The role of multidetector CT in the management of acute gastrointestinal bleeding
Shuvro H. Roy-Choudhury1
and Anthony Nicholson2
1Heart of England NHS Foundation Trust, Birmingham, UK
2United Leeds Teaching Hospital NHS Trust, Leeds, UK
Address for correspondence:
Dr Shuvro H. Roy-Choudhury MB, FRCS, FRCR
Consultant Abdominal Radiologist
Heart of England NHS Foundation Trust (Teaching)
Bordesley Green East, Birmingham, B9 5SS
Tel: +44 (0)121-424-3282 Fax: +44 (0)121-766-6919
There is no universal consensus on the pathways used in the diagnosis and management of acute gastrointestinal (GI) bleeding. Endoscopy forms the mainstay of diagnosis and management. This is supplemented by nuclear medicine studies in diagnosis, and surgery together with interventional embolisation, in management. Recent advances in multidetector computed tomography (MDCT) have made it a suitable, quick, non-invasive and easily available tool to rapidly locate and evaluate relevant parameters in patients with acute GI bleeding. Experimental studies have shown high sensitivity of contrast-enhanced MDCT in detecting active bleeding. In this article, we will discuss the evolving role of this new technique and compare it with other diagnostic tools. We will look at the technical aspects, discuss pitfalls and briefly review the current literature. A possible algorithm incorporating MDCT is also suggested.
Acute gastrointestinal (GI) haemorrhage is a relatively common cause for hospital admission, particularly in the elderly, and is associated with significant morbidity and mortality. In the USA, there are 300,000 annual admissions related to this condition and mortality rates between 2% and 10% are reported [1,2]. Where symptoms are significant, it can be a life-threatening emergency. Endoscopy forms the mainstay of diagnosis and treatment for GI haemorrhage, particularly if the source is above the ligament of Treitz. Imaging or radiological techniques are useful in the following clinical scenarios:
The traditional therapeutic options available in patients with GI haemorrhage are either endoscopic or surgical. In either case, an accurate localisation of the source of bleeding is critical, as surgery without localisation carries a high risk of mortality . A planned segmental resection after appropriate localisation has been proven to be better than blind colectomy and mortality rates can be reduced from 25% to 10% and rates of re-bleeding can be reduced from 38% to 15% .
The established imaging modalities used for localisation for acute LGI bleed include nuclear medicine studies like technetium99m-sulphur colloid or technetium99m-labelled red blood cells (RBC). Selective catheter angiography remains the gold standard for the diagnosis of active bleeding. However, this is invasive, requires specialised expertise, is not freely available out of hours and is often negative when bleeding is intermittent. In the recent past, the use of multidetector computed tomography (MDCT) angiography has increasingly been reported to detect, characterise and localise active bleeding from the gastrointestinal tract [3,6-8]. In this article, we will review the emerging role of this modality and consider its application in relation to other available diagnostic modalities. We will also study the technique and pitfalls of MDCT angiography and illustrate its applications. The primary focus of this article will be the application of MDCT in LGI haemorrhage, although similar principles apply for detection of UGI haemorrhage. We will also briefly look into the emerging role of MDCT in detecting low-grade obscure bleeding.
Lower GI haemorrhage
- Colon cancer and other neoplasms including polyps
- Inflammatory bowel disease
- Bowel ischaemia
- HIV and related diseases
- Meckel's diverticulum
- Anorectal diseases like haemorrhoids, fissures or rectal ulcers
Diagnostic modalities used for characterisation and localisation of bleeding source
Colonoscopy or sigmoidoscopy
This is usually the first diagnostic technique used  However, in the face of torrential bleeding, adequate visualisation is often impossible due to the presence of blood. This investigation may have to be performed in an unprepared colon given the relative emergency. In the face of such bleeding, the definitive yield from colonoscopy can be as low as 13%  Nevertheless, there is evidence that careful early colonoscopy (often repeated) can have high success in diagnosing and treating the source of bleeding. In a period of  years in which 85 patients were treated and 126 colonoscopies carried out, the cause of bleeding was correctly identified in 97% of patients. Control of haemorrhage was achieved in 17 out of 27 patients who were actively bleeding. As is often the case, bleeding stopped spontaneously in the other 58 patients in this study .
Nuclear medicine studies
These studies are highly sensitive (up to 90%) and can detect bleeding rates as low as 0.1 ml/min. It is usually thought that these tests have lower specificities than other techniques, although specificity as high as 95% has been reported . The biggest problem with nuclear medicine studies is poor localisation due to peristaltic movement of the tracer which can happen in up to 14% of cases. Therefore, segmental resection based on this investigation alone is not recommended. Usually, the use of technetium99m-RBC is preferred over technetium99m-sulphur colloid as the latter may obscure bleeding from the colonic flexures due to reticuloendothelial system uptake in the liver and spleen and because it has a shorter half-life. Technetium99m-RBC can be imaged over the following 24 hours, thereby increasing the potential for detecting intermittent haemorrhage. Using this technique, higher sensitivity than MDCT has been reported in a series of 31 cases. Further additional bleeding sites were identified, thereby improving the overall bleeding detection rate 
Digital subtraction angiography
The sensitivity of digital subtraction angiography (DSA) in detecting active GI bleeding ranges from 42-92% and the specificity is nearly 100% [14-18] In landmark experiments, Baum and Nusbaum demonstrated that the lower threshold for detection of bleeding with angiography was 0.5-1 ml/min  With modern DSA, this sensitivity has perhaps increased further. Angiography has the advantage of delivering a high iodine concentration via selective injection into a bleeding artery. However, its contrast resolution is poor, movement or peristalsis makes interpretation difficult, the technique is time consuming and invasive, and it needs more specialised expertise.
The main problem with angiography lies in the fact that it is an invasive test and that studies are often negative because of the intermittent nature of the pathology. Catheterisation of main vessels may miss small bleeding sources from the vasa recta and experience is needed in interpretation of angiographic images. Such expertise is limited, particularly ‘out of hours'.
Once the bleeding source is identified, a superselective embolisation can be performed with coils or Gelfoam  Vasopressin can be used as a temporary measure to stabilise a patient for surgery if embolisation is considered technically demanding (vasopressin is more commonly used for upper GI haemorrhage). In a large series of 420 patients with GI bleeding, 14% of cases received endoscopic treatment, 16% received medical treatment, 17% underwent an operation and 67% received interventional radiology with embolisation.9 It is important to remember that targeted super-selective studies may be necessary.
Where available, capsule endoscopy has been shown to be superior to CT or DSA in the detection of obscure haemorrhage. In a series of 28 patients, capsule endoscopy was successful in identifying the source in 72%, while CT was only successful in identification in 24% of patients and DSA identified 56% of patients . Sometimes these tests are complementary. The biggest problem of capsule endoscopy is the fact that the transit time is around 8 hours and that data interpretation takes a further 2 hours. The role of capsule endoscopy in the diagnosis of obscure small bowel haemorrhage is discussed further below.
The demonstration of active bleeding on CT was first reported in 1989 . Since then, this has been established as a critical sign in the diagnosis and triage of patients with abdominal and pelvic trauma that predicts the need for active intervention. When this sign is seen, it is assumed that the bleeding is occurring at a brisk rate. However, there is anecdotal and published evidence that has shown active bleeding being identified by CT but not by catheter angiography even when performed close to each other . High speed, narrow collimation MDCT allows large volume coverage, has fewer motion and respiratory artefacts and can be accurately timed to acquire data in arterial or venous phases. The excellent Z-axis resolution, near isotropic voxel resolution and ability to localise pathology accurately should make MDCT an excellent modality to localise GI bleeding. In a swine colon phantom, Kuhle et al. showed that helical CT can detect bleeding down to 0.3 ml/min (lower than the sensitivity reported with angiography) . In vivo bleeding can be intermittent even from minute to minute and can stop due to thrombosis, spasm or tamponade - therefore clinical comparisons may be invalid. In a comparative experimental study using a physiological cardiopulmonary flow phantom, we have shown that intravenous contrast enhanced MDCT has a lower threshold (0.35 ml/min) than selective catheter angiography (0.96 ml/min) to detect bleeding, unless the catheter position is highly super-selective and close to the bleeding site 
The rationale for using MDCT as a first-line test extends beyond its superior sensitivity over catheter angiography in the experimental setting:
(i) Accurate localisation may determine whether a
conservative, endoscopic, surgical or interventional
approach is most suitable, e.g. a diverticular bleed
may be treated conservatively, an anorectal
cause may be treated surgically while a bleeding
gastroduodenal artery pseudoaneurysm may be
managed by embolisation
(ii) 70-80% of GI bleeds stop spontaneously . In this setting, a non-invasive test like MDCT angiography is preferable, with catheter angiography only reserved for those cases where MDCT is positive
(iii) CT provides important aetiological and prognostic information that helps plan further management, even when active bleeding is not demonstrated
(iv) Accurate localisation reduces the time to control the source of bleeding, whether at surgery or during embolisation. It prevents unnecessary vessel cannulations and reduces both the contrast load and the radiation dose during embolisation
(v) It helps to plan the interventional approach, e.g. an occluded coeliac trunk may suggest a surgical approach to a bleeding gastroduodenal aneurysm
(vi) Practical issues such as ‘round-the-clock' availability, reproducibility, repeatability and fewer requirements for specialised training are important considerations
Many centres have incorporated MDCT as a part of the first-line diagnostic test to investigate GI bleeding. The clinical efficiency of this approach has yet to be reported in the published literature although several encouraging case series have been presented. A summary of the larger series [3,6-8,26-29] reporting on the use of MDCT in this setting is presented in the accompanying table (Table 1).
Clinical scenarios for performing acute MDCT angiography in our practice are as follows:
- Torrential LGI bleed where an UGI (endoscopy or nasogastric aspiration) and anorectal cause has been excluded
- Drop in haemoglobin by 4 g/dl
- Patient with GI bleeding who is hypotensive (systolic BP below 100 mm Hg) or tachycardic in spite of fluid replacement. Shock index >1
- Bleeding continuing for 3 days
- More than 4 units transfusion requirement per day
- 500 ml of blood lost per day (equates to loss of approximately 0.3 ml/min)
- Re-bleed within 7 days
- Failed endoscopic management of upper GI bleed
MDCT angiography is usually performed without any oral preparation, particularly if the patient is unstable. Where there is more time, water or other negative oral contrast can be used. Two large-bore cannulas are inserted: one to continue resuscitation while the other is used for the CT scan being performed. A non-contrast scan of the abdomen and pelvis is performed first. Following this, an arterial phase scan is performed (with a delay of 20-30 seconds depending upon the detector configuration). A late portal venous phase (90 seconds) and/or a delayed phase at 3- 5 minute scan is then performed. The latter was used in our initial experience, although we have found the delayed scan rarely contributes further information. Scanning parameters are summarised in the accompanying table (Table 2).
Signs of active bleeding
The cardinal sign on MDCT indicating active intraluminal bleeding is the same as in other abdominal pelvic organs. The presence of high attenuation material within the bowel lumen which was not present on the initial non-contrast scan is pathognomonic of acute GI haemorrhage (Figure 1). This can take the form of pools or jets. The extravasated contrast is usually within 10 Hounsfield units of a large artery in the same MDCT slice, although accurate measurement of numbers is less practical in the setting of GI haemorrhage.
In the non-contrast scan, the presence of clotted or thrombosed blood (Figure 1) may give an indication as to the location of bleeding, but it should be remembered that peristalsis can move clots up, or, more commonly, downwards. Occasionally, due to continuing pooling, the leaked contrast may be of higher density, particularly in the 90-second scan. Contrast can sometimes pool in a bleeding diverticulum.
Other associated signs may indicate the source of haemorrhage in a patient in whom bleeding has stopped. These include presence of a mass or tumour, a vascular nidus or large draining vein indicating angiodysplasia (Figure 2), hyperenhancing or thickened bowel wall indicating enteritis or early ischaemia and perienteric stranding in inflammatory conditions.
Pitfalls of CT angiography to detect GI haemorrhage
There are a number of pitfalls in the detection of GI haemorrhages using CT angiography:
(i) The predominant problem with any diagnostic
method in localising a GI bleed is that the GI
haemorrhage can often settle spontaneously (in up
to 80% of cases). Even severe GI bleeding is often
intermittent, or even minute to minute  Therefore,
the timing of the MDCT study is critical - the ideal
is for the patient to be scanned when borderline
hypotensive or tachycardic or when they require
fluid to maintain a stable blood pressure. MDCT
can and often needs to be repeated. Alternatively, a
red cell study can be performed once the initial CT
study is negative (see Algorithm)
(ii) Active bleeding can be very subtle (Figure 4) and therefore it is imperative that radiologists are scrupulous in trying to identify tiny areas of hyperattenuation within the bowel lumen, not present on the non-contrast scan
(iii) The lack of a non-contrast scan for comparison can often lead to areas of calcification or luminal content mimicking active bleeding.
(iv) A venous bleeding source or minute bleeding from a large surface area (for example diffuse duodenitis) may be difficult to diagnose although mucosal or mural signs on CT may help
(v) In an undistended bowel, intense mucosal enhancement in the arterial phase (normal) can lead to false positive diagnosis of active bleeding.
A potential pathway for current management of an acute GI bleed is shown in the accompanying flow chart. Given its potential superior sensitivity and high specificity, initial screening with the more rapid and available MDCT may represent a reasonable approach for early evaluation of LGI bleed . The management of intermittent lowgrade bleeding is more difficult to determine and can take several routes incorporating RBC scan, colonoscopy or MDCT depending on local preference.
- Peptic ulcer disease
- Oesophageal varices
- Haemorrhagic oesophagitis, gastritis or duodenitis
- Gastric duodenal or pancreatic neoplasms
- Aortoenteric fistulae
- Pseudoaneurysms usually following pancreatitis or iatrogenic causes
- Mallory-Weiss tear
Obscure low-volume GI bleed
This is defined as chronic low-volume bleeding from the GI tract, the cause of which is not immediately apparent following thorough clinical and standard endoscopic assessment. Clinically, it is often manifested as iron deficiency anaemia with a positive test for faecal occult blood. The source of this type of bleeding is often the small bowel. Imaging techniques that can be used in this setting include CT or MR enterography/ enteroclysis, barium small bowel enema or Tc99m-RBC studies. Capsule endoscopy (CE) has reported to have a higher diagnostic yield than the imaging studies. In a series of 100 patients with active bleeding, CE was reported to have sensitivity and specificity of 89% and 95% respectively  A more recent endoscopic technique - double balloon enteroscopy - may have a higher diagnostic yield than CE or the imaging techniques mentioned above. A catheter angiogram may also be used with some success.
In truly refractory cases, consideration should be given to intra-arterial CT mesenteric angiography. A catheter is placed in the suspected artery and an arterial phase CT scan is performed after pump injection. As early as 1984, Stanley et al. suggested the usefulness of postangiography CT in detecting occult gastrointestinal haemorrhage missed during plain angiography  Ettore et al. were able to establish the site of haemorrhage in 72% of patients with obscure gastrointestinal bleeding by performing CT after a mid-aortic injection of contrast medium, a technique that was shown to be more sensitive than conventional angiography  In our physiological phantom, this technique could demonstrate bleeding rates of 0.05 ml/min or lower 
Multidetector CT is rapidly emerging as a sensitive, specific and easily reproducible test in the diagnosis of acute lower GI bleed. This is fortuitous as the condition has relatively high mortality if mismanaged. Decisions in any particular patient should be collaborative and depend on local expertise. Multidisciplinary management involving the gastroenterologist, radiologist, the surgeon and the nursing staff is absolutely critical in triaging, diagnosing and localising acute GI bleed and its management.
- Acute GI bleeding should be managed in a multidisciplinary setting. Although endoscopy forms the mainstay in management, diagnostic and interventional radiology is playing an increasingly prominent role
- MDCT angiography is an extremely sensitive modality to diagnose active GI bleeding. It can be successfully used to localise bleeding site preoperatively/prior to angiography and thereby can reduce mortality and re-bleeding
- The investigation should be performed when the patient is bleeding. Provided there is safe ongoing resuscitation available, rarely is a patient not stable enough to undergo a CT study
- A non-contrast and an arterial phase scan form the basis of this study. The presence of high attenuation material within the bowel lumen, of the same density as vessels, that was not present on the non-contrast scan is diagnostic of active extravasation
- Extensive use of maximum intensity projections and multiplanar reconstruction images is critical. Findings can be subtle
- There are a number of pitfalls to be wary of
- Lee EW, Laberge JM. Differential diagnosis of gastrointestinal bleeding. Tech Vasc Interv Radiol 2005;7:112-22.
- Green BT, Rockey DC, Portwood G, et al. Urgent colonoscopy for evaluation and management of acute lower gastrointestinal hemorrhage: a randomized controlled trial. Am J Gastroenterol 2005;100:2395-402.
- Laing CJ, Tobias T, Rosenblum DI, et al. Acute gastrointestinal bleeding: emerging role of multidetector CT angiography and review of current imaging techniques. Radiographics 2007;27:1055-70.
- Setya V, Singer JA, Minken SL. Subtotal colectomy as a last resort for unrelenting, unlocalized, lower gastrointestinal hemorrhage: experience with 12 cases. Am Surg 1992;58:295-9.
- Heider J, Layer G, Textor HJ, et al. [Does angiographic localization of bleeding affect the management and mortality in gastrointestinal hemorrhage of unknown origin?] Aktuelle Radiol 1998;8:71-5.
- Yoon W, Jeong YY, Shin SS, et al. Acute massive gastrointestinal bleeding: detection and localization with arterial phase multi-detector row helical CT. Radiology 2006;239:160-7.
- Tew K, Davies RP, Jadun CK, et al. MDCT of acute lower gastrointestinal bleeding. AJR Am J Roentgenol 2004;182:427-30.
- Rajan R, Dhar P, Praseedom RK, et al. Role of contrast CT in acute lower gastrointestinal bleeding. Dig Surg 2004;21:293-6.
- Brackman MR, Gushchin VV, Smith L, et al. Acute lower gastroenteric bleeding retrospective analysis (the ALGEBRA study): an analysis of the triage, management and outcomes of patients with acute lower gastrointestinal bleeding. Am Surg 2003;69:145-9.
- Angtuaco TL, Reddy SK, Drapkin S, et al. The utility of urgent colonoscopy in the evaluation of acute lower gastrointestinal bleeding: a 2-year experience from a single center. Am J Gastroenterol 2001;96:1782-5.
- Chaudhry V, Hyser MJ, Gracias VH, et al. Colonoscopy: the initial test for acute lower gastrointestinal bleeding. Am Surg 1998;64:723-8.
- Zuckier LS. Acute gastrointestinal bleeding. Semin Nucl Med 2003;33:297-311.
- Diehl SJ, Ko HS, Dominguez E, et al. [Negative endoscopy and MSCT findings in patients with acute lower gastrointestinal hemorrhage. Value of (99m)Tc erythrocyte scintigraphy]. Radiologe 2007;47:64-70.
- Rahn NH 3rd, Tishler JM, Han SY, et al. Diagnostic and interventional angiography in acute gastrointestinal hemorrhage. Radiology 1982;143:361-6.
- Laine L. Acute and chronic gastrointestinal bleeding. In: Feldman M, Scharschmidt BF, Sleisenger MH, eds. Gastrointestinal and liver disease. 6th edition. Philadelphia, PA, USA, Saunders, 1997: 198-219.
- Kuo WT, Lee DE, Saad WE, et al. Superselective microcoil embolization for the treatment of lower gastrointestinal hemorrhage. J Vasc Interv Radiol 2003;14:1503-9.
- Funaki B, Kostelic JK, Lorenz J, et al. Superselective microcoil embolization of colonic hemorrhage. AJR Am J Roentgenol 2001;177:829-36.
- Defreyne L, Vanlangenhove P, De Vos M, et al. Embolization as a first approach with endoscopically unmanageable acute nonvariceal gastrointestinal hemorrhage. Radiology 2001;218:739-48.
- Baum S, Nusbaum M, Clearfield HR, et al. Angiography in the diagnosis of gastrointestinal bleeding. Arch Intern Med 1967;119:16-24.
- Nicholson AA, Ettles DF, Hartley JE, et al. Transcatheter coil embolotherapy: a safe and effective option for major colonic haemorrhage. Gut 1998;43:79-84.
- Saperas E, Dot J, Videla S, et al. Capsule endoscopy versus computed tomographic or standard angiography for the diagnosis of obscure gastrointestinal bleeding. Am J Gastroenterol 2007;102:731-7.
- Sivit CJ, Peclet MH, Taylor GA. Life-threatening intraperitoneal bleeding: demonstration with CT. Radiology 1989;171:430.
- Singer AA. Value of CT in localizing site of gastrointestinal hemorrhage following negative angiography. Abdom Imaging 1995;20:31-2.
- Kuhle WG, Sheiman RG. Detection of active colonic hemorrhage with use of helical CT: findings in a swine model. Radiology 2003;228:743-52.
- Roy-Choudhury SH, Gallacher DJ, Pilmer J, et al. Relative threshold of detection of active arterial bleeding: in vitro comparison of MDCT and digital subtraction angiography. AJR Am J Roentgenol 2007;189:W238-46.
- Junquera F, Quiroga S, Saperas E, et al. Accuracy of helical computed tomographic angiography for the diagnosis of colonic angiodysplasia. Gastroenterology 2000;119:293-9.
- Ko HS, Tesdal K, Dominguez E, et al. [Localization of bleeding using 4-row detector-CT in patients with clinical signs of acute gastrointestinal hemorrhage]. Rofo 2005;177:1649-54.
- Ernst O, Bulois P, Saint-Drenant S, et al. Helical CT in acute lower gastrointestinal bleeding. Eur Radiol 2003;13:114-7.
- Yamaguchi T, Yoshikawa K. Enhanced CT for initial localization of active lower gastrointestinal bleeding. Abdom Imaging 2003;28:634-6.
- Sos TA et al. Intermittent bleeding from minute to minute in acute massive gastrointestinal hemorrhage: arteriographic demonstration. AJR Am J Roentgenol. 1978;131:1015-7.
- Duchesne J, Jacome T, Serou M, et al. CT-angiography for the detection of a lower gastrointestinal bleeding source. Am Surg 2005;71:392-7.
- Pennazio M, Santucci R, Rondonotti E, et al. Outcome of patients with obscure gastrointestinal bleeding after capsule endoscopy: report of 100 consecutive cases. Gastroenterology. 2004;126:643-53.
- Stanley JH, Vujic I, Gobien RP, et al. Computed tomography as an angiographic aid in evaluation of intraabdominal hemorrhage. J Comput Tomogr 1984;8:301-6.
- Ettorre GC, Francioso G, Garribba AP, et al. Helical CT angiography in gastrointestinal bleeding of obscure origin. AJR Am J Roentgenol 1997;168:727-31.
06-2008 BUY1163660/JB3272/MB002753CMC Int'l English 16th Edition