Introduction
In the United States, there are roughly 6.5 million adults living with congestive heart failure (CHF) and prevalence is estimated to exceed 8 million by 2030. Each year approximately 960,000 new cases of CHF are diagnosed [1 ]
[2 ]. In the past, the only treatment available for CHF refractory to medical management was cardiac transplantation; however, many patients were left with inadequate options owing to the limited availability of donors. The advent of left-ventricular assist devices (LVADs) since the early 1990 s has filled this gap and is now used as a bridge to cardiac transplant, as a bridge to recovery, or even as destination therapy in advanced CHF patients [3 ].
Unfortunately, both pulsatile-flow and continuous-flow LVADs have been associated with an increased risk of gastrointestinal bleeding [4 ]. Reported prevalence of gastrointestinal bleeding in LVAD patients can range between 20 % and 42 %, with small-bowel bleeding accounting for the majority of gastrointestinal bleeding events. Although video capsule endoscopy (VCE) can be a useful initial tool to identify the site of bleeding within the small bowel, balloon-assisted enteroscopy (BAE) has the advantage of being both diagnostic and therapeutic. The two types of BAE are double-balloon enteroscopy (DBE) and single-balloon enteroscopy (SBE). There have been limited case reports [5 ]
[6 ] and case series [7 ] that have discussed the utility and safety of BAE for the evaluation of gastrointestinal bleeding in patients with LVADs. The main objective of our study was to assess the periprocedural safety of BAE in LVAD patients, and to quantify the diagnostic and therapeutic yields of BAE in this cohort.
Patients and methods
We performed a multicenter retrospective cohort study involving four centers: Saint Lukeʼs Hospital of Kansas City/University of Missouri Kansas City (UMKC), Kansas University Medical Center (KUMC), Cleveland Clinic in Ohio and Weston Florida. Institutional review board approval was obtained for each center. The study included adult patients (age ≥ 18 years) with either pulsatile-flow or continuous-flow LVADs who underwent BAE for any indication from January 1, 2007 to December 31, 2018. Further criteria for inclusion were availability of periprocedural information for the BAE with follow-up for at least 72 hours. Patients were excluded if they had only right ventricular assist devices.
Data was collected for patient demographics, LVAD type and goal of therapy, comorbidities (chronic kidney disease, chronic liver disease/cirrhosis, and diabetes mellitus), antithrombotic medications, history of prior gastrointestinal bleeding, and previous endoscopic procedures performed. Indications for BAE were also noted, along with periprocedural details such as transfusion of blood products, American Society of Anesthesiologists (ASA) classification, method of sedation, duration of procedure, identification of culprit lesion(s), and type of therapeutic intervention(s) performed. Overt gastrointestinal bleeding was defined as melena, hematochezia or hematemesis. Diagnostic, therapeutic and procedure-related complications within 72 hours were noted including respiratory (aspiration, and hypoxia), cardiac (LVAD dysfunction, cardiac arrhythmias, hypotension, and myocardial ischemia), and gastrointestinal (nausea, vomiting, abdominal pain, pancreatitis, perforation, and uncontrolled bleeding). Death within 72 hours of the procedure was also noted from the electronic medical record.
The decision to perform either DBE or SBE was made by the endoscopist based on the clinical status of the patient, scope availability (antegrade SBE was only available at one center), and prior endoscopic evaluations.
Outcomes
The primary outcome was to assess the periprocedural safety of BAE, based on the preanesthesia evaluation, any complications from the enteroscopy, complications from therapeutic interventions when applicable, and complications related to LVAD function.
The secondary outcome was the diagnostic and therapeutic yields of BAE (including DBE and SBE). Diagnostic yield was defined as the proportion of procedures in which the cause of bleeding was identified. Therapeutic yield was defined as the proportion of procedures in which therapeutic intervention was performed over the total number of procedures performed [8 ].
Statistical analysis
Categorical variables were summarized using frequencies (n) and percentages. Continuous variables were summarized using means and standard deviations (SD). Comparisons of diagnostic and therapeutic yields between subgroups were conducted using chi-square tests of independence. When expected cell frequencies were below 5.0, Fisher’s exact tests were used. P = 0.05 was used as the level of statistical significance. All analyses were performed using IBM SPSS Statistics for Windows, Version 23.0 (IBM Corp., Armonk, New York, United States).
Results
A total of 34 patients with LVADs who underwent 46 BAEs were identified and included in the study. The mean age of the patient population was 66.4 ± 8.3 years and were predominantly male (79.4 %) ([Table 1 ]). Thirty-two patients (94.1 %) had pulsatile-type LVADs, and the majority of patients (n = 28, 82.4 %) were utilizing the LVAD as a destination therapy. Approximately half of the patient population had chronic kidney disease (CKD) or diabetes mellitus. None of the patients had chronic liver disease or cirrhosis. There were 11 patients (32.4 %) with a previous history of gastrointestinal bleeding from small-bowel arteriovenous malformations (AVMs). Thirty-one patients (91.2 %) had a history of recent endoscopy, with both EGD and colonoscopy being more frequent than VCE which was performed in 21 patients (61.8 %). Twenty patients (58.8 %) were on antiplatelet therapy and 24 patients (70.6 %) were on warfarin.
Table 1
Baseline characteristics of patients in the study.
Patients, n
34
Demographics
66.4 (8.3)
27 (79.4)
Race, n (%)
28 (82.4)
6 (17.6)
LVAD History
Type of LVAD, n (%)
32 (94.1)
2 (5.9)
LVAD indication, n (%)
20 (58.8)
14 (41.2)
LVAD therapy goal, n (%)
6 (17.6)
28 (82.4)
Comorbidities, n (%)
13 (38.2)
0 (0.0)
16 (47.1)
Etiology of prior GI bleeding, n (%)
11 (32.4)
5 (14.7)
4 (11.8)
1 (2.9)
1 (2.9)
1 (2.9)
5 (14.7)
6 (17.6)
Previous endoscopy, n (%)
31 (91.2)
26 (76.5)
9 (26.5)
24 (70.6)
21 (61.8)
Antithrombotic medications, n (%)
20 (58.8)
24 (70.6)
13 (54.2)
10 (41.7)
1 (4.2)
1 (2.9)
AVM, arteriovenous malformation; CKD, chronic kidney disease; EGD, esophagogastroduodenoscopy; LVAD, left ventricular assist device; NOAC, novel oral anticoagulant; PPI, proton pump inhibitor; SD, standard deviation; VCE, video capsule endoscopy.
1 Antiplatelets include aspirin and clopidogrel.
Procedure details
Preanesthesia evaluation identified 37 patients (80.4 %) with American Society of Anesthesiologists (ASA) class IV ([Table 2 ]). General anesthesia was the predominant method (n = 34, 73.9 %) of sedation in comparison to monitored anesthesia care (MAC). All procedures were performed for evaluation of gastrointestinal bleeding. Indications for performing BAE were most frequently due to acute blood loss/anemia (n = 31, 67.4 %), or melena (n = 19, 41.3 %). A mean of 3.57 ± 2.6 units of packed RBC transfusions were administered before the procedures ([Table 3 ]). Most patients were admitted for more than 48 hours prior to undergoing the procedure. A total of 46 BAEs were performed, of which nine were antegrade SBEs and 37 were DBEs (32 antegrade, and 5 retrograde). The mean length of procedure was 52.5 (± 25.2) minutes.
Table 2
Characteristics of anesthesia for patients undergoing balloon-assisted enteroscopy.
Type of BAE
SBE (n = 9)
DBE (n = 37)
Total (n = 46)
ASA classification, n (%)
1 (11.1)
3 (8.1)
4 (8.7)
0 (0.0)
5 (13.5)
5 (10.9)
8 (88.9)
29 (78.4)
37 (80.4)
Type of sedation performed, n (%)
5 (55.6)
7 (18.9)
12 (26.1)
4 (44.4)
30 (81.1)
34 (73.9)
Location of procedure, n (%)
1 (11.1)
7 (18.9)
8 (17.4)
8 (88.9)
30 (81.1)
38 (82.6)
ASA, American society of anesthesiologists; BAE, balloon-assisted enteroscopy; DBE, double-balloon enteroscopy; MAC, monitored anesthesia care; SBE, single-balloon enteroscopy.
1 Involved propofol administration.
Table 3
Procedural details for balloon-assisted enteroscopy.
Type of BAE
SBE (n = 9)
DBE (n = 37)
Total (n = 46)
Number of blood products transfused, mean (SD)
4.00 (1.9)
3.46 (2.7)
3.57 (2.6)
0.11 (0.3)
0.00 (0.0)
0.02 (0.1)
1.78 (2.3)
0.41 (0.8)
0.67 (1.3)
Indications for performing BAE, n (%)
4 (44.4)
15 (40.5)
19 (41.3)
2 (22.2)
6 (16.2)
8 (17.4)
1 (11.1)
0 (0.0)
1 (2.2)
5 (55.6)
26 (70.3)
31 (67.4)
Time to procedure, n (%)
0 (0.0)
3 (8.1)
3 (6.5)
9 (100.0)
34 (91.9)
43 (93.5)
Type of procedure, n (%)
9 (100.0)
0 (0.0)
9 (19.6)
0 (0.0)
32 (86.5)
32 (69.6)
0 (0.0)
5 (13.5)
5 (10.9)
Length of procedure, mean (SD), min
36.11 (19.3)
57.23 (25.0)
52.5 (25.2)
ADBE, antegrade double-balloon enteroscopy; BAE, balloon-assisted enteroscopy; DBE, double-balloon enteroscopy; FFP, fresh frozen plasma; PRBCs, packed red blood cells; RDBE, retrograde double-balloon enteroscopy; SBE, single-balloon enteroscopy; SD, standard deviation.
1 Acute blood loss was defined as ≥ 2 g/dL drop in hemoglobin from the recorded baseline value.
Primary and secondary outcomes
The primary outcome was to assess the safety of the procedure. Patients tolerated anesthesia without any complications. In addition, there were no complications from the enteroscopy itself, such as perforation, pancreatitis or a need for surgical intervention. One patient had a progressive drop in hemoglobin following a nondiagnostic SBE and 4 days later underwent antegrade double-balloon enteroscopy (ADBE) with a finding of mid-ileal AVMs that were successfully treated with BICAP cautery and endoclips ([Table 4 ]). After ADBE, one patient developed paroxysmal supraventricular tachycardia (PSVT) requiring intensive care unit (ICU) transfer and medical management. Another patient who was already diagnosed with LVAD thrombosis underwent ADBE and died within 72 hours of the procedure due to worsening of the LVAD thrombosis.
Table 4
Outcomes of balloon-assisted enteroscopy procedures.
Type of BAE
SBE (n = 9)
DBE (n = 37)
Total (n = 46)
Adverse events[1 ]
1
2
3
Endoscopic findings, n (%)
4 (44.4)
16 (43.2)
20 (43.5)
0 (0.0)
2 (5.4)
2 (4.3)
0 (0.0)
4 (10.8)
4 (8.7)
0 (0.0)
1 (2.7)
1 (2.2)
0 (0.0)
5 (13.5)
5 (10.9)
1 (11.1)
0 (0.0)
1 (2.2)
0 (0.0)
1 (2.7)
1 (2.2)
4 (44.4)
10 (27.0)
14 (30.4)
Interventions performed, n (%)
5 (55.6)
10 (27.0)
15 (32.6)
4 (44.4)
20 (54.1)
24 (52.2)
0 (0.0)
2 (5.4)
2 (4.3)
0 (0.0)
1 (2.7)
1 (2.2)
0 (0.0)
3 (8.1)
3 (6.5)
0 (0.0)
1 (2.7)
1 (2.2)
APC, argon plasma coagulation; AVM, arteriovenous malformations; BAE, balloon-assisted enteroscopy; BICAP, bipolar circum-active probe; DBE, double-balloon enteroscopy; SBE, single-balloon enteroscopy.
1 None of these events were directly related to the enteroscopy procedure itself.
The secondary outcome was to quantify the diagnostic and therapeutic yields. The cause of bleeding was identified in 32 procedures (69.6 %), with 73.0 % for DBE and 55.6 % for SBE (P = 0.309). Small-bowel angioectasias/AVMs were the most frequent finding accounting for 20 cases (43.5 %) ([Table 4 ]). A therapeutic intervention was carried out in 67.4 % of cases, 73.0 % for DBE and 44.4 % for SBE (P = 0.102). Thermal therapy was used in the majority of procedures: argon plasma coagulation (APC) in 24 procedures (52.2 %), endoclips in two (4.3 %), both APC and endoclips in three procedures (6.5 %), both bipolar circum-active probe (BICAP) cautery and endoclips in one procedure, and APC, BICAP and endoclip in one procedure. [Fig. 1a ], [Fig. 1b ], [Fig. 1c ], and [Fig. 1d ] show therapeutic interventions performed during DBE in an LVAD patient with anemia.
Fig. 1 a In a patient with anemia and left-ventricular assist device (LVAD), antegrade double-balloon enteroscopy was performed and argon plasma coagulation (APC) was used to control a bleeding angioectasia in the jejunum. b Because the bleeding persisted after APC, three hemostatic clips were applied but the angioectasia continued to bleed. c A second round of treatment with APC was performed. d After these therapeutic interventions, hemostasis was finally achieved.
Diagnostic and therapeutic yields of BAE were similar for overt gastrointestinal bleeding (73.1 % and 69.2 %, respectively) and anemia (65 % and 65 %, respectively). However, for those who presented with overt gastrointestinal bleeding, DBE had a higher diagnostic yield compared to SBE (84.2 vs. 42.9 %; P = 0.057). In the same group of patients, the therapeutic yield for DBE was significantly higher compared to SBE (84.2 % vs. 28.6 %; P = 0.014).
Discussion
This is the largest multicenter study of patients with LVADs who underwent DBE for evaluation of gastrointestinal bleeding comprising 34 patients and 46 BAEs (37 DBE, 9 SBE). Our results show that BAE is a safe procedure for patients with LVADs with a good diagnostic and therapeutic yield.
The first and second-year survival benefits for advanced CHF patients on LVADs were demonstrated in the landmark REMATCH trial but also a 42 % incidence of gastrointestinal bleeding at 6 months of follow-up [9 ]. Other studies have reported incidence rates between 20 % to 30 % [3 ]
[10 ]. As noted in our results, the most common source of bleeding is small-bowel angioectasias/AVMs [4 ]
[11 ]
[12 ]
[13 ]
[14 ].
Multiple pathological mechanisms have been proposed to account for the increased incidence of gastrointestinal bleeding. Continuous impeller action and increased turbulence in the outflow and inflow cannula of the LVAD pump causes proteolysis of von Willebrand factor (vWF), thus impairing platelet mediated hemostasis [3 ]
[10 ]
[15 ]. Formation of AVMs may be due to the narrow pulse pressure generated by the LVAD, leading to decreased intraluminal pressure and dilation of the mucosal veins [16 ]. Depending upon the type of LVAD and comorbid conditions, patients may be on long-term anticoagulation and/or antiplatelets, predisposing them to an increased risk of gastrointestinal bleeding [11 ].
Apart from physiological changes induced by the LVAD, use of BAE for evaluation of gastrointestinal bleeding in patients with LVADs presents unique challenges. A multidisciplinary team approach is required, usually consisting of a gastroenterologist, cardiothoracic surgeon, cardiac anesthesia, and transplant team. Patients are often intubated under general anesthesia and supported by inotropes to counter the hypotension induced by anesthetic agents. It is also imperative to avoid excess upper abdominal pressure as it can impede venous return and decrease the preload required for optimal LVAD functioning [5 ]
[11 ]
[17 ]. More importantly, data regarding the safety and utility of BAE, in particular DBE, in the LVAD population are scant.
Approximately half of our population underwent VCE initially as part of the gastrointestinal bleeding workup. VCE is a safe and practical option in patients with LVADs to diagnose small intestinal bleeding [18 ]
[19 ]
[20 ]
[21 ]. Once a lesion has been identified on VCE, a BAE is then necessary to access the small intestine for visualization and treatment of the culprit lesions. With use of BAE, patients can avoid exposure to more invasive testing such as mesenteric angiograms or exploratory laparotomy [11 ]
[17 ]
[22 ]. But major AEs related to BAE have been previously reported, including pancreatitis, perforation, aspiration pneumonia, and death [23 ]
[24 ]. Others have reported no AEs associated with the procedure [5 ]
[6 ]
[10 ]. In our study, no major AEs were noted related to the enteroscopy procedure despite the majority of patients being classified as ASA class IV.
One patient developed an arrhythmia following the procedure, and it was medically managed in the intensive care unit (ICU). In another patient, hematocrit continued to drop following SBE, necessitating DBE that was diagnostic and therapeutic for mid-ileal AVMs. There was one death in our study population: a patient with a preexisting history of LVAD thrombosis who was admitted to the ICU with melena and acute drop in hemoglobin. The patient’s international normalized ratio was supratherapeutic, thus antithrombotics (aspirin and warfarin) were held. The patient underwent esophagogastroduodenoscopy and push enteroscopy, which were both negative. The next day, DBE was performed and it was diagnostic for a Dieulafoy’s lesion that was managed with argon plasma coagulation. But within a few hours of the procedure, LVAD thrombosis recurred. The patient was started on intravenous heparin and thrombolytic therapy was not administered due to the history of gastrointestinal bleeding. The patient was considered a poor candidate for LVAD exchange or urgent cardiac transplant due to multiple comorbidities, and after discussion with the patient and their family, palliative care was provided. The patient died 4 days after the DBE. Thrombosis is a well-known medical complication of LVADs; patients with LVAD thrombosis have higher rates of bleeding and an increased mortality rate, especially when managed medically (as was the case with our patient) versus surgically [25 ]. This case underscores the dilemma of balancing the fragile hemostatic state of patients with LVAD thrombosis who are on antithrombotics and present with gastrointestinal bleeding. The decision to hold antithrombotics should be made in consultation with a cardiologist in order to assess the risk of LVAD thrombosis [26 ].
The diagnostic yield for evaluation of gastrointestinal bleeding in patients with LVADs who underwent BAE was 69.6 % in our study. This is close to the previously reported diagnostic yield of 64 % to 94 % for BAE performed in the setting of gastrointestinal bleeding in patients with small-bowel diseases [27 ]. The diagnostic yield for DBE was 73.0 % in our study; this is comparable to the diagnostic yield of 69 % previously reported by Edwards et al for LVAD patients undergoing DBE [7 ]. The diagnostic yield for SBE was 55.6 % in our study; Koul et al have reported a diagnostic yield of 78 % for a similar cohort of patients. The timing of the procedure has been known to influence the diagnostic yield, with shorter times increasing the yield [7 ]. In our study, most procedures were performed after 48 hours, however, Koul et al did not provide data on timing of the procedures and they had a relatively larger cohort of patients undergoing SBE.
We did not observe a significant difference between the diagnostic or therapeutic yields for SBE and DBE in the entire cohort. However, in the subgroup of patients that presented with gastrointestinal bleeding, DBE had a higher diagnostic yield than SBE that bordered statistical significance and reached significance for therapeutic intervention. Other studies have also reported that DBE has a higher diagnostic yield in patients with overt gastrointestinal bleeding [28 ]
[29 ].
There are several limitations to this study. The study cohort was relatively small (34 patients), however, this is the only multicenter study to date and the largest cohort of LVAD patients undergoing DBE. In our study, patients were followed for 72 hours post-procedure to note any complications but it can be argued that a longer duration of follow-up might have revealed additional complications, including recurrent gastrointestinal bleeding events requiring repeat BAE. Small-bowel vascular lesions have been classified into four groups based on Yano-Yamamoto classification: angioectasias, Dieulafoy’s lesions, AVMs, and unclassifiable [30 ]. Based on this classification, angioectasias are venous lesions that should be treated with cauterization whereas Dieulafoy’s and AVM’s cause arterial bleeding that is likely to require hemostatic clips. In our study, lesions could not be classified based on the Yano-Yamamoto classification because this system was not used consistently at the different study centers. Although use of this classification could affect the treatment decision, it would not alter the diagnostic or therapeutic yields which were the outcomes of this study.
Conclusion
From our study, we can conclude that BAE in patients with LVAD is a safe and effective option in management of gastrointestinal bleeding. In patients that presented with overt gastrointestinal bleeding, DBE resulted in more therapeutic interventions than SBE. Prospective trials with larger patient populations and extended follow-up are needed.
