Neurotransmitter Imaging of the Human Brain: Detecting ?-Aminobutyric Acid (GABA) Using Magnetic Resonance Spectroscopy

Materials and Methods: In total, 164 participants were included and 272 MRS measurements were performed with the voxel placed in the medial prefrontal cortex (mPFC, 136), thalamus (32), and cerebellum (104) using two different but “identical” MR systems. Nineteen narcolepsy patients and 21 matched healthy controls performed an fMRI working memory task using a simultaneous EEG followed by an mPFC GABA-edited MRS measurement. Sixty-four IBS patients and 32 matched healthy controls underwent an mPFC GABA-edited MRS measurement followed by resting state fMRI. In addition, psychological symptoms were assessed using questionnaires. Ten ET patients and six matched healthy controls underwent four GABA-edited MRS measurements with the voxels placed in the thalamus and cerebellum. In this study, the symptom severity was investigated using the essential tremor rating scale (ETRS). All clinical MRS datasets were analyzed using conventional methods for post-processing and quantification. Furthermore, 12 volunteers were recruited for the purpose of investigating statistical retrospective approaches for artifact detection and elimination of MRS data. Each participant underwent three reference measurements and three measurements with induced head movements conducted according to a movement paradigm. Order statistic filtering (OSF) and jackknife correlation (JKC) were investigated as regards to the elimination of artifact-influenced spectra and reliability of the resulting concentrations. Finally, phantom measurements were performed for the purpose of investigating MEGA-edited MRSI.

Results: In narcolepsy, a trend-level association was observed between the mPFC MRS concentrations and increased deactivation within the default mode network during the working memory task. A significantly higher mPFC GABA+ concentration was observed in IBS patients with a high severity of comorbid anxiety. In ET, a positive correlation was observed between cerebellar GABA+/Glx ratios and tremor severity. Moreover, movements during the measurement decreased the concentration estimates due to signal loss in the spectra. Both OSF and JKC resulted in trend-level improvement of the signal- intense metabolites in spectrum when artifacts were present in the data, while performing equally as well as conventional analysis methodology when no intentional movements were present in the data. Finally, using the fast MEGA- edited multi-voxel sequence developed for a conventional clinical scanner, our phantom measurements showed that GABA was detectable using a 1:45 min acquisition time and an MRSI voxel size of 1 mL.

Discussion: Several challenges such as time constraints, large voxel sizes, and protocol design were encountered when performing SVS MEGA-PRESS in the clinical research settings. In addition, artifacts in the MRS data originating for example, from motions, negatively impacted the resulting averaged spectra, which was evident in both data from clinical populations and healthy controls. In the presence of artifacts in the data, both OSF and JKC improved the SVS MEGA-edited spectra. In addition, the implemented JKC method can be used not only for artifact detection, but also as a generally applicable retrospective technique for the quality control of a dataset, or as an indication of whether a shift in voxel placement occurred during the measurement. Using the MEGA-edited MRSI pulse sequence, there are many technical challenges, including detrimental effects from eddy currents, spurious echoes, and field inhomogeneities. Even though there are many technical challenges when using MEGA-edited MRSI, an optimized version of the MRSI sequence would be extremely valuable in clinical research applications where high spatial resolution and short acquisition times are highly desired.

Conclusions: OSF and JKC improved the metabolite quantification when artifacts were present in the data, and JKC was preferable. Although there are many technical challenges, MEGA-edited MRSI with full brain coverage in combination with a minimal voxel size for the purpose of GABA detection, would be extremely useful in clinical research applications where disorders such as narcolepsy, IBS, or ET, are investigated.