The SNMMI Procedure Standard/EANM Practice Guideline for Gastrointestinal Bleeding Scintigraphy 2.0

A. Request

At the time of the request, it is important for the referring clinician to have a management plan in place before GIBS. Through early coordination of clinical services (such as interventional radiology, gastroenterology, or surgery), the patient can be directed promptly to the next diagnostic or therapeutic procedure if GIBS has positive results (35). Any unnecessary delay increases the likelihood of negative angiography findings, as bleeding often stops spontaneously (33).

All pertinent clinical information should be reviewed before the study is started. Information specifically related to GIBS may include the following.

1. Clinical signs of gastrointestinal bleeding (frequency, volume, and character [hematochezia, melena, or hematemesis]; current and recent hemoglobin, hematocrit, and blood urea nitrogen findings; number of recent transfusions; current blood pressure and heart rate; presence of orthostatic vital signs)

2. History of prior abdominal or pelvic surgeries

3. Diagnostic studies (nasogastric tube aspiration, esophagogastroduodenoscopy, capsule endoscopy, double-balloon enteroscopy, sigmoidoscopy or colonoscopy, prior GIBS, angiography, CT enterography/angiography)

4. Therapeutic interventions (endoscopic epinephrine injection, coagulation [by cautery, heater probe, laser, or argon plasma coagulator] or mechanical therapy [clips, bands, or detachable loops], angiographic embolization, selective arterial infusion of vasoconstrictors such as vasopressin)

5. Current medications

6. Factors that may decrease RBC radiolabeling efficiency (3,36,37) (drug interactions [iodinated contrast material, chemotherapy, digoxin, calcium channel blockers, cyclosporin, metronidazole, ranitidine, propanolol, quinidine, dipyridamole, and heparin]; low hematocrit; sickle-cell disease or thalassemia; circulating antibodies from prior transfusion or transplantation)

7. Oral contrast agents such as barium used for other gastrointestinal imaging studies (can cause photopenic artifacts (38) but are not an absolute contraindication for GIBS (39))

B. Patient preparation and precautions

Patients with gastrointestinal bleeding who are considered hemodynamically unstable should be monitored by a physician or nurse while in the nuclear medicine department.

Reinjection of radiolabeled blood poses the risk of incorrect administration to a wrong patient. Written policies must be in place, with special safeguards regarding the handling and administration of blood to eliminate any possibility of administration to the wrong patient, particularly when two or more RBC labeling studies are performed simultaneously. Universal precautions must be followed to avoid staff exposure to blood products. Refer to the SNMMI Procedure Standard for Use of Radiopharmaceuticals.

Fasting is not required for GIBS. However, fasting may be required for subsequent procedures such as angiography or surgery.

Patients should be instructed to void immediately before imaging so they are comfortable during a potentially long scan and to facilitate scan interpretation.

C. Radiopharmaceuticals

Historically, two radiopharmaceuticals have been used for GIBS: ^99mTc-RBCs and ^99mTc-sulfur colloid. ^99mTc-RBCs are the radiopharmaceutical of choice for performing GIBS because of an intravascular half-life that allows continuous imaging of the gastrointestinal tract over many hours (40–42). The superior clinical utility of ^99mTc-RBCs over early studies using ^99mTc-sulfur colloid has been demonstrated in comparison studies (40–42).

^99mTc-RBCs can detect gastrointestinal bleeding at a rate of as low as 0.04 mL/min in experimental animal models and 0.1 mL/min in clinical studies (28,43). High efficiency of RBC labeling with minimal unbound ^99mTc is critical to producing optimal, artifact-free images. Three methods are available to label RBCs: in vitro, modified in vivo, and in vivo. The in vitro method using a commercially available kit yields the highest labeling efficiency (≥95%) and is the method of choice (44,45). A further advantage of the in vitro method is that a sample can be evaluated for radiolabeling efficiency with a centrifuge method as described in the manufacturer’s package insert. Additionally, if radiolabeling is substandard because of a drug interaction, low hematocrit level, or other factor (section VI.A.6), or because of a procedural deviation, a salvage technique may be attempted (46). This procedure involves transferring the in vitro ^99mTc-RBCs into a sterile 15-mL polypropylene centrifuge tube and centrifuging at 400g for 5 min. The supernatant is then removed with a sterile pipette, and the ^99mTc-RBCs are resuspended in 2 mL of 0.9% sodium chloride. If the radiochemical purity of the ^99mTc-RBCs is then more than 95% and adequate radioactivity remains, the salvage is deemed successful. The modified in vivo label (90% labeling efficiency) can serve as an alternative when the in vitro method is not available (47–49). The in vivo method is not recommended because of suboptimal labeling and a higher likelihood of free ^99mTc-pertechnetate. However, the in vivo method may be needed for patients who, because of religious convictions or other reasons, will not accept injection of blood.

The recommended administered activity of ^99mTc-RBCs is 555–1,110 MBq (15–30 mCi) in adults. In children under 18 y old, the recommended administered activity is based on the EANM Pediatric Dosage Card, which uses a baseline activity of 56 MBq (1.51 mCi) multiplied by a weight-based multiple (49,50). The resulting minimum administered activity is 80 MBq (2.16 mCi) for a 3-kg patient and the maximum administered activity is 784 MBq (21.19 mCi) for a 68-kg patient.

D. Protocol/image acquisition

1. Image acquisition

In the supine position, anterior images of the abdomen and pelvis are acquired (section VIII specifies equipment). A minimum image matrix of 128 × 128 is recommended. Any items on the patient that may produce imaging artifacts should be removed or moved out of the field of view. Care should be taken to keep patients’ upper extremities from overlying the abdomen and pelvis during imaging as the upper extremities can obscure findings and their movement can cause artifacts.

After the injection of ^99mTc-RBCs, rapid image acquisition at a rate of 1 frame per 1–3 s for 60 s (nuclear angiography) can be performed to visualize the distribution of vascular structures and may help differentiate between blood pool activity and bleeding on later images. However, these angiographic images seldom add to the overall study result and are considered optional.

Immediately after the angiographic study, dynamic imaging should be performed. Serial intermittent static images are not recommended. The maximum recommended frame rate should not exceed 1 frame per 60 s. As the frame rate becomes longer, the temporal resolution of the scan decreases, possibly leading to inaccurate localizing of the bleeding source.

Since intraluminal blood promotes rapid bowel peristalsis and movement of blood antegrade or retrograde from the bleeding site, faster frame rates such as 1 frame per 10–20 s allow for higher temporal resolution to better localize the gastrointestinal bleeding site (51). On the other hand, a small volume of intraluminal blood or slow gastrointestinal bleeding may be more difficult to detect when fast frame rates are used because of lower count densities (51). This shortcoming can be compensated by reformatting the acquired study into longer frames (51,52). Therefore, reformatting is recommended when using fast frame rates. However, the optimal dynamic frame rate for GIBS has not been established because there are no published clinical studies that have compared these various acquisition techniques.

Acquiring the dynamic images in 10- to 15-min sequences may facilitate review of these images by the physician because one series can be reviewed while subsequent sequences are still being acquired.

Since gastrointestinal bleeding occurs intermittently, the patient should be imaged continuously for as long as practical to identify the bleeding source (10,34,53–55). Initial imaging for a minimum of 60 min is recommended if no gastrointestinal bleeding is detected (14,19,42,52,56). If abnormal focal RBC activity is detected, image acquisition should continue for a sufficient time to confidently identify the bleeding site. Accurate localization of the bleeding site is dependent on identification of the initial location of extravasated blood and movement of blood from that site within the bowel lumen. Increased imaging time may be particularly needed to differentiate a small-bowel bleed from a large-bowel bleed. SPECT/CT may also be helpful in this context (57).

Urine activity in a full bladder may obscure sigmoid or rectal bleeding on a standard anterior view. Lateral, posterior, or subpubic views may help in identifying activity in the rectum that would otherwise not be detected because of bladder activity or soft-tissue attenuation. The entire abdomen and pelvis must be examined before one can conclude that no gastrointestinal bleeding is detected. When a dual-head camera is used, simultaneous imaging in the anterior and posterior views may improve the sensitivity for detecting rectal bleeding. Furthermore, lateral views are helpful in differentiating anterior vascular penile activity (which can move or change in intensity during imaging) from bleeding in the more posteriorly located rectosigmoid colon (35).

If the patient has a bowel movement during the scan, the stool should be imaged to assess for radioactivity. The presence of radioactivity in the stool would only confirm active gastrointestinal bleeding and does not necessarily localize the origin of the bleeding. If gastric activity is visualized, an anterior image of the head and neck should be obtained to evaluate for possible thyroid and salivary gland activity. Activity at these sites suggests the presence of free ^99mTc-pertechnetate as the cause of the gastric activity rather than gastric bleeding.

If no bleeding site is identified on the initial images, delayed images can be acquired by rescanning the patient for up to 24 h, especially if there is clinical evidence of recurrent gastrointestinal bleeding (section F). All delayed images should be acquired using the same dynamic method as the initial images. If available, the same camera as used for the initial images should also be used for the delayed images.

2. Processing

If motion correction software is available, it can minimize the effects of patient movement. Computer subtraction of background activity of early images from later frames in the imaging sequence has the following limitations: the patient must remain still during the examination or motion correction software must be applied, and the biodistribution of the ^99mTc-RBCs should be similar between the early frames and any image to be subtracted (51,58–62). Failure to control these factors can cause false-positive findings.

E. Interpretation

Accurate interpretation of GIBS requires knowledge of normal and abnormal anatomic variations in the abdomen and pelvis. Comparison with anatomic imaging studies (CT, MR imaging, or radiographs) of the abdomen and pelvis is useful in establishing gastrointestinal tract and vascular anatomy. Review of the dynamic images in cinematic display is essential to detect subtle gastrointestinal bleeding and avoid inaccurate localization of the bleeding site (1,2,52,63,64). Proper adjustment of gray-scale levels on the interpreting physician’s computer display also facilitates the detection of subtle abnormalities.

^99mTc-RBCs are rapidly distributed within the vascular space, including the heart, liver, spleen, and great vessels. Some excreted radioactivity may be seen in the urinary tract because of small amounts of free ^99mTc-pertechnetate and other ^99mTc moieties even when in vitro labeling is used (65). The initial angiographic-phase images rarely reveal the site of rapid gastrointestinal bleeding, which may be difficult to localize on the subsequent dynamic images, or a vascular blush in neoplasms, arteriovenous malformations, or angiodysplasia (52,66,67).

The key diagnostic criteria for scintigraphic gastrointestinal bleeding are the appearance of activity outside the expected anatomic blood pool structures, a change in the intensity of activity on consecutive images, and movement of activity in a pattern consistent with bowel. All 3 of these criteria must be satisfied to diagnose a site of active gastrointestinal bleeding.

Small-bowel bleeding usually can be distinguished from large-bowel bleeding by its rapid curvilinear movement and usual central location in the abdomen or pelvis. In comparison, large-bowel bleeding has a more linear pattern and typically occurs in the periphery of the abdomen or pelvis. Large-bowel bleeding can also be visualized as an S-shaped pattern in the central pelvis conforming to the distribution of the rectosigmoid colon. The origin of the site of gastrointestinal bleeding should be reported as the location of the initial site of detected activity rather than the most intense, largest, or most proximal site of activity.

GIBS may be used to estimate the severity of the bleeding. Factors associated with a low bleeding rate include visualization of blood after 1 h, activity less intense than that in the liver, and shorter bleeding durations (28). Higher bleeding rates are associated with the early appearance of blood in the bowel, intense activity equal to or greater than that in the liver, and a longer duration of bleeding (28).

F. Sources of error

Delayed imaging after a period of nonimaging can be problematic because bowel activity seen immediately on the first frame of delayed images indicates merely that bleeding originating elsewhere in the gastrointestinal tract has occurred during the interim; the activity should not be misinterpreted as the bleeding site. Therefore, the location of delayed bowel activity should be reported as the bleeding site only when an actual episode of RBC extravasation on dynamic imaging is observed. Digital subtraction may be helpful for identification of the actual site of active bleeding when delayed images have been obtained (section D2).

The benefits of delayed imaging, including its effect on patient management such as transfusion requirements, referrals to angiography or surgery, and clinical outcomes, are controversial (68,69). Many investigators have shown that delayed images are not as accurate as early images in localizing the site of gastrointestinal bleeding (30,34,56,68,70–72). Some authors advocate imaging patients for as long as possible during the initial phase rather than performing routine delayed imaging at arbitrary intervals hours after injection (10,56). Other investigators have demonstrated usefulness for delayed imaging in detecting a site of intermittent gastrointestinal bleeding not seen on the initial phase (1,40,53,55,61,72–74). Therefore, delayed imaging is considered optional.

There are some pitfalls to the interpretation of GIBS images.

1. Free ^99mTc-pertechnetate

Free ^99mTc-pertechnetate can be visualized in the upper gastrointestinal tract secondary to swallowed salivary gland activity or excreted gastric mucosal activity. Since free ^99mTc-pertechnetate can move from the stomach into the small bowel over time, it can be mistaken for upper gastrointestinal bleeding. Images of the neck to detect thyroid and salivary gland activity should be obtained to confirm the presence of free ^99mTc-pertechnetate as a source of an artifact.

Urinary tract activity due to free ^99mTc-pertechnetate may be seen in the abdomen or pelvis.

2. Increased RBC activity due to other causes

In the reproductive system, penile blood pool can be mistaken for rectal bleeding (75). Obtaining lateral images or changing the position of the penis can distinguish penile activity from rectal bleeding (35). In addition, variable uterine activity during the ovulatory cycle causes fixed increased perfusion due to endometrial proliferation (76). Finally, a uterine leiomyoma may show transient, fixed activity due to hypervascularity (77–79).

Renal activity is usually fixed but can confuse interpretation when the activity arises from an unexpected location such as a pelvic or ectopic kidney (79,80), a horseshoe kidney (81), or a renal transplant.

Movement or pooling of urine activity can mimic gastrointestinal bleeding located in the ureter, bladder, or bladder diverticulum (3) or can be caused by urinary diversion surgery.

Vascular causes of abnormal RBC distribution can include aneurysms of the abdominal aorta, gastroduodenal artery, iliac artery, and other arterial vessels. Vascular grafts can also alter the normal blood pool anatomy. There are several reports of arterial leaks mimicking gastrointestinal bleeding (82–87). In addition, the literature reports a number of case reports demonstrating aortoduodenal fistula rupture (88), hemangiomas in the liver or small bowel (89,90), and abdominal varices (92,94). Varices are most commonly seen as static blood pool structures, but they can also rupture and cause bleeding (91,93,95). The literature also contains a report of visualization of an abnormal ovarian vein (96).

Splenic variants and pathology can cause fixed activity in the form of accessory spleens and splenosis. They can mimic gastrointestinal bleeding if they rupture (97–99).

Activity may be seen in the gallbladder in patients with renal failure or prior transfusions from hepatobiliary excretion of radiolabeled heme. Less commonly, gallbladder activity can be seen with hemobilia (100–105).

There are several other possible causes of increased RBC activity. Bleeding can occur from a pancreatic pseudocyst through the papilla of Vater and into the duodenum (106). A catheter site can cause static activity in the abdominal wall (107). A blush of activity in the bowel may occur because of hyperemia after surgical resection or in Crohn disease (83). Nonenteric bleeding activity can move and accumulate and confuse interpretation, including intraperitoneal hemorrhage (108,109), mesenteric bleeding (110), and soft-tissue hematoma/hemorrhage (111–116). Both benign and malignant neoplasms and metastatic disease can cause hyperemia and bleeding when ulcerated or necrotic (117–126). Retroperitoneal bleeding can show focal uptake that grows in intensity but is not expected to move in a luminal pattern (127).

G. Issues requiring further clarification

1. SPECT

Using planar technique, GIBS may be able only to approximate the site of bleeding. The inherent 3-dimensional nature of SPECT with multiplane reconstruction may yield more accurate localization of a gastrointestinal bleeding site. Comparison of SPECT results to anatomic cross-sectional imaging such as CT and MR imaging can also help to identify the source of bleeding.

2. SPECT/CT

Software-fused SPECT and CT images can be beneficial but are limited by the potential interval change in bowel location between the 2 modalities (128). The use of dedicated SPECT/CT hybrid cameras can help overcome these shortcomings. Several early studies have suggested that SPECT/CT scanning is able to better pinpoint the site of bleeding that is not well localized or is equivocal on planar images or to differentiate physiologic uptake from pathologic activity (129–131). In one study in which abnormal activity on standard planar scans was evaluated by SPECT/CT, SPECT/CT was required in 37% of the patients either to precisely localize the site of gastrointestinal bleeding or to exclude gastrointestinal bleeding (131). Although the authors used a 30-min acquisition for the SPECT images, a shorter acquisition (∼15 min) may be adequate.

SPECT/CT can be particularly helpful in the case of slow gastrointestinal bleeding, for which one may have to wait a long time to see the bowel activity conform into a more specific pattern (57). SPECT/CT can also estimate the length of the gastrointestinal tract leading to the bleeding site and therefore help decide which endoscopic approach to use for further evaluation (57). Furthermore, SPECT/CT helps clarify and avoid the pitfalls that can mimic gastrointestinal bleeding (35). More studies are needed to validate these results. No data exist on the use of SPECT/CT when planar GIBS shows no evidence of gastrointestinal bleeding.

A. Goals of the report

Refer to the SNMMI Procedure Standard for General Imaging.

B. Direct communication

Refer to the SNMMI Procedure Standard for General Imaging.

C. Written communication

Refer to the SNMMI Procedure Standard for General Imaging.

D. Contents of the report

The first part of the report should identify the study; provide patient demographics, clinical information (indication for the study), and comparison/correlative imaging data; describe the procedure (radiopharmaceutical, administered activity, route of administration, radiolabeling method for RBCs [in vitro, in vivo, or modified in vivo], duration of the acquisition, frame rate, projections acquired, and whether delayed or special images were obtained); and mention the quality and any limitations of the study.

Next, the report should describe the findings.

1. Presence of any baseline vascular, gastrointestinal tract, or solid-organ variants.

2. Characteristics of any abnormal activity (time of onset in relation to injection, shape, intensity in relationship to background liver activity, extent, subjective volume [small, medium, or large], focal or diffuse, stationary activity or movement of activity in gastrointestinal tract [and, if the latter, whether it is curvilinear (small-bowel) or linear (large-bowel), rapid or slow, and antegrade or retrograde]).

3. Location of any abnormal activity (quadrant of abdomen and pelvis, gastric, small bowel [duodenum, jejunum, or ileum], or large bowel [cecum, ascending colon, hepatic flexure, transverse colon, splenic flexure, descending colon, sigmoid colon, or rectum]).

If the bleeding site cannot be definitively localized, then giving an approximate site based on the imaging characteristics is appropriate. If SPECT/CT is used to further localize small-bowel activity, an attempt should be made to approximate the distance from the ampulla of Vater to the bleeding site to help determine which endoscopic technique can be used should gastroenterology be consulted. The ampulla of Vater is located in the medial aspect of the second (descending) portion of the duodenum at the confluence of the common bile duct and pancreatic duct.

Finally the report should give the impression (whether the study was positive or negative for active gastrointestinal bleeding and, for a positive scan, the originating site of gastrointestinal bleeding, if known).

IX. QUALITY CONTROL AND IMPROVEMENT, SAFETY, INFECTION CONTROL, AND PATIENT EDUCATION CONCERNS

Refer to the SNMMI Procedure Standard for General Imaging.

X. RADIATION SAFETY IN IMAGING

Refer also to the SNMMI Procedure Standard for General Imaging. Radiation dosimetry in adults, a 5-y-old child, and the fetus are presented in Tables 1–3.

Administration of radiopharmaceuticals to pregnant, potentially pregnant, or lactating patients is addressed in the SNMMI Procedure Standard for General Imaging. ICRP publication 106, appendix D, suggests that lactating patients who receive in vivo ^99mTc-RBCs require a 12-h interruption of breast feeding. No cessation of breast feeding is required for patients receiving in vitro ^99mTc-RBCs. The physician must consider the indication for the test, the potential benefit the information may provide, and the potential radiation risk to the mother and fetus.

XI. ACKNOWLEDGMENTS

The Committee on SNMMI Procedure Standards consists of the following individuals: Kevin J. Donohoe, MD (Chair) (Beth Israel Deaconess Medical Center, Boston, MA); Sue Abreu, MD (Sue Abreu Consulting, Nichols Hills, OK); Helena Balon, MD (Beaumont Health System, Royal Oak, MI); Twyla Bartel, DO (UAMS, Little Rock, AR); David Brandon, MD (Emory University/Atlanta VA, GA); Paul E. Christian, CNMT, BS, PET (Huntsman Cancer Institute, University of Utah, Salt Lake City, UT); Erica J. Cohen, DO (Edward Hines VA Hospital, IL); Dominique Delbeke, MD (Vanderbilt University Medical Center, Nashville, TN); Vasken Dilsizian, MD (University of Maryland Medical Center, Baltimore, MD); Steven W. Falen, MD (Northern California PET Imaging Center, Sacramento, CA); Kent Friedman, MD (NYU School of Medicine, New York, NY); James R. Galt, PhD (Emory University Hospital, Atlanta, GA); Bennett S. Greenspan, MD (Medical College of Georgia, Augusta, GA); Jay A. Harolds, MD (Advanced Radiology Services, PC, Grand Rapids, MI); Aaron Jessop, MD (UT MD Anderson Cancer Center, Houston, TX); David H. Lewis, MD (Harborview Medical Center, Seattle, WA); Alan B. Packard, PhD (Boston Children’s Hospital/Harvard Medical School, Boston, MA); J. Anthony Parker, MD, PhD (Beth Israel Deaconess Medical Center, Boston, MA); James A. Ponto, RPh, BCNP (University of Iowa, Iowa City, IA); Darko Pucar, MD (Georgia Regents University, Augusta, GA); Lynne T. Roy, CNMT (Cedars/Sinai Medical Center, Los Angeles, CA); Heiko Schöder, MD (Memorial Sloan-Kettering Cancer Center, New York, NY); Barry L. Shulkin, MD, MBA (St. Jude Children’s Research Hospital, Memphis, TN); Michael G. Stabin, PhD (Vanderbilt University, Nashville, TN); and Mark Tulchinsky, MD (Milton S. Hershey Medical Center, Hershey, PA).

The EANM Board consists of the following individuals: Fred Verzijlbergen, MD, PhD (Erasmus MC Central Location, Rotterdam, The Netherlands); Arturo Chiti, MD (Istituto Clinico Humanitas, Rozzano Milan, Italy); Bernd J. Krause, MD, PhD (Universitätsklinikum Rostock, Rostock, Germany); Wim Oyen, MD, PhD (Radboud University Medical Centre, Nijmegen, The Netherlands); Valerie Lewington, MD, PhD (Guys Hospital, Kings College, London, United Kingdom); Savvas Frangos, MD (Bank of Cyprus Oncology Center, Strovolos, Nicosia, Cyprus); and Jure Fettich, MD (University Medical Centre Ljubljana, Ljubljana, Slovenia).

XIII. APPROVAL

This standard/guideline (version 2.0) was approved by the Board of Directors of the SNMMI on November 17, 2014, and by the EANM Board on September 22, 2014. Version 1.0 was approved on February 7, 1999.