Gastrointestinal Imaging
Original Research
Preliminary Estimate of Triphasic CT Enterography Performance in Hemodynamically Stable Patients With Suspected Gastrointestinal Bleeding
Audio Available | 
Share
OBJECTIVE. The objective of our study was to retrospectively evaluate the performance of triphasic CT enterography and identify causes of false-negative CT results in hemodynamically stable patients with suspected gastrointestinal bleeding.
MATERIALS AND METHODS. A retrospective review of 48 patients (male–female ratio, 22:26) with suspected gastrointestinal bleeding (first-episode gastrointestinal bleed, n = 19; obscure gastrointestinal bleed, n = 29) who underwent triphasic CT enterography was performed. All patients had endoscopic, pathologic, or other imaging confirmation within 3 months of triphasic CT enterography. The sensitivity and specificity of triphasic CT enterography were calculated using pathology, endoscopy, or other imaging confirmation as the reference standard. Results were retrospectively reviewed to determine the cause of missed findings at triphasic CT enterography.
RESULTS. The overall sensitivity and specificity of triphasic CT enterography for detecting gastrointestinal bleeding was 33% (7/21) and 89% (24/27), respectively. Sensitivity and specificity were higher in first-episode gastrointestinal bleed cases (42% and 100%, respectively) than in obscure gastrointestinal bleed cases (22% and 85%). In the subset of patients undergoing capsule endoscopy (n = 17), only triphasic CT enterography identified two of three bleeding sources. Triphasic CT enterography did not identify six ulcers, four vascular malformations, two hemorrhoids, a duodenal mass, and a bleeding colonic diverticulum. The missed findings at triphasic CT enterography were attributed to being CT occult (n = 9), perception errors (n = 4), and technical errors (n = 1). If perception errors are excluded, the sensitivity of triphasic CT enterography increases to 52% (11/21).
CONCLUSION. Triphasic CT enterography can be a useful and complementary test in the evaluation of clinically stable patients with suspected gastrointestinal bleeding by identifying the bleeding source in one third to one half of patients. Because of the potential for perception errors, radiologists should familiarize themselves with the appearance of bleeding sources at CT enterography.
Keywords: CT enterography, gastrointestinal bleeding, small bowel, triphasic CT enterography
| Introduction |   |
|---|
Gastrointestinal bleeding has many different presentations including massive upper or lower gastrointestinal hemorrhage, which can be life-threatening; microscopic blood loss with iron-deficiency anemia or positive results on Hemoccult (Beckman Coulter) fecal blood test; and overt bleeding with melena, hematochezia, or hematemesis [1]. If a patient has persistent, recurrent, or intermittent bleeding for which the cause remains unclear after esophagogastroduodenoscopy (EGD), colonoscopy, and radiographic small-bowel evaluation, it is referred to as “obscure gastrointestinal bleeding” [2, 3]. Determining the cause of intermittent (not massive) bleeding is often diagnostically challenging, requiring extensive and expensive diagnostic workups.
CT enterography is a technique using neutral oral contrast material, IV contrast material, and typically single-phase scanning to optimize small-bowel imaging. The initial clinical indication for CT enterography was Crohn's disease because the low-density intraluminal contrast material improved detection of abnormally enhancing and inflamed small-bowel mucosa [4, 5] compared with CT examinations using positive oral contrast material. Eventually, CT enterography was shown to not only improve the detection of Crohn's disease, but also enhance visualization of small-bowel gastrointestinal hemorrhage, tumors, and vascular malformations [6, 7].
Recently, Huprich et al. [8] introduced a triphasic CT enterography technique with the same oral contrast preparation as routine CT enterography but using multiple phases (arterial, enteric, delayed) instead of a single scanning phase. This protocol was designed to optimize detection of small vascular lesions, which are a common cause of obscure gastrointestinal bleeds. In that study of 22 patients with an obscure gastrointestinal bleed, triphasic CT enterography identified a bleeding site in 45% of the patients. The utility of triphasic CT enterography in patients presenting with their first episode of gastrointestinal bleeding has not yet been evaluated to our knowledge.
The purpose of this study was to retrospectively evaluate the sensitivity and specificity of triphasic CT enterography in clinically stable patients with suspected gastrointestinal bleeding (first-episode or obscure gastrointestinal bleed) and identify causes of false-negative CT enterography results using original clinical reports.
| Materials and Methods | |
|---|
Approval for this retrospective HIPAA-compliant study was obtained from the institutional review board, along with a waiver of informed consent.
A retrospective review of the electronic medical records between June 2006 and July 2007 identified 48 patients (male–female ratio, 22:26) who underwent triphasic CT enterography for suspected gastrointestinal bleeding (hematochezia, melena, iron-deficiency anemia). The average age of the patients was 69 years (range, 41–92 years). Nineteen patients presented with a first-episode gastrointestinal bleed and 29 patients with an obscure gastrointestinal bleed. All patients had endoscopic, pathologic, or other imaging confirmation at our institution within 3 months of triphasic CT enterography.
All patients undergoing triphasic CT enterography ingested a total of 1,350 mL of neutral oral contrast material (0.1% weight/volume barium sulfate suspension [VoLumen, Bracco Diagnostics]) over a 45-minute period, drinking 450 mL of oral contrast material at 15-minute intervals. Patients were scanned on a 16- or 64-MDCT scanner (Sensation 16 or 64, Siemens Healthcare; LightSpeed VCT, GE Healthcare). Glucagon (0.5 mg) was given IV immediately before scanning. One hundred fifty milliliters of IV contrast material (iohexol [Omnipaque 350, Amersham Health]) was given at a rate of 4 mL/s. Multiphase scanning was performed using bolus-phase triggering during the arterial phase, with enteric and delayed phases acquired 20 and 90 seconds, respectively, after the arterial phase as previously described [8]. Images were acquired with a section thickness of 3 mm and a reconstruction interval of 1.5 mm. Coronal reconstructions were performed at 2-mm-thick sections every 2 mm. An average effective dose of 63.4 mSv was calculated for patients undergoing triphasic CT enterography. The axial and coronal images were interpreted on a four-monitor PACS workstation (Centricity, GE Healthcare).
Reference standard data were obtained by an abdominal imaging fellow who did not participate in triphasic CT enterography image review or interpretation. A total of 115 correlative endoscopic, surgical, and imaging tests were performed in these 48 patients (EGD, n = 37; colonoscopy, n = 37; capsule endoscopy, n = 17; tagged RBC scan, n = 10; double-balloon enteroscopy, n = 7; angiography, n = 4; surgery, n = 2; push enteroscopy, n = 1). Most patients (45/48) had endoscopic correlation (upper only, n = 10; lower only, n = 5; both upper and lower, n = 30). Two patients had surgical and endoscopic correlation. One patient had angiographic correlation only.
The reference standard was pathology if available (8/48). If pathology was not available, then endoscopy (i.e., EGD, colonoscopy, double-balloon enteroscopy, or capsule endoscopy) was considered the reference standard (39/48). Conventional and balloon enteroscopy were considered more reliable than capsule endoscopy. If endoscopy and pathology were not available, other radiologic tests such as conventional angiography (1/48) were used to confirm or refute the triphasic CT enterography findings. Triphasic CT enterography results were not considered in creating the reference standard, which was based only on non–triphasic CT enterography results.
Triphasic CT enterography was performed after endoscopy in 31 patients and before endoscopy at our institution in 17 patients. The reasons triphasic CT enterography was performed before endoscopy included negative findings on an endoscopic or imaging workup performed at an outside institution (n = 10), comorbidities made emergent endoscopy difficult (n = 3), faster access to CT compared with other tests (n = 2), anticoagulation precluded endoscopy (n = 1), and need to rule out metastases in a patient with a gastrointestinal bleed and known malignancy before endoscopy (n = 1).
The original radiology interpretation was compared with the reference standard (endoscopy, n = 39; pathology, n = 8; angiography, n = 1) to determine the sensitivity and specificity of triphasic CT enterography. Triphasic CT enterography images for all patients were then reviewed by consensus of an abdominal imaging fellow and a board-certified abdominal radiologist with 7 years' experience to confirm that any positive triphasic CT enterography findings correlated to the reference standard, determine which phase or phases the triphasic CT enterography finding was visualized, and determine the cause of false-negative and false-positive triphasic CT enterography studies.
Causes of false-negative triphasic CT enterography studies were grouped in one of three categories: CT occult (finding not seen on CT even in retrospect), technical errors (errors during scan acquisition such as motion artifact or poor IV contrast bolus), and perception errors (finding identified in retrospect).
Thirty-seven of the 48 patients had follow-up information available in the electronic medical record at least 2 weeks after triphasic CT enterography that was used to determine if the gastrointestinal bleed resolved or recurred. The follow-up interval was an average of 10 months after the triphasic CT enterography examination (median, 9 months; range, 2 weeks–2.5 years).
| Results | |
|---|
The overall sensitivity and specificity of triphasic CT enterography were 33% and 89%, respectively (Table 1). The sensitivity and specificity of triphasic CT enterography was higher in patients presenting with a first episode of gastrointestinal bleeding (42% and 100%, respectively) than patients with an obscure gastrointestinal bleed (22% and 85%).
Overall, triphasic CT enterography correctly diagnosed seven of 21 positive cases for a sensitivity of 33% (Table 2). Triphasic CT enterography successfully identified hemorrhagic (actively bleeding colonic diverticula; Figs. 1A, 1B, 1C, 1D, and 1E), inflammatory (three duodenal ulcers, jejunal celiac sprue, active sigmoid ulcerative colitis; Fig. 2), and polypoid (jejunal Peutz-Jeghers polyp, Figs. 3A, 3B, 3C, and 3D) diseases. The reference standards used were pathology (n = 3), endoscopy (n = 3), and angiography (n = 1).
Fourteen patients had findings not detected by triphasic CT enterography, which are listed in Table 3. The missed findings included ulcers (n = 6), vascular malformations (n = 4), hemorrhoids (n = 2), a duodenal mass, and a colonic diverticular bleed. These missed findings at triphasic CT enterography were attributed to being CT occult (n = 9) (Figs. 4A, 4B, and 4C), perception errors (n = 4) (Figs. 5A, 5B, and 5C), and technical errors (n = 1, poor bolus timing). The reference standard in these cases were conventional endoscopy (n = 8), pathology and surgery (n = 5), and capsule endoscopy (n = 1).
Therefore, of the 21 positive cases by the reference standard, 11 of 21 (52%) were identifiable at triphasic CT enterography: the seven detected by readers plus the four missed due to perception errors. Of the 11 findings, most (9/11) were visible in all three phases. Two triphasic CT enterography findings were detected on the arterial and enteric phases only (gastric ulcer, hemorrhoids).
Of the 21 positive cases, most (17/21) were within reach of EGD or colonoscopy. In eight of 17 patients, triphasic CT enterography was performed after positive endoscopic findings to further assess endoscopic abnormalities (e.g., nonspecific findings; n = 5) or to exclude an additional small-bowel source of bleeding (n = 3). The remaining nine of 17 patients underwent triphasic CT enterography before repeat endoscopy at our institution to exclude a small-bowel source of bleeding after prior outside negative endoscopic or imaging workup (n = 6) or because they had comorbidities that made emergent endoscopy difficult (n = 3).
Clinical follow-up was available for 18 of 21 patients. Only two patients had recurrent symptoms. These consisted of iron-deficiency anemia after cauterization of a gastric vascular malformation and hematochezia after medical treatment of ulcerative colitis. The remaining 16 patients had resolution of their symptoms after interventional treatment (endoscopy, embolization, or surgery, n = 7), medical therapy (n = 7), or dietary changes (n = 2 [i.e., gluten-free diet or fiber supplements]).
Triphasic CT enterography correctly identified 24 of 27 negative cases for a specificity of 89%. False-positive triphasic CT enterography findings included one each of suspected gastric hemorrhage (Figs. 6A, 6B, and 6C), sigmoid vascular malformations, and hemorrhoids (Fig. 7)—all with negative follow-up endoscopy. The reference standards used to establish a negative diagnosis in all patients are listed in Table 4.
Clinical follow-up was available for 19 of 27 patients. Nine of 19 patients had recurrent bleeding but no definite diagnosis has been established to date. The remaining 10 patients had resolution of their symptoms after surgery (n = 1, colectomy for diverticulosis) or cessation of nonsteroidal anti-inflammatory drugs, aspirin, or clopidogrel bisulfate (Plavix, Bristol-Myers Squibb) (n = 5); symptoms spontaneously resolved without treatment (n = 4).
Capsule endoscopy was performed in 17 of 48 patients. In this subset, triphasic CT enterography had a better sensitivity and specificity than capsule endoscopy (CT enterography, 66% and 86%, respectively; capsule endoscopy, 33% and 78%). Of the three positive findings, two were identified only by triphasic CT enterography (celiac sprue and a duodenal ulcer), and one was identified only by capsule endoscopy (jejunal vascular malformation). The duodenal ulcer in a gastric bypass patient would have been impossible to detect by capsule endoscopy because the capsule did not traverse the bypassed duodenum (Figs. 8A, 8B, and 8C). The reference standards for these positive findings were push enteroscopy (celiac sprue), double-balloon enteroscopy (duodenal ulcer), and capsule endoscopy (jejunal vascular malformation).
 View larger version (115K) | Fig. 1A —Triphasic CT enterography true-positive finding in 91-year-old woman with active diverticular bleed. Axial arterial (A), enteric (B), and delayed (C) phase images show progressive extravasation of IV contrast material (circle) from diverticulum in lumen of descending colon. |
 View larger version (119K) | Fig. 1B —Triphasic CT enterography true-positive finding in 91-year-old woman with active diverticular bleed. Axial arterial (A), enteric (B), and delayed (C) phase images show progressive extravasation of IV contrast material (circle) from diverticulum in lumen of descending colon. |
 View larger version (118K) | Fig. 1C —Triphasic CT enterography true-positive finding in 91-year-old woman with active diverticular bleed. Axial arterial (A), enteric (B), and delayed (C) phase images show progressive extravasation of IV contrast material (circle) from diverticulum in lumen of descending colon. |
 View larger version (175K) | Fig. 1D —Triphasic CT enterography true-positive finding in 91-year-old woman with active diverticular bleed. Angiography image confirms finding (circle) shown in A–C. |
 View larger version (191K) | Fig. 1E —Triphasic CT enterography true-positive finding in 91-year-old woman with active diverticular bleed. Image obtained after embolization. |
Capsule endoscopy had three false-positive examinations for a specificity of 79% (11/14). The false-positives included two jejunal lipomas, neither of which was seen at subsequent double-balloon enteroscopy or triphasic CT enterography, and one colon arteriovenous malformation (AVM), which was not seen at subsequent colonoscopy or triphasic CT enterography.
Triphasic CT enterography had two false-positive examinations for a specificity of 86% (12/14). The false-positive triphasic CT enterography cases consisted of a suspected gastric bleed and colon AVM with subsequent negative capsule endoscopy in both, EGD and colonoscopy.
| Discussion | |
|---|
The use of CT in the diagnostic workup of patients with suspected gastrointestinal bleeding continues to grow because of improvements in scanning technology and favorable results in several recent articles [8–12]. Recent studies, however, vary with the type of scanning performed—that is, single phase [9], dual phase [10, 13, 14], or triple phase [8]. In addition, the types of patients vary from those with acute massive gastrointestinal hemorrhage [9, 13] to patients with an obscure gastrointestinal bleed [8, 14]. The sensitivity of CT for diagnosing the source of gastrointestinal bleeding has been shown to be higher in patients with active hemorrhage (91–92%) than in those with an obscure gastrointestinal bleed (45–47%) [8, 9, 13, 14].
Our study focused on the use of a specific CT technique, triphasic CT enterography, that was initially described by Huprich et al. [8] for the detection of an obscure gastrointestinal bleed. In clinical practice, however, we found that triphasic CT enterography was requested by physicians and radiologists alike for multiple indications relating to suspected gastrointestinal bleeding and not only in patients with prior negative endoscopies or imaging workup. Therefore, for this study, we retrospectively evaluated the use of triphasic CT enterography in any patient with suspected gastrointestinal bleeding. Our results indicate that triphasic CT enterography can identify the cause of gastrointestinal bleeds in one third of these patients with few false-positives (sensitivity = 33%, specificity = 89%). In addition, if perception errors were corrected, the potential sensitivity of triphasic CT enterography increases to 52%. These results are in the range of the results reported by Huprich et al.; they found triphasic CT enterography to show a bleeding site in 45% of patients presenting with obscure gastrointestinal bleeding [8].
 View larger version (154K) | Fig. 2 —Triphasic CT enterography true-positive finding in 78-year-old woman with active sigmoid ulcerative colitis. Sagittal arterial phase triphasic CT enterography image shows active sigmoid ulcerative colitis (arrow). Finding was confirmed at colonoscopy. |
Triphasic CT enterography is a complementary test that can detect causes of gastrointestinal bleeding that are missed by endoscopy. In this study, four patients had abnormalities detected at triphasic CT enterography that were not detected at standard endoscopy or capsule endoscopy (celiac sprue, ulcer in a surgically bypassed duodenum, a jejunal polyp, and a bleeding colonic diverticulum). Similarly, Huprich et al. [8] identified three findings missed by capsule endoscopy (cecal angiodysplasia, a jejunal vascular lesion, and a jejunal gastrointestinal stromal tumor). In that study, triphasic CT enterography identified more lesions with fewer false-positives than capsule endoscopy, also showing that triphasic CT enterography complements endoscopic assessment.
Although triphasic CT enterography was mainly developed to evaluate the small bowel, it can also detect abnormalities in the colon and stomach, largely because of improved distention with low-density oral contrast material in these areas that can make lesions more conspicuous. In this study, gastric ulcers, bleeding colonic diverticula, and ulcerative colitis were seen at triphasic CT enterography. Huprich et al. [8] also identified angiectasias in the stomach and colon. Oftentimes, however, the resolution of superficial ulcers and small angiectasias still remains occult at triphasic CT enterography as previously discussed. The ability, however, of triphasic CT enterography to potentially show abnormalities anywhere in the gastrointestinal tract makes it an attractive diagnostic option, particularly for evaluating patients at first clinical presentation of gastrointestinal bleeding or in the emergency setting. Whether this examination should be performed before or after other diagnostic studies remains to be determined.
Based on our results, only one patient with an actively bleeding colon diverticula avoided endoscopy based on triphasic CT enterography results. In other patients, however, triphasic CT enterography was useful to localize or further evaluate abnormalities identified on other tests and was an efficient option for the emergency setting because it did not require additional personnel to be called in at night as with nuclear medicine scanning, conventional angiography, and endoscopy. The implementation of triphasic CT enterography in the clinical algorithm is quite variable based on our findings.
 View larger version (202K) | Fig. 3A —Triphasic CT enterography true-positive finding in 73-year-old woman with jejunal Peutz-Jeghers polyp. Polyp was beyond reach of standard endoscopy. Arterial (A), enteric (B), and delayed (C) phase images show hyperenhancing polyp in proximal jejunum (circles). |
 View larger version (209K) | Fig. 3B —Triphasic CT enterography true-positive finding in 73-year-old woman with jejunal Peutz-Jeghers polyp. Polyp was beyond reach of standard endoscopy. Arterial (A), enteric (B), and delayed (C) phase images show hyperenhancing polyp in proximal jejunum (circles). |
 View larger version (211K) | Fig. 3C —Triphasic CT enterography true-positive finding in 73-year-old woman with jejunal Peutz-Jeghers polyp. Polyp was beyond reach of standard endoscopy. Arterial (A), enteric (B), and delayed (C) phase images show hyperenhancing polyp in proximal jejunum (circles). |
 View larger version (173K) | Fig. 3D —Triphasic CT enterography true-positive finding in 73-year-old woman with jejunal Peutz-Jeghers polyp. Polyp was beyond reach of standard endoscopy. Double-balloon enteroscopy image shows polyp (arrow). |
Similar to previous studies, the most commonly missed findings at triphasic CT enterography in this study were ulcers and vascular lesions. In the only other published study to date of triphasic CT enterography, four of five triphasic CT enterography false-negatives were attributed to the same findings [8]. Superficial ulcers that are not associated with wall thickening or increased enhancement remain challenging to detect at CT even with the increased bowel distention and multiphase imaging used with a triphasic technique. Vascular malformations are detected more commonly with triphasic CT enterography according to the results of the study by Huprich et al. [8] in which four of six vascular ectasias or angiodysplasias (two with active bleeding) were detected. In that study, however, only one of four vascular malformations was identifiable at triphasic CT enterography, and it was detected only in retrospect. None of the lesions was actively bleeding in our study, which may account for the lower detectability. Any detection of vascular lesions at CT is an improvement compared with an earlier study of routine CT in which 0 of 11 lesions were identified [15].
 View larger version (148K) | Fig. 4A —Angiodysplasias (arrows) that were occult at triphasic CT enterography. 44-year-old woman with 5-mm gastric arteriovenous malformation at esophagogastroduodenoscopy. |
 View larger version (130K) | Fig. 4B —Angiodysplasias (arrows) that were occult at triphasic CT enterography. 69-year-old man with jejunal angiodysplasia at capsule endoscopy (not measured). |
 View larger version (152K) | Fig. 4C —Angiodysplasias (arrows) that were occult at triphasic CT enterography. 73-year-old woman with vascular malformation in ascending colon at colonoscopy (not measured). |
 View larger version (136K) | Fig. 5A —Triphasic CT enterography false-negative finding in 79-year-old woman with gastric ulcer. False-negative was due to perception error. Esophagogastroduodenoscopy image shows gastric ulcer (circle). |
 View larger version (209K) | Fig. 5B —Triphasic CT enterography false-negative finding in 79-year-old woman with gastric ulcer. False-negative was due to perception error. Axial enteric image shows ulcer crater in gastric antrum (arrow) with surrounding wall thickening. |
 View larger version (186K) | Fig. 5C —Triphasic CT enterography false-negative finding in 79-year-old woman with gastric ulcer. False-negative was due to perception error. Sagittal multiplanar reformation image in arterial phase shows discontinuity of enhancing gastric mucosa (arrow). Triphasic CT enterography finding was missed prospectively. |
False-positive CT enterography results can be minimized by familiarizing oneself with potential pitfalls. In this study, precipitated barium from the oral contrast suspension was confused with gastrointestinal bleeding (Figs. 6A, 6B, and 6C). Overcalling vascular malformations can be minimized by requiring the abnormality be present on two or more phases. Familiarity with the inherent increase in attenuation of closely apposed normal mucosal folds is also necessary to avoid overcalling rectal hemorrhoids or mucosal inflammation (Fig. 7). This pitfall can be minimized with adequate bowel distention.
 View larger version (142K) | Fig. 6A —Triphasic CT enterography false-positive finding in 64-year-old woman with questioned gastric hemorrhage at triphasic CT enterography with negative esophagogastroduodenoscopy. Axial arterial (A), enteric (B), and delayed (C) phase images show persistent high density within gastric antrum and fundus (arrows) incorrectly interpreted as possible gastric hemorrhage. This finding is likely ingested contrast material, possibly concentrated amount of barium present in oral contrast material. |
 View larger version (149K) | Fig. 6B —Triphasic CT enterography false-positive finding in 64-year-old woman with questioned gastric hemorrhage at triphasic CT enterography with negative esophagogastroduodenoscopy. Axial arterial (A), enteric (B), and delayed (C) phase images show persistent high density within gastric antrum and fundus (arrows) incorrectly interpreted as possible gastric hemorrhage. This finding is likely ingested contrast material, possibly concentrated amount of barium present in oral contrast material. |
 View larger version (143K) | Fig. 6C —Triphasic CT enterography false-positive finding in 64-year-old woman with questioned gastric hemorrhage at triphasic CT enterography with negative esophagogastroduodenoscopy. Axial arterial (A), enteric (B), and delayed (C) phase images show persistent high density within gastric antrum and fundus (arrows) incorrectly interpreted as possible gastric hemorrhage. This finding is likely ingested contrast material, possibly concentrated amount of barium present in oral contrast material. |
 View larger version (134K) | Fig. 7 —Triphasic CT enterography false-positive finding (circle) in 65-year-old woman with apposed folds in rectum simulating hemorrhoids (negative colonoscopy). |
 View larger version (190K) | Fig. 8A —Triphasic CT enterography true-positive finding not detected by capsule endoscopy in 77-year-old man with actively bleeding ulcer in bypassed portion of duodenum after remote gastric bypass. Finding was confirmed at double-balloon enteroscopy. Arterial (A), enteric (B), and delayed (C) phase images show actively bleeding ulcer (circle) in bypassed portion of duodenum. Note progressive accumulation of extravasated contrast material. |
 View larger version (197K) | Fig. 8B —Triphasic CT enterography true-positive finding not detected by capsule endoscopy in 77-year-old man with actively bleeding ulcer in bypassed portion of duodenum after remote gastric bypass. Finding was confirmed at double-balloon enteroscopy. Arterial (A), enteric (B), and delayed (C) phase images show actively bleeding ulcer (circle) in bypassed portion of duodenum. Note progressive accumulation of extravasated contrast material. |
 View larger version (185K) | Fig. 8C —Triphasic CT enterography true-positive finding not detected by capsule endoscopy in 77-year-old man with actively bleeding ulcer in bypassed portion of duodenum after remote gastric bypass. Finding was confirmed at double-balloon enteroscopy. Arterial (A), enteric (B), and delayed (C) phase images show actively bleeding ulcer (circle) in bypassed portion of duodenum. Note progressive accumulation of extravasated contrast material. |
The main disadvantage of the triphasic CT enterography technique is its increased radiation exposure, up to 63 mSv per examination in this study and 59 mSv in a prior study [8]. These doses are increased in part because of the preliminary nature of these studies. Although this dose is high, it is important to note that mortality related to radiation-induced cancer is decreased considerably for patients older than 40 years, the age of most patients who present with suspected gastrointestinal bleeding (average age in this study, 69 years). Many patients with suspected gastrointestinal bleeding often undergo multiple imaging and endoscopic examinations and the radiation dose risk in this population may be offset by the benefit of minimizing invasive testing. However, the dose of this examination could probably be reduced significantly using a lower-dose technique, eliminating phases, or both. In the unique setting of acute massive gastrointestinal bleeding, performing a single angiographic phase, which has been shown to result in a high degree of sensitivity and specificity [9], may be sufficient. Our results in clinically stable patients with suspected gastrointestinal bleeding indicate that the delayed phase may not be required for diagnosis. In another study of patients with an obscure gastrointestinal bleed, although the delayed phase best showed the findings in three patients, the finding could still be identified on other phases as well [8]. Other ways to minimize radiation exposure besides eliminating phases is to lower the peak kilovoltage or tube current. Preliminary results have shown that 33–66% radiation dose reductions at single-phase CT enterography can detect Crohn's disease at similar diagnostic rates as standard-dose CT enterography in spite of increased image noise (Siddiki H et al., presented at the 2007 annual meeting of the Radiological Society of North America). Even more aggressive dose reductions may be possible with image noise reduction algorithms [16].
Limitations of this retrospective review include selection bias, a heterogeneous clinical population, and a heterogeneous reference standard. Clinical follow-up was not available in 11 patients. Because of the intermittent nature of gastrointestinal bleeding, it is possible that some of the true-negative examinations were incorrectly classified, although this risk was minimized with the clinical follow-up performed in most patients. Because of the retrospective nature of the study, interpretations were not blinded and it is possible that radiologists and endoscopists were aware of prior diagnostic results, which could have falsely elevated the sensitivity of both examinations. The sensitivity of triphasic CT enterography, however, does not appear obviously affected because even though it was performed after positive endoscopy in 14 cases, only four cases had corresponding positive triphasic CT enterography results.
In spite of the limitations described, this study found that triphasic CT enterography has the potential to identify the source of gastrointestinal bleeding in up to half of the patients if reader errors are eliminated. It has a higher performance in patients presenting with a first episode of gastrointestinal bleeding compared with patients with prior negative endoscopic workup. Educating one-self on the appearance of common lesions responsible for gastrointestinal bleeding, an awareness of CT enterography pitfalls, and knowledge of the limitations of this technique will improve diagnostic results.
Address correspondence to A. K. Hara ([email protected]).
We acknowledge the statistical expertise and help of Qing Wu.
| References | |
|---|
