In vivo magnetic resonance spectroscopy of human brain: the biophysical basis of dementia

Nuclear magnetic resonance spectroscopy (MRS) in low and medium magnetic fields yields well-resolved natural abundance proton and decoupled phosphorus spectra from small (1-10 cc) volumes of brain in vivo in minutes. With this tool, neurochemical research has advanced through identification and non-invasive assay of specific neuronal--(N-acetylaspartate), glial (myo-inositol)--markers, energetics and osmolytes, and neurotransmitters (glutamate, GABA). From these simple measurements, several dozen disease states are recognized, including birth injury, and white matter and Alzheimer disease. Addition of stable isotopes of carbon (in man) or nitrogen (in experimental animals) has provided in vivo assays of enzyme flux through glucose transport, glycolysis, TCA-cycle, and the glutamine-glutamate-GABA system. Finally, a number of xenobiotics are recognized with heteronuclear NMR techniques. Together, these tools are having a major impact on neuroscience and clinical medicine. Through diagnosis and therapeutic monitoring, a new generation of in vivo metabolite imaging is expected with the advent of conforming RF coils and higher field NMR systems.

In Vivo Magnetic Resonance Spectroscopy (MRS) of the Activated Human Brain at 7T

Benoît Schaller

The principal aim of this thesis was to investigate human brain energy metabolism during physiological activation using proton (1H) magnetic resonance spectroscopy (MRS) at 7 Tesla (T). High magnetic fields (≥ 7 T) provide several advantages in human brain imaging and spectroscopy. In particular, signal-to-noise ratio (SNR) and chemical shift dispersion increase at higher magnetic fields, allowing for improved accuracy and precision of metabolite quantification. To take profit of the benefits of MRS at high field, one needs to provide an advanced methodology including hardware, pulse sequences experimental setup and data analysis. Each of these aspects was thoroughly explored in this thesis. In the field of functional studies, functional MRS (fMRS) has recently been used as a research tool to investigate the neuroenergetic and metabolic basis of physiologic brain activation. It provides direct insight into brain metabolism by non-invasively determining concentrations of metabolites. This is critical for the basic understanding of overall brain function, and potentially of the pathogenesis of many neurodegenerative diseases. In general, an optimized MR system and accurate acquisition methodology are required for high sensitivity and reliable metabolite quantification. The metabolites of interest for fMRS studies (e.g. lactate, glucose) are present in low concentration and metabolite changes are small under activation. Hence, studies of metabolite changes, in particular lactate, have led to inconsistent reports in the literature over the last decades. To address these challenges, it was essential to develop a robust MR protocol for the quantitative measurement of metabolite changes. A fMRS study was performed to investigate metabolite changes during visual stimulation using the enhanced sensitivity of the SPin ECho full Intensity Acquired Localized (SPECIAL) sequence. Small but significant increases of lactate (19 ± 4 %, P < 0.05) and glutamate (4 ± 1 %, P < 0.001) were observed using a small number of subjects (n = 6). With the exception of glucose (12 ± 5 %, P < 0.001), no other significant metabolite concentration changes beyond experimental error were observed. Based on this successful fMRS study, we further investigated brain energy metabolism. A subsequent fMRS study was performed to determine metabolite changes occurring during motor activation in the human brain. This second study demonstrated that increases in lactate (17 ± 5 %, P < 0.001) and glutamate (2 ± 1 %, P < 0.005) during motor stimulation were small, but similar to those observed during visual stimulation. These metabolite changes were further analyzed, and they supported the hypothesis of an increase in the change of cerebral metabolic rate of oxygen, ∆CMRO2, that is transiently lower than that of glucose, ∆CMRGlc, during the first one to two minutes of stimulation. Finally, we hypothesized that the observed glutamate and lactate increases were a general manifestation of the blood-oxygenation level dependent effect. The accuracy and the reliability of metabolite quantification is particularly challenging at short echo times due to the presence of broad underlying resonances overlapping with those of metabolite. Such resonances arise from macromolecules, which are characterized by short T1 and T2 relaxation times. Two studies were performed at 3 T and 7 T to investigate the characteristics of the macromolecule signal. An important finding was that the mathematical approximation (spline baseline) was a sufficient estimation of the macromolecule contribution to 1H spectra at 3 T compared to the experimentally measured macromolecule signal for healthy subjects. As a caveat, small, but significant, differences in the quantification may need to be taken into account when comparing metabolite concentrations obtained with these two different approaches, such as for glutamate and for γ-amino-butyric, which were more reliably quantified when using the measured macromolecule baseline. In addition, a study on the tissue-specificity of the macromolecule signal in the healthy human occipital lobe at 7 T reported that a general average macromolecule signal ensured reliable metabolite quantification. In general, these studies helped to set up a robust and reliable methodology in order to investigate human brain metabolism with improved accuracy. Despite their great potential for MRS, high main magnetic fields also compose challenges. In particular, the wavelength of the proton spin resonance becomes comparable to the size of a human head (at 7 T, λ = 11 cm). The phase variation across the brain produces signal addition (constructive interference, central bright spot) and signal cancellation (destructive interference, signal drop). Transceive arrays have been developed to address this issue. In this study, we designed, tested, and built an eight-channel transceive array, which was used to acquire images of the human head at 7 T (297 MHz). The design aimed for a low mutual coupling between the microstrip elements (> -15 dB), a relatively large penetration depth in the loading sample, reduced radiation losses and load-invariant tuning and matching. Electromagnetic simulations were run using Microwave Studio (CST, Darmstadt, Germany) for design optimization. The model was further demonstrated with the construction of the transceive array. Finally, the resulting array coil was shown to be handling, stable and also suited for techniques in which the transmission signals from all channels with different amplitudes and phases are combined. To conclude, this thesis provided complementary and new metabolic information about cerebral energy metabolism during physiological activation. Furthermore, the macromolecule studies and the building of the transceive array allowed to improve the accuracy and sensitivity of NMR measurements.

EPFL2013

In vivo magnetic resonance studies of glycogen and hyperpolarized lithium in the rat brain

Ruud van Heeswijk

Nuclear magnetic resonance (NMR) is used for a large array of applications, ranging from chemical characterization to oil drilling to medical imaging. In these fields NMR is used as an investigational tool, but new techniques and applications are continuously being developed as well. The latter is also the subject of this thesis. After an introduction to NMR theory, it focuses on the development of new methodologies for two separate areas of investigation: the metabolism of brain glycogen and hyperpolarization via dynamic nuclear polarization (DNP). Brain glycogen is the main storage form of glucose in the brain. However, because it is present in much lower concentrations in the brain than in other tissues, the significance of this storage and other roles are subject to debate. To better understand the roles of brain glycogen, it is important to quantify its metabolic parameters. The only currently available method to detect brain glycogen in vivo is 13C NMR spectroscopy. Incorporation of 13C-labeled glucose is necessary to allow glycogen measurement, but might be affected by changes in the turnover between glucose and glycogen. We therefore established a protocol to measure the glycogen absolute concentration in the rat brain by eliminating label turnover as a variable. The approach is based on establishing an increased, constant 13C isotopic enrichment (IE). 13C-glucose infusion is then performed at this IE of brain glycogen. As glycogen IE cannot be assessed in vivo, we validated that it can be inferred from the N-acetyl-aspartate (NAA) IE in vivo: after [1-13C]-glucose ingestion, glycogen IE was 2.2 ± 0.1 times that of NAA. Glycogen concentration measured in vivo by 13C NMR (5.8 ± 0.7 µmol/g) was in excellent agreement with post-mortem biochemistry (6.4 ± 0.6 µmol/g). A second glycogen NMR protocol was then implemented to measure concentration and turnover simultaneously. After reaching isotopic steady state for glycogen C1 using [1-13C]-glucose administration, [1,6-13C2]-glucose was infused such that isotopic steady state was maintained at the C1 position, but the C6 position reflected 13C label incorporation. To overcome the large chemical shift displacement error between the C1 and C6 resonances of glycogen, 2D gradient-based localization using the Fourier series window approach was implemented. The glycogen concentration of 5.1 ± 1.6 µmol/g measured from the C1 position was in excellent agreement with concomitant biochemical determinations, while glycogen turnover measured from the rate of label incorporation into the C6 position of glycogen in the α-chloralose anesthetized rat was 0.7 µmol/g/h. The contrast agent Omniscan was added to a formic acid-filled glass reference sphere in the abovementioned protocols in order to shorten its T1 relaxation time and to accelerate the calibration procedure it is used in; therefore it was established that Omniscan in pure 13C formic acid has a relaxivity of 2.9 mM-1s-1. The study was expanded, and the relaxivity of several commercially available gadolinium (Gd)-based contrast agents was determined for X-nuclei resonances with long intrinsic relaxation times. The relaxivity of Omniscan on the glutamate C1 and C5 in aqueous solution was ~ 0.5 mM-1s-1. These relaxivities allow the preparation of solutions with a predetermined short 13C T1. Another isotope with a long relaxation time is lithium-6 (6Li). Lithium is an ion that is widely used in psychotherapy, and 6Li has an intrinsic relaxation time on the order of several minutes, which was strongly affected by the contrast agents. Its relaxivity ranged from ~ 0.1 mM-1s-1 for Omniscan to 0.3 for Magnevist, whereas the relaxivity of Gd-DOTP was 11 mM-1s-1. Overall, these experiments suggested that the presence of submicromolar contrast agent concentrations should be detectable, provided sufficient sensitivity is available, such as that afforded by hyperpolarization. This submicromolar contrast agent concentration detection was demonstrated in a subsequent study: first it was established that lithium-6 can be readily hyperpolarized through DNP within 30 min while retaining a long polarization relaxation time on the order of a minute. The shortening of this relaxation time was used for the detection of 500 nM contrast agent in vitro. Hyperpolarized lithium-6 was then administered to the rat, where its signal retained a relaxation time on the order of 70 s in vivo. Localization experiments implied that the lithium signal originated from within the brain and that it was detectable up to 5 min after administration. Next, ~ 1.5 µM of contrast agent was injected in the rat, and its effect on previously injected hyperpolarized lithium-6 was successfully quantified. It was concluded that the detection of submicromolar contrast agent concentrations using hyperpolarized NMR nuclei such as 6Li may provide a novel avenue for molecular imaging. In a separate set of of experiments, hyperpolarization through DNP was applied to 15N-labeled choline, 13C-labeled bicarbonate and 13C-labeled acetate. All three metabolites were successfully polarized and were detected both in vitro and in vivo.

EPFL2009

Social isolation stress and chronic glutathione deficiency have a common effect on the glutamine-to-glutamate ratio and myo-inositol concentration in the mouse frontal cortex

Alberto Corcoba, João Miguel das Neves Duarte, Rolf Gruetter

Environmental stress can interact with genetic predisposition to increase the risk of developing psychopathology. In this work, we tested the hypothesis that social isolation stress interacts with impaired glutathione synthesis and have cumulative effects on the neurochemical profile of the frontal cortex. A mouse model with chronic glutathione deficit induced by knockout (-/-) of the glutamate-cysteine ligase modulatory subunit (Gclm) was exposed to social isolation stress from weaning to post-natal day 65. Using magnetic resonance methods at high-field (14.1 T), we analysed the neurochemical profile in the frontal cortex, brain size and ventricular volume of adult animals. Glutathione deficit was accompanied by elevated concentrations of N-acetylaspartate, alanine, and glutamine, as well as the ratio of glutamine to glutamate (Gln/Glu), and by a reduction in levels of myo-inositol and choline-containing compounds in the frontal cortex of -/- animals with respect to wild-type littermates. Although there was no significant interaction between social isolation stress and glutathione deficiency, mice reared in isolation displayed lower myo-inositol concentration (-8.4%, P