Preserved Edge Convolutional Neural Network for Sensitivity Enhancement of Deuterium Metabolic Imaging (DMI).

PURPOSE: Common to most MRSI techniques, the spatial resolution and the minimal scan duration of Deuterium Metabolic Imaging (DMI) are limited by the achievable SNR. This work presents a deep learning method for sensitivity enhancement of DMI. METHODS: A convolutional neural network (CNN) was designed to estimate the 2H-labeled metabolite concentrations from low SNR and distorted DMI FIDs. The CNN was trained with synthetic data that represent a range of SNR levels typically encountered in vivo. The estimation precision was further improved by fine-tuning the CNN with MRI-based edge-preserving regularization for each DMI dataset. The proposed processing method, PReserved Edge ConvolutIonal neural network for Sensitivity Enhanced DMI (PRECISE-DMI), was applied to simulation studies and in vivo experiments to evaluate the anticipated improvements in SNR and investigate the potential for inaccuracies. RESULTS: PRECISE-DMI visually improved the metabolic maps of low SNR datasets, and quantitatively provided higher precision than the standard Fourier reconstruction. Processing of DMI data acquired in rat brain tumor models resulted in more precise determination of 2H-labeled lactate and glutamate + glutamine levels, at increased spatial resolution (from >8 to 2 $\mu$L) or shortened scan time (from 32 to 4 min) compared to standard acquisitions. However, rigorous SD-bias analyses showed that overuse of the edge-preserving regularization can compromise the accuracy of the results. CONCLUSION: PRECISE-DMI allows a flexible trade-off between enhancing the sensitivity of DMI and minimizing the inaccuracies. With typical settings, the DMI sensitivity can be improved by 3-fold while retaining the capability to detect local signal variations.

Mapping of exogenous choline uptake and metabolism in rat glioblastoma using deuterium metabolic imaging (DMI).

Introduction: There is a lack of robust metabolic imaging techniques that can be routinely applied to characterize lesions in patients with brain tumors. Here we explore in an animal model of glioblastoma the feasibility to detect uptake and metabolism of deuterated choline and describe the tumor-to-brain image contrast. Methods: RG2 cells were incubated with choline and the level of intracellular choline and its metabolites measured in cell extracts using high resolution 1H NMR. In rats with orthotopically implanted RG2 tumors deuterium metabolic imaging (DMI) was applied in vivo during, as well as 1 day after, intravenous infusion of 2H9-choline. In parallel experiments, RG2-bearing rats were infused with [1,1',2,2'-2H4]-choline and tissue metabolite extracts analyzed with high resolution 2H NMR to identify molecule-specific 2H-labeling in choline and its metabolites. Results: In vitro experiments indicated high uptake and fast phosphorylation of exogenous choline in RG2 cells. In vivo DMI studies revealed a high signal from the 2H-labeled pool of choline + metabolites (total choline, 2H-tCho) in the tumor lesion but not in normal brain. Quantitative DMI-based metabolic maps of 2H-tCho showed high tumor-to-brain image contrast in maps acquired both during, and 24 h after deuterated choline infusion. High resolution 2H NMR revealed that DMI data acquired during 2H-choline infusion consists of free choline and phosphocholine, while the data acquired 24 h later represent phosphocholine and glycerophosphocholine. Discussion: Uptake and metabolism of exogenous choline was high in RG2 tumors compared to normal brain, resulting in high tumor-to-brain image contrast on DMI-based metabolic maps. By varying the timing of DMI data acquisition relative to the start of the deuterated choline infusion, the metabolic maps can be weighted toward detection of choline uptake or choline metabolism. These proof-of-principle experiments highlight the potential of using deuterated choline combined with DMI to metabolically characterize brain tumors.

Preclinical evaluation of a brain penetrant PARP PET imaging probe in rat glioblastoma and nonhuman primates.

PURPOSE: Currently, there are multiple active clinical trials involving poly(ADP-ribose) polymerase (PARP) inhibitors in the treatment of glioblastoma. The noninvasive quantification of baseline PARP expression using positron emission tomography (PET) may provide prognostic information and lead to more precise treatment. Due to the lack of brain-penetrant PARP imaging agents, the reliable and accurate in vivo quantification of PARP in the brain remains elusive. Herein, we report the synthesis of a brain-penetrant PARP PET tracer, (R)-2-(2-methyl-1-(methyl-11C)pyrrolidin-2-yl)-1H-benzo[d]imidazole-4-carboxamide ([11C]PyBic), and its preclinical evaluations in a syngeneic RG2 rat glioblastoma model and healthy nonhuman primates. METHODS: We synthesized [11C]PyBic using veliparib as the labeling precursor, performed dynamic PET scans on RG2 tumor-bearing rats and calculated the distribution volume ratio (DVR) using simplified reference region method 2 (SRTM2) with the contralateral nontumor brain region as the reference region. We performed biodistribution studies, western blot, and immunostaining studies to validate the in vivo PET quantification results. We characterized the brain kinetics and binding specificity of [11C]PyBic in nonhuman primates on FOCUS220 scanner and calculated the volume of distribution (VT), nondisplaceable volume of distribution (VND), and nondisplaceable binding potential (BPND) in selected brain regions. RESULTS: [11C]PyBic was synthesized efficiently in one step, with greater than 97% radiochemical and chemical purity and molar activity of 148 ± 85 MBq/nmol (n = 6). [11C]PyBic demonstrated PARP-specific binding in RG2 tumors, with 74% of tracer binding in tumors blocked by preinjected veliparib (i.v., 5 mg/kg). The in vivo PET imaging results were corroborated by ex vivo biodistribution, PARP1 immunohistochemistry and immunoblotting data. Furthermore, brain penetration of [11C]PyBic was confirmed by quantitative monkey brain PET, which showed high specific uptake (BPND > 3) and low nonspecific uptake (VND < 3 mL/cm3) in the monkey brain. CONCLUSION: [11C]PyBic is the first brain-penetrant PARP PET tracer validated in a rat glioblastoma model and healthy nonhuman primates. The brain kinetics of [11C]PyBic are suitable for noninvasive quantification of available PARP binding in the brain, which posits [11C]PyBic to have broad applications in oncology and neuroimaging.

NMR visibility of deuterium-labeled liver glycogen in vivo.

PURPOSE: Deuterium metabolic imaging (DMI) combined with [6,6'-2 H2 ]-glucose has the potential to detect glycogen synthesis in the liver. However, the similar chemical shifts of [6,6'-2 H2 ]-glucose and [6,6'-2 H2 ]-glycogen in the 2 H NMR spectrum make unambiguous detection and separation difficult in vivo, in contrast to comparable approaches using 13 C MRS. Here the NMR visibility of 2 H-labeled glycogen is investigated to better understand its potential contribution to the observed signal in liver following administration of [6,6'-2 H2 ]-glucose. METHODS: Mice were provided drinking water containing 2 H-labeled glucose. High-resolution NMR analyses was performed of isolated liver glycogen in solution, before and after the addition of the glucose-releasing enzyme amyloglucosidase. RESULTS: 2 H-labeled glycogen was barely detectable in solution using 2 H NMR because of the very short T2 (<2 ms) of 2 H-labeled glycogen, giving a spectral line width that is more than five times as broad as that of 13 C-labeled glycogen (T2 = ~10 ms). CONCLUSION: 2 H-labeled glycogen is not detectable with 2 H MRS(I) under in vivo conditions, leaving 13 C MRS as the preferred technique for in vivo detection of glycogen.

Characterization of Kinetic Isotope Effects and Label Loss in Deuterium-Based Isotopic Labeling Studies.

Deuterium metabolic imaging (DMI) is a novel, 3D, magnetic resonance (MR)-based method to map metabolism of deuterated substrates in vivo. The replacement of protons with deuterons could potentially lead to kinetic isotope effects (KIEs) in which metabolic rates of deuterated substrates are reduced due to the presence of a heavier isotope. Knowledge of the extent of KIE in vivo and 2H label loss due to exchange reactions is required for DMI-based measurements of absolute metabolic rates. Here the deuterium KIE and label loss in vivo are investigated for glucose and acetate using a double substrate/double labeling strategy and 1H-decoupled 13C NMR in rat glioma cells and rat brain tissue metabolite extracts. The unique spectral patterns due to extensive 2H-13C and 13C-13C scalar couplings allow the identification of all possible metabolic products. The 2H label loss observed in lactate, glutamate, and glutamine of rat brain was 15.7 ± 2.6, 37.9 ± 1.1, and 41.5 ± 5.2% when using [6,6-2H2]-glucose as the metabolic substrate. For [2-2H3]-acetate, the 2H label loss in glutamate and glutamine was 14.4 ± 3.4 and 13.6 ± 2.2%, respectively, in excellent agreement with predicted values. Steady-state 2H label accumulation in the C4 position of glutamate and glutamine was contrasted by the absence of label accumulation in the C2 or C3 positions, indicating that during a full turn of the tricarboxylic acid cycle all 2H label is lost. The measured KIE was relatively small (4-6%) for both substrates and all measured metabolic products. These results pave the way for further development of quantitative DMI studies to generate metabolic flux maps in vivo.

¹H-[¹³C]-nuclear magnetic resonance spectroscopy measures of ketamine's effect on amino acid neurotransmitter metabolism.

Ketamine has recently gained significant attention owing to its psychotomimetic and more recently discovered rapid antidepressant-like properties. ¹H-[¹³C]-nuclear magnetic resonance studies were employed to explore potential physiological processes underlying these unique effects. [1-¹³C]glucose and [2-¹³C]acetate-nuclear magnetic resonance ex vivo studies were performed on the medial prefrontal cortex (mPFC) and hippocampus of rats acutely treated with 30 mg/kg or 80 mg/kg ketamine and compared with saline-treated animals to determine the effects of ketamine on amino acid neurotransmitter cycling and glial metabolism. A subanesthetic, but not anesthetic, dose of ketamine significantly increased the percentage of ¹³C-enrichments of glutamate, γ-aminobutyric acid, and glutamine in the mPFC of rats. Subanesthetic doses of ketamine increased mPFC amino acid neurotransmitter cycling, as well as neuronal and glial energy metabolism. These data add to previous reports suggesting increased mPFC levels of glutamate release, following the administration of subanesthetic doses of ketamine, are related to the drug's acute effects on cognition, perception, and mood.