CT examination of mesenteric vessels and small bowel
Dr. Horton is an Associate Professor of Radiology, Johns Hopkins University School of Medicine, Baltimore, MD.
Computed tomography (CT) plays an important role in the diagnosis of mesenteric ischemia and small-bowel neoplasms. It enables detailed examination not only of the mesenteric arteries and veins, but also of the small bowel itself, including the wall. Improvements in scanner technology, computer technology, and contrast media are markedly improving visualization and will further expand the role of CT in the management of these diseases.
One of the ﬁrst considerations when imaging the small bowel and mesenteric vessels is the choice of oral contrast agent. For routine abdominal imaging, positive contrast agents such as diatrizoate sodium meglumine or barium sulfate are sufﬁcient. Our protocol at Johns Hopkins Medical Institutions calls for patients to drink 500 to 750 mL 30 to 90 minutes before the exam, and another 250 mL immediately before the exam.
However, for detailed imaging of the small intestine and the mesenteric vessels, it is necessary to use a neutral or negative contrast agent. Neutral agents have an attenuation similar to water, and include water itself and methylcellulose solutions. We use water, 750 to 1000 mL 30 to 90 minutes before the exam, and 250 mL once the patient is on the table.
Negative contrast agents are fat- or oil-based. These are excellent for visualizing the small intestine, as fat in the lumen enables excellent visualization of the enhancing intestinal wall. The use of oil-based contrast agents is impractical, however, as most patients cannot tolerate drinking 2 to 3 cups of peanut oil before the study, and the body cannot digest and absorb that much fat.
Milk has also been used as a negative contrast agent1 and in some ways may represent a good compromise. Because water is emptied very quickly from the stomach and moves rapidly through the intestine, it fails to produce optimal distension. Whole milk is 4% fat. As a result, it is emptied more slowly from the stomach and moves more slowly through the intestine, resulting in better distention. We don’t use milk, however, as it is often impractical to have the patient drink several cups of milk.
Carbon dioxide would make a good negative contrast agent,2 but there is no practical way to deliver it noninvasively to the intestine. Therefore, water and the new methylcellulose agents are most often used for endoluminal enhancement during examination of the small intestine and mesenteric vessels.
We scan from the diaphragm through the pubic symphysis. We usually do not perform noncontrast scans. However, if the speciﬁc indication is suspected, small-bowel hemorrhage, a noncontrast scan can show high-density blood in the intestinal wall.
Dual-phase imaging is necessary for visualization of both the mesenteric arteries and mesenteric veins. The arterial-phase acquisition begins about 25 seconds after the start of the injection of intravenous contrast material, and the venous-phase acquisition, about 50 seconds after the start of the contrast injection.
We use nonionic contrast media, typically 120 mL of nonionic contrast material with a concentration of 350 mgI/mL, and inject at 3 to 5 mL/sec through a peripheral line. Many patients with mesenteric ischemia are elderly and have atherosclerotic disease and renal insufﬁciency. In such patients we use iodixanol (Visipaque), an isosmolar contrast agent.
For examination of the small bowel, detector collimation may be set at 0.75 mm or 1.5 mm, depending on the indication. We often use a 1.5-mm detector collimation, and then create 2-mm slices, reconstructing every 1 mm for 3-dimensional (3D) imaging. For review at the workstation or in hardcopy, we use 3-mm slices.
For evaluation of the mesenteric vessels, it is necessary to use thin collimation. Typically, we use a 0.75-mm collimation, and make 0.75-mm slices, reconstructing every 0.5 mm for 3D imaging. We print 3-mm slices for workstation or hardcopy review.
Viewing the dataset
Figure 1 shows a normal small bowel, with endoluminal water clearly highlighting the folds of the jejunum and ileum. Often the jejunum and ileum are not fully distended at the time of image acquisition. This should not be cause for concern, however, as the use of intravenous contrast ensures good enhancement of the wall and makes it unlikely that any abnormality will be missed, even if some of the intestinal loops are not completely distended.
At the same time, by simply changing the settings on the workstation monitor to highlight the vessels and make other tissues appear more radiolucent, it is possible to examine the mesenteric vessels, mainly the superior mesenteric artery (SMA), superior mesenteric vein (SMV), and celiac axis.
Examining the entire arterial and venous dataset generally takes less than 5 minutes. It is easy to scroll through the images, starting with the axial view, and then examining sagittal and coronal images. Typically, the sagittal view shows the origin of the celiac artery and the SMA. Atherosclerotic disease often occurs at the origin, so it is important to closely examine that area. Coronal or coronal oblique views show the branches of the SMA, including the jejunal, ileal, ileal colic, and right colic arteries. In some patients, it is possible to see the very terminal branches of these arteries.
To do a thorough study, it is important to examine 3D views and volume renderings. The inferior mesenteric artery (IMA), for example, is barely visible on axial images, appearing only as a dot. As shown in Figure 2, a sagittal volume rendering depicts the IMA as it courses down to the pelvis and bifurcates into the hemorrhoidal arteries. A coronal oblique view and coronal maximum-intensity projection further shows terminal branch vessels feeding the surface of the bowel.
It is also possible to document anatomic variants. This is particularly important in patients with chronic mesenteric ischemia, who may require corrective surgery. Examples of typical anatomic variants include a common origin of the celiac artery and the SMA, a replaced or accessory right hepatic artery coming off of the SMA, and a separate left gastric artery.
The mesenteric veins can be easily imaged during the second acquisition of a dual-phase study. Generally, a coronal oblique view will show the mesenteric, splenic, and portal veins at the conﬂuence (Figure 3).
Acute mesenteric ischemia
Mesenteric ischemia is a complex condition that is difﬁcult to diagnose clinically. It can be acute or chronic and can occur in a variety of conditions. The incidence of mesenteric ischemia increases with age.3
The etiology of acute mesenteric ischemia in 50% of patients is thrombosis, usually to the SMA. Another 30% of patients have nonocclusive arterial ischemia. The least common cause of acute mesenteric ischemia is mesenteric vein thrombosis, although at Johns Hopkins we see several patients with this condition every year.4 Depressed cardiac function, hypotension, and certain medications—essentially, anything that results in insufﬁcient blood ﬂow to the intestines—can also cause mesenteric ischemia.
Patients with acute mesenteric ischemia usually experience abdominal pain that appears out of proportion to the ﬁndings on physical examination. They may also experience nausea, anorexia, vomiting, and bloody diarrhea. Whereas in the past clinicians may have hesitated to send a patient with suspected mesenteric ischemia for invasive angiography, today they readily turn to CT for assistance in making the diagnosis.
In acute mesenteric ischemia, the CT examination will show one or more of the following: thrombus in the mesenteric arteries or veins, atherosclerotic plaque, focal narrowing of the vessels, tumor-related vascular occlusion, and small attenuated vessels.
Close examination of the bowel will reveal thickening of the bowel wall. The thickening is usually circumferential and results in a total bowel wall thickness of <1.5 cm. Submucosal edema and inﬂammation may also be evident, as well as intramural hemorrhage.
Bowel-wall thickening is a nonspeciﬁc ﬁnding, but when considered with other factors it can suggest the diagnosis of mesenteric ischemia in many patients, and certainly suggests the need for a contrast-enhanced scan. The distribution of bowel-wall thickening, evidence of atherosclerotic disease, and abnormalities of the mesenteric vessels should all raise a suspicion of mesenteric ischemia.
Some patients whose mesenteric ischemia is caused by vasculitis develop ascites, mucosal hyperemia, and mesenteric stranding. Vasculitis, hypotension, and nonocclusive ischemia may also result in markedly narrowed vessels and visualization of fewer arterial branches than usual.
According to the scientiﬁc literature, changes in bowel enhancement can also signal mesenteric ischemia.5 In my experience, however, such ﬁndings are often too subtle to be useful. Very few patients have a complete lack of bowel enhancement or clearly delayed enhancement.
CT may also show air in the bowel wall, or mesenteric or portal veins. Pneumatosis and portomesenteric venous gas are late ﬁndings, however, that usually identify an infarcted bowel requiring surgical resection (Figure 4). Both pneumatosis and portomesenteric venous gas can be seen in nonischemic conditions (including diverticulitis and chronic obstructive pulmonary disease), after endoscopy, and in association with steroid use. A patient with bowel infarction has a much worse clinical presentation, however.
Bowel infarction is also associated with a very poor prognosis. Wiesner et al6 studied 18 patients with pneumatosis or a combination of pneumatosis and portomesenteric gas. They found that 78% of patients with pneumatosis alone had transmural bowel infarction. The mortality in this group was 56%. A combination of pneumatosis and portomesenteric gas was associated with a 91% infarction rate, and a mortality of 72%.
Chronic mesenteric ischemia
Chronic mesenteric ischemia is the consequence of atherosclerotic disease. It occurs in older patients and develops slowly over time, progressing in some cases to acute ischemia.
CT ﬁndings that identify chronic mesenteric ischemia include atherosclerotic plaque, both calciﬁed and noncalciﬁed; focal narrowing of the mesenteric vessels; and small attenuated vessels. The presence of collateral vessels is a key ﬁnding of chronic ischemia (Figure 5). Patients with chronic ischemia who develop acute ischemia may have evidence of thrombus as well.
It is important to note that many elderly patients have calciﬁed atherosclerotic plaque in the mesenteric arteries that is not hemodynamically signiﬁcant and does not cause mesenteric ischemia.
Median arcuate ligament syndrome may also be associated with mesenteric ischemia. The median arcuate ligament is a ﬁbrous arch that unites the crura on either side of the diaphragm. It is a normal structure that usually crosses anterior to the aorta and a little above the celiac axis. In approximately 15% of people, the median arcuate ligament crosses lower, across the proximal celiac axis. In some cases, it causes signiﬁcant stenosis and symptoms, which are usually worse on expiration. CT ﬁndings include a characteristic hook appearance of the celiac axis and a very tight stenosis 5 mm to 1 cm from the origin (Figure 6).
Like mesenteric ischemia, the evaluation of small-bowel neoplasms involves thorough examination of the mesenteric vessels and bowel. Small-bowel neoplasms are uncommon, occurring in 1 in 100,000 people and accounting for fewer than 25% of gastrointestinal neoplasms.7
Usually, patients experience abdominal pain, weight loss, and anemia. Most small-bowel neoplasms are small when symptoms ﬁrst appear, but diagnosis can be delayed for several years because patients present with such nonspeciﬁc symptoms. In addition, good diagnostic exams are lacking. Endoscopy is of limited value, and radiologic studies have a high false-negative rate.8
Small-bowel neoplasms comprise 4 types of tumors: adenocarcinoma, carcinoid, lymphoma, and gastrointestinal stromal tumors (GIST), in that order of incidence.
The diagnosis of adenocarcinoma of the duodenum is usually made on endoscopy, with CT used primarily for staging. However, CT is used for detection and staging of jejunal and ileal lesions. Figure 7 shows the importance of examining not only the bowel but also the mesenteric vessels. In this patient, duodenal adenocarcinoma has expanded beyond the bowel to involve the superior mesenteric vein.
Carcinoid tumors constitute 33% of small-bowel malignancies. All carcinoid tumors have the potential for malignancy. There are no histologic differences between benign and malignant carcinoid tumors; therefore, early diagnosis is crucial. The smaller the tumor, the greater the chance that it is benign. Fewer than 2% of tumors <1 cm in diameter are malignant, whereas 80% of tumors <2 cm in diameter are malignant.9 Finding a carcinoid tumor early, when it is still a small hypervascular submucosal mass, requires good CT technique with contrast enhancement and very careful examination of the bowel wall.
Carcinoid tumors arise from the bowel wall, but once the tumor has extended into the mesentery, CT shows a characteristic mesenteric mass and desmoplasia. Calciﬁcation is present in 70% of patients. Angiographic studies are needed to determine whether the tumor has encased the mesenteric vessels. The patient may also develop mesenteric ischemia and bowel thickening, either from encasement of the mesenteric vessels or excretion of chemicals from hormonally active tumors, which damage the smooth muscle in the vessel wall (Figure 8).
Lymphoma is the third most common small-bowel malignancy, accounting for 10% to 15% of cases. This tumor may occur anywhere along the gastrointestinal (GI) tract, most often in the stomach but also in the small bowel, colon, and esophagus, in that order.10 It may involve the GI tract as a primary small-bowel lymphoma, or as a secondary manifestation of systemic disease.
CT examination usually involves just one scan, conducted during the portal-venous phase. Characteristic CTﬁndings in patients with lymphoma of the small bowel are thickening of the bowel wall, aneurysmal dilatation, adenopathy, and in some cases, encasement of the mesenteric vessels (Figure 9).
Gastrointestinal stromal tumors (GIST) are the fourth most common small-bowel malignancy, representing fewer than 10% of cases. They are equally common in the jejunum and ileum, and less so in the duodenum. Several types of tumors fall under the heading of GIST. Derived from smooth muscle, or neural or lymphatic tissue, these tumors can be distinguished histologically with immunophenotyping.
Three-dimensional CT plays an important role in diagnosing GIST. It also enables thorough mapping of the vessels and evaluation of neovascularity. CT is especially useful for distinguishing GIST from other types of pathology in adjacent organs. These tumors can grow so large that patients are often referred for CT evaluation of a suspected pancreatic or kidney tumor, or an abscess, for example.
In the early stages of GIST, CT shows a small mass in the bowel wall. As the mass grows, it becomes very bulky and, often, exophytic. The larger the tumor, the more likely it is to develop central ulceration or necrosis (Figure 10). Gastrointestinal stromal tumors, especially those derived from smooth-muscle tissue, can look like sarcoma. Careful examination of the vessels is essential, therefore, to differentiate GIST from tumors arising from the vessel wall itself.
Continued advancements in CT and computer technology enable better visualization of the mesenteric vessels and the small-bowel wall. As a result, CT angiography is playing an increasingly important role in the evaluation of patients with small-bowel pathology, especially ischemia.
New oral contrast agents have improved bowel distention. Neutral or negative agents enable better evaluation of the enhancing bowel wall.
ELLIOT K. FISHMAN, MD: Thanks very much, Karen. Do you believe that you need to use postprocessing in looking at the small bowel?
KAREN M. HORTON, MD: Yes, I think you deﬁnitely have to do some sort of 3D imaging. It’s very hard to get a good feel for the anatomy and pathology if you’re just using axial imaging. Also, I think that by doing the 3D volume rendering, you can kind of play with the settings to accentuate the bowel or to accentuate the folds a little bit better than if you are just looking at standard soft-tissue windows at a monitor or on ﬁlm.
MICHAEL P. FEDERLE, MD: There is a question I get a lot from people, and I don’t have it fully resolved in my own mind. That is, if you are seeing a patient with abdominal pain—mesenteric ischemia is one of the possibilities, but there is a whole world of possibilities for why they have abdominal pain. How do you decide when to give a negative contrast and when to give an oral contrast? It could be that it is really diverticulitis or perforated ulcer or something else causing the pain.
HORTON: A lot of times we get that too. But if the presentation were given just as abdominal pain, routinely we would give the high-densi-ty oral contrast agent. If the physician can limit it a bit more to what they are really thinking of, then we can tailor our protocol. For certain things, we would always give water—these include mesenteric ischemia and any type of gastrointestinal tumor, except colon cancer. So if it’s a gastric cancer and they wanted us to stage it, or if they’re looking for a small bowel tumor and a patient may have chronic blood loss, we give water. We use water any time we really think we are going to have to look at the bowel wall or look at the vessels in detail.
But I always look at the mesenteric vessels in patients with abdominal pain, even though there might be contrast in the bowel—you still may have to do a little editing. But you can still get a good look at the mesenteric vessels. You really need to look there, because people don’t think about it and I think it is underdiagnosed.
ALEC J. MEGIBOW, MD, MPH, FACR: Like you, I’ve thought about neutral contrast agents for a long time. As the images from these 16and 64-row scanners start becoming more prevalent, with the better use of IV contrast and all of the information that’s embedded there, I think that neutral contrast agents are going to be used more frequently. I think they will replace the positive contrast agents, but I do I think you need to use IV contrast with them. I would not use them on a noncontrast scan.
I would like to ask you about ischemia. Many patients who are suspected of ischemia end up going to MR, because some MR imagers claim that they can quantify the degree of ischemia by ﬂow, and, thereby, stratify patients by who may or may not beneﬁt from an intervention. Since you do a dual run and you are looking at both the arterial and venous phase, is there any way to get a handle on the physiologic consequence of the morphologic abnormality in these patients?
HORTON: At this point, I don’t think so. But early on, I looked at measuring the bowel wall and the enhancement of the bowel—it is very difﬁcult to do technically due to partial volume averaging. If you could do those kinds of measurements, then you could analyze blood ﬂow and enhancement. But, at this point, I don’t think it’s practical to do that. At Hopkins, I don’t think we do a lot of MR for this. I think the referring physicians really rely on the CT. If I say that everything looks normal, they’re not really worried about it. If something looks abnormal, they might go on to do another study. But I think if you get a good look at the bowel wall, and you look at the vessels, you’re not going to miss anything.
MEGIBOW: One other question. It seems that we do nuclear scans now for evaluation of the small bowel, for example, for GI bleeding. With the scan times that you can get now and the degree of contrast opaciﬁcation, it would appear that CT could replace those studies. Has anybody thought about what the minimum rate of bleeding would necessarily be?
We know, for instance, with the technetium scan, you need to be doing at least a minimum of bleeding. Is there any research on that for CT? It would seem that a 64-slice scanner should be the way you should work-up GI bleeding, period.
HORTON: We have been able to ﬁnd the source of GI bleeding in a few patients. At this point, I don’t think that you would be able to ﬁnd the source in the average patient. But, maybe it will be possible with the 64-slice scanner.
FISHMAN: I think with 64-slice, particularly if you have a 3- or 2-second abdomen scan, then you’d probably have the same detection rate that you’d have with classic angiography.
FEDERLE: My experience is just anecdotal also. But I’m pretty bullish on the possibilities of CT becoming important in GI bleeding. Think about the fact that we see active extravasation in anticoagulation patients. That’s on the venous side, usually. In the setting of trauma, we pick up active bleeding very often. We pick it up in 20% of our patients who have positive trauma CT scans.
So my guess is that if we routinely use a negative or neutral oral contrast medium and do CT, I bet we could detect the source of bleeding as well as angiography can. It’s a guess, but I think it is possible, based on our experience in other settings that should translate pretty well.
FISHMAN: I think also with the 64-slice scanner, and even the 16-slice now, one of the things we’re looking at is that roughly 70% of our scans use water. If there is a question about erring on one side or the other, we will err on the side of water. Again, you have to use IV contrast, there’s no doubt about that––otherwise it is a wasted study.
But you’re also looking at the small bowel and differential enhancement of bowel folds. We’ve been able to pick up early ischemia. Our surgeons have a lot of conﬁdence in patient management based on what we do or don’t ﬁnd.
The big problem Karen mentioned was with water. I think that Alec has some experience with some new agents, don’t you?
MEGIBOW: Well, we’ve been working with EZ-EM (Westbury, NY) to develop a 0.1% barium solution that has some of their suspension agents; it turns out that if you give this the way you give the water, you’ve got to give a large amount in a short period of time. You get superb distention throughout the small bowel and good visualization of the anatomic features that you need to see. This will be presented at the 2004 RSNA meeting.
We ran some statistics looking at the maximal luminal diameter using this agent, water, and conventional barium. A medical student measured the maximal diameter in many different patients. This solution had signiﬁcantly better distention of any segment along the GI tract, along with visualization of neural features, which was equivalent to that of water.
FEDERLE: Where does it fall in the density range?
MEGIBOW: It usually shows up at somewhere between 15 to 30 HU.
HORTON: What does it taste like?
MEGIBOW: It’s not bad. Actually I’ve tasted it, and it has sort of a minty taste. It doesn’t necessarily compete with water, but the problem with plain water is that it is absorbed really quickly. So by the time you get to the ileum, you don’t have much contrast left. If you put methylcellulose in there, that adds a taste to it that is fairly well tolerated. It has been used in some other centers; some experience at the Mayo Clinic and at Massachusetts General Hospital. They have been successful with it and have not seen any adverse effects. There have been some reports of some accelerated intestinal transit, shall we say? But that hasn’t been the experience in some other places. We’ll see what happens; it is an interesting agent.
JULIA R. FIELDING, MD: We just spent all this time talking about doing really careful bolus timing, etc. And you guys are doing 25 straight off the bat. Obviously, it’s an entirely different approach from the bolus timing.
FISHMAN: I think a lot of that relates to how much you are doing 3D postprocessing. Something that Alec said before is the reality. Whether you are using 270, 300, or 350 mgI/mL contrast, it is probably not going to make that much difference, particularly if you do volume rendering very well.
One of the things I’ve always tried to push is the strategy of moving aggressively into volume imaging and postprocessing. What Karen showed in the bowel is really critical. We do CT for all these potentially ischemic patients, and we’ve had very good results. A few of the patients have gone onto angiography and there has been a really good correlation with the CTA ﬁndings. Again, I think that if you stay in the axial world, you don’t do very well on these studies. Or you do not pick up the early cases. So it is a way of recalibrating how we do business.
FIELDING: But what I was trying to get at was that you don’t use bolus tracking and you scan very early for your arterial phase–25 seconds is pretty early on a 16-slice scanner.
FISHMAN: If you just had to scan the diaphragm to the pelvis with a delay of 25 seconds, you are hitting the superior mesenteric artery at roughly 31 seconds. You are hitting the distal vessels in about 36 to 37 seconds. So it’s not really as early as you would think. But you are getting really good opaciﬁcation, and you get an incredible angiographic map. So if you use CT the same as you use, for example, catheter-based angiography, the timing tends to be the same.
MEGIBOW: We have never really looked at our experience with this rigorously. But I sit at that workstation all the time doing this stuff, and you have such wide latitude in terms of the timing that it makes it much simpler than when we try to actually discuss it.
FIELDING: Well, I use ﬁxed timing, but I use empiric timing as well. I see lots of discussion in the literature all the time about this kind issue. People have very strong opinions about where it should go.
MEGIBOW: Well, clearly if we look at GI bleeding this way, the timing is going to be pretty rigorous so we have to be very speciﬁc.
DOMINIK FLEISCHMANN, MD: But one issue is still that as long as we deal with parenchymal organs, the bowel wall, or even bleeding from the bowel, we want to have a lot of contrast injected over a pretty long period of time. So the timing window is pretty large. So that is why it works.
Today spatial resolution is so much better that one can probably identify the renal artery pretty well even on a noncontrast study. So that may be a reason why today, even on the late venous-phase image, you can see an artery. I think that it is due, at least in part, to the progress of scanner technology, which helps us to compensate for some factors that we probably do not always perform optimally.
I think there’s always a trade-off between making a study more complicated versus making it practical. If you make it complicated, you can probably save contrast, but it will be more error-prone. In routine practice, the error margins have to be somewhat larger, and, in fact, I think everyone will probably ﬁnd a protocol that is practical or useful for his or her own practice.
FISHMAN: Right. As you go from, let’s say, the academic institutions, where we all are, to private practice, these techniques—a little bit more contrast, a little bit more ﬁxed delay—tend to work a bit better in just getting the studies done correctly.
JAY P. HEIKEN, MD: Elliot, I’d like to make a comment. First of all, I agree with Dominik. But there’s another point that should be made. That is, with these very fast scanners, we have an opportunity when we’re doing CTA-speciﬁc examinations. We’re not really concerned about parenchymal enhancement; we’re just looking at the vessel and we have the opportunity to signiﬁcantly decrease the amount of contrast we use.
If we are going to decrease the amount of contrast we use, then timing is critical. The effect that this can have on cost is actually quite signiﬁcant. If you look at the difference between using 75 mL versus 125 mL of contrast across all of the CTA studies you do over the course of a year, it translates into a very large cost saving for the department.
So those are issues that we need to think about as well. But when we’re looking at parenchymal enhancement and GI bleeding, I agree with Dominik that it makes sense to give a larger volume of contrast material, and, in those cases, the timing is less signiﬁcant. But with our very rapid scanners, we can signiﬁcantly decrease the amount of contrast we use and save a lot of money.
FEDERLE: This forum gives a chance to get a thought to the manufacturers of contrast media. Because I think they’re going to need to adapt, as will the manufacturers of injectors, to this new era.
For the decade leading up to the 16-slice scanners, I’d thought that there were almost too many variables and options as far as injectors and volumes of contrast, etc. Frankly, a single-chamber injector and 125 mL of contrast served my needs for 99% of the studies that I did. Now it is definitely not true, and we are moving into an era in which we need more variability.
Alec alluded to the fact that if you are going to use a saline chaser, then we will need some alternative packaging for saline for hospitals between 10-mL bottles of bacteriostatic saline and 1-L bags. We are going to need something in between.
Similarly, there is also a need now for something other than a standard injector that holds preﬁlled syringes of 125 mL of contrast to be used in everybody.
FISHMAN: Thank you, Dr. Horton.