4.2. Proton Magnetic Resonance Spectroscopy (1H-MRS) Acquisition

Proton magnetic resonance spectroscopy (¹H-MRS) is a non-invasive imaging method that provides spectroscopic information in addition to the image that is generated by magnetic resonance imaging MRI alone. The spectroscopic information obtained in a ¹H-MRS study can be used to infer further information about cellular activity (metabolic information).

The methodology used in the development of this study was ordered according to the importance of the studies carried out, which led us to the reported results. For the first, the consistency of ¹H-MRS was confirmed by detecting the metabolites present in the brain. The drawbacks in the application of ¹H-MRS in intact tissues ‘in vivo’ have been reported as an inherent limitation of spectroscopy is the narrow range (5 ppm) in the chemical shift of non-exchangeable protons, where the resonance of a large number of metabolites overlaps. Therefore, it was necessary to fine-tune the technique before proceeding to developing the paradigm contained in the methodology used to quantify the detectable metabolites.

The absolute concentration of the metabolites present in each voxel was calculated, in the spectra resolved in time, that is, 68 spectra for each stimulus observed (i.e., 272 spectra). The presentation of each stimulus lasted 145.52 s (582.08 s in total); TR = 1070 msg and NEX = 2. The quantification was performed with the LCModel program, and the Tukey’s multiple comparisons test was applied in the statistical analysis. Furthermore, to optimize the acquisition of the spectra, it was necessary to modify several parameters to decrease the noise and artifacts of the spectroscopic signal, the most relevant parameters being the TE and the TR, with the former being the most relevant; currently, the TE used ‘in vivo’ by most groups varies between 18 and 288 ms. In this regard, we talk about studies with a short or long TE, using most studies with short TEs (between 18 and 45 ms) and studies with long TEs (between 120 and 288 ms), adducing different arguments in favor and against each option. On the contrary, in short TEs, a greater number of resonances are visible because the signal of compounds with strong modulation can be lost at long TEs. Thus, a short TE is necessary for a better evaluation of some compounds such as myoinositol, glutamine, and glutamate. However, there is no unanimous decision on TE to be used in clinical diagnosis [71,72].

The spectra were obtained by varying both the TE and TR values, which allowed us to define which were the best times for the spectra obtained to show the highest number of metabolites, based on the reliability indicators or lower levels of Cramér–Rao (<20%). Due to the characteristics and limitations of the area of interest, a voxel with a volume of 2 × 2 × 2 cm³ (8 mL) was used to obtain the average of the signals, which allowed us to elucidate spectra with optimal signal/noise ratios. Likewise, the acquisition time of the spectra (TE = 23 ms; TR = 1070 ms) was necessary for a better assessment of some compounds such as myoinositol or glutamate, according to the interest of our study.

In this way, the paradigm that was used throughout the research trajectory was established, considering also in the post-processing the main peaks of interest and their corresponding metabolites processed by the LCModel, taking in hand only the metabolites with a Cramér–Rao level [73,74] less than 20% (blue color). Finally, the paradigm used in all of the spectroscopy studies was as follows: TR = 1070 ms, TE = 23 ms, longitudinal magnetization rotation angle (flip angle) = 90°, matrix size = 256 × 256, NEX = 2, R¹ = 13%, R² = 30% and TG = 135 in a 2 × 2 × 2 cm³ voxel (8 mL). The data of all of the studies carried out and their results are available in the database of our Neurochemistry and Neuroimaging Laboratory.

All subjects underwent an MRI and ¹H-MRS using a 3T Signa-HD MR scanner (GE Healthcare, Waukesha, WI, USA). T2-weighted images were used for positioning the volumes of interest (VOIs). The single voxel acquisition used a spin-echo sequence recorded within the following parameters: TE = 23 ms, TR = 1070 ms, NEX = 2, flip angle = 90°, and 256 acquisitions with the point-resolved spectroscopy (PRESS) technique. During data acquisition, the same experienced neuroradiologist, blind to the clinical data, placed the voxels (2 × 2 × 2 cm³) at the ACC and PCC. The main metabolite resonances were limited to 2.02 ppm for NAA, 2.04 ppm for NAAG, 3.03 ppm for Cr, 3.20 ppm for Cho, 3.55 ppm for mI, and, 3.77 ppm for Glu. We used a TE of 23 ms, as it is known that myoinositol can be readily detected in a short TE using ¹H-MRS spectra of the brain due to its high concentration of (4–8) mM [71]. Each voxel was positioned so as to exclude contamination of signal from the skull and subcutaneous fat. Morphological examination enabled us to exclude other pathologies, such as congenital abnormalities, lesions in cerebral palsy, tumors, and hydrocephalus.

Contradictory, the signal from the neuromodulator NAAG and that of the amino acid NAA are considered difficult to separate in spectroscopy [75]; however, some works, using the construction of the phase space of the particles developed in our research group, revealed that it is possible to separate them due to NAAG having four visible attraction regions in the phase diagram space, considering the resonating protons of the acetyl-CH3 groups (2.04 ppm). Therefore, it is easy to deduce that the short- and long-term variability of the resonance process of NAAG is greater than that of NAA (2.02 ppm) and Glu (2.10 ppm) [76]. The paradigm used to obtain the spectroscopy in our studies allowed us to detect this neuromodulator with a %SD of <20%. Moreover, it was previously shown that NAAG and NAA can be discriminated through appropriate application of the MEGA-PRESS method by in vivo MRS at 3 Tesla (3T) [76].

Although an intense peak at 2 ppm is generally assigned to NAA (which is responsible for the greater part of the signal), in this work, it was assumed to correspond to NAA and NAAG. However, any small N-acetyl molecules in the brain will contribute to the peak. Furthermore, it was possible to measure NAAG at short echo times, suppressing the signals of multiplets from strongly coupled spin systems near 2 ppm, thus minimizing the interfering signals for detecting the acetyl proton signal of NAAG. One possible weakness of our method is the reliance on accurate suppression of the NAAG signal in the NAA scan, and particularly the NAA signal in the NAAG scan (due to the higher concentration of NAA). However, small contributions from other N-Acetyl species (e.g., N-Acetyl-glutamate) could result in overestimation of the NAA and NAAG concentrations.

Glutamine and glutamate (i.e., Glx and Glu/Gln) create a series of signals that are grouped into two regions (2.1–2.5: multiplets; 3.6–3.8: triplets). Some works suggest that a detailed analysis of the region allows to determine the contribution of each of the two components, but most study the entire region together. Recent studies with animal models in vitro suggest the existence of two pools of Glu and Gln related to the neuronal and glial compartments. Finally, it should also be noted that various studies suggest considering Gln as a glial marker.