Improvements in the MR acquisition hardware coupled to the growing availability of ultra-high field magnets, are enabling the acquisition of in vivo MR spectra without water suppression.^1,2 Moreover, the use of selective spectral excitations allows one to exploit the abundant reservoir of magnetically unperturbed spins, to enhance the longitudinal recovery of the excited metabolic or macromolecular sites.^3,4 Several of the peaks resonating downfield from the water resonance originate from exchangeable protons of -NH and -OH groups present in metabolites⁵, and become visible only when avoiding water suppression. Previous studies have detected downfield histidine, homocarnosine and phenylalanine peaks in human and mouse brain⁶; glucose (Glc), glutathione (GSH) and NAD+ were also unambiguously assigned.^7,8,9 Based on chemical shifts databases, attempts have been made to assign resonances observed in the 6-9 ppm region to additional metabolites such as adenosine triphosphate (ATP), N-acetyl aspartate (NAA), glutamine (Gln), histamine, tyrosine and tryptophan.¹⁰ This paper explores the use of Relaxation Enhanced (RE) MRS techniques using selective pulses to avoid water suppression and to record in vivo localized spectra from the downfield ¹H region at 21.1 T. The study was carried out on rats subjected to cerberal ischemia or glioblastoma.

Brain ischemia model. To mimic brain ischemia, middle cerebral artery occlusion (MCAO) was performed on six Sprague-Dawley rats for 1.5 h followed by re-perfusion. The animals were imaged 24 h following the occlusion.

Glioblastoma model. 9L glioma rat cells were cultured and expanded in DMEM+10%FBS. 100,000 9L glioma rat cells were injected at 2 mm anterior, 2.5 mm lateral and 3.5 mm deep with respect to Bregma. Five male Sprague-Dawley rats were used.

MRS acquisitions. All experiments were performed at the NHMFL using an ultra-wide bore 21.1-T vertical magnet, a Bruker Avance III MRI console and a home-built probe incorporating a transmit/receive quadrature surface coil. in vivo ¹H RE-MRS experiments were performed using a localized spin-echo sequence (Figure 1) for which spectrally selective pulses were employed to excite and refocus the 5.5-9.5 ppm range. The excitation was performed using a 5.55-ms, 10-lobe sinc pulse, and a 4-ms refocusing Shinnar-LeRoux pulse. Spatial localization was carried out using 3D LASER¹¹ incorporating six consecutive 5-ms adiabatic 180° pulses leading to a minimum echo time of 55 ms. The upfield spectral region (0-4 ppm) also was explored by adjusting frequency offsets. 512 to 1024 averages were acquired (TR=1.5-2.5 s) from two different voxel localizations (Figure 2) with an average volume of 90 µL, centered on diseased and healthy regions.

Data quantification. The spectroscopic data were processed and quantified with customized MATLAB software; the quantification stage relied on a GAMMA-library-based algorithm¹² originally developed for in vivo 2D MRS time-domain quantification.¹³ This fitting model incorporates a predefined set of 1D metabolic spectral traces with known frequency shifts, keeping concentrations and linewidths as adjustable parameters. As most of the potential metabolites contributing to the downfield spectral region have too low a concentration to be detectable (≤100 µM), the metabolite basis set used here contained only ATP, Gln, GSH and NAA--all of these present at ≥1 mM concentrations--and four broad gaussian resonances to model the spectral baseline.

in vivo downfield ¹H spectra show rich metabolic information. Figure 3 compares upfield and downfield spectral regions acquired on normal and diseased tissues. An average SNR of 40 was calculated on the 7.8-ppm peak. The result of a typical spectral quantification is displayed in Figure 4; this procedure was applied to all spectra.Concentrations appear to be globally lower in ischemic and glioma tissue than in normal brain tissue for all metabolites (Figure 5). Exceptions in this trend are given by (i) GSH, whose 23% increase in ischemic tissue could reflect a neuroprotective measure against oxidative stress; and (ii) Cho, whose 30% increase in glioma tissue is probably a consequence of tumor cell proliferation. Moreover, a 5.9-ppm peak shows a marked, statistically significant increase (45%) in the tumor tissue. Previous NMR studies on cells^14,15 assigned this resonance to UDP-NAc, which is known to be an abundant metabolite upon cancer proliferation.By avoiding water excitation, the RE-MRS sequence led to high quality, spectra with well-resolved resonances and low spectral baseline distortions--not only for the classical upfield peaks but also for the more elusive downfield resonances. The method’s main limitation is its relatively long echo time, a consequence of B₁ power limitations. This potential bias notwithstanding, led to quantifiable ¹H downfield MR spectra that highlighted systematic differences between healthy and non-healthy tissues, which could serve as potential biomarkers as well as providing better understanding of disease development.