Biphasic CT with Mesenteric CT Angiography in the Evaluation of Acute Mesenteric Ischemia: Initial Experience1

MATERIALS AND METHODS

Patients

From November 2000 to July 2002, all patients clinically suspected of having AMI at two tertiary care hospitals who were examined with CT were prospectively studied. Prior to CT, all patients were evaluated by a specialist (either a general or vascular surgeon or an internist) who believed there was clinical concern for AMI. This conclusion was based on the finding of abdominal pain with at least one of the following symptoms: pain out of proportion to clinical findings; biochemical evidence of ischemia, such as elevated lactate levels or unexplained metabolic acidosis; or patient risk factors for ischemia, such as previous mesenteric ischemia or symptoms of chronic ischemia, severe vascular disease, atrial fibrillation without therapeutic anticoagulation, low flow (a history of hypotension or vasopressor therapy), or hypercoagulable states. Eight patients with a contraindication to the intravenous contrast agent, such as risk factors for a severe allergic reaction or renal insufficiency, were excluded from the study. Two other patients were eligible for the study, but they were ultimately not included, as the arterial phase of the CT protocol was inadvertently not performed at the time of the CT examination due to operator error. After informed consent was obtained, 62 patients were examined, including 21 men (age range, 32–87 years; mean, 60 years) and 41 women (age range, 24–95 years; mean, 65 years). All patients were selected and examined with the same protocol and were later divided into a study group composed of patients with proven AMI and a control group composed of patients with final diagnoses that did not include AMI, for the purposes of statistical analysis. Patients were observed for any evidence of a contrast agent reaction (change in vital signs with administration of contrast agent, skin rash, hives, pruritis, or change in renal function after the study that was not readily attributable to another cause). Institutional Health Research Ethics Board approval and Medical Records Department approval were obtained for the study.

Imaging and Image Evaluation

All patients were examined with a four-row multi–detector row CT scanner (Lightspeed; GE Medical Systems, Milwaukee, Wis). Water was initially given as an oral contrast agent (500–750 mL, as tolerated). All patients received 140 mL of nonionic intravenous contrast agent (Omnipaque 300; Nycomed, Princeton, NJ) through an 18-gauge antecubital intravenous line at a rate of 4 mL/sec. Initially, a CT angiographic phase acquisition was performed 25 seconds after contrast agent injection began; scanning was performed with 1.25-mm collimation, 7.50-mm pitch, and high-speed mode. Images in this phase were obtained from above the level of the celiac axis to below the level of the aortic bifurcation into the common iliac arteries. After this, a conventional portal phase acquisition was obtained 60–70 seconds after injection of the contrast agent began; scanning was performed with 5-mm collimation, 11.25-mm pitch, and high-quality mode. Images in this phase were obtained from above the level of the diaphragm to below the level of the symphysis pubis. Images were retrospectively reconstructed as needed, with standard reconstructions of contiguous 1.25-mm transverse images in the angiographic phase and 5-mm transverse images that overlapped every 2.5 mm in the portal venous phase.

All images were evaluated at a workstation (Advantage; GE Medical Systems) in the standard transverse plane. All CT angiographic images were also evaluated with real-time multiplanar reformatted images, maximum intensity projection images, and volume-rendered images. Multiplanar and maximum intensity projection reformatted images were reviewed for the portal phase data sets as necessary. Image reconstruction was performed by radiologists (I.D.C.K., M.A.K., H.M.G.) in all patients. All images were reviewed independently by two radiologists (I.D.C.K., M.A.K., or H.M.G.), with three initial disagreements resolved by consensus review. The disagreements consisted of two instances in which it was believed that one radiologist did not measure bowel wall thickness in the area of most pronounced thickening, and one instance in which it was believed that one radiologist overlooked and did not record bowel dilatation. The first radiologist to review the scan after it was obtained dictated the official report for the patient’s chart. The urgency of the examinations and logistical limitations made it impossible for both radiologists to review the study prior to the issuing of an official report. The final impressions from the official radiology reports were recorded, and an impression of findings that were consistent with, concerning for, or diagnostic of mesenteric ischemia was considered positive for the diagnosis of AMI. The second radiologist reviewed the images after the patient left the department, but was blinded to the first radiologist’s interpretation. The prospective data collected at that time from both radiologists were used for all statistical analyses.

Arterial opacification in the CT angiographic phase was judged to be satisfactory by each radiologist if there was visualization through at least the second-order vessels of both the celiac axis and the SMA both in transverse and in reconstructed images. CT scans were evaluated for evidence of bowel wall thickening (defined as a wall thickness of more than 3 mm in noncollapsed small or large bowel running perpendicular to the transverse plane), mucosal enhancement (defined as focal enhancement out of proportion to the rest of the bowel in noncollapsed small or large bowel), focal lack of bowel wall enhancement, bowel dilatation (defined as a small-bowel diameter of more than 2.5 cm or colon diameter of more than 8 cm), bowel obstruction (defined as bowel dilatation with evidence of a transition point and collapsed bowel distally), mesenteric stranding, ascites, solid organ infarction, free intraperitoneal air, pneumatosis intestinalis, superior mesenteric or portal venous gas, or superior mesenteric or portal venous thrombosis. The bowel wall was measured with electronic calipers, and the mean of three individual measurements from the region of greatest thickness in either the small or the large bowel was recorded. Additionally, CT angiograms were assessed for evidence of arterial stenosis (defined as luminal narrowing of >50%) or occlusion in the celiac, superior mesenteric, and inferior mesenteric arterial systems. Occlusions that appeared to be secondary to an embolus with an intravascular filling defect were also noted. Findings at CT angiography prompted requests for conventional angiography to help confirm the initial findings in three patients, and angiographic findings were correlated.

Chart Review and Pathologic Reports

All patient charts and pathologic reports were reviewed to determine the final diagnosis (I.D.C.K.). All patients with AMI had either surgical proof (n = 14), autopsy proof (n = 5), or both (n = 6), except for one patient in whom the diagnosis was made on the basis of clinical and serial CT findings. Patients without AMI had an alternative diagnosis confirmed either at surgery (n = 8) or on the basis of appropriate laboratory values and clinical findings (n = 22), or they had no diagnosis and recovered spontaneously without treatment (n = 6). Charts were also reviewed for documented evidence that information provided by the CT angiogram affected ultimate patient care, either by influencing the decision to perform surgery or by influencing the planned surgical procedure. The patient risk factors for AMI and the treatment of each patient were recorded, along with the status of each patient throughout his or her hospital stay and eventual outcome.

Statistical Analysis

In patients with AMI, the sensitivity and specificity of each individual CT sign were calculated by using the 36 patients who did not have intestinal ischemia as a control group. CT criteria that optimized sensitivity and specificity for the diagnosis of AMI were also developed retrospectively by assessment of patients with possible combinations of CT findings through queries made from fields in a database (Access; Microsoft, Redmond, Wash) (no information that could be used to directly identify patients was entered into this database). The Fisher exact test was used to compare patient distribution by sex between the AMI and the control groups. An unpaired t test was used to compare mean bowel wall thickness for the ischemia group to that of the control group, and to compare mean ages of the two groups. A P value of less than .05 was considered to indicate a statistically significant difference.

RESULTS

All 62 patients had satisfactory opacification of the abdominal aorta and mesenteric vessels on the CT angiogram as judged by all authors. This occurred even in patients with proximal vessel occlusion due to filling of distal vessels via collateral vessels. No patients experienced a reaction to the contrast agent.

Final Diagnoses

Overall, AMI was diagnosed in 26 of the 62 patients. Ischemia was recorded as a diagnosis in the report of the initial interpreting radiologist for all 26 patients. The one patient without surgical or pathologic proof of diagnosis demonstrated superior mesenteric and portal venous thrombosis on CT scans, with small-bowel dilatation, mucosal enhancement, and wall thickening that appeared consistent with intestinal ischemia, but this patient was treated conservatively. Serial CT scans showed an improved appearance of the bowel as the venous thrombus decreased in size after anticoagulation therapy, and the diagnosis of AMI was made on this basis.

CT was used to make a definitive alternative diagnosis that was later proven with either pathologic examination or appropriate laboratory values and clinical findings in 21 (58%) of the remaining 36 patients. Two of these patients had pancreatitis, two had gastritis, and one each had splenic and portal vein thrombosis, isolated intrahepatic portal vein thrombosis, diverticulitis, ileus, choledocholithiasis, obstructing ureteric calculus, ovarian carcinoma, appendicitis, perforated gastric ulcer (Fig 1), retroperitoneal hemorrhage, abdominal wall hematoma, hemorrhagic ovarian cyst, pneumonia, lumbar disk herniation, mechanical small-bowel obstruction related to adhesions, intraperitoneal abscess related to Crohn disease, or Clostridium difficile colitis.

Among the remaining 15 patients, the findings of the CT examinations were normal in six. These patients recovered spontaneously without any specific treatment and, therefore, were presumed not to have AMI. The findings of the CT examinations were nonspecific in the nine other patients. Mesenteric ischemia was recorded as a diagnosis on the report of the initial interpreting radiologist in four of these patients, whose illnesses were later diagnosed as Crohn disease (n = 2), neutropenic enterocolitis (n = 1), and infectious enteritis (n = 1). The diagnoses were confirmed with surgery that was prompted by the report in three patients (the patient with infectious enteritis did not undergo surgery). Spontaneous bacterial peritonitis (n = 2), hepatitis (n = 2), and celiac disease (n = 1) were eventually diagnosed in the five remaining patients.

Demographics

The group of 26 patients in whom AMI was diagnosed consisted of nine men and 17 women with an age range of 25–95 years (mean, 66 years). The control group of 36 patients in whom AMI was not diagnosed consisted of 12 men and 24 women with an age range of 24–95 years (mean, 60 years). There was no statistically significant difference in the sex or age distribution of the two groups.

CT Findings

The number of patients in each group with each of the individual CT findings, along with the individual sensitivity and specificity of each finding, is given in Table 1 and listed in order of declining specificity. The individual findings of pneumatosis intestinalis, isolated SMA occlusion, celiac and inferior mesenteric artery (IMA) occlusion with distal SMA disease (branch vessel occlusions or narrowing), arterial embolism, or venous gas were each 100% specific for AMI but only 73% sensitive. Alternatively, it was determined that the presence of bowel wall thickening in addition to either focal lack of bowel wall enhancement, solid organ infarction, or venous thrombosis was 50% sensitive and 94% specific. If the diagnosis of AMI was made on the basis of either of these criteria, an optimal calculated sensitivity of 96% and specificity of 94% can theoretically be achieved in this patient population. The one patient that would have a false-negative diagnosis with these criteria had an episode of hypotension in the intensive care unit, and CT demonstrated small- and large-bowel wall thickening, mucosal enhancement, bowel dilatation, mesenteric stranding, and ascites. A colonoscopic biopsy showed ischemic changes, but the patient recovered with supportive therapy and the findings of a follow-up CT examination were normal. The two patients who would have false-positive diagnoses stemming from our use of these criteria were ultimately found to have C difficile pancolitis and sigmoid diverticulitis. The review of the initial interpreting radiologists’ official reports revealed a sensitivity of 100% and specificity of 89% for diagnosing AMI initially.

Twenty-two (85%) of the 26 patients with a diagnosis of ischemia had CT signs of bowel wall thickening compared with 10 (28%) of the 36 patients without a diagnosis of ischemia. Mean wall thickness was 0.8 cm for the patients with ischemia and 0.8 cm for the patients in the control group, with no difference between these two means.

CT Angiographic Findings

The CT angiographic phase images showed arterial abnormalities in eight patients with proven AMI, and only two patients (one with SMA occlusion and one with arterial embolism) had abnormalities that were readily identifiable on the portal phase image. All eight abnormalities were evident on angiographic phase transverse images, along with both maximum intensity projection and volume-rendered reconstructions. Two patients demonstrated isolated occlusion of the SMA with a patent IMA and celiac axis. Both patients underwent massive resections of gangrenous bowel but ultimately died. One patient had an SMA occlusion with stenoses of both the IMA and celiac axis that was confirmed with catheter angiography (Fig 2) and subsequently underwent successful surgical revascularization. Two patients had proximal occlusions of both the celiac axis and IMA with evidence of distal disease (small vessel occlusions or narrowing) in the SMA distribution. Findings were confirmed with catheter angiography in both patients (Fig 3). Both of these patients had no other CT findings to suggest the diagnosis of AMI, and the bowel appeared normal. In both patients, however, an ischemic bowel that required resection was found intraoperatively. Three patients had CT evidence of arterial emboli (Fig 4) that was confirmed at surgery or autopsy, including two of seven patients who had atrial fibrillation and either were not treated with anticoagulant therapy or their condition was subtherapeutic with anticoagulants. Two of these three patients underwent surgical revascularization rather than bowel resection. In total, five of six patients had vascular abnormalities that were appreciated on the CT angiographic images but not on the portal venous phase images. In these six patients, the findings on the CT angiogram clearly influenced patient care, either by suggesting AMI when there were no other consistent CT findings (n = 2) or by identifying a vascular event that could be treated with revascularization rather than simple bowel resection (n = 3). Overall, the mesenteric CT angiogram altered treatment in five (19%) of 26 patients with ischemia and in five (8%) of 62 patients overall.

Four patients with AMI had evidence of venous thrombosis. Three of these patients had thrombus depicted in both the portal and the superior mesenteric veins, and all three showed evidence of associated bowel wall thickening and bowel dilatation. Two of these three patients also showed areas of mucosal enhancement. The fourth patient demonstrated intrahepatic portal venous thrombosis, and while she also showed areas of mucosal enhancement and bowel dilatation, the bowel wall was not thickened and there were areas of small bowel pneumatosis intestinalis. This patient also had intraperitoneal free air at CT and was found to have extensive areas of necrotic small bowel at surgery. She ultimately died, and the thrombosis was confirmed at autopsy.

Patient Outcomes

Overall, 11 (42%) of the 26 patients with bowel ischemia died while in the hospital, including four men and seven women. There was no statistically significant difference in mortality by sex. The outcome and CT findings in patients with AMI are shown in Table 2.

DISCUSSION

Diagnosis of AMI remains a challenge. Mortality from this condition remains high despite medical advances, and a short time to diagnosis is one factor that has been shown repeatedly to improve survival (1,2). Performing exploratory laparotomy for any patient suspected of having AMI would subject a number of people to unnecessary surgery, although it may well benefit those who in fact have ischemia. Catheter angiography, while previously considered to be the standard for making an imaging diagnosis of AMI, is not available in many centers and takes time to arrange and perform (1). Multi–detector row CT offers an attractive diagnostic alternative.

To our knowledge, the study by Taourel et al (7) is the most comprehensive evaluation of the usefulness of CT in the diagnosis of AMI to date, and by using retrospectively applied optimal diagnostic criteria, the authors calculated a theoretical sensitivity of 64% and specificity of 92%. Our study differs from theirs in several respects. To our knowledge, ours is the first study to examine the effectiveness of multi–detector row CT technology in the diagnosis of AMI and the only study to use a prospective design with prospective control patients. Also, we examined the added effectiveness of mesenteric CT angiography in addition to the catheter portal venous phase transverse images.

We retrospectively applied optimal diagnostic criteria of any one of pneumatosis intestinalis, venous gas, SMA occlusion, celiac and IMA arterial occlusion with distal SMA disease, or arterial embolism as positive for AMI or, alternatively, bowel wall thickening combined with any one finding of focal lack of bowel wall enhancement, solid organ infarction, or venous thrombosis. By using these criteria, a sensitivity of 96% and a specificity of 94% can be achieved, percentages that are much higher than previously published values (1,7). This is partly due to the added effectiveness of mesenteric CT angiography. Not only are many CT angiographic findings very specific for the diagnosis, but in two patients these were the only abnormal CT findings and they could not be appreciated on catheter portal phase images. Furthermore, CT angiography influenced care in 19% of the patients with ischemia by either indicating the diagnosis or altering the surgical treatment through involvement of a vascular surgeon and subsequent revascularization procedures. Although the official reports of the initial interpreting radiologist demonstrated a high sensitivity (100%) at some expense of specificity (89%), there was variation in the strength of the diagnostic impression (ranging from “concerning for” to “diagnostic of” AMI) that, along with the specificity, could be improved with the use of the above diagnostic criteria.

The rigid use of these optimized diagnostic criteria is, however, potentially problematic. The two patients who would have had false-positive diagnoses had their abnormal bowel in a distribution (sigmoid colon in a region of extensive diverticulosis for the patient with diverticulitis and pancolonic for the patient with C difficile colitis) that would cast some doubt on the diagnosis of acute ischemia. One patient would have a false-negative diagnosis if these criteria were strictly adhered to, and, as described previously, this patient had a history that was strongly suggestive of nonocclusive ischemia. Many of the more specific CT findings such as venous thrombosis and arterial embolism or occlusion would not be expected with this cause unless the patient had underlying vascular occlusive disease. Hallmark angiographic findings such as narrowing of multiple branch vessels of the SMA, small vessel spasm, or impaired filling of intramural vessels would be more difficult to appreciate on CT scans relative to catheter angiography, which also offers the option of therapy with papaverine (13,14). A disadvantage of CT angiography is that it depicts the vessels at a single point in time, and the temporal changes in vascular filling that are seen on catheter angiographic images cannot be appreciated. A patient may demonstrate poor opacification of distal SMA branches at CT for a number of reasons, including generalized slow blood flow (as seen, for example, in cardiac disease) or diffuse spasm related to nonocclusive ischemia. Slow filling related to decreased cardiac output would clearly be much easier to appreciate with catheter angiography. The CT diagnosis in nonocclusive ischemia is much more dependent on coexistent changes in the appearance of the bowel and knowledge of a consistent clinical context. It should be considered that patients are rarely brought to the CT scanner while acutely hypotensive, and, as a result, findings of mucosal enhancement from reperfusion rather than lack of enhancement may be seen. Unfortunately, this gives a more nonspecific CT appearance, and correlation to clinical information becomes much more important. Also, since our study defined mucosal enhancement or lack of enhancement as a focal change relative to the rest of a patient’s bowel, a patient with globally increased or decreased enhancement of both the small and the large bowel could conceivably receive a false-negative diagnosis.

It is estimated that approximately one-third of cases of AMI are caused by arterial embolism, one-third are caused by acute arterial thrombosis, between two- and three-tenths are caused by nonocclusive ischemia, and the remainder are caused by venous thrombosis (14,15). If this distribution holds true for our study population, then some cases of arterial thrombosis and embolism were not directly depicted on our CT angiograms. Seven patients with atrial fibrillation who did not undergo anticoagulant therapy or whose levels of anticoagulants were subtherapeutic had AMI. While an embolic source was strongly suspected in these patients, arterial emboli were seen in only two of these patients. It is possible that some emboli are too small to be within the resolution of present-day CT angiography. Fortunately, due to other findings (Fig 4) such as focal lack of bowel wall enhancement, the diagnosis could be made in the rest of the patients.

Venous thrombosis is an important, but less common, cause of AMI. The majority of patients with venous occlusion showed bowel wall thickening, dilatation, and mucosal enhancement, similar to previously published findings (7,16). It is noted that AMI was not ultimately diagnosed in two patients with portal vein thrombosis, and their symptoms resolved with anticoagulant therapy alone. Possible explanations include that this was a chronic finding and that the acute abdominal pain was unrelated, that some degree of ischemia was present but not severe enough to cause abnormalities to appear on CT scans, or that collateral venous drainage was sufficient to prevent AMI as a result of this insult. Conversely, we also had one patient with portal venous thrombosis with frank bowel necrosis and pneumatosis, a rare but not unheard of finding with this condition (17). This patient was the only (25%) of four patients with venous thrombosis as a cause of AMI who died in the hospital. In contrast, four (50%) of eight patients with arterial abnormalities on the CT angiogram died in the hospital, including four (67%) of six with either proximal SMA occlusion or visualized arterial embolism. Other findings that were particularly more common in patients who died while in the hospital include pneumatosis intestinalis, venous gas, free intraperitoneal air, and solid organ infarction. This is in keeping with prior research and makes intuitive sense since the first three of these signs are seen in bowel necrosis and perforation and the fourth is a manifestation of arterial occlusion (which itself is associated with higher mortality both in our study and in the literature) (7).

We encountered no vascular abnormality on the mesenteric CT angiograms that could not be appreciated on both maximum intensity projection and volume-rendered reconstructions. The maximum intensity projection reconstructions are easier and faster to obtain, but they are subject to artifacts caused by vascular calcification, venous visualization (Fig 3), and the potential for less accurate estimation of stenoses with increasing slab thickness. Volume-rendered images, which require more time and labor, avoid these artifacts and give superior three-dimensional spatial relationships of vessels (19). We found that correlation with transverse and multiplanar reformatted images was very helpful in the confirmation of any findings.

While the rate of mortality from AMI in our study is substantially lower (42%) than previously reported, this may in part be related to patient selection bias (1,2). While it is standard practice to obtain a CT scan of patients suspected of having AMI at our institutions, some patients with unsuspected AMI may have undergone exploratory laparotomy for an acute abdomen, at which point AMI may have been found intraoperatively (although we could find no evidence of any such cases with a search of the health records database). As previously stated, all patients in the study had suspected AMI documented by a specialist. It is important to note that this did not include patients with a clinical diagnosis of mechanical bowel obstruction who may have had strangulation and ischemia as a complication. Three patients with proven primary AMI in our study did have CT evidence of obstruction with the transition point at the area of ischemia, which is thought to be related to a functional obstruction from a lack of normal peristalsis (18).

Patient selection bias may also influence the sensitivity and specificity data of our study; however, we feel that this accurately reflects real-world clinical practice, in which only patients with clinically suspected ischemia would undergo a biphasic CT examination with CT angiography.

We conclude that biphasic CT with mesenteric CT angiography is a valuable method of diagnosing AMI, with a sensitivity of 96% and a specificity of 94% calculated by using outlined diagnostic criteria. In addition, CT angiograms provide valuable information that is not obtained with conventional portal phase images alone and alters management in 19% of the patients with ischemia. As we are not aware of other studies of this nature, further investigations would be valuable to confirm these findings and diagnostic criteria.