Share this article on:

Early Implementation of Video Capsule Enteroscopy in Patients with Left Ventricular Assist Devices and Obscure Gastrointestinal Bleeding

Truss, Wiley D.*; Weber, Frederick*; Pamboukian, Salpy V.; Tripathi, Arvind; Peter, Shajan*

doi: 10.1097/MAT.0000000000000303
Adult Circulatory Support

Gastrointestinal bleeding (GIB) is a frequent challenge encountered in patients implanted with a left ventricular assist device (LVAD), affecting approximately 25% of this population. Many patients have no identifiable source of bleeding after routine esophagogastroduodenoscopy and colonoscopy and are labeled as obscure GIB (OGIB). Significant costs and invasive procedures are required to investigate and stop the source of bleeding in these patients. We performed a retrospective analysis at a single tertiary referral center to investigate the diagnostic yield and overall effectiveness of video capsule enteroscopy (VCE) in this population. Eight patients with LVADs underwent nine VCE studies for OGIB. The diagnostic yield was 100%, with intraluminal blood the most common finding. The jejunum was the most common location for pathology detected on VCE. Sixty-seven percent of the studies directly guided further endoscopy with successful cessation of bleeding in 100% of these patients. Finally, after an average follow-up of 46 weeks, the total number of endoscopic procedures and total units of transfused packed red blood cells (pRBC) were significantly less after the patient underwent the VCE study compared with before. Video capsule enteroscopy is a safe and high-yield investigative procedure in this population and should be implemented earlier to improve patient outcomes and reduce costs of care.

From the *Division of Gastroenterology and Hepatology, Department of Medicine, University of Alabama at Birmingham, Birmingham, Alabama; Division of Cardiology, Department of Medicine, University of Alabama at Birmingham, Birmingham, Alabama; and School of Public Health, Biostatistics, University of Alabama at Birmingham, Birmingham, Alabama.

Submitted for consideration July 2015; accepted for publication in revised form October 2015.

Disclosure: The authors have no conflicts of interest to report.

Correspondence: Wiley D. Truss, MD, MPH, Department of Gastroenterology and Hepatology, University of Alabama at Birmingham, BDB 395, 1720 2nd Avenue South, Birmingham, AL 35294. Email:

Left ventricular assist devices (LVADs) have led to significant improvements in quality of life and reduced mortality for patients suffering from end-stage systolic heart failure. For more than 3 decades, VAD therapy has been used, and there has been an evolution in technology from pulsatile-flow devices toward continuous-flow (CF) devices. Although CF technology has improved some of the early limitations of device therapy, long-term adverse events of these devices still exist including infection, pump thrombosis, thromboembolic events, and gastrointestinal bleeding (GIB).

Many studies have documented the association between LVADs and GIB, but few have examined the appropriate algorithm for diagnosing and treating the culprit lesion. In a recent meta-analysis by Draper et al.,1 the incidence of GIB in LVAD patients was documented at 23%, considerably higher than the 4% rate in patients undergoing antiplatelet/anticoagulation therapy for other cardiac conditions.2 Furthermore, diagnosing the source of GIB in these patients has historically been very difficult with diagnostic yields estimated as low as 45%.3,4 When identified, the bleeding source is most commonly confined to the stomach; however, the small bowel contains the source of bleeding in 29% of these patients.5,6 Video capsule enteroscopy (VCE) has been proven safe for use in this patient population, but its effectiveness and timing has never been fully evaluated.7 With recent studies by Meltzer et al.8 and Gralnek et al.9 documenting the utility and cost-effectiveness of early VCE in the emergency room setting, earlier utilization of this technology is now on the rise. The aim of this study is to examine diagnostic yield and overall effectiveness of VCE in the evaluation of obscure gastrointestinal bleeding (OGIB) in patients implanted with a LVAD.

Back to Top | Article Outline



This is a retrospective chart review conducted at a single tertiary referral center. Patients were identified from a registry of VCE studies performed at the University of Alabama at Birmingham (UAB) between November 2009 and August 2012. All LVAD patients who underwent VCE in the evaluation of OGIB were included. Average follow-up time was 46 weeks. The institutional review board at UAB approved the review of patient data for the purpose of this study.

Back to Top | Article Outline

Gastrointestinal Bleeding

OGIB can be divided into occult or overt OGIB, depending on whether there is a gross evidence of gastrointestinal blood loss. In our study, occult OGIB included patients presenting with positive fecal occult blood test, anemia, or iron deficiency. Overt OGIB included any patient who reported signs of melena, hematochezia, or hematemesis. For patients with recurrent GIB, a new case was defined after the patient had been successfully discharged from the previous hospitalization.

A clinically significant finding was defined as any mucosal lesion that could explain or correlate with the patient’s presenting complaint. These findings generally correlated with the P1 and P2 lesions first described by Saurin et al.10 in 2003. In our study, the final diagnosis was considered “clinically significant” if there was active bleeding or the mucosal lesion correlated with a P1 or P2 on the Saurin scale.

Back to Top | Article Outline


Video capsule enteroscopy is performed for a variety of diagnostic purposes and is now considered first-line therapy in the workup of OGIB. PillCam SB2 (Given Imaging, Yokneam, Israel) was used in all investigations. An incomplete study was defined as the PillCam not visualizing the cecum. Two experienced readers with more than 200 capsule encounters interpreted all capsule studies.

Back to Top | Article Outline

Data Collection

This was a retrospective chart review centered on VCE investigation of OGIB. Data including age, sex, race, device type, indication, hospitalization status, and comorbid conditions were collected. Use of aspirin, clopidogrel, warfarin, and proton-pump inhibitors at the time of the PillCam study was documented. Pre-VCE data were collected including the total number of endoscopic procedures, total units of packed red blood cells (pRBCs) transfused, number of prior hospitalizations, and the indication for the VCE. All information recorded in the official VCE report was also documented. Post-VCE data included VCE-directed endoscopic procedures, subsequent hospitalizations, medical treatment, length of follow-up, and final diagnosis.

Back to Top | Article Outline

Statistical Analysis and Outcome

For continuous variables, mean ± standard deviation was calculated. Data before and after the VCE were compared using the Student’s two-tailed t-test. Categorical data was analyzed with the χ2 test. A p value <0.05 was considered statistically significant. The primary outcome was diagnostic yield. Secondary outcomes included total number of endoscopic procedures, total number of hospitalization, and total units of blood transfused before and after the VCE study.

Back to Top | Article Outline



Between November 2009 and August 2012, there were 370 VCE studies conducted at UAB. Three hundred five (82%) studies were conducted for OGIB. Twelve of those studies (4%) were conducted in patients implanted with a LVAD. Two patients had multiple VCE studies performed. Therefore, there were 10 LVAD patients who underwent a total of 12 VCE studies for evaluation of OGIB. Three studies were excluded for incomplete small bowel evaluation, leaving a total of nine studies on eight patients for analysis. The average age of the participants was 63.4 ± 6.3 years (Table 1). All the participants were male, and 87% were Caucasian. At the time of the VCE, 78% of the participants were hospitalized. Antiplatelet therapy and anticoagulation use as well as pre-VCE international normalized ratio are shown in Tables 1 and 2.

Back to Top | Article Outline


All patients included in the study underwent LVAD implantation between April 2007 and February 2012. All patients included in the study utilized either the HeartWare (25%) or HeartMate II (75%) CF LVAD. Seventy-five percent received the device as destination therapy. Overall, at our institution, there were 87 patients implanted with CF LVADs during the study period (2009–2012). Of those 87 patients, 16 (18.4%) experienced a GIB within 6 months of device implantation and 14 (87.5%) underwent endoscopic evaluation with an overall diagnostic yield of 57%. The specific patients included in our study were those who subsequently required VCE. The initial VCE was performed an average of 428 ± 549 days after the implantation of the LVAD.

Back to Top | Article Outline


One hundred percent of the VCE studies were performed in evaluation of overt OGIB, with 89% of the participants presenting with melena. The average time that a patient had been struggling with GIB before undergoing VCE evaluation was 26 ± 32 weeks (range, 1–96 weeks; median, 8 weeks). For the inpatient population, the VCE was performed an average of 1.7 days after the last overt bleeding episode. For the outpatient subgroup, the number of days from last bleed ranged from 1 to 45. The average number of previous hospitalizations for GIB before the VCE was 3.8 ± 3.1. The average units of pRBCs transfused before VCE was 8.8 ± 6.2 units. Finally, the average number of endoscopic evaluations (esophagogastroduodenoscopy [EGD], colonoscopy, etc.) performed before the VCE was 6.44 ± 2.87.

Back to Top | Article Outline


Per the inclusion criteria, the completion rate of the VCE to the cecum was 100% (Table 3). Bowel preparation was considered adequate in 89% of the procedures. There were no adverse events or cases of retention in this population. The mean gastric transit time (GTT) was 23 minutes. The mean small bowel transit time (SBTT) was 207 minutes. The diagnostic yield of VCE was 100% (Figure 1). Intraluminal blood was noted in 89% of the studies, with fresh blood present in 56%. Angioectasias were clearly visualized in only 22% of the studies. The abnormality was localized to some part of the small bowel in 89% of the cases, including the jejunum (78%), duodenum (44%), and ileum (11%). The stomach (11%) and colon (11%) contained the source in two patients.

Back to Top | Article Outline


The VCE directly guided further endoscopic investigations in 67% of the cases (Table 2). The most common post-VCE endoscopic intervention was push enteroscopy (PE), performed in 44% of the patients, followed by repeat EGD/colonoscopy (33%) and double balloon enteroscopy (DBE; 11%). These subsequent procedures successfully identified the bleeding source in 67% of the cases, with angioectasias (67%) being the most common final diagnosis. Successful control of bleeding with endoscopic therapy occurred in 100% of these patients.

Finally, with an average follow-up of 46 weeks, the average number of post-VCE hospitalizations for GIB declined to 2.1, compared with 3.8 hospitalizations before the VCE (p = 0.169; Table 4). The average number of endoscopic procedures post-VCE was 2.22, compared with 6.44 before the VCE (p = 0.004). The average number of pRBCs transfused post-VCE was 3.8, compared with 8.8 before the VCE (p = 0.05).

Back to Top | Article Outline


As a minimally invasive study, VCE is an excellent tool for evaluating pathology of the gastrointestinal (GI) tract, specifically the small bowel. Few studies have evaluated the effectiveness of VCE in the investigation of OGIB in patients implanted with LVADs. In a narrative review of case reports and case series by Islam et al.,11 317 patients with LVADs and GIB were examined. In that report, only seven VCEs were performed. This scarcity of data may be related to safety concerns of VCE in this population. Initial worries related to radiofrequency interactions between the PillCam equipment and the LVAD initially prompted the US Food and Drug Administration and the PillCam manufacturer (Given Imaging) to place a relative contraindication on its use in this population. Since that time there have been several studies documenting the safety of VCE in patients with LVADs.12,13 Most recently, a study out of the Mayo Clinic evaluated 118 patients with implantable electromechanical cardiac devices (IEDs; LVADs, pacemakers, implantable defibrillators) undergoing VCE to determine the safety in this population (14 patients had LVADs). They concluded that VCE did not interfere with IEDs.14 In our data set, we excluded three patients who had incomplete studies. Of those three patients, one was excluded because of technical error of the belt device, which has been suggested to occur more frequently in this population. The remaining nine studies, however, had no adverse events and provided complete evaluation of the small bowel. Another possibility for the absence of VCE data in LVAD patients is the severity of disease in this population. Most patients with LVAD-associated GIB present with overt GIB (100% in our study), and VCE lacks the ability to intervene if necessary. Consequently, some endoscopists may prefer to repeat investigations capable of therapeutic modalities such as EGD, PE, or DBE.

Previous studies have documented the difficulties of source localization in this population. In a retrospective review by Stern et al., 17 LVAD GIB events were investigated with 36 diagnostic studies with a diagnostic yield of only 28%. Overall, they reported that no definitive bleeding source could be found in 65% of the episodes.3 However, in another study, by Kushnir et al., 82 endoscopic and radiologic studies were performed in evaluation of 44 separate GIB events. A source was eventually localized in 71.5% of these episodes with EGD providing the highest diagnostic yield (57%).4 In our study, a clinically significant finding was identified via VCE in 100% of the cases. Although the most common finding was intraluminal blood (either old or fresh), the actual mucosal defect was only captured in 22% of the capsule studies primarily because of blood obscuring the intraluminal mucosal surface. Despite this, these VCEs still proved invaluable, by accurately regionalizing the bleeding source for further therapeutic interventions. After VCE localization of bleeding, 67% of patients underwent guided therapeutic interventions, with successful bleeding cessation achieved in all of those patients. These findings highlight the value VCE can provide in this patient population, where localizing the source of bleeding has historically been elusive. From our experience over the past decade with GIB in LVAD patients, an algorithmic approach has slowly evolved (Figure 2). The UAB algorithm for GIB in LVAD patients places emphasis on earlier use of VCE and also recognizes the utility of PE more than EGD in initial evaluation of the upper GI tract. Specific details of this approach are largely based on previous studies demonstrating the stomach and upper small bowel (duodenum/proximal jejunum) as the most likely location of bleeding source in this population. One study, by Goldstein et al.,6 reviewed 382 patients with end-stage heart failure and GIB and identified the small bowel as the most common location for bleeding source at 29%.

Inherent in the problem of source localization, previous studies have documented the vast number of endoscopic procedures required in searching for the culprit lesion. In their review of 10 case reports and 22 case series, Islam et al.11 identified a total of 188 procedures performed in the workup of GIB events in LVAD patients. The most common endoscopic procedure documented was EGD (n = 103) followed by colonoscopy (n = 49). Again, VCE was utilized only seven times throughout their entire review. We noted a similar problem in our subset of patients; multiple endoscopic interventions were being repeated in the search for the suspect lesion with a very low diagnostic yield, especially with repeat endoscopy. We noted that before undergoing VCE; the average number of endoscopic interventions performed on each patient was approximately 6. When compared with after the VCE (with average follow-up of 46 weeks), the average decreased to approximately 2 (p < 0.05). The precise reason for this significant decline is difficult to ascertain. Three plausible explanations can be considered. First, the average follow-up time was not sufficient to capture all future bleeding events compared with the period before the capsule study, and therefore, fewer endoscopic evaluations would be expected. Second, because the VCE study successfully identified the bleeding source in 67% of the patients and all underwent successful therapy, the therapeutic benefit may be long term. Finally, a final diagnosis of angioectasia was found in 67% of the patients, and it cannot be excluded that identification of a source in itself reduced diagnostic evaluation during future rebleeding events. In any case, a significant decline in invasive endoscopic procedures (and likely costs of care) is an important finding of this study and further justifies early utilization of VCE in this population. The UAB algorithm reflects this finding and recommends VCE before colonoscopy in LVAD patients presenting with melena.

Another significant problem in LVAD patients with GIB is the vast number of pRBC transfusions they require. The pathogenesis of GIB in this population has been attributed to anticoagulation/antiplatelet therapy, predisposition to formation of angioectasias, destruction of Von Willebrand multimers through shear stress, and submucosal ischemia. Regardless of the etiology, patients require significant transfusions to sustain life once bleeding develops. Sponga et al.15 described a case of a 57-year-old man with HeartMate II LVAD who developed severe GIB requiring a total of 60 transfusions over 23 weeks. Similarly, the patients in our study required massive transfusions although their source of bleeding was localized. However, of clinical importance is the marked decline in transfusions noted in the month after the VCE investigation (~4 units) compared with the month before VCE investigation (~9 units). This decline was statistically significant, and we believe may have resulted from improved targeting of the bleeding source. Regardless of the cause, a significant decline in blood transfusions is an important finding and further supports early implementation of VCE in this population.

Back to Top | Article Outline


One of the limitations of our study was the small size and relatively unique population we investigated. By design, our study population was prejudiced toward more critically ill LVAD bleeders. This resulted from the fact that we only included LVAD patients with GIB who underwent VCE and, therefore, had already failed initial investigative measures with EGD and colonoscopy. This bias toward OGIB skewed our population to those LVAD patients who were bleeding from the small bowel. In a narrative review of case reports of LVAD patients with GIB, Islam et al.11 documented that the stomach (41%) was the most common source for the culprit lesion followed by the colon (35%) and small bowel (21%). However, Goldstein et al.6 found that the small bowel accounted for the majority (29%) of bleeding lesions in their cohort of LVAD patients with GIB. Given the high rates of small bowel lesions in this population, VCE should be considered earlier in the diagnostic algorithm for these patients (Figure 2).

Finally, the retrospective nature of this study is a limitation. Although the data was obtained retrospectively, the population studied had extremely close follow-up given the severity of their condition, and accordingly, their electronic medical records were saturated with all necessary data points of interest. Furthermore, the majority of these patients receive all their care at our institution, which decreases the chance of missing data (i.e., repeat hospitalizations) from outside institutions. Future randomized prospective trials with administration of VCE at initial presentation or immediately after initial nondiagnostic EGM/PE would be helpful to further characterize the utility of this technology in this unique population.

Back to Top | Article Outline


Acute GIB is a significant problem in patients implanted with LVADs and is estimated to occur in one quarter of this population.1 With the increasing use of CF LVADs in patients suffering from end-stage heart failure, this problem is likely to increase in the near future. Video capsule enteroscopy has proven to be safe in this population, but its effectiveness and timing have not been investigated fully. In our study, there were no adverse events or cases of retention with the use of VCE. The diagnostic yield was 100% with the majority of lesions noted in the proximal small bowel within reach of PE or antegrade DBE. This resulted in a VCE-directed therapeutic yield of 67%. The use of VCE was also associated with a decline in the total number of endoscopic procedures required and units of pRBCs transfused in these patients. Although not statistically significant, there was also an impressive trend toward fewer GIB-related hospitalizations after the VCE study compared with before. We conclude that VCE should be utilized earlier and more frequently in LVAD patients presenting with acute GIB.

Back to Top | Article Outline


1. Draper KV, Huang RJ, Gerson LB. GI bleeding in patients with continuous-flow left ventricular assist devices: A systematic review and meta-analysis. Gastrointest Endosc. 2014;80:435–446.e1
2. Shrode CW, Draper KV, Huang RJ, et al. Significantly higher rates of gastrointestinal bleeding and thromboembolism events with left ventricular assist devices. Clin Gastroenterol Hepatol. 2014;12:1461–1467
3. Stern DR, Kazam J, Edwards P, et al. Increased incidence of gastrointestinal bleeding following implantation of the HeartMate II LVAD. J Card Surg. 2010;25:352–356
4. Kushnir VM, Sharma S, Ewald GA, et al. Evaluation of GI bleeding after implantation of left ventricular assist device. Gastrointest Endosc. 2012;75:973–979
5. Demirozu ZT, Radovancevic R, Hochman LF, et al. Arteriovenous malformation and gastrointestinal bleeding in patients with the HeartMate II left ventricular assist device. J Heart Lung Transplant. 2011;30:849–853
6. Goldstein DJ, Aaronson KD, Tatooles AJ, et al.ADVANCE Investigators. Gastrointestinal bleeding in recipients of the HeartWare ventricular assist system. JACC Heart Fail. 2015;3:303–313
7. Daas AY, Small MB, Pinkas H, Brady PG. Safety of conventional and wireless capsule endoscopy in patients supported with nonpulsatile axial flow Heart-Mate II left ventricular assist device. Gastrointest Endosc. 2008;68:379–382
8. Meltzer AC, Ward MJ, Gralnek IM, Pines JM. The cost-effectiveness analysis of video capsule endoscopy compared to other strategies to manage acute upper gastrointestinal hemorrhage in the ED. Am J Emerg Med. 2014;32:823–832
9. Gralnek IM, Ching JY, Maza I, et al. Capsule endoscopy in acute upper gastrointestinal hemorrhage: A prospective cohort study. Endoscopy. 2013;45:12–19
10. Saurin JC, Delvaux M, Gaudin JL, et al. Diagnostic value of endoscopic capsule in patients with obscure digestive bleeding: Blinded comparison with video push-enteroscopy. Endoscopy. 2003;35:576–584
11. Islam S, Cevik C, Madonna R, et al. Left ventricular assist devices and gastrointestinal bleeding: A narrative review of case reports and case series. Clin Cardiol. 2013;36:190–200
12. Fenkel JM, Grasso MA, Goldberg EM, Feller ED. Capsule endoscopy is safe in patients with pulsatile Novacor PC left ventricular assist device. Gastrointest Endosc. 2007;65:559–560
13. Bechtel JF, Wellhöner P, Charitos EI, Bucsky B, Morshuis M, Sievers HH. Localizing an occult gastrointestinal bleeding by wireless PillCam SB capsule videoendoscopy in a patient with the HeartMate II left ventricular assist device. J Thorac Cardiovasc Surg. 2010;139:e73–e74
14. Harris LA, Hansel SL, Rajan E, et al. Capsule endoscopy in patients with implantable electromedical devices is safe. Gastroenterol Res Pract. 2013;2013:959234
15. Sponga S, Nalli C, Casonato A, Charbonneau E. Severe upper gastrointestinal bleeding in Heartmate II induced by acquired von Willebrand deficiency: Anticoagulation management. Ann Thorac Surg. 2012;94:e41–e43

video capsule enteroscopy (VCE); capsule endoscopy; left ventricular assist device; obscure gastrointestinal bleeding

Copyright © 2016 by the American Society for Artificial Internal Organs