Lower-grade glioma (LGG) is a malignant tumor that primarily occurs in the brain,1,2 and successfully optimizing its treatment process is still challenging.3,4 In general, MRI is used for its diagnosis, whereas its subtypes and grades are defined based on histological and molecular assessment of biopsy specimens.5,6 Specifically, if LGG is associated with favorable prognostic factors, including age younger than 40 years, no neurological deficits, and a low tumor burden, the recommended treatment strategies are resection followed by “wait-and-see” or radiation therapy with adjuvant procarbazine, carmustine, and vincristine (PCV) or temozolomide chemotherapy.4,7 Conversely, in the case of less favorable prognostic factors, such as age older than 40 years, the presence of neurological deficits, and residual tumor, the only treatment strategy is resection with radiation therapy, followed by PCV or temozolomide without the wait-and-see protocol.4,8
Most radiological assessments for glioma are usually conducted based on specific glioma symptoms, including seizures and neurological deficits.9,10 However, a few cases of LGG are incidentally discovered during medical checkups for conditions such as trauma or simple headache.11,12 For such incidental LGG (iLGG), it is more challenging to decide on the treatment strategy to apply. Thus, there is still controversy regarding whether the wait-and-see strategy or immediate treatment should be applied. Several studies conducted over several decades have shown that iLGG eventually progresses over time; however, its natural history shows a more favorable prognosis than that of symptomatic LGG (sLGG).11,13 Surgically treated iLGG is predominantly IDH mutated and 1p19q codeleted, with favorable prognosis, and patients with iLGG who were actively treated as fast as possible have better prognosis empirically.11,14,15
A previous study showed that, comparatively, the group in which early resection was favored was associated with better overall survival than the group in which biopsy and “watch and see” was favored.16 This survival benefit was maintained even after a 5-year follow-up period.17 Furthermore, based on several clinical investigations involving iLGG, several hypotheses regarding the nature of iLGG with respect to glioma progression, from initiation to malignant transformation, have been proposed. Among them, the most reasonable is that glioma can be divided into four stages: 1) the occult stage, which corresponds to the period between biological birth and radiological birth; 2) the silent stage, which corresponds to the period from radiological birth to the appearance of the first symptoms; 3) the symptomatic stage; and 4) the malignant transformation stage.18 Based on this hypothesis, sLGG is a more progressed and malignant stage of iLGG in the chronology of the disease’s progression.18 However, given that most previous studies only describe the clinical nature of iLGG, its other molecular characteristics are still unclear. Thus, the treatment strategy before the development of glioma-specific symptoms has not yet been optimized between immediate treatment after iLGG diagnosis and wait and see.18–21 Based on the results of clinical investigations, we hypothesized that iLGG precedes sLGG and investigated whether differences existed between their molecular characteristics. We also investigated whether these characteristics could lead to differences in response to treatment.
Methods
Glioma Database Description and Sample Selection
The Cancer Genome Atlas Data
To conduct the analyses realized in this retrospective study we used GBMLGG data, including mutation, clinical, survival, and RNA-seq data. The clinical and survival data were downloaded from the University of California, Santa Cruz’s XENA database (https://xena.ucsc.edu/), whereas mutation and raw count RNA-seq data were downloaded from the Genomic Data Commons website (https://gdc.cancer.gov/about-data/publications/mc3-2017). The selected samples in the GBMLGG data set comprised 662 samples, including primary WHO grade II, III, and IV glioma cases. Furthermore, the glioma subtypes were astrocytoma, oligodendroglioma, oligoastrocytoma, and glioblastoma. Additionally, the symptoms specified in the clinical data included the following: headache, seizure, visual change, sensory change, mental status change, and motor movement change. A glioma with no symptoms or only headache was selected as an iLGG. Thus, the number of iLGG and sLGG cases labeled based on symptom status were 31 and 248, respectively. Detailed information corresponding to these samples is provided in Table S1. Furthermore, the overall progress of this study is provided as a flowchart in Fig. S1.
Chinese Glioma Genome Atlas Data
Chinese Glioma Genome Atlas (CGGA) data (http://www.cgga.org.cn/), including clinical and RNA-seq data, were used in this study to validate the results of survival analysis performed using The Cancer Genome Atlas (TCGA) data. All the samples corresponding to the CGGA database were primary and WHO grades II, III, and IV glioma subtypes (astrocytoma, oligodendroglioma, oligoastrocytoma, and glioblastoma), and the number of included samples was 422.
Rembrandt Data
Rembrandt array gene expression data and clinical data were downloaded from the Rembrandt database (GSE108474, https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE108474) and used to validate the significance of the results of the survival analyses performed using TCGA and CGGA data. The data used in the survival analysis included 244 samples, which comprised primary and WHO grades II, III, and IV glioma subtypes (astrocytoma, oligodendroglioma, mixed glioma, and glioblastoma, respectively).
DepMap Data
Drug sensitivity data and RNA-seq expression data obtained based on the assessment of cancer cell lines from the Cancer Dependency Map (https://depmap.org/portal/) were used to analyze the correlation between gene expression and drug sensitivity. The drug sensitivity types analyzed in this study included procarbazine drug sensitivity area under the curve (Cancer Target Discovery and Development [CTD2]), vincristine drug sensitivity, carmustine drug sensitivity, and temozolomide drug sensitivity (PRISM Repurposing Primary Screen 19Q4 in all three instances). RNA-seq expression data were expression data released in 21Q4. Information regarding the cell lines used and the drug sensitivity tests performed is described in Table S2.
Mutation Analysis
Using TCGA data comprising 277 samples, we performed mutation analysis to investigate the differences between iLGG and sLGG. The entire mutation analysis process was conducted using maftools v2.10.05, which is one of the packages included in the R software.22 The command “plotmafSummary” was used to profile the mutation statuses in iLGG and sLGG. Furthermore, oncoplot was visualized using the “oncoplot” command, and cobar plot was used to compare the top 5 mutations in iLGG and sLGG using the command “coBarplot.” Tumor mutation burden (TMB) was analyzed and compared using the command “tmb.” Given that the length of the human exon is approximately 38 MB, TMB was calculated as the total mutations/38.23 Co-occurrence and mutually exclusive mutations were visualized using the command “somaticInteractions,” and the p values were calculated by performing the pairwise Fisher exact test. Additionally, the command “OncogenicPathway” was used to detect well-known oncogenic pathways.24
Differentially Expressed Gene Analysis
Differentially expressed genes (DEGs) between iLGG and sLGG based on TCGA data were analyzed using the DESeq2 package in R.25 Genes with an absolute value of log2-fold change > 1 and Benjamini-Hochberg adjusted p value < 0.05 were included. The conversion of gene names from ensemble ID to gene symbols was conducted using ShinyGO v0.741.26
Gene Ontology Enrichment Analysis
Gene ontology (GO; http://geneontology.org/) enrichment analysis was performed using ShinyGO software, and significantly enriched terms with FDR correction < 0.05 were extracted using ShinyGO v0.741.26 The p values of the enriched pathways were presented as −log10 values.
Survival Analysis
Log-rank tests and Cox regression analysis were performed, and Kaplan-Meier curves were generated to compare prognosis with respect to iLGG or sLGG. Survival analysis was then conducted using survival v3.2 (https://cran.r-project.org/web/packages/survival/index.html). Thereafter, the generated survival plots were visualized using survminer v0.4.8 (https://cran.r-project.org/web/packages/survminer/index.html).
Statistical Analysis
Statistical analyses were conducted to investigate the clinical significance of iLGG using the t-test and chi-square test, or Fisher’s exact test. Specifically, the t-test was conducted to determine differences between TMB, which were then visualized using the ggpubr package (https://cran.r-project.org/web/packages/ggpubr/index.html) in R. Post hoc power analysis was conducted to calculate the power of comparing TMB with G*power.27 Furthermore, chi-square and Fisher’s exact tests were conducted to compare the proportion of mutation statuses and degree of alteration in 10 well-known oncogenic pathways corresponding to iLGG and sLGG. Correlation analyses were also conducted to assess the correlation between gene expression and drug sensitivity.
Results
Selection of iLGG Cases From the TCGA Database and the Associated Clinical Characteristics
The iLGG and sLGG cases were classified based on symptom data in the TCGA clinical database. Specifically, based on seizure history, headache history, sensory change, visual change, mental status change, and motor movement change, iLGG samples were considered as glioma if there were no symptoms or headache only. Otherwise, in cases with seizure only, seizure accompanied by other symptoms, and other symptoms without seizures, sLGG was considered. Thus, a total of 31 and 248 iLGG and sLGG samples were identified in this study based on TCGA data (Table 1). Statistical analysis was conducted to compare clinical statuses of these 279 samples. Thus, two categories—age and Karnofsky Performance Scale (KPS) score—were compared by performing the Student t-test, whereas the other categories were compared using the chi-square test or Fisher exact test. All the different categories, with the exception of KPS score, showed no significant differences between the iLGG and sLGG samples (Table S3). However, the KPS score corresponding to sLGG samples was significantly lower than that corresponding to iLGG samples.
Criteria for the classification of iLGG and sLGG based on the TCGA data set
| Tumor Type & Symptoms | Sample Size (no.) |
|---|---|
| iLGG | 31 |
| No symptoms | 11 |
| Headache | 20 |
| sLGG | 248 |
| Only seizure | 99 |
| Seizure w/ other symptoms | 73 |
| w/o seizure | 76 |
| Overall total | 279 |
Similarity of the Genomic Landscapes Corresponding to iLGG and sLGG Samples
Both iLGG and sLGG samples showed similar missense mutation proportion tendencies, and their single nucleotide polymorphisms were also similar. However, sLGG showed additional oligonucleotide polymorphisms, and the proportion of C > A was larger in iLGG samples than in sLGG samples. Additionally, the distribution of mutation counts in each sample and the tendency of the occurrence of the top 7 mutations in iLGG and sLGG samples were similar (Fig. 1A and B). Considering the top 10 mutations, both iLGG and sLGG samples showed similar rankings for the top 7 mutations and the NOTCH1 mutation (ranked 10). However, mutations ranked at 8 and 9 in iLGG samples were ANKRD11 and DNMT3A, respectively, and in sLGG these were PIK3CA and EGFR, respectively (Fig. 1C). This notwithstanding, there were no statistically significant differences between iLGG and sLGG samples with respect to the proportion of mutations (Fig. 1D).
Description of the mutation profiles of iLGG and sLGG. A: Summary of mutation status in iLGG. B: Summary of mutation status in sLGG. C: Cobar plot of top 10 mutations in iLGG and sLGG. D: Chi-square analysis of 12 mutations between iLGG and sLGG. After Bonferroni correction, the results of the chi-square analysis showed no significant correlations. DEL = deletion; INS = insertion; MT = mutant; ONP = oligonucleotide polymorphism; SNP = single nucleotide polymorphism; SNV = single nucleotide variant; WT = wild type.
Furthermore, the medians of the TMBs in iLGG and sLGG samples were approximately 0.74 and 0.79, respectively (Fig. 2A). To compare the two iLGG and sLGG samples with respect to TMB, a t-test was conducted. Thus, no statistically significant difference was observed in this regard, but the power was 0.259 (Fig. 2B). The proportions of the top 10 mutations, including IDH1, TP53, ATRX, CIC, and TTN mutations in both iLGG and sLGG samples, were visualized as a bar plot. Furthermore, a comparison of the proportions of these top 10 mutations via chi-square and Fisher exact tests showed no statistically significant difference between the iLGG and sLGG samples.
TMB and interaction between somatic mutations in iLGG and sLGG. A: Distribution of TMB in each sample of iLGG and sLGG. B: Boxplot showing a comparison of the TMB results corresponding to iLGG and sLGG via t-test. C: Interactions between the top 10 mutations in iLGG. D: Interactions between the top 10 mutations in sLGG. ns = not significant.
sLGG showed a higher level of interactions between the top 10 somatic mutations than did iLGG. Specifically, in iLGG, the co-occurrence of TP53-ATRX, CIC-NOTCH1, ANKRD11-NOTCH1, and ANKRD11-DNMT3A and the mutual exclusiveness between TP53-CIC and ATRX-CIC were significant (Fig. 2C). In sLGG, the co-occurrence of IDH1-TP53, IDH1-ATRX, IDH1-CIC, TP53-ATRX, CIC-FUBP1, CIC-PIK3CA, and CIC-NOTCH1 and the mutual exclusiveness between IDH1-EGFR, TP53-CIC, TP53-FUBP1, TP53-PIK3CA, TP53-EGFR, ATRX-CIC, ATRX-FUBP1, ATRX-PIK3CA, ATRX-EGFR, and CIC-EGFR were significant (Fig. 2D). Thus, sLGG samples showed a higher degree of co-occurrence and mutually exclusive interactions between the top 10 mutations than did iLGG samples.
Degree of Oncogenic Pathway Alteration in sLGG and iLGG
To compare the degree of impairment in 10 oncogenic pathways, the fractions of affected pathways and samples were determined.24 For iLGG only 8 of the 10 considered oncogenic pathways were affected, whereas in sLGG all the oncogenic pathways were affected (Fig. 3A and B). Furthermore, a comparison of the proportion of affected pathways and samples corresponding to the iLGG and sLGG samples via chi-square analysis and Fisher exact test showed no significant difference with respect to the proportion of samples affected. However, 5 oncogenic pathways, including the RTK-RAS, WNT, Hippo, TP53, and PI3K signaling pathways, were significantly more affected in sLGG than in iLGG (Fig. 3C). Furthermore, to compare the included gene mutations, the fractions of the affected pathways were visualized (Fig. S2A and S2B), and considering that the difference between iLGG and sLGG with respect to TMB was not significant, sLGG had a greater number of mutated genes in 5 oncogenic pathways than did iLGG.
TMB, glioma signature mutation, and oncogenic pathway profiling of iLGG and sLGG. A: Affected oncogenic pathways in iLGG. B: Affected oncogenic pathways in sLGG. C: Statistical analysis of the correlation between the affected oncogenic pathways in iLGG and sLGG.
sLGG Showed Malignant Characteristics, a Worse Prognosis, and Higher Drug Resistance Than iLGG Based on Transcriptomics
The analysis of DEGs was performed based on transcriptomics to compare the characteristics of iLGG and sLGG. Thus, we observed that a total of 189 genes were more significantly upregulated in sLGG and 8 genes were more significantly downregulated in iLGG (Fig. 4A). Details regarding all the DEGs are provided in Table S4. In a previous study involving the use of transcriptomics of iLGG based on CGGA data, a total of 2331 and 484 significantly upregulated and downregulated genes, respectively, were identified.28 Overall, considering both TCGA and CGGA data, a total of 75 genes were co-upregulated (Fig. 4B), and based on these 75 genes, GO enrichment analysis was performed using the GO Biological Process database (Fig. 4C). Thus, we observed that the positive regulation of cell proliferation, cell migration, epithelial-to-mesenchymal transition, and angiogenesis, and also the negative regulation of apoptosis were more significantly upregulated in sLGG than iLGG. The response to ionizing radiation, including radiography, was also more significantly upregulated in sLGG. Additionally, the prognostic value of the expression of the selected genes was assessed via Cox regression analysis involving primary glioma grades II, III, and IV samples from three cohorts, including TCGA, CGGA, and the Rembrandt database (Fig. 4D). Most of the 75 coexpressed genes showed statistically significant prognostic value in at least two cohorts, with hazard ratios greater than 1, except HOXC8, which showed an adverse effect between TCGA, CGGA, and Rembrandt. Furthermore, C6 and LPAR3 did not show any significant prognostic value in every cohort, whereas WNT2 played a protective role in LGG samples from the Rembrandt cohort.
Comparison of biological process activity, prognosis in iLGG and sLGG. A: Volcano plot based on DEGs. The blue dots represent genes with log2-fold change ≤ 1 and adjusted p value < 0.05, whereas the red dots represent genes with log2-fold change > 1 and adjusted p value < 0.05. B: Significantly upregulated genes in sLGG within TCGA and CGGA. C: Bubble plot with GO biological process enrichment analysis based on 75 co-upregulated genes. D: Cox regression analysis involving 75 genes corresponding to all the glioma samples from the TCGA, CGGA, and Rembrandt cohorts. AS = significant in all; NS = not significant; RS = significant in Rembrandt; TCS = significant in TCGA and CGGA; TRS = significant in TCGA and Rembrandt; TS = significant in TCGA. Figure is available in color online only.
To predict the difference between iLGG and sLGG with respect to therapeutic sensitivity, statistical analysis was conducted based on drug sensitivity data and RNA-seq data from the DepMap portal. Given that PCV and temozolomide are frequently used in glioma chemotherapy, four categories of drug sensitivity data were selected. Thereafter, correlation analysis was performed between the expression of each gene and drug sensitivity in glioma cell lines. The genes that showed positive correlation between their expression and drug sensitivity data were found to be positively correlated with drug resistance. The threshold was |correlation coefficients| > 0.3 and p value < 0.05. Details in this regard have been provided in Tables S5–S8. Specifically, SERPINA3, GPR65, LTF, ADAMTS18, and GALNT5 were upregulated in LGG and showed positive correlation with resistance to PCV, and TNFRSF12A, FOSL1, and ANXA1 were upregulated in LGG and showed positive correlation with resistance to temozolomide (Fig. 5). GPR65 was positively correlated with both procarbazine and carmustine resistance. To assess the impact of these 8 genes on prognosis, log-rank tests involving iLGG and sLGG samples from the TCGA data set were used. Ultimately, all 8 genes showed significantly worse prognosis in the higher-expression group (Fig. S3). Furthermore, the expression levels of 9 genes showed negative correlation with drug resistance. These genes were upregulated in sLGG, and the impact of their expression on prognosis was also worse in the higher-expression groups (Fig. S4).
Comparing drug resistance features between iLGG and sLGG. Common upregulated genes in sLGG and genes correlated with drug sensitivity. NC = genes negatively correlated with drug resistance; PC = genes positively correlated with drug resistance. Figure is available in color online only.
Discussion
Based on existing literature regarding the clinical characteristics of iLGG, the prognosis of iLGG is better than that of sLGG.11,13,14,28 In this study we investigated the differences between the molecular characteristics of iLGG and sLGG samples from the TCGA cohort. Thus, we observed that except for KPS score, which was lower for sLGG than iLGG, both samples showed similar clinical characteristics. Furthermore, in terms of mutation landscape, both groups showed similar variant types, variant classifications, and TMB. However, transcriptomic analysis showed several differentially upregulated biological processes between the two groups and worse prognosis for sLGG cases.
Comparing the top 10 mutations in iLGG and sLGG, the general genetic features, including IDH1, TP53, ATRX, CIC, FUBP1, and NOTCH1 mutations, were consistent with observations made in previously reported studies.29,30 Additionally, the rankings of TTN and MUC16 mutations were the same in both iLGG and sLGG samples, and the two groups showed no significant difference with respect to the proportion of IDH1 mutations. EGFR mutations, which predominantly occur in IDH1 wild-type glioma, were among the top 9 mutations in sLGG.30 Conversely, ANKRD11 and DNMT3A mutations were ranked as 8 and 9, respectively, in iLGG. Furthermore, considering the top 10 mutations, the mutational characteristics of astrocytoma and oligodendroglioma were obvious in both iLGG and sLGG; however, the characteristics of IDH1 wild-type glioma were clearer in sLGG than in iLGG.
Even though the proportions of the mutations that are most frequently observed in glioma corresponding to sLGG and iLGG samples showed no statistically significant difference, the interactions between these mutations were increased in sLGG. The degree of interaction between mutations was predominantly higher for the signature mutations corresponding to represent each glioma histological subtype, including astrocytoma, oligodendroglioma, and IDH1 wild-type LGG. The tendency of co-occurrence mutation groups was classified as IDH1, TP53, and ATRX mutations, which are the signature mutations of astrocytoma; IDH1, CIC, FUBP1, and NOTCH1, which are the signature mutations of oligodendroglioma; and EGFR, NF1, and PTEN in sLGG.30 Notably, the mutually exclusive effect also increased between these mutation groups. Conversely, iLGG samples showed weak mutation interactions. From these results it can be inferred that the characteristics of the different glioma subtypes were clearer in sLGG. Furthermore, it has been reported that the interactions between somatic mutations, which are usually accompanied by the accumulation of mutations, become stronger with cancer progression.31,32 Specifically, in a previous study in which the genetic characteristics of glioma were described, it was hypothesized that different histological and molecular glioma subtypes develop through progression of different biological pathways.33 Taking these reports and our results into consideration, sLGG is possibly a progressed, malignant form of iLGG.
In this context, the impairment of the mutation of oncogenic pathway components increased to a greater extent in sLGG than in iLGG. Conversely, the proportions of the affected samples in each group were not significantly different; however, five pathways, including the RTK-RAS, WNT, Hippo, TP53, and PI3K pathways, were affected to a greater extent in sLGG than in iLGG. Specifically, the deregulation of the RTK-RAS, PI3K, and TP53 pathways—which are defined as the core signaling pathways—via genetic alteration is frequently observed in glioblastoma.34,35 Additionally, the WNT signaling pathway is also associated with glioma progression and resistance to radiation therapy.36 Regarding the Hippo signaling pathway, its activation induced by deregulation showed correlation with malignancy, poor prognosis, and drug resistance.37 Therefore, considering these features, sLGG may be more malignant in nature than iLGG in terms of the genetic alteration of oncogenic pathways. Additionally, in sLGG samples, the NRF2 and TGF-beta pathways were affected to a greater extent than in iLGG samples (Fig. S5).
With respect to transcriptomic characteristics, the number of significantly co-upregulated genes in sLGG compared with iLGG based on two cohorts (TCGA and CGGA) was 75, whereas no genes were co-downregulated in this regard. Furthermore, the analysis of a significantly enriched GO biological process using the 75 upregulated genes in sLGG showed cancer progression characteristics, including cell proliferation, cell migration, angiogenesis, epithelial-to-mesenchymal transition, and negative regulation of cell death. The response to ionizing radiation and to drug therapies suggested that sLGG could be resistant to radiation and chemotherapy. Additionally, the hazard ratios of 64 of the 75 upregulated genes were significantly greater than 1 in at least two cohorts. Only the expression of HOXC8 showed an adverse effect on prognosis in three cohorts, including the CGGA, TCGA, and Rembrandt cohorts.
Given that responses to radiation and drug therapy were upregulated in sLGG, the differences between sLGG and iLGG with respect to these therapeutic strategies were investigated. Correlation analysis between gene expression and area under the curve values from the DepMap portal was conducted based on drug sensitivity data corresponding to PCV and temozolomide, which were used in glioma chemotherapy, and intersections based on the results and the 75 upregulated genes were obtained. Thus, a total of 8 genes that were upregulated in sLGG and showed positive correlation with resistance were selected. Reportedly, among these 8 genes, higher expression levels of SERPINA3 and ANXA1 have been associated with resistance to procarbazine and temozolomide, respectively.38,39 Furthermore, 9 genes that were found to be upregulated in sLGG showed negative correlation with resistance to therapy. Even though this observation indicated that these 9 genes could sensitize sLGG to chemotherapy, groups with higher expression levels of these genes showed worse prognosis than their counterparts with lower expression levels (Fig. S4A–S4I).
In a previously reported study, 31 radiosensitivity signature genes were identified based on microarray data corresponding to NCI-60 cell lines.40 Comparing the radiation resistance features corresponding to iLGG and sLGG revealed ACTN1 as the only common gene that was significantly upregulated in sLGG, considering the 31 signature genes (Fig. S6A). Additionally, it has also been reported that ACTN1 expression is downregulated in radiosensitive cell lines.40 Even though the log2-fold change of ACTN1 was slightly less than 1 (approximately 0.93), the p value showed significance. Thus, sLGG may have more radioresistance characteristics than iLGG. Furthermore, the prognosis of glioma was worse in higher-expression groups than in lower-expression groups (Fig. S6B).
We also investigated the genomic and transcriptomic differences between iLGG and sLGG samples corresponding to grades II and III glioma. Thus, we observed that the mutational characteristics of grade II glioma were different from those of grade III glioma. The general tendencies of variant type and single nucleotide variant class were the same for both grade II iLGG (G2 iLGG) and grade II sLGG (G2 sLGG); however, glioma signature mutations were more frequently observed in G2 sLGG than in G2 iLGG (Fig. S7A–S7D and Table S9). Moreover, TMB was significantly higher in G2 sLGG than in G2 iLGG, and the interactions between the top 15 somatic mutations were also higher for G2 sLGG than for G2 iLGG (Fig. S7E–S7H). In G2 iLGG, the altered oncogenic pathways included NOTCH, RTK-RAS, TP53, and Hippo pathways. However, in addition to these, in G2 sLGG the WNT, PI3K, TGF-beta, MYC, and cell cycle pathways were also altered (Fig. S7I). However, no significant transcriptomic differences were observed (Fig. S7J). In case of grade III glioma, grade III sLGG (G3 sLGG) and iLGG (G3 iLGG) showed similar variant type and single nucleotide variant class tendencies (Fig. S8A and S8B). Regarding the top 10 mutations, just two genes each (MAGEE1 and TENM1 in iLGG, and EGFR and NF1 in LGG) showed significant differences. However, these common top 10 mutations showed no significant differences between G3 sLGG and G3 iLGG with respect to their proportions (Fig. S8C and S8D and Table S9). The two groups showed similar TMB; however, the interaction between the top 15 somatic mutations was higher in G3 sLGG samples than in G3 iLGG samples (Fig. S8E–S8H). The degrees of difference in oncogenic pathway alterations were lower in grade III glioma than in grade II glioma (Fig. S8I). However, the heterogeneity of transcriptomic characteristics was significantly higher in grade III glioma (Fig. S8J). Basically, genomic mutation alterations were higher in G2 sLGG than in G2 iLGG. Conversely, G3 sLGG and G3 iLGG showed more prominent differences of transcriptomic characteristics.
There are some limitations to this study. First, the symptoms were assessed retrospectively in a noncentralized manner, so there is some potential bias, because the criteria for classifying symptoms could be inconsistent. Second, some tumors grow slowly from the start and remain, so additional analyses are required. However, the TCGA LGG data set analyzed in this study does not contain adequate data for additional analyses. Third, the results for drug and radiation resistance were obtained in cell lines. Specifically, radiation resistance was measured in NCI-60 cell lines, so these results did not represent a glioma-specific radiosensitivity signature, but common signatures. Thus, additional experimental validation of a glioma-specific radiosensitivity signature and experiments in primary glioma cell lines from patients are needed.
Conclusions
We identified and compared the molecular features of sLGG and iLGG that may be responsible for the distinct characteristics of the different histological subtypes. Identified oncogenic pathway alterations play important roles in glioma progression, malignancy, and resistance to therapy. In terms of transcriptomic landscapes, biological processes associated with glioma malignancy were upregulated in sLGG. Additionally, our results indicated that sLGG may have a worse prognosis and may be more resistant to radiation and chemotherapy than iLGG. These findings suggested that sLGG could be a more progressed form of iLGG, and immediate management of iLGG before it becomes symptomatic could lead to better patient outcomes. Therefore, the immediate management of iLGG could be more favorable for patient survival than the wait-and-see strategy.
Acknowledgments
Funding for this work was provided by the Ministry of Science, Technology, and Information, Republic of Korea (2021R1F1A105780111 to Dr. Lim) and the Technology Development Program (S3030270 to Dr. Chang) funded by the Ministry of Small and Medium-Sized Enterprises and Startups (MSS, Korea). Parts of the results are based on data generated by the TCGA Research Network: https://www.cancer.gov/tcga (accessed on July 14, 2021).
Disclosures
The authors report no conflict of interest concerning the materials or methods used in this study or the findings specified in this paper.
Author Contributions
Conception and design: Lim, Moon, Sung. Acquisition of data: Park, Ahn. Analysis and interpretation of data: Lim, Park, Ahn. Drafting the article: Lim, Park, Sim. Critically revising the article: Lim, Park, Sung. Reviewed submitted version of manuscript: all authors. Approved the final version of the manuscript on behalf of all authors: Lim. Statistical analysis: Park, Ahn. Study supervision: Lim, Sung.
Supplemental Information
Online-Only Content
Supplemental material is available with the online version of the article.
Figures S1–S8 and Tables S1–S9. https://thejns.org/doi/suppl/10.3171/2022.6.JNS22967.
References
- 1↑
Ostrom QT, 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 2014–2018. Neuro Oncol. 2021;23(12)(suppl 2):iii1–iii105.
- 2↑
Ostrom QT, Gittleman H, Xu J, et al. CBTRUS statistical report: primary brain and other central nervous system tumors diagnosed in the United States in 2009–2013. Neuro Oncol. 2016;18(suppl 5):v1–v75.
- 3↑
Mohile NA, Messersmith H, Gatson NTN, et al. Therapy for Diffuse Astrocytic and Oligodendroglial Tumors in Adults: ASCO-SNO Guideline. Oxford University Press US; 2022.
- 4↑
Weller M, van den Bent M, Preusser M, et al. EANO guidelines on the diagnosis and treatment of diffuse gliomas of adulthood. Nat Rev Clin Oncol. 2021;18(3):170–186.
- 5↑
Jo J, van den Bent MJ, Nabors B, Wen PY, Schiff D. Surveillance imaging frequency in adult patients with lower-grade (WHO Grade 2 and 3) gliomas. Neuro Oncol. 2022;24(7):1035–1047.
- 6↑
Nabors LB, Portnow J, Ahluwalia M, et al. Central nervous system cancers, version 3.2020, NCCN clinical practice guidelines in oncology. J Natl Compr Canc Netw. 2020;18(11):1537–1570.
- 7↑
Pouratian N, Schiff D. Management of low-grade glioma. Curr Neurol Neurosci Rep. 2010;10(3):224–231.
- 8↑
Khan MN, Sharma AM, Pitz M, et al. High-grade glioma management and response assessment-recent advances and current challenges. Curr Oncol. 2016;23(4):e383–e391.
- 9↑
Posti JP, Bori M, Kauko T, et al. Presenting symptoms of glioma in adults. Acta Neurol Scand. 2015;131(2):88–93.
- 10↑
IJzerman-Korevaar M, Snijders TJ, de Graeff A, Teunissen SCCM, de Vos FYF. Prevalence of symptoms in glioma patients throughout the disease trajectory: a systematic review. J Neurooncol. 2018;140(3):485–496.
- 11↑
Pallud J, Fontaine D, Duffau H, et al. Natural history of incidental World Health Organization grade II gliomas. Ann Neurol. 2010;68(5):727–733.
- 12↑
Opoku-Darko M, Lang ST, Artindale J, Cairncross JG, Sevick RJ, Kelly JJP. Surgical management of incidentally discovered diffusely infiltrating low-grade glioma. J Neurosurg. 2018;129(1):19–26.
- 13↑
Potts MB, Smith JS, Molinaro AM, Berger MS. Natural history and surgical management of incidentally discovered low-grade gliomas. J Neurosurg. 2012;116(2):365–372.
- 14↑
Zhang ZY, Chan AKY, Ng HK, et al. Surgically treated incidentally discovered low-grade gliomas are mostly IDH mutated and 1p19q co-deleted with favorable prognosis. Int J Clin Exp Pathol. 2014;7(12):8627–8636.
- 15↑
Pignatti F, van den Bent M, Curran D, et al. Prognostic factors for survival in adult patients with cerebral low-grade glioma. J Clin Oncol. 2002;20(8):2076–2084.
- 16↑
Jakola AS, Myrmel KS, Kloster R, et al. Comparison of a strategy favoring early surgical resection vs a strategy favoring watchful waiting in low-grade gliomas. JAMA. 2012;308(18):1881–1888.
- 17↑
Jakola AS, Skjulsvik AJ, Myrmel KS, et al. Surgical resection versus watchful waiting in low-grade gliomas. Ann Oncol. 2017;28(8):1942–1948.
- 18↑
Lima GL, Zanello M, Mandonnet E, Taillandier L, Pallud J, Duffau H. Incidental diffuse low-grade gliomas: from early detection to preventive neuro-oncological surgery. Neurosurg Rev. 2016;39(3):377–384.
- 19
Ng S, Herbet G, Moritz-Gasser S, Duffau H. Return to work following surgery for incidental diffuse low-grade glioma: a prospective series with 74 patients. Neurosurgery. 2020;87(4):720–729.
- 20
Ius T, Cesselli D, Isola M, et al. Incidental low-grade gliomas: single-institution management based on clinical, surgical, and molecular data. Neurosurgery. 2020;86(3):391–399.
- 21↑
Zeng L, Mei Q, Li H, Ke C, Yu J, Chen J. A survival analysis of surgically treated incidental low-grade glioma patients. Sci Rep. 2021;11(1):8522.
- 22↑
Mayakonda A, Lin DC, Assenov Y, Plass C, Koeffler HP. Maftools: efficient and comprehensive analysis of somatic variants in cancer. Genome Res. 2018;28(11):1747–1756.
- 23↑
Chalmers ZR, Connelly CF, Fabrizio D, et al. Analysis of 100,000 human cancer genomes reveals the landscape of tumor mutational burden. Genome Med. 2017;9(1):34.
- 24↑
Sanchez-Vega F, Mina M, Armenia J, et al. Oncogenic signaling pathways in The Cancer Genome Atlas. Cell. 2018;173(2):321–337.e10.
- 25↑
Love MI, Huber W, Anders S. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol. 2014;15(12):550.
- 26↑
Ge SX, Jung D, Yao R. ShinyGO: a graphical gene-set enrichment tool for animals and plants. Bioinformatics. 2020;36(8):2628–2629.
- 27↑
Faul F, Erdfelder E, Buchner A, Lang AG. Statistical power analyses using G*Power 3.1: tests for correlation and regression analyses. Behav Res Methods. 2009;41(4):1149–1160.
- 28↑
Wang QW, Wang YW, Wang ZL, et al. Clinical and molecular characterization of incidentally discovered Lower-Grade gliomas with enrichment of aerobic respiration. OncoTargets Ther. 2020;13:9533–9542.
- 29↑
Suzuki H, Aoki K, Chiba K, et al. Mutational landscape and clonal architecture in grade II and III gliomas. Nat Genet. 2015;47(5):458–468.
- 30↑
Wen PY, Reardon DA. Neuro-oncology in 2015: progress in glioma diagnosis, classification and treatment. Nat Rev Neurol. 2016;12(2):69–70.
- 32↑
Dasari K, Somarelli JA, Kumar S, Townsend JP. The somatic molecular evolution of cancer: mutation, selection, and epistasis. Prog Biophys Mol Biol. 2021;165:56–65.
- 34↑
Venkatesan S, Lamfers ML, Dirven CM, Leenstra S. Genetic biomarkers of drug response for small-molecule therapeutics targeting the RTK/Ras/PI3K, p53 or Rb pathway in glioblastoma. CNS Oncol. 2016;5(2):77–90.
- 35↑
Wang H, Xu T, Jiang Y, et al. The challenges and the promise of molecular targeted therapy in malignant gliomas. Neoplasia. 2015;17(3):239–255.
- 36↑
Ali MY, Oliva CR, Noman ASM, et al. Radioresistance in glioblastoma and the development of radiosensitizers. Cancers (Basel). 2020;12(9):2511.
- 37↑
Casati G, Giunti L, Iorio AL, Marturano A, Galli L, Sardi I. Hippo pathway in regulating drug resistance of glioblastoma. Int J Mol Sci. 2021;22(24):13431.
- 38↑
Luo D, Chen W, Tian Y, et al. Serpin peptidase inhibitor, clade A member 3 (SERPINA3), is overexpressed in glioma and associated with poor prognosis in glioma patients. OncoTargets Ther. 2017;10:2173–2181.
- 39↑
Qian Z, Fan W, Meng F, et al. Molecular characterization and clinical relevance of ANXA1 in gliomas via 1,018 Chinese cohort patients. Front Cell Dev Biol. 2021;9:777182.
- 40↑
Kim HS, Kim SC, Kim SJ, et al. Identification of a radiosensitivity signature using integrative metaanalysis of published microarray data for NCI-60 cancer cells. BMC Genomics. 2012;13(1):348.
