Introduction

Gliomas are most the commonly occurring primary malignant tumor in the central nervous system. Mutations in the cytosolic isocitrate dehydrogenase 1 (IDH1) gene are present in up to 90% of primary low-grade gliomas and result in the neomorphic activity of the enzyme, which catalyzes conversion of α-ketoglutarate to the oncometabolite 2-hydroxyglutarate (2-HG).^1,2 Mutant IDH1 tumors are less aggressive than primary glioblastoma, but nonetheless always recur and ultimately lead to patient death. The treatment of IDH1 mutant patients with Temozolomide (TMZ), previously reserved for the treatment of the more aggressive primary glioblastoma, improves survival,^3,4 but there remains a need for complementary imaging methods to assess early response to therapy. The goal of this study was therefore to determine the magnetic resonance spectroscopy (MRS)-based biomarkers that can serve indicators of low-grade glioma response to treatment.

Methods

NHA and U87 cells genetically engineered to express mutant IDH1 (i.e. NHAIDHmut and U87IDHmut cells) were maintained in culture as previously described⁵ and treated either with their IC₅₀ value of TMZ (100 μM; N=5), or DMSO (0.2%; N=5) for 72 hours. SF10602, patient-derived astrocytoma cells, were treated with their IC₅₀ value of 400 μM TMZ (N=4) or DMSO (N=4) for 7 days. To assess metabolic fluxes cells were also labelled with [1-¹³C] glucose or [3-¹³C] glutamine. Cells were then extracted using the dual-phase extraction method and ¹H-MRS (1D water presaturation ZGPR sequence, 90° FA, 3s TR, 256 acquisitions) and proton-decoupled ¹³C-MRS (30° FA, 3s TR, 2048 acquisitions) spectra of extracted metabolites were acquired using a 500 MHz Bruker Avance spectrometer. Multivariate analysis was performed on the ¹H MRS data using SIMCA.⁶ Correlation values and variable importance for projection scores were used to identify significantly altered metabolites following treatment. ¹H and ¹³C MRS peaks were manually integrated and quantified using Mnova7. Spectral integrals were normalized to TSP and to cell number and statistical significance of differences determined using unpaired Student’s t-test (Graphpad Prism). For hyperpolarized [2-¹³C] pyruvate ¹³C-MRS studies, cells were encapsulated in agarose, perfused in an MR compatible perfusion system, exposed to hyperpolarized [2-¹³C] pyruvate and ¹³C spectra acquired at 11.7T (5° FA and 3s TR over 300s, 500MHz INOVA spectrometer) to monitor the production of hyperpolarized [5-¹³C] glutamate. For hyperpolarized [1-¹³C] α-ketoglutaric acid studies live cells were exposed to [1-¹³C] α-ketoglutaric acid and ¹³C spectra acquired at 1.47T in order to leverage the longer hyperpolarized T1 at lower field (13° FA and 3s TR over 300s, pulsar 1.47T) production of hyperpolarized [5- [1-¹³C] glutamate was monitored. Hyperpolarized glutamate signal was then quantified using Mnova7 and normalized to the respective substrates and cell number.

Results and Discussion

Treatment of NHAIDHmut, U87IDHmut and SF10602 cells with their IC₅₀ of TMZ resulted in 47.59±8.63%, 49.13±4.11% and 56.32 ± 6.48% reduction in cell number, respectively. Chemometric analysis of ¹H MR spectra showed the metabolic discrimination between TMZ-treated cells and controls. First, PCA visualized the inherent clustering of groups. Further, improved separation between the groups was obtained by supervised classification models, OPLS-DA (Fig.1). Following cross-validation of models for good predictive ability and accuracy, the OPLS-DA model was used to extract the S-line and VIP plots. Most significant metabolites contributing to class separation were successfully identified using the correlation coefficients and VIP with thresholds > ±0.60 and ≥ 1, respectively (Fig. 1). Next, univariate analysis was applied to the integral values of the metabolites identified in this manner. Several metabolites were altered, but the most notably change was an increase in glutamate following treatment and this metabolite was further investigated as a translatable metabolic markers of low-grade glioma response to TMZ treatment (Fig.2). An increase in glutamate was confirmed in the clinically relevant patient-derived cell model, SF10602 (Fig.2). To further assess whether the increase in glutamate could be explained by an increase in TCA cycle flux, synthesis of glutamate from [1-¹³C] glucose, as well as from [3-¹³C] glutamine were probed (Fig.3). Consistent with the increase in total metabolite levels, both glucose- and glutamine-derived glutamate were increased in TMZ treated cells compared to controls, together explaining the increase in total pools (Fig.3). Furthermore, dynamically probing the metabolism of hyperpolarized [2-¹³C] pyruvate in live cells revealed that build-up of [5-¹³C] glutamate, which is associated with flux via the TCA cycle, was significantly higher in TMZ-treated cells in both the NHAIDHmut and U87IDHmut models (Fig.4). Following injection of [1-¹³C] α-ketoglutaric acid, a detectable build-up of [1-¹³C] glutamate was also observed in both cell models and was significantly increased following TMZ treatment (Fig.5). In line with these findings, total cellular activity of enzymes known to catalyze the α-ketoglutaric acid to glutamate conversion, namely aspartate transaminase and alanine aminotransferase were found to be upregulated in TMZ-treated cells.

Conclusion

This study shows that ¹³C MRS detectable production of hyperpolarized [1-¹³C] and [5-¹³C] glutamate from hyperpolarized [1-¹³C] α-ketoglutaric acid and [2-¹³C] pyruvate, respectively, can be used to monitor response to TMZ therapy in mutant IDH1 cells. Further studies are warranted to confirm our findings in vivo in animal models. Nonetheless, these findings identify a possible new approach for improving currently available imaging methods for early detection of response to therapy in low-grade mutant IDH1 glioma.

Acknowledgements

This work is supported by NIH R01CA197254 and NIH center grant P41EB013598.