1381

Downloaded from www.ajronline.org by 68.100.197.246 on 04/04/17 from IP address 68.100.197.246. Copyright ARRS. For personal use only; all rights reserved

Enteroclysis in the Evaluation of

Obscure Gastrointestinal Bleeding

Anne Moch1 OBJECTIVE. The diagnostic yield of enteroclysis was retrospectively evaluated for
Hans Herlingen1 patients with obscure bleeding from the gastrointestinal tract.
Michael L. Kochman2 MATERIALS AND METHODS. A total of 128 patients with obscure gastrointestinal
bleeding were referred to our department for enteroclysis between 1 988 and 1993. The
Marc S. Levine1
original radiologic reports were reviewed to determine the radiographic findings in
Stephen E. Rubesin1
these patients. The radiographic findings then were correlated with medical, surgical,
Igor Laufer1
and pathologic findings.
RESULTS. Thirty-two patients had lesions found at enteroclysis. Nineteen of those
patients had confirmation of the radiographic diagnosis, primarily by pathologic
examination of the surgical specimen. Five other patients were found at surgery to
have had false-positive diagnoses at enteroclysis. Eight patients ceased to bleed
without pathologic corroboration, but their clinical presentation and course sup-
ported the radiographic diagnosis. Thus, 27 (21%) of the 128 patients had confirmed
or highly probable lesions seen at enteroclysis as the cause of obscure gastrointesti-
nal bleeding. Seventeen patients (1 3%) had tumors involving the small bowel, and
three (2%) had arteniovenous malformations in the jejunum.
CONCLUSION. This study corroborates earlier reports that enteroclysis is a useful
diagnostic test for examining the small intestine in patients with obscure gastrointes-
tinal bleeding.

AJR 1994;163:1381-1384

Gastrointestinal bleeding, usually chronic on intermittent, is defined as obscure,

on unexplained, if diagnostic examinations of the upper gastrointestinal tract and
the colon (including barium studies, endoscopy, on both) fail to reveal the cause of
the bleeding [1]. Angiognaphy, scintignaphy, entenoclysis, and entenoscopy are
additional diagnostic procedures that may be used in these patients in an attempt
to identify a source of bleeding in the small intestine. Two earlier studies found
entenoclysis to be a useful test for detecting lesions in the small intestine as the
cause of obscure bleeding [2,3].
Unexplained bleeding from the gastrointestinal tract accounted for 17% of all
referrals for entenoclysis at our hospital and was the third most common indication
for this procedure (after small-bowel obstruction and inflammatory disease). As a
result, we have had the opportunity to use entenoclysis on a significant number of
patients with this vexing clinical problem. The purpose of our study was to determine
Received May 3, 1994; accepted after revision the diagnostic yield of entenoclysis in patients with obscure gastrointestinal bleeding.
July 11, 1994.
tDepartment of Radiology, Hospital of the Uni-
versity of Pennsylvania, 3400 Spruce St., Philadel- Materials and Methods
phia, PA 19104. Address correspondence to H.
Herhnger. A total of 758 patients had enteroclysis examinations at our hospital between September
2Department of Medicine, Hospital of the Uni- 1988 and July 1993. In 128 (17%) of those 758 patients, the indication for enteroclysis was
versity of Pennsylvania, Philadelphia, PA 19104. unexplained gastrointestinal bleeding, manifested by rectal bleeding, melena, and/or heme-
0361-803X/94/1 636-1 381 positive stool. These 128 patients (74 men, 54 women) constituted our study group. Their
American Roentgen Ray Society average age was 62 years (range, 20-87 years).
 1382 MOCH ET AL. AJR:163, December 1994

Enteroclysis with barium and methylcellulose was performed on jejunum or ileum. The remaining six patients (22%) had
those 128 patients, applying the technique as previously described lesions in the ascending portion of the duodenum.
[4]. Because of the history of gastrointestinal bleeding, however, the Of the 96 entenoclysis examinations that showed no lesions,
examination was modified in several ways [5]. The catheter was we are aware of only one study with false-negative results in a
advanced only a short distance into the jejunum, with the balloon
patient with von Willebrands disease, who was found to have a
inflated near the ligament of Treitz, so that even the most proximal
small-bowel hematoma at later surgery. This was the only
Downloaded from www.ajronline.org by 68.100.197.246 on 04/04/17 from IP address 68.100.197.246. Copyright ARRS. For personal use only; all rights reserved

loops of jejunum could be included in the study. Because artenio-

venous malformations, a relatively common cause of gastrointesti- patient with no lesions seen at entenoclysis who had surgery.
nal bleeding, often are located in the proximal jejunum [6], these The remaining 95 patients with normal findings on enteroclysis
loops were carefully assessed by graded compression during both examinations had no pathologic corroboration (usually because
the single-contrast and the double-contrast phases of the examina- the bleeding stopped with conservative treatment), so it was
tion. If rapid flow of barium or methylcellulose into the colon pro- impossible in this study to determine the sensitivity of entenocly-
cluded adequate visualization of the ileum, 0.5 mg of glucagon was sis in detecting small-bowel disease in these patients.
given IV to increase ileal distention. Finally, the balloon on the cathe-
ten was deflated, and the catheter was withdrawn into the descend-
ing duodenum, where the balloon was reinflated. High-density Confirmed Lesions
barium, followed by air or methylcellulose, then was injected rapidly
through the catheter to outline the third and fourth portions of the Of the 19 patients with confirmed lesions at entenoclysis, 10
duodenum and the duodenojejunal flexure in order to detect poten- had malignant tumors, including three lymphomas (Fig. 1), two
tial bleeding sites that might not have been seen at upper endos- adenocarcinomas, two metastases, two carcinoids (Fig. 2), and
copy or routine upper gastrointestinal series.
one leiomyosarcoma. The malignant tumors appeared radio-
The original radiologic reports of these 128 patients were reviewed
graphically as ulcerated, annular, on exoenteric lesions. One
to determine the findings at enteroclysis. In all but three patients,
patient had a benign leiomyoma that appeared as a smooth sub-
whose radiographs were missing, the radiographs were reviewed to
further characterize the findings at enteroclysis. Thirty-two patients
mucosal mass in the terminal ileum. Two patients, one of whom
(25%) had radiographic evidence for sources of bleeding in the small was reported previously [6], had anteriovenous malformations in
intestine. The medical, surgical, and pathologic records of these the proximal jejunum, manifested by the relatively subtle finding
patients were reviewed to determine if the clinical and/or pathologic of slightly lobulated, focal widening of part of an otherwise non-
findings confirmed the radiographic findings. Pathologic records of the mal small-bowel fold (Fig. 3). One patient had a mural
remaining 96 patients were also reviewed to determine if any of these hematoma, and one had extensive jejunal disease due to throm-
patients had lesions in the small intestine that could have been bosis of the superior mesenteric and portal veins associated with
missed at enteroclysis. However, as only one of the patients with no an idiopathic hypencoagulability state (Fig. 4). One patient had a
lesions seen on enteroclysis examinations underwent surgery, it was
Meckels diventiculum that contained ectopic gastric mucosa.
impossible to determine the sensitivity of enteroclysis in detecting
One patient with heme-positive stool was found at enteroclysis to
lesions in the small intestine in this group of patients.
have celiac disease, a condition that may be associated with
occult gastrointestinal bleeding and iron malabsonption [7]. One
Results patient had typical findings of radiation entenopathy involving the
distal part ofthe ileum aftentreatmentfor cervical carcinoma. The
Thirty-two (25%) of the 128 patients had lesions diagnosed
remaining patient had a benign ulcer in the fourth portion of the
at entenoclysis that were considered to be responsible for the
duodenum that responded to medical treatment.
gastrointestinal bleeding. Ten patients (8%) had other findings
at enteroclysis that may have been incidental and unrelated
to gastrointestinal bleeding. The remaining 86 patients (67%) Highly Probable Lesions
had normal findings on enteroclysis examinations.
Nineteen of the 32 patients with lesions at entenoclysis had Of the eight patients with unconfirmed lesions at enteno-
pathologic confirmation of the radiographic diagnosis by sun- clysis, four were considered to have secondary malignant
gery, angiography, on subsequent enteroscopy. In another five tumors involving the small bowel, including two with malig-
patients, subsequent surgery failed to corroborate the radio- nant melanoma, one with a Ieiomyosancoma, and one with
graphic diagnosis. These five examinations with false-positive gallbladder carcinoma. Two patients had benign-appearing
results were thought to have shown one ulcerated nodule, one polyps in the distal part of the duodenum, one of which was
polyp, one annular lesion, one case of ischemia or hemon- pedunculated (Fig. 5). One patient had a probable artenio-
rhage, and one Meckels diverticulum. However, these five venous malformation in the proximal jejunum, manifested at
patients had technically suboptimal examinations. The remain- enteroclysis by focal widening and lobulation of part of a
ing eight patients had cessation of bleeding without pathologic small-bowel fold. The remaining patient had thickened folds
corroboration of the findings at entenoclysis, but the clinical in an ileal segment of limited distensibility, presumably due to
presentation and course supported the radiographic diag- a vasculitis associated with underlying lupus enythematosus.
noses. Thus, 27 (21%) of the 128 patients with obscure gas-
trointestinal bleeding had a confirmed on highly probable
Incidental Lesions
lesion shown by entenoclysis, which could have been the
cause of the bleeding. Seventeen patients (13%) had tumors In 1 0 patients, entenoclysis revealed abnormalities that
involving the small bowel, and three (2%) had arteniovenous were not thought to be a cause of gastrointestinal bleeding,
malformations in the jejunum. In 21 (78%) of the 27 patients including jejunal diverticula in eight patients, intestinal malno-
with lesions shown by enteroclysis, the lesions were in the tation in two, a nonulcenated submucosal tumor in one, and a
 AJR:163, December 1994 ENTEROCLYSIS IN THE EVALUATION OF OBSCURE GASTROINTESTINAL BLEEDING 1383
Downloaded from www.ajronline.org by 68.100.197.246 on 04/04/17 from IP address 68.100.197.246. Copyright ARRS. For personal use only; all rights reserved

Fig. 1.-Lymphoma detected at enteroclysis. Fig. 2.-Carcinoid detectedatenteroclysis. This Fig. 3.-Arteriovenous malformation detected
This 81-year-old man with celiac disease presented 33-year-old man had gastrointestinal bleeding. En- at enteroclysis. This 79-year-old man reported in-
with diarrheaand gastrointestinal bleeding. Entero- teroclysis shows 1.8-cm rounded, ulcerated termittent bleeding of lower gastrointestinal tract
clysis shows findings of celiac disease, with wide (arrow) tumor, causing distortion of surrounding Enteroclysis shows 5-mm lobulated widening of
separation of folds in jejunum. An annular, ulcerat- small-bowel folds. At laparotomy, evidence of fold (arrows) in proximaijejunum. Diagnosis of ar-
ed lesion isvlsible, oneofthreesuch iesionsin this spread to mesenteric nodes was found. teriovenous malformation was suggested on be
patient Laparotomy was performed with excision sis of radiographic findings. Enteroscopy con-
of all three tumors. Histologic diagnosis was non- firmed presence of lesion, which was fulgurated.
Hodgkins lymphoma, the mOStfreqUent malignant Patient had another episode of gastrointestinal
tumor complicating adult celiac disease. bleeding 6 months later, presumably because of
additionalvascular malformations more distally in
small bowel.

Fig. 4.-Mural hematoma detected at enteroclysis. This 58-year-old man complained of abdomi- Fig. 5.-Pedunculated polyp detected at en-
nal pain and gastrointestinal bleeding. teroclysis. This 72-year-old woman had a long
A, Enteroclysis shows segment of jejunum with thumbprinting (curved arrows). More extensive history of gastrointestinal bleeding. Fourth por-
indentation is seen on mesenteric border of bowel (straight arrow), probably due to hematoma cx- tion of duodenum was visualized at conclusion
tending into mesentery. of otherwise normal enteroclysis examination.
B, MR image shows evidence of thrombosis of superior mesenteric and portal veins. Ti-weighted Note 2-cm polyp (large arrows) on 1.5-cm stalk
spin-echo MR image shows flowing blood as dark and thrombus (arrow) as bright. Eventual diag- (small arrows) near ligament of Treltz. No surgi-
nosis was idiopathic hypercoagulability state. cal or endoscopic confirmation was obtained.

single small polyp in
even, the percentage
study group (6%) was
ula found among the
(2%). This observation
Discussion
In about 95% of all patients with gastrointestinal bleeding,
the bleeding site can be detected
the distal jejunum in one patient. How- endoscopy of the upper gastrointestinal tract and the colon
of patients with jejunal diverticula in the [8]. In the remaining 5%, these diagnostic procedures fail to
greater than the percentage of divertic- establish a cause for the bleeding. Often this obscure bleed-
630 patients who did not have bleeding ing originates in the small intestine, posing a difficult diag-
may suggest a causal relationship. nostic problem. Scintigraphy, angiognaphy, entenoclysis, and
enteroscopy are all procedures that have been used in an
attempt to show these bleeding sites in the small intestine.
Scintignaphy can be done when a Meckels diverticulum is
suspected on clinical grounds. In adults, this abnormality often
by barium studies and/on goes undetected on scintignams [8], but can be shown by cane-
 1384 MOCH ET AL. AJA:163, December 1994

fully performed entenoclysis [9]. However, a Meckels diventicu- were shown by enteroclysis in 27 of 128 patients (21 %).
lum is an unlikely cause of gastrointestinal bleeding in patients Although anteriovenous malformations were found in only three
more than 40 years old [10]. Angiognaphy also may be used to ofthose 1 28 patients (2%), tumors were found in 17 (13%). En-
show bleeding sites in the small intestine [11], but this invasive tenoscopy is therefore a considerably more sensitive technique
procedure usually is not part of the routine workup of patients than entenoclysis for detecting anteriovenous malformations in
with low-grade on intermittent gastrointestinal bleeding. the small intestine. In contrast, entenoclysis is capable of detect-
Downloaded from www.ajronline.org by 68.100.197.246 on 04/04/17 from IP address 68.100.197.246. Copyright ARRS. For personal use only; all rights reserved

The role of entenoclysis in evaluating obscure gastrointesti- ing about three times as many tumors involving the small bowel
nal bleeding was first reported by Maglinte et al. in 1985 [2]. as entenoscopy. This is probably related to the limited length of
Rex et al. [3] subsequently showed thatthe diagnostic yield of small bowel traversed by the push-type entenoscope, whereas
entenoclysis in these patients was about 10%. In a subgroup the entire small bowel can be visualized at entenoclysis.
of their patients with normal findings on barium studies and/or In conclusion, carefully performed enteroclysis allows
endoscopy of the upper gastrointestinal tract and the colon, detection of many lesions in the small intestine that cause
however, the diagnostic yield of entenoclysis was reported to gastrointestinal bleeding. If enteroclysis fails to demonstrate a
have approached 20%. The authors concluded that entenocly- lesion as a likely source of bleeding, an arteniovenous malfor-
sis was a valuable diagnostic test in patients with unexplained mation is the probable cause, and entenoscopy may be
gastrointestinal bleeding. required for diagnosis and treatment of these vascular malfor-
In our study, lesions were found at enteroclysis in 27 of 128 mations. Thus, we believe that entenoclysis is a valuable diag-
patients (21 %) with obscure gastrointestinal bleeding. Tumors nostic modality and that it should precede entenoscopy in the
were found in 1 7 patients, comprising 63% of the patients with examination of patients with obscure gastrointestinal bleeding.
lesions shown by entenoclysis. Fourteen of those patients had
malignant tumors. Three patients had arteniovenous malfon-
REFERENCES
mations, which made up 11 % of the lesions seen at entenocly-
sis. Our study corroborates the findings of earlier investigators 1 . Bunker CJ. In: Scully AE, ed. Case records of the Massachusetts General
[2, 3], and shows that entenoclysis is a useful diagnostic tech- Hospital, Case 1 5-1 993. N EngI J Med 1993;328:11 07-1114
2. Maglinte DDT, Elmore MF, Chernish SM, et al. Enteroclysis in the diagno-
nique in patients with unexplained gastrointestinal bleeding.
sis of chronic unexplained gastrointestinal bleeding. Dis Colon Rectum
However, about 22% of the bleeding lesions were located in 1985;28:403-405
the fourth portion ofthe duodenum near the ligament of Tneitz. 3. Rex DK, Lappas JC, Maglinte DDT, et al. Enteroclysis in the evaluation of
This indicates the importance of modifying the technique of suspected small intestinal bleeding. Gastroenterology i989;97:58-60
enteroclysis in patients with obscure bleeding to evaluate the 4. Herlinger H. A modified technique for the double contrast small bowel
enema. Gastrointest Radiol 1978;2:307-400
distal part of the duodenum and the duodenojejunal flexune,
5. Herlinger H. Barium examinations. In: Gore AM, Levine MS, Laufer I, eds.
areas that may not be visualized fully during routine endos- Textbook ofgastrointestinalradiology. Philadelphia: Saunders, 1994:771-788
copy on barium studies of the upper gastrointestinal tract. 6. Herlinger H, Levine MS, Furth EE, Moonka D. Arteriovenous malformation of
In recent years, peroral entenoscopy has become a tech- the small bowel diagnosed with enteroclysis. AJR 1992;159:1225-1226
7. Losowsky MS, Walker BE, Kelleher J. Malabsorption in clinical practice.
nique used increasingly for examining patients with obscure
Edinburgh: Churchill Livingstone, 1974:234
gastrointestinal bleeding. Numerous publications describe the 8. Thompson JN, Hemingway AP, McPherson GAD, et al. Obscure gastrointes-
instruments and techniques used [12-20]. The sonde-type tinal haemorrhage of small-bowel origin. BMJ 1984;288:1663-1 665
entenoscope requires 6-8 hr to reach the ileum and has been 9. Maglinte DDT, Herlinger H. In: Herlinger H, Maglinte DDT, eds. Clinical
largely replaced by the newer push-type entenoscope, which radiology of the smallintestine. Philadelphia: Saunders, 1989:249-273.
10. Leijonmarck CE, Bonman-Sandelin K, Frisell J, et al. Meckels diverticu-
has high-resolution video capability and a channel for forceps lum in the adult. BrJ Surg 1986;73:146-149
biopsy on cauterization [20]. This type of instrument is now 11. Rollins ES, Picus D, Hicks ME, et al. Angiography is useful in detecting
routinely in use at our institution. the source of chronic gastrointestinal bleeding of obscure origin. AJR
Currently, no consensus exists among gastroenterologists and 1991156:386-388
12. Lewis BS, Waye JD. Chronic gastrointestinal bleeding of obscure origin:
radiologists about the respective roles of enteroclysis and ent-
role of small bowel enteroscopy. Gastroenterology 1 988;94:111 7-1120
enoscopy in the evaluation of patients with obscure gastrointesti- 1 3. Lewis BS, Wenger JS, Waye JD. Small bowel enteroscopy and intraopera-
nal bleeding. Some authors favor enteroclysis, if it is available, as tive enteroscopy for obscure small intestinal bleeding. Am J Gastroenterol
the next diagnostic test when barium studies and/on endoscopy 1991 86:171-1 74
14. Lewis BS, Kombluth A, Waye JD. Small bowel tumors: yield of enteros-
of the upper gastrointestinal tract and the colon have failed to
copy. Gut 199132:763-765
show a potentialsource of bleeding [2,3, 16-18]. Others prefer 1 5. Gostout CJ, Schroeder KW, Burton CGC. Small bowel enteroscopy: an
entenoscopy as the first diagnostic procedure and do not even early experience in gastrointestinal bleeding of unknown origin. Gas-
mention entenoclysis as a practical alternative [12, 14, 15]. trointestEndosc 1991 37:5-8
Recently published data and our own experience suggest that 16. Bowden TA Jr. Endoscopy of the gastrointestinal tract. Surg Clin North
Am 1989;69:1237-1247
entenoclysis and entenoscopy are complementary techniques for
17. Foutch PG, Sawyer A, Sanowski RA. Push enteroscopy for diagnosis of
examining patients with obscure gastrointestinal bleeding. Three patients with gastrointestinal bleeding of obscure origin. Gastrointest
recent publications [17, 19, 20] have together reported a total of Endosc 1990;36:337-341
98 push-type entenoscopies carried out for the investigation of 18. Krevsky B. Enteroscopy: exploring the final frontier. Comment. Gastroen-
obscure gastrointestinal bleeding. Lesions were demonstrated in terology 1991100:838-839
19. Barkin JS, Lewis BS, Reiner DK, et al. Diagnostic and therapeutic jejunos-
55 of those 98 examinations (56%). Arteriovenous malforma-
copy with a new, longer enteroscope. Gastrointest Endosc i992;38:55-58
tions were found in 39 of the 98 examinations (40%), tumors in 20. Harris A, Dabezies MA, Catalano MF, Krevsky B. Eariy experience with a
four (4%), and other lesions in 12 (12%). In our study, lesions video push enteroscope. Gastrointest Endosc 1 994;40:62-64