Introduction
Gliomas are the majority prevalent primary tumors that develop of the nervous system in the brain [
1]. The classification of gliomas by the WHO is based on histological differences, resulting in four grades, with WHO II and III being classified as lower-grade gliomas (LGG). LGG have a more favorable prognosis compared to glioblastoma (GBM) [
2]. While surgery is the recommended treatment for LGG, the invasive nature or close proximity of these tumors to critical tissues can make complete removal challenging. The current standard treatment for LGG involves surgery followed by radiotherapy, with chemotherapy serving as a promising alternative therapy [
3]. Unfortunately, the findings suggest that all LGG survivors, regardless of their treatment approach (surgical only management or no treatment), are at risk of experiencing long-term cognitive impairments in various domains [
4]. Despite the current standard treatment, which yields a median survival time of 5–10 years for LGG patients [
5]. Hence, there is a pressing need to unravel the underlying molecular mechanisms and develop a dependable molecular classification model that can effectively evaluate prognosis and guide personalized treatment strategies for individuals diagnosed with LGG.
Programmed cell death (PCD) is a crucial physiological process that plays a pivotal role in maintaining tissue homeostasis and eliminating damaged or unwanted cells. PCD can occur through various mechanisms, including apoptosis, anoikis, autophagy, alkaliptosis, cuproptosis, entosis, entotic cell death, immunogenic cell death, ferroptosis, lysosome-dependent cell death, methuosis, necroptosis, netoticcelldeath, NETosis, oxeiptosis, pyroptosis, parthanatos, and paraptosis [
6]. Imagine PCD as a housekeeping system within the body. Just like we clean our houses to maintain cleanliness and order, PCD acts as an internal cleaning mechanism that removes damaged or unnecessary cells to keep the tissues healthy. Among these PCD mechanisms, mitochondrial dysfunction has been implicated in several of them [
7]. Mitochondria play a pivotal role in supplying energy for cellular functions, regulating cellular signaling pathways, and governing PCD. Studies have shown that mitochondrial dysfunction, characterized by changes in mitochondrial structure, function, and dynamics, is associated with decreased mitochondrial respiration, altered mitochondrial morphology, and impaired mitochondrial quality control in LGG [
8‐
10]. Apoptosis is a widely recognized mechanism of PCD, which serves an essential function in preserving tissue homeostasis and eliminating damaged or unnecessary cells. Apoptosis is typified by a sequence of biochemical and morphological alterations [
11,
12]. Pyroptosis is a form of PCD that occurs following inflammasome activation and caspase-1 cleavage. Cellular enlargement, membrane rupture, and the production of pro-inflammatory cytokines are its defining features [
13]. Ferroptosis is a recently identified type of PCD defined by iron-dependent cellular demise and lipid peroxidation [
14]. Autophagy is a cellular mechanism that is essential for preserving cellular equilibrium by breaking down impaired proteins and organelles. Autophagy can serve as a mechanism for either promoting cell survival or inducing cell death, depending on the specific context in which it occurs [
15]. Necroptosis is a form of PCD that is characterized by necrosis-like cell death and is triggered by the activation of RIPK1 and RIPK3 [
6]. Cuproptosis is a form of PCD that is triggered by copper overload and is characterized by lipid peroxidation and mitochondrial dysfunction. Entotic cell death occurs exclusively in viable cells and their adjacent regions. Unlike the traditional apoptotic pathway, entotic cell death does not require the activation of apoptotic executioner pathways [
16]. Netotic cell death is an additional type of PCD that occurs due to the discharge of neutrophil extracellular traps (NETs), commonly observed in response to infections or injuries [
17]. Parthanatos is a tightly controlled type of cell death triggered by excessive activation of the nuclease PARP-1 [
18]. The process of lysosome-mediated cell death involves the action of hydrolases that enter the cytosol through membrane permeabilization [
19]. Additionally, alkaliptosis, an emerging type of programmed cell death, is controlled by the process of intracellular alkalinization [
20]. Oxeiptosis, which utilizes the reactive oxygen sensing capabilities of KEAP1, is a recently discovered cellular pathway that is likely to operate in conjunction with other cell death pathways [
21].
In the context of LGG, tumor cells can evade PCD mechanisms through various strategies, similar to concealing garbage in hidden corners of a room. They may change their shape or activate specific signaling pathways to escape elimination. The increased understanding of PCD mechanisms has led to the development of numerous drugs that target these pathways. For instance, the FDA approved a BCL-2 inhibitor that regulates cell apoptosis, which is effective in treating lymphoma [
22]. GSDME-induced pyroptosis, a distinct form of programmed cell death, has shown potential as an anti-tumor immunotherapy [
23]. Furthermore, research has demonstrated that obstructing ferroptosis can trigger cellular resistance to anti-PD-1/PD-L1 treatment [
24]. These findings demonstrate the importance of PCD research in advancing our understanding of LGG and developing new therapies to combat them.
Disrupted mitochondrial morphology, such as changes in shape, size, or cristae organization, can disrupt normal mitochondrial function and trigger PCD [
25‐
27]. Structural abnormalities may affect the release of pro-apoptotic factors from the mitochondria, leading to caspase activation and subsequent apoptosis [
27]. Mitochondrial function is also closely linked to PCD mechanisms. Dysfunctional mitochondria with impaired oxidative phosphorylation and ATP production can induce cellular stress and initiate PCD pathways [
25]. To better link mitochondrial function to PCD as described above, a series of related biomarkers were screened in this study. Nowadays, several molecular markers have been identified as clinically significant in both the diagnosis and prognosis of LGG [
28‐
31]. Markers such as IDH1 mutation and MGMT promoter methylation play a critical role in determining the optimal postsurgical treatment, including adjuvant chemotherapy and radiotherapy, and are strongly associated with the prognosis of LGG [
32]. In addition to genetic factors, various environmental and lifestyle factors can contribute to the development of LGG [
33]. The carcinogens present in tobacco smoke have the potential to inflict DNA damage on brain cells, thus facilitating tumor formation [
34]. There have been studies examining the potential correlation between exposure to electromagnetic fields (EMF), such as those emitted by power lines or electronic devices, and the risk of brain tumors, including LGG [
35].
In the process of screening and proving numerous biomarkers, sources of bias such as sample selection bias, tumor heterogeneity, analytical bias, and publication bias can impact the accuracy and generalizability of the results [
36]. Hence, the identification of survival-associated genes through transcriptome-based databases is necessary for prognostic prediction and targeted treatment selection. For decades, researchers have demonstrated that mitochondrial dysfunction and PCD mechanisms are essential for the development and spread of malignant neoplasms. In order for malignant cells to progress, they must overcome various forms of cell death and mitochondrial dysfunction. Nevertheless, there is still a lack of comprehensive understanding regarding the interplay between mitochondrial dysfunction and PCD in LGG, and there are limited detailed functional studies of these processes in LGG. To fill these knowledge gaps, we introduced a novel metric called the mitochondrial programmed cell death index (mtPCDI) to forecast the efficacy and prognosis of therapeutic interventions in LGG. Through our investigation, we discovered the heterogeneity among LGG patients and evaluated their clinical outlook. While our findings rely on the hypothesis of an interaction between mitochondrial dysfunction and PCD, this provides valuable guidance for selecting the best treatment options.
Materials and methods
Data collection
We obtained clinical details and transcriptome data of individuals suffering from LGG from four databases: the Cancer Genome Atlas (TCGA,
https://portal.gdc.cancer.gov), Chinese Glioma Genome Atlas (CGGA,
http://www.cgga.org.cn/), Gene Expression Omnibus (GEO,
http://www.ncbi.nlm.nih.gov/geo), and ArrayExpress (
https://www.ebi.ac.uk/biostudies/arrayexpress). The total analysis included 1467 samples, with 506 samples from TCGA-LGG, 172 samples from CGGA-325, 420 samples from CGGA-693, 121 samples from Rembrandt, 121 samples from GSE16011, and 142 samples from E-MTAB-3892. To improve comparability across datasets, all RNA-seq data were converted to transcripts per million (TPM) format and corrected for batch effects using the “combat” function of the “sva” package. Prior to analysis, all data were log-transformed.
In data collection for differential expression analysis, we acquired RNAseq data in TPM (Transcripts Per Million) format from two distinct sources: the TCGA and the Genotype-Tissue Expression (GTEx) project. Specifically, we retrieved 506 LGG samples (WHO grade II and III) from TCGA and 105 normal cerebral cortex samples from GTEx. For consistency and standardized processing, all the data underwent uniform processing using the Toil process [
37] from UCSC XENA (
https://xenabrowser.net/datapages/). Harmonized data processing ensures that the datasets are compatible and reduces any potential bias arising from variations in data preprocessing methods.
Identification of prognostic mitochondria-related genes and PCD-related genes
We conducted a literature search [
38] and gathered 18 patterns of PCD and key regulatory genes, which included 580 genes related to apoptosis, 367 genes related to autophagy, 7 genes related to alkaliptosis, 338 genes related to anoikis, 19 genes related to cuproptosis, 15 genes related to enteric cell death, 87 genes related to ferroptosis, 34 genes related to immunogenic cell death, 220 genes related to lysosome-dependent cell death, 101 genes related to necroptosis, 8 genes related to netotic cell death, 24 genes related to NETosis, 5 genes related to oxeiptosis, 52 genes related to pyroptosis, 9 genes related to parthanatos, and 66 genes related to paraptosis. Additionally, there were 8 Methuosis genes and 23 Entosis genes, resulting in a total of 1964 PCD-related genes. We removed 416 duplicates, resulting in 1548 PCD-related cluster genes for our analysis (Additional file
9: Table S1). From MitoCarta 3.0 [
39], we extracted 1136 mitochondria-related genes (Additional file
10: Table S2).
We employed the “limma” package to identify genes with differential expression in LGG and their associated normal tissues. To determine differential expression, we set thresholds of log2 fold change (log2FC) greater than 2 and a false discovery rate (FDR) less than 0.05. Subsequently, we utilized the “VennDiagram” package to show visual representations in differentially expressed genes (DEGs) related to mitochondrial function and programmed cell death. Additionally, we conducted Pearson correlation analysis on the RNA-seq data of TCGA-LGG samples to identify mtPCD (mitochondrial programmed cell death) co-expressed genes that exhibit a correlation coefficient (R) greater than 0.6 and a p-value (P) less than 0.001.
Development of prognostic model
We integrated ten diverse machine learning algorithms and evaluating 101 algorithmic combinations [
40,
41]. These machine learning algorithms included Support Vector Machine (SVM), Least Absolute Shrinkage and Selection Operator (Lasso), Gradient Boosting Machine (GBM), Random Forest, Elastic Net, Stepwise Cox, Ridge, CoxBoost, Super Partial Correlation (SuperPC), and Partial Least Squares with Cox regression (plsRcox). We followed a sequential approach [
40] that involved identifying prognostic variables using univariate Cox regression modeling, developing prediction models on the TCGA-LGG cohort, validating these models on five external and independent datasets (CGGA-325, CGGA-693, Rembrandt, GSE16011, and E-MTAB-3892), and calculating the Harrell Consistency Index (C-index) for model selection. We defined the model with the highest average C-index in all cohorts as the optimal model. Based on previous descriptions in the references [
40,
42], we categorized LGG patients into high- and low-mtPCDI cohorts using the median score of the respective cohort. Subsequently, we assessed the prognostic significance using Kaplan–Meier curves. Additionally, Calibration curves and Receiver Operating Characteristic (ROC) curves were generated to evaluate the mtPCDI's prognostic efficacy.
Biological function and pathway enrichment analysis
To identify genes that exhibited significant differential expression between the high and low mtPCDI categories, we applied selection criteria of FDR < 0.05 and log2 fold change (FC) > 1. In order to explore the biological functions and pathway processes associated with mtPCDI, we conducted gene ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis using “clusterProfiler” package. Briefly, we used the above differential genes as inputs, converted them to Entrez id before GO and KEGG enrichment analysis, and used a adj. p-value < 0.05 as a criterion. Gene set enrichment analyses (GSEA) enrichment were conducted for differential genes between different mtPCDI groups. In addition, Gene Set Variation Analysis (GSVA) was performed to investigate the heterogeneity of various biological processes, using the “GSVA” package [
43].
Assessment of the immune microenvironment
To assess the degree of immunological infiltrate comprehensively, we utilized various bioinformatic algorithms, including single-sample Gene Set Enrichment Analysis (ssGSEA) [
44], Tumor Immune Estimation Resource (TIMER) [
45], Cell-type Identification by Estimating Relative Subpopulations of RNA Transcripts (CIBERSORT) [
46], CIBERSORT-ABS [
47], QUANTISEQ [
48], Microenvironment Cell Populations-counter (MCP-counter) [
49], Xcell [
50], and Estimation of Proportion of Immune and Cancer cells (EPIC) [
51]. Each algorithm employed unique strategies and gene expression signatures to estimate the abundance of different immune cell subpopulations. By computing the enrichment or relative abundance of marker genes, we accurately estimated the proportions of invading immune cell types in the LGG samples. Furthermore, we employed the “Estimation of Stromal and Immune cells in Malignant Tumor tissues using Expression data” (ESTIMATE) algorithm to generate overall immunological scores. Finally, we conducted an analysis of the expression patterns of 60 immunomodulatory genes [
52], including those involved in antigen presentation, cell adhesion, co-inhibitors, co-stimulators, ligands, and receptors. Finally, we utilized the Wilcoxon signed rank summation test. This statistical test allowed us to determine the significance of differences in immune infiltration. Additionally, we generated heat maps showing the abundance of immune infiltration for each LGG sample under the distinct algorithms used, providing a visual representation of these differences.
Development of nomogram scoring system
We employed the ‘rms’ package, a widely used tool in statistical analysis and modeling, to develop a nomogram. This nomogram aimed to integrate clinical and pathological features with mtPCDI to predict patient survival. Nomograms are graphical predictive models that provide a personalized probability of an outcome based on individual characteristics. To validate the precision of the projected survival rates at 1-, 3-, and 5-year intervals, we generated calibration plots.
The mtPCDI signature’s function in forecasting drug susceptibility
We utilized the ‘pRRophetic’ and ‘oncoPredict’ packages to determine the half-maximal inhibitory concentration (IC50) of chemotherapy drugs commonly administered for LGG treatment in both the high- and low-mtPCDI groups.
Cluster analysis of mtPCDI signature genes
We performed a consensus cluster analysis on tumor samples using the expression levels of 18 mtPCDI marker genes, employing the “ConsensusClusterPlus” package. The objective of this analysis was to classify the samples into distinct subgroups. In addition, we employed t-SNE analysis and principal component analysis (PCA) to visually evaluate the clustering patterns of the samples.
Pan-cancer analysis of mtPCDI signature genes
For each type of cancer, our analysis compared tumor expression with normal tissue surrounding the tumor. The gene expression patterns of signature genes (log2FC > 1.5, FDR < 0.05) were examined specifically. We utilized clinical data from 33 different cancer tumor samples obtained from TCGA to investigate the relationship between gene expression and survival. Moreover, to assess the impact of gene expression on patient survival, we constructed a survival landscape for the signature-derived genes. We categorized the tumor specimens into low- and high-expression groups based on mRNA values and utilized the “survival” package to analyze the survival times and status within these groups. Furthermore, we examined copy number variation (CNV) data from 11,495 samples within the TCGA database. Our objective was to identify instances of significant CNV amplifications or deletions within the cohort. We considered both amplifications and deletions to enhance the detection of alterations in each gene. We defined high-frequency CNVs as those with a frequency exceeding 5%. Additionally, we collected single nucleotide variant (SNV) data from a total of 10,234 samples across 33 different types of cancers. Our aim was to understand the overall mutation frequencies across pan-cancer. To summarize the SNV data, we generated percentage heat maps. Finally, we compared the methylation patterns of each gene between tumor and normal samples using the Wilcoxon signed rank test. We identified significantly hypo- or hypermethylated genes using a probability value threshold of 0.05.
Statistical analyses
Statistical analyses were performed to assess the significance of observed differences and correlations in the study. All data were expressed as mean ± SD (standard deviation). To evaluate the impact of risk factors on survival outcomes, Cox regression models and Kaplan–Meier (K-M) survival analysis were utilized. Pearson correlation analysis was conducted to explore the relationships between variables. Statistical analysis and scientific graphing were performed using R Studio (version 4.2.3). A significance level of p < 0.05 was considered statistically significant.
Discussion
LGG is a complex and heterogeneous disease, and predicting the prognosis of LGG patients is challenging. Prognostic factors play a crucial role in predicting disease progression, selecting appropriate treatment strategies, and estimating overall survival [
55]. Among the significant prognostic factors, histological subtype and grade hold paramount importance. Higher grades are associated with more aggressive tumor behavior, shorter progression-free survival, and reduced overall survival. Molecular markers have emerged as powerful predictors of prognosis in LGG. Age at diagnosis is another crucial factor influencing prognosis, with younger patients tending to exhibit longer overall survival than older patients. Tumor extent and location carry prognostic implications, with smaller tumor size and accessible locations associated with better outcomes. Residual disease after surgical resection is another important factor influencing prognosis. Other prognostic factors include O^6-methylguanine-DNA methyltransferase promoter methylation status, genetic alterations beyond IDH and 1p/19q, and the impact of adjuvant therapy. As our understanding of LGG biology continues to expand, emerging prognostic factors and novel molecular markers are being investigated to refine prognostication and personalize treatment approaches.
Machine learning techniques are becoming more prevalent in predicting patient survival. Nevertheless, effectively implementing these methods in clinical practice while maintaining accuracy remains a challenge. However, two questions worth considering are why particular machine learning algorithms should be used and which solution is the best one. A researcher's choice of algorithm may depend heavily on their own preferences and biases [
56]. In this research, we gathered expression files from 1467 LGG patients from six multicentre cohorts worldwide and used a novel computational framework to explore the crosstalk between mitochondrial function and 18 cell death patterns. The mtPCDI was based on the expression of 18 genes (ACAA2, ACACB, ANGPTL2, ANXA5, BRCA1, BRCA2, CTSL, ECHDC2, ERCC4, GNS, IFI16, MCUB, MRPS16, MSH6, MTPAP, NCOA4, PABPC5, and PDE2A) that were most potent in predicting patient survival. Our study findings indicate that mtPCDI can serve as a valuable tool in guiding treatment decisions and improving patient outcomes. By identifying these genetic alterations through mtPCDI, clinicians can gain important insights into the underlying molecular mechanisms driving the tumor growth. This information can help guide the selection of targeted therapies or treatment strategies that specifically address the identified genetic abnormalities.
ANGPTL2 is a secreted protein that has been demonstrated to facilitate tumor growth and invasion in various cancer types, including glioma. Both glioma tissues and cells have shown significant elevations of ANGPTL2 expression levels. Nevertheless, ANGPTL2 knockdown induced a significant decline in the invasive capacity and proliferation of glioma cells. Additionally, tumorigenesis assays demonstrated that inhibiting ANGPTL2 caused a decrease in glioma tumor growth in vivo. Noteworthy, ANGPTL2 silencing led to a reduction in the protein levels of p-ERK1/2 in glioma cells, consequently impeding the ERK/MAPK signaling pathway’s activity [
57]. These findings indicate that ANGPTL2 might have a critical role in glioma’s emergence and progression, and targeting this protein could be a potential therapeutic approach for glioma treatment. The ANXA5 gene, which is situated on 4q27 in humans, encodes for Ca2 + -regulated phospholipid- and membrane-binding proteins. Increasing evidence suggests that ANXA5 may be involved in carcinogenesis by inhibiting the activity of protein kinase C in the RTK-Ras/Raf/MEK/ERK signaling pathway [
58,
59]. Moreover, ANXA5 has a broad distribution and has been observed to exhibit abnormal expression in various cancer types, such as prostate cancer, cervical carcinoma, and cholangiocarcinoma [
60,
61]. According to research, non-angiogenic gliomas are characterized by express higher levels of ANXA5 (with a 2.1-fold increase) and angiogenic glioma (with a 3.4-fold increase) as compared to normal tissues [
62]. In LGG, mutations in BRCA1 and BRCA2 have been associated with a higher risk of tumor recurrence and a poorer prognosis. CTSL is a lysosomal protease that has been implicated in the regulation of apoptosis and autophagy. In LGG, methionine deprivation to downregulate CTSL and induce proliferation inhibition in glioma cells [
63]. IFI16 has been found to inhibit apoptosis and enhance cell survival by activating the NF-κB pathway. MSH6 is a DNA mismatch repair protein that is involved in the maintenance of genome stability. LGG has been linked to a greater risk of tumor recurrence and a poorer prognosis resulting from mutations in MSH6. NCOA4, a nuclear receptor coactivator, is involved in the regulation of autophagy and iron metabolism. Moreover, NCOA4 has been discovered to induce apoptosis by activating the caspase-3 pathway. On the other hand, PABPC5, a poly(A)-binding protein, has been implicated in regulating mRNA stability and translation. It has also been found to exert an inhibitory effect on apoptosis while to promote cell survival by activating the PI3K/Akt pathway. The organelles called mitochondria play a crucial role in ATP production and the regulation of cellular metabolism. There is a suggestion that dysfunctional mitochondria play a notable role in regulating cell survival, metabolism, and proliferation in cancer. In LGG, the regulation of mitochondrial function has been linked to several genes including ACAA2, ACACB, ECHDC2, MCUB, MRPS16, MTPAP, and PDE2A. ACAA2 is a key enzyme involved in the beta oxidation of lipid acids. It works closely with its isoenzyme, ACAA1, to catalyze the beta oxidation process through a complex mechanism [
64]. Inhibiting ACACB has demonstrated a decrease in the proliferation and de novo lipogenesis of EGFRvIII human glioblastoma cells [
65]. MCUB is involved in facilitating glioma invasion/migration under hypoxic conditions [
66]. The activation of the PI3K/AKT signaling pathway by MRPS16 leads to elevated levels of Snail protein expression, thereby promoting glioma progression [
67]. The main transcript of PDE2A/miR-139 has been recognized as a crucial factor in hindering the Wnt/β-catenin pathway, which eventually results in the regulation of GSC stemness and the development of tumors [
68]. In conclusion, the regulation of programmed cell death and mitochondrial function plays a critical role in the development and progression of LGG.
Our investigation revealed the presence of shared genes impacted by somatic copy number alterations and significant disparities in somatic mutations between the mtPCDI-high and low-mtPCDI cohorts. Interestingly, the high mtPCDI group demonstrated significantly higher TMB, which has emerged as a novel prognostic biomarker closely associated with the response to immunotherapy. We established a moderately positive correlation between TMB and mtPCDI and discovered a potential link between mutational burden and immunotherapy response, providing a new perspective on checkpoint blockade treatment. Moreover, our analysis revealed that the high-mtPCDI group demonstrated increased immune cell infiltration and higher expression levels of immunomodulators, including classical immune checkpoint molecules. However, patients classified in the high-mtPCDI group exhibited notably lower levels of StromalScore, ImmuneScore, and ESTIMATEScore compared to those in the low-mtPCDI group. This suggests a complex interplay between the mtPCDI and the tumor immune microenvironment. Overall, our findings suggest that mtPCDI may serve as a valuable biomarker for predicting the genomic pattern and response to immunotherapy in LGG patients.
As a result of our enrichment analysis, we found that mtPCDI-related genes were mainly enriched in various cellular processes, environmental information processing, and organismal systems. Of particular interest, our analysis uncovered a significant correlation between the high mtPCDI group and DNA replication, cell cycle, and other proliferation-related biological processes. These findings provide partial insight into the more unfavorable prognosis observed in this group.
One notable limitation of this study is the lack of in vitro or in vivo experiments to directly validate our findings. While we have extensively utilized bioinformatics analyses and computational methodologies, experimental validation remains an essential aspect of scientific research. In experiments could provide valuable insights into the functional implications of the observed patterns and strengthen the credibility of our results. Therefore, future research should focus on conducting targeted experiments that can corroborate and extend the observations from mtPCDI.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.