Abstract
Background
Mutations in the isocitrate dehydrogenase 1 (IDH1) gene that are frequently observed in low-grade glioma are strongly associated with the accumulation of 2-hydroxyglutarate (2HG), which is a valuable diagnostic and prognostic biomarker of IDH1 mutant glioma. However, conventional MR spectroscopy (MRS)–based noninvasive detection of 2HG is challenging. In this study, we aimed to determine the additional value of other metabolites in predicting IDH1 mutations with conventional MRS.
Methods
Forty-seven patients with glioma underwent conventional single voxel short echo time MRS prior to surgery. A stereotactic navigation-guided operation was performed to resect tumor tissues in the center of the MRS voxel. MRS-based measurements of metabolites were validated with gas chromatography–mass spectrometry. We also conducted integrated analyses of glioma cell lines and clinical samples to examine the other metabolite levels and molecular findings in IDH1 mutant gliomas.
Results
A metabolomic analysis demonstrated higher levels of 2HG in IDH1 mutant glioma cells and surgical tissues. Interestingly, glutamate levels were significantly decreased in IDH1 mutant gliomas. Through an analysis of metabolic enzyme genes in glutamine pathways, it was shown that the expressions of branched-chain amino acid transaminase 1 were reduced and glutamate dehydrogenase levels were elevated in IDH1 mutant gliomas. Conventional MRS detection of glutamate and 2HG resulted in a high diagnostic accuracy (sensitivity 72%, specificity 96%) for IDH1 mutant glioma.
Conclusions
IDH1 mutations alter glutamate metabolism. Combining glutamate levels optimizes the 2HG-based monitoring of IDH1 mutations via MRS and represents a reliable clinical application for diagnosing IDH1 mutant gliomas.
Keywords: 2-hydroxyglutarate, glioma, glutamate, IDH1 mutation, MR spectroscopy
Heterozygous somatic mutations in the isocitrate dehydrogenase (IDH) gene have been detected in many malignant tumors, including glioma, myelodysplastic syndrome, and acute myeloid leukemia.1,2 The frequent IDH mutations in multiple cancers suggest that IDH mutations are closely associated with tumorigenesis and that they have a marked effect on the molecular and genetic pathways of oncogenic progression. In particular, the finding of a 2- to 4-fold longer median survival in patients with IDH1 mutant glioma compared with those with IDH1 wild-type glioma has generated clinical interest in IDH mutations as diagnostic and prognostic indicators.3,4
IDH genes have 3 isoforms (IDH1, 2, and 3) in which the genes encoding IDH1, and to a lesser degree those encoding IDH2, are mutated in 50%–80% of World Health Organization (WHO) grades II and III gliomas and secondary glioblastomas (glioblastoma multiforme [GBM]). These mutations are found at a single amino acid residue of IDH1 in which arginine 132 (R132) is commonly replaced with histidine (R132H).3,5 The primary biochemical alteration of mutant IDH1 is the gain of activity in converting alpha-ketoglutarate (α-KG) to 2-hydroxyglutarate (2HG), which inhibits the normal activity of IDH1 to reduce NADP+ to NADPH.5,6 Mutant IDH1 leads to a very high accumulation of 2HG in the range of 10- to 100-fold more than that in IDH wild-type tumors or healthy tissue,5 suggesting that 2HG analysis is an alternative indirect method for identifying IDH status.
Magnetic resonance spectroscopy (MRS) is unique in that it predicts IDH mutations noninvasively by evaluating endogenously produced 2HG. Seminal studies have reported the utility of the noninvasive detection of 2HG using in vivo 3 T7 and 7 T MRS.8 In a conventional 3 T MRS study, this application was widely available for clinical use on all scanners without spectral editing; however, there was a high incidence of false positive results (18.5%–26%).9,10 Recently, to detect 2HG more reliably, customized MRS techniques employing spectral editing, such as 2-dimensional correlation spectroscopy and J-difference spectroscopy, have been reported.11,12 Additionally, an asymmetric long echo time (TE) MRS (TE = 97 ms) has been shown to better quantify 2HG and reduce false positive results.13 Although these methods are effective in minimizing macromolecule signals and improving 2HG detection, they require a high concentration of 2HG (>1.5–2 mM) for reliable 2HG measurement, and the sequencing is not readily available on mainstream clinical scanners. Therefore, further investigations into the conventional MRS-based detection of 2HG and improvements in the identification of IDH1 mutations in standard MRS-based measurements of other metabolites are necessary, and the findings would have considerable utility in clinical institutions.
Materials and Methods
Study Design and Study Population
This study was approved by the ethics review boards of our institutions (approval numbers: #1497 for MRS; #782 for gas chromatography–mass spectrometry [GC-MS] studies of glioma patients; #1579 for use of glioma samples). Informed consent was obtained from each patient. Detailed protocols are found in the Supplementary materials.
Subjects were 57 consecutive patients with an intracranial tumor and a suspected diagnosis of glioma who were treated at the Department of Neurosurgery, Kobe University Hospital, between August 2013 and August 2015 (Supplementary Fig. S1). All patients underwent preoperative MRI and MRS within 1 week prior to surgery. Of the 57 patients enrolled, 10 were excluded before analysis of the correlation between MRS results and IDH1 status. Three patients had a low signal-to-noise ratio of <5 (IDH1 wild-type n = 1, IDH1 mutant n = 2), 1 patient had a high full width at half maximum of more than 0.065 ppm (IDH1 wild-type n = 1), and 6 patients had a necrotic lesion with insufficient volume of the solid component (IDH1 wild-type n = 6). Surgical specimens resected during surgery at the exact targets indicated by MRS using an intraoperative navigation system (Brainlab) from 34 patients were analyzed using targeted GC-MS to measure the levels of 2HG and α-KG. Surgical samples from 23 patients were analyzed using real-time PCR. Comprehensive metabolomic analysis of surgical samples from 20 patients using GC-MS and immunohistochemical staining of glutamate dehydrogenase (GDH) was performed.
To validate the cutoff values obtained from in vivo MRS analysis which discriminate between IDH1 wild-type and mutant glioma, we prospectively collected another 10 intra-axial tumors from September 2015 to January 2016.
Magnetic Resonance Imaging and Magnetic Resonance Spectroscopy
The MRS signal was acquired with a 3 T MRI/1H-MRS scanner (Achieva, Philips Medical Systems). An 8-channel head MRI coil was employed for signal reception and the quadrature body MRI coil was used for transmission of the radiofrequency pulses. Single-voxel localized MR spectra were acquired using the double-echo point-resolved spectroscopic sequence (PRESS) with chemical-shift selective water suppression. The MRS acquisition parameters were as follows: volume of interest, 1.5 × 1.5 × 1.5 cm3; repetition time/TE = 2000/35 ms; number of acquisitions, 128 averages and 1024 complex points for the spectral data. Volumes of interest were localized to representative areas of the solid tumor. Regions of necrosis, hemorrhage, calcification, or peripheral edema were excluded from the corresponding region.
MRS data were quantified with LCModel version 6.3 (Stephen Provencher) using a simulated basis set (Supplementary Table S1) (GAMMA, Radiology, Duke University Medical Center).14 The absolute metabolite concentrations (mM) were estimated using an unsuppressed water signal as a reference. All detected 2HG concentrations and Cramér-Rao lower bound (CRLB) were compared with the absolute 2HG concentration in ex vivo GC-MS. Glutamate (Glu), glutamine (Gln), and glutathione (GSH) were excluded when CRLB was >30%; total choline (tCho, phosphocholine + glycerophosphocholine [PCh+GPC]), total NAA (tNAA, N-acetyl-asparate + N-acetyl-aspartylglutamate), lactate (Lac), total creatine (tCr, creatine + phosphocreatine [Cr+PCr]), and myo-inositol (mIns) were excluded when CRLB was >20%.
Histological Diagnosis and IDH Gene Status of Clinical Specimens
IDH1 R132H analysis was confirmed by immunohistochemistry and DNA sequencing as previously described.4 Paraffin sections of the intracranial tumor specimens were stained with IDH1 R132H mutation-specific antibodies (1:50; H09 clone, Dianova). The IDH1 forward primer (5′-ACC AAA TGG CAC CAT ACG A-3′) and reverse primer (5′-GCA AAATCA CAT TAT TGC CAA C-3′) were designed to amplify exon 4 (codon R132) of the IDH1 gene.
MGMT Promoter Methylation
Gene promoter methylation status of O6-methylguanine-DNA methyltransferase (MGMT) was determined using a quantitative methylation-specific PCR assay as described.15
Gas Chromatography–Mass Spectrometry Analysis
Metabolite extraction and derivatization from biological samples were performed as previously described.16 The GC-MS analysis was based on a published method with some modifications.17,18 Derivatized samples were analyzed on a GCMS-QP2010 Ultra (Shimadzu). The peak height intensity data of the m/z 247 (2HG), 198 (α-KG), and 275 (2-isopropylmalic acid [2-IMA]) ions were collected via selected ion monitoring. For comprehensive metabolomic analysis, 20 scans/s were recorded over the mass range of m/z 85–500 using the Advanced Scanning Speed Protocol (Shimadzu). All data were normalized to the peak height of 2-IMA.
Cell Culture
U87 malignant glioma (U87 MG) cell lines were obtained from the American Type Culture Collection and cultured in Dulbecco's modified Eagle's medium (Nacalai Tesque) supplemented with 10% fetal bovine serum (Biological Industries) and 100 U/mL penicillin and streptomycin (Nacalai Tesque) in a humidified 5% CO2 incubator at 37°C.
Plasmid Transfection
Transfection was performed with FuGENE HD (Promega) according to the manufacturer's instructions. Constructs were used of the wild-type human IDH1 (pcDNA DEST53/IDH1, which contains the green fluorescent protein [GFP] tag) and R132H mutant (pcDNA DEST53/IDH1, which contains the GFP tag).
RNA Extraction and Real-Time PCR Analysis
RNA extraction and real-time PCR analysis were performed as previously described.18 The quantitative mRNA expression data were analyzed using the ΔΔCt method. The following TaqMan Gene Expression Assays were used: SLC1A2 (EAAT2, GLT-1) (SMID: Hs01102423_m1), SLC1A3 (EAAT1, GLAST) (SMID: Hs00188193_m1), SLC7A11 (xCT) (SMID: Hs00921938_m1), BCAT1 (SMID: Hs00398962_m1), GDH 1 (SMID: Hs03989560_s1), GDH2 (SMID: Hs01649931_s1), GLS (SMID: Hs00248163_m1), GLUL (GS) (SMID: Hs00365928_g1), and 18S (SMID: Hs99999901_s1).
Western blot analysis
Western blot analysis was performed according to the method described in the previous reports.18
Antibodies and Reagents
The following antibodies were used: IDH1 R132H (H09 clone, Dianova), anti-IDH1 (Sigma-Aldrich), GDH1/2 (Cell Signaling), BCAT1 (Cell Signaling), and β-actin (Ambion). Reagents used were: GDH inhibitor (epigallocatechin gallate [EGCG]; Sigma-Aldrich), glutaminase (GLS) inhibitor (compound 968, Calbiochem), and dimethyl–α-KG (dm-αKG) (Sigma-Aldrich).
Assessment of Staining
Immunohistochemical staining of GDH was independently evaluated and scored using Image J 1.49 software (NIH) as previously described.19 Investigators performing the assessment were blinded to clinical information.
Statistical Analysis
The Mann–Whitney U-test and ANOVA were used to examine statistical differences between ex vivo tumor samples. For U87 glioma cells, significant differences between 2 groups were determined using an unpaired t-test; between 3 groups, significant differences were determined using the Tukey–Kramer honestly significant test. The results are shown as means ± SEM. In the box and whisker plots, the box spans represent the 25th and 75th percentiles with medians, and the whiskers represent the 10th and 90th percentiles. Statistical significance was indicated as P < .05, P < .01, and P < .001. All statistical analyses were performed using the JMP 11 package (SAS Institute; www.jmp.com). A principal component analysis and variable importance in the projection (VIP) scoring were performed via the MetaboAnalyst 3.0 web portal (www.metaboanalyst.ca). VIP scores are commonly used to estimate the importance of each metabolite.20 To investigate the diagnostic performance of MRS, the area under the curve (AUC) of the receiver operating characteristic (ROC) curve was used. The net reclassification improvement (NRI) and integrated discrimination improvement (IDI)21 were calculated using R (version i386.3.2.1 for Windows). The prediction model was assessed in terms of its prediction accuracy (the proportion of correctly classified objects) using the confusion matrix function of the “caret” package of R (R3.1.1; R Foundation for Statistical Computing; http://www.r-project.org).
Results
Quantitative 2HG Levels Using In vivo MRS and Ex vivo GC-MS
Of the 47 patients, 18 had IDH1 mutant tumors and 29 had IDH1 wild-type tumors. All IDH1 mutant gliomas have mutations in the hotspot codon R132. Thirteen patients (28%) had WHO grade II gliomas, 10 (21%) had WHO grade III gliomas, and 24 (51%) had WHO grade IV gliomas. Median patient age was not significantly different between IDH1 wild-type gliomas (17–85 y; median, 55 y) and mutant gliomas (32–80 y; median, 47 y). Forty of the 47 patients analyzed (85%) had new diagnoses. The patient characteristics are summarized in Supplementary Table S2.
Stereotactic navigation-guided sampling was performed in the center of the MRS voxel. Thirty-four glioma samples obtained from the 47 patients could be analyzed by GC-MS. 2-Hydroxyglutarate in each sample was quantified by GC-MS and compared with the 2HG level of the same target detected using in vivo MRS. Typically, tumors with IDH1 mutations had an extra 2HG peak at ∼2.25 ppm on MRS (Fig. 1A). In MRS, the IDH1 mutant gliomas showed a significantly higher accumulation of 2HG (P < .001; Fig. 1B). Quantification using GC-MS of the total 34 glioma samples (IDH1 wild: 23, IDH1 mutant: 11) showed significantly higher 2HG accumulation (up to 100-fold) in the IDH1 mutant tumors compared with the IDH1 wild-type gliomas (P < .001; Fig. 1C). Alpha-ketoglutarate did not significantly differ between the IDH1 wild-type and mutant gliomas (Supplementary Fig. S2). The 2HG concentration comparison between in vivo MRS and ex vivo GC-MS in the same region demonstrated that 2HG was overestimated by MRS in 6 of 23 (26%) IDH1 wild-type gliomas (Fig. 1D). In 3 of 11 (27%) IDH1 mutant gliomas, 2HG could not be detected.
Fig. 1.
Comparative detections of 2HG to identify the IDH1 mutation using in vivo MRS and ex vivo GC-MS. (A) Designating tissue targets from in vivo MRS analyzed using an LCModel and ex vivo GC-MS. Stereotactic navigation (Brainlab)-guided sampling was performed at the exact targets indicated by MRS. Spectra were obtained with PRESS at short TE (35 ms). IDH status was assessed via immunohistochemical staining and direct sequencing. (B) The 2HG concentration of MRS in 47 patients with cerebral glioma (IDH1 wild-type, n = 29; IDH1 mutant, n = 18). The line shows the median. P-values were calculated using the Mann–Whitney nonparametric U-test (***P < .001). (C) 2-Hydroxyglutarate concentration of GC-MS in 34 glioma samples (IDH1 wild-type, n = 23; IDH1 mutant, n = 11). The line shows the median. P-values were calculated using the Mann–Whitney nonparametric U-test (***P < .001). (D) Comparison of 2HG concentrations between in vivo MRS and ex vivo GC-MS.
Metabolic Difference Between IDH1 Wild-type and Mutant Glioma Cells
To identify other predictive biomarkers of IDH1 mutation, differences in cellular metabolism between U87 glioma cells with wild-type IDH1 and mutations were assessed as reported in the literature.22 The wild-type or mutant IDH1 gene was successfully transfected into U87 glioma cells, as confirmed via western blot (Fig. 2A). Next, to determine the metabolic profiling of IDH1 wild-type and mutant U87 glioma cells, 65 extracted metabolites were measured in a GC-MS analysis (Supplementary Table S3). Hierarchical clustering of metabolic changes between IDH1 wild-type and mutant glioma cells indicated a clear difference between the U87 IDH1 wild-type (n = 3) and mutant (n = 3) cells (Fig. 2B). In the IDH1 mutant glioma cells, 2HG was significantly elevated (P < .001) and glutamate was significantly reduced (P < .05). Glutamine and α-KG levels did not significantly differ between IDH1 wild-type and mutant glioma cells (Fig. 2C). The key differentiating metabolites were glutamate, 2HG, and pyroglutamic acid (VIP scores >2; Fig. 2D).
Fig. 2.
Alterations in metabolic profiles observed in U87 glioma cells expressing a mutant IDH1 (R132H). (A) Western blot analysis using IDH1 R132H antibodies and IDH1 antibodies in control, IDH1 wild-type, and IDH1 mutant (R132H)–transfected U87 glioma cells. (B) Heat map representation of 2-dimensional hierarchical clustering of metabolites identified as differentially expressed in IDH1 wild-type and IDH1 mutant U87 glioma cells. Each column represents an IDH status group, and each row represents a metabolite. Distance was measured using Minkowski measures; clustering was performed using Ward's minimum variance method. (C) GC-MS quantification of glutamine, glutamate, α-KG, and 2HG normalized to cellular protein. Data represent the means ± SEM of 3 independent experiments. P-values were calculated using 2-tailed Student's t-test (*P < .05, ***P < .001). (D) Metabolic profiling for IDH1 wild-type and IDH1 mutant U87 glioma cells assessed via ex vivo GC-MS normalized to 2-IMA as an internal control. Important features were identified using partial least squares discriminant analysis. Colored boxes on the right indicate the relative concentrations of the corresponding metabolite in this study using the level of significantly different metabolites determined by VIP score.
GDH and BCAT1 Expressions Regulate Glutamate Flux in Clinical Glioma Samples
To determine whether glutamate metabolism is affected by IDH1 mutation, further biochemical analyses of surgical samples were performed. GC-MS of surgical samples showed that glutamate and malate levels were decreased in IDH1 mutant gliomas (n = 10) compared with IDH wild-type gliomas (n = 10) (P < .05; Fig. 3A), whereas no differences between glutamine and tricarboxylic acid cycle metabolites were observed (with the exception of malate) (Supplementary Fig. S3). To determine how the IDH1 mutation influenced the glutamate metabolism pathway, mRNA expressions of 8 key enzymes were examined in 23 resected tumor samples (Fig. 3B). These included GDH1 and GDH2, GLS, glutamine synthase (GS), branched chain amino acid transaminase 1 (BCAT1), solute carrier family 1 member 2 (SLC1A2, excitatory amino acid transporter 2 [EAAT2]), solute carrier family 1 member 3 (SLC1A3, excitatory amino acid transporter 1 [EAAT1]), and solute carrier family 7 member 11 (SLC7A11, x-C-type transporter [xCT]). A gene expression analysis showed that expression of GDH1 and GDH2 mRNA was significantly higher in IDH1 mutant gliomas than in IDH1 wild-type gliomas (P < .001 and P < .01, respectively) (Fig. 3C). In contrast, IDH1 mutant gliomas expressed a lower mRNA level of BCAT1 than IDH1 wild-type gliomas (P < .001). The mRNA expression levels of other enzymes did not show any differences. Western blot analysis also demonstrated that the GDH1/2 level was higher in IDH1 mutant tumors. Furthermore, BCAT1 expression was essentially absent in IDH mutant tumors (Fig. 3D). In the immunohistochemical staining analysis, the IDH1 mutation led to increases in GDH-positive cells, which is consistent with the mRNA and protein level results (P < .001; Fig. 3E).
Fig. 3.
Differences in mRNA expressions, enzymes, and metabolites between IDH1 wild-type and mutant gliomas. (A) Relative metabolites of glutamate (Glu) and glutamine (Gln) assessed by ex vivo GC-MS normalized to 2-IMA as an internal control. Comparisons of relative metabolites in 20 gliomas (IDH1 wild-type, n = 10; IDH1 mutant, n = 10). The line indicates the median. P-values were calculated using the Mann–Whitney nonparametric U-test (*P < .05). (B) Schematic showing the enzymes, transporters, and metabolites related to 2HG and glutamate. TCA, tricarboxylic acid. (C) Expression of mRNA in 23 gliomas (IDH1 wild-type, n = 9; IDH1 mutant, n = 14) normalized to expression in normal brain. The line indicates the median. P-values were calculated using the Mann–Whitney nonparametric U-test (**P < .01, ***P < .001). (D) Western blot analysis showing IDH R132H, IDH1, GDH1/2, and BCAT1 protein expression in gliomas with wild-type IDH1 (lanes 1–5) and mutant IDH1 (lanes 6–9). (E) Immunohistochemical images of GDH1/2 obtained from 20 glioma samples (IDH1 wild-type, n = 10; IDH1 mutant, n = 10). Tissue was counterstained with hematoxylin. Original magnification: ×400. Scale bars, 50 mm. The line indicates the median. P-values were calculated using the Mann–Whitney nonparametric U-test (***P < .001).
To determine whether glutamate metabolism influenced IDH1 mutation-mediated 2HG upregulation, α-KG and 2HG levels were measured in IDH1 wild-type and mutant U87 glioma cells treated with EGCG, a GDH inhibitor.23 EGCG suppressed α-KG and 2HG levels in both IDH1 mutant and wild-type U87 cells. Of note, dm-αKG rescued the reduction in 2HG level in IDH1 mutant U87 glioma cells but not in IDH1 wild-type U87 glioma cells (Supplementary Fig. S4A). These results were confirmed by administering compound 968, a GLS inhibitor that blocks GLS activity in cancer cells24 (Supplementary Fig. S4B). Taken together, these findings demonstrated that glutamate may be consumed to compensate for the altered flux of α-KG to 2HG in IDH1 mutant gliomas.
Clinical Application to Identify the IDH1 Mutation Using MRS of Glutamate and 2HG
Due to the marked reduction in the glutamate level in IDH mutant gliomas in vitro, we reevaluated the clinical MRS data. Mean signal-to-noise ratio for spectra was 13.7 (range, 6–29). The level of 2HG was elevated in IDH1 mutant gliomas, as shown in Fig. 1B (P < .001). Glutamate, glutamine, and glutathione levels were lower in IDH1 mutant gliomas than in IDH1 wild-type gliomas (P < .001, P < .01, P < .001, respectively) (Fig. 4A). Tumor grade had a significant influence on glutamate concentrations (P < .01), but not on 2HG concentrations. Concentrations of 2HG and glutamate were not significantly different in histology or MGMT methylation status (Supplementary Fig. S5). Next, we investigated whether glutamate could be used to detect the IDH1 mutation in MRS using the ROC curve method. ROC curve analysis demonstrated that 2HG concentrations >1.8 mM were 56% sensitive and 96% specific for IDH mutant glioma in all 47 patients (AUC, 0.79). Glutamate concentrations <3.9 mM were 78% sensitive and 90% specific for IDH1 mutant glioma (AUC, 0.88). Using the 2HG concentration of 1.8 mM and the glutamate concentration of 3.9 mM, diagnostic sensitivity and specificity of the algorithm were improved to 72% and 96%, respectively (AUC, 0.92) (Fig. 4B and Table 1). The increased discriminative value after glutamate was added to the 2HG was also estimated using continuous NRI and IDI. The continuous NRI and IDI were 1.2107 (P < .001; 95% CI, 0.7383–1.6831) and 0.2169 (P < .01; 95% CI, 0.0803–0.3535), respectively (Table 2). These results demonstrated that combined biomarkers using glutamate and 2HG improved diagnostic performance. In subgroup analysis of lower-grade gliomas (grades II and III), glutamate concentration was also significantly lower in IDH1 mutant glioma (P < .01) (Supplementary Fig. S6). ROC analysis demonstrated that combined measurement of 2HG and glutamate had high detection accuracy (sensitivity 88%, specificity 100%; AUC, 0.98) for IDH1 mutation compared with measurement of 2HG alone (sensitivity 75%, specificity 100%; AUC, 0.84). The continuous NRI and IDI were −1.5625 (P < .001) and −0.2887 (P < .05), respectively.
Fig. 4.
Improving predictive values for IDH1 mutation by adding glutamate to the MRS model with 2HG measurements. (A) Metabolic profiling of tumor samples assessed via in vivo MRS. Comparisons of the amount of metabolites in 47 gliomas (IDH1 wild-type, n = 29; IDH1 mutant, n = 18). Box and whisker plots show the concentration range of metabolites; the box spans the 25th and 75th percentiles of the median; the whiskers represent the 10th and 90th percentiles. P-values were calculated using the Mann–Whitney nonparametric U-test (*P < .05, **P < .01, ***P < .001). (B) Receiver-operating characteristics (ROC) curve of the combination of 2HG and glutamate (red line) and 2HG alone (blue line) in differentiating IDH1 mutant glioma from IDH1 wild-type glioma.
Table 1.
Bivariate elevation of 2HG and reduction of glutamate using ROC curve analysis
| Biomarker | Sensitivity (%) | Specificity (%) | AUC | 95% CI |
|---|---|---|---|---|
| 2HG | 56 | 96 | 0.79 | 0.6188–0.8949 |
| Glu | 78 | 90 | 0.88 | 0.7345–0.9512 |
| 2HG and Glu | 72 | 96 | 0.92 | 0.7963–0.9679 |
Table 2.
Investigation using NRI and IDI
| 95% CI | P | |
|---|---|---|
| 1. Comparison of 2HG alone and Glu alone | ||
| NRI | 0.3103 (−0.2668–0.8875) | .29194 |
| IDI | 0.0845 (−0.1276–0.2965) | .43495 |
| 2. Comparison of 2HG alone and combination of 2HG and Glu | ||
| NRI | 1.2107 (0.7383–1.6831) | <.000001 |
| IDI | 0.2169 (0.0803–0.3535) | .00186 |
Next, we also applied the cutoff values of parameter (2HG > 1.8 mM or Glu < 3.9 mM) to the validation set that consisted of 10 prospectively collected intra-axial tumors. Using these cutoff values, 100% sensitivity and 100% specificity for the detection of IDH1 mutant glioma were achieved, which was statistically significant (P = .028). The detailed profile of the patients is summarized in Supplementary Table S4.
Discussion
We noninvasively detected 2HG using conventional MRS methods in subjects with gliomas and showed a relationship between 2HG levels and mutations in IDH1 in addition to the accumulation of 2HG in tumor tissue. Oncogenic mutations in IDH1 result in the production of 2HG as an oncometabolite.5,25 2-Hydroxyglutarate contributes to epigenetic reprogramming (eg, histone and DNA methylation) and degradation of hypoxia-inducible factor 1, leading to cellular malignant transformation and tumorigenesis.26–28 Additionally, the IDH mutation inhibits cell differentiation. Thus, IDH mutations and 2HG represent a potential therapeutic target. MRS is a non-invasive imaging method for detecting IDH mutations and monitoring 2HG levels as a treatment response.29 Therefore, MRS evaluation of metabolites, including 2HG, is of clinical interest in developing a noninvasive detection method for IDH mutant gliomas.
Conventional MRS is widely available for all clinical scanners; however, the false positive rate can be high. The signals of γ-aminobutyric acid, glutamate, and glutamine partially overlap at 2.25 ppm of the 2HG peak. These overlapping spectra and the spectral decomposition of the fitting algorithm may lead to a high false positive rate in detecting 2HG. On the other hand, the low 2HG concentrations cannot be precisely detected in IDH mutant glioma using MRS (due to false negatives). The confounding effect of NAA and glutamate may serve to mask 2HG at low concentrations.30 These findings suggest that 2HG evaluation alone is not sufficient for predicting IDH1 mutant glioma. A number of MRS studies have recently been reported8–13; however, MRS studies using metabolites other than 2HG in detecting IDH1 mutations have not yet been performed. Widespread metabolic changes associated with the IDH mutation in glioma have been observed.31 A metabolomic analysis using glioma cell lines and surgical samples showed that glutamate levels are both significantly reduced in IDH mutant gliomas.22,32,33 These findings are consistent with our results (Figs 2 and 3).
Several enzymes, including GLS, GS, GDH, BCAT1, and glutamate transporters (encoded by the SLC1A2 [EAAT2], SLC1A3 [EAAT3], and SLC7A11 [xCT] genes), influence the flux of glutamate.34–37 In the present study, the mRNA and protein levels of GDH1 and GDH2 were elevated and BCAT1 expression was significantly reduced in IDH1 mutant glioma. Notably, The Cancer Genome Atlas showed that mRNA for both GDH1 and GDH2 increased in IDH1 mutant GBM relative to IDH1 wild-type GBM according to RNA sequencing data.38 BCAT1 is a cytosolic enzyme that catalyzes α-KG to glutamate.35 Decreased BCAT1 activity, mRNA expression, and protein levels have been reported in IDH1 mutant glioma cell lines. These findings indicated that the IDH1 mutation leads to metabolic changes related to glutamate beyond the production of 2HG, and reduced glutamate may reflect an attempt to replenish α-KG lost by conversion to 2HG. Further flux analysis is required to investigate detailed glutamate metabolism.
These findings motivated us to determine whether glutamate could be used as a biomarker of the IDH mutation in glioma. A further advantage of glutamate measurement is its high concentration, which enables reliable detection by MRS. The predictive value of our results was improved by adding glutamate to 2HG in discriminating IDH status.
This study has several limitations. First, examination of the impact of the mutation on metabolism is limited by the potential confounder of tumor grade. Specifically, GBM is almost all IDH1 wild-type, so it is unclear if differences in metabolism are truly due to IDH1 status or simply metabolic differences between low- and high-grade glioma. Importantly, in vitro and ex vivo studies suggest that glutamate metabolism was altered in IDH1 mutant gliomas. Second, our study performed only short-TE (35 ms) PRESS sequences using conventional MRS. Although long-TE acquisitions (97 ms) with asymmetric sequences demonstrated higher sensitivity to 2HG compared with short-TE studies,13 the detection of 2HG in IDH1 mutant gliomas with a small size had low sensitivity (<50%), even in long-TE acquisitions.39 Importantly, short-TE using conventional MRS is widely implemented in clinical scanners and has the advantage of maximizing raw signal intensity, suggesting that short-TE acquisitions may be general and useful for predicting IDH mutation by evaluating multiple metabolites. Third, mutations in IDH2 were not examined. Fourth, there was a discrepancy of MRS and GC-MS results, which may be due to difference of metabolite identification between MRS and GC-MS. Also, difference of in vivo MRS voxel size and ex vivo GC-MS tumor size possibly caused this discrepancy. Finally, there were a small number of patients in the prospective validation analysis. Although the precision of the cutoff values for maximal diagnostic accuracy remains to be defined, 2HG and glutamate may be incorporated within diagnostic algorithms in clinical practice to facilitate the workup of patients with IDH1 mutant glioma. Further prospective MRS studies should include a large number of glioma patients.
In conclusion, we conducted integrated analyses of glioma cell lines and clinical samples to improve the feasibility of mutated IDH1 detection in gliomas using noninvasive 2HG measurements with conventional 3 T MRS. Results from the MRS-based measurement of metabolites were compared with GC-MS and molecular and histopathological findings. We demonstrated that glutamate levels significantly decreased in a large number of IDH1 mutant gliomas as the expression of GDH increased and the level of BCAT1 decreased. The combined measurement of glutamate and 2HG showed highly reliable results in diagnosing patients with IDH1 mutant glioma via MRS. These findings indicated that glutamate is a potential biomarker for IDH1 mutations. The combined detection of glutamate and 2HG by MRS may be useful in the clinical management of IDH1 mutant glioma patients.
Supplementary material
Funding
This work was supported by the Japanese Ministry of Education, Culture, Sports, Science and Technology (26462181); Takeda Science Foundation; and Mochida Memorial Foundation for Medical and Pharmaceutical Research grants to K.T., and Japanese Ministry of Education, Culture, Sports, Science and Technology grants 25462258 to T.S., 26830073 to Y.I., 15K10302 to K.H., and 25293309 to E.K.
Supplementary Material
Acknowledgments
We would like to thank those at the Brain Tumor Translational Resource at Kobe University for access to biospecimens and for biorepository support.
Conflict of interest statement. None declared.
References
- 1.Turkalp Z, Karamchandani J, Das S. IDH mutation in glioma: new insights and promises for the future. JAMA Neurol. 2014;71(10):1319–1325. [DOI] [PubMed] [Google Scholar]
- 2.Kosmider O, Gelsi-Boyer V, Slama L et al. Mutations of IDH1 and IDH2 genes in early and accelerated phases of myelodysplastic syndromes and MDS/myeloproliferative neoplasms. Leukemia. 2010;24(5):1094–1096. [DOI] [PubMed] [Google Scholar]
- 3.Yan H, Parsons DW, Jin G et al. IDH1 and IDH2 mutations in gliomas. N Engl J Med. 2009;360(8):765–773. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Tanaka K, Sasayama T, Mizukawa K et al. Combined IDH1 mutation and MGMT methylation status on long-term survival of patients with cerebral low-grade glioma. Clin Neurol Neurosurg. 2015;138:37–44. [DOI] [PubMed] [Google Scholar]
- 5.Dang L, White DW, Gross S et al. Cancer-associated IDH1 mutations produce 2-hydroxyglutarate. Nature. 2009;462(7274):739–744. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Yen KE, Bittinger MA, Su SM et al. Cancer-associated IDH mutations: biomarker and therapeutic opportunities. Oncogene. 2010;29(49):6409–6417. [DOI] [PubMed] [Google Scholar]
- 7.Andronesi OC, Rapalino O, Gerstner E et al. Detection of oncogenic IDH1 mutations using magnetic resonance spectroscopy of 2-hydroxyglutarate. J Clin Invest. 2013;123(9):3659–3663. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Emir UE, Larkin SJ, de Pennington N et al. Noninvasive quantification of 2-hydroxyglutarate in human gliomas with IDH1 and IDH2 mutations. Cancer Res. 2016;76(1):43–49. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Pope WB, Prins RM, Albert Thomas M et al. Non-invasive detection of 2-hydroxyglutarate and other metabolites in IDH1 mutant glioma patients using magnetic resonance spectroscopy. J Neurooncol. 2012;107(1):197–205. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Natsumeda M, Igarashi H, Nomura T et al. Accumulation of 2-hydroxyglutarate in gliomas correlates with survival: a study by 3.0-tesla magnetic resonance spectroscopy. Acta Neuropathol Commu. 2014;2:158. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Andronesi OC, Kim GS, Gerstner E et al. Detection of 2-hydroxyglutarate in IDH-mutated glioma patients by in vivo spectral-editing and 2D correlation magnetic resonance spectroscopy. Sci Transl Med. 2012;4(116):116ra114. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Choi C, Ganji SK, DeBerardinis RJ et al. 2-Hydroxyglutarate detection by magnetic resonance spectroscopy in IDH-mutated patients with gliomas. Nat Med. 2012;18(4):624–629. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Choi C, Ganji S, Hulsey K et al. A comparative study of short- and long-TE (1)H MRS at 3 T for in vivo detection of 2-hydroxyglutarate in brain tumors. NMR Biomed. 2013;26(10):1242–1250. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Provencher SW. Automatic quantitation of localized in vivo 1H spectra with LCModel. NMR Biomed. 2001;14(4):260–264. [DOI] [PubMed] [Google Scholar]
- 15.Vlassenbroeck I, Califice S, Diserens AC et al. Validation of real-time methylation-specific PCR to determine O6-methylguanine-DNA methyltransferase gene promoter methylation in glioma. J Mol Diagn. 2008;10(4):332–337. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Wise DR, Ward PS, Shay JE et al. Hypoxia promotes isocitrate dehydrogenase-dependent carboxylation of alpha-ketoglutarate to citrate to support cell growth and viability. Proc Natl Acad Sci U S A. 2011;108(49):19611–19616. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Tsugawa H, Bamba T, Shinohara M et al. Practical non-targeted gas chromatography/mass spectrometry-based metabolomics platform for metabolic phenotype analysis. J Biosci Bioeng. 2011;112(3):292–298. [DOI] [PubMed] [Google Scholar]
- 18.Tanaka K, Sasayama T, Irino Y et al. Compensatory glutamine metabolism promotes glioblastoma resistance to mTOR inhibitor treatment. J Clin Invest. 2015;125(4):1591–1602. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Sasayama T, Tanaka K, Mizowaki T et al. Tumor-associated macrophages associate with cerebrospinal fluid interleukin-10 and survival in primary central nervous system lymphoma (PCNSL). Brain Pathol. 2015. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Xia J, Sinelnikov IV, Han B et al. MetaboAnalyst 3.0—making metabolomics more meaningful. Nucleic Acids Res. 2015;43(W1):W251–W257. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Pencina MJ, D'Agostino RB Sr, D'Agostino RB Jr et al. Evaluating the added predictive ability of a new marker: from area under the ROC curve to reclassification and beyond. Stat Med. 2008;27(2):157–172; discussion 207–112. [DOI] [PubMed] [Google Scholar]
- 22.Izquierdo-Garcia JL, Viswanath P, Eriksson P et al. Metabolic reprogramming in mutant IDH1 glioma cells. PloS One. 2015;10(2):e0118781. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Li C, Allen A, Kwagh J et al. Green tea polyphenols modulate insulin secretion by inhibiting glutamate dehydrogenase. J Biol Chem. 2006;281(15):10214–10221. [DOI] [PubMed] [Google Scholar]
- 24.Wang JB, Erickson JW, Fuji R et al. Targeting mitochondrial glutaminase activity inhibits oncogenic transformation. Cancer Cell. 2010;18(3):207–219. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Xu W, Yang H, Liu Y et al. Oncometabolite 2-hydroxyglutarate is a competitive inhibitor of alpha-ketoglutarate-dependent dioxygenases. Cancer Cell. 2011;19(1):17–30. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Yang H, Ye D, Guan KL et al. IDH1 and IDH2 mutations in tumorigenesis: mechanistic insights and clinical perspectives. Clin Cancer Res. 2012;18(20):5562–5571. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Lu C, Ward PS, Kapoor GS et al. IDH mutation impairs histone demethylation and results in a block to cell differentiation. Nature. 2012;483(7390):474–478. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Tahiliani M, Koh KP, Shen Y et al. Conversion of 5-methylcytosine to 5-hydroxymethylcytosine in mammalian DNA by MLL partner TET1. Science. 2009;324(5929):930–935. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Andronesi OC, Loebel F, Bogner W et al. Treatment response assessment in IDH-mutant glioma patients by non-invasive 3D functional spectroscopic mapping of 2-hydroxyglutarate. Clin Cancer Res. 2015. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Bertolino N, Marchionni C, Ghielmetti F et al. Accuracy of 2-hydroxyglutarate quantification by short-echo proton-MRS at 3 T: a phantom study. Phys Med. 2014;30(6):702–707. [DOI] [PubMed] [Google Scholar]
- 31.Reitman ZJ, Jin G, Karoly ED et al. Profiling the effects of isocitrate dehydrogenase 1 and 2 mutations on the cellular metabolome. Proc Natl Acad Sci U S A. 2011;108(8):3270–3275. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Ohka F, Ito M, Ranjit M et al. Quantitative metabolome analysis profiles activation of glutaminolysis in glioma with IDH1 mutation. Tumour Biol. 2014;35(6):5911–5920. [DOI] [PubMed] [Google Scholar]
- 33.Elkhaled A, Jalbert LE, Phillips JJ et al. Magnetic resonance of 2-hydroxyglutarate in IDH1-mutated low-grade gliomas. Sci Transl Med. 2012;4(116):116ra115. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Csibi A, Fendt SM, Li C et al. The mTORC1 pathway stimulates glutamine metabolism and cell proliferation by repressing SIRT4. Cell. 2013;153(4):840–854. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Tonjes M, Barbus S, Park YJ et al. BCAT1 promotes cell proliferation through amino acid catabolism in gliomas carrying wild-type IDH1. Nat Med. 2013;19(7):901–908. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Robert SM, Sontheimer H. Glutamate transporters in the biology of malignant gliomas. Cell Mol Life Sci. 2014;71(10):1839–1854. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Hensley CT, Wasti AT, DeBerardinis RJ. Glutamine and cancer: cell biology, physiology, and clinical opportunities. J Clin Invest. 2013;123(9):3678–3684. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Chen R, Nishimura MC, Kharbanda S et al. Hominoid-specific enzyme GLUD2 promotes growth of IDH1R132H glioma. Proc Natl Acad Sci U S A. 2014;111(39):14217–14222. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.de la Fuente MI, Young RJ, Rubel J et al. Integration of 2-hydroxyglutarate-proton magnetic resonance spectroscopy into clinical practice for disease monitoring in isocitrate dehydrogenase-mutant glioma. Neuro Oncol. 2015. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.