The aim of the study was to develop a suitable animal model for validating dynamic contrast-enhanced magnetic resonance imaging perfusion measurements. A total of 8 pigs were investigated by DCE-MRI. Perfusion was determined on the hind leg musculature. An ultrasound flow probe placed around the femoral artery provided flow measurements independent of MRI and served as the standard of reference. Images were acquired on a 1.5 T MRI scanner using a 3D T1-weighted gradient-echo sequence. An arterial catheter for local injection was implanted in the femoral artery. Continuous injection of adenosine for vasodilation resulted in steady blood flow levels up to four times the baseline level. In this way, three different stable perfusion levels were induced and measured. A central venous catheter was used for injection of two different types of contrast media. A low-molecular weight contrast medium and a blood pool contrast medium were used. A total of 6 perfusion measurements were performed with a time interval of about 20–25 min without significant differences in the arterial input functions. In conclusion the accuracy of DCE-MRI-based perfusion measurement can be validated by comparison of the integrated perfusion signal of the hind leg musculature with the blood flow values measured with the ultrasound flow probe around the femoral artery.
Perfusion is the supply of tissues with blood. Specific perfusion is defined as the volume of blood flowing through an organ or tissue per unit of time and is usually given as mL/(
There are several methods for measuring perfusion in biological tissues. Most techniques use indicator agents that are added to the blood. The distribution of the indicator provides indirect information on the distribution of the blood. For the measurement of perfusion using positron emission tomography (PET), radioactive markers (e.g., H2 15O, 18F-fluorodeoxyglucose, and 133Xe) are used. For determination of perfusion with computed tomography (CT), X-ray contrast media are used. Both methods expose the patient to ionizing radiation.
Furthermore, microspheres can be used for perfusion measurements in animal models. Microspheres are colored or radioactively labeled polystyrene beads with diameters of 5–50
Increasingly, perfusion measurement is performed using magnetic resonance imaging (MRI). Several MRI techniques are available for perfusion measurement. The most important methods are dynamic contrast-enhanced MRI (DCE-MRI) and arterial spin labeling (ASL). Arterial spin labeling uses blood as an endogenous marker. Radiofrequency pulses are applied to invert or saturate the longitudinal magnetization of arterial blood upstream of the region of interest. The labeled blood causes a signal reduction in the target tissue. Images with and without labeled blood are acquired to generate subtraction images for determination of perfusion [
DCE-MRI uses a contrast medium (CM) that is injected as an intravascular bolus. Perfusion can then be determined from the resulting temporal changes in CM concentrations in blood and tissue. The accuracy of perfusion measurement by DCE-MRI is unclear. In addition to perfusion, vascular permeability as a functional parameter can be determined by measuring CM extravasation. An additional aim is to assess the interstitial and vascular volumes.
As a highly perfused organ, the kidney has been used to evaluate perfusion measurement with DCE-MRI [
Compared to other organs, such as kidney or brain, muscle perfusion is relatively low at rest and varies widely during physical activity. Therefore, muscles are well suited to investigate how well different perfusion levels can be detected and quantified by DCE-MRI. Pigs have a large muscle mass due to selective breeding for meat.
The gold standard used for blood flow measurement in the present study is an ultrasonic transit time flow probe that is fully independent of the perfusion imaging technique. It can be surgically attached to the artery of the supply area to be measured and it enables continuous measurement of pulsatile blood flow through this artery. Unlike electromagnetic or Doppler systems, the flow probe is insensitive to hematocrit and particulate matter. The transit time technique yields absolute flow values, whereas other systems measure only flow changes [
To perform measurements beyond baseline flow, blood flow is enhanced by a locally injected vasodilator. The antiarrhythmic drug adenosine is especially well suited for this purpose. Adenosine consists of the nucleotide base adenine and the sugar
We chose pigs of similar body size as humans in order to establish a well-suited model for clinical applications. A total of 8 female pigs, German landrace or hybrid form (body weight, 56–74 kg), were investigated to implement the method. Only healthy animals without known cardiovascular or musculoskeletal disorders were used. The pigs were not fed overnight prior to the experiment but had free access to water. All experiments were reviewed and approved by the local animal protection committee and performed according to the rules of the German animal welfare regulations.
All surgical procedures were performed under aseptic conditions in an operating room equipped for large animals.
Premedication was performed using 30 mg/kg ketamine (ketamine 10%, Ceva Tiergesundheit GmbH, Germany), 2 mg/kg azaperone (Stresnil Janssen-Cilag GmbH, Germany), and 0.02–0.05 mg/kg atropine sulfate (Atropinsulfat, B. Braun Melsungen AG, Germany) by intramuscular injection.
A peripheral 20-G venous catheter was placed in an ear vein and, about 30 min after premedication, total intravenous anesthesia was started. A perfusor was used to inject 4–7 mg/kg/h propofol (Propofol-ratiopharm, Ratiopharm, Germany), 0.1–0.5 mg/kg/h midazolam (midazolam injection solution 0.5%, Germany), and 0.0015 mg/kg/h fentanyl (fentanyl citrate solution 3.925 mL/50 mL, Germany). A tracheal tube was placed (Hi-Contour cuffed tracheal tube, ID 8.0, Mallinckrodt, Ireland), and the pig was ventilated with a respiratory device (Fabius, Draeger, Germany). The tidal volume was set at 10 mL/kg, the respiratory rate at 12–14 breaths/min, and the positive end-expiratory pressure was set at 5 mbar. The tidal volume and frequency were adjusted to keep end-expiratory CO2 within 35–40 mmHg and to maintain at least at 95% peripheral oxygen saturation.
A central venous catheter (3-Lumen-ZVK-Set, ARROWg+ard Blue, Arrow, Germany) was placed in the jugular vein on the right side of the neck of the pig. The central venous catheter was used for administration of contrast medium. Subsequently, the right or, if this was not possible, the left femoral artery was exposed. The femoral artery was catheterized proximally by the Seldinger technique to enable local administration of adenosine or CM. The metal-free, MRI-compatible catheter (Arterial Leader Cath, Vygon, France) was advanced proximally and fixed with several sutures. Distal to the catheter, an ultrasound flow probe (T206, Transonic Systems Inc., Ithaca, NY) was implanted around the femoral artery to measure blood flow (Figure
Positions of the ultrasound flow probe and catheter at the femoral artery. The arrows indicate the direction of blood flow.
First, the pig was transported into the scanner room. From the beginning of the transport the pig was ventilated with a portable medical ventilator (Oxylog, Dräger, Germany). The ventilator contained metal so it had to be placed outside the scanner room. The breathing tube therefore had to have length of 6 m. Due to the length of the tube, monitoring of end-expiratory CO2 with a monitoring device for capnography (Vamos, Dräger, Germany) was particularly important. The ventilator settings were adjusted for an end-expiratory CO2 level of 35–45 mmHg.
An MRI-compatible monitoring device (Veris, Medrad, Germany) was used for monitoring heart rate and oxygen saturation. The pig was placed in the supine position on the scanner table, which contains an integrated 32-channel spine coil (Siemens Magnetom Aera 1.5 T). A surface coil (Tim body coil, 18 RF channels) was placed on the hind limbs. The blood flow measurement data of the flow probe were continuously recorded using LabVIEW 2012, an A/D converter card (NI USB-6211), and a standard netbook with Windows XP.
First, morphological images were acquired without contrast medium administration for orientation (Figure
For vasodilatation, a perfusor was used for locally injecting adenosine via the Seldinger catheter in the femoral artery. The adenosine perfusion rate was individually and dynamically adapted to achieve similar blood flow levels in all pigs at each measurement time point. Steady blood flow occurred after approximately 5 min, and the first perfusion measurement was started. Thereafter, two further perfusion measurements with different perfusion levels and systemic CM administration were performed, followed by a perfusion measurement with local CM administration.
Subsequently, a rapidly extravasating low-molecular weight contrast medium (LMCM) was injected locally to visualize the area supplied by the femoral artery (Figure
Difference image. T1-weighted images before and after local injection of contrast agent into the right femoral artery were acquired. There is significant brightening of the supply area.
Photograph showing the supply area of the femoral artery after administration of Evans blue via the Seldinger catheter.
Representative example of workflow and timing in a single swine experiment.
First, unenhanced T1-weighted morphological images with and without fat suppression and T2-weighted images were acquired for orientation. The protocol included an axial T1-weighted turbo spin echo (TSE) sequence with fat suppression in the transverse plane with the following parameters: repetition time, TR = 625 ms, echo time, TE = 12 ms, flip angle,
The protocol for perfusion measurement included a 3D gradient-echo sequence (Siemens syngo TWIST) with a high temporal resolution of approximately 1.5 s and the following parameters: TR = 2.69 ms, TE = 0.86 ms,
The central 20% of the central
After acquisition of the pulse sequences described above, the next measurement was started using a blood pool contrast medium (BPCM) (0.25 mMol/mL gadofosveset trisodium, Vasovist Bayer Schering, Berlin, Germany/Ablavar Lantheus Medical Imaging, Inc., USA) that remains intravascular due to high protein binding. The BPCM was administered at a dose of 0.1 mL/kg body weight and an injection rate of 5 mL/s. CM administration was followed by injection of 20 mL 0.9% saline solution at the same rate. The dynamic acquisition was performed with identical parameters at 100 time points. Before each acquisition, static images were acquired using the TWIST sequence without the keyhole technique to create maps of baseline relaxivity and magnetization with flip angles of 5°, 10°, 20°, and 30°. After the first static acquisitions, adjustments were made and used for the subsequent static and dynamic acquisitions with the TWIST sequence. Time-dependent maps of relaxation rate changes were computed using the method of Li et al. [
Increased blood flow was induced by local injection of adenosine (Adenosin Life Medical, 5 mg/mL, Carinopharm, Germany) into the femoral artery using a syringe pump. The dose was chosen according to its flow-enhancing effect. After about 5 min, a steady blood flow, measured by the ultrasound flow probe, was achieved, and the measurement by MRI was started. After the fifth time step, the low-molecular weight CM was administered. Subsequently, a second measurement with the blood pool contrast medium (BPCM), which remains intravascular, was carried out. It was administered after the fifth time step as well.
A total of up to three perfusion measurements were performed per animal. After several measurement cycles, signal saturation due to contrast medium accumulation in tissues is expected to preclude reliable perfusion measurements. The order in which the different perfusion levels (baseline blood flow or increased blood flow induced by adenosine) were measured was varied in each pig. The first perfusion measurement was performed either at baseline blood flow or at blood flow enhanced by adenosine injection. The following measurements were carried out at either higher, lower, or identical blood flows compared with the first measurement. Over the entire series of experiments, the same number of different flow levels in each of the three measurements was achieved. This was done to compensate for systematic errors due to signal saturation resulting from contrast medium accumulation. After the perfusion measurements, the supply area of the femoral artery was determined by MRI. Anatomical images were used to generate difference images (Figure
To compare the data collected from the pig experiments with those in humans, arterial input function (AIF) relaxation rate change curves and tissue curves were generated from MRI data obtained in a 17-year-old female patient with a highly vascularized sarcoma in the left lower leg. MRI in this patient was performed on a 1.5 T system (Magnetom Symphony, Siemens, Erlangen, Germany). The DCE-MRI protocol included a 3D T1-weighted gradient-echo (GRE) sequence with TR = 3.64 ms, TE = 1.22 ms,
Eight pigs were used to establish the method of perfusion measurement by MRI with an ultrasound flow probe as a reference method and adenosine administration to enhance perfusion.
The flow measurements with the ultrasound flow probe were continuously recorded. During imaging with high flip angles, the signal measured by the probe was disturbed by the electromagnetic field. Therefore the flow probe measurements obtained before and after each MRI acquisition were averaged to estimate blood flow during MRI acquisitions.
The desired levels of increased blood flow were accomplished by adjusting the adenosine dose based on its observed effect (Figure
Increase in blood flow (mL/min) after local administration of adenosine in experiment number 5 (a). Distribution of adenosine doses (
Figure
Signal curves of arterial input functions (AIF) of the aorta. The figure shows curves of three measurements after administration of the BPCM. The curves of the second and third measurements have a higher baseline signal than the first curve because of persisting effects of CM administered for the preceding measurement (a). Relaxation rate change versus time of the arterial blood measured in the aorta. The relaxation rate change curves were calculated with the method of Li et al. using the same data as for (a). The CM preload has no evident effect on the curves (b). Signal tissue curves of semitendinosus muscle using LMCM. As far as possible, no bones, skin, or great vessels were included (c). Corresponding to the method used in (c), relaxation rate change versus time curves were generated (d). Relaxation rate change versus time of the AIF of a 17-year-old female sarcoma patient measured in a vessel in the lower extremity using a BPCM (e). Relaxation rate change versus time of a tissue curve measured in a muscle of the lower extremity of the sarcoma patient using an LMCM (f).
Figure
The curves of the patient (Figures
Furthermore, the time intervals between first and second pass of the CM in the vessels were determined in five patients enrolled in a clinical study and in the eight pigs. The time between first and second pass corresponds to the circulation rate. The mean interval was
The aim of our study was to develop an appropriate animal model to validate DCE-MRI-based perfusion measurements. The pig is a well-suited animal for experimental studies because its anatomy and physiology are similar to humans. This is confirmed by our findings. The pigs used in our experiments had heart rates of 60–80 beats/min, which corresponds to adult human heart rates. Blood volume of pigs is about 70–80 mL/kg and is also similar to that of humans (60–98 mL/kg) [
Continuous local adenosine administration caused a sustained and steady increase in blood flow in the femoral artery that could be well controlled. The total adenosine doses administered were different among the pigs. This was deliberately accepted to achieve comparable perfusion values in the various phases of the experiments. However, with a constant supply of adenosine using a syringe pump, it was possible to establish steady blood flow after a few minutes. The respective blood flows were reproducible for each pig.
Achieving a steady perfusion level was essential because DCE-MRI acquisitions at high flip angles interfered with blood flow measurement using the ultrasound flow probe.
To overcome this limitation, blood flow during MRI acquisition was calculated by averaging the probe-based flows measured before and after MRI acquisition. Our results show that the pig model used in the present study is well suited for DCE-MRI-based perfusion measurement.
The authors declare that there is no conflict of interests regarding the publication of this paper.