GASTROENTEROLOGY 982 ;83 :394-8 An Evaluation of 99mTc-Labeled Red Blood Cell Scintigraphy for the Detection and Localization of Gastrointestinal Bleeding Sites JOHN A. MARKISZ, DOV FRONT, HENRY D. ROYAL, BARRY SACKS, J. ANTHONY PARKER, and GERALD M. KOLODNY Divisions of Nuclear Medicine and Angiography, Departments of Radiology, Harvard Medical School and Beth Israel Hospital, Boston, Massachusetts 99mTechnetium-Iabeled red blood cell scintigraphy was performed upon 39 patients with clinical evidence for acute lower gastrointestinal bleeding from an unknown source. Seventeen of 39 patients (44%) had a scan that indicated active gastrointestinal bleeding. Scans became positive or more h after injection, consistent with intermittent bleeding, in 8 of 7 patients (47%). In the patients in whom the bleeding site was definitely identified by arteriography, surgery, or colonoscopy, scintigraphy correctly localized the bleeding site in 0 of patients (9 %). Four of patients (3%) had an active bleeding site identified by arteriography. Ten of 7 patients (58%) with a positive scan required either gelfoam embolization (4 patients) or surgery ( patients) to control the bleeding, whereas only of 22 patients (5%) required surgery when the scan was negative. Six deaths occurred in the scan-positive patients compared with no deaths in the scan-negative patients. None of the 8 patients who had arteriography and no active bleeding site by scintigraphy had arteriographically demonstrable active bleeding. Scintigraphy provides a reliable noninvasive test to screen patients in whom arteriography is being considered to localize active bleeding sites. If the arteriogram is negative, the scintigraphic findings alone may guide the surgical or arteriographic intervention. In addi- Received November 9, 98. Accepted March 2, 982. Address requests for reprints to: Henry D. Royal, M.D., Division of Nuclear Medicine, Department of Radiology, Beth Israel Hospital, 330 Brookline Avenue, Boston, Massachusetts 0225. Dr. Parker is the recipient of Research Career Development Award K04 HL 0045 from the National Institutes of Health. This study was presented at the Annual Meeting of the Inter American Association of Gastroenterology, Bal Harbour, Florida in October 98. 982 by the American Gastroenterological Association 00-5085/82/080394-05$02.50 tion, scintigraphy identifies two patient populations which have considerably different morbidity and mortality. To be diagnostic, arteriography usually must be performed when a patient is actively bleeding, In the upper gastrointestinal tract, active bleeding can be assessed by nasogastric aspiration; therefore the decision to perform endoscopy or arteriography to localize the bleeding site, or both, can usually be made easily. In the lower gastrointestinal tract, bleeding is often intermittent and the signs of active gastrointestinal (GIl bleeding (tachycardia, guiacpositive stools, orthostatic hypotension) are frequently unreliable. Because of these problems there is a need for a sensitive noninvasive test that could be used to detect and localize active GI bleeding in the lower gastrointestinal tract. Two radionuclide techniques have recently been described as possible sensitive noninvasive tests for active GI bleeding. Technetium sulfur colloid scintigraphy uses a radiopharmaceutical that is rapidly cleared from the intravascular space (T /2 = 3 min in normals). In animals, the sensitivity of this technique has been excellent, detecting and localizing bleeding sites with bleeding rates as low as 0.05 mll min (-3). The disadvantage of this technique is that the patients have to be actively bleeding during the few minutes in which the radiopharmaceutical is in the blood. An alternative technique is to use an intravascular tracer such as 99ffiTc-Iabeled red blood cells (4-7) or albumin (8-0). For the bleeding site to be localized, 5-70 cm 3 of blood is needed (,8). The advantage of this technique is that the patient can be monitored for active GI bleeding for as long as 24 h after a single injection.
August 982 RBC SCINTIGRAPHY AND GI BLEEDING 395 For the last 2 yr we have been performing 99mTc_ labeled red blood cell scintigraphy in patients with suspected GI bleeding. We have reviewed our experience in an effort to answer the following questions: (a) Is red blood cell scintigraphy sufficiently sensitive in detecting active GI bleeding to serve as a screening test for arteriography? (b) Can red blood cell scintigraphy reliably localize the site of bleeding so that it can be used to direct angiography or surgery, or both? (c) What is the long-term prognosis and management of patients whose red blood cell scans are negative vs. those whose scans are positive? Methods Thirty-nine consecutive patients (24 men, 5 women) suspected of having a GI hemorrhage from an unknown source underwent a total of 50 scanning procedures. The patients ranged in age from 2 to 97 yr (average age was 8 yr), with 27 of the patients ranging in age from 7 to 97 yr old. Autologous red blood cells were labeled with 99mTc in vitro using the Brookhaven kit (). An anterior abdominal flow study was obtained on a large field of view gamma camera. Thirty-two.5-s images were acquired immediately after injection of 5 mci (550 MBq) of labeled red blood cells. The total whole body radiation dose was 0.45 rem (2). Static anterior views were taken for 500,000 counts, at 0-min intervals for 90 min, with lateral or oblique views obtained to aid in localization once a bleeding site was evident (Figure ). If no bleeding site was observed, delayed images of up to 24 h were taken (Figure 2) Angiography was performed as soon as possible after the scintigraphic study was obtained. The scintigraphic findings were used to determine the site of the first selective injection. If the first injection did not reveal the bleeding site, the other arteries supplying the GI tract were examined. The 4 patients with a positive arteriogram were treated with gelfoam embolization. One patient with a negative arteriogram received selective arterial vasopressin infusion based on the scintigraphic findings. The charts of all patients were reviewed. The significance of various clinical findings were tested using either an unpaired Student's t-test (age, number of transfusions, length of hospital stay) or X 2 analysis (invasive therapy, deaths) (Table ). Results As summarized in Figure 3, 7 patients had positive scans. In 4 of these patients angiography also demonstrated sites of active bleeding. In of these 4 patients, the bleeding site was reported to be in the transverse colon on scintigraphy; however, arteriography showed it to be in the duodenum (Figure 4). Seven of the patients with positive scans had negative arteriograms and had no arteriographic studies. In 5 of 7 patients with a negative arteriogram, the site of bleeding was identified by direct visualization of active bleeding or fresh blood clots intraoperatively. One of the patients with a positive scan and no arteriography also had his bleeding site identified intraoperatively. In patient, active bleeding was seen on colonoscopy. Of the 22 patients with negative scans, 8 patients had arteriography; none of these 8 patients had an active bleeding site identified. Twenty-one of these 22 patients with negative scans required only medical management and 20 of the 2 medically treated patients remained free of bleeding symptoms for the duration of the study, ranging from 5 to 9 mo after scintigraphic examination. In patient the arteriography revealed angiodysplasia of the ascending colon. In this patient, both scintigraphy and arteriography were negative for active bleeding, and no active bleeding, or subsequent blood loss was reported by the surgeons. One patient, a 2 yr-old woman, re- Figure. Accumulation of activity in the transverse colon, observed between 30 and 0 min postinjection in this 83-yr-old patient with extensive diverticular disease and a 5-yr history of GI bleeding. Angiography was negative.
39 MARKISZ ET AL. GASTROENTEROLOGY Vol. 83, No. 2 Figure 2. Focus of activity first observed at 55 min, which has extended throughout the ascending and transverse colon by h. At surgery, bleeding diverticuli were found in this 93-yr-old woman, who remained asymptomatic following a right colectomy. turned twice with GI bleeding complaints, but no determination of a bleeding site was made. Assuming that the true positive cases are those established by arteriography, colonoscopy, or surgery (Table 2), the sensitivity of arteriography is only 4 of patients (3%) in our study, compared with of patients (00%) for 99ffiTc-Iabeled red blood cell scintigraphy. Of the 7 patients with positive scans, 8 of 7 bleeding sites in patients (47%) were first visualized between and 24 h postinjection (Figure 5). Nine of patients (82%) whose bleeding site was definitively identified by independent tests had positive scans within the first 90 min of the study. In addition to diagnosing the site of active GI bleeding, scintigraphy demonstrated ileocolic varices (Figure ) and an unsuspected abdominal aortic aneurysm in 2 additional patients. The long-term prognosis and management of patients who had positive scintigraphy differed dramatically from those who had negative scintigraphy (Table ). In the 7 patients with a positive scan, gelfoam embolization was used to control the bleeding in 4 patients and surgery was required in other patients. The average length of hospitalization was 33 days (range -75 days) and these patients required.9 U of blood (range 0-). The average age of the patients was 78 yr (range, 2-97 yr). Six of the patients with positive scans died during their hospitalization. Of the 22 patients who had a negative scintigram, only patient required surgery for angiodysplasia. One additional patient has had recurrent bleeding from an unidentified site. The average length of hospitalization was 7.0 days (range, 2-27 days) and these patients required 2.5 U of blood (range 0-8). The average age of the patients was 3 yr (range, 2-90 yr). Discussion Since arteriography is invasive, the intermittent nature of GI bleeding is the major obstacle to its routine use for the accurate detection and localization of lower GI bleeding sites. The results of our study suggest that red blood cell scintigraphy can be used effectively to screen patients who are having arteriograms for active GI bleeding. No patient with a negative red blood pool study had a positive arteriogram for active GI bleeding. The high sensitivity of red blood cell scintigraphy relative to arteriography has been reported by others (3-7). In addition to detecting the presence of active GI bleeding, scintigraphy correctly localized the bleeding site in 0 of patients (9%) whose bleeding site was definitively identified. The accuracy with which the bleeding site was localized was unexpected since reflux and movement of the activity due to persistalsis was anticipated. The one false localization occurred in a patient who was only imaged for 80 min (Figure 4). Activity in the duodenum was misinterpreted to be in the transverse colon. Had additional delayed views been obtained, this error may have been avoided. Accurate localization requires frequent scans of the patients. Computerized acquisition with a movie display of these images can be very helpful. Rapid serpiginous movement of activity is very characteristic of activity within the small bowel. If a long delay has occurred between scans, movies should be acquired when the scans become positive. The certainty of localization decreases as the interval between negative and positive Table. Summary of Clinical Characteristics Number of patients Average age (yr) Required invasive therapy" Number of transfusionso (U) Length of hospitalizationo (days) Number of deaths b a p < 0.0. b P < 0.05. Scan + 7 78 (2-97) 0.9 (0-) 33 (-75) Scan - 22 3 (2-90) 2.5 (0-8) 7.0 (2-27) o
August 982 RBC SCINTIGRAPHY AND GI BLEEDING 397 Number of Patients (39) Positive s c a ~ n N~ e g ascans t i V e ~ ( 7 ) /(22) positive negative \ negative \ arteriograms arteriograms no arteriograms arteriograms no arteriograms po,hive nl po,," ~ no o t : : : ~ : e- p~ e! a t e (4) ~ ( 7 ) () (8) (4) O.R. further O.R. colon- further on for recurrent episodes results work-up results oscopy work-up angiodysplasia G.!. bleeding no diagnosis (5) (2) () () (4) () (20) () Figure 3. Schematic representation of diagnostic results. scans increases. Nine of patients (82%) whose bleeding sites were localized by independent means had positive scans within the first 90 min of the study. Red blood cell scintigraphy may be used reliably to direct the arteriography. If the arteriogram is negative, vasopressin may be infused into the vessel most likely supplying the bleeding site. If surgery is required, the surgeon can use the results of the scintigraphic study to plan his operation. All of the patients in this study had documented GI bleeding which precipitated or occurred during hospitalization. Since the red blood cell scan was performed only on patients bleeding from an unknown site, most of the patients in this study were bleeding from lower GI sites. The long-term prognosis and management of the patients who had positive scintigraphy differed dramatically from those who had negative scintigraphy (Table 2). Radionuclide imaging should be particularly useful in the elderly when the hazards of arteriography are the greatest. Further studies are needed to determine the relative merits of gastrointestinal bleeding studies using radioactive agents that are rapidly cleared from the intravascular space, such as 99IDTc-Iabeled sulfur colloia, vs. agents that remain in the intravascular space, such as 99IDTc-Iabeled red blood cells. Rapidly cleared agents may be more sensitive than intravascular agents; however, they have the disadvantage of allowing patient monitoring for only a few minutes. Since bleeding is often intermittent, prolonged monitoring of patients is very desirable. Eight of 7 scans that were positive (47%) became positive only after or more h of monitoring. An additional disadvantage of sulfur colloid is that this agent normally accumulates in the liver and spleen, which would make bleeding sites near these organs difficult to identify. Figure 4. Duodenal bleeding, documented by arteriography, in a 97-yr-old man was thought to be in the transverse colon at the hepatic flexure. Had the study been continued past 80 min, it might have shown a small bowel pattern, allowing for correct localization of bleeding site. Table 2. Diagnoses (Number of Patients) Angiography Distal ileum (superior mesenteric artery) Duodenal-gastroduodenal artery Surgery Extensive diverticular disease Bleeding gastric varices Polypoid carcinoma-descending colon Lesion of distal ileum Colonoscopy Diverticular disease No. of patients 4 3 3
398 MARKISZ ET AL. GASTROENTEROLOGY Vol. 83, No. 2 00 80 Q) ~ ~ 'iii 0 Il. (jj 0 I- * ( 4( 2( 0 -----.-- ~_ / - - - -,., f le -..-------- 9 2 I ~ Time (hours) 8 2 24 Figure 5. Percentage of total positive studies vs. time of first scintigraphic appearance of bl eeding. To obtain accurate results using red blood cell scintigraphy, a high labeling efficiency of the red cells is needed. Free pertechnetate is normally secreted by the gastric mucosa; therefore, a falsepositive scan may result due to the unbound radioisotope. We achieved a high labeling efficiency by using an in vitro technique (). Winzelberg et al. (7) has described a partly in vivo, partly in vitro technique which also results in a high labeling efficien- Figure. Previously unsuspected ileocolic varices found scintigraphically and documented b y arteriography. During the latter procedure gelfoam embolization was carried out. cy. The most commonly available technique for labeling red blood cells (for cardiac studies) results in a lower labeling efficiency (3). If this in vivo technique is used, a nasogastric tube attached to continuous suction may reduce the number of falsepositive bleeding studies. In conclusion, red blood cell scintigraphy appears to be a reliable sensitive noninvasive test for active GI bleeding. In addition, valuable information regarding the site of bleeding may be obtained and should be used to aid the angiographer in a more selective examination. If the arteriogram is negative, scintigraphy alone may guide the surgeon to the correct location of the bleeding site. Finally, patients with positive and negative scintigrams had dramatically different clinical courses. References. Alavi A, Dann RW, Baum S, et al. Scintigraphic detection of acute gastrointestinal bleeding. Radiology 977;24:753-7. 2. Alavi A, McLean GI. Radioisotopic detection of gastrointestinal bleeding. An integrated approach with other diagnostic and therapeutic modalities. In: Freeman, LM, Weissman, HS, eds. Nuelear medicine annual 980. New York: Raven Press, 980:77-28. 3. Alavi A, Ring EJ. Localization of gastrointestinal bleeding: superiority of 99mTc sulfur colloid compared with angiography. Am J Radiology 98J ;3 7:74-8. 4. Winzelberg GG, McKusick KA, Straus HW, et al. Evaluation of gastrointestinal bleeding by red blood cells labeled in vivo with technetium 99m. J Nuel Med 979;20:080-. 5. Royal HD, Papanicolaou N, Bettman M, et al. Scintigraphic identification of bleeding duodenal varices. Am J GastroenteroI980;74:73-5.. Smith K, Arterburn G. Detection and localization of gastrointestinal bleeding using Tc-99m-pyrophosphate in vivo labeled red blood cells. Clin Nuel Med 980;5:55-0. 7. Winzelberg GG, Froelich JW, McKusick KA, et al. Radionuelide localization of lower gastrointestinal hemorrhage. Radiology 98;39:45-9. 8. Miskowiak. Neilsen SL, Munch 0, et al. Abdominal scintiphotography with 99mTc labeled albumin in acute gastrointestinal bleeding. Lancet 977;2:852-4. 9. Miskowiak. Nielsen SL, Munck 0, et iii. Acute gastrointestinal bleeding detected with abdominal scintigraphy using technetium 99m labeled albumin. Scand J Gastroenterol 979;4:389-94. 0. Miskowiak J, Nielsen SL, Munck O. Scintigraphic diagnosis of gastrointestinal bleeding with 99mTc labeled blood pool agents. Radiology 98 ;4 :499-504.. Smith TD, Richards P. A simple kit for the preparation of 99mTc-labeled red blood cells. J Nuel Med 97;7:2-32. 2. Malamud H. Dosimetry of 99n>Tc-labeled blood pool scanning agents. Clin Nuel Med 978;3:420-. 3. Pavel DG. Zimmer AM, Patterson VN. In vivo labeling of red blood cells with 99mTc. A new approach to blood pool visualization. J Nuel Med 977;8:305-8.