Department of Imaging, Division of Vascular Imaging & Intervention, Massachusetts General Hospital, Boston, MA
The authors declare that they have nothing to disclose.
Reprints: Sanjeeva P. Kalva, MD, Department of Imaging, Division of Vascular Imaging & Intervention, Massachusetts General Hospital, Boston, MA 02114 (e-mail: email@example.com).
Acute gastrointestinal (GI) bleeding represents a common medical emergency with severe potential clinical consequences. With an annual incidence of 40 to 150 episodes per 100,000 persons for upper GI bleeding, and 20 to 27 episodes per 100,000 persons for lower GI bleeding in the United States,1 this diagnosis is seen fairly frequently across the spectrum of all clinical practices. At the same time, the morbidity and mortality for this diagnosis remain high, with almost 300,000 hospital admissions every year,2 and death rates approaching 40% for patients with hemodynamic instability.3 Given its high clinical impact, early diagnosis and treatment remain a goal of all clinicians who need to manage this disease.
Unfortunately, current diagnostic algorithms vary widely from institution to institution and from clinician to clinician. Imaging modalities remain the mainstay of the diagnostic approach. They include endoscopy, videocapsule, radionuclide imaging, catheter angiography, and multidetector computed tomography (MDCT) imaging. For years, endoscopy has been the initial standard approach to patients who present with acute GI bleeding. Endoscopy represents a safe and effective method to diagnose and often treat GI bleeding in a controlled environment. The sensitivity and specificity of endoscopy for upper GI bleeding are very high, approaching 98% and 100%, respectively.2 However, while esophagogastroduodenoscopy clearly is a first-line tool for diagnosis and treatment of upper GI bleeding, colonoscopy remains challenging for lower GI bleeding, with the ability to identify a definite bleeding source in only 13% of patients.4 The necessity of adequate bowel preparation precludes the ability of colonoscopy to be utilized effectively in urgent clinical situations. Overall, conventional endoscopy suffers from an inability to directly visualize the distal duodenum and small bowel. Videocapsule represents a promising tool to overcome these anatomic areas, but it is inappropriate in the emergent setting. Traditionally, radionuclide imaging with technetium 99m-labeled red blood cells has been utilized as an adjunct tool to evaluate lower GI bleeding, and has been shown to detect active bleeding at a rate of 0.2 mL/min.5 The noninvasive nature of this modality and lack of necessary patient preparation are advantageous for many clinicians. However, the availability of nuclear medicine may be limited at some institutions in an overnight emergency room setting. In addition, the lack of accurate anatomic detail obtained by nuclear imaging may make subsequent intervention a challenge. Catheter angiography has proven to be a useful tool in both the diagnosis and treatment of GI bleeding. Experimental studies in dogs have established its ability to diagnose bleeding rates from 0.5 to 1.0 mL/s.6 In addition, catheter angiography offers the possibility of immediate treatment through embolization of active bleeding sites or through infusion of vasopressin, thereby saving the cost and time of subsequent interventions. However, angiography is often operator dependent, and represents an invasive tool that may not be appropriate for every patient.
In recent years, MDCT has emerged as a promising technology to evaluate GI bleeding. The modality's ubiquitous nature, ease of use, and rapid results favor its use in any emergent situation. In addition, today's high-speed, narrow collimation multidetector technology allows a large coverage area with minimal motion artifacts, with the ability to capture both arterial and venous phase with ease. Swine models have shown that MDCT can capture bleeding rates as low as 0.4 mL/min.7 Despite the theoretical promise of MDCT in a challenging clinical situation, published reports have been limited as to its true clinical effectiveness. Hara et al8 retrospectively reviewed 48 patients with suspected GI hemorrhage, and found that MDCT-detected GI bleeding with an overall sensitivity and specificity of 33% and 89%, respectively. Interestingly, in subgroup analysis, first time GI bleeding had an improved sensitivity and specificity of 42% and 100%, with a sharp decline in effectiveness in obscure GI bleeding, with sensitivity and specificity rates of 22% and 85%. In a recent article, Huprich et al9 prospectively compared multiphase computed tomography (CT) enterography with videocapsule, demonstrating a significantly greater sensitivity of CT (88% vs. 38%), highlighting its utility in diagnosing small bowel tumors in obscure GI bleeding. Finally, Wu et al10 conducted a meta-analysis reviewing 9 studies with a total of 198 patients. They found that MDCT had a pooled sensitivity and specificity of 89% and 85%, respectively, for detecting GI bleeding.
In this issue of Journal of Clinical Gastroenterology, Sun et al11 bring new evidence to bear on the high utility of MDCT for the detection and localization of active upper and lower GI bleeding. In a clinical prospective study utilizing MDCT (16 slice, 64 slice or dual source CT), the authors evaluated 113 patients with clinical symptoms of GI bleeding. In defining positive CT angiography (CTA) broadly with either active extravasation, focal or segmental bowel mucosal enhancement, detection of a vascular malformation, abnormally enhancing polyp/diverticulum, or tumor, the authors moved beyond the limits of what CTA has been traditionally used for. As a gold standard the authors used either digital subtraction angiography (DSA), endoscopy, surgery, or pathology reports. The CTA was positive in almost 71% of their patients, with no false-positive MDCTA examinations. Interestingly, of the 22 positive MDCTA cases defined by active contrast material extravasation, all patients had clinically severe GI bleeding. 29% of the patients had negative studies, but 13 (39%) of these patients were considered false negative. Of note, of these 13 patients, there were 6 patients who required further intervention after additional studies demonstrated positive findings. The authors also compared MDCTA and DSA studies, demonstrating 86% concordance in 28 patients who had both studies. Four patients, who had both studies, failed to have concordance and were initially negative on DSA after a positive MDCTA, were subsequently found to be true positive on additional studies. Overall, the sensitivity of MDCTA was 86%, the specificity 100%, the positive predictive value 100%, the negative predictive value 61%, and the accuracy 89%.
The authors present much needed data in an area of controversy as to how to approach patients with GI hemorrhage. Sun et al11 develop a novel approach to the definition of a positive MDCTA, with a broad categorization of a positive study. In this manner, the authors take advantage of the technical advances of CT scanners with their increase in detector number and speed, thereby allowing a more accurate diagnosis of a wider variety of etiologies for GI bleeding. We believe that the authors' practice reflects current technology platforms that are being adopted by the majority of institutions in the United States. By including findings beyond simple contrast extravasation, the authors are to be commended in their ability to capture sources of bleeding that are treatable, but may not normally be seen on standard reference studies such as DSA or technetium 99m red blood cell studies.
Sun et al11 also propose a novel diagnostic algorithm for patients with GI hemorrhage. In suggesting MDCTA as the first-line diagnostic modality, the authors believe that the positive predictive value and negative predictive value are sufficiently high that patients could be safely triaged based on either a positive or negative CTA study. However, we believe that while the authors demonstrate MDCTA to have a high specificity and positive predictive value, the low negative predictive value of 61% limits the modality's use as a first-line screening method for detection and localization of GI bleeding. Looking at the subgroup analysis, we note that in patients with clinically severe GI bleeding, the negative predictive value decreases to 50%. In addition, the authors report that almost half of the 39% of patients with false-negative MDCTA required intervention, suggesting that the risk of missing a critical diagnosis that could be treated remains too high for a first-line screening modality.
In short, clinicians caring for patients who present with GI bleeding have a wide variety of diagnostic modalities to choose from to accurately perform effective triage. Sun et al11 present novel data demonstrating the utility of MDCT as a rapid, easy method to accurately identify a potentially treatable etiology for GI bleeding. In particular, Sun et al11 broaden the scope of MDCTA by identifying a wider range of causes of GI bleeding as compared with previously published studies. However, while their data clearly demonstrate a desirable sensitivity and specificity for the modality, the authors do not convincingly show that a negative MDCTA can truly exclude some underlying etiology for patients with GI bleeding. At this point, we recommend more research to truly define the best first-line modality for patients presenting with GI bleeding.
1. Manning-Dimmitt LL, Dimmitt SG, Wilson GR. Diagnosis of gastrointestinal bleeding in adults. Am Fam Physician. 2005;71:1339–1346
2. Lee EW, Laberge JM. Differential diagnosis of gastrointestinal bleeding. Tech Vasc Interv Radiol. 2005;7:112–122
3. Walsh RM, Anain P, Geisinger M, et al. Role of angiography and embolization for massive gastroduodenal hemorrhage. J Gastrointest Surg. 1999;3:61–65; discussion 66
4. Angtuaco TL, Reddy SK, Drapkin S, et al. The utility of urgent colonoscopy in the evaluation of acute lower gastrointestinal tract bleeding: a 2-year experience from a single center. Am J Gastroenterol. 2001;96:1782–1785
5. Zuckier LS. Acute gastrointestinal bleeding. Semin Nucl Med. 2003;33:297–311
6. Nusbaum M, Baum S, Blakemore WS, et al. Demonstration of intra-abdominal bleeding by selective arteriography: visualization of celiac and superior mesenteric arteries. JAMA. 1965;191:389–390
7. Kuhle WG, Sheiman RG. Detection of active colonic hemorrhage with use of helical CT: findings in a swine model. Radiology. 2003;228:743–752
8. Hara AK, Walker FB, Silva AC, et al. Preliminary estimate of triphasic CT enterography performance in hemodynamically stable patients with suspectedgastrointestinal bleeding. AJR Am J Roentgenol. 2009;193:1252–1260
9. Huprich JE, Fletcher JG, Fidler JL, et al. Prospective blinded comparison of wireless capsule endoscopy and multiphase CT enterography in obscure gastrointestinal bleeding. Radiology. 2011;260:744–751
10. Wu LM, Xu JR, Yin Y, et al. Usefulness of CT angiography in diagnosing acute gastrointestinal bleeding: a meta-analysis. World J Gastroenterol. 2010;16:3957–3963
11. Sun H, Jin Z, Li X, et al. Detection and localization of active gastrointestinal bleeding with multidetector row CT angiography: a 5-year prospective study in one medical center. J Clin Gastroenterol. 2012;46:31–41