Online Issues
<< All Back-issues
<< This Issue's Table of Contents
ILAR Journal V42(3) 2001
Impact of Noninvasive Technology on Animal Research
Electron
Electron Paramagnetic Resonance for
Small Animal Imaging Applications
Murali C. Krishna, Nallathamby Devasahayam, John A. Cook, Sankaran Subramanian, Periannan Kuppusamy, and James B. Mitchell
| Murali C. Krishna, Ph.D., is Section Chief; Nallathamby Devasahayam, B.S.E.E., is Research Fellow; John A. Cook, Ph.D., is Staff Scientist; Sankaran Subramanian, Ph.D., is Research Fellow, and James B. Mitchell, Ph.D., is Branch Chief at the Radiation Biology Branch, Division of Clinical Sciences, National Cancer Institute, Bethesda, Maryland. Periannan Kuppusamy, Ph.D., is Assistant Professor at The EPR Center, Department of Medicine, The Johns Hopkins University, Baltimore, Maryland. |
Abstract
Magnetic resonance imaging (MRI) provides high-resolution morphological images useful in diagnostic radiology to differentiate between normal and abnormal/pathological states in tissues. More recently, emerging developments in MRI have added a functional/physiological dimension to anatomical images. Electron paramagnetic resonance (EPR), a magnetic resonance technique similar to nuclear magnetic resonance, detects paramagnetic species such as free radicals. Like MRI, EPR can be implemented as an imaging technique for small animals and potentially human applications. Because of the low abundance of naturally occurring paramagnetic species, exogenous paramagnetic species are needed for in vivo EPR imaging (EPRI). The image data from EPRI contain both spatial distribution of paramagnetic species and spectral information. Hence, spatially encoded functional information such as tissue oxygen status and redox status can be extracted and coregistered with the spatial distribution of the spin probe, to the anatomy, or both by suitable means. Ultimately, the images obtained from EPRI may be used to overlay the functional information (containing spatial tissue physiology information) onto detailed anatomical maps. With its ability to enable whole animal imaging in mice, EPRI will be a useful imaging technique that complements other techniques such as MRI and positron emission tomography in obtaining valuable functional/physiological images.
Key Words: EPR; imaging; nitroxides; oxymetry; physiology; radicals; redox; tumor
Introduction
Small animal models are being used increasingly as research tools in several human diseases such as neurodegenerative disorders, cardiovascular disorders, obesity, and cancer. Genetically engineered mice serve as useful models, aiding in the understanding of the molecular basis of pathologies and in the development of drugs and/or molecular reagents with therapeutic intent. One major limitation in realizing the benefits of small animal models is the need oftentimes to euthanize the animal to examine tissue for manifestation of phenotypical changes. In vivo imaging techniques, which can provide biochemical, physiological, or pharmacological information noninvasively on such genetically defined animals, will be of significant value in biomedical research.
Several radiological techniques, such as positron emission tomography, magnetic resonance imaging (MRI1), and ultrasound, provide detailed anatomical information noninvasively and are already in routine clinical use. These techniques can be scaled down to the small animal level to provide high-resolution anatomical and functional images. In addition, other imaging modalities, which may not be feasible for human use due to technical limitations, may be considered for small animal applications if the information they provide is useful, if the study can be performed with relative ease, or if both conditions are met (Golman et al. 1998; Lo et al. 1997).
Ideally, a small animal noninvasive imaging modality should (1) provide high-resolution morphological images, (2) contain functional information that can be coregistered with anatomy, (3) be performed with relative ease repetitively on the same animal, and (4) preferably employ nonionizing radiation. Imaging techniques based on magnetic resonance spectroscopy comply with these requirements. MRI provides detailed anatomical images in small animals. Additionally, functional MRI methods are being adapted increasingly to probe functional/physiological processes noninvasively (Griffiths et al. 1997).
Nuclear magnetic resonance (NMR1) and electron paramagnetic resonance imaging (EPRI1), based on similar principles, were discovered in the mid-1940s (Bloch et al. 1946; Zavoisky 1945). Both modalities detect resonance absorption: NMR detects nuclei with non-zero nuclear spin, such as 1H, 31P, and 13C; and electron paramagnetic resonance probes species with unpaired electrons, such as free radicals and transition metal complexes. In MRI and EPRI, spatial images are obtained by application of magnetic gradient fields. A simplified picture of the magnetic moments of electrons and protons can be envisioned as an ensemble of negatively (electron) or positively (proton) charged particles spinning without friction about their axes. Because moving electrically charged particles generate magnetic fields, both electrons and protons have their respective magnetic moments. The magnetic moment of the electron is 658 times greater than that of the proton. The magnetic moment is a vector quantity with an associated direction. Therefore, each electron or proton can be considered as magnets that have magnetic north and south poles. The spins of these species are either clockwise or anticlockwise. The magnetic moment vectors derived from these two spin states have directions opposite to each other and are denoted as parallel or antiparallel. In the absence of an external magnetic field, an ensemble of electrons or protons will not have any preferred orientation of their magnetic moments vectors, and all the spin states will have equal energy, resulting in a net zero macroscopic magnetization. However, in the presence of an externally applied magnetic field, the spin states corresponding to the individual magnetic moments within the ensemble of electrons or protons will have two discrete energy levels. The spin states with magnetic moments aligned antiparallel to the direction of the external magnetic field will have a higher level of energy than the spin states having magnetic moments aligned parallel to the external magnetic field direction. A small excess population exists in the lower level of energy state compared with the higher state due to Boltzmann distribution. This population difference confers to the spin ensemble, macroscopic magnetization, a property that can be probed by magnetic resonance techniques, such as NMR and electron paramagnetic resonance (EPR1). The higher the magnitude of the macroscopic magnetization, the greater will be the sensitivity of detection in magnetic resonance spectroscopy experiments. The magnitude of the macroscopic magnetization can be increased by increasing the strength of the external magnetic field or by decreasing the temperature of the spin ensemble. The ensemble of spins (protons or electrons) with magnetization can be probed by radiofrequency (RF1) radiation of the appropriate frequency to satisfy the resonance conditions causing resonance absorption.
Although the physical bases for EPR and NMR are similar, there are differences to be noted. A schematic description of the energy levels and populations of spin states at a given RF (approximately 200 MHz), with the corresponding magnetic field strengths, is shown in Figure 1. Because the magnetic moment of the electrons is approximately 658 times stronger than that of protons, the strength of the external magnetic field required to separate the energies of the two spin states of electrons to the same extent as that of protons is correspondingly 658 times lower in magnitude. For a given resonance frequency of RF radiation, the population difference of the spin ensemble of unpaired electrons will be equal to that of protons at an operating magnetic field that is 658 times lower. For a frequency of 200 MHz RF radiation, the magnetic field for MRI would be 4.7 T, whereas the operating field for EPRI would be 7 mT. The spectral responses of the two spin systems also show interesting differences. When perturbed from equilibrium conditions by a RF excitation pulse, the spin ensemble of protons return to equilibrium with characteristic time constants, termed the spin-lattice and spin-spin relaxation times, denoted as T1 and T2. For protons, T1 and T2 are in the range of 1.0 to 0.1 sec, with the corresponding spectral line widths in the range of Hz to kHz. In vivo, the T1 and T2 of protons vary depending on the tissue. Additionally, T2 values can be modulated using paramagnetic contrast agents such as gadolinium complexes. In the case of electrons, T1 and T2 are in the range of 5 µsec to 10 nanosec, with the corresponding spectral line widths in the range of kHz to MHz. Like protons, the T2 of electrons is influenced by molecular oxygen, which is paramagnetic; hence, oxygen can be used as an endogenous contrast agent to an exogenous paramagnetic spin probe suitable for in vivo imaging applications (Go da et al. 1995; Halpern et al. 1994).
The ability to measure oxygen concentrations in tissue noninvasively has several applications in biomedical research. Hypoxia, which often exists in tumors, is thought to be a barrier to effective cancer treatment with either radiation or chemotherapy (Teicher et al. 1981; Thomlinson and Gray 1955). Oxygen electrode data have shown that some (not all) human tumors do indeed contain hypoxic regions, the extent of hypoxia can influence tumor response to radiation treatment, and the presence of hypoxia is an indicator of tumor aggressiveness (Gatenby et al. 1988; Hockel et al. 1996). These findings point to the importance of a priori knowledge of pO2 levels in tumors in selecting the most appropriate and effective treatment options. The use of small animal models to assess the impact of tumor hypoxia on treatment response and evaluation of novel agents to circumvent this problem is extremely important to the advancement of cancer treatment. Tumor angiogenesis and its relation to metastatic disease, as well as the evaluation of agents designed to inhibit angiogenesis (Folkman 1999), comprise an emerging area of research in oncology. It would be expected that inhibitors of angiogenesis would influence tissue oxygen levels. The ability to assess the effectiveness of such agents noninvasively and repetitively in small animals would be useful to the progress and development in this field. Tissue damage as a result of ischemia with subsequent reperfusion, which occurs as a result of stroke and heart attack, is another area of research that could benefit from having a means of noninvasively monitoring tissue oxygen levels. In all of these examples, small animal models are crucial in evaluating treatment interventions and establishing mechanistic principles.
Considerations for in Vivo EPR Imaging in Small Animals
Spin Probes
MRI has been successfully implemented to study humans because of the abundance of tissue water protons with simple NMR spectra, which are probed for imaging. However, unlike in MRI, the naturally occurring paramagnetic species amenable for EPRI are below the detection limits. Thus, to implement EPRI in biological systems, exogenous paramagnetic spin probes must be introduced. Various factors involved in MRI and contrasted to EPRI are listed in Table 1. An EPRI probe used for in vivo imaging applications should (1) be chemically stable and water soluble; (2) be nontoxic; (3) have a simple EPR spectrum at ambient temperature; (4) have a pharmacological half-life that permits imaging times of at least 10 min; and (5) have spectral properties such as line width and/or intensity, which can be modulated by tissue oxygen status or redox status.
Based on the five requirements listed above, two classes of compounds were found to be suitable for use in small animal imaging applications. The first is the class of compounds known as nitroxide radicals (Figure 2). These agents are conferred with chemical stability by the substituent groups at the alpha position of the free radical site to prevent disproportionation. Additionally, nitroxides participate in well-defined redox chemistry, which makes them effective antioxidants in vitro and in vivo (Mitchell et al. 2000). Nitroxides undergo facile reduction in vivo via intracellular enzymatic processes at a rate that is inversely proportional to the oxygen status of tissue (Swartz 1990). Thus, EPRI experiments can noninvasively interrogate redox status differences in vivo, utilizing the oxygen-dependent pharmacological half-lives of nitroxides.
A second group of compounds that satisfy the requirements for in vivo EPRI applications are trityl-based stable free radicals (Figure 2). Like nitroxides, trityl radicals are extremely stable in solution. The molecule is derivatized to have no magnetic nuclei in close proximity to the free radical site so as to minimize spectral splittings, thus exhibiting an EPR spectrum with a single narrow line. However, the line width increases with oxygen concentration in a linear manner. The toxicity of these spin probes is minimal, and they are well tolerated in small animals such as mice, rats, rabbits, and pigs. Their pharmacological half-lives, in the range of 10 to 20 min, are convenient for collecting image data.
Optimum Frequency and Magnetic Field for EPRI
As in MRI, the sensitivity and resolution of EPRI are directly dependent on the operating frequency of the RF or, equivalently, the static magnetic field. For increased sensitivity, higher operating frequency is desirable provided the nonresonant absorption (which can cause heating of the object) of the RF at higher frequency is not significant. To operate at higher frequencies for sensitivity, magnet technology is a determinant factor in MRI but not for EPRI. However, the penetration and distribution of the RF fields into biological samples is a critical factor in the selection of the EPR frequency for a given sample size. In general, it is advisable to use the highest frequency at which uniform penetration can be achieved to maximize sensitivity.
In imaging, the limitations imposed on frequency can be even more severe due to the homogeneity requirements of B1, the magnetic field component of the RF. Although inhomogeneity in B1 can be mathematically corrected, it is desirable to avoid strong variations over the field of view. Numerous NMR and EPRI studies describe the properties of biological tissues in regard to the penetration of RF/microwave fields (Bottomley and Andrew 1978; Roschmann 1987). Several EPRI scanners have been designed and implemented for small animal imaging in the frequency range of 200 to 1200 MHz (8-40 milli-Tesla). Both theoretical predictions and documented experimental data suggest the choice of a 250- to 500-MHz frequency range for whole body spectroscopy/imaging of small animals such as rats and rabbits and 750 to 1200 MHz for mice and isolated organs from larger animals.
Signal Detection
Signal detection of magnetic resonance signals can be achieved either in the frequency domain or in the time domain of the spectra. When the magnetic resonance techniques were implemented originally, the signals in both NMR and EPR were detected experimentally in the frequency domain in the continuous wave (CW1) mode, in which the frequency or field was varied continuously and each individual point in a spectrum was collected sequentially (Bloch et al. 1946; Zavoisky 1945). In NMR and MRI, the frequency domain mode of data acquisition has been replaced by time domain methods in which the complete NMR spectrum is acquired simultaneously after pulsed excitation. In time domain methods, the spin system is irradiated by a microsecond RF pulse at resonance conditions. Thereafter the free induction decay (FID1) of the magnetization is detected in the time domain. Fourier transformation (FT1) of the FID results in the NMR spectrum in the frequency domain. FT-NMR sensitivity of detection has been increased significantly by coherent addition of several individual spectra in short periods of time to increase the signal-to-noise ratio. At the time of this writing, NMR spectroscopy and MRI use FT methods exclusively.
In EPR, most spectrometers/imagers utilize the CW methods of signal detection. In spite of the success of FT NMR in spectroscopic and imaging experiments, the reason FT EPR was not preferred for in vivo imaging studies was that most FIDs of commonly used paramagnetic spin probes last for times in the order of microseconds. These times are longer than the recovery times of the EPR spectrometers. However, EPR instrumentation with short recovery times have recently become available, making FT EPRI with spin probes such as the trityl radicals, whose FIDs last longer than 2 microsec, feasible (Afeworki et al. 2000; Murugesan et al. 1997). However, in the case of nitroxides, whose FIDs last for less than 0.5 microsec, CW EPRI methods may be appropriate.
CW EPRI methods have the advantage of detecting several paramagnetic species such as the nitroxides, and trityl radicals. Additionally, in imaging experiments, CW EPR permits the use of larger magnetic field gradients to obtain images with better spatial resolution. These positive aspects of CW EPR are offset by the longer imaging times. Therefore, CW EPRI methods compromise imaging studies in which high temporal resolution is desired. Additionally, in spectroscopic imaging experiments with CW EPR, where spatially resolved spectroscopic information must be extracted from the object, the imaging times may be significantly longer (1 hr), which may confound the physiological processes being interrogated.
Although restricted to the detection of paramagnetic spin probes such as the trityl radicals, FT EPRI methods have the advantage of collecting image data relatively rapidly (<5 min). Furthermore, the spatial image data in FT EPRI contain inherent spectroscopic information, which can be used to extract tissue oxygen concentration on a voxel-by-voxel basis (Afeworki et al. 2000). This information can be extracted using simple mathematical treatment of the image data, and the oxygen maps from such computations can be coregistered with the spatial images of the spin probe distribution.
Resonators
The resonator is a critical component for both EPRI and MRI, and it can determine the sensitivity of detection and image quality. The sample should occupy most of the resonator volume, and the resonator itself should use the space efficiently within the homogeneous volume of the magnet/gradient assembly. Furthermore, the resonator should permit access to the animal for anesthesia and placement of intravenous lines to deliver the contrast agent. In addition, the coupling of the conversion of the RF energy from the source to the corresponding magnetic component B1 should be efficient. In CW EPR operations, in which the resonator is critically coupled to the source to achieve maximum sensitivity in detection, it is necessary to compensate for any coupling changes due to object movement such as respiratory motion with automatic coupling control and automatic frequency control. However, in FT EPRI, the resonator is deliberately mismatched to the source to minimize the resonator response time after the RF pulse is delivered. The resonator dead time is a critical factor because it will determine the sensitivity of detection and the image quality.
Volume resonators such as the loop-gap resonator (Froncisz and Hyde 1982) and the re-entrant resonator (Sotgiu 1985) are two configurations that have been used successfully for small animal CW EPR imaging applications. Topical resonators for skin imaging or tumor xenografts have also been used recently (Kuppusamy et al. 1998a,b). For FT EPR imaging applications, volume resonators suitable for whole body imaging of mice have been developed with short response times (<0.5 μsec) and acceptable homogeneity. Imaging experiments with phantom objects using these resonators validate EPRI as providing images that faithfully represent the spin probe distribution.
Image Formation and Reconstruction
Spatial encoding of the paramagnetic spin probes in an object to obtain a two- or three-dimensional image is accomplished by magnetic field gradients as in MRI. The magnetic field strengths typically used in both CW EPRI and FT EPRI are in the range of 10 to 50 mT/m (1-5 G/cm). However, unlike in MRI where the use of pulsed gradients permits Fourier imaging, in CW EPRI and FT EPRI, static field gradients are used, requiring projection reconstruction methods such as filtered back-projection for image computation. Although static field gradients are mandatory in CW EPRI methods, in FT EPRI, in principle, pulsed magnetic field gradients can be used, as in MRI. However, they will not be practical because the gradient switching times (~1 msec) will be longer than the FIDs (-2 µsec) in EPRI. The filtered back-projection is by far the most commonly used reconstruction algorithm in EPR.
Image Resolution
The accuracy of an image can be assessed in terms of its quality and resolution. Image quality is the fidelity or faithfulness of the measured image in reproducing the original object. The quality depends mainly on signal-to-noise ratio. The resolution is defined as the minimum significant image element (distance, area, or volume) that can be meaningfully interpreted from the image. Resolution depends on several factors, both instrumental and computational. Hoch and Ewert (1991) have delineated the factors that affect resolution of an EPR image and discussed methods of estimating the resolution on an individual factor basis. The image resolution in EPRI is generally in the range of 1 to 2 mm for whole animal imaging in mice and <1 mm in topical applications where the resonators span a region of interest not greater than 1 to 2 cm.
Instrumentation
A typical arrangement of the resonator in the magnet/gradient assembly of a whole body CW EPRI and FT EPRI is show in Figure 3A, with a sketch of a mouse within the resonator along with the gradient axes directions (Afeworki et al. 2000). The Z axis coincides with the axis of the static magnetic field (Zeeman field). These instruments, capable of whole body imaging of mice, are performed at a RF of 300 MHz, with the corresponding magnetic field of 10.6 mT. The three axes gradient assembly can generate magnetic field gradients of up to 50 mT/m. Resonators with bore sizes in the range of 25 to 50 mm and length in the range of 25 to 50 mm have been successfully tested for small animal imaging experiments. For topical imaging applications where the field of view is smaller, higher resonant frequencies can be used. A typical arrangement of the animal in topical imaging experiments is shown in Figure 3B, where the operating frequency is 1200 MHz, with the magnetic field at 40 mT. Halpern and colleagues (1989), Kuppusamy and coworkers (1994), and Murugesan and colleagues (1998) have described details of the instrumentation.
Example of Redox Status Mapping and
Tissue Oxygen Levels with EPRI
EPR imaging studies using nitroxides as redox-sensitive spin probes could provide information noninvasively to discriminate between the differences of the pharmacokinetics of nitroxides in tumor versus normal tissue. The basis for the capability of EPRI in distinguishing such differences is the pharmacological behavior of nitroxides in vivo. Although nitroxides are chemically stable in solutions, they participate in redox reactions with reactive species in vitro and they undergo enzymatic redox reactions in vivo. Nitroxides undergo redox transformations to two other oxidation states. By undergoing 1-electron oxidation, they are converted to the transient oxoammonium species (diamagnetic); and by accepting an electron, they are reduced to the corresponding hydroxylamine, which is also diamagnetic and stable (Krishna et al. 1996). When nitroxides are administered in vivo, they are converted to the corresponding hydroxylamines, which themselves can be reoxidized back to the nitroxides.
The ratio of nitroxide/hydroxylamine depends on several factors of which tissue oxygen and redox status are important (Swartz 1990). Nitroxides are converted to hydroxylamines by intracellular enzymatic processes that are more efficient in hypoxic conditions than aerobic conditions (Figure 4). Solid tumors are known to differ in redox status, as well as oxygen status, compared with normal tissue. When nitroxides were administered to tumor-bearing mice and the tissue homogenates were examined by EPR spectroscopy ex vivo for nitroxide levels, greater quantities of hydroxylamine were found in tumors than in normal tissue consistent with the notion that tumors possess more reducing environment. Thus, EPRI studies of pharmacokinetic differences of nitroxide in tumors and adjacent normal tissue should reflect their respective redox status consistent with the reduction/clearance of nitroxides.
EPRI experiments have been performed using an EPR spectrometer operating at 1.2 GHz corresponding to a magnetic field of 40 mT with a specially built bridged-loop surface resonator (Kuppusamy et al. 1998a). The open structure of the resonator is ideal for localized study of metabolic activity in large objects. With this imaging system, a cylindrical volume of 10 mm diameter and 5 mm depth could be probed in both tumor and normal tissue. With this technique, it was possible to assess the levels of nitroxide accumulation in normal tissue and tumor as well as differences in their metabolism. Mice bearing ~1 cm-diameter tumors (radiation-induced fibrosarcoma) were anesthetized and the tail vein cannulated with a heparin-filled 30-gauge catheter to infuse 3-carbamoyl-proxyl (160 mg/kg). Either the right leg with tumor or the left leg with normal tissue (muscle, skin) was placed on the resonator and sampled for nitroxide content by EPR. The presence of nitroxide in both normal and tumor tissue was readily detected. A two-dimensional spatial EPRI of the distribution of 3-carbamoyl-proxyl in normal muscle and 1 cm-diameter radiation-induced fibrosarcoma tumor as a function of time is shown in Figure 5. The panels in the top row represent the decrease in the nitroxide levels in normal muscle as a function of time, and the corresponding images in the bottom row represent clearance from tumors. Comparisons of images obtained at 4.5 and 10.5 min indicate that the rate of clearance of nitroxide in tumors is faster than in normal tissue. This indication is consistent with previous observations that compared with normal tissue, tumors provide a strong reducing environment that results in faster reduction of the nitroxide. These observations agree with earlier studies suggesting that hypoxic cells within tumors reduce nitroxides more efficiently than well-oxygenated normal tissue. The EPR spectral line width of nitroxides is also dependent on oxygen concentrations (Figure 6A). These line widths can be calibrated to pO2 concentrations. Estimates of oxygen concentration in the tissues shown in Figure 6B using EPRI/oximetry indeed confirmed that the tumor tissue was much lower in oxygen concentration than normal tissue. Collectively, these studies establish the feasibility of detecting nitroxides in tumors and clearly show major differences in nitroxide distribution and metabolism between normal and tumor tissue.
Example of Oxygen Mapping by FT EPRI
Paramagnetic spin probes based on trityl radicals have been developed for use in Overhauser-enhanced MRI (Golman et al. 1998). Spin probes that exhibit narrow single EPR spectra were found to be optimal for these applications. Such properties in a spin probe make it feasible to apply FT EPRI methods for in vivo EPR imaging. The initial demonstration of FT EPRI in phantom objects and small regions in in vivo studies demonstrated the feasibility of these methods (Murugesan et al. 1997). Later studies focused on developing these techniques for whole body imaging in small animals as well as adding the functional imaging capabilities to EPRI. FT EPRI experiments were carried out in an anesthetized mouse after administering the spin probe intravenously. The mouse was placed in the resonator so that the field of view covered the abdominal region where the kidneys and liver are located. The thoracic and pelvic regions were deliberately left outside the field of view (Figure 7). After intravenous administration of the spin probe, an EPR spectrum in the absence of gradients was recorded to establish the center of gravity of the image. EPR projections were then collected in the presence of magnetic field gradients (4-12 mT/m). Projection data were collected by rotating the gradient vector direction every 15° in the polar and azimuthal angles to total 144 projections over the solid angle of a hemisphere concentric with the resonator. Three-dimensional images of the spin probe distribution were reconstructed from these projections by filtered back-projection methods and were surface rendered (some perspectives are shown in Figure 7A). The images obtained suggest that the spin probe localizes to a significant extent in the liver and kidney, consistent with the pharmacokinetic distribution and elimination via renal clearance of this agent. Distances measured from image data were in good agreement with the actual distances measured in the sacrificed animal.
Because the time domain mode of data acquisition (FT EPRI) permits the discrimination of differences in oxygen status by a T2* weighted image processing, it should be possible then to extract such differences in oxygen status between the liver and kidney of the EPR image shown in Figure 7A. The physical basis for such capabilities in FT EPRI is as follows: The paramagnetic relaxation brought about by molecular oxygen, which is paramagnetic, by physical interaction with the unpaired electron of the paramagnetic spin probe causes a shortening of its time domain response. Therefore, the time domain responses of the spin probe are expected to last longer in hypoxic regions compared with normoxic regions. Therefore, if the FIDs in an image data are weighed for O2-induced relaxation and then subjected to Fourier transformation, the resulting spectral projections will reflect oxygen status differences. A computational approach to reveal such differences from the same image data is to progressively delete the initial points in the FID of each projection and Fourier transformed. The images will reveal reduced intensity in relatively well-aerated regions compared with the original image. As shown in Figure 7 (B-D), the in vivo images obtained by processing the data sets correspond to the image displayed in Figure 7A after progressively deleting the initial 100, 200, and 400 time points in the FIDs. These images reflect a significant reduction in intensity from the kidneys compared with that from the liver (note: the kidney image disappears in Figure 7, C-D). This observation is further supported by independent studies (direct oxygen electrode measurements), which suggest that kidneys are relatively more oxygenated than the liver (data not shown). Such strategies may show promise in spatially resolving hypoxic regions and may represent a noninvasive approach to estimate oxygen status differences in organs or any other regions of interest.
Summary
EPRI at frequencies in the 200- to 1200-MHz range has been developed as a viable in vivo imaging tool for small animal functional imaging. Using nontoxic, oxygen- and redox-sensitive spin probes, it has been demonstrated that it is possible to obtain morphological images coregistered with physiological information using both the CW and FT mode of EPR imaging modalities. Although both the CW and FT modes have individual advantages and disadvantages, they act as complementary techniques when a variety of biocompatible in vivo spin probes are used for imaging. The CW mode allows higher resolution with broad-line redox-sensitive probes such as nitroxides in interrogating tumor and tissue redox status, and the FT modes provide rapid pO2 mapping with better temporal resolution. By improving the sensitivity of detection, which further reduces the incident electromagnetic flux density, and scaling up the resonator for topical imaging, we hope to develop the EPRI method as a functional and noninvasive imaging tool. This method has important implications in the study of tumor hypoxia, tissue heterogeneity with respect to oxygen and redox status, and vascular deficiencies in vivo.
1Abbreviations used in this article: CW, continuous wave; EPR, electron paramagnetic resonance; EPRI, electron paramagnetic resonance imaging; FID, free induction decay; FT, Fourier transformation; MRI, magnetic resonance imaging; NMR, nuclear magnetic resonance; RF, radiofrequency.
References
Afeworki M, van Dam GM, Devasahayam N, Murugesan R, Cook J, Coffin D, Larsen JHA, Mitchell JB, Subramanian S, Krishna MC. 2000. Three-dimensional whole body imaging of spin probes in mice by time-domain radiofrequency electron paramagnetic resonance. Magn Reson Med 43:375-382.
Bloch F, Hansen WW, Packard HE. 1946. Nuclear induction. Phys Rev 69:127-131.
Bottomley PA, Andrew ER. 1978. RF magnetic field penetration, phase shift and power dissipation in biological tissue: Implications for NMR imaging. Phys Biol Med 23:630-645.
Folkman J. 1999. Angiogenesis research: From the laboratory to clinic. Forum (Genova) 9:59-62.
Froncisz W, Hyde JS. 1982. The loop gap resonator: A new lumped circuit ESR sample structure. J Magn Reson 47:515-521.
Gatenby RA, Kessler HB, Rosenblum JS, Coia LR, Moldofsky PJ, Hartz WH, Broder GJ. 1988. Oxygen distribution in squamous cell carcinoma metastases and its relationship to outcome of radiation therapy. Int J Radiat Oncol Biol Phys 14:831-838.
Goda F, O'Hara JA, Rhodes ES, Liu KJ, Dunn JF, Bacic G, Swartz HM. 1995. Changes in oxygen tension in experimental tumors after a single dose of x-ray irradiation. Cancer Res 55:2249-2252.
Golman K, Leunbach I, Ardenkjaer-Larsen JH, Ehnholm GJ, Wistrand LG, Petersson JS, Jarvi A, Vahasalo S. 1998. Overhauser-enhanced MR imaging (OMRI). Acta Radiol 39:10-17.
Griffiths JR, Taylor NJ, Howe FA, Saunders MI, Robinson SP, Hoskin PJ, Powell ME, Thoumine M, Caine LA, Baddeley H. 1997. The response of human tumors to carbogen breathing, monitored by gradient-recalled echo magnetic resonance imaging. Int J Radiat Oncol Biol Phys 39:697-701.
Halpern HJ, Spencer DP, van Polen J, Bowman MK, Nelson AC, Dowey EM, Teicher BA. 1989. Imaging radio frequency electron-spin-resonance spectrometer with high resolution and sensitivity for in vivo measurements. Rev Sci Instrum 60:1040-1050.
Halpern HJ, Yu C, Peric M, Barth E, Grdina DJ, Teicher BA. 1994. Oximetry deep in tissues with low-frequency electron paramagnetic resonance. Proc Natl Acad Sci U S A 91:13047-13051.
Hoch MJR, Ewert U. 1991. Resolution in EPR imaging. In: Eaton GR, Eaton SS, Ohno K, eds. EPR Imaging and In Vivo EPR. Boca Raton: CRC Press. p 153-160.
Hockel M, Schlenger K, Aral B, Mitze M, Schaffer U, Vaupel P. 1996. Association between tumor hypoxia and malignant progression in advanced cancer of the uterine cervix. Cancer Res 56:4509-4515.
Krishna MC, Samuni A, Taira J, Goldstein S, Mitchell JB, Russo A. 1996. Stimulation by nitroxides of catalase-like activity of hemeproteins. J Biol Chem 271:26018-26025.
Kuppusamy P, Afeworki M, Shankar RA, Coffin D, Krishna MC, Hahn SM, Mitchell JB, Zweier JL. 1998a. In vivo electron paramagnetic resonance imaging of tumor heterogeneity and oxygenation in a murine model. Cancer Res 58:1562-1568.
Kuppusamy P, Chzhan M, Vij K, Shteynbuk M, Lefer DJ, Giannella E, Zweier JL. 1994. Three-dimensional spectral-spatial EPR imaging of free radicals in the heart: A technique for imaging tissue metabolism and oxygenation. Proc Natl Acad Sci U S A 91:3388-3392.
Kuppusamy P, Wang PH, Shankar RA, Ma L, Trimble CE, Hsia CJC, Zweier JL. 1998b. In vivo topical EPR spectroscopy and imaging of nitroxide free radicals and polynitroxyl-albumin Magn Res Med 40:806-811.
Lo LW, Jenkins WT, Vinogradov SA, Evans SM, Wilson DF. 1997. Oxygen distribution in the vasculature of mouse tissue in vivo measured using a near infrared phosphor Adv Exp Med Bio 411:577-583 .
Mitchell JB, Russo A, Kuppusamy P, Krishna MC. 2000. Radiation, radicals, and images. Ann NY Acad Sci 899:28-43.
Murugesan R, Afeworki M, Cook JA, Devasahayam N, Tschudin R, Mitchell JB, Subramanian S, Krishna MC. 1998. A broadband pulsed radio frequency electron paramagnetic resonance spectrometer for biological applications. Rev Sci Instr 69:1869-1876.
Murugesan R, Cook JA, Devasahayam N, Afeworki M, Subramanian S, Tschudin R, Larsen JHA, Mitchell JB, Russo A, Krishna MC. 1997. In vivo imaging of a stable paramagnetic probe by pulsed-radiofrequency electron paramagnetic resonance spectroscopy. Magn Reson Med 38:409-414.
Roschmann P. 1987. Radiofrequency penetration and absorption in the human body: Limitations to high-field whole-body nuclear magnetic resonance imaging. Med Phys 14:922-931.
Sotgiu A. 1985. Resonator design for in vivo EPR spectroscopy. J Magn Reson 65:206-214.
Swartz HM. 1990. Principles of the metabolism of nitroxides and their implications for spin trapping. Free Radic Res Commun 9:399-405.
Teicher BA, Lazo JS, Sartorelli AC. 1981. Classification of antineoplastic agents by their selective toxicities toward oxygenated and hypoxic tumor cells. Cancer Res 41:73-81.
Thomlinson RH, Gray LH. 1955. The histological structure of some human lung cancers and the possible implications for radiotherapy. Br J Cancer 9:539-549.
Zavoisky E. 1945. Spin-magnetic resonance in paramagnetics. J Phys E 9:245-249.
Figure 1 Schematic comparison of the energy levels and populations of the spin states of protons and electrons in a magnetic field. The magnetic moments of the protons (top) and electrons (bottom) are represented as rectangular bars. In the absence of an external magnetic field, individual magnetic moments are aligned randomly. In the presence of an external magnetic field (hatched rectangle), they occupy two energy levels. Irradiation with radiofrequency radiation of appropriate frequency causes resonance absorption, which is shown as an absorption spectrum. EPR, electron paramagnetic resonance; NMR, nuclear magnetic resonance.
Figure 2 Chemical structures of spin probes used in electron paramagnetic resonance imaging. Left, nitroxide. R is -C(=O)NH2 for the nitroxide, 3-carbamoyl-proxyl, used for imaging studies. Right, trityl radical. The three benzoic acid moieties, denoted by phenyl groups are bonded at the 4-position to form the stable trityl radical.
Figure 3 Schematics of instrumentation. (A) Placement of the mouse for whole body imaging at 300 MHz (Afeworki et al. 2000). (B) Arrangement for topical imaging at 1200 MHz. The magnet and gradient assembly are shown. Placement of the mouse on the resonator within the magnet/gradient assembly is shown on right.
Figure 4 Schematic description of the dependence of tissue redox status on the oxidation states of nitroxides.
Figure 5 Spatially resolved clearance of nitroxide in normal and tumor tissue (mouse). The nitroxide 3-carbamoyl-proxyl was administered by tail vein infusion (160 mk/kg), and a series of two-dimensional images were collected as a function of time from the normal muscle (top) and tumor (bottom) using the L-band electron paramagnetic resonance spectrometer/imager. RIF-1, radiation-induced fibrosarcoma. Data adapted with permission from Kuppusamy P, Afeworki M, Shankar RA, Coffin D, Krishna MC, Hahn SM, Mitchell JB, Zweier JL. 1998a. In vivo electron paramagnetic resonance imaging of tumor heterogeneity and oxygenation in a murine model. Cancer Res 58:1562-1568.
Figure 6 (A) Line width dependence of the spin probe per deuterated Tempone, 4-oxo-2,2,6,6-tetramethylpiperidine-1-oxyl d-17, in solutions at different oxygen concentrations in phosphate buffered solutions. The linearity of line width dependence on oxygen is satisfactory. (B) Noninvasive determination of oxygen concentration in muscle and tumor using electron paramagnetic resonance (EPR) oxymetry. After intravenous administration of 100 mg/kg of per deuterated Tempone, the tumor-bearing leg or normal leg was clamped to restrict blood flow. EPR spectra were continuously monitored during the ischemia induction, and the line widths were obtained. This figure reveals that normal muscle tissue has a much higher oxygen concentration than tumor and that oxygenation decreased rapidly when blood flow is restricted to the muscle and tumor. Data adapted with permission from Kuppusamy P, Afeworki M, Shankar RA, Coffin D, Krishna MC, Hahn SM, Mitchell JB, Zweier JL. 1998a. In vivo electron paramagnetic resonance imaging of tumor heterogeneity and oxygenation in a murine model. Cancer Res 58:1562-1568.
Figure 7 Discrimination of oxygen differences in liver and kidneys from Fourier transformation electron paramagnetic resonance imaging in a mouse given intravenous administrations with the trityl spin probe. (Top) Orientation of mouse within the resonator. (A) Image processed with all the points in the FID. Image processed after progressively deleting the initial 100 (B), 200 (C), and 400 (D) points (ns) in free induction decay. Note the rapid attenuation of image intensity in kidneys, suggesting that the liver is more hypoxic than the kidneys. Data from Afeworki M, van Dam GM, Devasahayam N, Murugesan R, Cook J, Coffin D, Larsen JHA, Mitchell JB, Subramanian S, Krishna MC. 2000. Three-dimensional whole body imaging of spin probes in mice by time-domain radiofrequency electron paramagnetic resonance. Magn Reson Med 43:375-382.
Table 1 Comparison of magnetic resonance imaging (MRI) and electron paramagnetic resonance imaging (EPRI)| MRI | EPRI | |
Magnetic field at 300 MHz | 7 T | 10 mT |
Radio frequency pulse width | μsec - msec | 10 -100 nsec |
3 MHz free induction decay response | msec - sec | nsec - μsec |
Endogenous probes | Water protons | None |
Exogenous probes | NAa | Nitroxides, trityl |
Concentration | >60 M | ~mM |
Stability | Stable | Minutes |
Line width | Hz - kHz | 0.1 MHz |
Copyright © 2008. National Academy of Sciences.
All rights reserved.
500 Fifth St. N.W., Washington, D.C. 20001.
Terms of Use and Privacy Statement