Multi-Center and Multi-Vendor Study of Long-TE 1H MRS at 3T for Detection of 2-Hydroxyglutarate in Brain Tumors In Vivo
Changho Choi1, Thomas Huber2, Anna Tietze3, Byung Se Choi4, Jung Hee Lee5, Seung-Koo Lee6, Alexander Lin7, and Sunitha Thakur8

1UT Southwestern Medical Center, Dallas, TX, United States, 2Technical University of Munich, Munich, Germany, 3Aarhus University Hospital, Aarhus, Denmark, 4Seoul National University College of Medicine, Seongnam, Korea, Republic of, 5Sungkyunkwan University School of Medicine, Seoul, Korea, Republic of, 6Yonsei University College of Medicine, Seoul, Korea, Republic of, 7Harvard Medical School, Boston, MA, United States, 8Memorial Sloan-Kettering Cancer Center, New York, NY, United States


The non-invasive identification of elevated 2-hydroxyglutarate (2HG) in IDH-mutated gliomas by 1H MRS in vivo is a major breakthrough in brain tumor research. Studies have shown that optimized long-TE approaches may confer advantages over short-TE MRS for detecting 2HG. Here we report an evaluation of the feasibility of long-TE 2HG MRS in Philips, Siemens and GE 3T scanners. Echo times were optimized, with numerical simulations and phantom validation, for the vendor-specific RF pulses. In-vivo data from IDH-mutated glioma patients, obtained in the three vendors, are discussed.


Following the discovery of 2-hydroxyglutarate (2HG) production in gliomas with mutations in isocitrate dehydrogenase 1 and 2 1, the capability of in-vivo detection of this onco-metabolite by MRS has become a central interest in cancer research. Recent studies 2,3 showed that optimized long-TE MRS may be advantageous over standard short-TE MRS for 2HG measurement at 3T. Here we aim to evaluate the performance of long-TE MRS for 2HG detection in Philips, Siemens and GE 3T scanners.


Among the five non-exchangeable J-coupled proton resonances of 2HG, the C4-proton resonances at ~2.25 ppm give rise to the largest signal in most experimental situations. The dependence of this 2.25-ppm resonance on the subecho times of PRESS, TE1 and TE2, was examined with numerical density-matrix simulations that incorporated the slice-selective RF and gradient pulses which are readily available in Philips, Siemens, and GE clinical scanners (Fig. 1). In-vitro experiments were conducted on a phantom solution with 2HG and Gly at three sites. In-vivo experiments were performed in patients with IDH-mutated gliomas at multiple centers (9 sites). Spectra were analyzed with LCModel using calculated spectra of 20 metabolites.


The slice-selective refocusing RF pulse envelopes of the PRESS sequences in Philips, Siemens and GE 3T scanners were quite different, as shown in Fig. 1. In the GE PRESS, the flip angle of the refocusing pulse was set at 137° to improve the frequency profile shape. Numerical simulations indicated that, in each of the Philips, Siemens and GE PRESS sequences, the 2HG signal at short TE (e.g., ≤ 30 ms) was large but it was broad and thus not very ideal for 2HG signal differentiation from neighboring resonances (spectra not shown). In all cases, the 2HG signal varied extensively with changing TE1 and TE2, showing asymmetric dependence on TE1 and TE2 (Fig. 2). When normalized to the signal at (TE1, TE2) = (12, 12) ms, the PRESS sequences of Philips, Siemens and GE had temporal maxima of 77%, 85% and 81% at (TE1, TE2) = (32, 65), (30, 71), and (22, 82) ms, respectively, ignoring T2 relaxation effects. In-vitro and in-vivo experiments were carried out using TE = 97 ms for the convenience of individual sites, whose subecho times were (TE1, TE2) = (32, 65), (17, 80), and (26, 71) ms for Philips, Siemens and GE, respectively. The 2HG signal agreed well between phantom experiment and volume-localized simulation (Fig. 3a,b). The 2HG 2.25-ppm signals were large and narrow, with the H3 and H3’ resonances attenuated. For comparison, spectra calculated with hard pulses (without volume localization) were very different from phantom spectra (Fig. 3c). Finally we tested the TE = 97 ms PRESS of the three vendors in patients with IDH-mutated gliomas. In each case, a signal was clearly discernible at 2.25 ppm (Fig. 4). The in-vivo spectra were well reproduced by LCModel fits, resulting in negligible residuals at ~2.25 ppm (Residuals-1). When 2HG was removed from the basis set however, unfit residuals were observed at ~2.25 ppm in all three spectra (Residuals-2), indicating that 2HG was detected with high selectivity. The 2HG CRLB was less than 10% in all cases.


This study reports multi-site test of optimized long-TE approaches in the MR scanners of three major vendors for detection of 2HG. Results show that the MRS methods have the capability of providing 2HG measurement with good precision in all three vendors. Since the RF pulses used are readily available in the vendor-supplied protocols, the data acquisition methods are easily transferable to other sites. The PRESS subecho times used for experiments were slightly different than suggested by simulations. This may not be problematic since the 2HG signals do not differ substantially between the simulation-suggested and experimental subecho time sets. Use of a proper basis is rather critical for reliable spectral analysis, as evidenced in Fig. 3. A major pitfall of this study is that, due to inconsistency of experimental conditions for reference water signal acquisition, the 2HG concentration is presented relative to the choline level, which is substantially altered in tumors. Future study requires standardization of the reference signal acquisition that will provide 2HG estimates directly comparable between vendors and sites. Lastly, the long-TE 2HG MRS can be easily extended to multi-voxel imaging with minimal interferences from macromolecules and lipids.


Data indicate that 2HG-optimized long-TE approaches perform well in the Philips, Siemens and GE MR systems, giving rise to narrowing of the 2HG 2.25-ppm multiplet and consequently conferring precise estimation of 2HG. The long-TE MRS methods may therefore have the great potential for non-invasive diagnosis/prognosis in IDH-mutated gliomas.


This research was supported by a Cancer Prevention Research Institute of Texas grant RP130427.


1. Dang L, White DW, Gross S, et al. Cancer-associated IDH1 mutations produce 2-hydroxyglutarate. Nature 2009;462:739-744.

2. Choi C, Ganji SK, DeBerardinis RJ, et al. 2-hydroxyglutarate detection by magnetic resonance spectroscopy in IDH-mutated patients with gliomas. Nat Med 2012;18:624-629.

3. Choi C, Ganji S, Hulsey K, et al. A comparative study of short- and long-TE 1H MRS at 3 T for in vivo detection of 2-hydroxyglutarate in brain tumors. NMR Biomed 2013;26:1242-1250.


FIG 1. Comparison of the envelopes and refocusing profiles between the PRESS refocusing RF pulses available in Philips, Siemens, and GE 3T MR scanners. The refocusing profile was obtained with a 2-step phase cycling (0 and 90°) in the Bloch simulations. Note that the flip angle is 137° for GE.

FIG 2. The amplitude of the numerically-calculated 2HG 2.25-ppm peak, normalized to that at TE=24 ms, was mapped versus TE1 and TE2 of the PRESS sequences for three vendors. A cross symbol (+) indicates a (TE1, TE2) pair, at which the 2HG signal is temporally maximum for TE > 50ms.

FIG 3. Phantom and calculated spectra of 2HG at selected (TE1, TE2) pairs for Philips, Siemens, and GE 3T scanners. Spectra were calculated using the PRESS volume-localization RF and gradient pulses (Calculation-1) and using 1-ns (non-localizing) 90° and 180° RF pulses (Calculation-2). Spectra were broadened to singlet linewidth of 5Hz.

FIG 4. In-vivo spectra from IDH-mutated glioma patients, obtained in three vendors (TR=2s; TE=97ms), are shown together with LCModel fit, residuals and metabolite signals. Residuals-1 and Residuals-2 were obtained from fittings with and without 2HG in the basis set, respectively. The numbers are metabolite/tCho ratios and CRLBs of the metabolites.

Proc. Intl. Soc. Mag. Reson. Med. 24 (2016)