Isocitrate Dehydrogenase Mutation Assessment in Human Glioma using Desorption Electrospray Ionization-Multiple Reaction Monitoring

Abstract

Mutation status of isocitrate dehydrogenase (IDH) in glioma patients has diagnostic and prognostic significance but is available only after surgery. Rapid and sensitive intraoperative molecular diagnosis is essential for distinguishing tumor from infiltrative and from normal brain tissue to maximize safe resection. IDH-mut tumors produce high concentrations of 2-hydroxyglutarate (2HG), which serves as a surrogate marker of IDH mutation. Here, we use desorption electrospray ionization-multiple reaction monitoring (DESI-MRM), a form of mass spectrometry (MS), for rapid prediction of IDH mutation status. A set of 30 human brain smears was analyzed using DESI-MRM experiments to quantify 2HG and predict the mutation status of IDH in glioma samples within 2 min (endogenous standard) or 7 min (external standard). Nano-electrospray ionization (nESI) was employed as a validation method. An isotopically labeled surrogate internal standard allowed quantitative measurement of 2HG and clear separation between IDH-mut and IDH-wildtype (wt) samples, showing 100% agreement with clinical diagnosis. 2HG concentrations in IDH-mut samples ranged from 88 to 1,445 ng mg⁻¹, while IDH-wt samples exhibited 2HG concentrations of 9 – 15 ng mg⁻¹. Calibration curves showed method accuracy within ≤ 15%. The quantitative results obtained using isotopically labeled standards showed a strong correlation between IDH mutation score and 2HG concentrations, as did an earlier semiquantitative (ratiometric, endogenous standard) method. These findings highlight the standalone capability of DESI-MRM for making measurements to justify assessments for intraoperative diagnosis and to support surgical decision-making to improve patient outcomes.

Infiltrative glioma is the most malignant primary tumor that forms in the adult central nervous system (CNS) and is associated with poor survival rates¹. Surgical resection is the mainstay of treatment, and extent of resection (EOR) is one of the strongest prognostic factors, significantly affecting patient survival². Given the diffuse nature of glioma, the infiltrated tumor residues often remain beyond surgical margins, leading to tumor recurrence despite chemotherapy and radiation³. Therefore, a critical need persists for highly sensitive, fast, efficient and accurate diagnostics to support surgical decision-making at the margins. Conventional microscopic evaluation of hematoxylin and eosin (H&E) stained tissue provides only morphological information, such as tumor cell density and grade, without molecular resolution. Intraoperative magnetic resonance imaging offers limited molecular information and is insensitive to microscopically infiltrated regions. The gold standards for molecular diagnostics, immunohistochemistry (IHC) and genetic testing, are time-consuming and not compatible with the intraoperative workflow.

Among glioma molecular markers, the mutation status of isocitrate dehydrogenase (IDH) is particularly important for accurate diagnosis, such as differentiating IDH mutant (IDH-mut) astrocytoma from glioblastoma (GBM) under the 2021 World Health Organization (WHO) classification of CNS tumors¹. IDH mutation status also carries prognostic value as patients with IDH-mut gliomas generally have longer overall survival rates⁴. The neomorphic gain-of-function mutation alters the nicotinamide adenine dinucleotide phosphate (NADP⁺)-dependent conversion of isocitrate to α-ketoglutaric acid (αKG), instead converting αKG to 2-hydroxyglutarate (2HG), which accumulates in cancer cells and serves as the surrogate biomarker of IDH-mut gliomas (Fig. 1a)^4–7. Multiple studies have confirmed 2HG accumulation in IDH-mut gliomas and demonstrated a correlation between elevated 2HG abundance at margins and tumor infiltration^8–10. Moreover, mapping 2HG concentration at tumor margins can guide surgical resection, especially at infiltrative boundaries. Importantly, with ongoing clinical trials targeting IDH-mut gliomas^11,12, knowledge of IDH mutation status can affect treatment planning and enable personalized medicine. Therefore, assessment of IDH mutation status is essential for accurately classifying gliomas, guiding therapy, and predicting patient outcomes.

Fig. 1.

a) IDH-mut (IDH1/2-mut) converts αKG to D-2HG which accumulates in glioma tissue. b) Schematic illustration of DESI-MRM measurement for detecting 2HG in glioma smears using a DESI source coupled to a TQ system. Drawings are not to scale.

Analytical techniques such as Raman and fluorescence imaging are being explored for rapid glioma diagnosis^13,14. Mass spectrometry (MS) is a sensitive analytical tool that stands out from other spectroscopic platforms for its superior molecular specificity. MS has been widely used in clinical chemistry and is increasingly being investigated in cancer diagnosis^15–17. The development of ambient ionization mass spectrometry (AIMS)¹⁸ opened a door to translating MS into the operating room (OR). AIMS enables rapid sample analysis largely due to minimal sample preparation requirements. Several AIMS approaches have been developed and used for intraoperative diagnosis, including desorption electrospray ionization (DESI)⁸, rapid evaporative ionization mass spectrometry (REIMS)¹⁹, MasSpec Pen²⁰, touch spray (TS)²¹, picosecond infrared laser mass spectrometry (PIRL-MS)²², and cavitron ultrasonic surgical aspiration/sonic spray ionization mass spectrometry (CUSA/SSI-MS)²³. DESI has been among the most utilized AIMS techniques in the field of cancer diagnosis and imaging²⁴. However, direct analysis of complex biological matrices without prior separation, such as chromatography, presents challenges due to matrix interferences²⁵. Integration of DESI with MS/MS can reduce interferences from matrix during ion measurement and so enable rapid analysis of minimally processed tissue. MRM a form of MS/MS measurement that is commonly performed using triple quadrupoles (TQ), first described by our group in the late 1970s and is known for its high analytical sensitivity in spite of a relatively low specificity^26,27. During MRM experiments, only selected precursor-product ion pairs are monitored, resulting in higher scan rates, improved signal-to-noise ratios, and enhanced analytical sensitivity (Fig. 1b). A detailed description of various MS/MS scan modes was provided by Synder et al.²⁸. Integration of DESI with MRM thus allows rapid and sensitive biomarker analysis directly from complex samples such as brain tissue.

In this study, brain samples were analyzed by DESI-MRM to (i) quantify 2HG concentrations in glioma tissue, (ii) assess IDH mutation status using a ratiometric (semiquantitative) approach, and (iii) correlate the ratiometric assessments of IDH mutation status to quantitative determinations of 2HG. We note that in previous studies, IDH mutation status in glioma samples was assessed using MS/MS methods aimed at improving clinical performance. However, no experiments employing AIMS for direct 2HG quantification from tissue have been reported^8,9,29,30. Ion suppression is a common phenomenon when analyzing complex matrices such as brain tissue, where readily ionizable molecules mask the analytes with poor ionization efficiency or at low concentrations. This effect can be addressed in two ways: (i) by adding an isotopically labeled molecule to the sample to serve as an internal standard for quantitation, or (ii) by using an intrinsic brain tissue constituent present at an approximately constant concentration so mitigating matrix effects on ionization (as described in earlier literature²⁴).

A set of 26 glioma and 4 non-glioma samples was smeared and analyzed by DESI-MRM. An isotopically labeled surrogate internal standard (IS) was added to tissue samples prior to smearing for 2HG quantification. Concentrations of 2HG were measured in IDH-mut and IDH-wt samples, and the clinical performance of DESI-MRM for predicting IDH mutation status was evaluated by comparison with IHC and genetic testing results. Assessment of IDH mutation status followed a previously reported semiquantitative (ratiometric) method^8,9,29, in which the relative abundance of 2HG to glutamate (Glu), an endogenous reference molecule, was measured. The performance of semiquantitative MS predictions was then compared with IHC and/or genetic test results. To further validate results, tissue extracts from a subset of brain samples were analyzed by nESI-MRM and compared with DESI-MRM data. Supplementary Table S1 summarizes the experimental details of DESI-MRM and nESI-MRM. Both quantitative and semiquantitative (ratiometric) predictions achieved complete agreement with clinical diagnosis, underscoring the promise of DESI-MRM for intraoperative glioma diagnosis.

Experimental Section

Brain samples

A total of 30 brain samples, including nine IDH-mut glioma, 17 IDH-wt glioma, and four non-glioma controls, were purchased from the Cooperative Human Tissue Network (CHTN) under a material transfer agreement with Purdue University. All samples were stored at −80 °C until analysis. The study was done in accordance with institutional review board (IRB) approval (IRB no. 2024–1417) and relevant regulatory guidelines. Details on tissue type, IDH mutation status, and tumor grade are provided in Supplementary Table S2.

Sample Preparation

For DESI-MRM experiments, small sections of tissue biopsies were weighed, and aliquots of N-acetylaspartic acid-d₃ (NAA-d₃) (CDN Isotopes, 1.5 mM) were added to each sample and mixed to achieve a final concentration of the IS of 1.5 μmol g⁻¹. For calibration curve preparation, mouse brain homogenates were spiked with varying amounts of 2HG (Sigma), ranging from 5 nmol g⁻¹ to 40 μmol g⁻¹. Samples were smeared onto microscopic glass slides and allowed to dry prior to analysis.

For nESI-MRM validation experiments, 12 brain samples (6 IDH-mut and 6 IDH-wt) were extracted according to the Bligh and Dyer protocol³¹. The aqueous phase of each extract was dried using an Eppendorf Vacufuge Plus (RRID:SCR_019876) and reconstituted in MeOH. NAA-d₃ was then added to each sample to a final concentration of 1.5 μmol g⁻¹.

MS experiments

After optimization of the DESI source (prototype, Waters), a solvent mixture of MeOH-H₂O (98:2) was delivered at a flow rate of 3 μL min⁻¹ from the sprayer, held at 0.86 kV. Product ion scans of Glu, 2HG, and NAA were performed in negative ion mode, and selected ion transitions were iteratively recorded during DESI-MRM experiments for 1.8 min (see Supplementary Fig. S1). The most abundant transitions were selected for the quantitative measurements (2HG m/z 147 → m/z 129 and NAA-d₃ m/z 177 → m/z 91). A TQ (Xevo^™ TQ-S micro, Waters) was used, and each experiment was repeated three times.

For nESI-MRM experiments, pulled borosilicate glass capillaries with 5 μm tip inner diameters were filled with 10 μL of brain extracts. Ionization was initiated by applying 0.86 kV potential, and the same MRM transitions were recorded for 1.8 min. The nESI experiments were repeated three times.

Assessment of IDH mutation status was made according to the previous reports^8,9. Briefly, product ion fragments of Glu and 2HG were recorded during DESI-MRM experiments and by correcting for the contribution of ¹³C-Glu to 2HG signal, the ratio of 2HG (m/z 129) to Glu (m/z 128) was calculated (Equation (1)).

$$\mathit{\text{IDH}}\mathit{\text{mutation}}\mathit{\text{score}}=\frac{I_{129}-\left(I_{128}\times 0.061\right)}{I_{128}}$$ Equation (1)

Data Analysis

MRM raw data were converted to .txt format using MSConvertGUI (Proteowizard (RRID:SCR_012056)). A customized MATLAB script (RRID:SCR_001622) was used to extract MRM transition intensities and process the data. Average concentrations and IDH mutation scores from three repeats were calculated for statistical analysis. ROC analysis was performed using MetaboAnalyst 6.0 (RRID:SCR_015539), and cutoffs between IDH-mut and IDH-wt samples were estimated using the Youden index. Sensitivity, specificity, and accuracy were calculated from the confusion matrix based on the optimal threshold. Statistical significance between IDH-mut and IDH-wt samples was assessed using the Mann–Whitney rank-sum test (StataSE 19, RRID:SCR_012763), with p < 0.05 considered significant. Supplementary Table S3 provides 2HG concentrations and IDH mutation scores for each sample.

Results and Discussion

2HG quantitation in glioma by DESI-MRM

DESI-MRM experiments were performed to quantify 2HG in glioma samples. For calibration curve preparation, rather than using standard solutions alone, IS and 2HG were spiked into mouse brain homogenates to account for tissue microextraction effects and matrix interferences present in real samples. Varying concentrations of 2HG, from 5 nmol g⁻¹ to 30 μmol g⁻¹ (equivalent to 0.7 – 4440 ng mg⁻¹), were added to mouse brain, and a calibration curve was prepared accordingly (Fig. 2a). The linear dynamic range was 50 nmol g⁻¹ to 15 μmol g⁻¹ (equivalent to 7.4 – 2,220 ng mg⁻¹) with excellent linearity (R² = 0.9957). The limit of detection (LOD) was 5 nmol g⁻¹ and the limit of quantification (LOQ) was 50 nmol g⁻¹. The relative standard deviation (RSD) across concentrations was ≤ 15%. From the calibration curve, 2HG concentrations were calculated for IDH-wt and IDH-mut samples. In IDH-mut samples, 2HG ranged from 0.597 to 9.767 μmol g⁻¹ (equivalent to 88 – 1445 ng mg⁻¹) with precision ≤ 15%. By contrast, IDH-wt samples showed 2HG concentrations of 9 – 15 ng mg⁻¹, not much above the LOD of 5 nmol g⁻¹. A clear separation between IDH-wt and IDH-mut samples was observed based on 2HG concentrations (Fig. 2b), where box plots display the interquartile range, median line, and whiskers at ±1.5 IQR (***p < 0.0001).

Fig. 2.

Quantitative DESI-MRM results based on use of exogenous internal standard: a) Calibration curve for mouse brain tissue spiked with 2HG at seven concentrations of 2HG and normalized to NAA-d₃ - using DESI-MRM (RSD ≤ 15%), b) Box plots with whiskers at ±1.5 interquartile range (IQR) and median lines, demonstrating clear separation between IDH-mut and IDH-wt human tissue samples based on 2HG concentrations (***p < 0.0001), c) ROC curve showing the optimal 2HG concentration (0.35 μmol g⁻¹) for distinguishing IDH-mut from IDH-wt samples, with AUC = 1.00, d) Confusion matrix evaluating the clinical performance of 2HG quantitation for predicting IDH status, showing 100% accuracy.

The receiver operating characteristic (ROC) is shown in Fig 2c, with an AUC of 1.00. The optimal cutoff between IDH-mut and IDH-wt samples was estimated at 0.35 μmol g⁻¹ (52 ng mg⁻¹). Sensitivity, specificity, and accuracy of the 2HG quantification method for predicting IDH mutation status were calculated using a confusion matrix, all yielding 100% (Fig. 2d). It should be noted that the reported performance parameters should be interpreted as proof-of-concept results rather than definitive clinical performance. A larger independent cohort is required for future validation of clinical performance. The IDH-wt samples included both glioma and non-glioma brain tissues. To further evaluate the accuracy of the DESI-MRM method, 2HG was spiked into mouse brain tissue, and recoveries had errors less than 15%.

Semiquantitative (ratiometric) assessment of IDH mutation status by DESI-MRM

In addition to the quantitative assessment of 2HG, a semiquantitative (ratiometric, endogenous standard) approach was applied to predict IDH mutation status in brain samples. In this method, the ratio of 2HG to Glu (used as the endogenous reference) was calculated, and the resulting values were defined as the IDH mutation score²⁹. As shown in Fig. 3a, the ratiometric method achieved clear separation between IDH-wt and IDH-mut samples (***p < 0.0001), demonstrating its robustness. The ROC curve yielded an AUC of 1.00, with a log(IDH mutation score) threshold of – 0.161 for discrimination between IDH-wt and IDH-mut samples (Fig. 3b). A confusion matrix constructed at the optimal cutoff demonstrated complete agreement between DESI-MRM predictions and IHC (and/or genetic testing), with sensitivity, specificity, and accuracy all at 100% (Fig. 3c). The correlation between log(IDH mutation score) and 2HG concentration (as determined in the method of Fig. 2b) is shown in Fig. 3d: IDH-mut samples exhibited high mutation scores associated with elevated 2HG concentrations, whereas IDH-wt samples showed low concentrations and low scores. These findings indicate that high IDH mutation scores are strongly correlated with increased 2HG abundance in gliomas. Therefore, in cases where direct 2HG quantification is not feasible, the ratiometric method provides an excellent alternative for predicting IDH mutation status.

Fig. 3.

Semiquantitative (ratiometric) DESI-MRM results based on use of endogenous internal standard: a) Box plots of IDH-mut and IDH-wt human brain samples showing clear separation using the ratiometric method. b) ROC curve demonstrating an AUC of 1.00 and the cutoff ratio distinguishing IDH-mut from IDH-wt samples. c) Confusion matrix comparing ratiometric predictions from DESI-MRM with clinical diagnosis, and showing complete concordance. d) Correlation of log(IDH mutation score) with 2HG concentrations in IDH-mut and IDH-wt samples, with dashed lines indicating the cutoff values for the quantitative and ratiometric methods.

2HG quantification by nESI-MRM

nESI-MRM experiments were conducted in parallel with DESI-MRM to validate the performance of DESI where tissue extraction was not performed. A series of standard 2HG solutions containing the IS was prepared and analyzed by nESI-MRM to generate a calibration curve. Due to the limited availability of human tissue, six IDH-mut and six IDH-wt samples were selected and extracted using the Bligh and Dyer protocol³¹. The aqueous phase of each extract was dried using a Vacufuge Plus and the metabolite extracts were reconstituted in methanol, and NAA-d₃ was added to achieve a final concentration of 1.5 μmol g⁻¹. The calibration curve obtained from nESI-MRM measurements showed a linear response from 0.05 to 15 μmol g⁻¹ (Fig. 4a). Box plots of IDH-wt and IDH-mut samples revealed clear separation between the two groups based on 2HG concentrations in tissue extracts (Fig. 4b). The correlation between IDH mutation scores and 2HG concentrations was consistent with the DESI-MRM results: IDH-mut samples displayed elevated 2HG levels and higher mutation scores (Fig. 4c). A comparison of 2HG concentrations obtained by DESI-MRM and nESI-MRM (Fig. 4d) showed good agreement across most samples. Minor deviations between the two methods may be attributed to tumor heterogeneity in gliomas, particularly since DESI-MRM analyzes smeared samples rather than homogenized tissue. All IDH-wt samples measured by both techniques exhibited very low 2HG concentrations, confirming the high specificity of the method.

Fig. 4.

Quantitative nESI-MRM results using exogenous internal standards: a) Calibration curve of 2HG at varying concentrations obtained from nESI-MRM experiments performed in the brain extracts, showing excellent linearity. b) Box plots with whiskers at ±1.5 IQR illustrating 2HG concentrations in IDH-wt and IDH-mut brain samples, where 2HG serves as the discriminatory biomarker. c) Correlation of 2HG concentrations in IDH-wt and IDH-mut samples with their corresponding IDH mutation scores. d) Comparison of 2HG concentrations measured by DESI-MRM and nESI-MRM in IDH-mut and IDH-wt samples.

DESI-MRM demonstrates capability for simultaneous quantitative and semiquantitative (ratiometric) assessment of glioma smears rapidly with minimal sample preparation. Analysis of minimally processed brain smears accelerates the workflow taking only 1.8 min per measurement while reducing the use of organic solvents required for extraction. The use of MS/MS, particularly in the MRM mode, enables sensitive detection of analytes while mitigating matrix interferences. IDH-mut glioma cells produce large amounts of 2HG that accumulate in tumor tissue due to the neomorphic gain-of-function mutation. Accordingly, 2HG concentrations measured in glioma samples serve as a surrogate marker of IDH mutation status. The relative abundance of 2HG to Glu (endogenous reference molecule) was also evaluated for IDH mutation prediction, and the performance of both quantitative and semiquantitative (ratiometric) methods was compared. Clear separation between IDH-mut and IDH-wt gliomas was observed using both 2HG concentrations and the ratiometric 2HG/Glu method (***p < 0.0001). IDH-mut samples exhibited 2HG concentrations ranging from 88 to 1445 ng mg⁻¹, whereas IDH-wt samples contained only 9–15 ng mg⁻¹, confirming that 2HG is highly specific to IDH-mut tissue. ROC curve analysis estimated a cutoff concentration of 52 ng mg⁻¹ for discrimination between IDH-mut and IDH-wt tumors. The quantitative DESI-MRM method demonstrated precision and accuracy ≤ 15% on minimally processed brain tissue under ambient conditions. The ratiometric method yielded a threshold log(IDH mutation score) of −0.161. Regardless of absolute 2HG concentration, the 2HG/Glu ratio provided a robust measure of IDH mutation status. Moreover, previous studies have shown that Glu levels are reduced in IDH-mut gliomas compared with IDH-wt tissue³², further enhancing the discriminative power of the ratiometric approach. Correlation analysis confirmed that IDH mutation scores and 2HG concentrations increased concurrently in IDH-mut samples, while both remained extremely low in IDH-wt tumors. Both the ratiometric and the quantitative analysis of 2HG offer valuable diagnostic information for glioma subtype classification, while absolute 2HG quantification may provide insights into patient response to targeted therapies. The primary difference between the two methods lies in the nature of the reference standard: the semiquantitative approach uses an endogenous molecule, whereas the quantitative method relies on an exogenous isotopically labeled IS. As a result, the semiquantitative method enables faster measurements, with a turnaround time of approximately 3 min, while sample preparation and standard addition for quantitative analysis extend the turnaround time to about 7 min. (compare Table S4). Quantitative measurements provide more accurate diagnostic information and serve to validate the semiquantitative results. Overall, DESI-MRM enables rapid assessment of IDH mutation status and simultaneous quantification of 2HG in brain smears, making it a promising tool for intraoperative and clinical diagnosis.

Conclusions

A DESI-MRM method was developed for the simultaneous assessment of IDH mutation status and quantification of 2HG in glioma samples. A total of 30 brain smears were analyzed to measure 2HG, Glu, and NAA-d₃ (as the IS) using MRM experiments. Both quantitative (isotopically labeled IS) and semiquantitative (ratiometric, endogenous standard) analyses showed clear separation between IDH-mut and IDH-wt samples (Mann–Whitney test), with 100% concordance with IHC and/or genetic testing. All IDH-mut samples exhibited 2HG concentrations above 88 ng mg⁻¹ and up to 1445 ng mg⁻¹, while 2HG levels in IDH-wt brain samples were close to LOQ, confirming the high specificity of 2HG as a surrogate marker of IDH mutation status. The semiquantitative (ratiometric) method demonstrated robustness, providing accurate IDH mutation assessment independent of absolute 2HG concentration, as elevated 2HG levels and reduced Glu concentrations enhanced its discriminative power. Validation of DESI-MRM results by nESI-MRM experiments showed strong agreement between the two techniques. Overall, DESI-MRM enables rapid quantitative and semiquantitative analysis of glioma tissue, making it a valuable approach for intraoperative and clinical applications where both molecular quantification and phenotypic information are essential for improving diagnostic precision and patient care.

Supplementary Material

MRM transitions of analytes, additional experimental details, and quantitative and semiquantitative data.

Data availability

MS data can be found at Purdue University Research Repository (PURR)³³.