Approximately 20–30 % of patients with acute GI bleeding who undergo capsule endoscopy and/or deep enteroscopy will have sources of bleeding within reach of standard upper or lower endoscopic examinations that might have been missed upon initial testing. In one study published in 2009, the prevalence of lesions outside the small bowel in patients undergoing double-balloon enteroscopy (DBE) was assessed. The study examined 143 DBE procedures in 107 patients over an interval of 3.5 years [2]. The patients presented with either obscure overt or obscure occult bleeding. The direction of the DBE examination was guided by prior video capsule endoscopy (VCE) findings, if available. In the absence of a VCE study, an anterograde approach was used for patients presenting with melena, whereas patients with hematochezia underwent initial retrograde examinations. Bidirectional enteroscopy was performed if no source was found on the initial DBE approach. Small bowel pathology was detected in 69 (65 %) patients and included angiodysplastic lesions in 34 (32 %), ulcerations in 12 (11 %), and small bowel neoplasms in 6 (6 %), in addition to other etiologies.

In order to classify bleeding lesions as “definite” versus “probable ,” the authors used a classification system based on clinical action where A1 was a lesion requiring immediate hemostasis and A2 was a lesion where close observation was recommended. In addition, none of the following lesions were considered as definite causes of bleeding, including grade A–B erosive esophagitis, small esophageal varices without stigmata of bleeding or red wale signs, nonspecific duodenitis, minimal gastric antral vascular ectasia (GAVE), single non-bleeding colonic angiodysplasias, and hemorrhoids without stigmata of bleeding. If the patient presented with occult GI bleeding, any lesion was categorized as a possible source of bleeding except for colonic diverticulosis without signs of hemorrhage, esophageal varices, and/or hemorrhoids.

The authors found a possible cause of bleeding in 51 (48 %) patients. These sources included colonic diverticulosis in 11, gastric or duodenal ulcerations in 6, non-bleeding colonic angiodysplastic lesions in 5, GAVE in 5, Cameron erosions in 2, and other sources in 18 patients. Twenty-six (24 %) patients were considered to have definite sources of hemorrhage detected, including gastric and/or duodenal ulcerations (n = 6), Cameron erosions (n = 2), GAVE (n = 4), radiation ileitis (n = 3), angiodysplastic lesions (n = 4), diverticulosis (n = 3), colonic Crohn’s disease (n = 1), anastomotic ulceration (n = 1), and hemorrhoids (n = 1). Based on these findings, the authors suggested that second-look endoscopic examinations be performed prior to small bowel evaluation, particularly if there was inadequate mucosal visualization or lack of documentation of landmarks on the initial examination.

The authors from another study published in 2008 queried whether repeat endoscopic examinations within 6 months of a VCE procedure were necessary when initial examinations were negative [3]. At the time of the study, the investigators were required to repeat standard upper and lower endoscopic examinations within 6 months of a VCE study for the purposes of obtaining Medicare reimbursement for the VCE study. Of the 198 patients referred for investigation of obscure GI bleeding, 50 underwent repeat endoscopic examinations solely to enable Medicare reimbursement. The average duration of obscure bleeding was 50 months, and the most recent upper and/or lower endoscopic exams had been performed 19 months prior to the VCE study. The authors found only two patients with probable causes of bleeding (GAVE and gastric ulceration). On the subsequent VCE study, probable bleeding sources were identified in 24 (51 %) patients and possible sources in 5 (11 %) patients, including angiodysplastic lesions (n = 17), mass lesions (n = 2), nonsteroidal anti-inflammatory drug enteropathy (n = 2), Cameron erosions (n = 2), and Crohn’s disease (n = 1). The authors concluded that repeat endoscopic examinations prior to VCE were associated with a low diagnostic yield. Differences between this study and the aforementioned 2009 study may be explained by the fact that the population in the 2009 study was referred for deep enteroscopy and the quality of the prior endoscopic examinations not rigorously assessed as in the 2008 study.

Once the decision is made to pursue a small bowel source of hemorrhage, the next recommended test is VCE testing. VCE was initially approved by the United States Food and Drug Administration in 2001 as an adjunctive test for the evaluation of small bowel disorders and then as a first-line modality for small bowel evaluation in 2003.

Currently available VCE systems in the United States include the PillCam SB (Given Imaging, Inc., Yokneam, Israel) and the Olympus Endocapsule (Olympus Corp., Center Valley, PA). Both capsule systems capture two frames per second and have a field of view of 160°. The PillCam SB is now available with a 12-h battery life, which may increase completion rates to the cecum. The Olympus Endocapsule has an 8-h battery life. The current capsules measure 11 × 26 mm in dimensions and contain a lens, white light-emitting diodes for illumination, silver oxide batteries, and an ultrahigh frequency band radio telemetry transmitter. The camera for the PillCam SB capsule is a complementary metal oxide semiconductor (CMOS) chip, whereas the Olympus Endocapsule uses a charge-coupled device (CCD) chip. A capsule-loading device is available to directly deliver the capsule into the duodenum and may be considered in patients with significant dysphagia secondary to esophageal motility disorders and gastroparesis, inpatients with limited mobility, patients on chronic narcotics, pediatric patients, and subjects unable to ingest large pills. For patients who are unsure about the ability to swallow the capsule, particularly children or young adolescents, the “jelly bean” test can be administered as a trial before attempting to swallow the capsule endoscope.

The video capsule endoscope requires activation prior to ingestion. A flashing light will be apparent when the device is ready for usage. Sensors are placed on the patient’s chest and abdomen in order to capture quadrant location for the VCE device; while recent software does not require sensor placement, most VCE readers find this information to be useful, and studies have demonstrated accuracy of the quadrant locator in the setting of subsequent surgical interventions. After capsule ingestion, the patient is able to leave the endoscopy unit wearing the waist belt holding the data recorder. For inpatients, ingestion can occur directly at the bedside. The patients are allowed to ingest a liquid meal 2 h later and a regular diet after 4 h.

Meta-analyses have demonstrated the efficacy of administering a purgative preparation prior to the VCE examination [4]. The studies have demonstrated that the administration of 2 l of polyethylene glycol (PEG) is equivalent to 4 l and results in improved small bowel visualization and diagnostic yield. In a 2009 meta-analysis encompassing 12 studies (6 prospective, 6 retrospective), the diagnostic yield was significantly improved in patients who received purgative preparations compared to those maintained on clear liquids alone (263 vs. 213 patients; odds radio (OR) 1.8, 95 % CI 1.3–2.6, p = 0.002). The quality of small bowel visualization was also significantly improved (OR 2.1, 95 % CI 1.3–3.6, p = 0.005). The administration of purgatives did not affect overall completion rates to the cecum or transit times in the stomach or small bowel.

For inpatients, the yield of VCE has been shown to exceed 90 % when administered within 48 h of hospital admission [5]. However, usage of VCE in the inpatient setting, while associated with higher diagnostic yields, carries increased rates of gastric retention (mainly due to the use of narcotics and other medications, in addition to the immobility state) and incomplete examinations to the cecum [6]. For inpatients undergoing VCE studies with a higher risk of incomplete studies, measures to avoid gastric retention include endoscopic placement of the capsule into the small bowel; use of promotility agents, such as intravenous metoclopramide or erythromycin before and during the study; and cessation of narcotics, if possible. If the patient is being discharged from the hospital, it may be advantageous to wait until discharge for the VCE study to occur so that the examination can occur while the patient is more active in an outpatient setting.

VCE is advantageous with a diagnostic yield of 25–50 % compared to the yields demonstrated using traditional small bowel radiography (3–20 %) [7, 8], push enteroscopy (3–30 %) [9–11], and elective angiography (5–15 %) [12, 13]. In patients with a negative capsule endoscopy, the usage of multi-detector computed tomographic (CT) or magnetic resonance (MR) enterography has been shown to detect pathology in some patients, particularly if bleeding is related to an underlying neoplasm [14].

Timing of the VCE examination correlates with diagnostic yield in patients with overt obscure GI hemorrhage. In the landmark study by Pennazio et al., the diagnostic yield in 100 patients undergoing VCE was 92 % in those with ongoing overt hemorrhage, 13 % in patients with bleeding that had stopped (intervals ranging between 10 days and 1 year), and 44 % in the iron-deficiency anemia cohort [11]. Subsequent studies reported higher diagnostic yields when VCE was performed within 2 weeks of an overt bleeding episode (detection rate 91 %) compared to 34 % when the VCE occurred more than 2 weeks later [15]. Similarly, higher diagnostic yields have been demonstrated when deep enteroscopy is performed within 2 weeks of an overt bleeding episode [16].

In a recent retrospective study from a tertiary medical center, inpatients with overt obscure GI bleeding undergoing VCE studies were analyzed for diagnostic and therapeutic yields in relation to timing of VCE administration. The diagnostic yield was significantly higher for inpatients (66 %) compared to outpatients (53 %) and greater if the VCE was administered within 3 days of hospitalization (yield for active bleeding and/or angiodysplasia of 44 % versus 28 % for VCE studies performed after 3 days, p = 0.05). If the VCE was administered early, the length of stay was shorter (6 versus 10 days, p < 0.0001), and there was a greater rate of therapeutic intervention (19 % versus 7 %, p = 0.05) [17].

When the VCE study is negative, a decision should be made whether to pursue ongoing diagnostic evaluation for a small bowel source or wait until another episode of bleeding occurs. The argument for the latter decision is that the diagnostic yield is higher within 2 weeks of a bleeding episode, so that waiting for another episode of hemorrhage to occur can be useful if the patient is stable. On the other hand, if the patient continues to demonstrate overt bleeding and/or require transfusions, then ongoing evaluation should occur.

Multiple studies have demonstrated increased diagnostic yields when VCE studies are repeated after an initial negative evaluation. In the study by Viazis et al., the investigators followed patients with initial nondiagnostic VCE studies and performed repeat VCE when the patients had recurrent overt bleeding or a drop in hemoglobin ≥ 2 mg/dl [18]. Of 104 patients with an initial nondiagnostic VCE followed for a mean of 25 months, 76 (73 %) received a second VCE study due to recurrent bleeding. Thirty-seven (49 %) subjects had positive findings on repeat VCE, with angiodysplastic lesions being the most common finding. On logistic regression analysis, significant predictors for an abnormal repeat study included change in bleeding type from occult to overt hemorrhage and drop in hemoglobin levels ≥ 4 mg/dl.

In another retrospective study, the authors analyzed 82 of 676 patients undergoing repeat VCE studies from 2001 to 2009 [19]. Overall, the diagnostic yield of repeat VCE was 55 %, leading to a change in management in 39 % of the cohort. Reasons for repeat VCE studies included incomplete initial VCE (n = 22, yield of repeat VCE 45 %), screening for polyposis syndromes (n = 4, yield 50 %), ongoing gastrointestinal symptoms (n = 26, yield 38 %), and prior VCE studies leading to therapeutic interventions but with ongoing symptoms (n = 30, yield 77 %).

Based on the abovementioned studies, close follow-up is advised in patients with an initial normal VCE study who remain clinically stable. If the patient has ongoing acute hemorrhage, urgent repeat VCE could be considered versus other options, including deep enteroscopy, CT or MR enterography, or angiography. If the patient remains stable without bleeding, the recommendation would be to consider repeat VCE study when hemoglobin falls or the patient experiences recurrent overt bleeding.

The diagnostic yield of mesenteric angiography in patients with obscure GI bleeding traditionally has been low, in the range of 5–15 % [13]. The question, therefore, arises whether there currently is a role for angiography and, if so, in what subset of patients.

The diagnostic algorithm suggests angiography for patients with “massive bleeding” (Fig. 5.1). The amount of hemorrhage required for a positive angiographic examination is in the order of ≥1 ml/min or bleeding in the setting of hemodynamic instability (hypotension and/or tachycardia). Angiography has typically been performed in patients with suspected ongoing diverticular hemorrhage where a source could not be identified during colonoscopy, or in patients with significant bleeding from mucosal or submucosal vascular lesions, where embolization might be an effective treatment modality (Videos 5.1 and 5.2).

A recent randomized controlled trial compared angiography (n = 30) to VCE (n = 30) in patients with obscure overt GI bleeding [20]. VCE was diagnostic in 16 (53 %) patients compared to 6 (20 %) patients undergoing angiography (p < 0.005). Findings on VCE examinations included tumors (n = 3), active bleeding (n = 4), ulcerations (n = 6), and active gastric bleeding (n = 3). Lesions found on angiography included Meckel’s diverticulum (n = 1), tumor (n = 1), and angiodysplastic lesions in the small bowel and/or colon (n = 4). None of the patients in the angiography group underwent embolization. Four patients continued to bleed after angiography and underwent VCE demonstrating diverticular hemorrhage in three patients (two small intestinal and one colonic). Five patients underwent surgical resection for tumors and Meckel’s diverticulum. Over a mean follow-up of 48 months, rebleeding occurred in 25 % of the overall cohort, mainly due to vascular lesions and diverticular disease. There were no significant differences between groups in terms of hospitalization or death rates. A limitation of the study was that it was likely underpowered to detect meaningful differences in diagnostic yields and outcomes.

In summary, angiography cannot be recommended as the first test in the setting of obscure overt GI bleeding, although it may be considered in the patient who presents with hemodynamic instability.

Push enteroscopy (using endoscopes up to 250 cm in length) was introduced in the 1990s. The technique is hampered by looping of the enteroscope resulting in patient discomfort and limiting the extent of the examination to 50–150 cm of visualized small bowel (corresponding to proximal jejunum) despite the use of an overtube [21]. The reported diagnostic yield of push enteroscopy in patients with obscure GI bleeding ranges from 15 to 75 %, with arteriovenous malformations being the most common lesions identified [22]. However, in 10–60 % of push enteroscopic examinations, detected lesions were described to be within reach of a standard endoscope. Push enteroscopy examinations can still be useful in patients with suspected bleeding of the proximal small bowel but have largely been replaced by utilization of VCE and deep enteroscopy. Push enteroscopy can be considered in patients requiring a “second look” examination of the upper digestive tract prior to video capsule endoscopy.