CO2-Enhanced CT Imaging for the Detection of Pulmonary Emboli

Preliminary Laboratory Investigation

CO2-enhanced CT Imaging for the Detection of Pulmonary Emboli: Feasibility Study in a Porcine Model1 Kyongtae T. Bae, MD, PhD, Cheng Hong, MD, PhD, Christoph R. Becker, MD Srini Prasad, MD, Mark A. Nolte, RT, Paul E. Eisenbeis, LAT, Jay P. Heiken, MD

Rationale and Objectives. The authors investigated the feasibility of using computed tomography (CT) with CO2 gas as a negative contrast agent for detecting pulmonary emboli in a porcine model. Materials and Methods. Seven pigs with or without pulmonary emboli underwent thoracic imaging with multi– detector row spiral CT. To identify optimal injection and scanning protocols, the first four pigs were scanned repeatedly in the supine and prone positions with different scan delays (10, 15, and 20 seconds) and different volumes of CO2 (60, 120, 180, and 240 mL), which were hand infused (each infusion took 10 –15 seconds). The last five pigs with emboli were scanned with iodinated contrast medium and then rescanned with 120 or 180 mL of CO2. The CO2 volumes and scan delays were qualitatively assessed. The supine and prone CT scans and the number and location of thrombi depicted in the CO2- and contrast material– enhanced CT scans were compared. Results. Because the pulmonary artery in pigs is in the posterior anatomy, the prone position was more effective than the supine position with CO2 enhancement. An infusion of 120 mL of CO2 was sufficient to enhance the entire pulmonary artery, and scanning timed to coincide with the completion of infusion was the most effective. Both the CO2- and contrast-enhanced CT scans demonstrated all thrombi. Thrombi were more apparent on the CO2-enhanced CT scans than on the contrast-enhanced scans because of the high contrast interface between soft tissue and gas. However, two of the seven pigs with thrombi experienced abrupt cardiac arrest after CO2-enhanced scanning and could not be resuscitated. The cause of these events was not determined in the current study. Conclusion. The CT depiction of pulmonary emboli is feasible with CO2 gas as a negative contrast agent and may even be superior to that with iodinated contrast media. Further studies are required to evaluate the safety of this method and to develop an improved delivery of CO2 gas for this application. Key Words. Computed tomography (CT), contrast enhancement; CT, contrast media; carbon dioxide; pulmonary arteries, stenosis or obstruction. ©

AUR, 2003

One of the fastest-growing applications for computed tomographic (CT) angiography is in imaging of the pulmo-

Acad Radiol 2003; 10:313–320 1 From the Mallinckrodt Institute of Radiology, Washington University School of Medicine, 510 S Kingshighway Blvd, St Louis, MO 63110 (K.T.B., C.H., S.P., M.A.N., P.E.E., J.P.H.); and Institute of Clinical Radiology, University of Munich, Germany (C.R.B.). Received October 14, 2002; revision requested November 18; revision received and accepted November 25. Address correspondence to K.T.B.

©

AUR, 2003

nary arteries to assess patients with symptoms of pulmonary embolism (1). With a bolus of iodinated contrast medium injected into a peripheral or central vein, pulmonary emboli are detected as nonenhanced intraluminal filling defects on a background of brightly enhanced blood in the pulmonary arteries. Despite the use of low-osmolar contrast media and of premedication in some patients, iodinated contrast media are still associated with nephrotoxicity and other adverse reactions in some individuals (2). As an alternative to 313

BAE ET AL

iodinated contrast media, CO2 gas has been used for conventional angiography of patients with a contraindication to iodinated contrast medium. CO2-enhanced digital subtraction angiography has been very effective for assessing peripheral vascular disease and guiding vascular intervention (3,4). When injected intravascularly, iodinated contrast medium mixes with blood and increases the x-ray beam attenuation, resulting in positive contrast enhancement. Intravascularly injected CO2, however, does not mix with blood but forms a moving gaseous column that displaces blood and thus provides negative contrast enhancement. The solubility of CO2 in serum is approximately 20 times greater than that of oxygen (5). As a result, a small amount of CO2 injected to enhance contrast at arterial angiography is typically cleared before it reaches the capillary level. Animal studies performed in the 1950s (5–7) assessed the feasibility of CO2 as an intravenous contrast agent for conventional angiography. These studies were conducted mainly to visualize the cardiac or pericardiac anatomy and to evaluate the safety profile of CO2. To our knowledge, no previous study evaluating CO2 gas as an intravenous contrast agent for use with CT has been reported. The objective of this study was to test the feasibility of using CO2 gas as a negative CT contrast agent for detecting pulmonary embolism in a porcine model. MATERIALS AND METHODS Animal Preparation and CT Scanning Parameters All animal care and procedures performed in this study were approved by the institutional animal study committee at Washington University. Seven pigs (pigs A–G), each weighing 30 –35 kg, underwent several spiral CT scans of the thorax before and after CO2 or iodinated contrast medium was administered through a plastic venous catheter (20 gauge) placed in an ear vein. All CT scanning was performed with a Plus 4 Volume Zoom scanner (Siemens Medical Systems, Iselin, NJ) and a standard chest CT protocol for pulmonary embolus detection with a multi– detector row CT scanner. The scan parameters included 4.0 ⫻ 2.5 mm collimation, 120 kV, 120 effective mAs, 0.5-second gantry rotation, and pitch of 1.7 (a 17-mm table increment per gantry rotation divided the x-ray beam collimation width). Scan duration was approximately 10 seconds for a 30 –35-cm scan length. Prior to each CT scanning session, the pig was intubated and anesthetized with isoflurane (Halocarbon Laboratories, River Edge, NJ). The electrocardiogram and oxygen satu314

Academic Radiology, Vol 10, No 3, March 2003

Figure 1. CO2 delivery system consisting of pressurized CO2 tank, tube, 60-mL syringes, and stopcocks. This photograph was taken from the back of the CT gantry as a pig lay prone on the CT table. Three syringes were linked together by three-way stopcocks and filled with CO2 released from the tank. The syringes were connected by an injection tube to a vein in the pig’s ear.

ration were monitored constantly. Ventilation was suspended at end expiration to achieve a breath hold during scanning. Preparation and Delivery of CO2 Gas The CO2 gas for digital subtraction angiography was drawn as needed from a pressurized metal storage container (Puritan Medical Products, St Louis, Mo). A CO2 gas delivery system was constructed by linking 60-mL syringes (up to four syringes, for 240 mL of gas) with stopcocks and a tube (Fig 1). After one of the stopcocks was connected to the container, the valve of the container was opened to release the pressurized CO2 gas, which filled the syringe and repelled the plunger. The stopcock was switched and the filled syringe was emptied into room air. We repeated these steps two or three times at the beginning to flush any remaining air from the tube and the container. Then we filled each syringe sequentially by opening and closing the stopcocks. The container valve was turned off as soon as the plunger of each syringe was completely repelled, to prevent the syringe from becoming pressurized. The filled syringes were connected to the peripheral line, which was then subsequently cleared with CO2 gas

Academic Radiology, Vol 10, No 3, March 2003

by gently pushing forward the plunger of one of the syringes. When CT scans were performed, the full amount of CO2 in the syringes was hand infused over 10 –15 seconds by quickly turning the stopcocks in sequence. We contemplated using our standard CT power injector as an alternative to hand injection, but we chose the latter because of the potential for contamination by room air and for unwanted delivery of pressurized CO2 with standard CT power injection. Preparation of Pulmonary Emboli Autologous thrombi were generated by using a slightly modified version of the technique described by Brink et al (8). Fresh porcine blood (15–20 mL) obtained from the right jugular vein was placed in polyvinyl chloride tubing 30 cm long with a 10-mm-diameter internal bore. The tube was sealed and placed on a turntable inclined at 60° to the horizontal plane and rotating at approximately 16 revolutions per second for 30 – 45 minutes, allowing the blood to flow circularly through the tube until an adequate thrombus was formed. The thrombus was then removed from the tubing and divided into segments of approximately 20 ⫻ 4 mm. Four or five thrombi were delivered into the right jugular vein through a 12-F catheter. The thrombus diameter was chosen because it was the largest diameter found to be readily deliverable through this catheter. The thrombus length was determined for the same reason as proposed by Brink et al (8). CT Scanning of Normal Pigs and Pigs with Pulmonary Emboli To determine optimal injection and scanning protocols, the first four pigs (pigs A–D) of the group of seven were scanned repeatedly in the supine and prone positions with different volumes of CO2 gas (60, 120, 180, and 240 mL) and different scan delays (10, 15, and 20 seconds). Next, thrombi were introduced into the right jugular vein in the last five pigs (pigs C–G) of the seven, and the pigs were placed in the prone position on the CT scanner table. CT scans were obtained first with the iodinated contrast medium and then repeated with the CO2 gas used as a contrast agent. For contrast medium– enhanced CT, ioversol (Optiray 320, 320 mg I/mL; Tyco Healthcare Mallinckrodt, St Louis, Mo) was injected (2.0 mL/kg at 2.5 mL/sec) with a standard CT power injector (Tyco Healthcare Mallinckrodt). A scan delay of 15 seconds from the start of injection was used. For the CO2-enhanced CT, CO2 volumes of 120 or 180 mL were injected at approximately 15 mL/sec. The scan delay was also 15 seconds (typically the start of CT scanning coincided with

CT METHOD FOR DETECTING PULMONARY EMBOLI

the completion of the injection). This specific protocol, which provided consistent, effective CO2 negative enhancement, was determined during the early phase of the experiment by scanning pigs without pulmonary emboli. Finally, balloon-occlusive pulmonary angiography was performed in three pigs (pigs C, F, and G) to confirm the CT findings. The pulmonary angiogram was obtained by injecting an iodinated contrast medium selectively into the right or left main pulmonary artery via a balloon catheter that was introduced into the right jugular vein. Contrast Enhancement Quality and Detection of Emboli In this feasibility study with a limited number of pigs, no systematic or blinded analysis to grade the image quality was attempted. The CO2-enhanced CT images were displayed with a lung-window setting (center, ⫺550 HU; width, 1,800 HU), which allowed us to distinguish the gas-filled vessels from the surrounding parenchyma. The iodinated contrast– enhanced CT images were displayed with a soft-tissue window setting (center, 30 HU; width, 400 HU) for the visualization of enhanced vessels. The amount of negative enhancement in the pulmonary arteries with different CO2 volumes and scan delays was qualitatively assessed by consensus of two radiologists (K.T.B., C.H.) using side-by-side display. In addition, the CT images acquired when the pigs were in the supine position were compared with the images acquired when the pigs were in the prone position. The number and location of thrombi depicted in the CO2- and contrast-enhanced CT scans were compared in both axial and three-dimensional reformatted (coronal and sagittal multiplanar and maximum intensity projection) angiographic images. The reformatted images were obtained with the Wizard software provided by the manufacturer of the CT scanner (Siemens Medical Systems). RESULTS Because of its buoyant nature, CO2 gas displaced the blood in the nondependent portions of the cardiac chambers and pulmonary arteries. In supine pigs, the gas was distributed mainly in the anterior portions of the right side of the heart and the pulmonary arteries (Fig 2a, 2b). CT images of supine pigs after the administration of 120 mL of CO2 showed the gas collected in the anterior right ventricle and only a minimal amount of gas collected in the anterior lumen of the pulmonary artery. The incomplete opacification of the pulmonary artery in supine pigs was not substantially improved by increasing the amount of gas to 180 –240 mL. 315

BAE ET AL

Academic Radiology, Vol 10, No 3, March 2003

Figure 2. CO2-enhanced chest CT images of a normal pig (pig C) scanned in the (a, b) supine and (c, d) prone positions after the infusion of 120 mL of CO2. (a) CT image of a midcardiac section shows CO2 occupying the anterior right ventricle and a minimal amount of the gas in the lumen of the anterior pulmonary artery. (b) CT image of the lower lung level demonstrates insufficient CO2 enhancement of the pulmonary arteries. (c, d) Complete displacement of the blood in the pulmonary arteries is seen on (c) the CT image obtained with a 15-second delay (ie, at the end of the infusion), and reduced displacement is visible on (d) the CT image obtained with a 20-second scan delay.

Since the porcine pulmonary artery is located posteriorly in the thorax, the CO2 enhancement of the pulmonary artery was more effective in prone pigs than in supine pigs. Near-complete CO2 filling of the pulmonary artery was achieved consistently when the pigs were scanned in the prone position with infusion of 120 or 180 mL of CO2 gas and when the scan delay (15 seconds) coincided with the completion of infusion (Fig 2c). Scan316

ning in the middle of the infusion or more than 20 seconds after the completion of the infusion resulted in reduced filling of the pulmonary arteries (Fig 2d). Repeat CT scans obtained approximately 10 minutes after the initial CO2-enhanced scans showed no residual gas in the cardiac chambers (Fig 3). Pulmonary emboli were depicted on contrast-enhanced CT images as dark intraluminal filling defects surrounded by

Academic Radiology, Vol 10, No 3, March 2003

CT METHOD FOR DETECTING PULMONARY EMBOLI

Figure 3. CO2-enhanced supine CT images of a normal pig (pig G) scanned (a) immediately and (b) 10 minutes after infusion of 120 mL of CO2. Gas is no longer seen in the cardiac chambers on the repeat CT image obtained 10 minutes after infusion.

brightly enhanced blood, whereas they appeared on CO2enhanced CT images as intraluminal soft-tissue structures surrounded by the gas, which displaced the flowing blood (Fig 4). Small emboli were clearly visible on both the CO2and the contrast-enhanced CT images. The pulmonary emboli were more conspicuous on the CO2-enhanced CT images than on the contrast-enhanced CT images because of the higher contrast interface between soft tissue (embolus) and gas than between embolus and enhanced blood. Maximum intensity projections and multiplanar reformatted images aided the visualization of pulmonary emboli and overall thoracic anatomy (Fig 5). The number and location of pulmonary thrombi depicted on CT scans obtained with the contrast medium were the same as those depicted on CT scans obtained with CO2. These findings also were confirmed in some cases by balloon-occlusive angiography. The number and location of pulmonary emboli varied among the pigs. Most of the emboli were located in the lower lobe segmental pulmonary arteries. In pig C, no definite embolus was seen on either CO2- or contrast-enhanced CT images. We speculate that this apparent absence of emboli was due to a difference in the embolus delivery technique used in this pig, compared with the technique used in others. Emboli were introduced

while this pig was in the prone position but while the other pigs were in the supine position. The prone positioning might have contributed to a compression of the thrombus delivery catheter in the right jugular vein, resulting in the complete fragmentation of the introduced thrombi. Pulmonary emboli were detected in all pigs that were supine when the thrombi were introduced (pigs D–G). In pigs D and F, three emboli were detected in each pig (one in the right lung, and two in the left lung), and in pigs E and G, five emboli were detected in each (pig E had three in the right lung and two in the left, and pig G had four in the right lung and one in the left). Repeat CO2- and contrast-enhanced CT scanning was well tolerated in five pigs (A, B, C, F, G). However, two pigs (D and E) with pulmonary embolism experienced abrupt cardiac arrest after CO2-enhanced scanning and could not be resuscitated. No sign of acute distress was observed before or during the CT scanning. Repeat CT of the chest performed immediately after death did not reveal any noticeable unusual gas distribution in these animals that might have caused cardiac arrest. No autopsy was performed in the current study to investigate further the cause of death, because it was not a part of the approved protocol. 317

BAE ET AL

Academic Radiology, Vol 10, No 3, March 2003

Figure 4. CO2- and iodinated contrast-enhanced CT images of pulmonary emboli in (a, b) pig E and (c, d) pig F, acquired while the pigs were in the prone position. Small emboli (solid arrows) in the lower lobe segmental pulmonary arteries of both pigs are clearly depicted on CT images obtained both with CO2 (a, c) and with the iodinated contrast medium (b, d). As a normal reference for comparison, the right lower lobe segmental pulmonary artery in pig F, which does not have a filling defect, is depicted with full enhancement, as a black dot on the CO2-enhanced CT image (c, dotted arrow) and as a white dot on the iodinated contrast-enhanced CT image (d, dotted arrow).

DISCUSSION The role of CO2 as an alternative to iodinated contrast media for conventional anigography has been well established (3,4,9). CO2-enhanced digital subtraction angiography has provided accurate and useful vascular images in patients with a contraindication to iodinated contrast media. When CO2 is injected into a blood vessel, it displaces the blood and maintains an interface with it. Optimal CO2 injection volumes and injection rates depend on the blood volume and blood flow in the region of interest to be imaged. A small volume of injected CO2 may incompletely 318

displace the blood and result in inadequate negative enhancement. Insufficiently displaced blood in the pulmonary arteries may obscure pulmonary emboli or may result in a nondiagnostic study. In our porcine experiments, 4 mL/kg CO2 infused over 10 –15 seconds provided adequate filling of the pulmonary arteries in pigs scanned in the prone position. With this protocol, normally flowing blood in the pulmonary arteries was completely displaced by CO2 gas, and pulmonary emboli were clearly outlined. Because of its potential neurotoxicity (10,11), CO2 generally has not been used for contrast enhancement of blood vessels above the diaphragm—for example, in im-

Academic Radiology, Vol 10, No 3, March 2003

CT METHOD FOR DETECTING PULMONARY EMBOLI

Figure 5. Images of pulmonary emboli in pig F. Emboli (solid arrows) in the bilateral lower lobe segmental pulmonary arteries are depicted in the coronal plane in (a) a maximum intensity projection of a CT image obtained with contrast medium and (b) a curved multiplanar reformatted representation of a CT image obtained with CO2 contrast enhancement. The small embolus (dotted arrow) depicted in b does not appear in a because it has been obscured by the intensity of the surrounding, brightly enhanced blood. (c) A balloon-occlusive pulmonary angiogram obtained with the pig in the prone position confirmed the presence of two emboli in the left lung (arrows).

aging of coronary circulation, cerebral circulation, or the thoracic aorta. However, research studies performed with CO2-enhanced imaging of these vessels in animals have provided conflicting evidence of neurotoxicity. Coffey et al (10) reported neurologic deficits and pathologic changes of infarction after the injection of CO2 into the carotid arteries of rats, leading them to hypothesize that the injected CO2 had disrupted the blood-brain barrier. Conversely, other investigators have demonstrated that thoracic aortography and selective carotid arteriography may be performed safely with CO2, without pathologic changes detectable in neurologic examination or electroencephalographic monitoring, or gross pathologic changes, in dogs (12) and rabbits (13). CO2 injected for arterial angiography either is carried downstream by the blood to the level of the arterioles or capillaries or dissipates before reaching these levels. However, CO2 injected for venography may pool in the right atrium and the pulmonary arteries. Kerns et al (4) reported that inferior vena cavography performed with 70 mL of CO2 per injection was well tolerated in humans, and dissolution of the gas occurred within seconds. BoydKranis et al (14) also reported successful performance of CO2 vena cavography in most patients but noted an oc-

currence of transient systemic hypotension in one patient who had pulmonary hypertension. In general, no more than 200 mL of CO2 is recommended for venous injection in humans (4). The main contraindication for use of CO2 gas is severe respiratory disease (9). Patients with impaired CO2 excretion—for example, those with advanced pulmonary emphysema—may be at risk for CO2 narcosis (15). The physiologic effects of CO2 infused intravenously into the cardiac chambers and pulmonary arteries have been studied (5), and no statistically significant change in blood gases or hemodynamics has been detected even at 8 mL of CO2 per kilogram of body weight. However, Oppenheimer et al (5) found an elevation of right ventricular systolic pressure and acute systemic hypotension for several minutes after intravenous infusion of CO2. These changes may be due to CO2 lodged in the fine pulmonary vessels, leading to an increase in pulmonary peripheral resistance and pressure. The acute systemic hypotension may be due to a decrease in blood flow through the lungs to the left side of the heart when the right atrium is distended by CO2. It is plausible that these pulmonary physiologic changes, combined with the exacerbation of CO2 retention, may have contrib319

BAE ET AL

uted to the deaths of two pigs with pulmonary embolism during our study. Use of CO2 gas as a contrast medium for CT or magnetic resonance (MR) imaging has been very limited. Recently, CO2 has been tested for inferior vena cavography with MR imaging of a phantom and a pig (16). Because the CO2 bolus in a vessel lumen is rapidly replaced by incoming blood, the temporal resolution of data acquisition must be very high to cover a large volume of anatomy during adequate CO2 enhancement. Thus, to facilitate the use of CO2 gas for CT angiography, multi– detector row CT certainly has advantages over single detector row CT because of its faster data acquisition and superior spatial resolution. A major technical limitation of our study was our CO2 delivery system. Hand injection of CO2 gas was accomplished with minimal effort because the viscosity of CO2 is much lower than that of the iodinated contrast medium. Nevertheless, the operation of our syringe-stopcock CO2 delivery system was somewhat cumbersome. This system also is vulnerable to contamination by room air. Such contamination, we speculate, may have contributed to the deaths of two pigs with pulmonary emboli. We contemplated using a standard CT power injector for CO2 infusion but decided not to do so because we were skeptical of its reliability for air-tight sealing of CO2 gas from room air and for controlling the amount of gas delivered. Dedicated CO2 injectors have been manufactured in other countries but are not available in this country. In conclusion, our feasibility study demonstrates that CO2 is an effective CT contrast agent for detecting pulmonary embolism in a porcine model. It is capable of depicting pulmonary emboli with greater tissue contrast than that obtained with iodinated intravenous contrast media. However, further animal studies are required to evaluate its safety and to improve the delivery of CO2 gas for this application.

320

Academic Radiology, Vol 10, No 3, March 2003

REFERENCES 1. Bloomgarden DC, Rosen MP. Newer diagnostic modalities for pulmonary embolism: pulmonary angiography using CT and MR imaging compared with conventional angiography. Emerg Med Clin North Am 2001; 19:975–994. 2. Morcos SK, Thomsen HS. Adverse reactions to iodinated contrast media. Eur Radiol 2001; 11:1267–1275. 3. Huber PR, Leimbach ME, Lewis WL, Marshall JJ. CO2 angiography. Catheter Cardiovasc Interv 2002; 55:398 – 403. 4. Kerns SR, Hawkins IF Jr, Sabatelli FW. Current status of carbon dioxide angiography. Radiol Clin North Am 1995; 33:15–29. 5. Oppenheimer MJ, Durant TM, Stauffer HM, Stewart GH III, Lynch PR, Barrera F. In vivo visualization of intracardiac structures with gaseous carbon dioxide: cardiovascular-respiratory effects and associated changes in blood chemistry. Am J Physiol 1956; 186:325– 334. 6. Paul RE, Durant TM, Oppenheimer MJ, et al. Intravenous carbon dioxide for intracardiac gas contrast in the roentgen diagnosis of pericardial effusion and thickening. Am J Roentgenol Radium Ther Nucl Med 1957; 78:224 –225. 7. Scatliff JH, Kummer AJ, Janzen AH. The diagnosis of pericardial effusion with intracardiac carbon dioxide. Radiology 1959; 73:871– 883. 8. Brink JA, Woodard PK, Horesh L, et al. Depiction of pulmonary emboli with spiral CT: optimization of display window settings in a porcine model. Radiology 1997; 204:703–708. 9. Yang X, Manninen H, Soimakallio S. Carbon dioxide in vascular imaging and intervention. Acta Radiol 1995; 36:330 –337. 10. Coffey R, Quisling RG, Mickle JP, Hawkins IF Jr, Ballinger WB. The cerebrovascular effects of intraarterial CO2 in quantities required for diagnostic imaging. Radiology 1984; 151:405– 410. 11. Caridi JG, Hawkins IF Jr. CO2 digital subtraction angiography: potential complications and their prevention. J Vasc Interv Radiol 1997; 8:383–391. 12. Shifrin EG, Plich MB, Verstandig AG, Gomori M. Cerebral angiography with gaseous carbon dioxide CO2. J Cardiovasc Surg (Torino) 1990; 31:603– 606. 13. Dimakakos PB, Stefanopoulos T, Doufas AG, et al. The cerebral effects of carbon dioxide during digital subtraction angiography in the aortic arch and its branches in rabbits. AJNR Am J Neuroradiol 1998; 19:261–266. 14. Boyd-Kranis R, Sullivan KL, Eschelman DJ, Bonn J, Gardiner GA. Accuracy and safety of carbon dioxide inferior vena cavography. J Vasc Interv Radiol 1999; 10:1183–1189. 15. Turner AF, Meyers HI, Jacobson G, Lo W. Carbon dioxide cineangiocardiography in the diagnosis of pericardial disease. Am J Roentgenol Radium Ther Nucl Med 1966; 97:342–349. 16. Maes RM, Matheijssen NA, Pattynama PM, Krestin GP. The use of carbon dioxide in magnetic resonance angiography: a new type of black blood imaging. J Magn Reson Imaging 2000; 12:595–598.