This publication was supported by an educational grant from Amersham Health, Princeton, NJ. The opinions expressed in this publication are those of the authors and not necessarily those of Amersham Health.

Dr. Bae reports relationships with Tyco Healthcare and Mallinckrodt through patent agreements and as a consultant. Dr. Fishman reports relationships with Siemens Medical Solutions and Amersham Health as a consultant. Dr. Foley reports a relationship with GE Medical Systems through an investigator agreement. Dr. Naidich reports a relationship with Siemens Medical Solutions through its Advisory Board and as a consultant. Dr. Saini reports a relationship with GE Medical Systems through research support. Dr. Becker, Dr. Sahani, Dr. Siegel, Dr. Tahktani, and Dr. Zinreich report that no such relationships exist.

Dr. Zinreich is a Professor of Radiology in the Division of Neuroradiology, and Dr. Takhtani is an Assistant Professor of Radiology in the Department of Radiology, Johns Hopkins Medical Institutions, Baltimore, MD.

The most important neuroimaging application of computed tomographic angiography (CTA) is in the evaluation of patients with stroke. Each year, an estimated 730,000 to 760,000 people suffer a stroke. It is the third leading cause of death in the United States. Approximately one-third of stroke patients are left permanently disabled, making it the leading cause of disability as well. ^1,2

It is imperative that stroke be diagnosed in the acute phase, preferably within the first 3 hours, so that aggres-sive therapy can be instituted. Imaging plays a critical role in patient management, evaluating the size of the injury to the brain, determining whether revas-cularization can minimize the size of the stroke, detecting hemorrhage, and excluding other pathologies.

Stroke imaging

Evaluation of the neurovascular tree is an important component of stroke imaging. Noninvasive methods, such as color Doppler ultrasound, MR angiography (MRA), and CTA are routinely used to evaluate the presence and extent of stenosis. Color Doppler ultrasound is readily available and in some studies has shown a 92% accuracy when compared with digital subtraction angiography (DSA). ³ However, it is limited in its ability to evaluate high carotid bifurcations, and is unable to accurately scan the intracranial vessels or evaluate the origin of the vessels from the aorta. It is still being used today, but has not been widely embraced by clinicians. In many centers, it serves as a screening tool to determine whether a patient needs any further testing.

Soon after its introduction, DSA became the gold standard for the evaluation of the carotid bifurcation and the intracranial compartment. It nicely depicts the vascular lumen and documents the rate of blood flow. This approach has shortcomings, however. For example, DSA does not show background tissue. It is a two-dimensional (2D) technology that necessitates the acquisition of additional orthogonal views to fully examine an area of narrowing. It is time-consuming and expensive to perform. It requires a specialized team and the angiography suite is generally remote from the emergency room. Intra-arterial injection of contrast material results in stroke in 1.3% of patients. In older patients, a group more likely to have a stroke, the complication rate increases to 1.8%. ⁴

MR angiography emulates DSA visually and has the advantage of being noninvasive. It nicely depicts the intraluminal area, but because of its sensitivity to the rate of proton movement, it does not adequately visualize the arterial wall. A great deal of research is directed toward enabling MRA to depict wall thickness, as well as the composition of plaque narrowing at the carotid bifurcation, however. Another shortcoming of MRA for the evaluation of stroke is that the patient must be very cooperative and able to remain motion-free for long periods of time, if the examination is to be meaningful.

CTA offers several advantages in the evaluation of patients with stroke. Most stroke patients undergo a conventional CT examination to evaluate the intracranial compartment for hemorrhage or a mass. It is easy to perform CTA at the same time. CTA emulates the DSA display through the use of multiplanar volumetric reconstructions. It is able to show the vascular lumen, as well as the vascular wall. It is noninvasive, readily available in the emergency room, and does not require extensive training to perform. In addition, CTA takes just a few seconds to acquire the data (range 15 to 30 seconds, depending on the scanner speed and the detector width, pitch, and collimation). Data can be processed in 10 to 20 minutes. CTA does require intravenous administration of contrast material, however.

CTA of stroke

Table 1 outlines the acquisition parameters for CTA of the extracranial carotid arteries and intracranial circulation. At Johns Hopkins, we typically use 80 to 120 mL of contrast material when using a scanner without bolus triggering capability, and 60 mL of contrast when using a scanner equipped with bolus triggering, injecting at 3 to 4 mL/sec and beginning image acquisition after a delay of 12 to 18 seconds, depending on the area being examined. For imaging the carotid bifurcation alone, we use a scan delay of 15 seconds. When imaging the circle of Willis, we use a scan delay of 18 seconds. We prefer to use higher concentration contrast material (350 mgI/mL).

In the past, we advocated scanning from the seventh cervical vertebra to the skull base when imaging the carotid bifurcation. With advanced 4- or 16-detector scanners, we can scan from the aortic arch through the skull base with great ease. Faster scanners are able to image at the peak of contrast enhancement in the arterial system before venous contamination occurs.

One of the problems in imaging the carotid bifurcation is the simultaneous display of arterial and venous channels. This can be minimized by using bolus triggering, as well as by minimizing the acquisition time through proper selection of collimation, pitch, and rotation time. Simultaneous display of arterial and venous channels can also be overcome in postprocessing through the use of segmentation software, enabling display of the carotid bifurcation (Figure 1). Through the use of windowing techniques, it is also possible to zero in on arterial plaque and determine its composition, as well as precisely judge the cross-sectional area of the lumen and the extent of narrowing. In Figure 2, a volumetric display demonstrates both soft and calcified portions of the plaque.

In CTA, the cross-sectional view best displays luminal stenosis. The axial image shown in Figure 3 displays prominent narrowing of the proximal internal carotid artery. The lumen is visualized, as are both soft and calcified plaque. The contralateral carotid bifurcation and internal jugular vein are both normal.

CTA versus DSA

There is a strong correspondence between the findings of DSA and CTA. MR angiography, by comparison, often fails to show poststenotic flow or overestimates the severity of stenosis. MR angiography may show occlusion of the proximal internal carotid artery, for example, whereas CTA will demonstrate a patent but severely stenotic artery.

Early studies showed that the multiplanar nature of CTA enables it to differen-tiate calcified plaque from contrast material, providing information about plaque calcification, ulceration, and size that cannot be obtained with conventional angiography. ⁵ More recent studies comparing CTA and DSA have shown that CTA is sensitive and specific for the diagnosis of occluded large intracranial arteries. ⁶ Shrier et al ⁷ demonstrated a 99% agreement between CTA and DSA in diagnosing internal carotid arterial occlusions. In a study by Knauth et al, ⁸ CTA was able to detect complete occlusions of the internal carotid artery, middle cerebral artery, and basilar arteries with 100% accuracy. Most important, the Prolyse in Acute Cerebral Embolism (PROACT) study demonstrated that a negative CTA screening examination could eliminate the need for conventional angiography. ⁹

Unlike 2D DSA, CTA offers an unlimited number of views from the data set. This represents a significant advantage, as the arterial branches can be rotated on the screen to provide the best possible view. This could be challenged in the future by increasing use of 3D rotational DSA, however.

Reading CTA studies requires some familiarity with postprocessing techniques, including their relative strengths and weaknesses. Although volume rendering (VR) uses all the pixels in the data set to provide excellent details, it does not show the lumen. Visualization of the lumen may be important in partially thrombosed aneurysms. Three-dimensional VR images must be correlated with source images.

Similarly, when assessing stenosis of the carotid or vertebral arteries, it is important to combine VR images with maximum-intensity projection (MIP) images and oblique and curved reformats. Volume rendering tends to incorporate calcified plaques into the images, obscuring the severity of the narrowing. Some anatomy, such as tortuous origin of the vertebral arteries and a cavernous segment of the carotid arteries, necessitates the use of curved reformats.

CTA is an efficient and rapid method for the evaluation of intracranial and carotid arteries in patients with symptoms of stroke. A negative CTA may prevent unnecessary urgent conventional angiography. CTA also identifies the site of vascular occlusion and the degree of leptomeningeal collateralization, and accurately evaluates the degree of cervical luminal stenosis.

Intracranial aneurysm

Approximately 5% of people who undergo autopsy are found to have an intracranial aneurysm. More than half of ruptured aneurysms rehemorrhage, and multiple aneurysms are found in 20% to 30% of patients. ^10,11 Together these factors demonstrate the critical role of imaging in reaching an accurate and timely diagnosis of intracranial aneurysm.

As in the case of stroke, DSA represents the gold standard for the evaluation of intracranial aneurysm. This is being challenged, however. The risks associated with DSA prevent its use in screening, however, and it is instead reserved for the imaging of patients in whom the probability of finding subarachnoid hemorrhage (SAH) is high. In addition, the results of DSA are negative in 5% to 15% of patients with SAH as a result of a thrombosed aneurysm, vasospasm, poor vascular opacification, or inadequate angiographic views. ¹¹

The strengths and shortcomings of MRA for the evaluation of intracranial aneurysm are the same as for the evaluation of stroke: This technique is noninvasive, but its use is appropriate only in reasonably healthy patients capable of remaining motionless for extended periods of time.

By comparison, CTA offers several advantages. Subarachnoid hemorrhage is usually diagnosed by CT, so the addition of CTA extends the imaging exam by only 20 to 30 seconds. Three-dimensional reconstructions show the neck of the aneurysm and determine the largest neck diameter, dome size, and maximum aneurysm diameter, all of which aid in treatment planning (Figure 4). CTA also shows adjacent bony morphology.

Korogi et al ¹² compared 3D volume-rendered CTA with DSA in 49 patients with or without intracranial aneurysm. Conventional angiography detected 47 aneurysms in 35 patients. The sensitivity of CTA was 64% for aneurysms with a diameter of <3 mm; 83% for aneurysms with a diameter of 3 to 4 mm; 95% for aneurysms with a diameter of 5 to 12 mm; and 100% for aneurysms with a diameter >13 mm. ¹²

A recent study by Villablanca et al ¹³ demonstrated CTA to have a sensitivity of 98% to 100% for very small aneurysms (those <4 mm in diameter), as compared with DSA, which had a sensitivity of 95%. The accuracy of the CTA was 99% and 100% for aneurysms 3 and 4 mm in diameter, respectively. The investigators concluded that CTA can provide high-quality, noninvasive detection and characterization of an-eurysms, and can be performed using routine clinical CT scanners and commercially available image processing workstations. ¹³ Recent studies are more relevant, as they have been conducted using thinner collimation and better postprocessing methods, and benefit from an increasing familiarity in evaluating 3D data. With fast scanners, it should become increasingly common to get excellent opacification of the arterial system and achieve very reliable results.

For detection of aneurysm, CTA must depict the circle of Willis region in exquisite detail. This can be achieved by timing the contrast bolus properly, using the thinnest available collimation, and selecting a pitch factor of <1, with a 20% to 30% overlap.

Decision algorithms

Figures 5 and 6 outline the ap-proach advocated by the Barrow Neurological Institute in Phoenix, AZ, for imaging patients who have nontraumatic SAH or are at risk for cerebral aneurysm. According to the protocol, if SAH is demonstrated on head CT, these findings are compared with those of CTA. If the CTA data are congruous with the findings of standard CT, there is sufficient information to proceed with therapy. If the findings of standard CT and CTA are incongruous, particularly if the CTA
is negative in the face of SAH hemorrhage on standard CT, the patient undergoes DSA.

Patients who are at risk for cerebral aneurysm first undergo MRA. If MRA is positive, the patient either goes directly to DSA, or instead undergoes CTA. If the CTA and MRA findings are congruous, there is sufficient information to proceed with therapy. If the findings are incongruous, the patient undergoes DSA so a decision can be made about therapy. In this way, a combination of CTA and MRA can obviate conventional angiography, thereby eliminating the risk associated with an invasive intra-arterial study.

Additional applications

CTA is also useful in the evaluation of arterial dissection, a condition that typically is seen in younger patients and is associated with trauma, vasculopathy, high blood pressure, or idiopathic causes. Imaging involves evaluation of the extracranial carotid and vertebral arteries for blood penetration into the arterial wall. CTA can demonstrate not only a change in the caliber of the dissected vessel, but also in the intimal membrane.

Digital subtraction angiography is the gold standard for the evaluation of arterial dissection. In the acute stage, signal isointensity can cause MRA to miss arterial dissection. CTA, by comparison, easily shows the enlarged external arterial diameter (Figure 7). CTA can be beneficial in the evaluation of vascular malformations, angiomas, and other vascular abnormalities.

Finally, a CTA-like approach plays a critical role in the evaluation of head and neck lesions. Image data are not reconstructed in three dimensions to produce an angiographic display, but contrast is administered and the carotid sheath is examined from the skull base to the anterior mediastinum, just as in CTA. The examination, which separates vascular channels from pathologic nodes, is the only accurate way to detect the presence of pathologic nodes and to evaluate the relationship between pathologic nodes and/or malignancy surrounding the vascular channels.

Conclusion

The need for rapid and accurate diagnosis of stroke mandates the use of rapidly applicable, noninvasive radiographic techniques. CTA with or without MRA appear to be the modalities that will answer this need. With the currently available improved technology, there is no reason why these two modalities should not be the principal diagnostic instruments in evaluating the presence of intracranial vascular pathologies. Further developments during the next decade will certainly make this come true.

Figure Captions

FIGURE 1. Carotid bifurcation. Segmentation software overcomes the challenge presented by simultaneous display of arterial and venous channels.

FIGURE 2. Volumetric display demonstrates both (A) soft and (B) calcified portions of carotid plaque.

FIGURE 3. In CTA, the cross-sectional view best displays luminal stenosis. (A and B) These axial images display prominent narrowing of the right proximal internal carotid artery, as well as soft and calcified plaque. The contralateral carotid bifurcation and internal jugular vein are normal.

FIGURE 4. Aneurysm at the tip of the basilar artery. CTA demonstrates neck diameter, dome size, and maximum aneurysm diameter, all of which aid in treatment planning.

FIGURE 5. Imaging protocol for nontraumatic subarachnoid hemorrhage (SAH). If SAH is demonstrated on head CT, these findings are compared with those of CTA. If the CTA data are congruous with the findings of standard CT, there is sufficient information to proceed with therapy. If the findings of standard CT and CTA are incongruous, particularly if the CTA is negative in the face of SAH on standard CT, the patient undergoes digital subtraction angiography (DSA). (Protocol developed at Barrow Neurological Institute, Phoenix, AZ.)

FIGURE 6. Imaging protocol for patients at risk of cerebral aneurysms. Patients at risk for cerebral aneurysm first undergo magnetic resonance angiography (MRA). If MRA is positive for cerebral aneurysm, the patient either goes directly to digital subtraction angiography (DSA), or instead undergoes CTA. If the CTA and MRA findings are congruous, there is sufficient information to proceed with therapy. If the findings are incongruous, the patient undergoes DSA so that a decision can be made about therapy. (Protocol developed at Barrow Neurological Institute, Phoenix, AZ)

FIGURE 7. (A) CTA of a dissection of the vertebral artery. The image shows the proximal portion of the raising of the intima. (B) Dissection of the vertebral artery is confirmed by DSA.

Discussion

ELLIOT K. FISHMAN, MD: Thanks, Jim. I'll start with the first question. Since you're talking about strokes, typically we are dealing with older patients. Often they're diabetics with borderline renal functions. What cut-offs do you use for using contrast, in terms of dealing with creatinine in these patients?

S. JAMES ZINREICH, MD: We really look at the creatinine, and the most important thing when dealing with these debilitated patients is that we want to make sure that the creatinine level is not rising. If the creatinine level is decreasing, but still a little bit high, we'll still give the intravenous (IV) contrast.

FISHMAN: What's a little bit high?

ZINREICH: Well, 3 or 4 mg/dL is our cut-off. But the question is, is it coming down from 4 to 3 to 2 mg/dL. Now, in a patient with a serum creatinine level of 5 or 6 mg/dL, we will not give IV contrast at all. So there are several relevant questions. How important is it to give IV contrast? How important is the contrast to diagnose the entity? What is the status? Is the patient's medical condition worsening or improving?

FISHMAN: Right. But in terms of contrast selection, do you have any specific suggestions? There was an article in the New England Journal of Medicine a few months ago about the use of iso-osmolar contrast, for example, in patients with higher serum creatinine levels. They found little change in post-study renal function.

I'd like to open this question to the panel. Are there some creatinine levels that you worry about? Which specific numbers do you look at?

W. DENNIS FOLEY, MD: In our practice, we start to decrease the total load of contrast above a creatinine level of 1.5 mg/dL. But many of the CT angiographic studies now are done with reduced amounts of contrast to begin with. So if you can do almost a total body scan, at least a thorax/abdomen/pelvis scan, with 60 to 70 mL of contrast, are you putting a patient at risk? We don't know the answer to that. But, in a patient with a creatinine level above 1.5 mg/dL, it would be prudent to use a less nephrotoxic agent, even at reduced dose.

ZINREICH: That's something we definitely do. Even if the creatinine is high, but coming down, for intracranial study you really don't need 120 mL of contrast, especially for a CTA. You could easily do it with about 60 mL.

SANJAY SAINI, MD: For us, the issue of contrast is in a bit of flux. I've learned recently that the nephrologists of the world don't use the term creatinine anymore. They use the term creatinine clearance. With creatinine, there is an input sort of like an equation, that involves patient weight, and gender, amongst other factors. So they're telling us to stop thinking in terms of creatinine, and to think instead in terms of creatinine clearance.

The second issue is acetylcysteine, which we started thinking about when that study came up. Again, the nephrologists think about this a little more critically than we do. It turns out that acetylcysteine works only because, in the patient population studied, the increase in creatinine for high-risk patients was offset by a decrease in creatinine in the low-risk patients.

Another issue is that not all of these studies are done in the CT population. Most of them are done in the angiographic populations, all of whom have hydration as a critical component. So, looking as a nephrologist, even at the acetylcysteine study, the patients had 12 hours of hydration pre- and postprocedure. I don't know if hydration was involved in it, but when we tried to do this kind of study, the nephrologists say you have to use hydration as a component of any study. So, it turns out that a lot of these things are impractical in the radiology world.

So I don't know whether an isosmolar contrast agent in the CT world, where hydration really is the component of the patient management, will have the impact that we hope it will. But having said that, when somebody does have a high creatinine, we do use Visipaque.

FISHMAN: In terms of creatinine clearance or creatinine in outpatients, do you require a creatinine level before you do a study?

SAINI: Yes, we have become very protocol-oriented. Our protocol ad-dresses the two main risk factors for renal injury from contrast material: pre-existing renal insufficiency and diabetes. The combination is particularly bad. We have a screening form, and if any of those factors are positive, we get a creatinine level in those patients.

The nonproven risk factors for renal injury include advanced age, dehydration, etc. Until recently, our age cutoff was 55 years. It drove our techs crazy because almost everybody who has a CT scan done is over age 55. The only relevant study we could find used the age of 70 as a cut-off. So we changed that, as well, recently.

FISHMAN: Well, I want to comment on these issues with the outpatients. It's outpatients where you typically don't have the numbers. I know at Hopkins if you said, I want a creatinine level, about 6 hours later you would be doing the CT scan, on a good day. I don't know how it is in Boston, but it's not a 30-second study.

SAINI: For example, consider the case of a patient with normal renal function in whom you were to give 45 g of iodine, and you need to get a good-quality liver study. If they had a creatinine of 1.8 mg/dL and you decrease the contrast dose, are you getting less of a quality CT? Can you just live with that?

FOLEY: Yes, because the liver parenchymal enhancement will not be what you would like to maximize the portal venous phase. On the other hand, if your major issue is detecting a hypervascular tumor, even with the reduced amount of contrast, I think your sensitivity should still be good, because you're requiring data in that late arterial phase.

SAINI: So this comes up very often, but it turns out that most of them had normal renal function. So it is not obvious to me whether you should cut down the contrast dosages.

ZINREICH: There's another issue here. That is, when you talk about creatinine, the proper answer ought to be, especially if I am in a teaching institution, I certainly have CT and MR available. Therefore, if I am dealing with a patient with renal failure, and I want to take a look at the carotid arteries or the circle of Willis, I should do an MRA in that case and to see what that shows me.

In general, when we do head CTs, and the question is whether I'm dealing with renal failure or not, if I need to administer contrast, the first thing I ask myself is: Do I do CT with contrast, or should I first do an MRA?