Introduction
Cerebrospinal fluid (CSF) has various functions, such as protecting the brain, transporting biological substances, and excreting toxic and waste substances. CSF is in direct contact with the extracellular fluid of the brain. Although the CSF composition reflects that of the blood plasma, active transport from the blood and secretions from the brain contribute to the CSF composition. Therefore, the CSF composition can indicate biological brain processes, and CSF analysis is indispensable for diagnosing and understanding central nervous system (CNS) disorders [
1‐
4]. A chemical examination of CSF is an important tool for the diagnosis of some types of brain tumors [
5,
6].
Gliomas, the most common primary CNS tumors, are classified as grades I to IV based on the histopathological and clinical criteria established by the WHO [
7]. Grade I gliomas, which are considered benign, are generally curable with complete surgical resection [
8]. In contrast, Grades II and III gliomas are invasive, progress to higher-grade lesions, and have poor clinical outcomes. Grade IV gliomas are glioblastomas (GBMs), which are the most invasive and have a dismal prognosis [
9,
10]. Metabolic remodeling is a predominant phenotype of malignant tumor cells and refers to the alteration of the utilization and/or synthesis of important metabolites, including glucose, fatty acids, and amino acids, by tumor cells [
11]. The levels of several metabolites, such as lactic acid and choline, are elevated in malignant gliomas [
12]. Recently, mutations in the
isocitrate dehydrogenase (
IDH)
genes have been identified in gliomas [
13,
14]. Both IDH1 and IDH2 are NADP
+-dependent dehydrogenases and convert isocitrate into α-ketoglutarate. Previous reports have indicated that these mutations are frequently observed in astrocytic and oligodendroglial tumors of grades II and III [
14‐
17]. Because these enzymes catalyze reactions of energy metabolism,
IDH mutations may alter global cellular metabolism [
18].
Metabolomics has recently undergone rapid development. Metabolomics includes the analysis of metabolites from biofluids or tissues using nuclear magnetic resonance (NMR) or mass spectrometry (MS)-based approaches, including liquid chromatography/mass spectrometry (LC/MS) or gas chromatography/mass spectrometry (GC/MS). To date, the global metabolic profiling of human biofluids, such as urine and sera, has been used to visualize the distinct metabolic profiles of patients with cancer and gastroenterological disease [
19]. Furthermore, the metabolic profiling of tissue specimens from several cancer patients has revealed significant variations in the metabolites detected in tumors versus normal tissue [
20]. There have also been a small number of reports on the metabolomic analysis of CSF in CNS disorders [
21‐
26]. However, few metabolomic studies using MS-based methods have been performed on the CSF of glioma patients. In this study, we conducted a GC/MS-based metabolomic analysis of CSF samples from 32 glioma patients. We examined the differences in the metabolites of the CSF samples using various clinical parameters, such as WHO grades and
IDH mutation. Our study indicates that a metabolomic analysis of CSF from glioma patients may be useful for predicting the malignancy grade and
IDH mutation status.
Discussion
Metabolic remodeling is a predominant phenotype of malignant tumor cells (Warburg effect) and refers to the alteration of the utilization and/or synthesis of important metabolites, including glucose, fatty acids, and amino acids, by tumor cells [
31]. Aerobic glycolysis involves the generation of substrates such as fatty acids and nucleotides that are required for rapidly proliferating cells and is associated with a survival advantage. The 13C-nuclear MR spectroscopy measurements have demonstrated that glioblastoma cells convert as much as 90 % of the glucose that they acquire into lactic acid in vitro [
32]. Clinically, it is known that the levels of several metabolites, such as choline and lactic acid, are elevated in malignant glioma tissue compared with contralateral normal brain tissue. Additionally, lactic acid levels are higher in grades III and IV gliomas than in low-grade gliomas, and lipids are significantly elevated in grade IV gliomas [
33]. Additionally, Colavolpe reported that malignant glioma cells absorb a large quantity of F18-2-fluoro-2-deoxy-
d-glucose (FDG) and that pre-treatment with FDG-PET provides significant additional prognostic information for high-grade gliomas [
34]. Thus, glioma cells have different metabolic patterns in terms of utilizing important metabolites such as glucose and fatty acids when compared with normal glial cells or among different WHO grades.
In the present study, we identified 61 metabolites in the CSF from glioma patients using GC/MS. Previous reports detected ~40–90 metabolites using GC/MS in normal human CSF [
35‐
37], and the number of metabolites identified in this study was similar to that in the previous reports. The lactic acid level in the CSF was significantly elevated in the GBMs compared with the grades I–II gliomas. Lactic acid is often more prominent in the highest grade of glioma in MRS studies [
38]. Furthermore, Yamasaki et al. [
12] reported that lactic acid expression in glioma statistically correlated with a shorter OS among the MRS parameters. Consistent with their results, the higher CSF level of lactic acid tended to be associated with a shorter OS in all glioma patients and was statistically associated with a shorter OS in malignant glioma alone in our metabolomic study. Interestingly, the present study demonstrated that the CSF levels of lactic acid were significantly increased in gliomas with a mutant
IDH compared with wild-type
IDH in only low-grade gliomas (grades I–III). We sought to determine whether the higher lactic acid levels were associated with a poor outcome in grade I–III gliomas. However, this could not be determined because most of the patients were still alive at the end of the study. Although gliomas with a mutant
IDH are known to be associated with improved survival, further studies are required to determine the association between the CSF lactic acid levels and the prognosis in low-grade gliomas. Lactic acid is produced by lactate dehydrogenase (LDH) and is usually an anaerobic metabolic product that occurs when the oxygen demand of a rapidly growing tumor exceeds what its neovasculature supplies. In addition, unlike most normal tissues, malignant tumor cells convert most glucose into lactic acid, regardless of whether oxygen is available to support mitochondrial oxidative phosphorylation [
11]. Active glycolysis increases the cytosolic NADH/NAD
+ ratio and thereby accelerates lactate dehydrogenase activity [
39]. In the present study, LDHA was predominantly expressed in tumor cells rather than in vascular tissues. In particular, the tumor cells near the necrotic tissues markedly expressed LDHA (Supplementary Fig. 3i). From these findings, tumor cells, particularly hypoxic tumor cells, seem to predominantly produce and release lactic acid. Additionally, LDHA expression is stronger in GBMs than in other grades of gliomas. This is consistent with the CSF lactic acid level. Thus, the CSF lactic acid level appears to reflect the metabolic condition of the glioma tissues.
The CSF levels of citric and isocitric acid, which are TCA cycle metabolites, were significantly elevated in GBMs compared with grades I–II or III gliomas. Citric acid accumulates in tissue in which the glycolytic rate exceeds the TCA cycle activity [
40]. Furthermore, it is well known that citric acid is required to produce cytoplasmic acetyl-CoA for lipid synthesis, and this step is essential to support cell growth [
31,
41]. In an MRS study of gliomas, citric acid levels were reported to be high in pediatric pontine gliomas [
42]. In addition, citric acid levels were higher in aggressive pediatric astrocytomas than in indolent astrocytomas [
43]. Based on this evidence, it appears that high-grade gliomas have higher levels of citric acid than low-grade gliomas. In the present study, the higher CSF level of citric acid tended to be associated with a shorter OS in all glioma patients. Thus, measuring the amount of CSF citric acid may be useful for predicting the prognosis of gliomas.
IDH mutations were frequently observed in astrocytic and oligodendroglial tumors of grades II and III [
14]. In the present study, the differences in the metabolic profiles by the
IDH status were analyzed in grades I–III gliomas, excluding GBMs, because all GBMs did not have a
IDH mutation and because several metabolites were significantly different from other grades of gliomas. Reitman et al. [
44] reported that amino acid, choline lipid, and TCA cycle metabolite levels were altered in cells expressing IDH mutants. They described that the late TCA intermediates fumarate and malate were reduced in IDH mutant-expressing cells. Additionally, Lazovic et al. [
45] recently reported that the ratio of lactate/choline was significantly increased in IDH1R132H-transfected U87 cells compared with control cells in an MRS study. These results are consistent with our results. Metabolic changes by
IDH mutations are reported to be different among cell types [
46,
47]; however, further studies are necessary. We discovered that CSF concentrations of pyruvate + oxaloacetic acid were significantly lower in the gliomas with
IDH mutations compared with those with a wild-type
IDH. In contrast, the CSF concentrations of lactic, citric, and isocitric acid were significantly higher in gliomas with
IDH mutations compared with those of a wild-type
IDH. In addition, the levels of the late TCA cycle metabolites that act downstream of isocitric acid decreased in the gliomas with
IDH mutations. IDH1 converts isocitrate into α-ketoglutarate, which may explain the observed lower levels of late TCA metabolites (which are downstream from α-ketoglutarate) and increased isocitrate, pyruvate, and/or citrate levels (which are upstream from isocitrate). From these data, CSF metabolite levels may reflect the metabolic changes caused by the
IDH mutations in the glioma cells. Recently, biochemical studies revealed that mutant IDH1 protein gains the function to catalyze the reduction of α-ketoglutarate to 2-hydroxyglutarate (2-HG) in a NADPH-consuming manner [
48]. Sahm et al. [
49] reported the success of the detection of 2-HG in glioma tissue using GC/MS. However, our study, which used the GCMS-QP2010 Ultra and Plus, did not detect 2-HG in the CSF of the glioma patients with the
IDH mutation. Further studies are required to detect 2-HG in the CSF of glioma patients. Most recently, Locasale et al. [
50] analyzed the CSF metabolic profiles from the 10 patients with malignant gliomas using LC–MS/MS and identified 124 polar metabolites. They also identified significant differences in the CSF metabolite composition between patients with malignant gliomas and controls. Thus, they described that the CSF metabolite composition may provide clinically relevant biomarkers and insights into the mechanisms underlying the pathogenesis of malignant gliomas.
The present study has several limitations. First, there are several histological types in the same WHO grade. The low-grade gliomas in this series contained four different histological types (i.e., diffuse astrocytoma, oligodendroglioma, ependymoma and pilocytic astrocytoma), and the grade III gliomas contained three different histological diagnoses (i.e., anaplastic astrocytoma, anaplastic oligodendroglioma, and anaplastic ependymoma). There may be significant differences in the metabolic profile among different histological types of gliomas. Therefore, the differences in the metabolite profiles of different histological types could have contributed to the differences observed among the different grades. Second, the relatively small sample size limited our power to detect potentially important associations between CSF metabolites and the clinical outcomes of interest. Despite these and other limitations, this is the first comparative report of the CSF metabolite profile among the various grades of gliomas. However, further studies will be needed to determine the CSF metabolome of various histological types of gliomas.
In conclusion, the CSF levels of several metabolites, such as citric, isocitric, and lactic acid, were altered between glioma grades and the IDH mutation status, which may reflect the glioma cell metabolism. Our study indicates that an analysis of the CSF metabolite levels may be useful for predicting glioma malignancy and prognosis.