To optimize the MEGA-PRESS pulse sequence for the detection of the oncometabolite, 2-Hydroxyglutarate (2HG).2HG is formed in some brain tumors as a result of mutation of isocitrate dehydrogenase (IDH), and has important implications for tumor classification and prognosis. 2HG is present at very low levels in healthy brain tissue, but may be present in the mM range in tumors and is therefore potentially detectable in vivo using magnetic resonance spectroscopy (MRS). Several different MRS approaches to measuring 2HG have been proposed, including direct detection in conventional PRESS spectra using appropriate spectral fitting routines [1] , or by using J-difference editing to remove overlying signals and selectively detect 2HG. In this abstract, optimum conditions for 2HG spectral-editing are investigated using density-matrix simulations and phantom experiments. 2HG, whose structure is shown in Figure 1, has a 1H spectrum with signals at 4.01 ppm (CH), 2.2 ppm (CH2) and 1.9 ppm (CH2). In theory, editing could be performed either by applying frequency-selective pulses at 4.01 ppm and detecting difference-edited signal at 1.9 ppm, or vice versa. Detection of the 1.9 ppm signal is boosted by representing two protons, but hindered by the unpredictable evolution of (strong) coupling to the signal at 2.2 ppm.

Simulations: MEGA-PRESS simulations were performed for TEs 70 ms-240 ms in 10 ms increments with parameters: simulated B0 3T; 10 ms sinc-Gaussian editing pulses (comparing ON at 1.9 ppm to ON at 4.01 ppm); ‘GTST’ slice-selective refocusing pulses; 2 kHz spectral width; 2048 points; 8-Hertz exponential line-broadening; zero-filling to 8192 datapoints. The first PRESS spin-echo duration (TE1) was fixed at 13.4 ms.

Phantom experiments: Experiments with matched parameters to the above simulations were performed on a 26 mM 2HG phantom using a Philips Achieva 3T scanner using 20 Hz exponential line-broadening. The full TE series was acquired with 14-ms editing pulses applied at 1.9 ppm. Phantom experiments were also performed at a TE of 100 ms using a range of slice-selective refocusing pulses including sinc-Gaussian (bandwidth 800 Hz), GTST (bandwidth 1.3 kHz) and frequency-modulated (bandwidth 2.2 kHz), with editing pulses applied at 1.9 ppm.

Comparing the two simulated TE-series in Figure 2a, it can be seen that the strong coupling between spins at 1.9 ppm and 2.2 ppm results in complex behavior with editing of these signals when editing pulses are applied at 4.02 ppm. In contrast, the 4.02 ppm signal follows a more conventional TE-dependence. The integrals of these spectra (shown in Figure 2b) show that the maximal integral is similar in both; differences between the simulations and phantom experiments at longer TEs probably reflect the absence of any T2 relaxation in the simulations. Phantom data acquired with editing pulses at 1.9 ppm agree qualitatively with simulations, both in lineshape as shown in Figure 3a and TE-dependence as shown in Figure 3b. The benefits of high-bandwidth refocusing is apparent with GTST (high-bandwidth amplitude-modulated) refocusing pulses giving 47% more signal than sinc-Gaussian (medium-bandwidth amplitude-modulated) refocusing. High-bandwidth frequency modulated refocusing pulses give a further 26% improvement in signal yield.

Simulations show that, although detecting the single proton at 4.02 ppm is at a disadvantage compared to detecting the two 1.9 ppm protons, the maximal edited signal is similar in both cases. Since the integral plotted in Figure 3b represents both the 1.9 ppm and 2.3 ppm signals, the efficiency of editing of the 4.02 ppm signal is approximately four times higher. It is also noteworthy that the 4.02-detection timecourse has a maximum at a lower echo time (100 ms compared to 150 ms) which will be beneficial under in vivo conditions of more rapid transverse relaxation.

High-bandwidth refocusing pulses benefit MEGA-PRESS detection as they minimize signal losses associated with spatial inhomogeneity of coupling evolution [2]. In addition the high bandwidth refocusing pulses used in the current study have better slice profiles, which also contributes to the SNR improvement seen.

In conclusion, the MEGA-PRESS detection of 2HG is best performed at an echo time of approximately 100 ms (at least in a phantom; slightly shorted TE might be beneficial in vivo, depending on the in vivo 2HG T2 relaxation time) applying editing pulses to the 1.9 ppm spins to detect signal at 4.02 ppm, and employing high-bandwidth refocusing pulses.