ORIGINAL ARTICLE 
Annals of Nuclear Medicine Vol. 9, No.4, 203-208, 1995 
Sequential subtraction scintigraphy with 99mTc-RBC for the early detection of gastrointestinal bleeding 
Hikaru SETO Masanan KAGEYAMA, Yi-wei WU, Masashi SHIMIZU, Toshiro NAGAYOSHl, Naoto WATANABE, Kuninori NOMURA and Masao KAKISHITA 
Department of Radiology, Toyama Medical and Pharmaceutical University 
To improve the early detectability of gastrointestinal (GI) bleeding, we have developed a new subtraction scintigraphic technique using in vivo labeled 99*Tc red blood cells (RBC). Use of this new sequential subtraction method every 5 min makes it possible to detect early the bleeding site and calculate the bleeding rate. In phantom experiments, focal activity was detected on subtraction images at pump rates of more than 0.20 ml/min at any urethane thickness, but was negative on conventional scintigrams. The calculated pump rate by the subtraction method correlated well with the actual syringe pump rate (y = 1.07x - 0.02, r = 0.99, p < 0.01) in the case of an 8-cm urethane board as an abdominal wall. Twenty-four of 50 patients with suspected GI bleeding were confirmed to be positive within 24 hr. The subtraction method showed 1 5 true positive and 2 false positive cases within 60 min . On the other hand, the conventional scintigraphic method showed only 7 true positive and I false positive cases. Sensitivity of the subtraction method was 62.5%, which was significantly higher (p < 0.05) than that of the conventional scintiraphic method (29.2%). 
These results suggest that sequential subtraction scintigraphy with 99mTc-RBC is an effective method for the early detection of GI bleeding and shortening the examination time as compared with conventional scintigraphy. 
Key words: gastrointestinal bleeding, bleeding rate 99mTc-RBC, subtraction scintigraphy 
INTRODUCTION 
GASTROINTESTlNAL (GI) bleeding sometimes poses a serious diagnostic challenge to physicians, especially when it occurs intermittently. Invasive techniques such as endoscopy and contrast angiography may provide accurate diagnoses, but have limitations in that they have to be performed during active hemorrhage. GI bleeding tends to be intermittent,1 and therefore a study should be per-formed repeatedly over a period of 24 hr. 999mTc-labeled red blood cells (RBC) which remains in the intravascular blood pool for a prolonged period of time, is an ideal radiopharmaceutical for detecting the site of intermittent GI bleeding.2-7 In the advanced aged group, the patient's condition is sometimes very serious, and therefore shortening the time of the study and enhancing the detectability of the bleeding site are important considerations.8 
We have developed a new noninvasive scintigraphic technique to quickly detect the bleeding site and calculate the bleeding rate by the sequential subtraction scintigraphy method and have compared results obtained with it to those of conventional scintigraphy. This technique is based on the following theoretical considerations. 
THEORETICAL CONSIDERATIONS 
When a radiopharmaceutical remaining in a vascular space is injected intravenously into a patient with active bleeding, blood activity gradually reaches a plateau in I O min. The fraction of injected activity extravasated at the bleeding site gradually increases with the passage of time, but background activity is not significantly changed be-cause of the relatively small amount of extravasated activity as compared to the total blood volume. This phenomenon is repeated each time blood recirculates, adding another fraction of the extravasate at the bleeding site. If scintigrams are taken every 5 min and are stored in a computer for 60 min, sequential 5 min subtraction scintigrams are easily produced by a digital gamma camera. These subtraction scintigrams facilitate the early detection of the bleeding site as compared to conventional scintigrams because of the very low background activity and high contrast of the lesion to background (Fig. 1). Moreover, it is possible to calculate the bleeding rate, if the patient's blood sample is counted with the same camera correcting for the depth of the bleeding site. 99mTc-RBC is considered to be the most suitable agent for this purpose. 
MATERIALS AND METHODS 
Phantom experiments As a simulation of GI bleeding, we used a syringe pump containing '99mTc-pertechnetate solution similar to the blood activity of a patient administered 740 MBq of 99mTc. Urethane boards of various thicknesses as abdominal walls and a flood source of 99mTc as abdominal back-ground activity were used on a large-field-of-view gamma camera (Fig. 2). 
Data were obtained In a 512 x 512 matrix at various rates of the syringe pump from 0.05 to 1.00 ml/min. Subtraction images were produced every 5 min at various pump rates at various urethane thicknesses. Detectable 
activity on the subtraction images was then assessed. Pump rates were calculated from the detectable activity per 5 min with correction for urethane thickness. The calculated rates were then compared to the actual syringe pump rates. 
Clinical application Fifty consecutive patients (26 men and 24 women) aged 5-93 years were studied. All patients referred with clinical evidence of GI bleeding as documented by anemia, melena and persistent maroon stool underwent 99mTc-RBC scintigraphic studies to localize sites of GI bleeding. 
Radionuclide imaging was performed in the supine position with a large-field-of-view gamma camera inter-faced with a computer. Energy was centered at 140 keV and a 20% window. Red blood cells were labeled in vivo by the method of Pavel et al.9 A commercially available kit (Dai-ichi Radiopharrnaceutical Laboratory Inc., Japan) containing 4 mg stannous chlorlde, and 20 mg sodium pyrophosphate was diluted with 3 ml of sterile saline and injected into an antecubital vein. Thirty min later, 740 MBq of 99mTc-pertechnetate as an adult dose was injected and then sequential images of the abdomen were taken at 5-min intervals for the first 60 min and then at 2-4 hr. Delayed images up to 24 hr were obtained when early images were negative and/or recurrent bleeding was suspected. 
Data were obtained in a 512 x 512 matrix. Scintigrams were produced in a 256 x 256 matrix at 5-min intervals, and then 5-min images were subsequently subtracted from each other up to 60 min. Each 5-min image had more than 5,000 k counts. These sequential subtraction scintigrams with 99mTc-RBC were retrospectively interpreted by three nuclear physicians Independently with regard to the detectability of a bleeding site and were compared to the results of conventional scintigrams within 60 min. The 99mTc-RBC scintigrams were considered positive if a focus of activity occurred corresponding to the location in the GI tract, which either increased in Intensity or changed in location within the gut. Final diagnoses were established by endoscopy, angiography, operation and clinical findings such as persistent melena and maroon stool for 24 hr. 
For statistical analysis, we used Student's t-test, and a p value less than 0.05 was considered to be significant. 
RESULTS 
Phantom experiments Subtraction Images were produced at various pump rates at various urethane thicknesses by means of phantom simulating active bleeding. Focal activity was detected on the subtraction images at pump rates more than 0.20 ml/ min at any urethane thickness (Fig. 3). Pump rates were calculated from detectable focal activity per 5 min on the subtraction Images at various urethane thicknesses and were compared with the actual syringe pump rates (Table l). A pump rate more than 0.20 ml/min could be calculated at any urethane thickness. Good relationships between the calculated pump rates and actual syringe pump rates were obtained (Fig. 4). 
Clinical application 
Of the 50 patients with suspected bleeding, 24 showed active hemorrhage up to 24 hr. The subtraction sclntigraphic method showed 15 true positive and 2 false positlve cases. On the other hand, the conventional scintigraphic method showed only 7 true positive and l false positive cases within 60 min (Table 2). Sensitivity of the subtraction method was 62.5%, which was significantly higher (p < 0.05) than that of the conventional method (29.2%). No significant difference in specificity between the two methods was noted. The bleeding sites of l 5 positive cases detected by the subtraction method were as follows; stomach 3, jejunum 7, ileum l, ascending colon 2, and undetermined 2. As to the timing of positive scintigrams, eight of the 15 patients became positive at less than 30 min. The remaining patients became positive at less than 60 min. The earliest case was at 10 min in a patient with gastric bleeding. Representative cases are shown in Figures 5 and 6. 
DISCUSSION 
Endoscopy is usually the method of choice to determine a site of bleeding In the upper GI tract, but the procedure is difficult to perform in the small intestine and poorly prepared co]on. Contrast angiography is of great use in both locating and controlling active bleeding from the GI tract. These procedures are, however, invasive and of limited use during active hemorrhage. In contrast, radio-nuclide techniques are noninvasive and can be used to follow up a patient with intermittent GI bleeding. The possibility of detecting bleeding sites with stable blood labeling agents was first suggested by animal studies with 51Cr-labeled red blood cells and 131I-labeled albumin.Io Subsequently, 99mTc-labeled human serum albumin (HSA) was used to localize bleeding sites In patients.ll At the present time, the use of HSA as the carrier is, however, largely abandoned owing to the superior binding of 99mTc with red blood cells and the problems caused by diffusion of circulating HSA into inflammatory and neoplastic lesions. Recently, 99mTc-sulfur colloidsl2 and 99mTc-RBC2-7 have been used for this purpose. 99mTc-sulfur colloids is rapidly cleared by the RES system and its half-life in blood is approximately 2.5-3.5 min, 13 A smal1 amount of bleeding is therefore detectable because of the low back-ground activity, but this procedure must be performed during active bleeding and is not suitable for intermittent bleeding.5 On the other hand, 99*Tc-RBC remains in the blood pool for a long time and is suitable for intermittent bleeding, although background activity is high and the minimum detectable bleeding Is limited.6,14 
In order to improve the high background activity and detectability of the minimum bleeding on 99mTc-RBC images, we have developed a new noninvasive scintigraphic technique, in which sequential subtraction scintigraphy enables us to detect a small amount of bleeding within 60 min because of the low background activity. In phantom experiments, subtraction scintigrams easily showed a focus of bleeding, despite negative conventional scintigrams (Fig. 3). The minimal detectable bleed-ing rate was O. 1-0.2 ml/min by the subtraction method (Table l). This value was superior to the 0.20-0.40 ml/ min reported by Chandeysson et al.15 with conventional scintigraphy. In clinical application, we arbitrarily used the initial 60 min data for comparison because one large multiinstitutional study showed that a maximum of 80% of studies performed to detect bleeding of labeled red blood cells become positive within 60 min of the start of examinations.5 The sensitivity of the subtraction method was 62.5%, which was significant]y higher (p < 0.05) than that of the conventional scintigraphic method (29 .2% ) . There were, however, 2 false positive cases, in which the subtraction method might produce artifacts due to the movement of the intestines. As to labeling red blood cells with 99mTc, we adopted an in vivo method.9 This method yields more free 99mTc-pertechnetate than in vitro methods. Free 99mTc-pertechnetate might be another cause of false positive artifacts because of its secretion from the salivary glands and gastric mucosa, but another advantage of the subtraction method is the ability to calculate the bleeding rate every 5 min, as shown in Figure 3. 
Our data from these experimental and clinical studies suggest that sequential subtraction scintigraphy with 99mTc-RBC is an effective method for the early detection of GI bleeding as compared to conventional scintigraphy. Re-cent technical advances in digital gamma cameras enable us to make subtraction images easily in routine clinical practice and to shorten the examination time in pediatric and elderly patients. 
ACKNOWLEDGMENTS 
We thank Mr. Syuichi Inagaki, senior technologist at the division of Nuclear Medicine for technical support, Ms. Kiyomi Dorin for secretarial assistance and Mr. John S. Gelblum for proofreading. 
REFERENCES 
l. Sos TA, Lee JG, Wixson D, Sniderman KW. Intermittent bleeding frorn minute to minute in acute massive gastro-intestinal hemorrhage: arteriographic demonstration. Am J Roentgenol 131 : 1015-l017, 1978. 
2. Winzelberg GG, McKusick KA, Strauss HW, Watsan AC, Greenfield AJ. Evaluation of gastrointestinal bleeding by red blood cells labeled in vivo with technetium-99m. JNucl Med 20: 1080-l086, 1979. 
3. McKusick KA, Froelich JW, Callahan RJ, Winzelberg GG, Strauss HW. 99nrrc-red blood cells for detection of gastro-intestinal bleeding: experience with 80 patients. Am J Roentgenol 137: I 1 13-1 1 18, 1981. 
4. Miskowiak J, Nielson SL, Munck O. Scintigraphic diagno-sis of gastrointestinal bleeding with 99~Tc-labeled blood pool agents. Radiology 141 : 499-504, 1 981 . 
5. Bunker SR, Lull RJ, Tanasescu DE, Redwine MD, Rigby MD, Brown JM, et al. Scintigraphy of gastrointestinal 
hemorrhage: superiority of 99*Tc red blood cells over 99"Tc sulfur colloid. Am J Roentgenol 143: 543-548, 1984. 
6. Smith R, Copely DJ, Bolen FH. 99utTc RBC scintigraphy: correlation of gastrointestinal bleeding rates with scinti-graphic findings. Am J Roentgenol 1 48: 869-874, ,1 987. 
7. Akashi T, Aizawa N. Suzuki Y, Tsuda K, MoriyaT. Mitsui T. Clinical evaluation of technetium-99m labeled red blood cell scan in the detection of gastrointestinal hemorrhage. KAKU IGAKU (Jpn J Nucl Med) 25: 23-29, 1988 (in Ja anese). 
8. Maurer AH, Rodman MS. Vitti RA, Revez G, Krevsky B. Gastrointestinal bleeding: improved localization with cine scintigraphy. Radiology 185: 187-192, 1992. 
9. Pavel DG, Zimmer AM. Patterson VN. In vivo labeling of red blood cells with 99~Tc: a new approach to blood pool visualization. J Nucl Med 1 8: 305-308, 1977. 
10. D'Addabbo A, Fersini M. Detection of intra-abdominal bleeding by radioisotope scanning after experimental etravasation. J Nucl Biol Med 1 8: 170-174, 1974. 
ll. Miskowiak J, Nielssen SL, Munck O, Andersen B. Ab-dominal scintiphotography with 99~technetium-labelled al-bumin in acute gastrointestinal bleeding. Lancet 2: 852-854, 1977. 
12. Alavi A, Dann RW, Baum S, Biery DN. Scintigraphic detection of acute gastrointestinal bleeding. Radiology 1 24: 753-756, 1977. 
l3. Alavi A. Detection of gastrointestinal bleeding with 99~'Tc-sulfur colloid. Semin Nucl Med 1 2: 126-1 38, 1982, 
l4. Robinson P. The role of nuclear medicine in acute gastrointestinal bleeding. Nucl Med Commu 14:849-855, 1993
15. Chandeysson PL, Hanson RJ, Watson CE, Henesy GT. Minimal gastrointestinal bleeding rate detected by abdomi-nal scintigraphy. J Nucl Med 24: 97, 1983 (abstr).