CTA of the abdominal aorta guides stent-grafting
Dr. Heiken is a Professor of Radiology and Director of Abdominal Imaging, Mallinckrodt Institute of Radiology, Washington University School of Medicine, St. Louis, MO. To understand the best approach to computed tomographic angiography (CTA) of the abdominal aorta, it is necessary to examine how CT technology has changed over the last 10 to 15 years. As scanners have improved from single- to 4- to 16-slice, the speed at which patients move through the gantry has increased nearly 8-fold, from 8 mm/sec to 60 mm/sec. As a result, scan times are much shorter than in the past. An aortoiliac examination that took 44 seconds to complete with a single-slice scanner was accomplished in 15 to 35 seconds with a 4-slice scanner (Table 1). Now, with the 16-slice scanner, scan duration is only 6 to 12 seconds, depending on the image acquisition protocol. Multislice CTA has created certain challenges in aortoiliac imaging. One of the most important is that scan timing is more critical than ever. Another is that optimization of contrast enhancement necessitates modification of the scanning technique to suit the type of scanner used for the examination. Aortoiliac CT Multislice CT is the best overall technique for preoperative and postoperative evaluation for endoluminal stent grafting. The primary goal of the preoperative CT examination is to assess the patient’s anatomic suitability for an endoluminal stent-graft. Specifically, we look at the aneurysm sac and the proximal and distal aneurysm neck. We also examine the access vessels—the iliac and femoral arteries—through which the endoluminal stent-graft will be passed prior to deployment in the aorta. Assessment of the aneurysm includes measurement of its maximum diameter and identification of branch vessels arising from the aneurysm. Branch vessels that most commonly arise from the aneurysm sac are the inferior mesenteric artery, lumbar arteries, and renal arteries. We measure the proximal neck length from the lowest renal artery to the proximal end of the aneurysm, and the distal neck length from the distal end of the aneurysm to the aortic bifurcation. We measure the proximal neck diameter to assess the size of the stent-graft needed for an adequate proximal seal. The proximal neck is also evaluated for the presence of atheromatous plaque and intimal calcification, both of which can affect the adequacy of the seal. Evaluation of the access vessels includes assessment of iliac artery diameter and tortuosity. In addition, we assess the iliac and common femoral arteries for the presence of aneurysms, stenoses, atherosclerotic plaque and calcification. We determine the patency of the internal iliac arteries and measure the lengths from the aortic bifurcation to the origins of both internal iliac arteries. The goal of postoperative evaluation is examination of the aneurysm sac and the stent-graft. We measure the diameter or volume of the aneurysm and look for evidence of endoleak. In evaluating the stent-graft itself, we look at the following: the diameter of the graft; the position of the stent-graft, specifically its proximal and distal attachment sites; the integrity of the graft components; the configuration of the stent-graft, specifically whether there is severe angulation or stenosis within any of its components; occlusion of the stent-graft; and, on follow-up visits, the position of the stent-graft, to ensure that it has not migrated. This article will review techniques for performing aortoiliac CT, including: scan timing; methods for individualizing scan timing, including the test bolus method and bolus tracking; issues related to contrast enhancement, including iodine dose, injection rate, and patient weight; and techniques that may improve enhancement for CT angiography of the aorta, including saline flush and bolus shaping. Scan timing Injection duration is the most important technical factor influencing scan timing. Injection duration, in turn, is determined by the volume of contrast medium used and the rate at which it is administered. The most important physiologic factor influencing scan timing is cardiac output, or more accurately, cardiovascular transit time. Figure 1 shows the effect of cardiac output on scan timing, as gauged by enhancement of the aorta. The white aortic enhancement curve represents a patient with normal cardiac output. The remainder of the curves represent patients with cardiac outputs that are reduced by 20% to 60%.
Figure 1 It is clear from the graph that as cardiac output decreases, there is a progressively longer delay in the arrival of the contrast bolus within the aorta, which translates into a delay in peak aortic enhancement. Because of the variation in cardiac output among patients, it is important when doing a time-sensitive examination such as CTA to individualize the scan delay for each patient. The method most radiologists use to individualize the scan delay is the test bolus. With the test-bolus method, a small amount of contrast material, usually 15 to 20 mL, is injected at the same rate as will be used during the diagnostic examination. The radiologist then reviews the images and determines the time-to-peak aortic enhancement. With 8- and 16-slice scanners, an additional delay after the peak of aortic enhancement on the test bolus should be included. The test bolus technique works very well, but it is inefficient. It requires the use of an additional 15 to 20 mL of contrast material, and additional radiologist time. A much more efficient and elegant method is bolus tracking. We first select a region-of-interest on the aorta. Then, starting 10 seconds after initiation of contrast injection, we begin a series of low-dose scans to monitor contrast enhancement. Once aortic enhancement reaches a predetermined threshold—in the case of aortoiliac studies, 100 HU—diagnostic images are triggered, and scanning begins a few seconds later. Arterial enhancement The magnitude of contrast enhancement during CT angiography is determined by the rate of iodine delivery into the vascular system. Arterial enhancement, therefore, depends on the number of grams of iodine that enter the vascular system per second, which in turn is determined by the injection rate and the concentration of the contrast material. The volume of contrast medium also plays a role, primarily through recirculation. Figure 2 shows the importance of injection rate on arterial enhancement. Keeping the volume of contrast material constant, there is a progressive increase in the magnitude of arterial enhancement as the injection rate increases from 1 mL/sec to 5 mL/sec.
Figure 2 Figure 3 shows how the concentration of contrast material influences aortic enhancement. Given a hypothetical 150-pound male, the graph simulates the injection of 42 grams of iodine at 5 mL/sec, using 3 different concentrations of contrast media: 300 mgI/mL at a volume of 140 mL, 350 mgI/ mL at a volume of 120 mL, and 400 mgI/mL at a volume of 105 mL.1
Even though the amount of iodine remains the same in all 3 injection schemes, use of the highest-concentration contrast material achieves the greatest vascular enhancement, because the rate at which iodine is delivered into the vascular system is increased. Note that hepatic parenchymal enhancement is not influenced by the rate of iodine delivery, but primarily by the total iodine dose. The most important physiologic factor that influences the magnitude of contrast enhancement is patient weight. As patient weight increases, the magnitude of aortic enhancement decreases proportionately. Therefore, when imaging large patients, it is necessary to improve arterial enhancement by increasing either the injection rate, the volume, or the concentration of contrast material.2,3 For example, if a scan protocol calls for 75 mL of contrast material for a 150-pound patient, 100 mL would be needed for a 250-pound patient. Similarly, a standard 100-mL contrast dose would be increased to 130 mL, and a standard 125-mL contrast dose would be increased to 165 mL. Scan protocols Both contrast administration and scan timing must be modified according to the type of CT scanner. The scan duration for an aortoiliac CTA using a single-slice scanner, or a 4-slice scanner with narrow detector collimation, is 35 to 44 seconds. Although the scan delay for each patient is individualized, a typical scan delay would be about 20 to 25 seconds to ensure that imaging takes place during peak aortic enhancement. With faster scanners and shorter scan durations, it becomes necessary to further delay the onset of scanning in order to image during peak aortic enhancement. Assuming the contrast injection protocol remains the same, one approach is to add 5 to 20 seconds to the arrival time of the contrast bolus in the aorta, in order to image during peak arterial enhancement (Table 2). In that way, it is possible to use the same contrast injection protocol and achieve the same aortic enhancement curve, but simply adjust the scan delay according to the scanner speed. When using smaller volumes of contrast material, however, the magnitude of enhancement is reduced, and it is necessary to compensate by increasing the injection rate and/or increasing the concentration of contrast material. With shorter scan durations, therefore, the best approach is to decrease the amount of contrast material, increase the injection rate, and also use a relatively high-concentration contrast material. Table 3 outlines our standard protocol for aortoiliac CT angiography. With a 4-slice scanner, we inject 125 mL of 350 mgI/mL contrast material at 4 mL/sec. With a 16-slice scanner, we have been able to reduce the volume of contrast material to only 75 mL for standard-sized patients. At the same time, we have increased the injection rate to 5 mL/sec (Figure 4).
In patients with an endoluminal stent-graft, we also acquire a set of delayed images 60 to 120 seconds (typically, 90 seconds) after the start of the contrast bolus to assess for possible endoleak. In a small percentage of patients, type 2 endoleak is evident only on delayed images and cannot be observed during the arterial phase. There are additional ways to modify contrast injection that may improve arterial enhancement. The first is the use of a saline flush. This technique has several potential advantages. It slightly prolongs arterial enhancement, as it pushes contrast material into the vascular system that otherwise would have remained in the intravenous tubing and brachiocephalic vein. As a consequence, it is possible to reduce contrast volume by approximately 15% to 20%, if desired. The saline flush also slightly increases the magnitude of arterial enhancement by maintaining a tight contrast bolus. (In chest imaging, though not in abdominal aortic imaging, the saline flush can also reduce streak artifact from dense contrast material left behind in the brachiocephalic vein and superior vena cava.) Another method that can improve arterial enhancement is bolus shaping (Figure 5). The typical aortic enhancement curve that is achieved using a standard uniphasic injection protocol is not ideal for CT angiography. It results in a single peak of aortic enhancement that generally is of far greater magnitude than necessary, but of relatively short duration.
Figure 5 Instead, it would be ideal to achieve an adequate, uniform, and prolonged level of vascular enhancement. Uniformity of enhancement is helpful for image postprocessing, and prolonged enhancement is very useful when acquiring a larger imaging volume or, with slower scanners, using a narrower collimation. With a 16-slice scanner, it enables a reduction in the volume of contrast material. Bolus shaping is accomplished through an exponentially decelerated injection technique.4,5 Rather than injecting at a uniform rate, as in a uniphasic injection, we begin with a relatively high injection rate and gradually decrease the injection rate throughout the injection period, at an exponentially decelerating rate. This injection pattern enables either a reduction in contrast volume or an increase in coverage, depending on the application and type of CT scanner. The prolonged period of adequate vascular enhancement also reduces the chances of incorrectly timing image acquisition and “missing” the contrast bolus. Figure 6 is an example of vascular imaging from the neck through the proximal thighs using the exponentially decelerated contrast medium injection technique. Region-of-interest analysis shows that there is no more than a 15 HU difference in attenuation throughout the entire examination.
Figure 6 Conclusion Scan timing is critical for CTA of the abdominal aorta. Individualizing the scan delay for each patient can be done using either test bolus or bolus-tracking techniques. The magnitude of arterial enhancement is determined primarily by the rate of iodine delivery into the vascular system and is, therefore, strongly correlated with contrast injection rate, concentration, and volume. With fast scanners—16-slice and beyond—it is best to reduce the volume of contrast material while using a rapid injection rate and high-con-centration contrast medium. Bolus shaping can improve CTA by producing prolonged, uniform vascular enhancement using an exponentially decelerated injection protocol. REFERENCES
Discussion ELLIOT K. FISHMAN, MD: Thanks very much, Jay, for that terrific talk. Now we’ll open discussion to the panel. You talked about high-con-centration contrast and said that you are using 350 mgI/mL now. Theoretically, at least, on curves, even higher concentration might be better. But in terms of regular practice, do you think there will be a need for that? JAY P. HEIKEN, MD: It’s difficult to say. But, for the scanners that are currently available, I think 350 mgI/mL is adequate for just about every application we have. It remains to be seen as we move to 64-detector-row scanners, whether it might be beneficial to go to even higher concentrations. Theoretically, the higher concentration does have an advantage, primarily for CTA, although not as much for parenchymal imaging. But, specifically for CTA, as we begin to image even faster, the higher concentration might be beneficial. MICHAEL P. FEDERLE, MD: I’m just wondering, in follow-up to Elliot’s question, whether with cardiac CT and 64-slice scanners, we may have to re-evaluate that. Even higher concentrations of contrast may turn out to play an important role. But I agree that for the body indications as we currently use them, I haven’t found higher concentrations particularly advantageous. Let me play the devil’s advocate, Jay. Now that we’re using 16-slice scanners, and the next generation is bolus shaping still as critical? Because we can now scan through the entire thorax and abdomen, and so forth, so quickly, do we need to tailor the shape of the bolus quite so elegantly? HEIKEN: Well, with our faster scanners, I think there is less of a need for bolus shaping. But I think that it still can be very important, particularly when we are doing prolonged examinations, such as vascular runoff examinations, which Dominik will discuss shortly. Because of the hemodynamics in patients with peripheral vascular disease, for that examination it is important to do a relatively prolonged examination in order to get good opacification of the proximal abdominal aorta all the way down to the peripheral runoff vessels. Bolus shaping still plays an important role in that group of patients by giving us a nice prolonged uniform level of enhancement within the vessels. For other applications, it won’t be as important. But it will still allow us to decrease the volume of contrast material we use. Even though we are imaging during a shorter and shorter period of time, we can achieve our imaging with a smaller volume of contrast material if we inject it using bolus shaping. I think that is also an important issue. Finally, I think the uniformity of enhancement with bolus shaping is helpful. If we use a uniphasic injection, in which we get a single peak of enhancement, especially if we are using high injection rates, we get a very sharp peak of enhancement. If our timing is off slightly so that we’re catching the enhancement during either the rise or the decline in enhancement, that lack of uniformity may have some effect on our postprocessing, because most of the postprocessing techniques that we use are based on a certain threshold of enhancement. So I think the uniformity in general has an advantage, but I don’t disagree with you that there is somewhat less of a need for bolus shaping with the faster scanners. But I think it still can play an important role. Alec? ALEC J. MEGIBOW, MD, MPH, FACR: When you talk about a lower volume of contrast with the faster scanners, does the total grams of iodine to the patient remain constant because you are using more concentrated medium? HEIKEN: No, it doesn’t have to because, for instance, in our protocol, we have always been using 350 mgI/mL concentration for CTA. So we’ve simply decreased the volume using the same concentration, but we’ve increased the injection rate to help compensate for the decreased enhancement to a certain extent. MEGIBOW: Have you played with 80-kV scanning as a potential? HEIKEN: We have not, but that is a very good question. What Alec is alluding to is the fact that with a lower kV, the attenuation value of the contrast material within the vessels is increased. So we may be able to decrease the contrast material we use if we use a lower kV. We have not done that. But maybe someone else on the panel has. CHRISTOPH R. BECKER, MD: As a matter of fact, we tried to use 80 kV; 80 kV is somewhat critical because, particularly in bigger patients, the X-rays do not penetrate enough to get really good signal. But as a good trade-off, we have used 100 kV, and this is optimal to get a really good bright enhancement, and to have a good signal-to-noise ratio. So we found this superior to 80 kV. We got a really good enhancement and we were able to decrease the radiation amount by about 20%, as I recall. JULIA R. FIELDING, MD: I was going to ask if you use a fixed mA when you are doing these studies, or if you use a variable mA. HEIKEN: We currently are using a fixed mA, but it varies from patient to patient and depends upon the patient’s size. FIELDING: Okay. BECKER: Actually bolus shaping sounds fascinating to me, but, nevertheless, it is somewhat complicated and is probably not really practical right now. HEIKEN: What do you mean it’s really not practical? BECKER: You need to calculate something in advance, probably on the basis of the test bolus. HEIKEN: Actually, you do not need a test bolus. It is just a matter of simply pressing a button on an injector console and switching to that type of technique. So it’s very easy to implement. BECKER: Nevertheless, all the curves you have shown here are on the basis of the calculation probably obtained at a certain level of the body. Nevertheless, if you are extending to a long-range scan, like the whole body or at least the entire aorta, it may be that we need to take into account that the blood and the contrast column moves and that the scanner moves at a different speed. Does this affect this shaping then? HEIKEN: I don’t think it really has a significant effect. In our experience, this has worked very well in all patients. The fact that the scanner is moving and the duration of the scans that we’re doing is so short, I don’t think there really is much effect from the factors you just described. BECKER: On the other hand, the kind of exponential decay of contrast administration that you were suggesting actually makes perfect sense in a kind of drug that is administered within hours or days. But for contrast media, in particular, with this short period of administration time, wouldn’t it be better to reduce the injection rate incrementally every time the recirculation is taking place and the contrast is coming back after the first and second free circulation? It would be a step-wise reduction of peak flow rate to compensate for this free circulation. HEIKEN: No, I don’t think so. I think it makes more sense to do it this way. What happens with a step-wise decrease is that, first of all, we don’t know what the patient’s circulation time is and that becomes complicated to determine. So if you have a patient and you don’t know what his/her circulation time is, it makes even more sense to do it exponentially, because any time you make an abrupt change in the injection rate, it gives you a dip in the enhancement curve. Dominik has done some very nice work using biphasic imaging. In fact, in those enhancement curves, there does tend to be a dip at the change in injection rate. DOMINIK FLEISCHMANN, MD: It makes sense mathematically and physiologically to use an exponential decay of the injection flow rate. It corrects for the otherwise continuous increase of arterial opacification observed with uniphasic, constant rate injections. I would like to comment on the topic of the speed at which the contrast moves down to the aorta. Blood flow in the normal aorta is relatively fast, so it’s not necessary to take into account the longitudinal distance change between the proximal aorta and in the pelvis, for example. Except, of course, if the patient has aneurysms or large aneurysms, then it makes a difference. We know this from angiography, from cases in which one can see slow flow and swirling of contrast within an aneurysm sack. But for a normal-size aorta, it doesn’t make a large difference and it is within maybe a second or 2. FEDERLE: Jay, I have a question, and maybe I’m just misunderstanding something here. You have a graph that shows the effect of contrast volume on the aortic enhancement and scan timing. On the bottom line of the graph, you have the injection duration. It seemed to me that the aortic enhancement should almost always start falling off as soon as you stop injecting. But, according to that, for the faster scanners we would need to delay the amount of time after the injection because the peak, according to the graph, is still continuing to go up in the aorta after the injection has stopped. Is that true? HEIKEN: Yes.FEDERLE: It continues to go up after the injection stops? BECKER: It may. This just depends on the time it takes for the bolus to travel from the IV site to the artery. Let’s just suppose that we have a 40-second injection; we can expect that the arterial enhancement will also increase roughly for almost 40 seconds. So if your contrast media transit time is only 10 seconds, then you will still have a rise in enhancement when you’re in the aorta. But if your delay is much longer, then you will have already reached the end of injection, and then it starts to increase in the aorta, at that point of time. When you compare these, you have to take into account that the transit time drops relative to the injection. That would shift at least 40 seconds between those two. If the delay has to be very long, then you may have the end of injection even before the contrast arrives in the aorta. You can compare the end of the injection with the end of the plateau. FEDERLE: It’s still not intuitive to me why we should have to wait longer to initiate the scanning with the 16-slice scanner versus the 4-slice scanner relative to the beginning of the injection rate, assuming all other things are equal, including the injected volume and injection rate and so forth. FISHMAN: At Hopkins, we’re very simplistic. It’s like an article published by Macari at NYU about preset boluses; things tend to even out, particularly when you do postprocessing. I think one of the things that happens when the process gets too complicated is there are more errors, because people are overthinking. MEGIBOW: I was going to say the same thing. We kind of take a Luddite approach to this. If you look at your curve with the cardiac output, you really have to reduce 60% of the cardiac output before you really see a noticeable effect at scanning, if you are beginning scanning at 25 seconds. The clinical impact of an aorta enhanced at 300 HU versus 200 HU units, in terms of the imaging and diagnosis, seems to me to be questionable. If you get the aorta up over 200 HU, I don’t think you can visually tell, and the software can handle the rest. FISHMAN: One of the things about implementing CTA in clinical practice is that it has to be simple. If there is a lot of calculating, there is a tendency for people not to do it. We are doing 30 of these cases a day. Sometimes we’ll do 8 CTAs in a row. I think we do a pretty good job. I’ve not heard anybody complain. FLEISCHMANN: But even if the theoretical calculations are complicated, the practical consequences are very simple. The main consequence when adjusting the injection protocol to a faster scanner would be that your scanning delay should be a little longer. This is a simple rule that you can include in any protocol. You don’t have to do any calculations on the fly at all. If you start to scan a little later, relative to when the contrast arrives, your enhancement will be better. HEIKEN: These are not things that require any calculations. If you use either a test bolus or bolus tracking, you are just automatically forced to change the scan timing based on the aortic enhancement and the scan duration. MEGIBOW: I just have a comment about the end of graft follow-ups. Our traditional protocol has been to do a noncontrast study for postendograft follow-up. Then we do an arterial-phase and then a delayed-phase scan to pick up endoleaks that we missed. We are going to report our findings at the RSNA this year. We have found that the arterial-phase image was adding nothing; and we’re thinking about dropping it from the protocol. We’d save people radiation, approximately 7 mSvs a pass, and that does build up over time. Has that been the experience of other people? Although you get pretty pictures in the postendograft follow-up, we’ve seen most of the endoleaks, and, in our experience, we can eliminate a run. FEDERLE: So you do noncontrast and then how long do you delay out? MEGIBOW: I can’t tell you exactly; it’s approximately 60 to 70 seconds. We do just the two runs. FLEISCHMANN: We usually do arterial- and late-phase images. I don’t think that radiation dose is an issue in a patient who has an aortic endograft. So it comes down to the question of whether you would miss anything important when dropping the arterial-phase acquisition. Intuitively, I would prefer to know if an endoleak is filled by an early-enhancing feeder, because this might influence therapy. Also, an endoleak that is only detected on a late-phase image is very unlikely to be a type I endoleak––so I feel there is relevant information from obtaining both early and late postcontrast phases. FEDERLE: But on the other hand, if you’re going to do a delayed phase and an arterial phase, you could probably eliminate the noncontrast study, right? FLEISCHMANN: No.HEIKEN: No, because sometimes the perigraft calcification could be mistaken for an endoleak. MEGIBOW: Right, but if you have both arterial- and delayed-phase images, wouldn’t it be easy to tell what’s calcification and what’s not? FLEISCHMANN: Sometimes it’s not that easy, particularly when we do the late phase, usually it is somewhat thicker collimation. Although someone would need to analyze this systematically, intuitively, I would feel more comfortable comparing it with the noncontrast study. FISHMAN: Okay, thanks very much, Jay. |
