Research

Improving and exploring the capabilities of Arterial Spin Labeling (ASL) is a major focus in the Division. ASL exploits the spatial selectivity of MRI to change the sign of the nuclear spins (responsible for the MRI signal) of the water in inflowing arterial blood. This “labeling” of arterial spins is achieved only with magnetic fields; no injection or radiation is required. Images acquired after allowing for the labeled blood to enter tissue reflect perfusion (the blood flow through the capillaries of tissue). Quantitative images of perfusion can readily be generated using this technique. Since perfusion reflects the activity of tissue along with the health of its blood supply, ASL perfusion MRI has found wide use in studies of the brain, kidneys, and cancer. ASL can also be used to produce high quality images of the arteries, otherwise known as angiography. The unique noninvasive capabilities of ASL to selectively label individual arteries and create high temporal resolution images of arterial inflow are just now being appreciated.

Dr. Alsop has been a pioneer of techniques and applications of ASL for clinical research and clinical use. His group has developed improved methods for labeling and evaluated its sensitivity in applications from newborns to Alzheimer’s Disease, and stroke to cancer. Partnering with MRI vendors, his group has influenced the commercialization of ASL for more widespread use. With funding from the National Institutes of Health and GE Healthcare, his group is currently working to improve the reliability and precision of ASL in the brain and expand its use to applications in the body. Some recent areas of focus include

1. Improving the reproducibility of brain ASL perfusion across time and subjects to enhance the sensitivity of research and clinical use.
2. Quantifying correlated fluctuations of brain networks using ASL and understand the relationship between network fluctuations and time averaged network perfusion.
3. Characterizing alterations in the blood brain barrier with aging using ASL.
4. Measuring the effects of antiangiogenic therapy on perfusion and vascular supply in glioma and renal cancer and their relationship to treatment response.
5. Studying blood flow to the eye in the elderly and its relationship to age related macular degeneration.
6. Developing improved methods to display and analyze blood flow images by incorporating information from other images with greater anatomical resolution.
7. Developing clinically robust 3 dimensional imaging methods for blood flow imaging of the kidneys and other abdominal organs.
8. Characterizing perfusion changes induced by transcranial direct current stimulation.

Our lab is currently actively pursuing a new technology for quantitative imaging of myelin, in-vivo. Myelin is a sheath, located around nerves, which is essential to the rapid communication of neural signals. Myelination occurs rapidly during infant development as the child experiences new stimuli. Even in adults, myelination has been shown to contribute to long term skill acquisition and possibly other long term memory processes. Loss of myelin during aging is a major contributor to cognitive slowing with age. A number of disorders also specifically affect myelin including Multiple Sclerosis, Progressive Multifocal Leukoencephalopathy, and even traumatic brain injury.

Quantitative imaging of myelin would be a valuable tool for the study of brain development, aging, and many disorders. Several techniques for imaging myelin, including magnetization transfer (MT) imaging, separation of T2 relaxation components, and Diffusion Tensor Imaging (DTI) have been explored in the literature. Often these are affected by factors which in turn, limit their specificity to myelin.

Our work builds upon the concept of magnetization transfer imaging. MT is a technique to make MRI more sensitive to the large, slow moving macromolecules in tissue. Because these molecules move so slowly, their signal decays too rapidly for imaging with standard MRI techniques, that typically focus on the signal from water. Attenuated signal, or magnetization, from the macromolecules can exchange with the water signal, however, and thus become visible on water imaging. Saturation of the macromolecule signal can be achieved by applying radiofrequency magnetic fields away from the water frequency. The MT attenuation of the water proton signal in MRI produced by off-resonance radiofrequency irradiation has shown some relationship to the degree of myelination or damage to myelin. Unfortunately, many other molecules contribute to the MT signal such that the specificity to myelin is limited.

We have developed and reported a refinement of the MT method that makes the signal much more specific to myelin. By subtracting MT images produced by saturating simultaneously at identical positive and negative frequency offsets from the water frequency to those produced by applying all the power at either the positive or negative frequency, we produce an MT image that is selective for macromolecules that have restricted motion but still enough mobility to weaken the interactions between neighboring atoms. This weak interaction can be quantified in terms of the relaxation time for dipolar order, or T1d. T1d is much longer in myelinated tissues than any other tissue. Because molecules with long T1d are said to be inhomogeneously broadened, we refer to this technique as inhomogenous MT or ihMT.

We continue to work on improving and characterizing the ihMT method. We have demonstrated strong specificity for myelinated tissues in the brain and spine, developed methods to quantify T1d and other ihMT relevant parameters, and are exploring medical research and clinical applications. Validation against histological measures is still required to prove the myelin specificity of iHMT. This work is currently supported through internal funds and by GE Healthcare, but we are actively seeking additional funding to accelerate development and characterization of this promising imaging method.

We are now in the process of optimizing the method and performing some of the first studies with our myelin imaging technique in patient populations. Validation studies in model systems are also being explored. Specific topics include:

1. Increasing the myelin signal by brief concentrated applications of radiofrequency power.
2. Implementing motion insensitive methods to increase the robustness of the myelin signal.
3. Comparing our myelin methods with other proposed myelin imaging methods, including myelin water imaging.
4. Demonstrating the sensitivity of myelin imaging in animal models of multiple sclerosis and potentially other pathologies.

Magnetic resonance can be used to separate the signal from different molecules based on their slightly different frequencies. While some of the molecules remain separate from the water signal, some others exchange their nuclear magnetization with surrounding water. Applying radiofrequency power at the frequency of the nonwater molecules will lead to reduction in the water signal. Since some of the molecules with such properties are of intrinsic interest, including glucose, glycogen, glutamate, and creatine, and the exchange rate may be sensitive to pH, imaging using this chemical exchange saturation transfer (CEST) technique may be of great utility for measuring the properties of the cellular microenvironment.

At BIDMC MRI Research, we have developed new approaches to measuring the CEST signal that are simple and robust for clinical imaging. We have carefully characterized the different contributions to the CEST signal and compared the CEST signal in glioma to that in normal brain. Ongoing work seeks to characterize pH changes accompanying some forms of cancer therapy and seeks to provide a more robust measure so it can be used more readily in clinical applications.

In conventional MRI, the strong magnetic field of the scanner is used to align the magnetization of individual nuclei. However, the magnetic field interaction is weak compared to thermal energy so the net alignment is only a few millionths. Because of this low alignment, only very highly concentrated molecules, especially water in the body, can be imaged with MRI.

An alternative to thermal polarization is to use special techniques outside the body to hyperpolarize the nuclei so that nearly all of them are along the magnetic field. This increases the signal by a factor of 10000 or more. When injected in the body, these nuclei can be observed as molecular tracers. In particular, molecules labeled with stable carbon-13 can be used to probe membrane transport, enzyme activity and metabolism in-vivo. BIDMC is fortunate to have one of the few Dynamic Nuclear Polarization (DNP) systems and has been pursuing research on hyperpolarization for years.

The BIDMC team, led by Dr. Aaron Grant, has been developing new technologies and applying them to the study of metabolic therapies in cancer. Technical developments include new balanced SSFP acquisition methods for high spatial and temporal resolution imaging while separating the different metabolic products of hyperpolarized tracers, and the development of new tracers such as hyperpolarized t-butanol that can be used an excellent perfusion tracer. Dr. Grant has a particular focus on metabolism in cancer, including the study of glycolysis and its inhibition along with alternative fuels for cancer metabolism when glucose metabolism is blocked by enzyme target drugs.

MRI of the Tumor Microenvironment: The role of metabolic and perfusion heterogeneity in determining tumor behavior

Leo L Tsai, MD PhD, Principal Investigator
Patricia Coutinho DeSouza, DMV, PhD, Postdoctoral Fellow

Recent advances in our understanding of the pathogenesis of cancer have led to a major push towards the development of personalized medicine, with therapies targeting specific molecular pathways for tumor proliferation, survival, or treatment resistance. Tumor grading and molecular profiling require tissue sampling and histological analysis, which is impractical for tracking therapeutic response and resistance and subject to sampling error. Imaging is often used to track tumor progression and treatment response, but measures far less specific features, such as size. A major goal in molecular imaging is to bridge the diagnostic gap between radiology and pathology by noninvasively mapping specific intratumoral pathways, from expression of proliferative factors to metabolic activity, that would more accurately identify local variations in a tumor’s aggressiveness and treatment sensitivity.

Our group is interested in using MRI techniques to probe the microenvironment of tumors, with the goal of revealing new methods to accurately track tumor behavior and therapeutic response. In order to study this effectively we have had to improve techniques comparing radiologic to histologic data. These are our current projects:

1. We are using h13C-MRI to investigate regional differences in cellular metabolism and local perfusion within tumors. Our preliminary data show that lactate production is increased in hypoxic areas of low-perfusion, where there is greater tumor cell proliferation. We demonstrate this using a renal cell carcinoma mouse xenograft and are exploring how the intra-tumoral dynamics are altered during antiangiogenic therapy with sunitnib, and at resistance. This project is being performed in collaboration with Rupal Bhatt, MD PhD , of the Department of Medicine at BIDMC.


2. We are using Gd-based MR markers and h13C-MRI to demonstrate intra-tumoral and peri-tumoral changes in response to radiofrequency ablation (RFA) in a hepatocellular carcinoma (HCC) model. We are focused on the c-Met pathway in HCC and how its expression is altered in response to RFA. This project is being performed in collaboration with Muneeb Ahmed, MD , of the Laboratory for Minimally Invasive Therapies at BIDMC.


3. Whole-slice scanning microscopy and virtual histology of antibody-labeled specimens allows for analysis of tumor marker expression or changes in tissue morphology on a macroscopic scale similar to radiologic images. There is a need to improve quantitative comparisons between these two types of images to test for a multitude of radiologic or histologic parameters that could better predict tumor behavior. Our goal is to develop a robust, semi-automated, software-based platform to perform these whole-sample rad-path correlation analyses to improve the diagnostic yield of radiologic images. This technique will be useful in current studies of novel MRI biomarkers in preclinical models of renal and hepatic carcinomas. This method will also be immediately translatable to clinical use, where rad-path analysis between CT/MR and whole surgical specimens would more easily identify imaging parameters predictive for grade and genotype.