways was enriched in the low-risk group ([81]Figures 4A–F).
FIGURE 4.
[82]FIGURE 4
[83]Open in a new tab
GSEA pathway enrichment analysis. (A–F) Immune-related pathways that
are significantly enriched in the high-risk group. (G) Top ten pathways
for GSEA enrichment analysis.
Correlation Analysis of Immune Subtypes and Immune Response
We used the ssGSEA method to quantify 16 immune cell subsets and 13
immune-related functions to clarify the correlation between the risk
score and immune status. We found that the infiltration of 12 immune
cell subsets in the high-risk group was significantly more than that of
the low-risk group (p < 0.05), including B-cells, CD8 + T cells, iDCs,
pDCs, macrophages, neutrophils, T-helper cells, Tfh, Th1 cells, Th2
cells, TIL, and Tregs ([84]Figure 5A, [85]Supplementary Tables 6, 7,
and 9). Further analysis showed that compared with the low-risk group,
the scores of 13 immune functions and pathways, including APC
co-inhibition, APC co-stimulation, check point, and T-cell
co-inhibition, were significantly higher in the high-risk group (p <
0.05) ([86]Figure 5B). From the abovementioned results, we can see that
the immune response is more active in the high-risk group than in the
low-risk group, and the poor prognosis of LGG patients in the high-risk
group may be related to negative immune regulation. In order to assess
the differences between LGG patients with different risk values and
immunophenotyping, according to the distribution of immunophenotyping
of different types of tumor samples in the TCGA database, we combined
inflammatory (Immune C3), lymphocyte-depleted (Immune C4), and
immunologically quiet (Immune C5) responses. Three types were included
in the LGG study, and the results showed that there were significant
differences between C3 and C5, and C4 and C5 (p < 0.05), and Immune C3
had the largest risk score for LGG patients, whereas considering Immune
C5 in LGG patients, the risk was minimal ([87]Figure 5C). In order to
evaluate the feasibility of the inflammation-related risk model in
predicting the response of immunotherapy, we conducted a correlation
study between the risk score and three inhibitory immune checkpoints.
The results showed that the expressions of PD-1, PD-L1, and CTLA4 were
significantly upregulated in the high-risk group relative to the
low-risk group, and the expressions of PD-1, PD-L1, and CTLA4 were
positively correlated with the risk score (p < 0.05) ([88]Figures
5D–I). Therefore, the abovementioned results indicate that patients in
the high-risk group can benefit from immunotherapy more clinically.
FIGURE 5.
[89]FIGURE 5
[90]Open in a new tab
Correlation analysis of the immune infiltration pattern. (A)
Differences of immune cell subsets in high- and low-risk groups of an
inflammation-related prognosis model. (B) The difference of the immune
function and pathway in the high- and low-risk groups of
inflammation-related prognosis models. (C) The difference between LGG
patients with different risk values and immune classification. (D) The
expression difference of PD-1 in LGG high- and low-risk groups. (E)
Scatterplot of PD-1 correlation with the risk score. (F) Expression
difference of PD-L1 in LGG high- and low-risk groups. (G) Scatterplot
of PD-L1 correlation with the risk score. (H) The expression difference
of CTLA4 in LGG high- and low-risk groups. (I) Scatterplot of CTLA4
correlation with the risk score.
Correlation Analysis of Tumor Microenvironment
In order to clarify the impact of tumor microenvironment on the
prognosis of LGG patients, we conducted a correlation analysis of the
risk score and tumor microenvironment. From the scatterplot, it can be
seen that both the immune score and the matrix score are significantly
positively correlated with the patient’s risk score (p < 0.001)
([91]Figures 6A,B), which indicates that the higher the content of
immune cells and stromal cells in LGG patients, the higher the
patient’s risk score. The greater the risk, the shorter the survival
period. Next, the results of the stem cell correlation analysis showed
that the risk score of LGG patients was significantly positively
correlated with the stem cell score (DNAs) (p < 0.001), and was
significantly negatively correlated with the stem cell score (RNAs) (p
< 0.001) ([92]Figures 6C,D). Therefore, the risk score of this
prognostic model may be closely related to the activity of cancer stem
cells.
FIGURE 6.
[93]FIGURE 6
[94]Open in a new tab
Correlation analysis of tumor microenvironment and stem cells. (A)
Scatterplot of correlation between the immune cell score and risk score
of LGG patients. (B) Scatterplot of correlation between the stromal
cell score and risk score in LGG patients. (C) Scatterplot of the
association between the stem cell score (DNAss) and risk score for LGG
patients. (D) Scatterplot of the association between the stem cell
score (RNAss) and risk score for LGG patients.
Drug Sensitivity Analysis
We obtained the top 16 drugs with the most statistically significant
differences, by performing a separate drug sensitivity analysis on
inflammation genes in the prognostic model. The results showed that the
expression of SELL was positively correlated with the sensitivity of
nelarabine, ifosfamide, bendamustine, dexamethasone Decadron,
melphalan, pipobroman, and lomustine. It is indicating that the higher
the expression of SELL, the stronger the sensitivity to the
abovementioned drugs. The expression of MMP14 was positively correlated
with the sensitivity of vemurafenib, cabozantinib, zoledronate,
simvastatin, encorafenib, and dabrafenib, but it was negatively
correlated with the sensitivity of dexrazoxane and palbociclib. In
addition, the higher the expression of CALCRL in LGG patients, the
patient’s sensitivity to imiquimod is stronger ([95]Figure 7). In order
to further improve the clinical value of inflammation-related prognosis
models for the treatment of glioma, we analyzed the commonly used drugs
in the clinical treatment of glioma, which include temozolomide,
procarbazine, nitrosourea, vinblastine, podophyllotoxin, platinum, and
molecular-targeted drugs targeting VEGF. The results showed that
cisplatin, etoposide, vinorelbine, pazopanib, and sorafenib were more
sensitive in the high-risk group than in the low-risk group, and
axitinib was relatively weak in the high-risk group (p < 0.05)
([96]Figures 8A–F, [97]Supplementary Tables 3, 4, and 8).
FIGURE 7.
[98]FIGURE 7
[99]Open in a new tab
Gene–drug sensitivity analysis based on the CellMiner database; the top
16 drugs with high correlation with gene expression in
inflammation-related prognosis models were screened.
FIGURE 8.
[100]FIGURE 8
[101]Open in a new tab
Model drug sensitivity analysis based on the “pRRophetic” R package;
LGG therapeutic drugs related to inflammation-related prognosis models
are screened. The drug sensitivity of (A) Cisplatin, (B) Etoposide, (C)
Vinorelbine, (D) Pazopanib, (E) Sorafenib, and (F) Axitinib in High and
Low Risk Groups respectively.
Discussion
The immortal proliferation of glioma cells continuously breaks the
balance of the stable internal environment of normal brain cells and
shapes the tumor microenvironment that is characterized by an
immune-inflammatory response. In recent years, many studies have shown
that some inflammatory cells (such as neutrophils and macrophages),
inflammatory factors (IL-8, IL-1beta, IL-6, CD8^+, and CXCL16) in the
inflammatory microenvironment of glioma, and related signal pathways
(NF-κB and STAT-3) are closely related to the progression and prognosis
of glioma ([102]Müller et al., 2017; [103]Albrengues et al., 2018;
[104]Greten and Grivennikov, 2019; [105]Zha et al., 2020). The
inflammatory microenvironment can provide a good material preparation
for tumor cell expansion and mutation, which makes the abnormally
activated inflammatory response part of the reason why the overall
survival rate of LGG is still low. Therefore, it is very meaningful to
carry out in-depth research on the prognosis of LGG patients from the
perspective of inflammation-related factors. Although the rapid
development of high-throughput sequencing in recent years has brought
new hope and direction for the accurate diagnosis and prognostic
judgment of LGG ([106]Kiran et al., 2019), so far, the number of
biomarkers used to predict LGG in clinical practice is poor, which
limits our early diagnosis and prediction of the therapeutic effect for
LGG patients. It is shown that finding reliable and effective
biomarkers is of great significance to prognosis prediction and
clinical treatment of LGG. It has been confirmed that new biomarkers
including glycoprotein YKL-40, microRNA, 21-mRNA, lncRNA, and BRAF gene
mutations have significant relevance to the prognosis of LGG
([107]Gandhi et al., 2018; [108]Zhang et al., 2019; [109]Song et al.,
2020; [110]Maimaiti et al., 2021; [111]Gluexam et al., 2019; [112]Zhang
et al., 2020). In addition, serum biomarkers related to inflammation
(neutrophil–lymphocyte ratio), adipokines, immune-related gene markers,
autophagy-related genes (ARGs), energy metabolism genes, and repeated
mutation genes (IDH, MGMT, EGFR, and chromosome 1p/19q deletion) are
important prognostic factors for LGG ([113]Chen et al., 2021; [114]Zhou
et al., 2018; [115]Sun et al., 2014; [116]Vachher et al., 2020;
[117]Wang et al., 2019; [118]Ali et al., 2021; [119]Zheng et al., 2018;
[120]Wu et al., 2019). However, these clinical pathologic and genetic
factors are not specific, and may not achieve an accurate assessment of
the survival rate of LGG. Therefore, it is important to conduct more
comprehensive studies to increase the prognostic and predictive
accuracy of the current assessment system. This study systematically
analyzed the expression of 200 inflammation-related genes in LGG
tissues and their relationship with the overall survival (OS).
Thirteen differentially expressed inflammatory genes were screened from
the TCGA cohort, and after single-factor Cox analysis was conducted,
nine inflammatory genes were observed to be related to LGG OS. Then, a
prognosis model integrating three inflammatory response–related genes
was finally constructed by LASSO regression analysis. According to the
median risk score, patients were divided into the high-risk group and
low-risk group. We found that the high-risk group was significantly
associated with higher tumor grade, advanced TNM stage, and shorter OS
stage. Independent prognostic analysis showed that the risk score was
an independent predictor of OS.
The prognosis model established in this study consisted of three
inflammatory response–related genes (CALCRL, MMP14, and SELL), which
were upregulated in LGG tumor tissues. The CALCRL gene codes the
calcitonin receptor–like receptor which is a seven-transmembrane domain
G-protein–coupled receptor ([121]Larrue et al., 2021 and [122]Angenendt
et al., 2019). Its expression products are closely related to CALCRL
and RAMP expressed on the cell surface by co-expression with three
receptor active modification proteins (RAMPs). CALCR can act as
calcitonin gene–related peptide receptor 1 (CGRP1) when co-expressed
with RAMP1, and when RAMP2 or RAMP3 exists, CALCR and its formed
complexes can act as an adrenomedullin (ADM) receptor ([123]Hay et al.,
2011; [124]Russell et al., 2014). CGRP is a neuropeptide which can
promote tumor-related angiogenesis and tumor proliferation by
regulating the signal transduction of the downstream molecule VEGF,
which plays a key role in the occurrence and development of tumors
([125]Toda et al., 2008). In addition to having the function of
vasodilation to regulate blood pressure, ADM can be used as a hypoxia
regulator to avoid the death of malignant tumor cells due to hypoxia
and promote tumor cell growth ([126]Russell et al., 2014). Therefore,
CALCRL mediates the occurrence and development of tumor cells such as
CGRP and ADM. Relevant studies have proved that the mRNAs of
CALCRL/RAMP2 and CALCRL/RAMP3 are highly expressed in glioblastoma cell
lines ([127]Metellus et al., 2011). In addition, Benes L et al. have
demonstrated that CRLR is widely distributed in human gliomas of
different grades; at the same time, their team revealed that the
possible mechanism of CRLR in gliomas is related to its influence on
the formation of blood vessels, which assists the growth of gliomas
([128]Ouafik et al., 2002; [129]Benes et al., 2004). However, its
mechanism of action in LGG is still unclear. Based on previous studies,
we suspect that CRLR in low-grade glioma may boost the occurrence and
development of LGG by influencing angiogenesis-related factors. Of
course, this hypothesis still needs to be verified by experiment.
MMP-14 is a subfamily of the matrix metalloproteinase family (MMPs).
Studies have shown that MMP-14 can be used as a prognostic marker for
patients with glioma ([130]Wang et al., 2013). And some researchers
found that the mechanism of MMP14 in glioma mainly works by cutting
CD44 ([131]Kajita et al., 2001), or via the combination of TIMP-2 and
MMP14 to activate MMP-2 and MMP9 to enhance tumor invasion and tumor
cell proliferation ([132]Chernov et al., 2009). In addition, it also
plays an important role in angiogenesis ([133]Ulasov et al., 2014;
[134]Chen et al., 2016; [135]Rooprai et al., 2016). Thus, MMP-14 is
extremely important in predicting the prognosis of patients with
glioma. The SELL gene is a gene encoding L-selectin with the smallest
relative molecular mass in the selectin family. It is mainly
distributed on the surface of white blood cells, endothelial cells, and
platelets. SELL plays an extremely important physiological role in the
development and metastasis of tumors. Tanriverdi et al. found that
compared with low-grade glioma patients, selectins (E, L, and
P-selectins), leukocyte adhesion molecules (ALCAM), and platelet
endothelial cell adhesion molecules-1 (PECAM-1) are highly expressed in
patients with high-grade gliomas. L-selectin pushes tumor plasma cell
metastasis, which may be its main mechanism in glioma ([136]Tanriverdi
et al., 2017). In addition, recent evidence suggests that L-selectin
may be an important target for cancer immunotherapy. Watson et al.
discovered that in the immunotherapy of adoptive T-cell carcinoma in
mouse models, L-selectin, which is overexpressed in T-cells, is related
to the infiltration and enhanced proliferation of T-cells in tumors and
controlling the growth of the tumor to a certain extent ([137]Watson et
al., 2019). In this study, our team found that SELL is differentially
expressed in LGG, which indicates that it may be closely related to the
degree of immune cell infiltration in tumors.
With deeper understanding of the relationship between inflammation and
tumors, in recent years, studies have found that the relationship
between inflammation and the immune system of tumor cells cannot be
ignored. Zong Z et al. found that the inflammatory cytokine IL-1β in
liver cancer can induce PD-L1 expression through the transcription
factors p65 and IRF1, which creates opportunities for tumor cells to
escape the immune system ([138]Zong et al., 2019). Zhang W et al.
reported that the upregulated IL-6 in liver cancer can downregulate the
O-type protein tyrosine phosphatase receptor (PTPRO) by activating
JAK2/STAT3 signal transduction, resulting in high PD-L1 expression,
inducing immunosuppression and promoting tumor growth ([139]Zhang et
al., 2020). These studies show that when a powerful foreign factor
invades, the body needs to activate a stronger defense system–immune
response, and activated immune cells will produce inflammatory factors,
which can pass immune- and inflammation-related cell signaling pathways
to induce tumor cells to highly express immunosuppressive signal
molecules, and further induce the occurrence of tumors. This study
investigated the inflammation-related risk model to predict the
immunotherapy response and found that the expression of PD-1, PD-L1,
and CTLA4 was significantly upregulated in the high-risk group, and the
expression and risk score of PD-1, PD-L1, and CTLA4 are positively
correlated. Therefore, the abovementioned results indicate that
patients in the high-risk group can benefit from immunotherapy more
clinically. Besides, Berghoff AS et al. reported that the
pro-inflammatory cytokine IFN-γ can drive the high expression of PD-L1
in tumor cells ([140]Berghoff et al., 2015). Bloch O et al. found that
glioma promotes immune suppression by regulating IL-10 signal
transduction and then upregulating the expression of B7-H1 in tumor
infiltration–related macrophages, resulting in high PD-L1 expression
([141]Bloch et al., 2013). [142]Liu et al., 2019 showed that
inflammatory factors can upregulate the expression of PD-L1/PD-1 in
tumor cells to assist tumor cell immune escape and promote tumor cell
proliferation ([143]Liu et al., 2020). On these bases, our research
found that based on GSEA analysis, tumor-related signaling pathways
such as antigen processing and presentation, primary immunodeficiency,
natural killer cell–mediated cytotoxicity, intestinal immune IGA
production network, B-cell receptor signaling pathway, and T-cell
receptor 64 pathways such as body signaling pathways are significantly
abundant in the high-risk group, and the continuous activation of these
pathways has been confirmed to be related to LGG. A total of 13
immune-related functional pathways, including APC co-inhibition, APC
co-stimulation, check point, and T-cell co-inhibition, were more
significant in the high-risk group; B-cells, CD8^+ T-cell, iDCs, pDCs,
macrophages, neutrophils, T-helper cells, Tfh, Th1 cells, Th2 cells,
TIL, Tregs, and other immune cell subsets were significantly enriched
in the high-risk group, further verifying that inflammation is closely
related to tumor progression. The increased activity of Tfh, Th1 cells,
Th2 cells, TIL, and Tregs in the high-risk group indicates that the
immune regulation function of the high-risk group is disturbed, and the
antitumor immunity is weakened. In addition, the proportion of
macrophages, neutrophils, and Treg cells is higher in the high-risk
group, indicating that their increase can promote immune invasion,
which is closely related to the poor prognosis of LGG patients. In
order to gain a deeper understanding of the relationship between risk
scores and immune components, we examined the role of risk scores in
the types of immune infiltration. We found that the high-risk score was
significantly correlated with C3, whereas the low-risk score was
correlated with C5, indicating that C3 promotes the occurrence and
development of tumors. This finding is consistent with the results of
previous studies. By blocking the PD-1/PD-L1 pathway, immune checkpoint
inhibitors can relieve the tumor microenvironment’s inhibitory effect
on immune cells and activate the body’s immune function to achieve
antitumor effects. Immunotherapy based on immune checkpoint inhibitors
has been widely used in other types of tumors. Although, so far,
immunotherapy has not been approved for glioma treatment, some
preclinical studies have shown its therapeutic potential. For example,
a randomized trial was conducted in 35 patients with recurrent
glioblastoma (GBM). The survival time of the neoadjuvant pembrolizumab
group was longer than that of the adjuvant pembrolizumab group (median
PFS was 2.4 and 3.3 months, and median OS was 13.7 and 7.5 months,
respectively). The trials have shown the therapeutic potential of
immunotherapy in the neoadjuvant treatment of GBM ([144]Cloughesy et
al., 2019). A phase II clinical trial ([145]NCT03890952) is currently
underway to evaluate PD-L1 and other immune biomarkers. It mainly
compares the combination of nivolumab and bevacizumab in patients who
have not undergone surgery and those undergoing surgery. It is expected
that nivolumab can be obtained from recurrent GBM ([146]Zimmer et al.,
2020). Also, Hideho Okada et al. used glioma-associated antigen (GAA)
immune-vaccine therapy to target and activate cytotoxic T cells (CTL)
on the cell surface to treat LGG ([147]Sanders and Debinski, 2020).
Studies have confirmed GAA. It has strong specificity and good
tolerance, and can effectively control the occurrence and development
of LGG. Although most of the current clinical studies are mainly
targeting glioblastoma with a high degree of malignancy, there are few
clinical trials of immunotherapy in LGG patients, but a relatively high
specific immune checkpoint has been found to effectively screen the
benefits of immunotherapy. The population will be of great significance
in the treatment of LGG patients and the prevention of malignant
transformation and recurrence of low-grade gliomas.
It has been confirmed that inflammatory cytokines can induce
epithelial–mesenchymal cell transformation (EMT) and cancer stem cell
(CSC) production and related molecular regulation to establish an
inseparable connection with the tumor microenvironment
([148]Markopoulos et al., 2019). Cancer stem cells (CSCs) are often
referred to as tumor-initiating cells. Due to their self-renewal
ability and heterogeneity, they are the main cause of tumor resistance,
recurrence, and metastasis ([149]Scheel and Weinberg, 2012). In our
study, the correlation between the prognostic gene expression and
cancer stem cell score suggests that the risk score of the model
composed of CALCRL, MMP14, and SELL is significantly positively
correlated with the stem cell score (DNAss), suggesting that the risk
of this prognostic model may be closely related to the activity of
cancer stem cells. At the same time, this study clarifies that the
increased expression of SELL is related to elevation in the sensitivity
of cancer cells to lomustine. Lomustine is a drug containing a
classical chemotherapy regimen in the guide for WHO grade II gliomas
PCR regimen (promethazine + lomustine + Changchun Neobase) ([150]Shaw
et al., 2012). Therefore, the high expression of the SELL gene can
predict the sensitivity of LGG patients to this drug. The expression of
MMP14 is positively correlated with the sensitivity of vemurafenib.
Vemurafenib is a competitive small-molecule BRAF V600E inhibitor that
can act on BRAF V600E mutations in low-grade gliomas. It has been
proven to be effective in treating metastasis melanoma which is prone
to BRAF mutation. However, recent clinical trials have shown that the
drug has good efficacy in BRAF V600E mutant malignant astrocytomas and
low-grade gliomas; patients with high expression of MMP14 may predict
the better curative effect of vemurafenib treatment ([151]Del et al.,
2018; [152]Van et al., 2018). Therefore, the genes in our prognostic
model can be used as targets to predict drug sensitivity.
In summary, our study has determined a new prognostic marker model
composed of three inflammatory response–related genes. In the TCGC
database, this model has been proven to be independently related to OS,
and it has been proven to be of great significance in regulating the
immune microenvironment, tumor microenvironment, and drug sensitivity.
It provides a novel idea and method for LGG prognosis, immunotherapy,
and drug sensitivity evaluation. However, the specific underlying
mechanism between LGG inflammation–related genes and tumor immunity is
still unclear, and further research is needed.
Conclusion
The inflammatory gene–related prediction model proposed in this study
is of great significance for the screening of prognostic markers in LGG
patients, especially in the exploration of immune response, tumor
microenvironment, and immunotherapy sensitivity. It is expected to be
the basic research and immunotherapy of LGG immunity. The choice of
method provides an important reference and clinical transformation
value.
Data Availability Statement
The datasets presented in this study can be found in online
repositories. The names of the repository/repositories and accession
number(s) can be found in the article/[153]Supplementary Material.
Author Contributions
TH and ZZ contributed to the conception and design of this study. YZ,
ZZ, and MQ performed data collection, statistical analysis, and
interpretation. QL and HW contributed to administrative support. All
authors wrote the first draft of the manuscript. All authors
contributed to the revision and proofreading of the manuscript. All
authors contributed and approved the submitted version of the
manuscript.
Funding
This study was supported by the grants of the Natural Science
Foundation of Liaoning Province (20170540251).
Conflict of Interest
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.
Publisher’s Note
All claims expressed in this article are solely those of the authors
and do not necessarily represent those of their affiliated
organizations, or those of the publisher, the editors, and the
reviewers. Any product that may be evaluated in this article, or claim
that may be made by its manufacturer, is not guaranteed or endorsed by
the publisher.
Supplementary Material
The Supplementary Material for this article can be found online at:
[154]https://www.frontiersin.org/articles/10.3389/fphar.2021.748993/ful
l#supplementary-material
[155]Click here for additional data file.^ (1.1MB, XLSX)
[156]Click here for additional data file.^ (11.6KB, XLSX)
[157]Click here for additional data file.^ (32.2KB, XLS)
[158]Click here for additional data file.^ (44.1KB, XLSX)
[159]Click here for additional data file.^ (213.5KB, XLSX)
[160]Click here for additional data file.^ (383.7KB, XLSX)
[161]Click here for additional data file.^ (10.7KB, XLSX)
[162]Click here for additional data file.^ (2.3MB, XLSX)
[163]Click here for additional data file.^ (592.6KB, XLSX)
[164]Click here for additional data file.^ (27.9KB, XLSX)
[165]Click here for additional data file.^ (171.1KB, XLSX)
[166]Click here for additional data file.^ (296.7KB, XLSX)
References