The method presented here uses simultaneous positron emission tomography and magnetic resonance imaging. In the cerebral hypoxia-ischemia model, dynamic changes in diffusion and glucose metabolism occur during and after injury. The evolving and irreproducible damage in this model necessitates simultaneous acquisition if meaningful multi-modal imaging data are to be acquired.
Dynamic changes in tissue water diffusion and glucose metabolism occur during and after hypoxia in cerebral hypoxia-ischemia reflecting a bioenergetics disturbance in affected cells. Diffusion weighted magnetic resonance imaging (MRI) identifies regions that are damaged, potentially irreversibly, by hypoxia-ischemia. Alterations in glucose utilization in the affected tissue may be detectable by positron emission tomography (PET) imaging of 2-deoxy-2-(18F)fluoro-ᴅ-glucose ([18F]FDG) uptake. Due to the rapid and variable nature of injury in this animal model, acquisition of both modes of data must be performed simultaneously in order to meaningfully correlate PET and MRI data. In addition, inter-animal variability in the hypoxic-ischemic injury due to vascular differences limits the ability to analyze multi-modal data and observe changes to a group-wise approach if data is not acquired simultaneously in individual subjects. The method presented here allows one to acquire both diffusion-weighted MRI and [18F]FDG uptake data in the same animal before, during, and after the hypoxic challenge in order to interrogate immediate physiological changes.
Worldwide, stroke is the second leading cause of death and a major cause of disability 1. The cascade of biochemical and physiological events that occur during and acutely following a stroke event occurs rapidly and with implications for tissue viability and ultimately outcome 2. Cerebral hypoxia-ischemia (H-I), which leads to hypoxic-ischemic encephalopathy (HIE), is estimated to affect up to 0.3% and 4% of full-term and preterm births, respectively 3,4. The mortality rate in infants with HIE is approximately 15% to 20%. In 25% of HIE survivors, permanent complications arise as a result of the injury, including mental retardation, motor deficits, cerebral palsy, and epilepsy 3,4. Past therapeutic interventions have not proven worthy of adoption as standard of care, and consensus has yet to be reached that the most advanced methods, based on hypothermia, are effectively reducing morbidity 3,5. Other issues of contention include method of administration of hypothermia and patient selection 6. Thus, strategies for neuroprotection and neurorestoration are still a fertile area for research7.
Rat models of cerebral H-I have been available since the 1960s, and subsequently were adapted to mice 8,9. Due to the nature of the model and the location of the ligation, there is inherent variability in the outcome due to difference in collateral flow between animals 10. As a result, these models tend to be more variable compared to similar models such as middle cerebral artery occlusion (MCAo). Real time measurement of physiological changes has been demonstrated with laser Doppler flowmetry as well as diffusion-weighted MRI 11. The observed intra-animal variability in cerebral flow blood during and immediately after hypoxia, as well as in acute outcomes such as infarct volume and neurological deficit, suggest that simultaneous acquisition and correlation of multimodal data would be beneficial.
Recent advances in simultaneous positron emission tomography (PET) and magnetic resonance imaging (MRI) have allowed for new possibilities in preclinical imaging 12-14. The potential advantages of these hybrid, combined systems for preclinical applications have been described in the literature 15,16. While many preclinical questions can be addressed by imaging an individual animal sequentially or by imaging separate animal groups, certain situations – for example, when each instance of an event such as stroke manifests itself uniquely, with rapidly evolving pathophysiology – make it desirable and even necessary to use simultaneous measurement. Functional neuroimaging provides one such example, where simultaneous 2-deoxy-2-(18F)fluoro-ᴅ-glucose ([18F]FDG) PET and blood-oxygen-level dependent (BOLD) MRI has recently been demonstrated in rat whisker stimulation studies 14.
Here, we demonstrate simultaneous PET/MRI imaging during onset of a hypoxic-ischemic stroke in which brain physiology is not at steady state, but instead is rapidly and irreversibly changing during hypoxic challenge. Changes in water diffusion, as measured by MRI and quantified by the apparent diffusion coefficient (ADC) derived from diffusion-weighted imaging (DWI), has been well characterized for stroke in clinical and preclinical data 17,18. In animal models such as MCAo, diffusion of water in affected brain tissue drops rapidly due to the bioenergetic cascade leading to cytotoxic edema 18. These acute changes in ADC are also observed in rodent models of cerebral hypoxia-ischemia 11,19. [18F]FDG PET imaging has been used in stroke patients to assess changes in local glucose metabolism 20, and a small number of in vivo animal studies have also used [18F]FDG 21, including in the cerebral hypoxia-ischemia model 22. In general, these studies show decreased glucose utilization in ischemic regions, although a study using a model with reperfusion found no correlation of these metabolic changes with later infarction development 23. This is in contrast to diffusion changes which have been associated with the irreversibly damaged core 21. Thus, it is important to be able to obtain the complementary information derived from [18F]FDG PET and DWI in a simultaneous manner during the evolution of stroke, as this is likely to yield meaningful information about the progression of injury and the impact of therapeutic interventions. The method we describe here is readily amenable to use with a variety of PET tracers and MRI sequences. For instance, [15O]H2O PET imaging along with DWI and perfusion-weighted images (PWI) from MRI may be used to further explore the development of the ischemic penumbra and validate current techniques within the stroke imaging field.
All animal handling and procedures described herein, and according to the Animal Research: Reporting In Vivo Experiments (ARRIVE) guidelines, were performed in accordance with protocols approved by the Association for Assessment of Accreditation of Laboratory Animal Care (AAALAC) International accredited Institutional Animal Care and Use Committee at the University of California, Davis. Proper surgery should not result in signs of any pain or discomfort in the animal, but proper steps should be taken if these signs are observed, including administration of analgesics or in some cases, euthanasia. The right side of the animals was chosen arbitrarily for the unilateral procedure described.
1. Unilateral Common Carotid Artery (CCA) Ligation
2. Preparation for Imaging: System and Hardware Checks
3. Imaging workflow
After all necessary equipment checks are completed, proceed to imaging as follows:
Figure 1 demonstrates the result of a proper ligation of the common carotid artery, prior to closing the wound with 6-0 silk suture.
In this method, data obtained from imaging is highly dependent upon the temporal arrangement of the experiment, which in turn dictates and is also dictated by experimental limitations including image acquisition schemes and equipment setup. These and other considerations are further explored in the Discussion section. With the protocol described herein, the physical setup of the equipment (Figure 2A) allows for uninterrupted multi-modal image acquisition before, during, and after (Figure 2B) rapid introduction of the hypoxic challenge (Figure 2C).
In this animal model, as with many ischemic stroke models, changes in diffusion are detectable rapidly after insult (see Figure 3A for a representative example). As our method does not fundamentally alter the cerebral H-I model, diffusion changes can be reproduced in a robust manner – Figure 3B demonstrates the evolving percent differences in ADCz (ADC in the z-direction) between the contralateral (non-occluded, left) and ipsilateral (occluded, right) sides of the brain, %L-R, (n = 6 for scan H2, n = 5 for all other time points). As expected, ADC values on the occluded side of the brain decrease as the injury progresses. Figure 3C shows an example coronal slice from the DWI sequence, as well as a sagittal slice demonstrating the limited axial extent of the FOV (8 mm) for the sequence used. Details regarding limitations imposed upon the echo planar imaging (EPI) sequence used for DWI are described in the Discussion section. In short, image quality obtained with the proposed imaging framework is dependent on system performance characteristics, and EPI-based DWI sequences in particular may expose suboptimal hardware conditions or acquisition parameters (see Figure 5B). That significant differences were observed between baseline and subsequent ADC %L-R values (p < 0.05, unpaired t-test) suggests that this is a robust parameter to interrogate using our experimental setup.
Concurrent with changes in ADC, hemispherical differences were observed in the uptake of [18F]FDG after beginning the hypoxic challenge and during scan H2 (%11 mean L-R difference, n=3). In two of three cases, ipsilateral [18F]FDG uptake decreased relative to contralateral uptake after hypoxia (see Figure 4 for a representative example), though this was not true in all cases likely due to animal variability. Figure 5A shows an example where the relative difference in [18F]FDG uptake between the two hemispheres was not as expected in one animal (blue). Figure 5A also shows an example where, while [18F]FDG uptake was as expected following hypoxia, the animal died at the end of scan H2.
Figure 1. Example of the right common carotid artery ligated with 6-0 silk suture. The animal is supine with its head pointed towards the bottom of the image. The area around the incision has been depilated, and the incision is being held open with forceps for visualization. Please click here to view a larger version of this figure.
Figure 2. (A) Representative diagram of the physical arrangement of equipment. The PET insert is positioned in the bore of the magnet, and the MRI coil is in turn positioned in the bore of the PET insert. The animal bed, along with physiological monitoring (respiration pad not shown), anesthesia line, and IV catheter runs into the bore as shown. The dotted ring denotes a safety margin for the stray magnetic field – it may be necessary to place equipment with magnetic components outside of this region but within the MRI room (following all safety precautions). (B) Diagram summarizing the temporal progression of the experiment. (C) Representative results of initial changes in O2 level delivered to the animal immediately after the start of the hypoxia challenge. Within approximately 1 min, hypoxic conditions can be achieved, as measured by an O2 meter placed in a 0.5 L induction box (not shown), in-line with the anesthesia system. Please click here to view a larger version of this figure.
Figure 3. (A) Example of parametric ADCz maps acquired at baseline and through post-hypoxia. (B) Plot showing %L-R difference in ADCz from baseline to post-hypoxia. Asterisks indicate a significant difference (p < 0.05, unpaired t-test) compared to baseline value. Error bars represent +/- one standard deviation. (C) Example of an EPI-DWI acquisition (axial, sagittal, and 3D views to show extent of the FOV). Please click here to view a larger version of this figure.
Figure 4. (A) Coronal and transverse slice of an animal showing [18F]FDG uptake. The PET image is in the foreground and is registered and fused with an anatomical MRI image in the background for visualization. The PET data are summed across all frames. (B) In the same animal, [18F]FDG time activity curve for the contralateral hemisphere (blue) and ipsilateral hemisphere (red). Please click here to view a larger version of this figure.
Figure 5. (A) Time activity curves of contralateral (solid) and ipsilateral (dotted) hemisphere [18F]FDG uptake – shown on the same axis are examples of an unexpected [18F]FDG time activity curve (blue) and animal death at the end of H2 (at 45 min, green). (B) Ghosting artifacts due to potential hardware-based RF faults. Please click here to view a larger version of this figure.
Imaging Acquistion Parameters and Hardware Acquisition | |
Diffusion MRI (EPI-DWI) | |
Acqusition time | 15 min |
Matrix size | 256 x 64 |
Slices | 10 |
FOV | 30 x 14 x 8 mm |
Voxel size | 0.117 x 0.219 x 0.8 mm |
Effective spectral bandwidth | 150 kHz |
TE | 41 msec |
TR | 3,000 msec |
Averages | 6 |
k-space segments | 16 |
b-values | 0, 400, 800 sec/mm2 |
Anatomical MRI (MSME) | |
Acquisition time | 5 min |
Matrix size | 256 x 256 |
Slices | 16 |
FOV | 30 x 22 x 12.8 mm |
Voxel size | 0.117 x 0.086 x 0.8 mm |
TE | 14 msec |
TR | 1,000 msec |
Averages | 1 |
Repetitions | 1 |
Point-Resolved Spectroscopic Scan (PRESS) |
|
Acquisition time | 15 s |
Voxel size | 3.9 x 6 x 9 mm |
TE | 20 msec |
TR | 2,500 msec |
Averages | 6 |
FieldMap | |
Acquisition time | 1 min 21 sec |
1st TE | 1.49 msec |
2nd TE | 5.49 msec |
TR | 20 msec |
Averages | 1 |
PET Acquisition, Histogram, and Reconstruction Parameters |
|
Tracer | [18F]FDG |
Infusion rate | 4.44 µL/min |
Acquisition time | 60 min |
Image size per slice | 128 x 128 |
Slices | 99 |
Voxel size | 0.4 x 0.4 x 0.6 mm |
Dynamic framing | 12 x 300 sec |
Reconstruction type | OS-MLEM (6 subsets, 6 iterations) |
Table 1. MRI pulse sequence parameters for scans described in protocol, and PET acquisition, histogram, and reconstruction parameters.
Simultaneous anatomic MRI, and dynamic DWI-MRI and [18F]FDG PET data were successfully acquired from experimental animals during hypoxic challenge following common carotid artery ligation. This represents a powerful experimental paradigm for multimodal imaging of the rapidly evolving pathophysiology associated with ischemic insults in the brain and could readily be extended to study other PET radiotracers (for example markers of neuroinflammation) and MRI sequences, as well as the impact of interventional strategies during or shortly after ischemic challenge.
For successful execution of simultaneous PET/MRI imaging during hypoxic challenge in the cerebral H-I model, logistics must be considered and the methods adjusted accordingly. Factors potentially affecting the temporal arrangement of the experiment include, but are not limited to: 1) source of radioactivity – depending on the radiotracer used, half-life of the radionuclide, and specific activity requirements, this may affect the possible total number of animals imaged; 2) room layout – this may affect the lengths of tubing used and thus the injected dose, or may require additional steps to maintain injected dose. This may also have a small effect on the time to reach equilibrium for gas mixtures in the anesthesia tubing; 3) animal weight – some institutions may impose a limit on the total injected volume for survival procedures (e.g., less than 1% of body weight), in turn potentially affecting tubing length and infusion pump rate settings; 4) tracer delivery – a bolus, infusion, or bolus plus infusion delivery may be used, as determined by radiotracer pharmacokinetics and expected observable changes – the latter two are especially useful to follow dynamic changes 24.
Design of the PET and MRI image acquisition protocols, particularly given the limited time with which to work, is another crucial factor in this experiment. If using an echo-planar imaging (EPI)-based DWI sequence (EPI-DWI) as presented here, important considerations include scan duration, field of view, and diffusion gradient weighting and directions. While adjusting these parameters, inherent issues with EPI-DWI must also be addressed, including ghosting, signal dropout, and gradient duty cycle limitations. The use of respiratory gating may be used to address issues due to motion. Table 1 describes the MRI acquisition parameters used along with information on the PET hardware, acquisition parameters, and tracer delivery parameters. For quantification of PET data, detector normalization must be applied. Though not done in our case, further steps can be taken to achieve more accurate quantification, including attenuation correction using segmented MRI data and scatter correction. The former may not be necessary in small animals as the degree of attenuation is small and can be accounted for using similar-sized calibration objects. Depending on the MRI sequence used, it may also be necessary to consider any significant BOLD effects on T2* 25. In addition, the effect of anesthetic and carrier gas on blood glucose may need to be considered when using [18F]FDG 26.
Checks should be carried out to ensure there is no significant mutual interference between the PET and MRI systems, or between the imaging systems and other instrumentation used in the experiment. In our experience, there was no significant difference in the PET or MRI image quality when acquired individually or simultaneously, although we have observed momentary loss in counts in the PET system due to spurious signals in the PSAPD-based detectors induced by rapid gradient switching, an effect that has been noted by others 12. Another issue observed was RF noise from the infusion pump power supply disturbing PET detector acquisition resulting in loss of data. This was resolved by replacing the original AC adaptor with a laboratory-quality power supply. More PET/MRI hardware configurations are described in the literature, and adjustments to this protocol may be required to accommodate unique setups 12,27.
The imaging workflow may be modified in order to optimize conditions for different MRI pulse sequences or PET tracers and acquisition schemes. For instance, severity of injury in the cerebral H-I model has been shown to be modulated by, among other conditions, the duration of hypoxia 11. Increasing the length of the hypoxic challenge may allow acquisition of DWI data at finer temporal resolution, or allow for more robust hemispherical uptake comparisons for PET tracers. Other aspects of the protocol may be adjusted based on available resources and personnel. For instance, surgeries may be staggered and run parallel to imaging sessions in order to reduce the variability in the time between CCA ligation and hypoxia.
In this protocol, simultaneous PET and MRI acquisition, in addition to the physiological challenge, imposes mutual limits on one another in terms of timing. In optimizing the EPI-DWI sequence, it was found that having additional diffusion directions while maintaining image quality would increase acquisition time beyond acceptable limits for performing multiple acquisitions during the hypoxic challenge. Thus, diffusion gradients were applied solely along the z-axis. In addition, the adaptation of animal models to an imaging protocol may require some modification – in our case the standard cerebral hypoxia-ischemia model was altered by the injection of additional fluid (0.2 ml of the radiotracer) during the hypoxic challenge.
Given the complexity of timing in this experiment, there are many failure modes at different steps along the way that may delay the experiment at best, or end the experiment without salvageable data in the worst case. Consistency at each step, from animal model generation to imaging, is crucial and may only be achieved by preparation and practice. Mastery of the techniques presented in this paper will allow for the robust application of simultaneous PET/MRI imaging to a variety of animal models and PET and MRI contrast methods.
The authors have nothing to disclose.
The authors would like to acknowledge the Center for Molecular and Genomic Imaging at UC Davis and the Biomedical Imaging Department at Genentech. This work was supported by a National Institutes of Health Bioengineering Research Partnership grant number R01 EB00993.
Surgery | |||
Surgical scissors | Roboz | RS-5852 | |
Forceps | Roboz | RS-5237 | |
Hartman mosquito forceps | Miltex | 7-26 | |
2x McPherson suturing forceps, 8.5 cm | Accurate Surgical & Scientific Instruments | 4473 | It is useful to reduce the opening width with a band on the forceps used to hold the carotid artery |
6-0 silicone coated braided silk suture with 3/8 C-1 needle | Covidien Sofsilk | S-1172 | |
Homeothermic blanket system | Harvard Apparatus | 507220F | |
Super glue | (Generic) | ||
Hypoxia | |||
Flowmeter for O2 | Alicat Scientific | MC-500SCCM-D | |
Flometer for N2 | Alicat Scientific | MC-5SLPM-D | |
O2 meter | MSA | Altair Pro | |
Imaging | |||
7.05 Tesla MRI System | Bruker | BioSpec | 20 cm inner bore diameter with gradient set. Paravision 5.1 software. |
Volume Tx/Rx 1H Coil, 35mm ID | Bruker | T8100 | |
PET system | (In-house) | 4×24 LSO-PSAPD detectors, 10×10 LSO array per detector, 1.2mm crystal pitch and 14mm depth. 14 x 14 mm PSAPD. FOV: 60x35mm. 350-650 keV energy window. 16 ns timing window. |
|
Vessel cannulation Dumont forceps | Roboz | RS-4991 | |
PE-10 polyethylene tubing | BD Intramedic | 427401 | |
Infusion pump | Braintree Scientific | BS-300 | |
Animal monitoring & gating equipment | Small Animal Instruments Inc. | Model 1025 | Only respiration monitoring used |
Animal bed with temperature regulation | (In-house) |