Adverse Cerebrovascular Effects of Intraarterial CO 2 Injections: Development of an In Vitro/In Vivo Model for Assessment of Gas-based Toxicity David B. Kozlov, MD, Elvira V. Lang, MD, William Barnhart, RTR, Albert Gossler, PhD, and Umberto De Girolami, MD PURPOSE: To assess whether and how CO 2 can cause ischemic injury in the central nervous system after internal carotid artery injection. MATERIALS AND METHODS: In 14 adult pigs, both internal carotid arteries were catheterized via a transfemoral approach. One carotid artery served as control and the other was injected via a prototype gas injector with defined volumes and pressures of gas. Effects were assessed by clinical observation, repeated magnetic resonance (MR) imaging, histopathology, and vital staining. An in vitro flow circuit was used to model injection parameters. RESULTS: Single injections of CO 2 did not produce persistent clinical symptomatology. In vitro conditions were created in which bubbles adhered to the tubing of the circuit, creating functional stenoses, or coalesced into larger bubbles that became trapped, thereby reducing flow and augmenting potential embologenic effects of subsequent injections. With in vitro–derived dual injection parameters, seven pigs underwent two sequential injections of CO 2 . All did well after the first injections, but all had adverse effects after the second injections, including involuntary tonic-clonic muscular movements, cardiopulmonary arrest, recurrent intractable seizure activity during recovery, hemorrhagic venous infarcts on gross and histopathologic examination, and blood–brain barrier breakdown on vital staining. MR imaging was not sensitive even after symptomatic intraarterial air injection. CONCLUSIONS: Absence of adverse effects after single bolus injections in pigs does not prove the safety of intracranial CO 2 injections in human patients. Considering the possible deleterious effects of repeat intravascular injectionsinthehighlysensitivesystemofthebrain,itmaybeprudentforclinicalapplicationatotherapprovedsites to let time pass between boluses sufficient to permit absorption of wall-adherent and coalescent bubbles that could cause gas embolic events. J Vasc Interv Radiol 2005; 16:713–726 Abbreviations: CNS = central nervous system, DWI = diffusion-weighted imaging, PSP = peak systolic pressure CO 2 is nonallergenic and considered less nephrotoxic than conventional io- dinatedcontrastmediaforthediagno- sis of vascular disease under angiog- raphy(1–6),buttheriskofcentralner- voussystem(CNS)toxicityofCO 2 and the conditions under which it occurs are poorly defined (7–11). Although wide clinical application of CO 2 in subdiaphragmatic vascular beds is consideredsafeincurrentpractice,su- pradiaphragmatic intraarterial use raises concerns (8,10–15). Anecdotal human reports of complications after CO 2 angiographyhavebeenattributed to persistent macrobubbles resulting in “vapor lock” (eg, gas embolic phe- nomena) and subsequent end-organ ischemia (12,14–17). Cerebral arterial CO 2 injections have yielded widely disparate results in animal studies (7– 11).Proposedmechanismsfortoxicity includelocalacid–baseimbalance,di- rect vessel trauma or excessive shear stress, chemical breakdown of the blood–brain barrier or alteration in vascular permeability, gas embolic– mediated ischemia, remote and local reflexes, activation of granulocyte ad- herence, and complement activation (7–11,13,16–38). The possibility of gas embolic–mediatedischemiasecondary to a persistent gas bubble poses the greatest concern because of its imme- diate vasoocclusive effect. We there- fore focused our research on whether persistentbubbleformationcouldbea major source of CNS toxicity after in- traarterial injection of CO 2 and whether the formation of a persistent bubble could be expected to occur only under specific physical condi- tions, thereby explaining the differing From the Department of Radiology (D.B.K., E.V.L.), Beth Israel Deaconess Medical Center, and the De- partment of Pathology (U.D.G.), Harvard Medical School, 330 Brookline Avenue, Boston, Massachu- setts 02215; and Departments of Radiology (W.B.) and Hydraulics (A.G.), University of Iowa Hospital andClinics,IowaCity,Iowa.ReceivedMay27,2004; revisionrequestedAugust19;finalrevisionreceived November 19; accepted November 20. From the 2004SIRAnnualMeeting. Address correspondence to E.V.L.; E-mail: elang@bidmc.harvard. edu This work was supported by a grant from the SIR Foundation. None of the authors have identified a conflict of interest. ©SIR,2005 DOI: 10.1097/01.RVI.0000153114.05700.61 713