Intracarotid Delivery of Drugs: The Potential and the Pitfalls

Some key anatomical features of the cerebral circulation affect intracarotid drug delivery to the brain. These include the size of the brain relative to the body weight, anatomical configuration of cerebral arteries, the state of the BBB, and the anatomical compartments within the brain.

In general, relative to body size, primates have a much larger brain weight than nonprimate animals, and even among primates, humans have a much greater brain-to-body weight ratio.18As a result, primates can tolerate greater intracarotid doses of drugs based on body weight. For example, an intracarotid dose of 8 mg/kg of an antineoplastic drug, carmustine, produces severe hemorrhaging necrotizing lesions in dogs, but the same dose is well tolerated in rhesus monkeys and in humans.19–21 

Anatomical variations of the cerebral arteries can influence regional distribution and concentration of drugs after intracarotid delivery. The configuration of cerebral arteries varies (1) across animal species,22(2) within individuals of the same species, and (3) even in an individual over time.

The extent of communication between the external carotid artery and the internal carotid artery (ICA) greatly varies across animal species. Rabbits and primates show a clear separation of the intracranial and extracranial irrigations.23–25On the other hand, many animals, such as goats and pigs, have extensive collateral communications between the intracranial and extracranial arterial irrigations.22Even within animals of the same species, there can be significant differences in the origin and the size of the ICAs.24Smaller animals have larger external carotid arteries, and the sizes of the ICAs, relative to the vertebral arteries, differ across species. Therefore, ICA occlusion is fairly well tolerated in rodents and causes a transient decrease in blood flow in rabbits.26However, ICA occlusion often causes neurologic symptoms in humans.27,28The ability of human subjects to tolerate ICA occlusion depends largely on the configuration of the circle of Willis. Only 18% of human subjects have a balanced, symmetrical circle of Willis (fig. 1); therefore, there is a need to clinically test whether a patient can tolerate ICA occlusion before ICA sacrifice.29–31The configuration of the circle of Willis also determines the resistance of each arterial segment that could influence the distribution of intracarotid drugs.32 

Furthermore, the blood flow in a given artery can change with time, which could affect the regional tissue concentrations of intracarotid drugs. It has been shown that the pharmacologic effects of intracarotid drugs, such as anesthetics, are directly related to cerebral blood flow (CBF).33,34If changes in blood flow have a significant effect on tissue drug concentrations, the arterial concentrations generated after injection of the same dose of a drug might be different over time even in the same individual. Consider, for example, superselective intracarotid infusion of papaverine for cerebral vasospasm. Initially, when blood flow is low, the concentration of the drug is high because of minimal dilution with arterial blood. As vasospasm resolves, blood flow increases and arterial concentration declines because of greater dilution.35In theory, the vasodilatory effects of intracarotid drugs that augment local blood flow may paradoxically decrease regional deposition of the drugs.33 

The BBB is primarily constituted by tight junctions between the endothelial cells that severely restrict migration of molecules from the blood into the brain tissue. In other capillary beds, there are clefts between the endothelial cells that permit passage of molecules. However, in the brain, the tight endothelial junctions permit only transcellular migration of molecules by diffusion, by carrier transport, or rarely by pinocytosis.36,37Abnormalities of endothelial tight junctions are seen with brain tumors and could be responsible for tumor-related brain edema.38Morphologically, endothelial cells of brain arteries, compared with other regions of the body, are rich in mitochondria. These mitochondria are also larger in size. This suggests that endothelial cells are metabolically very active.39–41Endothelial cells also are rich in metabolic enzymes such as catecholamine methyltransferase, monoamine oxidase, and adenosine deaminase,42which create a chemical barrier to drugs, such as catecholamines and adenosine.43In general, small nonpolar lipid soluble drugs can easily penetrate the BBB by passive diffusion. In contrast, the effective delivery of macromolecules that include antineoplastic drugs and nucleotides is possible only when the BBB is disrupted.44 

The brain is best described in terms of three compartments: blood, cerebral spinal fluid, and brain parenchyma. The brain tissue itself can be further divided into gray and white matter. Additional compartments might be created within the brain parenchyma that have their own kinetic characteristics in pathologic states (fig. 2).10Consequently, under physiologic conditions, there are three interphases between the compartments in the brain of which the BBB is the most significant.44The BBB has been regarded as the gatekeeper to the brain. Compared with the blood–cerebrospinal fluid interphase, the BBB has 5,000-fold greater area and has much lower permeability.45 

Figure 3shows the simple kinetic framework described by Dedrick10demonstrating the advantage of intracarotid injection. If C¹and C²are the regional and systemic drug concentrations, the pharmacokinetic advantage of intracarotid (ic) over intravenous (iv) infusion can be defined as

In figure 3, I is the rate of drug infusion, Q is the regional blood flow, E is the first-pass extraction of the drug from the region, and CL¹and CL²are the clearances from the region and from the rest of the body. CL^TBis clearance from the whole body (CL¹+ CL²). R can also be derived as

If CL¹= 0, then

Conversely, if CL²= 0, then

Equation 1and its derivatives 1a and 1b show that the maximum advantage of intracarotid drug infusion is evident (1) with drugs that have high CL^TB, (2) when the regional arterial flow (Q) is low, or (3) when the regional extraction is high. For example, we can illustrate the advantage of intracarotid delivery of propofol using equation 1a. Assuming an ICA flow to be 200 ml/min and CL^TBof propofol to be 2,100 ml/min for a 7-kg individual, R propofol will be 1 + 10.5 = 11.5. This theoretical number is fairly close to the observed experimental results showing a 10-fold reduction in dose of propofol required to produce electrocerebral silence with intracarotid delivery compared with intravenous infusion.46Conversely, the possibility of achieving high intracarotid concentrations also suggests that intracarotid infusion might result in locally toxic drug concentrations without any evidence of systemic toxicity.

There are several assumptions in the kinetic model proposed by Dedrick.10These include uniform mixing of drug in the blood, determining free drug concentration based on conventional steady state drug–protein interactions, constant clearances over time, constant regional blood flow, homogenous distribution within the arterial irrigation, and homogenous behavior of the brain compartment. Two of these assumptions merit close scrutiny:

According to the “free-drug hypothesis,” it is the free drug in the plasma that is available for diffusion across the BBB. However, in vivo  during intracarotid injections, the tissue brain concentrations are higher than those predicted by the free-drug hypothesis. Jones et al.  47compared the uptake of different benzodiazepines after intracarotid bolus injection in rats. They found that the brain uptake of this class of drugs was largely determined by their lipophilicities. However, in vitro  modeling grossly underestimated the in vivo  uptake of these drugs. They determined that observed in vivo  data deviated from the more theoretical model with higher doses of albumin that were injected along with the drug. The free drug concentration seemed to be 5- to 25-fold greater in vivo  than predicted by in vitro  observations.47Several possible explanations have been offered for this observation. First is the “free-intermediate hypothesis,” which states that rapid uptake of a drug into the brain enhances the release of drug bound to plasma proteins.48,49The second possibility is that other proteins in the blood have higher drug affinity, such that in vitro  estimates of free drug concentrations based on albumin binding alone underestimate the blood concentrations.49,50The third possibility is that there might be some drug–protein binding competitor in the capillaries that increases the drug–protein dissociation rate.51Fourth, when drugs bind to proteins, they are absorbed by a peptide transporter system, which may increase transfer across the BBB.52 

It is best to consider the brain parenchyma in terms of separate compartments with distinct units as described previously: (1) gray matter, (2) white matter, and (3) pathologic compartments, such as tumor or ischemic regions, within the brain parenchyma.53,54These compartments will differ with respect to their blood flow, BBB functions, and ability to extract and metabolize drugs. For example, the blood flow to the gray matter (50 ml · 100 g⁻¹· min⁻¹) is approximately twofold greater than in the white matter (25 ml · 100 g⁻¹· min⁻¹) and fourfold greater than in the ischemic regions (12 ml · 100 g⁻¹· min⁻¹).55Baseline variability in regional blood flow could result in differences in drug concentrations in the brain tissue compartments after intraarterial injection.10Eventually, effective modeling of intracarotid drug delivery will require an understanding of the micropharmacokinetic factors that determine drug distribution within the brain tissue (fig. 4).56 

Intracarotid infusion of drugs offers the following theoretical advantages:

Intracarotid drug infusion can restrict the initial volume of distribution to one cerebral hemisphere. The volume of blood flow through the ICA in humans has been estimated to be approximately 200 ml/min, whereas the cardiac output is 5–6 l/min. Consequently, drugs that are infused into the small volume of ICA flow achieve relatively high initial arterial concentrations at low total doses, thereby decreasing systemic toxicity.46,57,58 

Although the exact nature of the interactions between the drug and binding proteins after intracarotid infusions remains to be fully understood, limited data suggest that intracarotid administration of drugs results in disproportionately higher free drug concentrations than those predicted by steady state kinetic models.47Recent videomicroscopic images have revealed that bolus injections of drugs transiently overwhelm the blood flow and may deliver virtually undiluted drug to the brain.59Such injections, in theory, would considerably attenuate the decrease in free drug concentration due to protein binding or the uptake by blood cells. Manipulating the key parameters of bolus injection, volume, concentration, and frequency can have a significant effect on the tissue concentrations of intracarotid drugs.59Bolus injections, when combined with regional blood flow manipulation, could therefore significantly enhance intracarotid drug delivery.34 

Intracarotid delivery can achieve rapid, virtually instantaneous, high drug concentrations in the brain. Such an ability to instantaneously generate high tissue concentrations may be critical in some clinical situations, such as preventing reperfusion injury after intracarotid thrombolysis.60–62 

The theoretical disadvantages of intracarotid infusions can be summarized as follows:

As the target organ, the brain, with a relatively high resting blood flow that amounts to 15–20% of the cardiac output, is at a disadvantage compared with some other less-well-perfused organs.63High blood flow decreases the peak drug concentrations due to greater dilution by the arterial blood. Increased blood flow to the brain could decrease the drug transit time through cerebral circulation if the cerebral blood volume was not proportionately increased. Transit time seems to have a direct effect on tissue concentrations of highly lipid-soluble anesthetic drugs after intracarotid injections.56In addition, high blood flow will augment drug efflux from the brain tissue.

For example, under physiologic conditions, the brain has at least three compartments: gray matter, white matter, and cerebrospinal fluid. These compartments demonstrate unique kinetic characteristics. Pathologic tissue, such as cerebral edema or brain tumor, may have significantly different kinetic properties. These factors may make it difficult to model drug kinetics in physiologic and pathologic states and has led to the failure of theoretical models of intracarotid drug delivery.64 

The advantages of intracarotid versus  intravenous drug delivery decrease with prolonged infusions due to lower peak regional concentrations and the eventual redistribution of the drug. Computer simulations suggest that intracarotid drug infusion over 2 h versus  the same dose infusion over 10 h results in regional concentrations that are 5.5- and 3.1-fold greater than intravenous delivery, respectively.10Experimental data also suggest that the comparative dose advantages of intracarotid versus  intravenous drug delivery decrease when infusions are used for prolonged periods of time. In rabbits, intracarotid propofol achieves transient electrocerebral silence at one tenth the intravenous dose. On the other hand, to maintain electrocerebral silence for 1 h, intracarotid delivery is only 5-fold as effective as the intravenous injection.46 

Measurements of drug concentrations are particularly challenging with intracarotid drug delivery to the brain. Postmortem samples do not provide time histories.65Multiple tissue biopsies, though feasible, tend to injure the preparation and are not site specific.66Insights into the field have been limited to a few magnetic resonance imaging positron emission tomography studies and a few radiolabeled drug–based studies.47,67–70The α half-life of many lipid-soluble drugs, such as carmustine and propofol, is exceedingly short, ranging between 1.4 and 7 min.68,71Microdialysis requires time to obtain sufficient sample for analysis, is invasive, and may alter tissue characteristics.72,73Therefore, the exceedingly rapid changes in tissue drug concentrations, particularly after bolus injections, are beyond the time resolution of microdialysis. Novel optical techniques, such as diffuse reflectance spectroscopy, can noninvasively measure tissue drug concentrations in virtual real time, which could provide better understanding of intracarotid drug kinetics.74,75 

From the practical standpoint, intracarotid drug delivery could be improved by the following means:

Controlled disruption of the BBB can be achieved by either hyperosmotic agents or chemical means.76,77 

Intracarotid injection of hypertonic substances such as mannitol and arabinose leads to a transient loss of BBB functions (fig. 5).78Hypertonic agents cause vasodilation and shrinkage of the endothelial cells, resulting in increased diffusivity and bulk flow across the BBB.77Cell shrinkage, along with the contraction of the endothelial cytoskeleton, results in widening of the tight junction to approximately 20 nm, which results in a 10-fold increase in permeability of some compounds. Alternately, some believe that mechanisms other than cell shrinkage, such as alterations in Na⁺–Ca²⁺exchange, might play a role during osmotic disruption of the BBB.79The effects of hypertonic mannitol were previously thought to be exceedingly transient, although data suggest that in humans, hypertonic BBB disruption lasts for 30 min and BBB functions remain impaired for several hours afterward.80Zunkeler et al.  showed that the healthy BBB was almost 15-fold more susceptible to osmotic disruption than the BBB in glioma tumors.81,82Biomechanical factors such as systemic hypotension seem to mitigate osmotic disruption of the BBB, whereas hypertension enhances disruption.83,84In experimental settings, the efficacy of hypertonic mannitol can also be enhanced either by cooling the solution to 4°C or with a Na⁺–Ca²⁺exchange blocking drug, KB-R7943.85,86There is evidence to suggest that treatment with steroids, anesthetics, and magnesium can mitigate BBB disruption by intracarotid mannitol.87–90Restoring BBB functions after intracarotid mannitol may help in limiting the complications of hypertonic mannitol therapy.88,91–93 

Chemical disruption of BBB by intracarotid injection of bradykinins such as Cereport (RMP-7) or leukotrienes (LTC-4) results in transient opening of the BBB.76Contrary to the disruption by mannitol, disruption of the BBB by Cereport and leukotrienes seems to be limited to pathologic lesions and not normal brain. The effects of the drug are reversible within 30–60 min. Cereport can increase the uptake of intraarterially delivered antineoplastic drugs.94,95 

One of the ways to minimize the side effects of intracarotid drugs is to restrict their delivery to pathologic lesions by superselective cannulation of the feeding arteries. With the development of highly flexible flow directed catheters with external diameters of 1.5–2 mm, it is possible to cannulate distal branches of the carotid artery without the risk of obstructing the blood flow (fig. 6).96The problem with distal cannulation is the increased risk of streaming (see Streaming of Drugs section).97However, side port catheters and diastolic pulse delivery can minimize the chances of streaming.98The theoretical advantages of superselective delivery of drugs include minimizing the total dose, decreasing the risk of regional toxicity, and limiting potential vascular complications to the pathologic regions of the brain. In case of intracarotid vasodilators, superselective infusions will decrease the chances of cerebral steal or the likelihood of an increase in intracranial pressure (ICP). A key concept recently introduced by Gobin et al.  14was to determine the dose of intracarotid drugs based on a spatial fractionation algorithm. The algorithm calculates the intracarotid dose requirements during superselective drug injections based on the volume of tissue perfused by a given intracranial artery. Such an approach decreases the risks of neurologic complications during intracarotid chemotherapy.

Intracarotid infusions can be further targeted to the brain by enhancing their removal from the systemic circulation, by extracorporeal removal, by forced alkaline diuresis, or by neutralizing the effects of a drug by the coadministration of a systemic antidote.3,99The use of drugs, such as adenosine, with exceedingly short biologic half-lives greatly minimizes the risks of systemic side effects and could potentially lead to “fire-and-forget” types of intraarterial drug interventions.58 

Early computer simulations showed that intracarotid drug delivery to the brain is particularly useful in low-blood-flow states.10In a series of experiments, CBF was altered by changing ventilation, by augmenting blood flow with intracarotid verapamil pretreatment, or by producing transient flow arrest. The dose of an intracarotid anesthetic, propofol, required to produce electroencephalographic silence, was directly related to the blood flow.33Both hypercapnia and verapamil pretreatment, which increased CBF, increased the dose requirements of intracarotid anesthetics, whereas flow arrest significantly increased the duration of electroencephalographic silence after bolus anesthetic injection (figs. 7 and 8).33,34 

There are several significant practical concerns with regard to intracarotid drug delivery to the brain:

Recent review suggests that the incidence of neurologic complications after endovascular procedures varies a great deal, from 5.6% during the placement of a carotid stent to 19% during occlusion of a parent artery in the treatment of cerebral aneurysms.100A significant factor contributing to this morbidity is cerebral embolism due to thrombi, dislodged atheromatous plaques, or air emboli. A number of strategies are being developed to interrupt the thromboembolic pathway. Soft, flow-guided catheters can decrease the risk of endothelial injury. Adhesion, activation and aggregation of platelets can be inhibited with agents such as aspirin, ticlopidine, and clopidogrel. The coagulation cascade can be blocked by inhibiting the activation of fibrinogen by thrombin. The antithrombin effects (as well as the potential adverse effects) of heparin are well known, can easily be monitored by measuring the activated clotting time, and can be easily reversed with protamine. Therefore, heparin provides convenient thromboembolic prophylaxis in clinical settings. Although heparin has been the mainstay of thromboembolic prophylaxis during endovascular surgery, newer strategies, such as hirudin analog or platelet receptor antibodies, may find wider application in the future.101Endovascular interventions in the background of significant anticoagulation carry a greater risk of hemorrhagic stroke as a complication. Catastrophic bleeds during the procedure necessitate immediate and aggressive endovascular and surgical interventions. Rapid reversal of the anticoagulant effects of heparin with protamine is particularly useful in such situations. A well-thought-out management strategy should be in place before anticoagulation.102 

The risk of thromboembolism is also directly related to the duration of endovascular catheterization. Prolonged catheterization of the cerebral arteries can permit repeated cerebral angiography to monitor response or facilitate delivery of drugs. Retrograde cannulation of the superficial temporal arteries was used to permit cerebral angiography over a week without any complication in patients with intracranial aneurysms.103Kallmes et al.  104have used a swine model to demonstrate that catheters with hydrophilic surfaces were less thrombogenic than those with hydrophobic surfaces. The risk of thromboembolism also depended on the catheter material not just surface coating. In this model, long-term implantation of a microcatheter was well tolerated for the longest duration of the study, 35 days.

One of the less-recognized risks of intracarotid drug therapy is the accidental injection of air, sometimes dissolved in fluids. Studies with transcranial Doppler flow measurements have revealed a frequent occurrence of microscopic air emboli during angiography. Microscopic air entrainment is likely to occur at two times during the angiographic procedure: during the aspiration of contrast into the syringe and during the injection of contrast. Increased viscosity of contrast increases the risk of air entrainment. The chances of air emboli being injected are decreased if the syringes are allowed to stand. The risk of air embolism is directly related to the rate of injection and inversely related to its viscosity. Therefore, decreasing the rate of injection decreases the chances of air embolism.105 

Streaming of drugs  refers to the uneven distribution of drugs in the arterial stream that results in different drug concentrations within that arterial distribution.106Streaming has been observed in vivo  and in vitro  and has been invoked to explain focal drug toxicity of chemotherapeutic agents within arterial distribution (figs. 9 and 10). Using a polystyrene cast of human cerebral arteries, it has been shown that streaming depends on the rate of drug infusion, the type of catheter used for drug delivery, and the position of the catheter in relation to the arterial branching. This in vitro  model suggests that there can be an almost 5-fold difference in drug concentrations due to streaming.107In a human study using O¹⁵positron emission tomography, Saris et al.  98demonstrated that there can be an 11-fold variation in regional drug concentration within the arterial irrigation of the supraophthalmic segment of the ICA.

The delivery of drugs directly into the brain has been associated with regional toxicity. There is an increased incidence of white matter lesions with intracarotid chemotherapy compared with intravenous therapy.108,109Streaming is also thought to be responsible for retinal injury after intracarotid infusion of antineoplastic drugs.110Therefore, in defining toxicity, one has to factor in the total dose of the intraarterial drug, as well as the highest possible concentration that might result from inadequate mixing. In addition, possibility of injury to blood vessels, such as to the vascular endothelium, should also be considered.111,112 

One of the major problems in intracarotid drug therapy is the standardization of doses within individuals of the same species or during the extrapolation of data between animals of different species. Doses have been described in terms of weight/unit body surface area,113total milligram dose, or infusion rates in mg/min, and dose/kg body weight. Investigators have also assessed the response to intracarotid drugs as a function of estimated molar concentrations.114–116In the absence of any consistent way to present dose data, it is sometimes difficult to compare drug response in different studies. We recommend that to optimally interpret intracarotid dose–response data, one has to consider all three infusion parameters: (1) The rate of drug infusion (mg/min, or mol/min), which, along with the blood flow, determines peak concentrations. (2) Estimated concentration in arterial blood, i.e. , dose of drug/volume of blood flow in the infused artery. With superselective infusions, drug concentrations may be estimated by the volume of tissue infused based on angiographic measurements.117(3) The total dose when it is sufficient to cause systemic toxicity.

The exact description of intracarotid doses is particularly important in extrapolating the doses from one animal species to another. Because of the differences in relative sizes of various organs, scaling of intracarotid doses merely on the basis of body weight or body surface area can lead to errors. For example, within individuals of a given species, parameters such as blood flow in the ICA can be assumed to be relatively constant, but across animal species, those assumptions may not be valid. Therefore, the same amount of drug will generate different arterial blood concentrations. Because of the relatively large size of the brain and higher carotid arterial blood flow, primates tolerate a much higher dose of intracarotid drugs compared with rodents or dogs. To undertake intraspecies comparisons, Dedrick118described kinetic models based on organ size and surface areas. These relative sizes of organs differed between species but their function and surface characteristics were similar. Even when such allometric corrections are made, it is still difficult to project intracarotid kinetic data from one animal species to another.

More than 50 yr ago, Wada developed the technique of injecting sodium amytal into the carotid artery.1The procedure was originally developed to permit unilateral electroconvulsive therapy but subsequently became a standard method for localizing language function and memory. Other anesthetic drugs that have been used for Wada testing include methohexitone and, recently, propofol.1,119–122Compared with amytal, intracarotid methohexitone seems to have a shorter duration of effect, thereby making it possible to test multiple arterial territories in the same setting.122However, there seems to be poor justification for the selection of intracarotid doses.71Most radiologists use a graded dose of 75–125 mg amytal mixed with contrast media. After an initial dose of 75 mg, additional boluses of 25 mg are injected until there is an upper limb drift. Speech, language, and memory functions are assessed as the deficit resolves. Intracarotid amytal in doses as high as 3 mg/kg or 200 mg has been used during the Wada test. Such high intracarotid doses of anesthetic drugs on recirculation suppress the contralateral hemisphere. Any additional baseline sedation provided by an anesthesiologist in attendance would further complicate the interpretation of the Wada test. Therefore, it has been recommended that when high doses of amytal are used, the contralateral hemisphere should be tested on a subsequent day.123 

In recent years, there has been considerable interest in the use of propofol for Wada testing.120In a recent study, however, as many as 19 of 58 patients developed transient neurologic symptoms. These consisted of tonic movement, confusion, and pain. The symptoms were seen when more than 10 mg of the drug was rapidly injected in patients older than 55 yr. None of these patients had any permanent complication, but the authors recommend limiting the dose of propofol to 10 mg and careful monitoring in patients older than 55 yr.124Despite these recent reports with propofol, the safety of the Wada test is underscored by the fact that the test has been widely used for a very long time. Although neurologic complications have been reported due to arterial emboli, they do not seem to be frequent.125 

Superselective Wada testing is also used before embolization of cerebral arteriovenous malformations (fig. 11). Because embolization of the lesion could injure both gray and white matter, injection of amytal is combined with injection of lidocaine to test for both the gray and the white matter. Amytal is injected first to suppress gray matter activity, followed by the injection of lidocaine to suppress nerve conduction through the white matter. Anesthesiologists involved in the care of patients undergoing Wada tests should use judicious amounts of sedation. Both propofol and dexmedetomidine have been used for the purpose. Midazolam may impair memory testing and therefore has to be used cautiously.

Cerebral vasospasm  is a pathologic narrowing of the cerebral arteries. The fundamental method to overcome this added arterial resistance, whether proximal or distal, is to increase the cerebral perfusion pressure and the cardiac output by using hypervolemic hemodilution with induced hypertension (triple-H treatment).126A significant number of patients with cerebral vasospasm do not respond to triple-H treatment. Furthermore, triple-H treatment has to be carefully applied to patients with untreated or multiple aneurysms. The alternate strategy in such cases is to decrease cerebrovascular resistance by superselective intracarotid or intracarotid infusion of vasodilators.58,117,127,128Intracarotid papaverine has been the mainstay of such a treatment, although a variety of drugs such as calcium channel blockers, mannitol, and prostaglandin E have also been used intraarterially for treating vasospasm (fig. 12). Other drugs, such as adenosine, have been proposed for the treatment of vasospasm and, despite their significant vasodilator effects, have not yet been used in clinical settings, probably because of their short duration of effect and poor penetration of the BBB (fig. 13).

Table 1shows the published outcomes of intracarotid vasodilator therapy. Clinically, approximately 30–60% patients seem to benefit from intracarotid vasodilators. Angiographic improvement, however, is seen much more frequently. Although there is a suggestion that intracarotid vasodilators may augment CBF, improve cerebral oxygenation, and reverse metabolic acidosis, the effect of vasodilator therapy seems to be transient and resolves completely over the next 24 h. By measuring the time it takes for radiocontrast to transit through the cerebral arteries and the capillary bed, one can estimate the differential effects of intracarotid vasodilators on the proximal and distal cerebral circulation. In the case of papaverine, the resistance of both the proximal and the distal cerebral arteries decreases during drug infusion.35,128 

Intracarotid vasodilator therapy carries with it two general risks: cerebral steal and increase in ICP. However, neurologic complications can also result from specific drugs, e.g. , neurotoxicity with papaverine.129 

Cerebral steal has been invoked to explain the decrease in regional blood flow in proximity to focal cerebral pathology after the administration of intravenous or intracarotid papaverine.130,131Any decrease in hemispheric cerebrovascular resistance, whether pathologic or iatrogenic, can cause a redistribution of blood flow away from those ischemic areas where there is already maximal vasodilation. In recent years, the clinical significance of steal has been challenged as with regard to large arteries.132There are, however, examples of microcirculatory steal during hypercapnia or during administration of volatile anesthetic agents, such as halothane, in the setting of focal cerebral ischemia.133,134Unless vasodilator therapy is restricted to the arteries afflicted by cerebral vasospasm, intracarotid vasodilator therapy carries the theoretical risk of cerebral steal, due to vasodilation in the normal vascular beds.

Increase in ICP has been reported with intracarotid infusion of papaverine. Controversy surrounds the effect of intracarotid vasodilators on ICP. Experiments in dogs and monkeys that do not have cerebral vasospasm suggest that intracarotid vasodilators such as sodium nitroprusside increase ICP but may not increase CBF, implying that nitroprusside affects cerebral capacitance arterioles to increase cerebral blood volume with no effect on the resistance arterioles.135On the other hand, intracarotid infusion of adenosine has a relatively benign effect on ICP but augments CBF, suggesting that adenosine primarily affects the resistance arterioles.136It remains to be seen whether intracarotid therapy could selectively target proximal or distal cerebral arteries to minimize the risk of increased ICP.

Cross et al.  137measured changes in ICP during intracarotid infusion of papaverine in 28 patients. The increase in ICP ranged from 0 to 60 mmHg. A baseline ICP of 15 mm or greater was associated with a greater risk of increasing ICP. However, Hunt and Hess scores, Fisher grades, age, and Glasgow Coma Scale scores on admission and immediately before treatment did not correlate with ICP increases. ICP increases result in a decrease in cerebral perfusion pressure. Therefore, hypertensive-hypervolemic interventions need to be continued during intracarotid vasodilator infusions. ICP should be monitored during intracarotid vasodilator therapy whenever feasible.

Papaverine is perhaps the most potent cerebral vasodilator. Papaverine acts through several intracellular pathways, which include cyclic guanosine monophosphate, cyclic adenosine monophosphate, phosphodiesterase inhibition, Ca²⁺channel blockade, and histamine release. Table 1summarizes the efficacy of papaverine and other vasodilators in the settings of cerebral vasospasm. Intracarotid papaverine can relieve angiographic narrowing, augment CBF, and mitigate neurologic symptoms in patients with cerebral vasospasm resistant to medical treatment. Typically, during the treatment of cerebral vasospasm, a dose of 300 mg papaverine is infused over 1 h.138Unfortunately, intracarotid papaverine infusion also results in significant neurologic complications,139attributable to several factors: cerebral steal, increased ICP,137,140microembolization of papaverine crystals,141,142neurotoxicity of the preservative (chlorobutanol),142proconvulsive properties of the drug,143and paradoxical vasoconstriction.141 

Cerebral steal, i.e. , redistribution of blood flow away from the ischemic region, has often been implicated in the etiology of neurologic symptoms after intravenous or intracarotid administration of papaverine.130,144There is evidence that intravenous papaverine increases blood flow in ischemic regions of patients with cerebrovascular insufficiency.145However, experiments in cats with middle cerebral artery occlusion suggest that steal does not always occur with cerebral vasodilation. Intravenous papaverine in the feline middle cerebral artery occlusion model did not decrease regional CBF to ischemic areas as long as the blood pressure is maintained.146 

Intracarotid calcium channel blockers have been used to treat cerebral vasospasm for more than a decade; however, their intracarotid administration is still not considered to be the mainstay for the treatment of cerebral vasospasm. Intracarotid verapamil, nicardipine, and nimodipine have been used for the treatment of vasospasm (table 1). There are no human studies as yet that compare the relative potencies of calcium channel blockers against papaverine. However, studies in similar groups of patients who do not have cerebral vasospasm suggest that intracarotid verapamil, compared with papaverine, seems to be less efficacious in augmenting CBF. Verapamil in small doses (3 mg) results in angiographic improvement without systemic hypotension, although only 5 of 29 patients in this retrospective study improved clinically.147The modest efficacy of the calcium channel blocker might still be advantageous because it is likely to decrease the risk of cerebral steal or increasing ICP. Experimental vasospasm triggered by topical application of endothelin 1 seems to be more responsive to intracarotid nicardipine than verapamil.148Intracarotid nicardipine is effective in reversing cerebral vasospasm, but no study has as yet compared the potency of the various intracarotid calcium channel blockers.149 

Recently, intracarotid infusions of liposomes with papaverine and prostaglandin E2 have been used for the treatment of cerebral vasospasm. Two thirds of the patients demonstrated an increase in CBF after such a treatment.150 

To enhance the clinical efficacy of intracarotid vasodilators and decrease the risks of complications, a number of adjuvant strategies have been suggested. First, transluminal angioplasty has been used to treat cerebral vasospasm when it affects the ICA or the proximal segment of the middle cerebral artery. Angioplasty, compared with intracarotid drugs, results in a more profound and persistent increase in blood flow but it can only be used in the proximal arteries. Drugs such as papaverine are therefore useful in treating distal spasm and could be of benefit even after angioplasty.128Hypothermia has also been used to provide brain protection and increase the time window for intracarotid vasodilator therapy.151Techniques of selective brain cooling either by the infusion of cold saline or by endovascular cooling devices are currently being developed to minimize injury after vasospasm and ischemic stroke.152–154Finally, cocktails of vasodilators have been used so as to increase both the potency and duration of effects of intracarotid vasodilators. Therefore, papaverine has been used in conjunction with nimodipine, nicardipine, prostaglandins to enhance its safety and efficacy.150 

The key in the anesthetic management of patients for intracarotid vasodilator treatment is to recognize the potential for exacerbating neurologic injury. These patients may present either with untreated ruptured aneurysms or after clipping of the lesion. At many institutions, triple-H therapy is instituted before intracarotid vasodilator therapy. Patients with subarachnoid hemorrhage may require inotropic support because of impaired myocardial functions secondary to hemorrhage. Ideally, management of these cases requires monitoring of the ICP so as to treat any increase by mechanical drainage, hyperventilation, or supplemental intravenous anesthesia. Furthermore, because of recirculation of intracarotid vasodilators, there may be significant systemic hypotension. Therefore, anesthesiologists should be prepared to induce hypertension to ensure adequate perfusion pressure.

The major driving force for the development of intracarotid therapies in the 1980s was the treatment of brain neoplasms. However, it soon became evident that the BBB prevented effective transfer of chemotherapeutic agents.2Therefore, attempts to disrupt the BBB began fairly early on.77,94Currently, intracarotid chemotherapy is used for the treatment of primary central nervous system lymphoma, primitive neuroectodermal tumor, germ cell tumor, cancer metastasis to the brain, and low- or high-grade glioma (fig. 14).111The initial response to intracarotid chemotherapy is impressive, with as many as 70–81% tumors regressing, occasionally with a significant prolongation of survival.155,156However, most studies have been anecdotal or uncontrolled case series. The major concern of intracarotid therapy is unexplained local toxicity, likely attributed to unusually high concentrations of the drug due to streaming. One strategy that has minimized streaming is using pulsed intracarotid infusion, which delivers the drug in the diastolic phase of cardiac cycle to increase mixing.106,157 

The role of regional blood flow in enhancing intracarotid delivery of chemotherapeutic drugs is ill understood at this time. Some groups have suggested an increase in drug doses in proportion to the regional blood flow to safely achieve higher total doses and better tumor regressions.14,156At the same time, it has recently been shown that transient decrease in CBF could significantly increase delivery of anticancer drugs to the brain in a rabbit model.15Therefore, the exact role for the manipulation of CBF in enhancing cancer drug delivery remains to be fully understood at this time. The fundamental problem confronting investigators is the inability to rapidly measure tissue concentrations of chemotherapeutic drugs. Insights into intracarotid delivery of some anticancer drugs could be generated by optically tracking their concentrations in the brain tissue by novel spectroscopic techniques.65,74,75 

The disruption of the BBB with intracarotid mannitol that is often needed to improve regional delivery of anticancer drugs also carries significant clinical risks.158Because of pain associated with the procedure, the disruption is undertaken during general anesthesia. In a recent report involving 17 patients and 210 treatment cycles, focal seizures occurred in 9 patients and in 10% of the treatment cycles. In some cases, these seizures were generalized (2 patients and 3% of treatment cycles). Chances of seizure were higher with intracarotid than with vertebral artery injections. Impaired consciousness was seen in 7.6% of patients with recovery in 24 h. One patient in this series had an intractable increase in ICP necessitating decompressive hemispherectomy a day after treatment. Attempts to augment CBF to enhance drug delivery with intravenous atropine probably contributed to ST-segment changes seen in 4.3% of the cases, requiring β-blockade. Other complications included transient neurologic deficits (6%), postoperative nausea and vomiting despite 20 mg odansetron (11.9%), and headache (4%). Management of these patients requires a well-thought-out plan to include close monitoring of the patient’s neurologic and hemodynamic status and to address significant complications that arise in the postoperative periods.159Patients with brain tumors may require preoperative embolization with n-butyl-cyanoacrylate to decrease blood loss during surgery (fig. 15). These procedures are relatively painless and can usually be performed with minimal sedation.

The object of intracarotid thrombolysis is to deliver high concentrations of thrombolytic drugs locally into the brain.160Intravenous thrombolysis can achieve a significant degree of recanalization within 3 h of onset of neurologic symptoms.160,161However, it is often difficult to provide thrombolytic therapy within such a narrow time window. Intracarotid delivery of thrombolytic drugs can extend the intervention time window to 6 h.162Intracarotid urokinase, streptokinase, and more recently recombinant tissue plasminogen activator have been used for the purpose (table 2).163Development of endovascular clot retrieving devices, laser or ultrasonic clot lysis techniques, and placement of endovascular stents can supplement or provide alternatives to intracarotid thrombolytic therapy.164Intracarotid thrombolysis is superior to intravenous thrombolysis in so far as restoring tissue perfusion, although as yet there is no clear-cut evidence of improved neurologic outcome with intracarotid thrombolysis in controlled trials (fig. 16).165,166Intracarotid thrombolysis today provides the most compelling reason to investigate intracarotid drug delivery to the brain. The inability to translate the results of preclinical studies with pharmacologic treatments of ischemic stroke into effective therapies compels us to develop better methods of drug delivery.167–169Intracarotid thrombolysis provides a unique opportunity to deliver drugs at the very site of reperfusion injury, and the catheter to do so is already in situ .

Because the urgency for interventions, detailed anesthetic assessment may not be feasible in stroke setting. Centers offering intracarotid thrombolysis should have an intervention suite ready to provide immediate anesthesia and should have well-established lines of communication between the stroke team and the anesthesiologists. A rapid review of medical history, last meal time, drug treatments, and allergies, particularly immediate anticoagulant treatment, has to be undertaken. Airway interventions could be challenging with background anticoagulation because of the potential for hemorrhage from even minor trauma, such as placement of a nasal airway. Wherever feasible, these interventions should be performed with minimal sedation to permit neurologic examination. However, the condition of the patient could rapidly deteriorate because of the evolution of neurologic symptoms or because of complications of treatment, such as a hemorrhagic transformation of the infarct, necessitating urgent conversion from sedation to general anesthesia. Blood pressure manipulation, particularly induced hypertension, may be required to improve perfusion.29 

Clinicians have frequently used intracarotid drugs to treat life-threatening brain diseases, but the following applications of intracarotid drugs have not been widely recognized.

Yokota et al.  170used low-dose bilateral intracarotid infusion of mannitol to treat severely increased ICP in 18 human subjects with head trauma. They observed that intracarotid mannitol significantly decreased ICP and caused no hemodynamic side effects or disturbances in electrolyte status.

Medical treatment of purulent meningitis and encephalitis is sometimes complicated by the inability to deliver sufficient antibiotics to the infected site.171In experimental animals, intracarotid hyperosmolar disruption of BBB has been shown to enhance the delivery of tobramycin and vancomycin to the brain.172Even without hypertonic disruption, intracarotid antibiotics have yielded beneficial results in treating intracranial infections.171,173–175 

Perhaps the most potent application of intracarotid drugs in the future is to deliver genes that could alter the course of a disease or help us to understand disease processes. Intracarotid delivery of viral vectors, DNA-bearing liposomes, and stem cells have all been successfully demonstrated.176,177Selective delivery of herpes viral vectors to tumor regions by concurrent use of Cereport (RMP-7) is one such example.178Bone marrow stem cells have been successfully delivered to traumatized regions of the brain after intracarotid injection.179Although the field of gene therapy is in its infancy, intracarotid gene delivery could play a key role in the future treatment of brain diseases due to degeneration, ischemia, trauma, or neoplasia. Therefore, intracarotid drug delivery promises to play a critical role in emerging areas of molecular and restorative neurosurgery.180,181 

During the past decade, compared with the technical advances in endovascular surgery, the field of intracarotid drug therapies has remained relatively ignored. There are ample anecdotal data to suggest that intracarotid drug therapy is effective for cerebral vasospasm, thromboembolic strokes, and neoplasms. However, there have been few attempts to systematically understand the kinetics of intracarotid drugs. The major efforts to deliver drugs selectively to the brain in the past decade have relied on molecular techniques to selectively penetrate the BBB (fig. 17). These advances in brain tissue drug targeting by novel neuropharmaceuticals come at a time of rapid advances in endovascular techniques, material sciences, and brain imaging that compel us to reevaluate intracarotid drug delivery. Intracarotid drug delivery can instantaneously generate exceedingly high local concentrations of the novel neuropharmaceuticals and thereby assist their regional delivery. Intracarotid drug delivery might be the primary route of drug delivery, or it could be used in conjunction with other brain tissue targeting technologies. Therefore, it is time to better understand the kinetics of intracarotid drugs and to develop techniques that could enhance the safety and reliability of such infusions.

This review is dedicated to our parents, who, despite their hardships, always encouraged us to pursue our dreams. Acknowledgments are due to Mei Wang, M.P.H. (Staff Researcher, Department of Anesthesiology, Columbia University, New York, New York), for her help in preparing this manuscript and the ongoing intracarotid research, and Edward A. Neuwelt, M.D. (Professor of Neurology and Neurosurgery, Oregon Health and Science University, Portland, Oregon), for his help with illustrating this review (fig. 14).