Jullie W. Pan - US grants

DESCRIPTION (provided by applicant): Although substantial data exist on the pathophysiology of hypoglycemia, relatively less work has examined how human cerebral functional activation is modulated by hypoglycemia. However, developments in vivo MR spectroscopy and functional MRI provide important new avenues to evaluate the metabolic dynamics of functional activation. This application proposes to assess the (patho)physiology of functional activation in euglycemia and hypoglycemia (insulin induced vs. fasting induced), and to examine the effects from ketones. In particular, in this application we hypothesize that ketones can provide substrate for cerebral activation, and to that extent, can be evaluated through the dynamics of lactate generation and extent of fMRI activation. These studies will be performed in both normal control and type 1 diabetic subjects. This is an R21, rather than a RO1 application because although recent data from Amiel et al and Veneman et al have shown that ketones can improve neurological symptomatology and cognitive performance in hypoglycemia, we do not know how ketones contribute to the metabolic physiology of functional activation. In some aspects, the hypothesis of ketones being directly contributing to functional activity is risky, because of the models suggesting that glucose is an obligate fuel. However, much existing imaging data (PET and MR) can be consistent with the view that oxidation is an important component of cerebral activation. We believe that this proposal will provide data to determine the potential role for ketones in hypoglycemia both in normal subjects and in type 1 diabetes mellitus (T1 DM) since we believe that the problem of cerebral hypoglycemia in T1DM relates in part to how the brain is able (or not) to draw on alternate fuels in activation. Work from this proposal will provide new information that may significantly change the perception as to how the brain works, as well as provide impetus towards development of ketones as an option in the clinical management of hypoglycemia.

DESCRIPTION (provided by applicant): While challenges of SNR, hardware, and pulse sequence have limited the penetration of MRSI into clinical use, it remains among the most sensitive avenues towards assessing cerebral function and an important motivation for ongoing 7T development. However at any field strength, MRSI has challenges for spectral quality, acceptable acquisition time and spatial coverage. Specifically, while 3T MRSI has reported excellent SNR for NAA in supraventricular locations, there remain acknowledged problems for spectral quality in critical brain regions including the temporal and frontal lobes. 7T MRS has shown the expected doubling in SNR, which with the >2-fold greater spectral resolution effectively gives a total 16x reduction for scan time in comparison to 3T. However, problems at 7T focus on rf coil technology and B0 inhomogeneity. At 300MHz, the dielectric constant of tissue results in marked axial and longitudinal B1 inhomogeneities, simultaneous to a linear increase in required power for equivalent B1 generation. With a goal of developing and implementing MR spectroscopic imaging at 7T, our group has developed a transceiver detector which as used with RF shimming, has shown excellent performance at 7T. In collaboration with Resonance Research Inc., we have also shown that with higher order shim mapping and corrections, outstanding field homogeneity can be achieved over extended brain regions. Thus far this success has been primarily achieved over single slice regions. In this project, we will continue to develop this work for wide brain and multi-slice MRSI at 7T. This will be achieved through Aim 1 that extends the longitudinal coverage of the transceiver and further improves large volume Bo homogeneity, and Aim 2 which develops the pulse sequences (B1 based localization, Hadamard and SENSE encoding with the J-refocused acquisition), our goal being high SNR multi-slice spectroscopic imaging with low SAR (~2W/kg). Because methodologic development ideally occurs with real-world targets, we will test these developments with the challenging problem of neocortical epilepsy (NE). Since many NE patients are clinically complex, their evaluation commonly requires intracranial EEG (icEEG), a neurosurgical procedure where intracranial electrodes are used to localize seizures. For this process, it is clear that as much advanced knowledge on where to place electrodes is needed, so as to not miss the seizure onset zone. Yet even with this complex process, the post-surgical outcome is that ~40-50% of patients continue with significant seizures. With the variable etiologies in NE, there are major challenges for MRSI coverage (seizures can arise from any cortical location), volume resolution (typical size of ictal onset zone), and optimal metabolite pattern (is glutamate better than NAA). These unknowns likely explain why MRSI is not routinely used at 3T, but even in anatomically well defined medial temporal lobe epilepsy, there are spectral quality problems at 3T. In Aim 3, we will test the hypothesis that in regions of seizure onset and propagation (as defined by icEEG) the NAA/Cr and Glu/Cr will be abnormal, thus determining the typical voxel size needed for such identification, and whether NAA or glutamate may be more accurate. To bring this work into greater implementation, Aim 4 will take the parameters identified at 7T into a collaboration with O Gonen PhD, New York Univ., a leader in the development and application of 3T wide brain coverage MRSI. We will compare extended volume coverage MRSI at 3T and 7T in healthy controls and in a limited group of patients, allowing us to define the optimum methods at 3T to achieve identification of ictogenic regions. This project proposes a coordinated development in hardware and pulse sequences for 7T MRSI. We believe that this project's impact is broad, not just for improved neurosurgical management of NE, but also for improved imaging and MRSI at 3 and 7T. As stated, 3T MRSI, while successful for supra- ventricular regions, is inconsistent in the temporal lobes. This will improve with our proposed work in higher order shims and algorithms that optimally correct for and redistribute B0 homogeneity. At 7T, the transceiver work is critical as presently there is no clear solution to the problem of homogeneous and extended rf (~20uT) coverage. Thus while the impact of this project is clearly for 7T MRSI, the proposed work in B1 methods and B0 shimming will be highly relevant for many aspects of high field MR, both 7 and 3T.

DESCRIPTION (provided by applicant): Magnetic resonance spectroscopic imaging (MRSI) has been developed and used by many groups to localize the regions of injury and seizure onset in human epilepsy. Intrinsic to its success is the common finding of metabolic dysfunction even in the absence of tissue volume loss, e.g., as has been shown in MRI-negative cases of temporal lobe epilepsy. Although this successful detection is useful for subsequent targeted and surgical management of epilepsy, it has been recognized that the treatment of epilepsy could be substantively changed and improved if early detection were feasible and accurate. From both human and animal model experience, the pathophysiology of acquired epilepsy is initiated by a cerebral insult (e.g., the fairly common event of fever-induced or febrile seizures, head trauma or infection), followed by a latent period that precedes the onset of overt spontaneous recurrent seizures (SRS), i.e., epilepsy. If early detection were available, there are several strong potential therapeutics that might decrease the likelihood of developing clinical epilepsy (distinct from therapy that simply decreases the likelihood of seizures), e.g., erythropoietin, anti-inflammatory compounds, anti- oxidants. In this study we examine whether MRSI can detect incipient epilepsy during this latent period. We do this using the rat perforant path stimulation (PPS) model, which replicates many of the defining features of acquired human medial temporal lobe epilepsy. Importantly this model does not show widespread brain injury commonly seen in chemoconvulsant models. Our preliminary brain extract data from this 24hour PPS model show that by 4days after initial insult, substantial changes in myo-inositol, glutamine, GABA and N-acetyl aspartate are seen. In this project, we will test the ability of in vivo MRSI to detect 4 das after the stimulation injury the level of injury and the development of incipient epilepsy using a severe and mild version of the PPS model. This will be done in a blinded fashion. As all animals will also be video and EEG monitored, we will also be able to test the ability of the MRSI to identify those animals who go on to develop overt recurrent seizures.

DESCRIPTION (provided by applicant): The relatively high success rates (~70%+) for 1 to 2 year seizure freedom in patients undergoing temporal lobe surgery for epilepsy indicate that with accurate localization of seizure onset, respective surgery works. While there are clearly uncertainties of longer term success, the 2 year timeframe argues clearly that the surgery successfully removed a key node of the seizure network, thus frequently enabling more effective medical therapy. However for localization-related epilepsies which are not clearly temporal lobe (or medial temporal lobe) in onset, the rates of success for the 1 year time frame are substantially less, e.g., at 25%. Given the understanding of the pathophysiological basis of localization related epilepsy, the key step in these more difficult cases remains adequate localization and network characterization MR spectroscopic imaging (MRSI) has been suggested to have the potential sensitivity to assist in this localization problem. In this projectwe will implement key steps originally developed at 7T to make MRSI clinically robust at a clinical field (3 Tesla) for epileptogenic locus and network localization. This will be performed at the University of Pittsburgh and New York University, both sites of which have the needed experience and collaborative interest in epilepsy and imaging. Aim 1 will result in a robust dataset of pre-operative fronto-parietal-temporal and temporal-occipital MRSI from a large group (a total of n=60 each year, 240 total) of surgically treated patients. The MRSI-abnormal regions will define candidate regions of seizure onset, to be compared against the current clinical localization paradigm and clinical outcome of seizure control. This will be done in several subgroups of epilepsy patients segregated based on lesion types and semiology to evaluate this approach in adequate sampling of epilepsy types. In comparison with the MRSI, Aim 2 will evaluate structural imaging abnormalities with whole brain T1W and FLAIR MRI using quantitative maps of cortical thickness, white-gray matter contrast and FLAIR intensities between patients and controls. This Aim will assess the extent to which these quantitative imaging abnormalities will be synchronous with MRSI-determined metabolic dysfunction. Finally in Aim 3, the conclusions of the MRSI and quantitative imaging analysis for localization of seizure onset will be correlated with region of seizure surgery and 2year post-surgical outcomes using ILAE classification. As a result of this project, we believe that we will have evaluated the role of extended volume brain metabolic MRSI evaluation for epilepsy localization and integrated its pathophysiological use with structural imaging. The health relatedness of this study: MRSI integrated with quantitative structural imaging analysis will be able to augment current seizure localization accuracy beyond the temporal lobe group. This may potentially improve outcome, reduce the cost of disease management, and hopefully make this complex yet highly effective treatment strategy more realistic and available to patients with medically refractory epilepsy.

Magnetic Resonance Spectroscopic Imaging (MRSI) has proven to provide unique information for the diagnosis and management of brain tumors, epilepsy, multiple sclerosis and traumatic brain injury. Despite the obvious advantage of imaging approaches over single volume measurements, clinically, most MRS studies are still performed as single voxel studies. The reluctance to include MRSI in clinical evaluations arises primarily from four factors: 1) increased acquisition times; 2) limitations in spectral quality when data is acquired over larger brain regions; 3) limitations in SNR and 4) challenges in sampling the cortical periphery. To overcome these limitations we will develop a fast MRSI method (5-10min.) which uses: 1) two dimensional rosette encoding trajectories to rapidly sample the brain in two dimensions while minimizing gradient demands and improve spectral quality; 2) Hadamard encoding in the third dimension to minimize localization artifacts and provide excellent slice profiles for smaller numbers of partitions (4-8) covering the most relevant brain region; 3) a high degree shim insert to maximize magnetic field homogeneity and improve spectral quality and 4) dynamic spatially selective dephasing to maximize SNR and sample the cortical periphery. Consistent with what is the most widely accepted MRS clinical application currently, we will evaluate the methods in patients with high-grade brain tumors receiving immunotherapy. Although immunotherapy is a highly promising new therapeutic approach for brain tumors, treatment effects can mimic tumor progression on conventional MRI, compromising our ability to effectively monitor and manage these patients. MRSI offers an alternative means to monitor progression, based on tumor metabolism and physiology as opposed to relaxation properties of tissue water (conventional MRI). Thus we believe that MRSI may provide additive values and significantly aid in the management of these patients. This work will be performed at 3T and 7T.