Carbon Dioxide Digital Subtraction Angiography (CO₂ DSA)

A Comprehensive User Guide for All Operators 

VASCULAR DISEASE MANAGEMENT 2014;11(10):E221-E256

Abstract

In 1971 during a routine celiac axis injection, 70 cc of room air was inadvertently injected into a patient instead of iodinated contrast. Fortunately, there were no ill effects and despite the use of cut film at the time, Hawkins visualized the celiac axis and its branches as a negative image. Because of this incident, in combination with his previous knowledge of carbon dioxide (CO₂) in venous imaging, he began to study the intra-arterial use of CO₂ in animals. Following the safe, successful use in animals he applied the same principles to humans. Unfortunately, technology lagged behind his genius and the initial imaging was poor. Later, during the 1980s, there was the development of digital subtraction angiography, tilting tables and a safe, reliable CO₂ delivery system. As technology continued to improve, CO₂ evolved into a viable vascular imaging agent. Although used initially for renal failure and iodinated contrast allergy, the many unique properties of CO₂ yielded multiple advantages, which are now used in a multitude of scenarios alone or in combination with traditional contrast. It has now been used with great success in both adults and children for more than 3 decades with only limited reportable complications. Its safe use in children has been described and when performed in this age group the same principles apply as for adults. This paper describes the history and technique of CO₂ angiography for vascular procedures.

History

It wasn’t long after the discovery of x-rays by Conrad Roentgen in 1895 that gas was first used as an imaging agent. In 1914 room air was used with radiographs in an attempt to visualize the abdominal viscera and its abnormalities.¹ Less than a decade later room air, oxygen, or carbon dioxide was insufflated in the retroperitoneum to evaluate for masses.^2,3 Because of the problem of air emboli, room air and oxygen were eventually replaced with  carbon dioxide (CO₂), the solubility of which is 20 times to 30 times that of O₂.

Extrapolating from Cormak who in 1837 reportedly blew two large breaths into the jugular vein of a horse with no ill effects, CO₂ was later used in the 1950s and 1960s as a venous contrast agent to evaluate for pericardial effusion.^4-7 Patients were placed in the left lateral decubitus position and a cross table lateral x-ray could outline the width of the right atrium. Bendib et al performed 1,600 cases without complications, using a peripheral injection of 100 cc to 200 cc of CO2 to evaluate for pericardial effusion (Figure 1). Additionally in 1969 Hipona reported the safe use of CO2 for the evaluation of the inferior vena cava (IVC).⁸

Subsequently, the biggest advancement in the use of CO₂ as a contrast agent was fortuitous. In 1971 during a routine celiac axis injection, 70 cc of room air was inadvertently injected into a patient instead of iodinated contrast. Fortunately, there were no ill effects and despite the use of cut film at the time, Hawkins visualized the celiac axis and its branches as a negative image (Figure 2).⁹ Because of this incident and with his previous knowledge of CO₂ in venous imaging, he began to study the intra-arterial use of CO₂ in animals. Following the safe, successful use in animals he applied the same principles to humans.¹⁰ Unfortunately, technology lagged behind his genius and the initial imaging was poor. Later, during the 1980s, there was the development of digital subtraction angiography, tilting tables, and a safe, reliable delivery system.

As technology continued to improve, CO₂ evolved into a viable vascular imaging agent. Although used initially for renal failure and iodinated contrast allergy, the many unique properties of CO₂ yielded multiple advantages, which are now used in a multitude of scenarios alone or in combination with traditional contrast. It has now been used with great success in both adults and children for over three decades with only limited reportable complications. Its safe use in children has been described and when performed in this age group the same principles as the adult apply (Figure 3).^11,12

 Properties

To use CO₂ appropriately, a basic knowledge of its properties is essential. CO₂ is a nontoxic, nonflammable, buoyant, compressible gas that has low viscosity and is produced endogenously at approximately 200 cc to 250 cc per minute. It is a natural byproduct and there are approximately 120 liters of CO₂ stored in the soft tissues at one time.¹³ It is transported in the blood to the lungs by three mechanisms: dissolution directly in the blood (7%), bound to hemoglobin (10%), or predominantly carried as a bicarbonate ion (85%) (Figure 4).  

Because CO₂ is present endogenously there is no concern for allergy or renal toxicity, which has been confirmed by numerous animal and human studies.^9,10 Its viscosity is 1/400 that of iodinated contrast and it is also highly soluble, roughly 20 times to 30 times greater than O₂. Therefore it is less occlusive than other gases (Table 1). When administered intravascularly, it tends to dissolve within a vessel in 30 seconds to 60 seconds. In intravenous administration it is also removed from the lungs in one pass. If CO₂ persists in a vessel for more than 30 seconds it is either trapped (Figure 4) or there is room air contamination.

As opposed to traditional liquid agents, carbon dioxide does not mix with blood. In fact, CO₂ is lighter than blood and floats anterior to it. To render a representative image it must displace the blood in the vessel. As a result, the vessel is less dense and a negative image is obtained with digital subtraction angiography. The quality and accuracy of the image will depend on the amount of blood displaced by the CO₂(Figure 4). Typically, smaller vessels, especially those 10 mm or smaller, demonstrate a better correlation with iodinated contrast (Figure 5).^14-16 The difficulty sometimes lies in the much larger vessels where a higher volume of CO₂ is required. Furthermore, once the total volume of blood has been displaced, using a higher volume of CO₂ does not improve the vascular image.

Buoyancy must also be considered when using CO₂. Typically, a gas will rise to the nondependent surface of the vessel. If all blood is not displaced from the vessel it will readily demonstrate the anterior structures but potentially generate a spuriously smaller image of the larger feeding vessel. Again, it is imperative to displace as much blood as possible to generate a comparable image. 

Conversely, buoyancy can be used as an advantage when attempting to visualize anterior vessels such as the celiac as well as the superior mesenteric artery (SMA) and inferior mesenteric artery. Where CO₂ is particularly favorable is in renal transplants and reimplanted renal arteries. These vessels are easily demonstrated because of their anterior position. CO₂ has the added advantage that it does not present a risk of renal toxicity in this group of patients who are particularly susceptible to renal insufficiency (Figure 6).^17,18 Conversely, in vessels such as the native renal arteries, the left renal artery lies slightly posterior. In this scenario as well as others where vessels are more posterior, the patient can be placed in the partial decubitus position with the site of interest situated anteriorly. CO₂ will preferentially demonstrate these repositioned vessels.

In addition to its buoyancy, when CO₂ is administered into the vessel via a catheter it has the potential to fragment into random bubbles depending on how it is delivered. In an attempt to avoid this, the catheter should be purged prior to definitive delivery and a continuous, controlled delivery of the volume of choice should be given. Dr. Cho studied the best catheter to administer a uniform, organized bolus of gas to minimalize the bubbling effect. He found that an end-hole catheter yielded the best results (Figure 7).¹⁹

Again with respect to buoyancy, there is one instance in which it can be a significant detriment. When a blood gas interface is not present in an anterior structure such as an abdominal aortic aneurysm (AAA), the CO₂ may sit without dissolving and “trap” in that position (Figure 8). This can cause a problem in two different ways. First, if there is an anterior vessel arising from the trapped gas portion of the aneurysm such as the inferior mesenteric artery, it may preclude the flow of blood and potentially lead to ischemia. Secondly, if a bolus of CO₂ is stagnant, more occlusive gases such as N₂ and O₂ will be exchanged into the bolus due to partial pressure differences. Again this could potentially lead to ischemia. If trapping occurs it is usually evident on fluoroscopy or may produce symptoms of pain. It can easily be remedied by repeatedly changing the patient’s position to stimulate gas movement. Some operators describe aspirating the gas with a catheter, but the author has not found this useful. Regardless, despite this potential, the reported incidence of this occurrence leading to significant clinical problems is negligible. 

Although trapping is exceedingly rare, it can be exacerbated by the administration of excessive volumes of CO₂. An excessive volume can be caused by one exceptionally large volume or multiple smaller volumes delivered repetitively without allowing enough time for the CO₂ to dissolve. As described below a single sizeable bolus-causing trapping is extremely unlikely. The incidence of gas trapping most commonly arises if a typical large tank (usually 3 million cc) of CO₂ under pressure is misconnected to the delivery catheter allowing unfettered flow of gas into the vessel. Regarding repetitive small doses, it is best to wait at least 30 seconds to 60 seconds between injections to allow for CO₂ to be dissolved.

Concerning dose, Cho addressed the effect of various intravenous CO₂ volumes on heart rate, respiration rate, blood pressure (BP), pulmonary artery pressure, arterial blood saturation of O₂, partial pressure of CO₂, partial pressure of O₂, pH, and bicarbonate at 1, 3, 5, and 10 minutes.¹⁹ He concluded that a single dose up to 1.6 cc/Kg resulted in no changes in cardiopulmonary parameters (Figures 9-11).¹⁹ The dose can subsequently be repeated with impunity after 30 seconds to 60 seconds. This amounts to a single dose of 112 cc for a 70-Kg person. That volume is more than necessary for any clinical scenario. Following the study, Cho mentioned a caveat that intravenous CO₂ should be performed cautiously in patients with known pulmonary hypertension. 

As with any interventional procedure, patients undergoing CO₂ digital subtraction angiography (DSA) should have routine monitoring of ECG, pulse oximetry BP, respiratory rate, and heart rate. As a safety measure, Cho recommends monitoring BP 1, 2, and 3 minutes after the first CO₂ injection.

Another important divergent property of CO₂ when compared with liquid agents is its compressibility. This compressibility can affect delivery, including volume, in various ways. Because CO₂ is a gas it is extremely compressible under pressure. A 20-cc syringe can hold a volume of 200 cc if compressed sufficiently. This can affect intravascular injections. Initially, if you are trying to keep the individual dose below the volume causing changes in parameters set forth by Cho, it is imperative to know the amount of CO₂ delivered. Therefore the delivery system should be purged to atmospheric pressure to eliminate the compression and to avoid administering a larger dose than the amount indicated on the syringe. 

Additionally, we have found that when a compressed dose is delivered it can be explosive. This type of delivery can expand the vessels, which results in adverse symptoms. In the abdomen symptoms of abdominal pain, urge to defecate and nausea can occur. The veins, which have less media in their walls than arteries, are more susceptible to pain especially when rapidly expanded. It is hypothesized that the vascular stretch receptors are activated with larger volumes and more explosive administration.²⁰ Aside from causing undue pain to the patient, this pain usually results in motion that significantly degrades the acquired images. Moreover, compression, which can lead to explosive delivery, can cause CO₂ to reflux into inappropriate vessels, potentially resulting in complications. This is especially true of the cerebral vessels in which CO₂ should be avoided. Finally, to completely avoid explosive delivery, the delivery catheter should be purged of saline or blood prior to the definitive dose. Hawkins showed that by not purging the catheter, the CO₂ in the delivery syringe can compress and 95% of the dose is explosively delivered in the last 0.5 seconds.

To recapitulate, for avoiding these rare but potentially untoward events, compressed explosive delivery should be avoided. The delivery system should not be under pressure. Whatever system is being used should be purged to the atmosphere to equilibrium. Following this, the delivery syringe plunger should be gently advanced to purge the diagnostic catheter of saline or blood. Subsequently the appropriate dose can be delivered in a controlled and nonexplosive manner. Note that unlike a liquid contrast syringe, in which a smaller syringe generates more pressure, a larger syringe (usually 20 cc to 35 cc) should be employed with CO₂. If not, the gas may simply compress in the syringe without delivery, especially if the catheter is not purged. 

Probably the most significant property of CO₂ and the one that causes the most concern and discourages competent operators from using it is that it is invisible. Because it is invisible, contamination with more occlusive room air is a major concern of many operators. Knowledge of the sources of contamination and how to avoid them is imperative but relatively simple. During the inception of intravascular use of CO₂, Hawkins found that routine, reusable cylinders contained carbonic acid, rust, particulate matter and water.¹⁰ It is essential, therefore, that disposable sources of at least medical-grade CO₂ be utilized (Table 2). We suggest research grade, which is more pure. Not only does this avoid inappropriate embolization but it also avoids pain for the patient. 

Contamination may also result when a syringe of CO₂ is left open to room air. Because of partial pressure differences, room air will enter the syringe and replace the pure CO₂ at the rate of .2 cc per minute. So a 20 cc open syringe will contain 12 cc of room air after an hour. This can be avoided with the use of a closed delivery system that is not open to room air. Delivery systems, especially those that use a bag reservoir, should be purged three times to rid the system of residual room air.

Finally, delivery system connections should be secured, either with glue or luer lock. Loose connections can lead to the aspiration of room air. Some individuals also hypothesize that stopcocks are not impervious to gas introduction with aspiration. Therefore, stopcocks and aspiration should be kept to a minimum. 

In summation, contamination can be easily avoided by utilizing medical-grade CO₂, disposable sources (cylinder or cartridge), and a closed, purged delivery system with secured connections, utilizing few stopcocks and minimal to no aspiration. Delivery systems are discussed below.

Contraindications, Caveats, and Disadvantages

When utilizing gas as an imaging agent the biggest concern is the possibility of embolic occlusion and secondary ischemia. The rapid solubility of CO₂ allows for intravascular use but the coronary, thoracic aortic, and cerebral vessels are less forgiving and delivery into these vessels should be avoided. Likewise, CO₂ should never be delivered in or adjacent to the thoracic aorta especially in the prone position. Theoretically the buoyancy of CO₂ can dissipate into spinal vessels and cause ischemia from embolus or trapping. Although there are literature discussions to the contrary, most operators will avoid direct injections in these sensitive vessels.^21-25 Additionally, because of the tendency of CO₂ to reflux, it is prudent to avoid intra-arterial injections above the diaphragm. Similarly, when imaging dialysis interposition grafts or fistulas, the arterial limb should be examined cautiously. To reduce the possibility of central cerebral reflux the patient can be placed in the Trendelenburg position or a microcatheter can be inserted into the artery and a gentle antegrade angiogram can be performed. In fact it is good practice, regardless of the procedure, to refrain from arterial delivery with the patient’s head in the elevated position. Again, this reduces the potential for central reflux into the arterial cerebral and thoracic circulation. 

Other clinical scenarios that theoretically predispose a patient to untoward embolization include right-to-left shunts and the combination of pulmonary artery hypertension and a patent foramen ovale. These instances are extremely rare and mostly hypothetical but operators should be knowledgeable of this possibility. In the patients with known right to left shunts, CO₂ can be safely injected into the venous system with the patient’s position right side up (left lateral decubitus) to trap the gas in the right atrium away from the septal defect.

An additional rare contraindication is the use of nitrous oxide general anesthesia when using intravenous CO₂. When using this anesthetic there is the potential for N₂ residing in the soft tissues to diffuse into the CO₂ bubble, causing it to be 5 times to 6 times more occlusive. This potential scenario arises in transjugular intrahepatic portosystemic shunt (TIPS) patients in which CO₂ is used in general anesthesia for the procedure.

A recurrent concern for novice operators is the use of CO₂ in patients with chronic obstructive pulmonary disorder (COPD). Considering the small amounts of CO₂ necessary for imaging compared to endogenous production, it is unlikely that a clinical dose will cause a problem as long as the patient is breathing spontaneously. As a precautionary measure in COPD patients, it is suggested to allow more time between injections. Instead of the recommended 30 seconds to 60 seconds between injections in most routine patients, those with COPD should be increased to 2 minutes to allow for definitive dissolution.

It is important to remember the potential increase in radiation exposure to the operator and the patient when using CO₂ DSA. It is recommended that the frame rate for acquiring CO₂ images approximate 6 per second or more. This is double the frame rate for typical iodinated contrast imaging, which could potentially result in an increase in radiation by a factor of 4. However, this number is difficult to measure exactly because each scenario is different and finding the diagnosis or accomplishing the desired result using CO₂ may decrease the overall necessity for more runs. Finally, the consequences of not using CO₂ DSA may far exceed the potential adverse effects of a slight increase in radiation from CO₂ DSA use. Considering the potential to increase radiation exposure when performing CO₂ DSA, typical radiation protection steps should be employed.

In addition to the above contraindications, there are a few minor disadvantages that exist when changing from a fluid-based vascular contrast to carbon dioxide. The primary disadvantage is learning how to employ a gas-based delivery system as opposed to iodinated contrast. CO₂ is invisible, colorless, odorless, and cannot be seen or felt so the comfort level for use is much less than iodinated contrast. The operator must learn and feel confident in the type of delivery system. 

Additionally, CO₂ vascular imaging is typically not as dense, and patient motion can seriously affect the final product. Likewise, bowel gas motion can deteriorate abdominal images. For these reasons and because of the fact that gas properties are much different than traditional liquid agents, rendering diagnostic images and post-processing may be more labor intensive when using CO₂. Reducing patient motion by injecting CO₂ gently without explosive delivery is extremely beneficial. To reduce bowel gas motion some operators use glucagon 1 mg/mL intravenously prior to injection. Using a gastrointestinal compression device to assist in displacing the bowel has been suggested.²⁵ 

Nonallergic Quality of CO₂

Because CO₂ is disseminated throughout the body’s soft tissues, it is nonallergenic. This advantage is extremely beneficial when an operator is confronted with an emergent intravascular procedure in a patient with a severe allergy to iodinated contrast. Currently most operators utilize low osmolar nonionic contrast which has an incidence of allergic reaction of .7% to 3% and severe anaphylaxis of .02% to .04%.^26-28 Rarely, even with appropriate premedication, breakthrough can occur. One of the classic procedures for this indication is IVC filter placement, particularly when there is no time for classic premedication with prednisone. 

In addition to the emergent situation, there are scheduled allergic patients who accidentally are not pretreated or simply forget to take the prep. Often, a scheduled procedure takes a great deal of preparation, travel, and cost, so cancelling would be counter-productive and expensive. CO₂ DSA can be employed in many of these cases to avoid the additional cost and chaos.

Non-nephrotoxic Quality of CO₂

Undoubtedly the best advantage of CO₂ DSA is its lack of nephrotoxicity.^10,29-34 Contrast-induced nephropathy (CIN) is the third leading cause of hospital-acquired renal failure behind decreased renal perfusion and nephrotoxic medications.³⁵ The incidence of hospital-acquired contrast-induced nephropathy approximates 7%.³⁶ This is based on the definition of CIN representing an increase of serum creatinine of 0.5mg/dL or a 25% elevation over baseline. This usually begins within the first 24 hours and peaks up to 5 days following the offending incident. 

The significance of CIN was underscored by McCullough.³⁷ He compared individuals who developed hospital-acquired CIN to those who received contrast and maintained stable creatinine. Those with CIN had 5.5 times the incidence (34%) of in-hospital mortality. Those requiring dialysis had even higher rates of mortality. The duration of hospitalization was twice those without CIN accompanied by increased morbidity and cost. In addition, those developing CIN demonstrated chronic effects with increased (2 times) 1- and 2-year mortality. Cardiovascular events were the leading cause of increased morbidity and mortality in these patients.

Rihal et al prepared a retrospective analysis on approximately 7,500 in-hospital patients, of whom 3.3% developed CIN.³⁸ The in-hospital mortality was 22% for those developing CIN vs 1.4% who did not. The 1- and 5-year mortality rates were almost 4 times greater in the CIN group. Therefore the risk of death persists long term following discharge. It must be stated that the incidence of CIN in these two groups was all-inclusive and that the incidence of CIN was much higher proportionately in those with underlying renal insufficiency. The majority of patients with CIN do not undergo dialysis. Today with improved iodinated contrast the incidence of CIN requiring dialysis approximates 4% for those with renal insufficiency³⁹ and 3% for those undergoing percutaneous coronary intervention.⁴⁰

Early animal studies by Hawkins and others showed that CO₂ as an intravascular contrast agent did not affect renal function.⁹ Hawkins later went on to demonstrate this in humans as well. Comparing iodinated contrast, gadolinium, and CO₂ in renal insufficient patients, CO₂ was the only agent not demonstrating an elevation in creatinine.^41,42 It should be the first-line imaging agent in patients with renal insufficiency requiring vascular evaluation or intervention. Even if there are limitations to the CO₂ imaging it can be used in conjunction with limited diluted doses of iodinated contrast.

Although hydration and cessation of nephrotoxic drugs are helpful, reducing the volume of iodinated contrast is by far the best method for eliminating CIN.⁴³ CO₂ DSA can usually accomplish this alone or as an adjunct to dilute liquid contrast. Moreover, because of the solubility of CO₂ and the fact that it is eliminated via one pass through the lungs when given IV, there is no limit to the total dose delivered. As a result, unlimited volumes can be given provided that individual doses are given as previously discussed.

Low Viscosity

The viscosity of CO₂ is 1/400 that of iodinated contrast, permitting its delivery through smaller, less invasive catheters and needles. This is advantageous when using microcatheters in which sufficient volumes of thicker contrast may be difficult to deliver.

Carbon dioxide can easily be administered in significant doses and has the advantage of central reflux resulting in opacification of the entire vascular structure (Figure 12). As opposed to CO₂, contrast will only demonstrate distal to the catheter. This is especially useful in procedures such as renal stent deployment in which intervention is central to the catheter tip. The position of the stenosis and ostium can always be visualized using CO₂ reflux (Figure 13).⁴⁴ 

Likewise, CO₂ can be injected through (ultra fine) needles as small as 27 gauge. These needles are much less invasive and have been used successfully with CO₂ in the liver and spleen as well as peripheral venography (Figure 14). When injected in the splenic and hepatic parenchyma CO₂ will opacify and exit via the low-pressure veins. Considering the size and flexibility of the smaller needle, use in the liver and spleen can be done in the presence of ascites and does not require the correction of an underlying coagulopathy.

Because of its low viscosity, CO₂ can be injected through a catheter with the wire in place using a Y-adapter (Figure 15). This is advantageous when performing invasive procedures and at times when it is preferable to maintain wire access. The result of the procedure can be assessed while leaving the wire in place. The more viscous contrast would require removal of the wire and loss of access.

Similarly, CO₂ is extremely beneficial in the demonstration of hemorrhage (Figures 16-18).^45-49 It tends to exit the arterial rent much more readily than its thicker counterpart (described later). Although less consistent this can also be seen with arteriovenous (AV) fistulas and tumor vessels. 

Another potential benefit of CO₂ is the fact that it doesn’t mix with blood so it is not diluted. This property permits excellent central venous visualization from peripheral injections using small needles (Figure 19).

Finally, one of the biggest advantages of CO₂ is its cost. The typical cost for CO₂ is 3 cents per 100 cc, which is exponentially cheaper than iodinated contrast.

Indications: Alone or as an Adjunct to Iodinated Contrast Allergy

Iodinated contrast allergy is explained above. Emergent patients or those who for a variety of reasons did not receive prednisone prep can use CO₂ DSA if necessary.

Contrast-Induced Nephropathy

There are many clinical scenarios that predispose a patient to CIN. These include myeloma, diabetes, acute cardiac abnormalities, hypotension, nephrotoxic drugs and underlying renal disease. Unfortunately, one or more of these conditions is commonly present in patients requiring invasive vascular procedures. In these patients, the incidence of CIN is increased, but can be lessened with the use of CO₂ DSA. Although rare, CIN can also occur in patients without comorbidities. Most commonly, the incidence of CIN is related to the volume of iodinated contrast, the route of administration (intra-arterial carries higher risk than intravenous) and pre-existing renal insufficiency.^17,50-52

Of these 3 variables, only volume of contrast is controllable. Therefore, when performing procedures requiring significant volumes of contrast, CO₂ can be used alone or as an adjunct to decrease this possibility. More importantly, CO₂ should be the preferred contrast in evaluation and intervention of the renal artery in renal transplants and renal arterial reimplantation cases. CO₂ not only avoids CIN in these highly susceptible patients, but the position of the renal artery anatomy is ideal for visualization. These arteries typically arise anteriorly where CO₂ is best at imaging (Figure 20).

Hemorrhage

One area in which CO₂ DSA has been extremely beneficial is in the diagnosis of acute arterial hemorrhage (Figure 21). Regardless of the etiology, whether iatrogenic, traumatic, or GI bleed, delineating the origin of bleeding and treating it precipitously can lead to significantly less morbidity and mortality. Iodinated contrast has typically been ineffective at demonstrating the offending vessel and as a result large volumes are often utilized with no diagnosis. Hawkins reported that the use of CO₂ DSA has approximately 2.5 times the sensitivity for defining the acute hemorrhage when compared to iodinated contrast.⁴⁷ There are several reasons for this, the primary one being its low viscosity. It is much easier for a gas to pass through a pinhole as opposed to the more molasses-like iodinated contrast. In addition, CO₂ tends to be under pressure in the vasculature so it expands and appears cloud-like when it exits. Once out of the vessel it does not mix with the pool of blood and dilute away like liquid contrast. Finally, there is no capillary phase with CO₂ to obscure a small potential bleed. In combination, these factors lead to the improved sensitivity. 

Again, CO₂ is not the ultimate contrast agent, and there are shortcomings when it involves acute GI hemorrhage. Bowel motion can significantly degrade a CO₂ DSA image and preclude diagnosis. Some operators have addressed this by administering glucagon 1 mg/mL intravenously prior to the injection of CO₂. Compression of the abdomen by the image intensifier or paddle may also help. Occasionally, the offending vessel is small or posterior and the CO₂ refluxes centrally and does not define the entire vessel. In this instance, there are two options. One can place the catheter in a superselective position and allow reflux, or it is sometimes better to administer the bolus in the larger central vessel, such as the aorta, so that all vessels fill. Once identified, the offending vessel can be embolized appropriately. It is critical to realize that when performing post embolization follow-up with CO₂ DSA, it should be done extremely gently. If CO₂ is forced into the embolized vessel, the explosive delivery can cause the embolic thrombus to migrate and result in rebleeding. To prevent this, a small amount of dilute contrast can be utilized, instead of CO₂.

Peripheral Arterial Occlusive Disease

The incidence of peripheral arterial occlusive disease (PAOD) and endovascular repair is increasing. CO₂DSA is vital and extremely useful in this clinical scenario but it is currently underutilized. Successful correlation of CO₂ and iodinated contrast in PAOD was demonstrated by Seeger’s group in the mid 1990s.⁵² They demonstrated a 92% correlation with an increase to 100% when a small amount of iodinated contrast was administered. It is well known that PAOD exists as part of a systemic process. Patients with intermittent claudication and rest pain undergoing angiography and intervention commonly have concomitant renal artery disease and insufficiency. In a review of 127 patients presenting with intermittent claudication or lower-limb ischemia, approximately 45% had coexistent renal artery stenosis.  Seventeen percent had mild, 16% had severe, and 12% had bilateral renal artery disease.⁵³ In another review of 100 patients, greater than 50% had renal artery occlusion and unilateral or bilateral stenosis.⁵⁴ More importantly than the actual stenosis is the occult renal impairment that accompanies PAOD even with a normal serum creatinine. In one series of 76 PAOD patients requiring angiography with normal serum creatinine, 86% had subnormal creatinine clearance with 65% below 60 mL/min.⁵⁵ Another author reported that serum creatinine is inaccurate in 33% of patients 40 years to 49 years of age and 90% of patients older than 70.⁵⁶ This latter group is where most of the PAOD occurs. Considering these facts and the susceptibility of this group to CIN it would seem intuitive to use CO₂ as a contrast agent whenever possible. Again it can either eliminate or lower the total volume of iodinated contrast thereby lessening the potential for CIN.

Most operators begin by performing a common injection at the aortic bifurcation. If evaluation of the aorta is also warranted, this can be performed before or after the aortic bifurcation. Typically, a somewhat forceful injection of 30 cc to 60 cc CO₂ at the level of the renal arteries is sufficient. The left renal artery lies slightly posterior and if not visualized can be reexamined with a mild right lateral decubitus position, placing the left renal artery in a nondependent position (Figure 22). Some operators administer glucagon 1 mg/mL intravenously to reduce bowel gas motion. Using an end-hole catheter will provide the best CO₂ bolus. For better visualization it is also recommended to employ a catheter with a radiopaque tip if contrast is to be avoided. As described above, with aortic injections it is imperative to monitor for AAA and potential trapping of gas. CO₂ can be visualized fluoroscopically and if a stagnant bolus is seen the patient should be rotated from side to side to dissipate the gas.

After evaluating the abdominal aorta, the central axial and nonaxial arteries of the pelvis and lower extremities can usually be evaluated with nonselective injections at the bifurcation. Volumes of 15 cc to 30 cc are usually adequate. Elevation of the lower extremities by 15 degrees to 30 degrees assists in peripheral delivery and visualization.⁵⁷ When performing this maneuver with the presence of advanced PVD, it may be prudent to wait longer and lower the patient’s legs between injections to avoid trapping. Furthermore, selective injections can be made to improve imaging. It is not unusual to perform selective injections for visualization below the knee. Microcatheters are ideal for this purpose. Initially the catheter should be placed at the common femoral artery to include the profunda and collateral vessels. The catheter can then be advanced more peripherally towards the trifurcation as necessary. Gentle, nonexplosive doses of 5 cc to 10 cc are commonly sufficient (Figures 23-26). Regarding the dose, if it does not appear appropriate, it can be tailored to more or less without concern, as none of these injections will jeopardize renal function. If there is poor filling of the vessels, 1 mL of nitroglycerin (100-200mcg/mL) can be administered intra-arterially just prior to CO₂ injection. 

Stacking software can also be employed with post processing. This program allows the superimposition of multiple frames to generate one composite image (Figure 27). It takes the “bubble” effect out of CO₂ DSA and creates a solid vascular image. Finally, once the infrapopliteal vasculature is evaluated with CO₂ many authors in the literature advise administering one-quarter to one-half strength iodinated contrast in small volumes to confirm their diagnosis. Table 3 is a list of hints at obtaining better images during a peripheral run-off study. It is important to note that, very rarely, when peripheral vascular disease is so severe there can be pre capillary shunting of CO₂ from artery to vein. This mimics, but does not represent, an AV fistula. Close examination in multiple views and the appearance of peripheral vasculature will determine the true nature of the shunting.

Endovascular Abdominal Aneurysm Repair

One currently employed use of CO₂ DSA has been in the placement and evaluation of abdominal aortic endografts. This use has been proliferating among numerous operators for a variety of reasons, including decrease of iodinated contrast load, more sensitive evaluation of endoleaks, safety, and cost.^28-31 

With the maturation of endovascular abdominal aneurysm repair (EVAR), it has become clear that there is a propensity for developing renal failure that is permanent and cumulative.^58-61 The number affected exceeds those having open repair and is not limited to patients with underlying renal insufficiency. The typical EVAR patient is usually greater than 70 years of age. This group of individuals has approximately a 30% incidence of abnormally low glomerular filtration rate that may not be reflected by serum creatinine levels alone. They also tend to have comorbid conditions that predispose them to renal insult. The incidence of renal insufficiency in patients undergoing EVAR approximates 7% to 25% with acute renal failure occurring in 2% to 16%, resulting in an associated mortality of 30% to 50%.  Even when there is no pre-existing renal dysfunction, there is a 2.5% incidence of renal impairment.

Associated with this renal compromise is an immediate and delayed increase in mortality. Although these changes may be multifactorial, iodinated contrast has been implicated as the major precipitating factor. CO₂ can be used as the exclusive contrast agent or in addition to smaller volumes of iodinated contrast, allowing accurate and complete endovascular repair without inducing renal compromise. In addition, because of its low viscosity, it has also been noted by several of the above authors that CO₂ is more sensitive for detecting endoleaks. These benefits, plus safety, make it the ideal agent for the placement and evaluation of abdominal aortic endografts.

Typical doses of 20 cc to 50 cc per injection are utilized when evaluating endografts. One author described performing the entire exam without a catheter by simply injecting CO₂ through the groin access sheath side ports (Figure 28).³¹ 

Interventional Oncology

Interventional oncology is a recent catch-all phrase for a specialty that includes a multitude of minimally invasive procedures in the treatment of a variety of tumors. One aspect of this specialty is catheter-directed and requires angiography/venography and embolization. The contrast load for these patients is often high. This, combined with a host of other compromising factors, places these patients at risk for CIN.

Transarterial Chemoembolization, Drug-Eluting Beads, and Yttrium-90

There are procedures that use angiographically directed therapy predominantly for liver tumors. Transarterial chemoembolization (TACE), drug-eluting beads, and yttrium-90 therapy are utilized with liver tumor patients, who tend to be older and usually have other comorbidities such as renal insufficiency, diabetes, and hypertension. Patients with primary hepatic tumors also have significant underlying liver disease and are at risk for hepatorenal compromise. Approximately 75% of patients with cirrhosis will have renal insufficiency at some time during the course of their disease. At the same time, these individuals are often on renal-compromising medications and tend to be more hypovolemic. Cirrhosis, plus nonsteroidal anti-inflammatory drugs (NSAIDS) plus hypovolemia are a conduit to acute renal failure (ARF). Commonly, a large volume of contrast is required for interrogation of the vasculature, treatment, and follow-up exam. These risks can be exacerbated by post-embolization syndrome followed by decreased oral intake or very rarely non-target embolization and tumor lysis syndrome. All of these factors, especially the high volume of contrast, place the patient at risk for CIN. Alternatively, and perhaps more significantly, some of these patients may be denied life-prolonging therapy because of their elevated creatinine and risk for CIN (Figure 29). 

The problem appears to be under diagnosed; however, there are several reports that describe the incidence of CIN in this group of patients.^62-64 Huo, in a group of 140 TACE patients with a creatinine 1.1±0.2 of had an incidence of ARF of 8.6% and irreversible renal failure in 2.8%. All of those in the latter category had diabetes. Acute renal failure correlated with the number of TACE procedures, severity of liver disease, and presence of post-embolization syndrome. Jang prospectively evaluated 319 patients who had 463 TACE procedures and resultant ARF in 3.2%. Fifty-eight patients were denied TACE due to a creatinine of greater than 2 mg/dL. Acute renal failure correlated with a creatinine of greater than 1.5 mg/dL and the presence of ascites. Similarly, Yamazaki prospectively looked at 120 patients with 180 TACE procedures and a normal creatinine. There was ARF in 6% of patients.

Another procedure, uterine fibroid embolization (UFE), is usually performed in young, healthy individuals. Less invasive transarterial therapy can be precluded if there is an elevation of the serum creatinine. In fact, renal insufficiency is listed as one of the relative contraindications⁶⁵ and has been described.⁶⁶ Because of chronic bleeding these patients may have a decreased effective arterial blood volume and may also be subjected to NSAIDS for periprocedural therapy. As with TACE procedures, they may also be subjected to high volumes of iodinated contrast and develop post embolization syndrome (Figure 30).

Similarly, some patients with renal cell carcinoma require embolization of the primary tumor or the hypervascular metastasis. Obviously, many of these patients are susceptible to CIN. 

In each of the clinical scenarios delineated above, CO₂ can usually be used as the predominant contrast agent with the addition of a small or limited amount of dilute iodinated contrast. The vessels necessary for evaluation and treatment are less than 10 mm in diameter and have a strong correlation with liquid contrast. Doses ranging from 5 cc to 20 cc are more than sufficient. Occasionally the low viscosity of CO₂ will make tumor vessels more apparent. Carbon dioxide, because of reflux, can evaluate the celiac artery and its branches at the same time as the SMA. To visualize the patency of the portal vein, the microcatheter can be wedged peripherally in the artery and 10 cc to 20 cc of CO₂ administered. In the case of UFE, the uterine arteries are anterior and therefore easily demonstrated. Carbon dioxide can also be more readily injected through the commonly utilized microcatheters and give a better demonstration of the feeding vasculature. The reflux attribute of CO₂ will also permit visualization of the entire system of vessels to determine if the catheter is in the appropriate position and all culprit vessels are embolized. Most significantly, the use of CO₂ DSA not only permits inclusion of a group of patients that may be denied treatment but also reduces the incidence of CIN in all individuals undergoing the procedure.

Venous Evaluation and Treatment

As noted, CO₂ was used safely in the venous system early in the 1960s. Since the discovery and refinement of DSA, the myriad uses for CO₂ as a venous vascular contrast agent have proliferated. The gaseous properties noted above make it an ideal venous contrast agent. 

In the extremities, the low viscosity of CO₂ permits delivery of sufficient volumes through smaller 25 ga needles. This is less invasive and less painful to the patient. Because CO₂ doesn’t mix with blood, it is not diluted and a peripheral hand injection will yield good opacification centrally. This is unlike contrast. In the presence of venous occlusion, the properties of low viscosity and reflux will often demonstrate collateral cervical and thoracic veins bilaterally (Figure 31). This is critical in patients with renal insufficiency and chronic venous occlusive disease. Because of CIN and nephrogenic systemic fibrosis respectively, venous CT and venous MR cannot be utilized for a roadmap prior to venous access procedures. This can be circumvented nicely with CO₂ extremity venography. It can help determine venous patency as well as potential access sites. CO₂ venography can also be employed in patients with iodinated contrast allergy. It is unnecessary to prep and wait and the procedure can be performed immediately. Slow, controlled, nonexplosive doses of 5 cc to 20 cc usually produce excellent DSA images. It is important to remember that the thinner media in veins cause more sensitivity and pain to more forceful injections. The efficacy of venous CO₂ DSA has been documented by numerous authors.^67-69 

Carbon dioxide DSA can be extremely useful in the examination of dialysis fistulas or interposition grafts (Figure 32). It helps preserve what little renal function reserve some of these patients have. It is excellent at demonstrating the veins and central circulation. One caveat is the evaluation of the arterial anastomosis. It should not be evaluated via compression of central veins with forceful injection. Reflux to the cervical arterial supply can occur.¹⁶ When examining the arterial anastomosis it is helpful to place the patient in the Trendelenburg position and maneuver a microcatheter into the feeding artery. A slow gentle injection of 5 cc to 10 cc will usually show the anastomosis and pathology if present. With extremity and/or dialysis venous evaluation, if pathology is present, percutaneous transluminal angioplasty and stenting can also be performed using CO₂ DSA as the primary imaging agent.

Venous CO₂ DSA is advantageous in IVC evaluation, especially for filter placement. Many times these are emergent and patients may have allergy, have elevated creatinine, or do not need the additional volume of iodinated contrast boluses. Carbon dioxide has been shown to be a very effective imaging agent for the evaluation of the IVC, even at the bedside (Figure 33).^70-73 There is close correlation regarding size when compared to liquid contrast. Because of occasional undersizing with CO₂, borderline diameters should be confirmed with dilute contrast, if computed tomography (CT) is not available. 

As with contrast, the renal veins may not be readily apparent. Occasionally, the right renal vein lies slightly posterior and may not be visualized. If there is no prior imaging to locate the origin of the renal veins, the patient can be rotated so the ipsilateral side is up. Another simple method the authors have used rarely is to place a wire in the renal veins to determine the level of origin. 

When a CT is not available to review anatomy beforehand, we usually inject 20 cc to 30 cc into an end-hole catheter placed in the left common iliac vein to exclude left-side IVC. Some operators use 50 cc to 60 cc. If the renal veins are not evident, it helps to move the catheter more centrally, closer to their origins. If the anatomy is still uncertain, and the renal veins have not been identified, a selective catheter is advanced into the renal veins and CO₂ venography performed. 

Carbon dioxide renal venography can also be performed in this and in other procedures (adrenal vein sampling, balloon occluded retrograde transvenous obliteration of varices) where evaluation of these veins is essential. The one caveat, as addressed above, is caution in the patient with pulmonary hypertension. 

The next venous use of CO₂ involves a variety of clinical scenarios in the liver and spleen. It has been shown by Culp and Hawkins that, as opposed to iodinated contrast, CO₂ does not have any negative effect on the splenocytes or hepatocytes when injected directly into the respective parenchyma.⁷⁴ Hawkins showed that an injection of as little as 12 cc per second had a negative effect as compared to CO₂ at 200 cc/second.

One of the more unusual but effective uses of CO₂ is in patients with abnormalities of the splenoportal system (Figure 34).^12,75-77 Occasionally, three-dimensional imaging is not sufficient in defining the presence or absence of an abnormality. This is more common in children. In these cases, the splenic parenchyma can be accessed with a 25 or 27 gauge spinal needle using ultrasound guidance. Carbon dioxide of 10 cc to 20 cc can be administered. The low viscosity of CO₂ will cause opacification the splenoportal system, identifying the normality, pathology, or collaterals. Due to the small caliber of the access needles plus the use of CO₂, this has been shown to be safe without significant bleeding.^12,75-77

More commonly, it has become routine for many operators to use CO₂ in TIPS procedures (Figure 35). The incidence of renal compromise post TIPS approximates 2% to 3%. The entire procedure, including hepatic and portal venography as well as tract measurement, portal localization, and post-TIPS placement, can be performed with CO₂, reducing the possibility of CIN.¹⁰ Occasionally, in about 75% of patients, a tractogram can be performed with CO₂ with a measuring catheter in place. The low viscosity of CO₂ will sometimes simultaneously visualize the portal vein, tract, and IVC for stent measurement. As in many other clinical scenarios, a dose of 5 cc to 20 cc is sufficient for each step in TIPS.

Where CO₂ is most helpful in TIPS is in the localization of the portal vein. Since the inception of TIPS, localizing the portal vein has been the most difficult and invasive step. Many different methods, including ultrasound, CT, and angiography were attempted. Subsequently, it appeared that hepatic vein wedged catheter or later balloon occlusion injections of iodinated contrast were the most efficient method. 

Unfortunately, these injections caused hepatocyte congestion and infarction as well as elevation of transaminases, obscuring parenchymal staining and even rupture of the capsule and its consequences (Figure 36). Additionally, these injections unreliably visualized only a small portion of the portal vein.⁷⁸ Reese described using CO₂ in the same fashion.⁷⁹ The low viscosity permits excellent visualization of the portal vein in greater than 80% of cases and visualizes more of the system than iodinated contrast without most of the consequences. Carbon dioxide passes very easily through the sinusoids into the portal system against the direction of blood flow. An occasional benefit is that CO₂ remains in the portal vein as a target. However, because of a few complications of capsular rupture with catheter-wedged CO₂ injections, balloon occlusion of the hepatic vein with venous CO₂ injections became the preferred method of choice (Figure 37).^79-82 Complications are fewer with the advantage of better portal visualization. 

To simplify the process, Hawkins later demonstrated and described the parenchymal injection of CO₂ through the same needle used in localizing the portal vein (Figure 38).¹⁰ Typically, the hepatic vein is accessed and the needle angled in the direction of the portal vein. The needle is advanced to access the portal vein. Once in the parenchyma, 10 cc to 20 cc of CO₂ can be injected and the portal vein will be visualized almost 100% of the time without the adverse effects of other methods, especially capsular rupture.

If anterior or posterior location of the portal vein is not clear, repeat injections can be made with and without rotation of the image intensifier (Figure 39). Adopting this method eliminates the extra step and expense of the occlusion balloon, yields accurate and consistent visualization of the portal vein, and can be used with any TIPS needle. Visualizing the portal vein has also become more important with the rise in interventional oncology. Transhepatic portal vein access is necessary for certain procedures. More specifically, it is critical in performing selective portal vein embolization (PVE) to generate hepatic parenchymal hypertrophy of the remaining future liver remnant when extended hepatectomy is necessary.

Portal vein access can often times be difficult and can extend the duration of the procedure. Quick, safe, peripheral portal vein access without transgressing the hepatic artery or neoplasm is essential. Appropriate access also permits easier catheter manipulation and can reduce non-target embolization. Once again, to expedite this a 22-gauge needle can be placed peripherally in the hepatic parenchyma followed by injection of 10 cc to 20 cc CO₂ to demonstrate the portal vein anatomy (Figure 40). This can be used as a guide for advancement of the needle into the most appropriate portal vein radical and follow-up PVE procedure. Once accessed, CO₂ administration to the portal vein will reflux and demonstrate the entire vein as opposed to contrast, which only travels peripherally (Figure 41). This same approach can be utilized in any portal vein intervention, for example lysing portal vein thrombus, venous sampling, or treatment of portal vein strictures.

Rarely, three-dimensional imaging has failed to diagnose certain patency of the portal vein prior to liver transplant. Many of these patients also have concurrent ascites and coagulopathy. In this scenario, a 25-gauge needle can be placed with impunity into the liver without correcting ascites or the coagulopathy. Injection of 10 cc to 20 cc CO₂ can determine the critical presence or absence of patency of the portal vein (Figure 42).

Finally, there have been additional reports of limited numbers of patients using CO₂ for atypical procedures. One of these is in pain management. Carbon dioxide has been used instead of iodinated contrast prior to neurolysis when the patient is highly allergic.⁸³ Another newer area of use requiring additional investigation is for intraosseous venography in percutaneous vertebroplasty.⁸⁴ Reportedly the CO₂ does not obscure the procedure like iodinated contrast. Carbon dioxide can also be used in angioscopy.⁸⁵ It persists longer than saline, giving a clearer view of the vessel. Similarly it can be used percutaneously to separate organs prior to and during ablative procedures. The CO₂mmander/AngiAssist (AngioAdvancements) is ideal for this. It can be connected to the displacement needle and CO₂ is puffed in as needed. Another area of interest is using CO₂ with CT, especially in the angio suite.^86-88 Basically, there are a variety of interventional applications for CO₂ and the repertoire of procedures is increasing as operators become increasingly familiar with its capabilities and safety. 

Carbon Dioxide Delivery

Since the advent of intravascular CO₂ delivery there have been a number of innovative methods of delivery developed.^89-93 CO₂ delivery begins with a source. As stated above, a research-grade CO₂ should be used and because of potential impurities over time, the source should also be disposable. Next there must be a mechanism to deliver the CO₂ from the source. One must be aware that many of the commercial canisters contain 3 million cc of pressurized gas. Initial applications simply took a syringe of CO₂ from the source and then delivered it into the catheter. The downside of this approach is that a compressed volume of CO₂ is commonly within the syringe, which can result in the problems delineated above. Additionally, if the syringe is inadvertently left patent, over time, CO₂ can be replaced with more occlusive room air. 

To circumvent this, CO₂ delivery was initially performed with a typical liquid contrast injector. This set up had a number of weak links and to address the problems the Angioject dedicated CO₂ injector was designed by Angiodynamics (Figure 43). The cost was exorbitant and the efficacy and safety did not exceed homemade systems. The next tactic was to connect tubing from the source to the patient with a series of stopcocks. Again the problem with this is human error. If the stopcocks are positioned inappropriately, the highly pressurized gas in the cylinder can take the path of least resistance and overload the vascular system. 

To avoid the inadvertent overload of pressurized gas and the cumbersome presence of a large canister we introduced the use of a flaccid reservoir (bag) with a series of one-way valves (Figures 44, 45). This method used a converted fluid management system by Angiodynamics called the Angioflush III system. A similar system by Merit Medical is also used successfully (Figure 46). The theory behind these systems was that they would be a nonpressurized flaccid reservoir of CO₂ that would avoid explosive delivery and excessive volumes. There is a sufficient reservoir volume and the gas can be easily aspirated and delivered.

The one-way glued valves are intended to eliminate stopcocks, prevent room air contamination, and eliminate the necessity to remove the delivery syringe. These systems also are smaller, more mobile, and user friendly as opposed to classical large pressurized CO₂ canisters. The inherent problem in each apparatus is that the systems required assembly. Regardless of the training and simplicity incorrect assembly can result in air embolus. Additionally, the bag must be filled and purged three times to remove residual room air. This step is somewhat cumbersome and time consuming, especially when the decision to use CO₂ occurs spontaneously during the procedure. Other similar systems do exist (Figure 47).

For years, we have reported that the patient should never be connected directly to the cylinder. This has been modified recently due to the development of a K-valve stopcock that precludes the possibility of CO₂ passing directly from the canister to the patient (Figures 48, 49).

The next generation of delivery systems employs a compact regulator that uses a small 10,000-cc canister of pharmaceutical-grade CO₂(Figure 50). The smaller system can be placed in a sterile sleeve or left beneath the sterile drape. It is connected to a series of 2 tubes with one-way valves, as well as a K-valve and a reservoir and delivery syringe. The K-valve prevents direct communication with the patient. Carbon dioxide is introduced into the reservoir syringe. From the reservoir, the gas should be pushed, not aspirated, to the delivery syringe to avoid the unlikely possibility of air contamination through the K-valve. Equilibrium with the atmosphere can be achieved with a 3-way stopcock on the delivery catheter. The system does not require assembly and is extremely user friendly. Set-up for use takes approximately 1 minute. Another recently developed type of delivery is the Angiodroid CO₂ injector (Angiodroid) which utilizes digital vs hand injection (Figure 51).

Finally, some operators filter the CO₂ before it enters the vasculature. Carbon dioxide can be delivered and has been delivered without filtration for arteriography and venography, as the gas is not injected as an arterial contrast agent above the diaphragm. However, the use of a filter (0.2 micrometer pore size) can effectively remove particulate contamination and bacteria (0.5 to 5.0 micrometer). At the University of Michigan, infrared gas filtration (FDA approved filter, Syringe Pharmassure .2 micron with HT Tuffryn Membrane Pall Corporation #HP1002) is used for CO₂ delivery, with a new filter used for each procedure as part of the quality and safety program. 

Conclusion

Carbon dioxide is not the quintessential imaging agent but it offers unique properties when used alone or in combination with iodinated contrast that can expand diagnostic and therapeutic options in a variety of clinical scenarios. Used appropriately it is safe and can not only prevent CIN but also also offer life-extending procedures to patients who would otherwise have been precluded because of their underlying renal status. It can also be used less invasively or when traditional contrast fails to make diagnoses and avoids more significant intervention. Current technology permits simple, safe administration with images comparable to liquid contrast. It is an inexpensive, versatile tool that should be added to every interventionalist’s toolbox.

Disclosure: The authors have completed and returned the ICMJE Form for Disclosure of Potential Conflicts of Interest. Dr. Caridi reports consultancy to and reimbursements for educational programming from AngioAdvancements. All remaining authors report no related disclosures. 

Manuscript submitted July 9, 2014; manuscript accepted September 2, 2014. 

Address for correspondence: James G. Caridi, MD, FSIR, Tulane University School of Medicine, Department of Radiology SL-54 1430, Tulane, LA 70112, United States. Email: jimcaridi@gmail.com

REFERENCES

1. Rotenberg E. Rontgenophotographie der leber, der milz, und des zwerchfells. Deutsch Med Wschr. 1914;40:1205.
2. Rosenstein P. Pneumoradiology of kidney position-a new technique for the radiological representation of the kidneys and neighboring organs (suprarenal gland, spleen, liver). J Urol. 1921;15:447.
3. Carelli HH, Sorddelli E. A new procedure for examining the kidney. Rev Asoc Med Argentina. 1921;34:18-24.
4. Scatliff JH, Kummer AJ, Janzen AH. The diagnosis of pericardial effusion with intracardiac carbon dioxide. Radiology. 1959;73:871-883.
5. Paul RE, Durant TM, Oppenheimer MJ, Stauffer HM. Intravenous carbon dioxide intracardiac gas contrast in the roentgen diagnosis of pericardial effusion and thickening. Am J Roentgenol RadiumTher Nucl Med. 1957;78(2):224-225.
6. Bendib M, Tourni M, Boudjellab A. [CO₂ angiography and enlarged CO₂ angiography in cardiology (author’s transl)]. Ann Radiol (Paris). 1977;20:673-686.
7. Phillips JH, Burch GE, Hellinger R. The use of intracardiac carbon dioxide in the diagnosis of pericardial disease. Am Heart J. 1961;61:748-755. 
8. Hipona FA, Ferris EJ, Pick R. Capnocavaography: a new technique for examination of the inferior vena cava. Radiology. 1969:92:606-609.
9. Hawkins IF. Carbon dioxide digital subtraction arteriography. AJR Am J Roentgenol. 1982;139(1):19-24.
10. Hawkins IF, Caridi JG. Carbon dioxide (CO2) digital subtraction angiography: 26 year experience at the University of Florida. Eur Radiol. 1998;8(3):391-402.
11. Kriss VM, Cottrill CM, Gurley JC. Carbon dioxide (CO2) angiography in children. Periatr Radiol. 1997;27(10):807-810.
12. Caridi JG, Hawkins IF Jr, Cho K, et al. CO2 splenoportography: preliminary results. AJR Am J Roentgenol. 2003;180(5);1375-1378.
13. Roussos C, Koutsoukou A. Respiratory failure. Eur Respir J. 2003;22: Suppl. 47, 3s–14s.
14. McLennan G. Moresco KP, Patel NH, et al. Accuracy of CO₂ angiography in vessel diameter assessment: a comparative study of CO₂ versus iodinated contrast material in a porcine model. J Vasc Interv Radiol. 2001;12(8):985-989.
15. Moresco KP, Patel N, Johnson MS, Trobridge D, Bergan KA, Lalka SG. Accuracy of CO₂ Angiography in vessel diameter assessment: A comparative study of CO₂ versus iodinate contrast material in a aortioiliac flow model. J Vasc Interv Radiol. 2000;11(4):437-444.
16. Ehrman KO, Taber TE, Gaylord GM, Brown PB, Hage JP. Comparison of diagnostic accuracy with carbon dioxide versus iodinated contrast material in the imaging of hemodialysis access fistulas. J Vasc Interv Radiol. 1994;5(5):771-775.
17. Moresco KP, Patel NH, Namyslowski Y, Shah H, Johnson MS, Trerotola SO. Carbon dioxide angiography to the transplanted kidney: technical considerations and imaging findings. AJR Am J Roentgenol. 1998;171(5):1271-1276.
18. Hawkins IF, Cho KJ, Caridi JG. Carbon dioxide in angiography to reduce the risk of contrast-induced nephropathy. Radiol Clin North Am. 2009;47(5):813-825.
19. Cho KJ. CO₂ as a venous contrast agent: safety and tolerance. In: KJ Cho, IF Hawkins, eds. Carbon Dioxide Angiography: Principles, Techniques and Practices. New York: Informa Healthcare; 2007:37-44.
20. Rolland Y, Duvauferrier R, Lucas A, et al. Lower limb angiography: a prospective study comparing carbon dioxide with iodinated contrast material in 30 patients. AJR Am J Roentgenol. 1998;171(2):333-337.
21. Dimakakos P, Stefanopoulos T, Doufas A, et al. The cerebral effects of carbon dioxide digital subtraction angiography in the aortic arch and its branches in rabbits. AJNR Am J Neuroradiol. 1998;19(2):261-266.
22. Shifrin E, Plich M, Verstandig AG, Gomori M. Cerebral angiography with gaseous carbon dioxide CO₂. J Cardiovasc Surg (Torino). 1990;31(5):603-606.
23. Wilson AJ, Boxer MM. Neurotoxicity of angiographic carbon dioxide in the cerebral vasculature. Invest Radiol. 2002;37(10):542-551.
24. 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(2):405-410.
25. Stram E, Molgaard CP. Use of a compression paddle to displace bowel gas for carbon dioxide digital subtraction angiography. J Vasc Interv Radiol. 1999;10(4):405-408.
26. Trcka J, Schmidt C, Seitz CS, Bröcker E-B, Gross GE, Trautmann A. Anaphylaxis to iodinated contrast material: nonallergic hypersensitivityor IgE-mediated allergy? AJR Am J Roentgenol. 2008;190(3):666-670. 
27. Lieberman PL, Seigle RL. Reactions to radiocontrast material: anaphylactoid events in radiology. Clin Rev Allergy Immunol. 1999;17(4):469-496.
28. Caro JJ, Trindade E, McGregor M. The risks of death and of severe nonfatal reactions with high- vs low-osmolality contrast media: a metaanalysis. AJR Am J Roentgenol. 1991;156(4):825-832. 
29. Hawkins IF, Cho KJ, Caridi JG. Carbon dioxide angiography to reduce the risk of contrast-induced nephropathy. Radiol Clin North Am. 2009;47(5):813-825.
30. Hawkins IF Jr, Wilcox CS, Kerns SR, Sabatelli FW. CO₂ digital angiography: a safer contrast agent for renal vascular imaging? Am J Kidney Dis. 1994;24(4):685-694.
31. Beese RC, Bees NR, Belli AM. Renal angiography using carbon dioxide. Br J Radiol. 2000;73:3-6.
32. Caridi JG, Stavropoulos SW, Hawkins IF Jr. CO₂ digital subtraction angiography for renal artery angioplasty in high-risk patients. AJR Am J Roentgenol. 1999;173(6):1551-1556.
33. Hawkins IF, Mladinich CJ, Drane WE, et al. Effects of CO₂ angiography on renal function. J Vasc Interv Radiol. 1992;3:6. 
34. Thomson KR, Tello R, Sullivan R, Dixon RG, Becker G, Mitchell PJ. Carbon dioxide angiography. Asian Oceanian J Radiol. 1996;1:20-23.
35. Nash K, Hafeez A, Hou S. Hospital-acquired renal insufficiency. Am J Kidney Dis. 2002;39(5):930-936.
36. Bartholomew BA, Harjai KJ, Dukkipati S. Impact of nephropathy after percutaneous coronary intervention and a method for risk stratification. Am J Cardiol. 2004;93(12):1515-1519.
37. McCullough PA. Contrast-induced acute kidney injury. J Am Coll Cardiol. 2008;51(15):1419-1428.
38. Rihal CS, Textor SC, Grill DE, et al. Incidence and prognostic importance of acute renal failure after percutaneous coronary intervention. Circulation. 2002;105(19):2259-2264.
39. Nikolsky E, Mehran R, Turcot DB, et al. Impact of chronic kidney disease on prognosis of patients with diabetes mellitus treated with percutaneous coronary intervention. Am J Cardiol. 2004;94(3):300-305.
40. Marenzi G, Lauri G, Assanelli E, et al. Contrast-induced nephropathy in patients undergoing primary angioplasty for acute myocardial infarction. J Am Coll Cardiol. 2004;44(9):1780-1785.
41. Spinosa DJ, Matsumoto AH, Angle JF, et al. Gadolinium-based contrast and carbon dioxide angiography to evaluate renal transplants for vascular causes of renal insufficiency and accelerated hypertension. J Vasc Interv Radiol. 1998;9(6):909-916.
42. Spinosa DJ, Angle JF, Hagspiel KD, Kern JA, Hartwell GD, Matsumoto AH. Lower extremity arteriography with use of iodinated contrast material or gadodiamide to supplement CO₂ angiography in patients with renal insufficiency. J Vasc Interv Radiol. 2000;11(1):35-43.
43. Liss P, Eklof H, Hellberg O, et al. Renal effects of CO₂ and iodinated contrast media in patients undergoing renovascular intervention: a prospective, randomized study. J Vasc Interv Radiol. 2005;16(1):57-65
44. Caridi JG, Stavropolous W, Hawkins IF Jr. CO₂ digital subtraction angiography for renal artery angioplasty in high-risk patients. AJR Am J Roentgenol. 1999;173(6):1551-1556.
45. Hashimoto S, Hashimoto K. Soto M. CO₂ as an intra-arterial digital subtraction angiography (IADSA) agent in the management of trauma. Semin Intervent Radiol. 1997;14:163-173.
46. Hawkins Jr IF, Caridi JG, LeVeen RF, Klioze SD, Mladinich CRJ. Use of carbon dioxide for the detection of gastrointestinal bleeding. Tech Vasc Interv Radiol. 2000;3(3):130-138.
47. Hawkins IF, Caridi JG, Wiechman BN, Kerns SR. Carbon dioxide (CO₂) digital subtraction angiography in trauma patients. Semin Intervent Radiol. 1997;14:175-180.
48. Sandhu C, Buckenham TM, Belli AM. Using CO₂-enhanced arteriography to investigate acute gastrointestinal hemorrhage. AJR Am J Roentgenol. 1999;173(5):1399-1401.
49. Krajina A, Lojík M, Rejchrt S, et al. Carbon dioxide arteriography in the detection of acute massive gastrointestinal bleeding. Folia Gastroenterol Hepatol. 2004;2(1):8-12.
50. Kooiman J, Pasha SM, Zondag W, et al. Meta-analysis: serum creatinine changes following contrast-enhanced CT imaging. Eur J Radiol. 2012;81(10):2554-2561.
51. Murakami R, Hayashi H, Sugizaki K, et al. Contrast-induced nephropathy in patients with renal insufficiency undergoing contrast-enhanced MDCT. Eur Radiol. 2012;22(10):2147-2152.
52. Seeger JM, Self S, Harward TR, Flynn TC, Hawkins IF Jr. Carbon dioxide gas as an arterial contrast agent. Ann Surg. 1993;217(6):697-698.
53. Missouris CG, Buckenham T, Cappuccio FP, MacGregor GA. Renal artery stenosis: a common and important problem in patients with peripheral vascular disease. Am J Med. 1994;96(1):10-14.
54. Choudhri AH, Cleland JG, Rowlands PC, Tran TL, McCarty M, al-Kutoubi MA. Unsuspected renal artery stenosis in peripheral vascular disease. BMJ. 1990;301(6762):1197–1198.
55. Rashid ST, Salman M, Agarwal S, Hamilton G. Occult renal impairment is common in patients with peripheral vascular disease and normal serum creatinine. Eur J Vasc Endovasc Surg. 2006;32(3):294-299.
56. Kappel J, Calissi P. Nephrology: safe drug prescribing for patients with renal insufficiency. CMAJ. 2002;166(4):473-477.
57. Hawkins IF. Aortogram and runoff. In: Cho KJ, Hawkins IF, eds. Carbon Dioxide Angiography: Principles, Techniques, and Practices. New York: Informa Healthcare; 2007:53-68.  
58. Walsh SR, Tang TY, Boyle JR. Renal consequences of endovascular abdominal aortic aneurysm repair. J Endovasc Ther. 2008;15(1):73-82.
59. Chao A, Major K, Kumar SR, et al. Carbon dioxide digital subtraction angiography-assisted endovascular aortic aneurysm repair in the azotemic patient. J Vasc Surg. 2007;45(3):451-460.
60. Gahlen J, Hansmann J, Schumacher H, Seelos R, Richter GM, Allenberg JR. Carbon dioxide angiography for endovascular grafting in high risk patients with infrarenal abdominal aortic aneurysms. J Vasc Surg. 2001:33(3):646-649.
61. Criado E, Kabbani L, Cho K. Catheter-less angiography for endovascular aortic aneurysm repair: A new application of carbon dioxide as a contrast agent. J Vasc Surg. 2008;48(3):527-534.
62. Huo TI, Wu JC, Lee PC, Chang FY, Lee SD. Incidence and risk factors for acute renal failure in patients with hepatocellular carcinoma undergoing transarterial chemoembolization: a prospective study. Liver Int. 2004;24(3):210–215.
63. Jang BK, Lee SH, Chung WJ, et al. Incidence and risk factors of acute renal failure after transcatheter arterial chemoembolization for hepatocellular carcinoma. Korean J Hepatol. 2008;14(2):168-177.
64. Yamazaki H, Oi H, Matsushita M, et al. Renal cortical retention on delayed CT and nephropathy following transcatheter arterial chemoembolisation. Brit J Radiol. 2001:74(884);695-700.
65. Stokes LS, Wallace MJ, Godwin RB, et al. Quality improvement guidelines for uterine artery embolization for symptomatic leiomyomas. J Vasc Interv Radiol. 2010;21(8):1153-1163.
66. Rastogi S, Wu YH, Shlansky-Goldberg, Stavropoulos SW. Acute renal failure after uterine artery embolization. Cardiovasc Intervent Radiol. 2004;27(5):549-550.
67. Sullivan K, Bonn J, Shapiro M, Gardiner GA. Venography with carbon dioxide as a contrast agent. Cardiovasc Intervent Radiol. 1995;18(3):141-145.
68. Heye S, Maleux G, Marchal GJ. Upper-extremity venography: CO₂ versus iodinated contrast media. Radiology. 2006;241(1):291-297.
69. Moos JM, Ham SW, Han SM, et al. Safety of carbon dioxide digital subtraction angiography. Arch Surg. 2011;146(12):1428-1432.
70. Pessanha de Rezende M, Massiere B, Von Ristow A, et al. Carbon dioxide use as contrast for vena cava filter implantation: case series. J Vasc Bras. 2011;11(1):18-21.
71. Holtzman R, Lottenberg L, Bass T, Saridakis A, Bennett VJ, Carrillo EH. Comparison of carbon dioxide and iodinated contrast for cavography prior to inferior vena cava filter placement. Am J Surg. 2003;185(4):364-368.
72. 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(9):1183-1189.
73. Sing RF, Stackhouse DJ, Jacobs DG, Heniford BT. Safety and accuracy of bedside carbon dioxide cavography for the insertion of inferior vena cava filters in the intensive care unit. J Am Coll Surg. 2001;192(2):168-171.
74. Culp WC, Mladinich CR, Hawkins IF Jr. Comparison of hepatic damage from direct injections of iodinated contrast agents and carbon dioxide. J Vasc Interv Radiol. 1999;10(9);1265-1270.
75. Burke CT, Weeks SM, Mauro MA, Jaques PF. CO₂ splenoportography for evaluation the splenic and portal vein after liver transplantation. J Vasc Interv Radiol. 2004;15(10):1161-1165.
76. Teng GJ, Deng G, Liu ZS, et al. Ultrafine needle CO₂ splenoportography: a comparative investigation with transarterial portography and MR portography. Eur J Radiol. 2006;59(3):393-400.
77. Cho KJ, Cho DR. CO₂ digital subtraction splenoportography with the “skinny” needle experimental study in a swine model. Cardiovasc Interv Radiol. 2003;26(3):273-276.
78. Castaneda-Zuniga WR, Jauregui H, Rysavy JA, Formanek A, Amplatz K. Complications of wedge hepatic venography. Radiology. 1978;126(1):53-56.
79. Rees CR, Niblettt RL, Lee SP, Diamond NG, Crippin JS. Use of carbon dioxide as a contrast medium for transjugular intrahepatic portosystemic shunt procedures. J Vasc Interv Radiol. 1994;5(2):383-386. 
80. Semba CP, Saperstein L, Nyman U, Dake MD. Hepatic laceration from wedged venography performed before transjugular intrahepatic portosystemic shunt placement. J Vasc Interv Radiol. 1996;7(1):143-146.
81. Theuerkauf I, Strunk H, Brensing KA et al. Infarction and laceration of liver parenchyma caused by wedged CO₂ venograpy before TIPS insertion. Cardiovasc Intervent Radiol. 2001;24(1):64-67.
82. Taylor FC, Smith DC, Watkins GE, Kohne RE, Suh RD. Balloon occlusion versus wedged hepatic venography using carbon dioxide for portal vein opacification during TIPS. Cardiovasc Intervent Radiol. 1999;22(2):150-151.
83. Hirata K, Higa K, Shono S, Hirota K, Shinokuma T. Splanchnic neurolysis using carbon dioxide as the contrast agent. Reg Anesth Pain Med. 2003;28(1):68-69.
84. Tanigawa N, Komemushi A, Kariya S, Kojima H, Sawada S. Intraosseous venography with carbon dioxide contrast agent in percutaneous vertebroplasty. AJR Am J Roentgenol. 2005;184(2):567-570.
85. Silverman SH, Mladinich CJ, Hawkins IF Jr, Abela GS, Seeger JM. The use of carbon dioxide gas to displace flowing blood during angioscopy. J Vasc Surg. 1989;10(3):313-317.
86. Sonoda A, Nitta N, Ushio N, et al. 320-row multidetector CT angiography for hepatocellular carcinoma using CO₂ gas instead of iodinated contrast agents: Experiment and preliminary clinical study. Poster presented at: European Congress of Radiology, March 3-7, 2011, Vienna, Austria. Poster C-1123.
87. Penzkofer T, Slebocki K, Grommes J, et al. Carbon dioxide-Contrasted computed tomography angiography: high pitch protocols and adapted detection parameters improve imaging quality. Rofo. 2013;185(2):128-135.
88. Wong AA, Charalel RA, Louie JD, Sze DY. Carbon dioxide contrast enhancement for c-arm CT utility for treatment planning during hepaic embolization procedures. J Vasc Interv Radiol. 2013;24(7):975-980.
89. Alexander JQ. CO2 angiography in lower extremity arterial disease. Endovascular Today. 2011;27-34. Available at https://evtoday.com/2011/09/cosub2sub-angiography-in-lower-extremity-art…;
90. Cherian MP, Mehta P, Gupta P, Kalyanpur TM, Jayesh SR, Rupa R. Technical note: A simple and effective CO2 delivery system for angiography using a blood bag. Indian J Radiol Imaging. 2009; 19(3):203-205. 
91. Mendes CA, Wolosker N, Krutman M. A simple homemade carbon dioxide delivery system for endovascular procedures in the iliofemoral arteries. Circ J. 2013;77(3):831. 
92. Cronin P, Patel JV, Kessel, DO, Robertson I, McPherson SJ. Carbon dioxide angiography: a simple and safe system of delivery. Clin Radiol. 2005;60(1):123-125.
93. Hawkins IF Jr, Caridi JG, Kerns SR. Plastic bag delivery system for hand injection of carbon dioxide. AJR Am J Roentgenol. 1995;165(6):1487-1489.