|Home | About | Journals | Submit | Contact Us | Français|
The purpose of this study is to develop a rodent functional magnetic resonance imaging (fMRI) survival model with the use of heparin-coated vascular access devices. Such a model would ease the administration of sedative agents, reduce the number of animals required in a survival experiment, and eliminate animal-to-animal variability seen in previous designs. Seven male Sprague-Dawley rats underwent surgical placement of an MRI-compatible vascular access port, followed by implantable electrode placement on the right median nerve. Functional MRI during nerve stimulation and resting-state functional connectivity MRI (fcMRI) were performed at times 0, 2, 4, 8, and 12 weeks postoperatively using a 9.4 T scanner. Anesthesia was maintained using intravenous dexmedetomidine and reversed using atipamezole. There were no fatalities or infectious complications during this study. All vascular access ports remained patent. Blood oxygen level dependent (BOLD) activation by electrical stimulation of the median nerve using implanted electrodes was seen within the forelimb sensory region (S1FL) for all animals at all time points. The number of activated voxels decreased at time points 4 and 8 weeks, returning to a normal level at 12 weeks, which is attributed to scar tissue formation and resolution around the embedded electrode. The applications of this experiment extend far beyond the scope of peripheral nerve experimentation. These vascular access ports can be applied to any survival MRI study requiring repeated medication administration, intravenous contrast, or blood sampling.
Since its initial description in 1990 (Ogawa et al., 1990), blood oxygen level dependent (BOLD) fMRI has provided a reliable noninvasive method for indirectly studying cerebral blood flow and neuronal activity. More recently, BOLD fMRI and fcMRI have been applied to the study of brain plasticity following nerve injury (Lundborg, 2000; Pawela, 2009). Because of the homology existing between the human and rat upper extremity (Bertelli et al., 1992; Bertelli et al., 1995), the rat forelimb has served as an excellent model for human upper extremity peripheral nerve injury (Parkins et al., 2009). As fMRI and fcMRI technology continue to grow, the development of long-term rodent survival models of nerve injury has become a major focus. One adjunct to any long-term survival study is a robust method for providing intravenous access. Such access allows repeated medication administration and repeated blood sampling, if necessary. While the use of vascular access devices is not novel, there is little documentation of their use in the MRI literature. In this paper, we describe our technique for maintaining vascular access in a rodent fMRI survival study using an MRI-compatible vascular access port.
Vascular access devices have been a mainstay in animal research for decades. While successful use can be regularly achieved in large animals, smaller access devices have shown only variable success in rodents. Although earlier techniques described in the 1950s and 1960s were pivotal for the development of long-term access, many of them externalized the intravenous catheter, requiring methods to prevent catheter tampering (Cox and Beazley, 1975; Dalton et al., 1969; Edmonds and Thompson, 1970; Kleinman et al., 1975; Mendelsohn, 1962). This led to the development of multiple restraining methods along with devices designed for rotational movement (Eve and Robinson, 1963; Lemmel and Good, 1971; Rose and Nelson, 1955; Wittgenstein and Rowe, 1965). Despite modest success with these cumbersome apparatuses, improvements in surgical technique, perioperative care, anesthesia, and device-design have led to more widespread use of vascular access ports in many fields.
In the 1960s, Weeks and Davis reported on the use of indwelling catheters placed within the internal jugular vein. The group concluded that the catheter tip needed to be placed within the right atrium of the heart in order to avoid occlusion. In 1980, Kaufman directed a silastic catheter through the femoral vein to the inferior vena cava for long-term venous access. Kaufman used the femoral vein as opposed to the internal jugular vein based on previous reports of catheter occlusion (Kaufman, 1980). As a result of the increased flow in the femoral/inferior vena cava system, more reliable patency was established (Burvin et al., 1998; De Jong et al., 2001; Kaufman, 1980; Koeslag et al., 1984; Pages et al., 1993). In a comprehensive catheter study in 2005, Yang et al. showed patency in 100% of 3 French (Fr) catheters 9 weeks after placement in the rat femoral vein (Yang et al., 2005). As surgical techniques improved, so too did device design. Since the late 1990s, focus on port and catheter technology has resulted in a variety of biocompatible materials, along with multiple antithrombotic and anti-inflammatory medicated coatings (Nolan and Klein, 2002).
In rodent fMRI, the ideal survival study allows the same animal to be followed over time, with imaging obtained at specific time points. In order to maintain appropriate levels of sedation while scanning, intravenous sedatives that do not affect cerebral BOLD signal are required (Pawela et al., 2009a). Previous fMRI rodent models of nerve injury and repair required large numbers of animals to generate the statistical power necessary to make scientifically valid conclusions (Parkins et al., 2009). In these studies, all animals would undergo a nerve operation, after which groups of animals would be randomly assigned to undergo imaging at a specific time point. As a result of paralytic administration, tracheostomy, mechanical ventilation, and invasive intravascular access, the animals would be euthanized immediately after imaging. The data would then be averaged for each time point, providing a picture of BOLD activation within the population at that time point. One of the inherent weaknesses of these designs is that the same animal is not followed over time, allowing animal-to-animal variability. In contrast, a survival model allowing the same animal to be followed would reduce the overall number of animals required in a survival experiment and eliminate the animal-to-animal variability seen in previous study designs. With this survival model, a repeated-measures ANOVA can be performed. This will allow researchers to measure, test for differences in, and account for subject variation.
As a result of the requirements for an fMRI survival study and data provided from previous catheter studies, we have developed a long-term vascular access model in the rat that is compatible with MRI. In this paper, we describe our methods using a Carmeda® BioActive Surface (CBAS)-coated, 3 Fr polyurethane catheter (Harvard Apparatus, Holliston, MA) placed in the femoral vein. The catheter is then attached to a plastic port (Harvard Apparatus, Holliston, MA), which is placed subcutaneously in the back. This technique advances our ability to conduct survival MRI studies in rats with minimal disruption to the rat between experiments, eliminating the stress of repeated vein puncture, which alters levels of multiple circulating hormones (Diehl et al., 2001; Joint Working Group on Refinement, 1993; Krinke, 2000).
All experiments were performed in compliance with federal regulations and the guidelines of our institutional animal care and use committee. Seven male Sprague-Dawley rats, weighing 150–200 g, were used in this study. All seven rats underwent surgical placement of a subcutaneous vascular access port. The details of this procedure are described below. Following a recovery period of one week, each rat underwent surgical placement of an implantable electrode on the right median nerve. Time of nerve surgery was considered 0. Each rat then underwent cerebral fcMRI followed by fMRI during direct nerve stimulation at times 0, 2, 4, 8, and 12 weeks postoperatively.
Following an acclimation period of one week, seven male Sprague-Dawley rats, weighing 150–200 g, underwent vascular access port placement (Fig. 1). To obtain general anesthesia, the animal was placed in a transparent chamber, where isoflurane (1.4%; Halocarbon Laboratories, River Edge, NJ) was administered through a vaporizer. The animal was then transferred to the operating table where isoflurane was continuously administered through a nose cone. After obtaining adequate general anesthesia with lack of response to tail pinch, the right groin, flank, and back were clipped with an electric razor. The areas were then prepped with povidone-iodine (Triad Disposables, Brookfield, WI). To begin the procedure, the animal was placed in a supine position, and a 1.5 cm incision was made over the right groin. After dissecting the subcutaneous tissues, the femoral vessels were identified. The femoral vein was then isolated and ligated distally with a 6-0 silk tie. Proximally, a 6-0 tie was looped around the vein, and the vein was placed under tension. A venotomy was then performed, and the proximal end of the 3 Fr CBAS-coated catheter (model no. 724394, Harvard Apparatus, Holliston, MA), primed with heparinized saline (14 U/kg), was passed intraluminally 3 cm to the inferior vena cava. The catheter was then tied down with the remaining 6-0 silk tie.
The rat was then shifted to the left lateral decubitus position, and attention was directed to the back. A 1.5 cm incision was placed longitudinally in the midline. A subcutaneous pocket was then dissected toward the right flank. A Kelly clamp was used to create a subcutaneous tunnel from the back incision to the groin incision. The distal end of the catheter was then retrieved and delivered into the back incision. The distal end was connected to the port (Harvard Apparatus, Holliston, MA), which was first primed with heparinized saline (14 U/kg). The catheter was then securely fixed to the port with the use of the provided locking sleeve. The port was accessed with a 22-gauge Huber needle (Smiths Medical, St. Paul, MN), drawn back to confirm patency and intraluminal position, and then flushed with heparinized saline (14 U/kg). The port was then placed within the subcutaneous pocket previously dissected (Fig. 2). It is important to maintain slack in the catheter to allow for mobility of the port as the animal grows. Furthermore, the port itself was not fixed to the underlying fascia to allow for mobility during growth. The skin was then closed in both the back and groin incisions using interrupted 3-0 nylon suture. Following the surgical procedure, the animal was allowed to recover in an isolated cage. Once the animal was alert, no further isolation was required.
One week after vascular access port placement, all seven rats underwent surgical placement of an implantable electrode (AISI 304, Plastics1, Roanoke, VA) on the right median nerve (Fig. 3). Time of electrode placement was considered 0.
To induce anesthesia at each time point, the animal was placed in a transparent holding chamber where vaporized isoflurane (1.4%) was administered. Following induction, the animal was transferred to the operating table and placed in the left lateral decubitus position. The port was then accessed percutaneously with a 22-gauge Huber needle using sterile technique (Fig. 4). The port was flushed with heparinized saline (14 U/kg) to confirm patency. The animal was then transferred to the 9.4 T MRI scanner. Once in the scanner, an infusion of dexmedetomidine (Orion Corp., Espoo, Finland) was started at a rate of 100 mcg/kg/hr. Dexmedetomidine, the S-enantiomer of medetomidine, is an alpha-2 adrenergic agonist. Medetomidine has been shown to provide adequate depth of anesthesia while not inhibiting the BOLD signal in fMRI experimentation (Cho et al., 2007; Cho et al., 2008; Pawela et al., 2009a). Several minutes after the infusion was started, isoflurane was discontinued. Delaying the discontinuation of isoflurane for several minutes after starting the dexmedetomidine infusion allows the dead space of the cannula to fill and ensures intravascular circulation of dexmedetomidine. This circumvents the necessity for induction using an intravenous bolus. Rapid intravenous bolus of dexmedetomidine can lead to significant bradypnea and even transient apnea and, therefore, should be avoided.
Methodology for fMRI using embedded electrode stimulation generally followed previous papers from this laboratory (Cho et al., 2007; Cho et al., 2008; Parkins et al., 2009; Pawela et al., 2009a; Pawela et al., 2009b). During scanning, the animal was allowed to breathe spontaneously through a nose cone with a fraction of inspired oxygen equal to 30%. Chest respiration (Model 1025, SA Instruments, Stony Brook, NY), end tidal gases (POET IQ2, Criticare Systems, Waukesha, WI), and pulse oximetry (Model 8600 V, Nonin Medical, Plymouth, MN) were continuously monitored.
The BOLD response to nerve stimulation in the primary sensory and motor regions was studied in seven animals over 12-weeks duration to determine the reproducibility of fMRI results in a longitudinal study. A Bruker AVANCE 9.4 T MRI scanner (Billerica, MA) with a 30-cm bore was used in this study. Images were acquired using a Bruker receiving surface coil (T9208) and a linear transmit coil (T10325). A rapid acquisition with relaxation enhancement (RARE) anatomic image was obtained prior to each fMRI experiment. RARE parameters were TR = 2.5 sec, TE = 12.5 ms, ETL = 8, FOV = 35 mm, and a 256 × 256 matrix. A gradient-recalled echo-planar imaging (EPI) sequence was used for fMRI and fcMRI experiments. EPI parameters were TR = 2 sec, TE = 18.76 ms, FOV = 35 mm, and a 96 × 96 matrix, with the same slice geometry as the RARE images. Ten 1-mm-thick contiguous slices were acquired, with slice 3 located directly over the anterior commissure (−0.36 mm from bregma). This slice profile allowed adequate coverage of the sensorimotor system. One-hundred-ten images were obtained during each fMRI experiment for a total acquisition time of 3 min 40 sec. The fcMRI study was done before the BOLD fMRI study with no electrical stimulation applied during the scan. A square-wave electrical generator (S88, Grass Telefactor, Warrick, RI) equipped with a constant current unit (CCU, Grass Telefactor, Warrick, RI) was used for all nerve stimulations during fMRI scanning. A standard block-design stimulation protocol of 40 sec rest followed by three periods of 20 sec on and 40 sec off was used (total of 3 min 40 sec). Electrical stimulation was computer-controlled and triggered by a transistor-transistor logic pulse from the scanner. Direct nerve stimulation was performed with a current of 0.5 mA, frequency of 10 Hz, and duration of 1 ms.
Following image acquisition, the animal was returned to the operating suite. Dexmedetomidine was then reversed using atipamezole (Orion Corp., Espoo, Finland), 100 mcg/kg given intramuscularly.
To promote port patency, the vascular access device was accessed with a Huber needle biweekly and flushed with heparinized saline (14 U/kg) as previously described. All incisions were inspected and assessed for signs of infection. Furthermore, behavior (including feeding habits and ambulation) was monitored to identify any signs of postoperative complications.
Motion correction was performed for all individual scans before any further analysis (AFNI, 3dvolreg). Datasets were detrended to eliminate linear drifts, and separate slices were aligned to the same temporal origin. The EPI scans were then registered to the anatomic images (FSL, FLIRT). For BOLD fMRI acquisitions, the registered EPI acquisitions were averaged using Analysis of Functional Neuroimaging (AFNI) software. These averaged EPI datasets were used to create BOLD activation maps. The activation maps were created by performing an F-test on the time series with the block design as the only regressor (AFNI, 3dDeconvolve). Activation was determined by an F-test (3dDeconvolve) with a p-value threshold of 0.005 (using AFNI). For the fcMRI study, all resting-state acquisitions were analyzed individually after registration using seed-voxel-based fcMRI analysis. The seed region was chosen from the left granular insular cortex (GI) and consisted of 25 voxels. The same seed was applied to each fcMRI scan to create a reference time course for each resting-state scan. A band-pass filter was used for all resting-state EPI acquisitions, with a low-pass filter of 0.1 Hz and a high-pass filter of 0.01 Hz on a voxel-by-voxel basis covering the entire brain (Hamming, 1983). The resting-state, voxel-wise correlation coefficient analysis was done using the reference time course of each individual scan. Results underwent fisher-Z transform and then were pooled based on the time point of data acquisition. The fcMRI results of each time point were generated by averaging individual fcMRI results. A histogram of voxel counts was created to demonstrate the results at a p-value threshold of 0.05.
Seven animals underwent vascular access port placement along with fMRI during direct nerve stimulation at time 0, 4, 8, and 12 weeks postoperatively. Image acquisition time was 45.3 min at each time point, and the total time on the scanner approximated 70 min at each time point. During all scanning, the mean heart rate was 243 bpm (162–320 bpm), the mean respiratory rate was 87 breaths/min (43–147 breaths/min), and pulse oximetry ranged from 89–99%. There were no fatalities or infectious complications during this study. There were no physical findings of lymphedema or venous stasis in the right lower extremity. All vascular access ports remained patent over the 12-week time period. Patency was determined by ease of injection, adequacy of and ability to obtain general anesthesia, and physical examination of the port site.
BOLD activation was seen within the forelimb sensory territory (S1FL) of the contralateral cortical hemisphere during direct nerve stimulation for all animals at all time points. Statistically significant BOLD fMRI signal activation during direct nerve stimulation is shown in Fig. 5. Figure 5A shows the averaged voxel activation of seven animals across five time points. Figure 5B is a graph showing the average number of activated voxels observed in the seven rats for each time point. Raw data are shown in Table 1.
The results of the fcMRI study are demonstrated in Fig. 6. Due to the highly selective seed region we used in this study (GI), a refined resting-state network can be seen within the bilateral insular cortex with great consistency across five different time points (Fig. 6A). The histogram shown in Fig. 6B further demonstrates this consistency, with the area under the curve being the sum of the voxel counts contained in the network of each time point. ANOVA was also performed using 0-week resting-state data as a reference, and comparing it to the results at 2, 4, 8, and 12 weeks. There is no statistically significant change in the fcMRI network.
By providing superior soft tissue contrast and detail, MRI has become a standard imaging technique for a host of central nervous system, musculoskeletal, cardiovascular, and oncologic disorders. As mouse and rat models have been developed to study these and other disorders (Beaumont et al., 2009; Stuckey et al., 2008; Valable et al., 2007), rodent MRI has been used with increasing frequency in multiple fields. Furthermore, as greater emphasis has been placed on outcomes, the need for longitudinal survival models has increased in rodent experimentation. Nearly all survival rodent studies require repeated medication administration and/or frequent blood sampling. Therefore, the maintenance of intravascular access is a helpful adjunct to any rodent survival study.
While intravenous access can be obtained from various methods, multiple studies suggest that repeated vein puncture increases animal stress and alters circulating levels of glucose, corticosteroids, prolactin, epinephrine, growth hormone, insulin, and renin (Diehl et al., 2001; Joint Working Group on Refinement, 1993; Krinke, 2000). Furthermore, vein puncture is technique-dependent and often requires a period of warming for vessel dilatation. Eliminating the morbidity associated with repeat vein puncture, intravascular access devices provide for easy administration of medications and repeated blood sampling. Despite reproducible success in larger animal models, only recently has long-term patency been shown in the rodent literature (De Jong et al., 2001; Yang et al., 2005).
In the case of fMRI, the ideal survival study should allow the same animal to be imaged at multiple time points. Such a study requires sedation that does not interfere with the BOLD signal, yet permits safe animal recovery following each scan. Dexmedetomidine, an alpha-2 adrenergic agonist, has been shown to be effective in multiple BOLD fMRI studies (Cho et al., 2007; Cho et al., 2008; Pawela et al., 2009a). This sedative allows adequate sedation for fMRI imaging, affords the use of spontaneous breathing, and has a potent reversal agent with a quick onset of action. Currently, with the use of MRI-compatible subcutaneous vascular access ports and the sedative dexmedetomidine, we have demonstrated a safe and effective model for a rodent survival fMRI study. Furthermore, with reliable intravenous access, sedative doses can easily be adjusted to accommodate longer scanning times. This protocol circumvents the morbidity associated with mechanical ventilation, eliminates animal-to-animal variability, and reduces the overall number of animals required for a longitudinal study.
In this study on healthy rats, S1FL activation drop-off can be seen at time points of 4 and 8 weeks. At longer time points, the fMRI signal recovers to a relatively normal level. This progression corresponds with the natural progression of scar tissue formation following electrode placement, which peaks at about 4–8 weeks with subsequent gradual resolution. The phenomenon observed here could be the consequence of compression of the nerve trunk by developing scar tissue after electrode placement, thereby affecting the conductivity of nerve fibers via axonotmesis. Behavior tests were performed to test this hypothesis. Decreases in grip strength and in Von Frey monofilament response were observed at 4–8 weeks, consistent with this hypothesis. Behavior test results as well as fMRI results return to normal ranges at 12 weeks. These experiments support the idea that there is a coupling between scar tissue formation around peripheral nerves that have been mildly stressed by electrode placement and reduced neuronal activity.
As a control for axonotmesis, fcMRI of the insular cortex was introduced to the study. A highly consistent fcMRI insular network was acquired at the various time points after surgery. A similar insular network was reported previously using BOLD fMRI and independent component analysis (ICA) in a nonsurvival rat model (Zhang et al., 2010; Liang et al., 2011). Our study shows that this network pattern is conserved across the different time points. Due to the highly specific seed we used in this study, only the insular cortical network was demonstrated.
Based on the atlas (Paxinos and Watson, 2007), the network consists of the granular insular cortex (GI), the dysgranular insular cortex (DI), the agranular insular posterior cortex (AIP), and the piriform cortex. These areas are related to signal integration, synaptic plasticity, and memory (Blair et al., 2001; Sweatt, 2001), which can also be modified by task (Kida et al., in press 2011). It is hypothesized that fcMRI of granular insular cortex can be used as a control in fMRI survival experiments of various kinds including cortical plasticity studies.
In this study, seven vascular access ports remained patent over a period of 12 weeks. This experiment is exciting in that long-term vascular access can be reproducibly achieved and used safely in a rodent survival MRI study. The applications of such a port extend far beyond the scope of peripheral nerve experimentation. Such access ports can be applied to any survival MRI study requiring repeated medication administration, intravenous contrast, or blood sampling.
This work was supported by grant EB000215 of the National Institutes of Health.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.