Negative capsule endoscopy in patients with obscure GI bleeding predicts low rebleeding rates

ORIGINAL ARTICLE: Clinical Endoscopy

Negative capsule endoscopy in patients with obscure GI bleeding predicts low rebleeding rates Jonathan Macdonald, BSc (Hons), BM (Hons), MRCP, Victoria Porter, MSc, Deirdre McNamara, MD, MRCPI, FRCP (Edin) Aberdeen, Scotland, UK

Background: Wireless capsule endoscopy (CE) is the investigation of choice in obscure GI bleeding (OGIB), with a high diagnostic yield when compared with other modalities. It enables specific treatment to be instigated in a significant proportion of cases, with the aim of reducing rebleeding rates. Little evidence is currently available on the clinical value of a negative study. Objective: To determine the long-term (O1 year) outcome in patients who underwent CE to investigate OGIB. Methods: Consecutive patients with OGIB referred for CE to a single center over an 18-month period were identified. Follow-up data were obtained by reviewing case notes and an internal CE database. Rates of rebleeding were established, and factors associated with rebleeding were assessed by means of univariate and multivariate analysis. Results: Forty-nine patients underwent CE for investigation of OGIB (obscure overt in 25 patients [51%]) between April 2005 and October 2006. Long-term data were available for 42 patients (86%), with a mean ( SD) follow-up of 17.3  6.2 months. Significant (P2) lesions were identified in 24 patients (57%). The overall rebleeding rate was 28%. There was a statistically significant difference in rebleeding between patients with a positive and patients with a negative study, 42% versus 11%, respectively, P ! .01. Anticoagulant use was also associated with an increased risk of rebleeding. Conclusion: A negative CE study in patients with OGIB is associated with a low rate of recurrent bleeding in the long term (11%). It is reasonable to take an expectant approach with these patients, thus avoiding the need for unnecessary additional investigations. (Gastrointest Endosc 2008;68:1122-7.)

Wireless capsule endoscopy (CE) is a highly effective means of examining the entire small bowel. Its clinical use as a diagnostic tool in patients with obscure GI bleeding (OGIB) has been proven in several prospective trials. CE has been shown to be superior to alternative diagnostic tools, including push enteroscopy (PE) and radiologic examinations in this setting. In the meta-analysis by Triester et al,1 the diagnostic yields for CE versus PE were calculated as 63% versus 36%, respectively, and, for CE versus small-bowel barium studies, the diagnostic yields were 56% versus 26%, respectively. The estimated number of patients needed to test to yield one additional Abbreviations: APC, argon plasma coagulation; OGIB, obscure GI bleeding; PE, push enteroscopy. See CME section; p. 1147. Copyright ª 2008 by the American Society for Gastrointestinal Endoscopy 0016-5107/$34.00 doi:10.1016/j.gie.2008.06.054

1122 GASTROINTESTINAL ENDOSCOPY Volume 68, No. 6 : 2008

significant finding for CE compared with the 2 other modalities was only 3. Based on such evidence, CE is recommended as a firstline investigation for OGIB, after at least 1 negative gastroscopy and colonoscopy. The overall effect on patient management is still debated. The long-term value of CE studies remains a key determinant of its relevance to clinical practice. In an article by Saurin et al2 that examined this question, it was suggested that, although CE was highly sensitive when the clinical outcome at 1 year was used as the criterion standard, it was less specific than PE (48% vs 69%). These data suggest that a negative CE examination does not accurately predict a good long-term outcome, implying that some patients at significant risk of rebleeding or who may benefit from further investigation or treatment will be overlooked. This is highly relevant, because it suggests that a negative CE report should instigate additional investigation, often by means of more-invasive procedures. There is some evidence to the contrary. Additional data www.giejournal.org

Macdonald et al

on the long-term validity and efficacy of CE are required to address this issue. The aim of our study was to determine the long-term outcome (O1 year) in patients who underwent a CE as an investigation for OGIB.

PATIENTS AND METHODS Over an 18-month period, consecutive patients with OGIB referred for CE to a single center and with a subsequent follow-up after CE of at least 12 months were identified. OGIB was defined as having evidence of melena, hematochezia, or a drop in hemoglobin (Hb) of at least 2 g/dL and a positive fecal occult blood test. All patients had both a negative gastroscopy and colonoscopy, with a negative barium enema if the colonoscopy was incomplete. CE studies were performed according to our unit’s protocol, which includes an overnight fast, the use of bowel preparation (2 sachets of sodium picosulphate 10 mg), and a prokinetic agent (metoclopramide 10 mg) for patients with longstanding diabetes mellitus or known slow transit. Studies were read separately by 2 individuals with experience in CE (D.M., V.P.). If discrepancies occurred, then findings were reviewed by both examiners simultaneously, and a consensus was reached. Follow-up data were obtained by case note review and review of an internal CE database. Patient demographics were recorded, including sex, age, obscure overt (melena, hematochezia) or obscure occult bleeding, and current medications. CE findings were classified according to standard practice as highly relevant lesions (P2) or less-relevant lesions (P1, P0). An abnormal finding was classified as a P2 lesion when it was considered to be the cause of or explanation for OGIB. When a definite abnormality was identified but was not thought to be the cause of or explanation for blood loss, it was assigned a P1 status. Minor mucosal changes or abnormalities that were not diagnostic were also categorized as P1 lesions. Examinations that demonstrated one or more P2 lesions were recorded as positive studies, whereas those with only P1 lesions or no abnormality at all (P0) were negative. The main outcome measurement was rebleeding. A recurrent bleeding episode was defined as evidence of recent or active bleeding at least 30 days after the index bleed. Again, bleeding was defined as a documented fall in Hb of 2 g/dL from baseline, evidence of melena or hematochezia, and the need for blood transfusion.

Statistics Actuarial rates of rebleeding during follow-up were calculated, and factors associated with rebleeding were assessed by means of univariate and multivariate analysis. A P value !.05 was considered significant.

RESULTS In all, between April 2005 and October 2006, 49 patients underwent 50 CE studies for investigation of www.giejournal.org

Predicting low rebleeding rates in obscure GI bleeding

Capsule Summary What is already known on this topic d

Because of its high diagnostic yield, capsule endoscopy (CE) is the procedure of choice in obscure GI bleeding (OGIB).

What this study adds to our knowledge d

Long-term data in 42 patients who underwent CE for investigation of OGIB showed that a ‘‘negative’’ study was associated with only an 11% rate of recurrent bleeding, whereas a ‘‘positive’’ study was associated with a 42% rate of recurrence.

OGIB. Complete visualization of the small bowel was achieved in 46 cases (93%). A repeated procedure, with the addition of a prokinetic, was performed in 1 patient. In total, 25 patients (51%) were referred with obscure overt bleeding. A significant abnormality thought to be a source of bleeding was found in 27 (55%). One-year follow-up data were available in 42 patients (86%): 19 women (45%), mean (SD) age 59  13 years, range 21 to 88 years. In 4 cases, patients had been referred from hospitals outside our geographical health care region, and additional clinical information was unavailable, and 3 patients were lost to follow-up. Twenty-four patients (57%) were referred with evidence of overt bleeding. All 42 subjects had a negative gastroscopy, and 41 (98%) had a negative colonoscopy within 1 month of CE. The completion rate for CE was 100% in the 42 patients with more than 12 months of follow-up. Twenty-four patients (57%) had a positive study that demonstrated one or more P2 lesions. Angiodysplastic lesions were detected in the majority (19 [79%]); 2 patients had visible small-bowel ulcers (8%), 2 had small-bowel tumors (8%), and 1 had a polypoidal lesion (4%) (Fig. 1). There was no statistical difference in the prevalence of P2 lesions according to presentation, with 15 of 24 patients with overt bleeding having a positive CE (63%) compared with 9 of 18 patients who underwent CE for occult blood loss (50%) (Table 1). Of the 18 negative studies, 13 had no abnormality detected (P0), the other 5 patients had P1 lesions (3 red spots and/or minor mucosal changes, 1 red spot and multiple atheromas, and 1 aphthous ulceration and nodular hyperplasia). The mean (SD) duration of follow-up was 17.3  6.2 months. Twelve patients (28%) had one or more rebleeding episodes during the follow-up period. This group comprised 10 patients whose CE was positive and 2 with a negative CE (Fig. 1). There was a statistically significant difference in rebleeding between those with a positive and a negative CE study, 42% (10/24) versus 11% (2/18), P ! .01. No patients with P1 lesions at CE rebled. Volume 68, No. 6 : 2008 GASTROINTESTINAL ENDOSCOPY 1123

Predicting low rebleeding rates in obscure GI bleeding

Macdonald et al

Figure 1. Study population by presentation, CE result, and rebleeding events.

TABLE 1. Study population Total population No. CE

49

Complete small-bowel visualization, no. (%)

46 (93)

Diagnostic yield, no. (%)

27 (55)

O12-mo follow-up, no. (%)

42 (86)

Women, no. (%)

19 (45)

Mean  SD age, y

59  13

Obscure overt bleeding, no. (%)

24 (57)

P0 or P1 lesions detected, no. (%)

18 (43)

Figure 2. CE in patients with OGIB.

The initial mode of presentation (overt vs occult) was not a predictor of recurrent bleeding. Of 30 patients with bleed-free follow-up, 16 initially presented with overt bleeding (53%), whereas the group of patients with rebleeding comprised 8 who were initially referred with obscure overt bleeding (67%), and 4 with occult loss of blood (33%) (Fig. 2). Subjects with recurrent bleeding tended to be older, 66 years versus 57 years, although the difference did not reach statistical significance (P ! .06). The mean (SD) transfusion requirement was 7.3  4 units of red cell concentrate (range 2-18 units).

Fifty percent of patients with a positive study (n Z 12) received specific treatment. The majority (8 [67%]) received argon plasma coagulation (APC); additional treatments included surgery in 2 patients (17%) and 2 received medical therapy. Six patients (14%) were prescribed anticoagulation with warfarin, and its use was strongly associated with an increased risk of rebleeding. Of 12 patients who rebled, 33% (4/12) received warfarin treatment throughout follow-up versus 7% (2/18) who did not rebleed (P !.05, odds ratio 7.8). There was no association found between the use of low-dose aspirin (n Z 4) and bleeding. The overall mortality rate during this period was 8.5% (n Z 4); 3 subjects had positive CE studies, 2 had angiodysplasia, and one had cancer. Bleeding was not the primary cause of death in any subject; there was 1 subacute bacterial endocarditis, 1 hepatorenal syndrome, 1 hematologic malignancy, and 1 cancer-related death.

1124 GASTROINTESTINAL ENDOSCOPY Volume 68, No. 6 : 2008

www.giejournal.org

Anticoagulant use, no. (%)

6 (14)

Macdonald et al

Predicting low rebleeding rates in obscure GI bleeding

TABLE 2. Long-term follow-up in patients with OGIB No. patients

Follow-up (mo)

% Total recurrent bleeding

% Recurrent bleeding and negative diagnostic test

35

12

23

0

30

12

17

N/A

47

12

26

N/A

Kovacs et al, 2006

66

20

18

N/A

10

49

12

33

6

44

12

11

0

105

29

31

N/A

49

17

28

11

Study Lorenceau-Savale et al,11 2007* 7

Redondo-Cerezo et al, 2007* 5

Hartmann et al 2007 6

Lai et al,

2006 9

Delvaux et al, 2005 8

Landi et al, 2002y Macdonald et al, 2008z N/A, Not applicable. *Abstract presentation. yPE study. zCurrent study.

DISCUSSION In patients with OGIB, the bleeding source is frequently located in the small bowel (30%-40%).3 Angiodysplasia accounts for a significant proportion of cases, approximately 70% to 80%.4 The natural history of these lesions is poorly understood, but it has been estimated that fewer than 50% will rebleed. CE is an established tool for visualization of the entire small bowel and may assist in the follow-up evaluation of patients with OGIB. The clinical value of CE largely depends on the efficacy of management of diagnosed lesions and on the negative predictive value of a CE test. In this study, we evaluated the efficacy of CE by determining the long-term rebleeding rates in patients with OGIB. In our group of patients, CE was found to be highly effective: 93% of studies were complete, the diagnostic yield was 55%, and there were no associated complications. The overall rebleeding rate with a mean of 17 months of follow-up for this population of patients with OGIB was 28%. This is in keeping with previously reported figures. Published overall rebleeding rates range from 6% to 31%.5-8 The overall population size in most studies was small and was 49 in this study. This may affect the generalization of findings. However, the considerable degree of homology between most studies to date regarding overall rebleeding rates would tend to refute this and further add validity to our findings. We found a statistically significant difference in the rebleeding rates between positive (42%) and negative studies (11%): P ! .01. These data suggest that the specificity and negative predictive value of CE are valid in the long term. The negative predictive value of CE over the duration of our study was 89%. Few previous studies clearly documented the outcome in patients with a negative basewww.giejournal.org

line CE study.9-11 Documented rebleeding rates range from 0% to 6%, which is lower than what was found in this population (Table 2). The duration of follow-up in both studies, which found no rebleeding in patients with a negative CE tests, was only 12 months. The longer follow-up in our patient group and in the study of Lai et al10 (17 and 19 months, respectively) may explain the increased rebleeding rates seen. In addition, in the study by Lorenceau-Savale et al,11 published in abstract form, 1-year follow-up data were only available in 35 patients of a total population of 124. This may represent a source of bias. In contrast, O12-month follow-up data were available for 86% of our population. The fact that our rebleeding rate following negative CE was higher than the 6% reported by Lai et al10 could be a result of differences in population or subsequent management. Both studies included 49 patients. The diagnostic yield for CE in the study by Lai et al10 was only slightly higher (63% vs 57%), and the overall rate of recurrent bleeding was 32.7%, similar to the 28% described in this article. It, therefore, is reasonable to presume that both studies report on similar populations; however, more invasive interventions were undertaken in the population in the study by Lai et al,10 with almost a third undergoing additional PE or laparotomy. The possibility exists that this more-invasive approach led to additional small-bowel diagnoses, which reduced the potential number of false negatives that may have occurred if CE alone were used as in our study. As expected, the majority of patients with a positive test had angiodysplasia (n Z 19 [79%]). The long-term outcome of patients with small-bowel angiodysplasia remains uncertain, and the clinical benefit derived from interventional endoscopic therapy is debated. There appears to be a poor correlation between the number, size, and Volume 68, No. 6 : 2008 GASTROINTESTINAL ENDOSCOPY 1125

Predicting low rebleeding rates in obscure GI bleeding

Macdonald et al

TABLE 3. Classification of angiodysplastic lesions detected at CE

Patient no.

No. lesions

Size of lesions

Anatomical location

Active and/or recent bleeding?

Treatment modality

Rebleeding?

1

2

Small

Distal

No

APC

Yes

2

1

Large

Proximal

No

Medical

No

3

1

Small

Proximal

No

APC

No

4

Unknown

Unknown

Distal

No

None

Yes

5

Multiple

Small

Diffuse

No

None

No

6

1

Small

Proximal

No

APC

Yes

7

Multiple

Small

Diffuse

No

None

Yes

8

1

Small

Proximal

Yes

APC

Yes

9

Multiple

Small

Proximal

No

APC

No

10

Multiple

Large

Proximal

No

None

Yes

11

Multiple

Small

Diffuse

No

APC

Yes

12

1

Large

Proximal

No

None

No

13

1

Small

Distal

Yes

None

No

14

1

Small

Distal

No

None

Yes

15

Multiple

Small

Diffuse

No

None

No

16

Multiple

Small

Diffuse

No

Medical

No

17

Multiple

Small

Proximal

No

APC

No

18

Multiple

Small

Diffuse

No

APC

Yes

19

1

Large

Distal

No

None

No

Small, !5 mm; Large, O5 mm; Medical, medical treatment.

location of angiodysplastic lesions and the propensity for recurrent episodes of GI bleeding. The lesions identified in our patient group are classified in more detail in Table 3. There were no features or characteristics identified that predicted the likelihood of rebleeding events in this group. Ten patients (53%) received treatment for angiodysplastic lesions: 8 APC and 2 medical therapy (1 patient bismuth sucrulfate, 1 patient high-dose proton pump inhibitor). Rebleeding rates did not vary significantly after treatment. Of 10 patients with positive CE findings who had recurrent episodes of bleeding, 9 (90%) had a CE diagnosis of angiodysplasia. Five of these patients (56%) received tailored treatment, and 4 (44%) did not. This is not wholly unexpected, and these rebleeding rates are consistent with previous reports, which use a variety of treatment modalities. Almost a third of patients relapsed over the study period, and there was a substantial transfusion requirement (average 7.3 units of blood per patient); however, there was zero bleeding-related mortality. Of note, the mode of presentation (occult vs overt OGIB) did not seem to influence the likelihood of rebleeding in the long term. This

supports a paradigm in which overt and occult cases are managed in a similar fashion. It would be of interest to determine whether the number of previous episodes of bleeding had an effect on outcome. However, this was beyond the scope of our study, because patients with obscure bleeding were referred from a variety of specialties and often had a protracted history. Clear documentation of previous events was often lacking. In addition to the main findings, we also demonstrated a clearly increased risk of rebleeding in patients who were taking warfarin. This is unsurprising, because anticoagulation is a well-recognized risk factor for bleeding peptic ulcers and is likely to have the same effects and mode of action in a population with small-bowel bleeding, mainly as a result of angiodysplasia.12 In conclusion, CE is a highly effective diagnostic tool in cases of OGIB, and its early use is recommended. A benefit of CE is that it accurately predicts the risk of future bleeding. Patients with a negative CE test have a low risk of relapse. Patients can be reassured and an expectant approach taken whereby additional invasive investigations are not performed without further clinical indications.

1126 GASTROINTESTINAL ENDOSCOPY Volume 68, No. 6 : 2008

www.giejournal.org

Macdonald et al

Unfortunately, the majority of patients with a positive study have angiodysplasia and almost half will have recurrent episodes of bleeding despite intervention. DISCLOSURE The authors report that there are no disclosures relevant to this publication.

REFERENCES 1. Triester SL, Leighton JA, Leontiadis GI, et al. A meta-analysis of the yield of capsule endoscopy compared to other diagnostic modalities in patients with obscure gastrointestinal bleeding. Am J Gastroenterol 2005;100:2407-18. 2. Saurin JC, Delvaux M, Vahedi K, et al. Clinical impact of capsule endoscopy compared to push enteroscopy: 1-year follow-up study. Endoscopy 2005;37:318-23. 3. Raju GS, Gerson L, Das A, et al. American Gastroenterological Association (AGA) Institute Medical position statement on obscure gastrointestinal bleeding. Gastroenterology 2007;133:1694-6. 4. Foutch PG. Angiodysplasia of the gastrointestinal tract. Am J Gastroenterol 1993;88:807-18. 5. Hartmann D, Schmidt H, Schilling D, et al. Follow up of patients with obscure gastrointestinal bleeding after capsule endoscopy and interoperative enteroscopy. Hepatogastroenterology 2007;54:780-3. 6. Kovacs M, Nemeth A, Pak P, et al. Evaluation of diagnostic yield and clinical impact of capsule endoscopy in patients with obscure gastro-

www.giejournal.org

Predicting low rebleeding rates in obscure GI bleeding

7.

8.

9.

10.

11.

12.

intestinal bleeding [Hungarian with English abstract]. Orv Hetil 2006;147:1827-33. Redondo-Cerezo E, Gomez-Ruiz CJ, Sanchez-Manjavacas N, et al. Long term follow up of patients with small bowel angiodysplasia on capsule endoscopy. Determinants of a higher clinical impact and rebleeding rate [abstract]. Endoscopy 2007;39(Suppl 1):A385. Landi B, Cellier C, Gaudric M, et al. Long-term outcome of patients with gastrointestinal bleeding of obscure origin explored by push enteroscopy. Endoscopy 2002;34:355-9. Delvaux M, Fassler I, Gay G. Clinical usefulness of the endoscopic video capsule as the initial intestinal investigation in patients with obscure digestive bleeding: validation of a diagnostic strategy based on the patient outcome after 12 months. Endoscopy 2005;37:1067-73. Lai LH, Wong GL, Chow DK, et al. Long-term follow up of patients with obscure gastrointestinal bleeding after negative capsule endoscopy. Am J Gastroenterol 2006;101:1224-8. Lorenceau-Savale C, Ben Soussan E, Ramirez S, et al. Obscure gastrointestinal bleeding after negative capsule endoscopy: results of a 1 year follow up [abstract]. Endoscopy 2007;39(Suppl 1):A385. Hallas J, Dall M, Andries A, et al. Use of single antithrombotic therapy and the risk of serious gastrointestinal bleeding: population based case-control study. BMJ 2006;333:726-30.

Received February 21, 2008. Accepted June 23, 2008. Current affiliations: Department of Gastroenterology, Aberdeen Royal Infirmary, Aberdeen, Scotland, U.K. Reprint requests: Jonathan Macdonald, BSc, Department of Gastroenterology, Aberdeen Royal Infirmary, Foresterhill Rd, Aberdeen AB25 2ZN, U.K.

Volume 68, No. 6 : 2008 GASTROINTESTINAL ENDOSCOPY 1127