SEARCH

SEARCH BY CITATION

Keywords:

  • pulmonary perfusion;
  • hyperpolarized carbon-13 MRI;
  • parahydrogen induced polarization;
  • angiography;
  • functional lung imaging

Abstract

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. CONCLUSIONS
  7. REFERENCES

The study of lung perfusion in normal and diseased subjects is of great interest to physiologists and physicians. In this work we demonstrate the application of a liquid-phase hyperpolarized (HP) carbon-13 (13C) tracer to magnetic resonance imaging (MRI) of the pulmonary vasculature and pulmonary perfusion in a porcine model. Our results show that high spatial and temporal resolution images of pulmonary perfusion can be obtained with this contrast technique. Traditionally, pulmonary perfusion measurement techniques have been challenging because of insufficient signal for quantitative functional assessments. The use of polarized 13C in MRI overcomes this limitation and may lead to a viable clinical method for studying the pulmonary vasculature and perfusion. Magn Reson Med 57:459–463, 2007. © 2007 Wiley-Liss, Inc.

The primary function of the lung is to supply oxygen to the body and eliminate carbon dioxide to meet its metabolic needs. In order to accomplish this, the lungs must bring sufficient quantities of air and blood into close enough proximity to allow for adequate exchange of gases. Efficient operation of the respiratory system requires a large gas-exchange surface area and a constant supply of both fresh air and blood to the exchange membrane. The study of pulmonary ventilation, perfusion, and gas-exchange characteristics is therefore of great interest to physiologists and physicians alike.

Because of their ability to supply both spatial and temporal information, radiographic methods are well suited for studying the pulmonary vasculature and lung perfusion. Effective radiographic techniques for performing these tasks have been well received by the medical community and rapidly assimilated into clinical practice. For example, pulmonary angiography, computed tomographic angiography (CTA), and nuclear medicine ventilation/perfusion studies are routinely used by physicians to diagnose and manage diseases that affect pulmonary perfusion, such as pulmonary emboli, chronic obstructive pulmonary disease, pulmonary tumors, and pulmonary hypertension. However, the development of non-radiographic techniques could aid in longitudinal studies for drug development (to show disease modification), as well as in disease management. Such techniques would also benefit the subpopulation of patients who have a suspected pulmonary embolism and for whom a standard CT scan is either disallowed or inadvisable, such as pregnant women and patients with iodine allergies or renal insufficiency. The first group should not be exposed to radiation, and patients with the other two conditions may not be able to tolerate large volumes of iodinated contrast agents. This is not a trivial subpopulation. In Perrier et al.'s (1) recent prospective management trial regarding the role of multidetector CT in detecting pulmonary embolism, more than 5% of the patients had to be excluded from the trial for one of these reasons.

MRI methods for measuring pulmonary perfusion and visualizing the pulmonary vasculature are available and can produce clinically acceptable results. However, these techniques have not found widespread clinical acceptance due to practical difficulties with the scan time and ease of interpretation. The limitations of current MRI techniques might be overcome by the development of a water-soluble, nonextravasating contrast agent with a particularly high signal-to-noise ratio (SNR) performance, which would make MRI more attractive for clinical use. This would allow for the rapid acquisition of measures of pulmonary perfusion with high spatial and temporal resolution, and make MRI an ideal candidate for detecting pulmonary emboli in patients for whom CT and angiography are contraindicated. We note the particular applicability of these methods in the lung because of its low tissue density, although a suitably high-SNR agent could also find application in other organ systems and studies of physiology and pathophysiology, and could greatly enhance the utility of MRI for monitoring function during clinical trials.

Recently there has been considerable interest in the development of water-soluble hyperpolarized (HP) 13C MRI contrast agents. With the use of controlled, spin-dependent interactions, it is possible to externally polarize these compounds, administer them to subjects, and obtain images with superior SNR. In this work we demonstrate the feasibility of imaging pulmonary perfusion in a large species using an exogenous HP 13C contrast agent: 2-hydroxyethylpropionate. The results and imaging characteristics of several examples are presented.

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. CONCLUSIONS
  7. REFERENCES

Contrast Agent Production

HP hydroxyethylpropionate was produced by the method of parahydrogen-induced polarization using a prototype polarizer (GE Healthcare, Malmö, Sweden). Partially deuterated hydroxyethyl propionate (CD2H*-CDH*-C*(=O)-O-CH2-CH2-OH, where * refers to the 13C and parahydrogen labeling positions) has a measured carbonyl 13C T1 of 138 ± 8 s, and T2 of 5.3 ± 0.2 s under the following conditions: T = 20°C, concentration = 100 mM, and D2O solvent. T1 was measured as the decay of the HP signal in a syringe in the 1.5T imaging magnet after RF-induced polarization loss was accounted for. T2 was measured using a standard CPMG spin-echo sequence at thermal polarization in a 7T spectrometer. Under in vivo conditions, T1 and T2 may be shortened through interaction with hemoglobin, serum proteins, etc., but we do not yet have a measurement or estimate of that effect.

The parahydrogen-induced polarization method was previously described by Goldman et al. (2) and Golman et al. (3), and is summarized briefly as follows: Parahydrogen enrichment is accomplished by passing hydrogen gas through a commercial catalyst (Ionex-type O-P catalyst; C*Chem, Lafayette, CO, USA) at cryogenic (25°K) temperatures, resulting in nearly pure parahydrogen gas (∼97%). A reaction chamber is then pressurized with parahydrogen and warmed to 60°C to speed the hydrogenation process. 13C-labeled and partially deuterated 2-hydroxyethylacrylate and a rhodium catalyst are then injected into the reaction chamber. After hydrogenation, the resulting 2-hydroxyethylpropionate is subjected to an Insensitive Nuclei Enhanced by Polarization Transfer (INEPT)-like RF pulse sequence (4–6) to transfer the spin order from the para-protons on the hydrogenated molecule to a vicinal, quaternary 13C. Briefly, the polarization transfer sequence consists of 1) a “free” evolution period of 28.28 ms, 2) a 180°x13C pulse, 3) an evolution period of 36.2 ms, 4) a 90°y13C pulse, 5) a delay of 50.34 ms, and 6) a 90°x13C pulse. The “free” evolution periods contain 180° decoupling or refocusing pulses as described in the referenced work.

The resulting solution was filtered to remove the catalyst, resulting in 5 ml of 300 mM aqueous, HP 2-hydroxyethylpropionate. The net polarization was approximately 11%.

Animal Preparation and Imaging Protocol

We chose a porcine model to study HP 13C imaging of the pulmonary vasculature and pulmonary perfusion. Since the pig's pulmonary anatomy and physiology are similar to those of humans, information gained from this animal study will be applicable to studies of perfusion pathophysiology in humans. We expect that the results from these pig experiments will be useful for quantifying regional and global perfusion relationships/defects in the lung in future studies. Although the compound used in this experiment is not suitable for human imaging, the other techniques established here are fully generalizable.

The following experiments were performed under a protocol approved by the animal-use committee. Yorkshire pigs were induced, intubated, paralyzed, and maintained on isoflurane anesthesia. Vital signs were monitored during the procedure. The pigs were placed in a birdcage coil tuned to the 13C frequency and positioned in a 1.5T whole-body imager (Sonata; Siemens, Erlangen, Germany). The lungs were localized in space using a series of proton images. HP 13C MRI angiography was then performed using 5 ml of a 300-mM solution of HP hydroxyethylpropionate injected at a rate of 1 ml/s. Images were obtained at 1-s intervals. We began the MRI 5 s after initiating the HP tracer injection using a trueFISP pulse sequence with the following imaging parameters: TR/TE = 4.6/2.3 msec; FOV = 32 × 32 cm; matrix size = 128 × 128; flip angle = 180°, and slice thickness = 2.5–3 cm.

RESULTS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. CONCLUSIONS
  7. REFERENCES

Results from typical HP 13C MRI pulmonary angiography experiments performed in healthy animals are shown in Figs. 1 and 2. Figure 1 depicts a time-lapse series of coronal 13C MR images of a healthy pig's abdomen and thorax. The images show the time-dependent behavior of the magnetization of the HP contrast agent after it was injected into the femoral vein. The contrast agent can be seen flowing through the inferior vena cava and into the pulmonary arteries before it becomes distributed throughout the lung parenchyma. A second series of coronal time-lapse images obtained from a different test animal is shown in Fig. 2 for comparison. These images are centered on the right and left inferior pulmonary arteries, and, unlike those shown in Fig. 1, are tilted slightly in the anterior–posterior direction to better define the pulmonary vasculature.

thumbnail image

Figure 1. HP 13C MRI angiograms depicting pulmonary blood flow through a coronal slice just posterior to the heart. Images were taken at 1-s intervals. HP 2-hydroxyethyl propionate (5 ml, 300-mM solution) was injected into the femoral vein at a rate of 1 ml/s. The inferior vena cava, pulmonary arteries, inferior pulmonary arteries, and perfused lung parenchyma are clearly depicted in the images, whereas air-filled structures and tissues that are not supplied by the pulmonary artery are devoid of signal. We note that the initial images display a high SNR, which drops as time progresses since polarization is lost due to the imaging process and wash-out of the contrast agent. The slice thickness was 30 mm.

Download figure to PowerPoint

thumbnail image

Figure 2. HP 13C MRI angiograms of another porcine subject. The images depicted are of a coronal slice slightly tilted in the anterior–posterior direction to better visualize the lower right and left pulmonary arteries. The injection parameters were identical to those shown in Fig. 1. The slice thickness was 25 mm.

Download figure to PowerPoint

We note that in the images presented, the signal intensity is proportional to magnetization, and that magnetization is a strong function of the contrast agent's net polarization and concentration. Large magnetization typically occurs in pulmonary regions with high blood flow and vascular density. In our case, this corresponds to areas of the lungs occupied by the pulmonary arteries, their major tributaries, and the lung parenchyma. As with Gd-MR and CT-based angiography, low SNR occurs in regions of the lung with low concentrations of contrast agent, such as the pulmonary airways, poorly perfused sections of the lung, areas of the lung with low vascular density, structures not directly perfused by the pulmonary arteries, and regions of the lung where contrast agents have not yet arrived or have been washed out. Low signal is also observed in regions where the contrast agent's polarization has been lost due to various relaxation mechanisms (e.g., interactions with paramagnetic species and RF pulsing due to the imaging processes). We note that due to the low intrinsic sensitivity of 13C MRI, the magnetization of naturally occurring, thermally polarized 13C atoms in the pig's corpus so far below the threshold of detectability that only visible regions of the vasculature are those where magnetization in enhanced by the addition of a HP contrast agent.

It is helpful to examine the temporal and spatial signal dynamics in these series of images more carefully. The predictable and systematic fashion in which the images evolve is a direct consequence of the conserved nature of the cardiovascular architecture and the principles of fluid mechanics, in that the flow of blood and the contrast agents dissolved in it is driven by pressure gradients along contiguous segments of the vasculature. In our example, the contrast agent is injected into the venous system, drains into the heart, and is pumped through the pulmonary vasculature and lung parenchyma. These structures become visible sequentially as the contrast agent builds above the threshold of detectability in each region. We note that in the short time limit, the contrast agent is localized to the vena cava, since the HP hydroxyethylpropionate was injected directly into the femoral vein. Blood flow in the vena cava is slow due to the small pressure gradients encountered in this portion of the venous system, and it takes approximately 5–6 s for the front of the contrast agent to clear the vena cava. An irregularity in the smooth shape of the vena cava's profile is noted at the level of the renal veins. Retrograde flow of contrast agent into the renal vein is limited by the high perfusion rate of the renal system, and thus only a small section of the renal vein is visualized. In the long time limit, the inferior vena cava, the pulmonary vasculature, and the lung parenchyma are all clearly depicted. The near-homogenous enhancement of the pulmonary lung fields suggests near-uniform perfusion. This is as expected, since these are coronal images taken in healthy supine pigs, and gravity and disease have not had a chance to perturb tissue perfusion. The pulmonary arteries and their tributaries are most clearly defined in the intermediate time points, when the contrast between the enhanced vessels and the surrounding lung parenchyma is greatest, since the HP agent has not had time to reach the lung parenchyma and cause tissue enhancement. We note in passing that the overall SNR values degrade in the long time limit. This is a result of contrast agent dilution, loss of polarization due to imaging, and loss of polarization through various relaxation processes, such as intermolecular interactions with oxygen and other dissolved paramagnetic species.

To better understand the temporal behavior of HP 13C MRI pulmonary angiography, it is informative to plot the relative MRI signal intensities as a function of time for various structures of interest. In Fig. 3 we plot the relative signal intensities measured in the inferior vena cava, pulmonary artery, and lung parenchyma as a function of time for the imaging experiment described by Fig. 1. We note that the signal intensity in the inferior vena cava shows a sharp rise to its maximum followed by a rapid decay back to baseline once contrast-agent injection ceases and the HP contrast agent is washed out of the vena cava. The pulmonary artery displays a similar behavior, but its time dependence is somewhat modified due to cardiac mixing. Cardiac mixing is a product of the incomplete emptying of the heart with each contraction. This allows residual blood from a previous cardiac cycle to mix with fresh blood delivered to the heart for pumping to the pulmonary arteries. The result is a broadening of the input function presented to the venous vascular system. Wash-in and wash-out of the contrast agent occur most slowly in the lung parenchyma because of the long capillary dwell time and presumed free diffusion of the agent into the lung tissue. Furthermore, the peak intensities are much lower in regions of interest (ROIs) in the lungs compared to those in the pulmonary arteries and inferior vena cava, due to volume dilution.

thumbnail image

Figure 3. Relative signal intensities measured in the inferior vena cava (solid line), pulmonary artery (dash-dot line), and lung parenchyma (dashed line).

Download figure to PowerPoint

DISCUSSION

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. CONCLUSIONS
  7. REFERENCES

Most pulmonary pathologic states, such as pulmonary embolism, pulmonary hypertension, and chronic obstructive pulmonary disease, affect pulmonary perfusion. These diseases are associated with significant morbidity and mortality, and many are showing a growing incidence around the world. The ability to obtain more sensitive and quantitative measures of pulmonary blood-flow dynamics and tissue perfusion using intrinsically safe methods, such as MRI, should enhance the detection and evaluation of these disorders and lead to a radical evolution in pulmonary disease screening, treatment, and drug development.

The application of traditional MRI methods to lung imaging for quantitative perfusion measurement has so far been hindered by several problems. The proton T2* value in lung parenchyma is very short (on the order of 1 ms) due predominantly to heterogeneous magnetic susceptibility gradients arising from multiple air–tissue interfaces (7, 8). Furthermore, one must account for cardiac and respiratory motion, molecular diffusion, and pulmonary blood flow. As a result, MRI of the lungs remains challenging and its impact on the clinical assessment of lung disease is relatively small.

Several techniques have been proposed and implemented to address different aspects of this problem. Traditional 1H imaging of blood and tissue with ultrashort TE has been used to greatly increase signal and allow adequate structural imaging of lung parenchyma (8–11). Although it is promising, this technique benefits greatly from respiratory/cardiac gating and signal averaging to produce useful images, and is therefore challenging to perform in a clinical setting. Alternatively, investigators have demonstrated excellent imaging of airspaces using HP (12–14) or thermally polarized (15, 16) gases. With these methods one can not only obtain superior structural images, one can also derive measures of pulmonary blood flow, gas uptake (17, 18), and alveolar microstructure (12, 19). These techniques generally can overcome the T2* and signal limitations of traditional MRI, and by their use of rapid imaging can address the problems of subject motion; however, they are largely focused on lung ventilation and indirect measures of perfusion.

The use of HP 13C agents for pulmonary imaging supplements these techniques by directly addressing the pulmonary blood flow. In this way HP 13C MRI is similar to Gd-enhanced perfusion imaging (20, 21), but in contrast it is background-free and provides an inherently greater SNR. Because of the unusual characteristics of the HP 13C agent, the details of the imaging sequences differ from sequences involving 1H and other HP agents. In traditional imaging of HP noble gases, a fast, low-flip-angle pulse sequence, such as fast low-angle shot (FLASH), is used and a small portion of the longitudinal magnetization is utilized with each phase-encoding step. The remaining transverse magnetization is spoiled between each excitation, leading to an inefficient use of magnetization. This strategy is chosen because of the short diffusion times of the HP noble gases and the relatively short T2 times, which make the recovery of transverse magnetization technically challenging. The T2 relaxation times of HP 13C MRI contrast agents are considerably longer, which suggests that in contrast to HP noble gas imaging, it may be possible to recycle the transverse magnetization from one repetition to the next. It is logical to perform this recycling by using a steady-state free precession (SSFP) scheme, such as trueFISP. The trueFISP pulse sequence offers a significant improvement in SNR over traditional FLASH sequences, and was therefore used for the perfusion studies presented in this paper. A more complete discussion of relevant pulse-sequence considerations was previously presented by Svensson et al. (22). In the current study we found that repeated imaging with high SNR is feasible. In the results presented, the SNR values remain above 100 to 1 in the major vessels that contain contrast agent during the first six images. After five images, around the time it takes the bulk of the injected bolus to pass through the heart and distribute itself through the lung fields, the total signal intensity starts to decay in a monoexponential fashion with an average time constant of roughly 0.3 s. This gives us a lower-bound estimate for the T2 time of our compound in blood of 2 s. The actual value is expected to be higher, since some of the magnetization is washed out of the imaging field by blood flow, and magnetization is lost by imperfections in implementing the trueFISP imaging sequence.

Similarly to the evolution of HP gas imaging, we expect that variants of this technique will make additional quantitative aspects of the pulmonary vasculature, such as blood-flow direction and speed (23, 24), capillary dwell time, and quantitative tissue perfusion, accessible to investigators. However, further study is required to assess the feasibility of these techniques, and in particular to quantitatively understand the effect of reduced T2* in the lung capillary and perfused tissue.

CONCLUSIONS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. CONCLUSIONS
  7. REFERENCES

There is considerable interest in and need for noninvasive MRI-based methods to measure pulmonary perfusion and visualize the pulmonary vasculature. We have shown that HP 13C MRI angiography of the pulmonary vasculature in a large animal model is not only feasible, it can also produce images with excellent SNRs. These results suggest that HP 13C angiography is a promising new MRI-based modality for evaluating pulmonary perfusion and the pulmonary vasculature.

REFERENCES

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. CONCLUSIONS
  7. REFERENCES