State-of-the-Art Instrumentation for PET and SPECT Imaging in Small Animals

Published on 02/03/2015 by admin

Filed under Internal Medicine

Last modified 22/04/2025

Print this page

This article have been viewed 2053 times

Chapter 11 State-of-the-Art Instrumentation for PET and SPECT Imaging in Small Animals

David K. Glover, Bijoy Kundu, Heinrich R. Schelbert

INTRODUCTION

Cardiovascular molecular imaging is an emerging field that will have tremendous impact on the diagnosis and treatment of heart disease in the years ahead. While traditional cardiac imaging provides important information regarding myocardial perfusion, function, or heart structure and dimensions, cardiovascular molecular imaging utilizes tracers or contrast agents to measure physiologic and pathologic processes at the cellular and molecular level. These tracers are designed to target specific cell types, receptors or transporters, peptides, or even the expression of a particular gene product. Targeted tracers are currently being used or investigated for imaging cellular processes such as glucose transport or metabolism (i.e., FDG, BMIPP), apoptosis (annexin-V), sympathetic innervation (MIBG), as well as inflammation, hypoxia, angiogenesis, matrix metalloproteinase expression, and the transition of atherosclerotic plaques from the stable to the unstable state. In addition, cardiovascular molecular imaging approaches are being investigated as a tool for monitoring cell trafficking and tissue localization of injected therapeutic stem cells and the delivery and efficacy of targeted gene therapy.

Small animals such as mice, rats, and guinea pigs have long been utilized as models in most areas of molecular biology, toxicology, and drug discovery research, and well-characterized models of specific diseases have been developed using these animals. With the relatively recent development of transgenic mice, mouse models have gained even greater importance. The ability to manipulate the mouse genome to produce accurate models of many human diseases has resulted in significant progress in understanding these diseases. Historic methodologies for assessing molecular changes require euthanizing the animal and performing invasive and destructive assays such as autoradiography and in vitro histopathologic assays on specific tissue types. These techniques are not ideal, because they do not provide information in real time and in situ. As a result, large numbers of animals must often be euthanized to screen for the desired phenotypic changes, resulting in significantly higher cost as well as lost time. The full potential of these new transgenic mouse models can only be realized with the ready availability of accurate, noninvasive imaging methods to investigate disease progression and its response to therapeutic agents. For these reasons, there has been increasing interest in developing noninvasive imaging techniques that allow repeated in vivo imaging at the cellular and molecular level in small animals such as mice. Importantly, repeated imaging studies allow for the delineation of temporal changes of these functional processes.

While other imaging modalities like magnetic resonance imaging (MRI), myocardial contrast echocardiography (MCE), and the newer optical imaging modality are being utilized for small-animal imaging and have their particular strengths, imaging techniques based on radiolabeled probes and positron emission tomography (PET) and single-photon emission computed tomography (SPECT) instrumentation will remain at the forefront because of their unparalleled sensitivity. Unfortunately, however, the spatial resolution of commercially available SPECT and PET cameras designed for clinical imaging is on the order of 4 to 6 mm, which limits their usefulness in small animals such as mice in which the heart is typically 3 to 4 mm across the short axis and the left ventricular wall thickness is less than 1 mm (~0.8 to 0.9 mm). Accordingly, there has been a tremendous effort in recent years to develop high-resolution SPECT and PET cameras specifically designed for imaging small animals in biomedical research. At the present time, there are multiple commercial versions of both microSPECT and microPET cameras available, and intensive work is underway in both academic settings and industry to further develop and improve these cameras.

MICRO–POSITRON EMISSION TOMOGRAPHY

PET extends the investigative possibilities of radionuclide approaches that have historically focused mostly on the evaluation of myocardial blood flow. With PET tracers, it is now possible to explore and, more important, to noninvasively quantify functional processes such as substrate metabolism, oxygen consumption, receptor density and occupancy, and activation and expression of transfected genes. Because of PET’s quantitative and high temporal resolution capability, it is possible to accurately measure regional activity concentrations in the myocardium and blood as well as their changes over time. As described later, the radiotracer input function and the tissue response can be determined from serially acquired PET images. When combined with appropriate tracer kinetic models, absolute estimates of rates of functional processes can be derived (e.g., mL blood or mol substrate per minute per gram myocardium).

Dedicated, high spatial resolution small-animal PET imaging devices have become available. Some are in-house-developed systems at several research institutions,¹^–⁵ but more recently, dedicated small-animal systems have become available through several commercial suppliers. The spatial resolution of these imaging devices is typically in the range of 1.0 to 1.5 mm full-width at half-maximum (FWHM) and enables mapping of the distribution of positron-emitting radiotracer and its changes over time throughout the body of small animals. These imaging systems have been used quite extensively for measuring substrate metabolism and cell replication rates of tumors as indices of tumor growth rates and tumor aggressiveness, and for determining responses of tumor growth to investigational therapeutic agents.⁶^–¹⁰ Other research has utilized these small-animal imaging devices for the development and validation of novel noninvasive assays for identifying activation and expression of transgenes.^6,⁹^–¹¹

POSITRON EMISSION TOMOGRAPHY IMAGING OF THE RODENT HEART

The small dimensions of the rodent heart pose formidable challenges for PET studies of the cardiovascular system in rodents. In general, dedicated small-animal PET systems employ two different technical concepts: (1) crystal detector–based and (2) multiwire proportional chamber–based systems.^12,¹³ As shown in Figure 11-1, multiwire proportional chamber–based systems may offer superior spatial or volume resolution but generally are less sensitive and require longer image acquisition times. Conversely, crystal-based systems offer higher sensitivity (count rate per unit radioactivity) and offer superior temporal resolution but at the expense of somewhat lower spatial resolution (Fig. 11-2).

Figure 11-1 ¹⁸F-deoxyglucose uptake in a 29-g mouse imaged with a multiwire proportional chamber-based system (QuadHIDAC-PET, Oxford Positron Systems, Cambridge, England). The 15-minute whole body image was recorded 1 hour after injection of 10 MBq ¹⁸F-deoxyglucose. A, Coronal whole body slice. B, Reoriented myocardial slices.

(Courtesy of K. Schäfers, University of Münster, Münster, Germany.)

Figure 11-2 Sagittal, coronal, and transaxial images slices of ¹⁸F-deoxyglucose uptake in a mouse, obtained with a crystal-based, dedicated small-animal PET system (Concorde Microsystems, Knoxville, TN; Focus microPET system) and reconstructed using maximum a posteriori probability at 60 minutes after a tail-vein injection of ¹⁸F-deoxyglucose.

(Courtesy of David Stout, University of California, Los Angeles.)

Several commercial small-animal scanners are now available since the first prototype scanner was built at University of California, Los Angeles. Figure 11-3 shows one such small-animal PET scanner, the microPET Focus 120.¹⁴ This system consists of 96 scintillation detectors arranged in four rings with a 14.7-cm diameter and 7.6-cm axial extent. Each detector consists of a 12 × 12 array of 1.52 × 1.52 × 10 mm lutetium oxyorthosilicate (LSO) crystals coupled to a position-sensitive photomultiplier tube (PS-PMT) via optical fiber bundles. Data are acquired in three-dimensional (3D) mode. With standard filtered backprojection reconstruction algorithms, the system achieves an isotropic (in all three directions) spatial resolution of 1.5 mm in the center of the field of view (FOV) and remains less than 2.0 mm within 5 cm of the center of the FOV. Imaging of the myocardium in mice and rats with the flow tracer ¹³N-ammonia or the glucose analogue ¹⁸F-deoxyglucose typically requires activity doses of 0.5 to 1.5 mCi.¹⁵ Count rates at these radioactivity doses remain relatively low so that dead-time losses are small. Image reconstruction with the more recently developed maximum a posteriori probability (MAP) algorithm, now employed for cardiac studies in mice, provides images of myocardial perfusion and glucose uptake of good to excellent diagnostic quality (see Fig. 11-2).¹⁶ The MAP reconstruction algorithm improves the spatial resolution in the center of the FOV to 1.2 mm FWHM so that a volumetric resolution of less than 5 microliters is now possible.

Figure 11-3 MicroPET Focus 120 scanner.

(Image courtesy of Siemens Medical Solutions USA, Inc.)

The emission data are reconstructed initially into a set of contiguous transaxial slices with a 0.75-mm interplane spacing (Fig. 11-4).¹⁷ The transaxial images are typically reoriented into short-axis and vertical and horizontal long-axis cuts (Fig. 11-5)¹⁷ that can be assembled into polar maps analogous to those of myocardial perfusion studies in humans (Fig. 11-6).¹⁷ Images of glucose uptake obtained with ¹⁸F-deoxyglucose are of greater contrast and higher quality than images of perfusion obtained with ¹³N-ammonia. This is most likely because of the more complete clearance of ¹⁸F-deoxyglucose from blood and higher tracer activity concentrations in the myocardium. Another reason is the slower physical decay of ¹⁸F-deoxyglucose so that over the total image acquisition time, more counts are recorded. On the other hand, the shorter physical half-life of ¹³N-ammonia affords the ability to perform repeat examinations of myocardial blood flow during the same study session so that effects of interventions can be evaluated. ECG gated image acquisition is also possible with small PET systems so that left ventricular function can be assessed.¹⁸ For example, in normal 300-g rats, the left ventricular ejection fraction was about 60% to 70%, and the stroke volume was about 300 μL.¹⁸

Figure 11-4 Contiguous transaxial images of the chest in a mouse (inferior to superior from top left to bottom right, interplane spacing 0.75 mm) obtained 30 minutes after intravenous ¹⁸F-deoxyglucose.

(Reproduced with permission from Schelbert HR, Inubushi M, Ross RS: PET imaging in small animals, J Nucl Cardiol 10:513–520, 2003.)

Figure 11-5 Reoriented images of the myocardial ¹⁸F-deoxyglucose uptake in the same mouse as depicted in Figure 11-4. Selected short-axis (upper row) and vertical and horizontal long-axis cuts are shown.

(Reproduced with permission from Schelbert HR, Inubushi M, Ross RS: PET imaging in small animals. J Nucl Cardiol 10:513–520, 2003.)

Figure 11-6 Polar map display of the ¹⁸F-deoxyglucose uptake in the left ventricular myocardium in the mouse depicted in Figures 11-4 and 11-5.

(Reproduced with permission from Schelbert HR, Inubushi M, Ross RS: PET imaging in small animals. J Nucl Cardiol 10:513–520, 2003.)

The image reconstruction algorithm corrects for radioactive decay and for dead-time losses. It also includes routines for photon attenuation correction. These routines utilize transmission images that are typically acquired prior to radiotracer administration and are performed using a Co57 point source or, for hybrid microPET/CT systems, an x-ray attenuation map. Despite the small body size of rats, photon attenuation causes about 30% systematic underestimation of the true myocardial radioactivity concentration.¹⁹ Albeit somewhat less in mice, photon attenuation still amounts to about 17% underestimation of the true activity. Accordingly, even in small animals, proper attenuation correction is important for the measurement of the true absolute intensities in cardiac imaging. Chow et al.¹⁹ have shown that CT-based attenuation correction methods result in less noise correction to emission images compared to the transmission-based attenuation corrections.

The stationary imaging gantry of microPET permits rapid serial (or “dynamic”) acquisition of images. Theoretically, images can be recorded at sampling rates of less than 1 second. For practical purposes, however, image acquisition rates depend on the number of counts required for statistically adequate image frames. Framing rates in the range of 2 to 10 seconds per image following an intravenous bolus of ¹³N-ammonia, ¹⁸F-deoxyglucose, or ¹¹C-palmitate produce statistically adequate images for generating tissue time-activity curves and thus are appropriate for radiotracer kinetic studies.²⁰ Figure 11-7 depicts selected frames acquired during and shortly after an intravenous bolus injection of ¹³N-ammonia in a rat. High temporal-resolution tissue time-activity curves for the right and left ventricular blood pools and the left ventricular myocardium in a rat following an intravenous bolus of ¹⁸F-deoxyglucose into a tail vein are shown in Figure 11-8.

Figure 11-7 Selected serially acquired images during and following a bolus administration of ¹³N-ammonia into the tail vein of a rat. The initial two images depict the tracer transit through the inferior vena cava through the right and then to the left ventricle, and the third image depicts the arrival and retention of tracer in both kidneys. Delayed (13.5 and 17.5 minutes) images demonstrate declining tracer activities in the kidneys and continued accumulation of tracer in the bladder. Tracer activity is also noted in the myocardium and throughout the liver.

(Reproduced with permission from Schelbert HR, Inubushi M, Ross RS: PET imaging in small animals, J Nucl Cardiol 10:513–520, 2003.)

Figure 11-8 Time-activity curves derived from the right and the left ventricular (RV, LV) blood pool and the left ventricular myocardium in a mouse. Serial images were recorded beginning with a tail vein injection of ¹⁸F-deoxyglucose. Note the high temporal resolution of the time-activity curves, with clearly separated activity peaks in the right and left ventricular blood pool.

(Courtesy of Michael Kreissl.)

IMAGING THE CARDIOVASCULAR SYSTEM IN MICE AND RATS

Experimental Coronary Occlusions in the Rat

The extent and severity of regional myocardial perfusion defects that are experimentally induced by transient coronary occlusions in rats can be accurately determined with microPET.¹⁵ In Figure 11-9, myocardial perfusion was imaged with intravenous ¹³N-ammonia at baseline, 45 minutes later during a 20-minute coronary occlusion, and again 45 minutes following reperfusion. Figure 11-10 shows the corresponding polar map displays. The location and extent of the image perfusion defects corresponded well with the actual regions observed on postmortem stained tissue. Using a threshold of 50% of the maximum myocardial activity concentration as the lower limit of normal, the size of the flow defect derived from polar maps of myocardial perfusion correlated linearly with the defect sizes determined by postmortem examination (Fig. 11-11).

Figure 11-9 Serial images of the regional myocardial perfusion with intravenous ¹³N-ammonia in a rat. Selected short axis slices through the left ventricular myocardium are shown. The initial image obtained at baseline depicts the homogenously distributed myocardial blood flow. The images in the center were recorded after a second ¹³N-ammonia injection during a transient coronary occlusion. The severe perfusion defect is seen in the base portion of the anterior wall (arrows). The defect has resolved completely at 45 minutes after release of the coronary occlusion, as seen on the third image set (right panel) obtained after another intravenous dose of ¹³N-ammonia.

(Reproduced with permission from Kudo T, Fukuchi K, Annala AJ, et al: Noninvasive measurement of myocardial activity concentrations and perfusion defect sizes in rats with a new small-animal positron emission tomograph, Circulation 106:118–123, 2002.)

Figure 11-10 Polar map display of the distribution of myocardial blood flow recorded serially before, during, and following a transient coronary occlusion in the rat depicted in Figure 11-7. Note the perfusion defect on the polar map during the occlusion, as indicated in blue.

Figure 11-11 Extent of perfusion defects induced by coronary occlusions in nine rats. Note the close linear correlation between the PET image–derived defect extents and the perfusion defect sizes determined following blue dye staining and postmortem measurements.

Studies in Genetically Modified Mice

Initial studies have demonstrated the possibility of phenotyping genetically modified mice. For example, cardiac myocyte-specific excision of the murine β-1 integrin gene results in extensive myocardial fibrosis.²¹ In adulthood, the mice develop a form of dilated cardiomyopathy and symptoms of congestive heart failure. MicroPET imaging with ¹⁸F-deoxyglucose in β-1-integrin knockout mice revealed heterogeneously distributed tracer uptake consistent with either replacement of myocytes by fibrotic tissue or as a consequence of regional alterations in substrate metabolism (Fig. 11-12). Assessment of myocardial perfusion with ¹³N-ammonia in the β-1-integrin knockout mouse further revealed heterogeneously distributed myocardial blood flow in addition to patchy regions of reduced tracer uptake.

Figure 11-12 Transaxial images of myocardial perfusion and of exogenous glucose utilization, obtained with intravenous ¹³N-ammonia and ¹⁸F-deoxyglucose in a β-1 integrin knockout mouse. Note the left ventricular enlargement and the heterogeneous tracer retentions.

(Reproduced with permission from Schelbert HR, Inubushi M, Ross RS: PET imaging in small animals, J Nucl Cardiol 10:513–520, 2003.)

Development of Novel Radiotracer Assay Approaches

Noninvasive, image-based probe systems for studying location, extent, magnitude, and duration of gene expression have also been explored and are now being used in small animals. One approach used in oncologic studies in mice has been the PET reporter gene/reporter probe system.⁹–^11,²² In one of these systems, a mutant herpes simplex virus thymidine kinase gene is delivered by an adenoviral vector and represents the PET reporter gene. If expressed in tissue, thymidine kinase phosphorylates the radiolabeled 9-(4-[¹⁸F]-fluoro-3 hydroxymethylbutyl) (guanine) (¹⁸FHBG) that serves as the PET reporter probe. Because the phosphorylated product of ¹⁸FHBG is relatively impermeable to the cell membrane, it becomes effectively trapped in tissue and thus can be imaged with PET. Details on this in vivo assay approach and on other probe systems are described in several recent reviews.^6,^9,^10,²³

Pilot studies with in vitro imaging approaches had explored the feasibility of adapting these probe systems to the heart.²⁴ Optical and microPET imaging approaches have more recently been utilized to examine the feasibility of using these probe systems for in vivo studies of the heart. In these investigations, the PET reporter gene was introduced into the myocardium by direct needle injection during a thoracotomy. Images of the PET reporter probe were then obtained several days after transfection.^25,²⁶ Another study in rats explored quantitative aspects of the probe system (Fig. 11-13).²⁷ The site of the reporter gene expression in the myocardium as identified by microPET correlated closely with that determined postmortem. In vivo measured activity concentrations of the PET reporter probe accurately represented those measured by in vitro tissue counting and, importantly, correlated with the tissue activity of the expressed thymidine kinase. Further, as shown in Figure 11-14, repeat imaging up to 17 days following transfection of the PET reporter gene defined the time course of the gene expression. The expression was highest on days 3 and 5, was still identifiable in some animals between days 10 and 14, but was no longer observed on day 17.

Figure 11-13 Retention of the PET reporter probe ¹⁸FHBG in the myocardium of a rat 3 days following transfection of herpes simplex virus type into the myocardium. The upper row shows contiguous ¹³N-ammonia images of the myocardium, the middle row depicts the retention of ¹⁸FHBG in the myocardium, and the bottom row, the superimposed images.

(Reproduced with permission from Inubushi M, Wu JC, Gambhir SS, et al: Positron-emission tomography reporter gene expression imaging in rat myocardium, Circulation 107:326–332, 2003.)

Figure 11-14 Repeat images of the myocardial retention of ¹⁸FHBG and myocardial perfusion in a rat. The images show the ¹⁸FHBG retention highest on day 3 of the transfection; the activity subsequently declines and is no longer seen on day 14.

(Reproduced with permission from Inubushi M, Wu JC, Gambhir SS, et al: Positron-emission tomography reporter gene expression imaging in rat myocardium, Circulation 107:326–332, 2003.)

PET imaging studies with the reporter gene/reporter probe system in rats also appear promising for investigations on cardiac cell therapy.²⁸ Embryonic rat cardiomyoblasts were transfected with the PET reporter gene and were delivered by direct needle injection into the myocardium of rats. Engraftment and survival of the transplanted myoblasts were then demonstrated noninvasively and repeatedly by microPET imaging of the myocardial retention of the intravenously delivered PET reporter probe.

Quantification of Functional Processes

Adoption of PET’s advantages and full quantitative capabilities to the cardiovascular system of mice and rats remains difficult and is still incomplete. The potential of PET for measuring rates of functional processes is based on the ability to accurately measure tracer activity concentrations in the myocardium and in arterial blood or left ventricular blood pool and to dynamically monitor their changes over time. Major limitations have been the small size of the left ventricular cavity, the thin left ventricular wall, and the rapid heart rates in rodents (~400 to 600 beats/min). Therefore, measurements of tracer activity concentrations in the less than 1-mm-thick myocardium require substantial corrections for partial volume effects that might lead to considerable errors. The small inner diameter of the LV cavity together with the rapid heart rate causes considerable spillover of activity between myocardium and blood, and results in considerable contaminations of counts recovered from regions of interests assigned to the myocardium and the blood pool. Difficulties in serial sampling of arterial blood needed for validation of noninvasively derived radiotracer input functions, mostly due to the small blood volume in mice or rats, has further limited development and implementation of tracer kinetic approaches in small animals. Nevertheless, significant advances have been made over the last couple of years in obtaining accurate measurement of the blood input function, either by direct arterial blood sampling ^29,³⁰ or by derivation of the input function from PET images. A recent study by Fang et al.³¹ validated a new method for deriving the arterial input function from dynamic PET image data that is less susceptible to spillover (SP) and partial volume (PV) effects. This method accounts for the SP and PV effects in a physiologic model to obtain a model-corrected input function that compared well with the physically derived arterial blood samples. These investigators speculated that the addition of cardiac and respiratory gating, in combination with iterative algorithms such as MAP, would further reduce the SP and PV effects in the image-derived input function. This is supported by recent studies at the University of Virginia that demonstrated that high-resolution iterative reconstruction algorithms, combined with cardiac gating, minimized SP from the blood pool to the myocardium and vice versa.³² Partial volume corrections for the blood pool were small because the dimensions were approximately two times the intrinsic spatial resolution of the Siemens Focus 120 scanner.³³ Performing PV corrections for the myocardium enabled accurate estimates of the kinetic rate constants and hence the rates of glucose uptake, K_i, and utilization, rMGU, using a three-compartment model approach in a mouse model of pressure-overload left ventricular hypertrophy produced by transverse aortic constriction (TAC) over 4 weeks (Fig. 11-15).

Figure 11-15 A, End-diastolic PET transverse slices from sham-operated and transverse aortic constricted (TAC) mice that developed left ventricular hypertrophy (LVH) over 4 weeks. B, Measured K_i (mL/g/min) and rMGU (micromoles/g/min) values obtained from PET images, showing increased rates of myocardial glucose uptake and utilization in the TAC mice with LVH. *P < 0.001 compared to sham.

Measurements of Myocardial Activity Concentrations

Regional radiotracer activity concentrations can be measured in absolute units in the myocardium of rats.¹⁵ Following intravenous injection of ¹⁸F-deoxyglucose and tracer accumulation in the myocardium, regions of interest were assigned to the left ventricular myocardium on the attenuation-corrected images. The PV-related underestimation of true tissue activity concentrations was corrected for with a recovery coefficient of 0.59, a value derived from postmortem measurements of the myocardial wall thickness and from the performance characteristics of the imaging system. Additional adjustments were needed for method-related differences in activity measurements between microPET and in vitro well counting of tissue samples. A calibration factor adjusted for differences in in vitro and in vivo measurements in tissue concentrations was obtained from images of a “rat phantom” that mimics the distribution of radioactivity in the body of the rat, and that contained in the region of the heart a solution of known radioactivity concentration.¹⁵ Once corrected for PV effect and for differences in measurement sensitivities, values of tissue activity concentrations determined in vivo correlated closely and linearly with those obtained by postmortem well counting of myocardial tissue samples and ranging from 0.89 to 10.46 MBq/g. (Fig. 11-16).

Figure 11-16 Comparison of ¹⁸F-deoxyglucose concentrations in vivo and by well counting.

(Reproduced with permission from Kudo T, Fukuchi K, Annala AJ et al: Noninvasive measurement of myocardial activity concentrations and perfusion defect sizes in rats with a new small-animal positron emission tomograph, Circulation 106:118–123, 2002.)

Assays of Time-Dependent Changes in Tissue Activity Concentrations

Other studies have explored semiquantitative PET approaches for identifying alterations in the myocardium’s substrate metabolism as phenotypic consequences of gene alterations in mice. In addition to ¹⁸F-deoxyglucose, these studies also employed ¹¹C-palmitate as a tracer of fatty acid metabolism in transgenic mice with cardiac-specific overexpression of peroxisome proliferator–activated receptor alpha (PPARα).²⁰ Beginning with the intravenous tracer administration, serial images were recorded for up to 40 minutes, and whole heart time-activity curves were generated. Consistent with the expression of PPARα target genes involved in free fatty acid and glucose metabolism, myocardial ¹⁸F-deoxyglucose uptake in these transgenic mice was found to be diminished, whereas ¹¹C-palmitate uptake and clearance as indices of myocardial fatty acid uptake and oxidation were found to be increased. The images and derived time-activity curves in these mice clearly indicated trends of the alterations in substrate metabolism that were in agreement with findings by in vitro assays of the heart.

Measurements of Absolute Blood Flow in the Myocardium of Rats

The possibility to extract time-dependent activity concentrations from the serially acquired images in both arterial blood and myocardium has also been demonstrated in the somewhat larger heart of rats.³⁴ Time-activity curves for myocardium and left ventricular blood pool were derived in these studies with ¹³N-ammonia and ¹¹C-acetate from regions of interest assigned to the serially acquired images. Model fitting of the time-activity curves yielded estimates of myocardial blood flow that averaged 4.56 mL/m/g, a value consistent with that reported for rat myocardium.³⁵ Further, the myocardial clearance rate of ¹¹C-acetate as an index of regional myocardial oxidative metabolism averaged 1.48 min⁻¹ and thus was markedly higher than clearance rates of 0.48 ± 0.004 min⁻¹ reported for human myocardium at baseline.³⁶ Although the noninvasively derived estimates of myocardial blood flow and oxidative metabolism lacked independent validation in this study in rats, and although heart rate and blood pressure were not reported, the high values of blood flow and oxidative metabolism, at least relative to those in humans, are consistent with the high heart rate and consequently high levels of cardiac work in the rodent heart. The findings of this study support the possibility of extracting quantitative information on functional processes of the myocardium in rodents. However, estimates of regional myocardial blood flow and oxidative metabolism varied considerably between myocardial regions, as for example the septum and the lateral wall, highlighting the difficulties and current limitations in extracting accurate and uncontaminated regional tracer activity concentrations from the serially acquired images.

Semiquantitative Estimates of Changes in Myocardial Blood Flow in Mice

Experimentally induced genetic alterations in mice may affect myocardial blood flow homogenously rather than regionally. Demonstration of phenotypic consequences of genetic alterations therefore requires estimates of myocardial blood flow in absolute units rather than identification of regional alterations in the relative flow distribution of blood flow. Additional factors complicating such measurements include the already high levels of myocardial blood flow at rest. Even though myocardial blood flow has not been measured in vivo in mice, values may be as high as 4 mL/m/g, judging from the high resting heart rates of 400 to 600 beats/min in mice and from estimates of 3.5 to 4.0 mL/min/g as observed in rat myocardium.³⁵ Consequently, the myocardial net retention of most tracers of myocardial blood flow resides in the plateau region of the nonlinear relationship between the tracer net retention and myocardial blood flow. Further increases in blood flow, as for example induced by pharmacologic vasodilatation, would be associated with only small increments in radiotracer uptake. Because of this, one microPET study in mice with ¹³N-ammonia attempted to lower rather than increase myocardial blood flow, and examined whether associated decreases in flow could be demonstrated noninvasively with microPET imaging.³⁷ With an approximate 50% reduction in the heart rate after intraperitoneal administration of the α²-adrenoreceptor agonist clonidine, the myocardial retention of ¹³N-ammonia expressed as percent of the total injected dose declined significantly but by modest 14%. Even if the 50% reduction in heart rate produced more substantial decreases in myocardial blood flow, the observed modest decrease in tracer retention most likely was because of the nonlinear relationship between tracer net uptake and myocardial blood flow, at least in the hyperemic flow range.

Limitations and Challenges

With the recent development of high-resolution and high-sensitivity small-animal PET scanners, accurate imaging of the murine heart is now possible. Obtaining the arterial blood input function has been a challenge in dynamic PET imaging of the mouse heart in the past, but adoption of the factor analysis approach to serially acquired microPET images appears promising for determining accurate estimates of the arterial tracer input function and likely also of the myocardial tissue response.³⁸ The high-resolution iterative ordered subset expectation maximization (OSEM)-MAP reconstruction algorithms improve the reconstructed image resolution as compared to the conventional analytic filtered backprojection algorithm.³⁹ Nevertheless, even the relatively high spatial resolution of the dedicated microPET systems remains low relative to the myocardial wall thickness of about 0.9 mm in the mouse. Accordingly, the necessary partial volume corrections have to be applied for accurate estimates of the myocardial tracer uptake. This along with cardiac and/or respiratory gating enables accurate estimates of the blood input function.³²

Other refinements to existing methodologies include small-animal PET/CT and PET/MRI approaches and thus coregistration of functional with anatomic information (Fig. 11-17).^40,⁴¹ Besides offering improved accuracy in localizing regional functional processes, the addition of anatomic information may prove useful for more accurate determination of regional tracer activity concentrations. With these method- and instrumentation-related refinements, it will likely become possible to fully utilize the potential of PET in mice as a noninvasive probe of a broad range of tissue assays. Such assays include tracer kinetic approaches already fully validated and employed in large-animal studies and in humans for measurements of myocardial blood flow, oxidative metabolism and consumption, and myocardial substrate metabolism.

Figure 11-17 Transaxial PET/CT images in a mouse in the prone position. The computed tomography (CT) image is shown on the left, the corresponding transaxial positron emission tomography (PET) image in the middle, and the PET/CT fusion image on the right. Note the ¹⁸F-deoxyglucose uptake in the left ventricular myocardium.

MICRO–SINGLE-PHOTON EMISSION COMPUTED TOMOGRAPHY

Imaging with microSPECT offers a number of advantages over microPET for cardiovascular molecular imaging using small-animal models. One advantage is the ready availability of isotopes with excellent properties for imaging. Commonly used isotopes for medical imaging such as ²⁰¹Tl, ^99mTc, ¹¹¹In, and ¹²³I can be ordered from local radiopharmacies or commercial vendors and are fairly inexpensive. Furthermore, new single-photon, radiolabeled probes can usually be developed quite easily because the chemistry techniques for iodination (¹²³I) or for chelating metals such as ^99mTc or ¹¹¹In and linking them to a specific protein or ligand are well established. Many of the radiolabeled probes can be prepared when needed in the research laboratory using simple kit-based labeling procedures. PET probes, on the other hand, require access to an on-site cyclotron for isotopes with short half-lives (i.e., ¹⁵O, ¹³N, ¹⁸F). Although ¹⁸F-FDG can also be ordered from regional suppliers, the synthesis of other PET probes on-site can be difficult or require expensive synthesis equipment. Another advantage of microSPECT is that the isotopes typically have longer half-lives (hours to days) that permit repeated monitoring of tracer activities over time. This is especially useful for longitudinal studies or for tracking the movement of specific cell types (e.g., inflammatory cells or stem cells). Another major advantage of microSPECT is the ability to discriminate between the energy photopeaks of different isotopes, allowing one to image multiple isotopes simultaneously in the same animal. For example, ^99mTc-sestamibi perfusion could be imaged at the same time as an ¹¹¹In-labeled tracer of hypoxia. With PET, if multiple processes are to be measured in the same animal, imaging must be performed sequentially rather than simultaneously. Finally, another reason for the growing interest in SPECT-based imaging systems is the ability to achieve greater spatial resolution with SPECT. PET spatial resolution is ultimately limited by the diffusion of high-energy positrons emitted from the radioisotope before annihilation (i.e., generation of two 511-keV gamma rays). As mentioned earlier in this chapter, the spatial resolution of the current generation of microPET scanners is on the order of 1.2 to 1.5 mm FWHM; microSPECT systems do not suffer from this inherent limitation and ultimately can achieve even higher spatial resolution. Commercial microSPECT systems from several vendors are currently achieving submillimeter spatial resolution.

MicroSPECT Instrumentation

There have been tremendous advances made in small-animal SPECT instrumentation over the past several years in both hardware and software. As with clinical SPECT cameras, a typical microSPECT system often consists of a rotating gantry with one or more single-photon detectors. Hybrid systems are also available whereby the gantry is equipped with a microfocus x-ray source and detector that allow dual microSPECT/CT studies to be performed in the same animals. The single-photon detectors can be based on scintillation crystals (NaI) coupled with photomultiplier tubes, or they can be solid-state such as cadmium telluride detectors. Single-photon gamma cameras require some form of collimation for accurate localization of the position of origin of a gamma photon. The collimators are interchangeable, depending upon the resolution and FOV required, and can be either parallel-hole for whole body imaging or, to achieve high spatial resolution, pinhole collimators with 0.5- to 2.0-mm apertures; the latter are most commonly employed. Because the photon detection efficiency is diminished with the use of a collimator, state-of-the-art high resolution microSPECT systems often employ multiple detectors, with each detector having more than one pinhole to improve sensitivity. Sensitivity and magnification are also enhanced by the use of a relatively tight radius of rotation whereby the animal is in close proximity to the pinhole apertures.

One drawback to gantry-based microSPECT systems is that they are not optimal for imaging rapid, dynamic processes, because there is a finite time required to rotate the detector head to the next position. The investigators from the Radiology Research Laboratory at the University of Arizona Health Sciences Center developed a unique “gantryless” stationary SPECT system called FASTSPECT for small-animal imaging. The system consists of 24 modular gamma cameras arranged in two circular arrays: one array of 11 cameras and a second array of 13 cameras. Each of the modular cameras consists of a 10 cm × 10 cm NaI(Tl) scintillation crystal coupled to a photomultiplier tube with an optical light guide. The collimator consists of 24 1-mm-diameter pinholes drilled into a cylindrical aperture in such a manner that an object placed in the center of the FOV is simultaneously projected onto the center of each of the 24 detectors. Compared with a single-pinhole aperture and detector system, having multiple pinholes and detectors markedly improves the overall system sensitivity. Importantly, because the detectors are placed in a circular pattern around the animal, there is no need to rotate a gantry. A full 3D dynamic image set can be acquired in only 5 to 10 minutes. Another unique stationary approach to rapid dynamic microSPECT imaging is a second-generation system called U-SPECT-II that was developed by Beekman and colleagues and is now available commercially (Fig. 11-18A).⁴² The system consists of interchangeable tungsten ring collimators with 75 pinholes (see Fig. 11-18B) that yield non-overlapping projections of the imaging volume onto three 508 × 381 × 9.5 mm detectors. Using the mouse collimator with 0.35-mm pinhole apertures, a resolution of less than 0.35 mm was obtained using a rod phantom filled with ^99mTc (see Fig. 11-18C). The U-SPECT-II system acquires the data set simultaneously in list mode, and both ECG and respiratory gating can be performed. Figure 11-19 is an example image of myocardial perfusion (^99mTc-tetrofosmin) in a mouse heart at end-diastole and end-systole that was obtained with U-SPECT. The ultra-high-resolution images clearly show both papillary muscle and the thin right ventricular wall of the mouse heart. With U-SPECT-II, CT imaging can also be performed using a separate CT module, and the reconstructed image data sets can be co-registered and fused with postacquisition software.

Figure 11-18 A, U-SPECT-II system from Molecular Imaging Labs Inc. B, Cylindrical tungsten collimators available in two different sizes for mice and rats, respectively. C, High-resolution technetium-99m hot rod phantom acquired using USPECT-II with a cylindrical collimator consisting of 75 0.35-mm pinholes.

(Image courtesy of MILabs Inc.)

Figure 11-19 End-diastolic (ED) and end-systolic (ES) images of perfusion (technetium-99m-tetrofosmin) in a mouse heart obtained using the U-SPECT camera. The ED and ES frames were obtained from a gated SPECT study.

(Image courtesy of MILabs Inc.)

Uses of MicroSPECT Instrumentation for Cardiovascular Molecular Imaging

In the remainder of this chapter, we will give a few specific examples of how microSPECT imaging is playing an important role in cardiovascular research using rodent models.

Myocardial Perfusion and LV Function

State-of-the-art, standalone microSPECT systems have many attractive features, but for some applications, excellent results have been obtained using clinical SPECT cameras with pinhole collimators. In one study, Maskali et al. used such a system to image myocardial perfusion and LV function by gated SPECT in a rat model of postinfarction LV remodeling over 12 weeks.⁴³ The rats were serially imaged over 12 weeks using 400 to 700 MBq of ^99mTc-sestamibi injected intravenously (300-500-μL volume) at each time point. The camera system consisted of a single-head clinical SPECT camera with a 3-mm diameter pinhole and a magnification factor of approximately 4.5 (195 mm focal length, 43 mm radius of rotation). Image reconstruction was performed using an OSEM iterative algorithm, and LV function was determined using clinical QGS software. The investigators demonstrated that in the post-MI setting, the initial (48-hour) infarct size measured by sestamibi SPECT predicts the extent of subsequent LV remodeling in rats (Fig. 11-20).⁴³

Figure 11-20 Sestamibi gated SPECT in a rat model of postinfarction LV remodeling. The left panel depicts perfusion at 48 hours and 12 weeks in rats with large or limited infarct size at 48 hours. The right panel depicts changes in LV volume and functional parameters in the same rats, as compared with sham-operated controls (open triangles), as well as in rats with no detectable infarcts (open circles).

(Reproduced with permission from Maskali F, Franken PR, Poussier S, et al: Initial infarct size predicts subsequent cardiac remodeling in the rat infarct model: An in vivo serial pinhole gated SPECT study, J Nucl Med 47:337–344, 2006.)

Quantitative Assessment of Myocardial Viability Using Technetium-99m-Glucarate

In a recent study by Liu et al. at the University of Arizona,⁴⁴ the FASTSPECT camera (described earlier) was used to image ^99mTc-glucarate uptake in the hearts of rats as a measure of the extent of myocardial necrosis following experimental coronary occlusion (Fig. 11-21). The investigators demonstrated that administration of the α1 adenosine receptor agonist 2-chloro-N6-cyclopentyladenosine reduced infarct size to the same extent as that obtained with ischemic preconditioning (IPC), and that the degree of protection was lost in the presence of a nonselective adenosine receptor antagonist, SPT.⁴⁴ This study is an excellent example of how quantitative microSPECT imaging can be used as an endpoint when assessing the efficacy of new therapies in experimental models.

Figure 11-21 Representative heart slices stained with triphenyl tetrazolium chloride (TTC) to delineate areas of necrosis and corresponding short-axis slice images of technetium-99m-glucarate. Infarct size quantification by analysis of hot spot activities demonstrated that infarct size in untreated animals (open bar) was reduced by ischemic preconditioning (IPC). Administration of the adenosine A1 agonist CCPA reduced infarct size to the same degree as IPC (light grey bar), whereas the IPC protection was abrogated in the presence of the nonselective adenosine receptor antagonist SPT (dark grey bar).

(Image courtesy of Zhonglin Liu, PhD, University of Arizona Health Sciences Center, Dept. of Radiology.)

Dual-Isotope Imaging of Myocardial Perfusion and Inflammation

Sato et al. reported on the use of a radiolabeled anti-tenascin-C monoclonal antibody for imaging myocardial inflammation.⁴⁵ In a rat model of autoimmune myocarditis produced by immunization with porcine myosin twice over a 7-day interval, these investigators demonstrated high accumulation of ¹¹¹In-labeled anti-tenascin-C antibody in the inflamed myocardial region. Using a dual-isotope approach with ¹¹¹In-anti-tenascin-C and ^99mTc-sestamibi (Fig. 11-22) and SPECT imaging, these investigators demonstrated that the focal uptake of ¹¹¹In-anti-tenascin-C antibody in the septal wall could easily be visualized in vivo. The camera system utilized for these experiments was a three-headed clinical SPECT camera with 1-mm pinholes on each detector. The fused image shown in panel C of Figure 11-22 highlights the multi-isotope advantage of SPECT instrumentation, whereby the sestamibi perfusion image can serve as an anatomic marker that is helpful for localizing the focal uptake of the “hot spot” tracer in the myocardium. Even with dual-modality hybrid microSPECT/CT systems where the CT image is extremely valuable as an anatomic reference point, the use of different isotopes to simultaneously image different aspects of cardiovascular system under study is advantageous.

Figure 11-22 Dual-isotope SPECT of inflammation (¹¹¹In-anti-tenascin C Fab), perfusion (^99mTc-sestamibi), and fused images in a rat model of experimental myocarditis. The use of a myocardial perfusion agent provides anatomic localization for a “hot spot” tracer that is labeled with a second isotope.

(Figure adapted and reproduced with permission from Sato M, Toyozaki T, Odaka K, et al: Detection of experimental autoimmune myocarditis in rats by ¹¹¹In monoclonal antibody specific for tenascin-C. Circulation 106:1397–1402, 2002.)

Monocyte Trafficking in Atherosclerosis

In another recent study, Kircher et al. used a commercially available dual-modality microSPECT/CT (X-SPECT, Gamma Medica Ideas, Inc.) to follow monocyte trafficking in a murine model of atherosclerosis.⁴⁶ ApoE^−/− subjects were fed for several weeks on the Western diet to produce atherosclerotic plaques. Monocytes were isolated from donor wild-type mice and labeled ex vivo with ¹¹¹In-oxine. Approximately 3 × 10⁶ labeled cells were then injected into a tail vein, and serial microSPECT/CT imaging was performed over 5 days. Figure 11-23 depicts the CT (row A), fused microSPECT/CT (row B), and postmortem phosphor plate autoradiographs of the excised aortas (row C). The panel on the left shows an ApoE^−/− mouse that received the indium-labeled monocytes. As shown, the monocytes homed in on the atherosclerotic region of the ascending aorta, which was confirmed by immunohistochemistry. The binding specificity was demonstrated using a mouse that received a fraction of cells that had been depleted of monocytes (middle panel) and with a negative control wild-type mouse with no atherosclerosis (right panel). In both of these negative controls, no uptake of the indium-labeled tracer was observed in the aortic arch. The longer half-life of the single photon isotope ¹¹¹In (2.7 days) allowed for the monitoring of the same population of labeled cells over time. As mentioned earlier, this ability to track cells over a period of days is one of the advantages of using microSPECT over microPET.

Figure 11-23 Example of imaging monocyte cell trafficking in a murine model of atherosclerosis. Radiolabeled monocytes homed to the ascending aorta in ApoE^−/− mice (left column), but not in mice where the injected cell fraction had been depleted of monocytes or in wild-type mice with no atherosclerosis. Cells labeled with ¹¹¹In can be followed over several days with microSPECT imaging to monitor their fate and ultimate site of localization.

(Figure reproduced with permission from Kircher MF, Grimm J, Swirski FK, Libby P, et al: Noninvasive in vivo imaging of monocyte trafficking to atherosclerotic lesions, Circulation 117:388–395, 2008.)

Imaging Matrix Metalloproteinase Expression in Vascular Remodeling

Zhang et al. utilized high-resolution microSPECT/CT to image activated matrix metalloproteinases (MMPs) using a novel ¹¹¹In-labeled tracer (RP782).⁴⁷ ApoE^−/− mice that had been fed a high-cholesterol diet were subjected to left common carotid injury. RP782 was administered at 1, 2, 3, and 4 weeks after injury. The microSPECT camera utilized 1-mm-diameter pinhole collimators. To enhance CT contrast, an infusion of an iodinated contrast agent (iohexol, 100 μL/min) was given. As shown in Figure 11-24, at 3 weeks postinjury, an RP782 hotspot was observed in the injured left carotid artery but not in the uninjured right carotid artery. Importantly, image quantification (panel D) allowed for the determination of the time course of MMP activation. This study provides an excellent example of the importance of using a second modality, in this case CT, to help localize the uptake of the radiolabeled probe. It also demonstrates that the use of an iodinated CT contrast agent can greatly enhance the CT image for this purpose.

Figure 11-24 An example of (A) RP782 microSPECT, (B) CT angiography, and (C) fused microSPECT/CT in vivo imaging at 3 weeks after carotid injury. Arrows point to the injured left (L) and noninjured right (R) carotid arteries. The small hot spot in the abdomen is likely the upper pole of the kidney on the edge of the SPECT field of view and/or a pinhole imaging artifact. S indicates sagittal; C, coronal; and T, transverse slices. D, Image-derived quantitative analysis of background-corrected RP782 carotid uptake. N = 5 to 7 in each group. w, week.

(Figure reproduced with permission from Zhang J, Nie L, Razavian M, et al: Molecular imaging of activated matrix metalloproteinases in vascular remodeling. Circulation 118:1953–1960, 2008.)