Usefulness of small bowel reexamination in obscure gastrointestinal bleeding patients with negative capsule endoscopy findings: Comparison of repeat capsule endoscopy and double-balloon enteroscopy

Abstract

Background

Although capsule endoscopy (CE) is a noninvasive diagnostic tool for patients with obscure gastrointestinal bleeding (OGIB), bleeding lesions are often not detected. No strategies have been established to determine whether CE or double-balloon enteroscopy (DBE) should be performed after negative CE.

Methods

Among 652 patients who underwent CE for OGIB, a total of 359 patients had negative findings. Of these, 41 and 48 patients underwent repeat CE and DBE for small bowel reexamination, respectively. We compared the rate of positive findings between the two groups. The findings were considered positive if mucosal break, vascular lesion, tumor, or active bleeding was observed.

Results

The rate of positive findings in the repeat CE group was significantly higher than that in the DBE group (73.2% vs. 39.6%; p = 0.001). Logistic regression analysis showed that the repeat CE group was significantly associated with positive findings (odds ratio (OR), 4.2; 95% confidence interval (CI), 1.7–10.2; p = 0.002). Inverse probability of treatment-weighted analysis revealed that repeat CE was significantly associated with positive findings (OR, 4.4; 95% CI, 1.6–12.2; p = 0.004).

Conclusions

Reexamination via CE appears to have more diagnostic value than DBE for OGIB patients with negative CE findings.

Key summary

1. Summarize the established knowledge on this subject.

   • Positive finding rates of repeat capsule endoscopy (CE) were significantly higher than that of double-balloon enteroscopy (DBE) in obscure gastrointestinal bleeding (OGIB) patients with negative CE findings.
   • Among the positive findings of repeat CE, mucosal break was the most common lesion, followed by vascular lesion, tumor, and active bleeding.
   • CE should be repeated even if the initial CE did not find any potential bleeding lesions in OGIB patients with suspected continuous bleeding.
2. What are the significant and/or new findings of this study?

   • Reexamination via CE has more diagnostic value than DBE for OGIB patients with negative findings in the initial CE.

Introduction

Obscure gastrointestinal bleeding (OGIB) is defined as gastrointestinal bleeding of an unknown source that persists despite the normal findings of esophagogastroduodenoscopy and colonoscopy. OGIB can be further categorized into overt OGIB, which presents as clinically evident bleeding (i.e. hematemesis, melena, or hematochezia), or occult OGIB, which indicates iron-deficiency anemia or a positive fecal occult blood test. Although most OGIB-associated lesions present within the small intestine, bleeding lesions cannot be identified for long periods of time because of the difficulty of physical accessibility.¹ Recently, the development of capsule endoscopy (CE) and double-balloon enteroscopy (DBE) has facilitated the process of diagnosis, and those procedures have evolved to become approaches to treating OGIB.^2,3 They have complementary roles in the management of OGIB. CE is minimally invasive and has a high diagnostic yield in patients with OGIB. On the contrary, DBE has the additional advantage of biopsy collection for pathological diagnosis and therapeutic intervention.

The American College of Gastroenterology clinical guideline recommends that CE should be considered first in order to determine the bleeding source of OGIB.⁴ Okamoto et al. considered that initial CE should be recommended in OGIB patients with no strong risk factors associated with the presence of bleeding lesions.⁵ The most common causes of bleeding in the small intestine in OGIB patients are erosions/ulcers and vascular lesions,^6–8 and CE is the best modality to detect these lesions. It is reasonable to perform CE early in the workup course of OGIB because positive CE findings could enhance the diagnostic yield of a subsequent DBE and serve as a guide for selecting the optimal DBE route.^9,10 Patients with positive CE findings can undergo therapeutic intervention such as hypertonic saline epinephrine injection, argon plasma coagulation, and hemostatic clipping. However, there are no clear guidelines for the further investigations of patients with negative CE findings, and no strategies have been established to date. Considering the low re-bleeding rate after negative CE findings,¹¹ some patients in stable general condition may require follow-up only. Further examinations should be considered in patients with recurrent bleeding or progressive anemia, but it is unknown whether CE should be repeated or DBE should be performed after the initial CE. In this study, we compared the rate of positive findings between repeat CE and DBE and aimed to propose a new strategy for OGIB patients with negative findings in the initial CE.

Methods

Study design

This is a single-center, retrospective, observational study.

Study population

Among 652 patients who underwent CE for OGIB from March 2004 to October 2015 at Osaka City University Hospital (Osaka, Japan), 18 patients (2.8%) were excluded for the following reasons: inability to swallow the capsule (n = 1); capsule retention at the esophagus (n = 3) and the stomach (n = 10); and machine trouble (error related to battery, adhesive sleeve, or recorder) (n = 4). Data from the remaining 634 patients were analyzed, and a total of 359 patients with negative findings in the initial CE were enrolled in this study. Repeat CE (n = 41) or DBE (n = 48) was performed in OGIB patients who had evidence of sustained bleeding or severe anemia. On the other hand, stable patients (n = 270) received observation only. A schematic diagram of this study design is shown in Figure 1.

Figure 1.

Schematic diagram of the study design.

OGIB: obscure gastrointestinal bleeding; CE: capsule endoscopy; DBE: double-balloon enteroscopy.

CE

CE procedure was performed using a Pill Cam® SB1 or SB2 Plus (Given Imaging Ltd, Yoqneam, Israel). Patients fasted for 12 hours before swallowing the capsule. Fluid was allowed two hours later, followed by a light meal another two hours later. Data were collected for up to eight hours after capsule ingestion. The data were downloaded to a workstation with RAPID® (Given Imaging Ltd) and reviewed by experienced endoscopists using the RAPID® software.

DBE

The DBE procedure was performed using two types of enteroscopes, the EN-450 P5 for regular use and the EN-450 T5 for therapeutic use (Fujifilm Corp, Tokyo, Japan). The 200-cm-long enteroscope has an attachable balloon at the tip, and a 145-cm-long overtube has another balloon on the distal end. These two inflatable balloons enable the insertion of the enteroscope by fixing and shortening the small intestine. An experienced endoscopist controlled the balloon with an automatic pump under monitoring of the balloon pressure, and an assistant operated the overtube. Patients were sedated using midazolam and pethidine hydrochloride. Per oral or per anal approach was used to evaluate the entire small intestine, and submucosal tattoo injection was performed in order to mark the maximum length of the small bowel inspected or the location of the lesion. For per anal examination, one-liter electrolyte lavage preparation (Niflec®; EA Pharma Co Ltd, Japan) was performed to clean the ileum and colon.

Assessments

The finding was considered positive if mucosal break, vascular lesion, tumor, or active bleeding was observed. Mucosal breaks include erosions/ulcers. Vascular lesions include angioectasia, Dieulafoy's lesion (arterial bleeding), and arteriovenous malformation. Tumors include benign lesions such as polyps as well as malignant lesions such as gastrointestinal stromal tumor, neuroendocrine tumor, and adenocarcinoma. Typical images of these bleeding lesions in CE are shown in Figure 2.

Figure 2.

Typical images of bleeding lesions using capsule endoscopy. (a) Mucosal break (arrow). (b) Vascular lesion (arrow). (c) Tumor (arrows). (d) Active bleeding.

Clinical factors, including age, sex, OGIB type (overt or occult), previous lowest hemoglobin (Hb) level, alcohol intake, smoking habit, nonsteroidal anti-inflammatory drug (NSAID) use, antithrombotic agent (antiplatelet drug or anticoagulant) use, proton pump inhibitor (PPI) use, Charlson comorbidity index (CCI), and interval between initial CE and repeat CE/DBE were assessed for association with positive findings in the small intestine. CCI is an indicator of comorbidity in which a weight is assigned to each category according to the features and severity of the comorbidity to yield a total score.¹²

Study end point

The primary endpoint was the rate of positive findings of small bowel reexamination in OGIB patients with negative CE findings. The secondary endpoint was clinical factors associated with the prevalence of positive findings.

Ethical approval

The study protocol was approved by the Osaka City University Graduate School of Medicine ethics committee (No. 2534) on June 23, 2016. Written, informed consent was obtained from each patient included in the study. The study protocol conforms to the ethical guidelines of the 1975 Declaration of Helsinki as reflected in prior approval by the institution's human research committee.

Statistical analysis

Continuous variables are expressed as median and interquartile range (IQR). Continuous variables were compared using t test, and categorical variables were compared using Pearson's chi-square test or Fisher's exact test. Clinical factors associated with positive findings were investigated using univariate and multivariate logistic regression analysis, and the odds ratio (OR) and 95% confidence intervals (CIs) were estimated for each variable. Furthermore, inverse probability of treatment-weighted (IPTW) analysis based on propensity scores was used to reduce selection bias by creating a “pseudo-population” in this study. After generating propensity scores using the clinical background factors (age, sex, OGIB type, previous lowest Hb level, alcohol intake, smoking habit, NSAID use, antiplatelet drug use, anticoagulant use, PPI use, CCI, and interval between initial CE and repeat CE/DBE) as covariates by logistic regression analysis, the inverse of the propensity score was used as the weight to adjust the clinical background factors between the two groups. The statistical calculations and analyses were performed using IBM SPSS Statistics, version 20 (IBM Corporation, Armonk, NY, USA) and R version 3.4.1 (The R Foundation for Statistical Computing, Vienna, Austria). P values < 0.05 were considered statistically significant.

Results

Comparison of clinical background factors and positive finding rates between the repeat CE group and the DBE group

In a total of 359 patients with negative findings in the initial CE, 41 and 48 patients underwent repeat CE and DBE for reexamination of the small intestine, respectively, and we compared the clinical background factors and positive finding rates between the two groups.

Among the clinical background factors, including age, sex, OGIB type, previous lowest Hb level, alcohol intake, smoking habit, NSAID use, antiplatelet drug use, anticoagulant use, PPI use, CCI, and interval between two inspection times, interval between initial CE and repeat CE was significantly longer than that between initial CE and DBE (121.0 days vs. 31.5 days, respectively; p = 0.003). There were more overt OGIB cases in the DBE group than in the repeat CE group, but the difference was not significant (75.0% vs. 56.1%, respectively; p = 0.060). Positive findings were identified in 30 cases of the repeat CE group and in 19 cases of the DBE group, and the positive finding rate in the repeat CE group was significantly higher than that in the DBE group (73.2% vs. 39.6%, respectively; p = 0.001).

Among the positive findings in the small intestine, mucosal break was the most common lesion (53.3%), followed by vascular lesion (23.3%), tumor (16.7%), and active bleeding (6.7%) in the repeat CE group. In contrast, mucosal break and vascular lesion were noted equally (31.6%), followed by tumor (26.3%) and active bleeding (10.5%) in the DBE group. Mucosal break was more frequently found in the repeat CE group than in the DBE group, but there were no significant differences in the bleeding lesions between the two groups (p = 0.519) (Table 1).

Table 1.

Baseline characteristics of the repeat capsule endoscopy (CE) group and the double-balloon enteroscopy (DBE) group.

Variables	Repeat CE (n = 41)	%	DBE (n = 48)	%	p value
Age (years, median (IQR))	65.0 (53.0–72.5)		63.0 (54.0–71.0)		0.949
Sex
Male	20	48.8%	26	54.2%	0.612
Female	21	51.2%	22	45.8%
OGIB type
Overt	23	56.1%	36	75.0%	0.060
Occult	18	43.9%	12	25.0%
Lowest Hb (g/dl, median (IQR))	7.5 (6.0–10.6)		7.9 (5.8–9.2)		0.780
Alcohol intake
Yes	11	26.8%	11	22.9%	0.670
No	30	73.2%	37	77.1%
Smoking habit
Yes	6	14.6%	10	20.8%	0.582
No	35	85.4%	38	79.2%
NSAID use
Yes	7	17.1%	3	6.2%	0.177
No	34	82.9%	45	93.8%
Antiplatelet drug use
Yes	14	34.1%	11	22.9%	0.240
No	27	65.9%	37	77.1%
Anticoagulant use
Yes	6	14.6%	5	10.4%	0.748
No	35	85.4%	43	89.6%
PPI use
Yes	8	19.5%	17	35.4%	0.096
No	33	80.5%	31	64.6%
CCI
≥5	2	4.9%	6	12.5%	0.279
<5	39	95.1%	42	87.5%
Interval between initial CE and repeat CE/DBE (days, median (IQR))	121.0 (28.0–683.5)		31.5 (3.3–92.8)		0.003
Positive findings
Yes	30	73.2%	19	39.6%	0.001
No	11	26.8%	29	60.4%
Bleeding lesions
Mucosal break	16	53.3%	6	31.6%	0.519
Vascular lesion	7	23.3%	6	31.6%
Tumor	5	16.7%	5	26.3%
Active bleeding	2	6.7%	2	10.5%

IQR: interquartile range; OGIB: obscure gastrointestinal bleeding; Hb: hemoglobin; NSAID: nonsteroidal anti-inflammatory drug; PPI: proton pump inhibitor; CCI: Charlson comorbidity index.

Complete small bowel examination was accomplished in 34 cases of the repeat CE group. Both per oral and per anal DBE were performed in 30 cases (enteroscope reached the tattoo marked by the enteroscope from the other side in six cases), per oral DBE only was performed in 15 cases (per oral DBE reached the ascending colon in three cases), and per anal DBE only was performed in three cases of the DBE group. When DBE found the obvious bleeding lesion, inspection from the other side was not performed.

Clinical factors associated with positive findings

We examined the clinical factors associated with positive findings, including age, sex, OGIB type, previous lowest Hb level, alcohol intake, smoking habit, NSAID use, antiplatelet drug use, anticoagulant use, PPI use, CCI, interval between initial CE and repeat CE/DBE, and repeat CE. Univariate logistic regression analysis showed that repeat CE was significantly associated with positive findings (OR, 4.163; 95% CI, 1.691–10.249; p = 0.002). Alcohol intake had a tendency to be associated with positive findings (OR, 2.080; 95% CI, 0.752–5.750; p = 0.158). Increased interval between initial CE and repeat CE/DBE also had a tendency to be associated with positive findings (OR, 1.001; 95% CI, 1.000–1.002; p = 0.106). Multivariate logistic regression analysis including age, sex, alcohol intake, interval between initial CE and repeat CE/DBE, and repeat CE revealed that the repeat CE was an independent factor associated with the increased prevalence of positive findings (OR, 3.623; 95% CI, 1.385–9.478; p = 0.009) (Table 2).

Table 2.

Univariate and multivariate analyses of factors associated with the prevalence of positive findings.

Variables	Univariate		Multivariate
	OR (95% CI)	p value	OR (95% CI)	p value
Age (years)	1.004 (0.979–1.029)	0.761	1.008 (0.980–1.036)	0.584
Sex
Male	0.785 (0.340–1.815)	0.572	0.760 (0.299–1.935)	0.565
Female	1		1
OGIB type
Overt	0.906 (0.374–2.197)	0.828
Occult	1
Lowest Hb (g/dl)	0.943 (0.807–1.101)	0.457
Alcohol intake
Yes	2.080 (0.752–5.750)	0.158	2.495 (0.810–7.686)	0.111
No	1		1
Smoking habit
Yes	0.574 (0.193–1.710)	0.319
No	1
NSAID use
Yes	2.056 (0.495–8.529)	0.321
No	1
Antiplatelet drug use
Yes	1.055 (0.416–2.674)	0.911
No	1
Anticoagulant use
Yes	2.407 (0.594–9.753)	0.219
No	1
PPI use
Yes	0.843 (0.333–2.129)	0.717
No	1
CCI
≥5	1.402 (0.314–6.260)	0.658
<5	1
Interval between initial CE and repeat CE/DBE (days)	1.001 (1.000–1.002)	0.106	1.001 (0.999–1.002)	0.394
Repeat CE or DBE
Repeat CE	4.163 (1.691–10.249)	0.002	3.623 (1.385–9.478)	0.009
DBE	1		1

OGIB: obscure gastrointestinal bleeding; Hb: hemoglobin; NSAID: nonsteroidal anti-inflammatory drug; PPI: proton pump inhibitor; CCI: Charlson comorbidity index; CE: capsule endoscopy; DBE: double-balloon enteroscopy; OR: odds ratio; CI: confidence interval.

IPTW analysis

Even after the adjustment of clinical background factors in the two groups using IPTW analysis, repeat CE was significantly associated with positive findings (OR, 4.423; 95% CI, 1.599–12.232; p = 0.004) and the robustness was observed (Table 3).

Table 3.

Inverse probability of treatment weighting (IPTW) analysis.

	Before IPTW		After IPTW
	OR (95% CI)	p value	OR (95% CI)	p value
Repeat CE	4.163 (1.691–10.249)	0.002	4.423 (1.599–12.232)	0.004
DBE	1		1

CE: capsule endoscopy; DBE: double-balloon enteroscopy; OR: odds ratio; CI: confidence interval.

Discussion

In this study, we revealed that the positive finding rate was significantly higher in the repeat CE group (73.2%) than in the DBE group (39.6%) in OGIB patients with negative findings in the initial CE. Therefore, we contend that CE should be repeated even if the initial CE did not find any potential bleeding lesions in OGIB patients with suspected continuous bleeding.

The positive CE finding rate in the initial CE was 43.4%, which was not different from a previous report in which the positive CE finding rate was 42% in patients with OGIB.¹³ Although repeat CE was performed for patients who had negative findings in the initial CE, repeat CE detected more potential bleeding lesions than the initial CE. This does not signify that CE should always be performed twice because CE often misses the bleeding lesions. As patients who underwent small bowel reexamination had evidence of sustained bleeding or severe anemia, it is conceivable that they were more likely to have potential bleeding lesions in the small intestine and this led to the high rate of positive findings.

There are several clinical risk factors for the presence of bleeding lesions considered positive in CE. Age, anticoagulant therapy, low Hb level, a history of NSAID use, PPI use, and severe comorbidities are known significant predictors of positive CE findings in patients with OGIB.^14–16 It is known that NSAIDs cause erosions/ulcers in not only the stomach but the small intestine, and we consider that intestinal microbiota plays a crucial role in NSAID-induced small intestinal damage.¹⁷ PPI may induce dysbiosis of intestinal microbiota and exacerbate the small intestinal damage.¹⁸ In addition, it is reported that the frequency of positive CE findings is higher in patients with overt OGIB than in those with occult OGIB (59.5% vs. 46.2%, respectively; p = 0.03).¹⁹ However, we revealed that these risk factors were not associated with positive findings, and repeat CE only was the only independent factor associated with the increased prevalence of positive findings in OGIB patients with initial negative CE findings in univariate and multivariate analyses. As patients with multiple risk factors had positive findings in the initial CE, these risk factors would not be associated with positive findings in the repeat CE. Comparing the clinical background factors of the repeat CE and the DBE groups, the interval between initial CE and repeat CE was significantly longer than that between initial CE and DBE, and DBE was performed more frequently in patients with overt OGIB. These findings may be because DBE was likely to be performed to rush hemostasis for overt OGIB cases in clinical practice. To adjust the clinical background factors between the two groups, we applied IPTW analysis in this study. IPTW analysis revealed that repeat CE was more effective than DBE in OGIB patients with negative CE findings even after the adjustment of all clinical background factors.

Among the bleeding lesions, mucosal breaks were found more frequently in the repeat CE group than in the DBE group. Since CE can closely view the small intestinal mucosa in the deaeration state, it is superior for the detection of minor bleeding lesions than DBE. Kameda et al. demonstrated that CE yielded more erosions/ulcers than DBE in patients with OGIB in a prospective study (90.6% vs. 65.6%, respectively; p = 0.03).²⁰ Nakamura et al. also showed that CE was superior to DBE at detecting small erosions.²¹ As most large bleeding lesions were easily detected in the initial CE and only minor bleeding lesions remained in the small bowel reexamination, repeat CE could detect more minor bleeding lesions than DBE as resulted in this study. It is reported that CE has a blind spot to miss the tumors that could be detected by DBE, especially in proximal small intestine.^22,23 However, tumors were detected almost equally in the repeat CE group and the DBE group.

Our findings in this study reveal a novel approach to treating OGIB patients with no bleeding lesions in CE. Subsequent examinations that should be performed after the initial CE include repeat CE, DBE, scintigraphy, or angiography. DBE or Meckel's scan (Tc-99m pertechnetate scintigraphy) is recommended if there are no abnormal findings in the initial CE and further workup is necessary according to the clinical practice guideline for enteroscopy by the Japan Gastroenterological Endoscopy Society.²⁴ There are reports that repeat CE appears to be beneficial because of its high diagnostic yield and noninvasive nature. Bar-Meir et al. revealed that 20% of patients with severe iron-deficiency anemia in whom the initial CE was negative had bleeding lesions on repeat CE.²⁵ Viazis et al. demonstrated that patients with negative CE findings would benefit from a repeat CE if the OGIB type changes from occult to overt or if the Hb level dropped to ≥4 g/dl.²⁶ Thus, patients with evidence of recurrent bleeding or progressive anemia should potentially undergo repeat CE to identify the bleeding lesions even if the initial CE was negative. To the best of our knowledge, no previous reports have compared the utility of repeat CE and DBE in OGIB patients with negative findings in the initial CE. CE can view the entire small intestine in a single examination with a high diagnostic yield. On the other hand, DBE sometimes fails to achieve total enteroscopy even if the per oral and per anal routes are used. We consider that CE has more diagnostic value than DBE, including the difficulty to visualize the complete small bowel in DBE. CE should be performed first to visualize the complete small bowel and predict the appropriate insertion route of DBE. Randomized controlled trials will be necessary for further clarification.

This study has several limitations. First, this design was retrospective and the data were collected from a single university hospital; therefore, selection bias is unavoidable. Second, it is unknown whether small erosions could be the true source of bleeding. The most common causes of bleeding in OGIB patients are known to be erosions/ulcers and vascular lesions in the small intestine, and it is frequently experienced that bleeding persists from minor erosions of the small intestine. However, it is hard to identify whether these minor lesions are the actual bleeding sources if the bleeding has already stopped. Third, it is difficult to identify the accurate date on which bleeding occurred, especially in occult OGIB cases, and the period from bleeding to examination varies. As the effectiveness of earlier CE was reported previously,^27,28 the interval between the bleeding episode and CE examination may have affected our results.

In conclusion, reexamination via CE has more diagnostic value than DBE for OGIB patients with negative findings in the initial CE. CE should be repeated prior to DBE in OGIB patients with suspected continuous bleeding even if the initial CE is negative. Further studies are needed to establish the best strategy for evaluating OGIB.

Declaration of conflicting interests

None declared.

Funding

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

Ethics approval

The study protocol was approved by the Osaka City University Graduate School of Medicine ethics committee (No. 2534) on June 23, 2016. The study protocol conforms to the ethical guidelines of the 1975 Declaration of Helsinki as reflected in prior approval by the institution's human research committee.

Informed consent

Written, informed consent was obtained from each patient included in this study.