Carbon Dioxide Angiography

Article Author:
Michael Young
Article Editor:
Jay Mohan
9/14/2019 6:38:28 PM
PubMed Link:
Carbon Dioxide Angiography


Room air was used as a radiographic contrast (Rotenberg 1914) prior to carbon dioxide (CO2) (Rosenstein 1921). Intravascular contrast was first used in 1924 as liquid (Brooks 1924). CO2 was studied in the arteries and veins of human patients, first via needle injection (Barrera 1956) and then via catheter delivery.[1]

With the development of digital subtraction angiography (DSA), CO2 angiography has became a useful diagnostic test, particularly in situations where the patient is hypersensitive to iodinated contrast material or those who have compromised renal function. As a result of high-resolution DSA and more reliable gas delivery systems, CO2 angiography is now commonly used for vascular imaging and endovascular procedures.

Since CO2 angiography is not associated with allergic reactions or nephrotoxicity, it is commonly used as a contrast agent for aortography, outflow assessment, visceral angiography, and renal arteriography. This contrast agent is most commonly used:

  • Central venography of the upper extremity
  • Wedged hepatic venography in order to visualize the portal venous system before the transjugular intrahepatic portosystemic shunt
  • For fine-needle TIPS procedures

CO2 is also used to guide vascular interventions, transcatheter embolization, and endovascular abdominal aortic aneurysm repair.


The main benefits of using CO2 for angiography are that CO2 has no adverse effect on the kidneys or the immune system, and it is the least expensive contrast medium.

CO2 can be used for a variety of procedures:


  • Renal arteries
  • Mesenteric arteries
  • Uterine arteries
  • Peripheral arteries

Wedged portal venography 

  • During a transjugular liver biopsy or transjugular intrahepatic portosystemic shunt (TIPS) creation


  • Prior to inferior vena cava filter placement
  • Dialysis fistula/shunt repair               
  • During vertebroplasty/kyphoplasty

Given its lower viscosity, CO2 is less likely than a fluid contrast to cause hepatic capsular rupture on a physiologic level,[2] but no prospective trial has evaluated whether there is a statistical difference between the 2 in clinical use.


As with liquid contrast, CO2 raises pulmonary artery pressure and can exacerbate pulmonary hypertension. It is important to check the pulmonary arteries for the accumulation of gas and dissipation within 30 to 45 seconds.

CO2 should not be combined with nitrous oxide sedation because N2 mixes with CO2 and reduces the solubility of CO2.

CO2 arteriography should not be used above the diaphragm to avoid the possibility of causing a cerebral air embolism. There are 2 mechanisms for this to occur:

  • From a known or unknown right to left cardiopulmonary shunt (cardiac septal defect, pulmonary arteriovenous malformation) 
  • From reflux retrograde within an artery to the brain

It is therefore prudent to have the patient in Trendelenburg position whenever not working deep in the mesenteric system.


Capnography (ETCO2) provides a way to monitor both ventilatory and intravascular CO2 in real time.

CO2 Preparation

There is only 1 FDA-approved medical CO2 delivery system. It holds 10,000 mL of CO2, which is enough to use with hundreds of patients. The same manufacturer sells a stopcock system (also called a K-valve, because of its shape). This apparatus has:

  • Valves directing gas in a single direction from the CO2 source toward the patient that prevents contamination of CO2 by room air,
  • A 60-mL reservoir syringe that helps the CO2 depressurize, and
  • A 30-mL injection syringe that helps deliver CO2 in a nonexplosive fashion into an angiographic catheter. 

Techniques that derive CO2 from large compressed CO2 cylinders have been used for decades but are not FDA-approved). Similar to a medical O2 cylinder, a medical grade CO2 cylinder has a metal diaphragm to keep the gas inside pure, a release valve, and a pressure gauge and pressure regulator. Single cylinders for purchase are typically sold in quantities of pounds of compressed CO2 and contain millions of mL of the CO2 set to around 18 PSI. Use of such CO2 cylinders requires a "homemade" simulation of the K valve system of the AngiAssist.[5] Whatever technique is chosen must allow passive unidirectional flow (via a series of valves) of CO2 from the cylinder into the syringes, tubing, and/or reservoir bags. This process purges room air from the system and allows CO2 to expand until it equalizes with room atmospheric pressure. The entry and exit points of the system are then sealed until the physician is ready to connect the system to a catheter.[3]

CO2 injection systems can be used in conjunction with

  • An additional gas purifying filter
  • A digital injector 
  • A Y-shaped connector to perform angiography without having to remove the guidewire from the catheter.

CO2 System to Catheter Hook Up

The fluid (blood/saline) in the angiographic catheter must be purged to prevent vessel dissection from explosive delivery of fluid during CO2 angiography. It is performed in a manner similar to hookup of a catheter to other angiographic contrast. One technique, the stopcock and waste syringe technique is as follows:

  • A 3-way stopcock is applied such that one end is connected to the CO2 source, one end is connected to a waste syringe, and one end is connected to the catheter. 
  • The stopcock is turned off to CO2 to allow blood to flow retrograde from the patient into the waste syringe. 
  • The stopcock is closed to the waste syringe. 
  • CO2 is injected slowly until it completely fills the catheter, at which time there is usually a palpable sensation of decreased resistance to injection.  

If the injection is performed by hand power, a large syringe is less likely than a small syringe to result in CO2 compression in the syringe followed by explosive delivery into the artery or organ.

An end-hole catheter yielded the best results, even in the aorta, IVC, and pulmonary arteries (where pigtail catheters are traditionally used for liquid contrast).


Injection Parameters

As with fluid contrast, the injection rate depends on the caliber and size of the vessel accepting the bolus and the size of the downstream vascular bed. The following volumes are ranges for amounts that are "usually" sufficient. 

Abdominal aortogram/inferior vena cavogram

  • Thirty to 40 mL; up to 60 mL may be needed in the aorta

Aortic branches (celiac, superior mesenteric, renal arteries), common iliac arteries and veins

  • Twenty to 30 mL

Wedged portal venography via the superior mesenteric artery

  • Ten to 20 mL

Common femoral arteries, second order arteries off the aorta, vessels requiring the use of a microcatheter, other veins, wedged venography (in the liver or spleen)

  • 5 to 10 mL; up to 20 mL may be needed

Proximal arteries can be imaged by refluxing CO2 from a more peripheral catheter location.  Digital subtraction angiography (DSA) should be used for CO2 imaging.  Veins should be injected more gently than arteries.

In an animal model, (Cho 2007) concluded that a single CO2 dose up to 1.6 mL/kg resulted in no changes in cardiopulmonary parameters. This amounts to 112 mL for a 70-Kg person, which is more than necessary for any clinical scenario.

Time Between Injections

CO2 tends to dissolve within a vessel in 30 seconds to 60 seconds.  If being cautious, injections should be performed at least 2 minutes apart.  For mesenteric and advanced disease peripheral artery imaging, at least 3 minutes should be considered or longer if there are symptoms.  Continued visualization of CO2 should be suspected to indicate a trapped bubble and/or room air contamination. Delayed absorption may also occur in persons with COPD (who have a high baseline CO2 level).

Maintaining Image Quality

The following techniques can optimize CO2 digital subtraction angiography:

  • Reduce respiratory motion and peristalsis artifact.
    • Intravenous glucagon (0.5 to 1 mg) can decrease peristalsis. 
    • Obtain one or more additional mask images for mask imaging subtraction to eradicate motion artifact.
  • Employ the imaging software to stack multiple images of the same location to create a composite picture of the anatomy that is more accurate than an image from a single run. 
  • Select the artery of interest with the catheter instead of imaging with the catheter tip outside.
  • Obtain rapid exposures (4 frames per second or more).
  • Elevate the target artery 15 to 20 degrees above the level of the table.
  • Improve filling of small arteries with intraarterial nitroglycerin (100-150 micrograms) therapy.


The most feared complication for intravascular use is air embolism, which can result in stroke, myocardial infarction, paralysis, amputation, or death, although this risk across all patients is less than 1%.  A large amount of CO2 trapped in the pulmonary artery or right side of the heart (only of concern during venography) obstructs venous return resulting in bradycardia and hypotension. The patient should be rotated into a left lateral decubitus position if this happens to attempt to separate the CO2 into a gas layer floating "on top of" and no longer interfering with the flow of the liquid and solid components of blood.  Large gas bubbles full of CO2 can remain "trapped" in the heart and/or a relatively reduced distribution of right pulmonary arterial tree until such time as their component molecules entirely dissolve into the bloodstream.

In an animal model,[4] CO2 was injected into the coronary arteries without adverse outcome, and laboratory physiology experimentation suggests CO2 can be delivered via catheter to the coronary arteries without reflux into the cerebral circulation (Corazza 2018), but there have been no case reports of use in the coronary system in humans.

Some people experience side effects of paresthesia, tenesmus, or nausea. Normally, nausea is only encountered when high flow rates are used for angiography.  There have been no reports of CO2 poisoning that presents as hypotension and hypoventilation.[5]

Abdominal pain during mesenteric arteriography usually can be handled by rotating the patient from side to side and massaging the abdomen.  However, it may signal the presence of a vapor lock. This phenomenon is when gas, which may also include endogenous nitrogen and oxygen, becomes trapped intraarterially due to having a diffusion constant the prevents its dissolving in blood while simultaneously having a high enough partial pressure relative to blood that no blood can be pumped through the gas into the capillaries. The result if not treated is mesenteric infarction. This is reported most commonly in the scenario of a large amount of CO2 collecting in an abdominal aortic aneurysm sac and then migrating into a mesenteric branch. First-line treatment involves attempting to dislodge the gas bubble mechanically via massage, patient rotation, and/or catheter aspiration.

Additional management strategies for CO2 adverse events are discussed elsewhere.[6]

Clinical Significance

CO2 floats on the gravity-nondependent surface of blood that it does not displace entirely; abnormalities of the dependent portions of vessels may be missed. Air-fluid levels of significance in practice only occur in the aorta, the IVC, and their first order branches.

Arteries that assume a posterior course, such as lumbar and some renal arteries, may require positioning the patient more decubitus to fill.

CO2 can cause an underestimation or overstimulation of vessel caliber compared with liquid contrast or IVUS[7] and have greater inter-observer variability in determining vessel caliber.[8]  CO2 has a lower accuracy for characterizing stenoses than liquid contrast. For example, in 27 lower extremities of adult men, CO2 opacified 86.2% (188/195) of arteries of concern and depicted stenosis adequately in 84.5% (191/226) of arterial segments. Infrapopliteal arteries were even less adequately visualized.[9]

As CO2 passes through vascular bifurcations, the bolus dissipates and can simulate a stenosis. If there is a physiologic shunt, then CO2 injection can mimic a fistula in the absence of an anatomic fistula needing mechanical correction.

Given the somewhat inferior appearance of CO2 to liquid contrast, patients should consent that a small volume of iso-osmolar contrast (such as 10 to 20 mL) may still be used for increasing the sensitivity of detecting stenoses or confirming equivocal findings.


CO2's lower viscosity compared with liquid contrast can allow it to more easily escape a vessel and more rapidly disperse along a slow flow system (Hawkins 1997). This can make CO2 more sensitive for detecting:

  • Arteriovenous fistula
  • Slow hemorrhage
  • Slow flow in a vessel (e.g., a bypass graft)
  • Slow flow endoleaks
  • Tumor vascularity. 

Multiple reports described the detection of gastrointestinal tract bleeding [6] and fistula when liquid contrast did not detect the finding.

Enhancing Healthcare Team Outcomes

CO2 angiography is used by a minority of angiographers and at a minority of institutions, even though it can offer advantages in diagnostic accuracy and patient outcomes in some settings. Improving health care practitioner team understanding of how to promptly evaluate and treat patients using digital subtraction carbon dioxide angiography will lead to better patient outcomes. [Level V]


[1] Hawkins IF, Carbon dioxide digital subtraction arteriography. AJR. American journal of roentgenology. 1982 Jul     [PubMed PMID: 6807073]
[2] Culp WC,Mladinich CR,Hawkins IF Jr, Comparison of hepatic damage from direct injections of iodinated contrast agents and carbon dioxide. Journal of vascular and interventional radiology : JVIR. 1999 Oct     [PubMed PMID: 10527206]
[3] Funaki B, Carbon dioxide angiography. Seminars in interventional radiology. 2008 Mar     [PubMed PMID: 21326495]
[4] Burko H,Klatte EC, Renewed interest in gases for contrast roentgenography. The American journal of roentgenology, radium therapy, and nuclear medicine. 1967 Mar     [PubMed PMID: 4289858]
[5] Ikeda N,Takahashi H,Umetsu K,Suzuki T, The course of respiration and circulation in death by carbon dioxide poisoning. Forensic science international. 1989 Apr-May     [PubMed PMID: 2504656]
[6] Sharafuddin MJ,Marjan AE, Current status of carbon dioxide angiography. Journal of vascular surgery. 2017 Aug     [PubMed PMID: 28735955]
[7] McLennan G,Moresco KP,Patel NH,Trobridge A,Dreesen J,Tennery J,Seshadri R,Johnson CS, Accuracy of CO(2) angiography in vessel diameter assessment: a comparative study of CO(2) versus iodinated contrast material in a porcine model. Journal of vascular and interventional radiology : JVIR. 2001 Aug     [PubMed PMID: 11487680]
[8] 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 iodinated contrast material in an aortoiliac flow model. Journal of vascular and interventional radiology : JVIR. 2000 Apr     [PubMed PMID: 10787201]
[9] Madhusudhan KS,Sharma S,Srivastava DN,Thulkar S,Mehta SN,Prasad G,Seenu V,Agarwal S, Comparison of intra-arterial digital subtraction angiography using carbon dioxide by 'home made' delivery system and conventional iodinated contrast media in the evaluation of peripheral arterial occlusive disease of the lower limbs. Journal of medical imaging and radiation oncology. 2009 Feb     [PubMed PMID: 19453527]