Multicenter Evaluation of Octreotide as Secondary Prophylaxis in Patients With Left Ventricular Assist Devices and Gastrointestinal Bleeding

Introduction

Gastrointestinal (GI) bleeding is a frequent complication after continuous-flow left ventricular assist device (CF-LVAD) implantation.^1,2 In some reports, over a third of patients with CF-LVAD implantation will experience a significant GI bleeding episode at some time point after implantation.³ The causes and sources of bleeding are multifactorial.^4–6 The only consistently reported demographic risk factor for LVAD-related GI bleeding seems to be age >65 years⁵; however, other clinical correlates include a history of previous GI bleeding, preoperative use of antiplatelet or anticoagulant medications, poor kidney function, and device strategy for LVAD implantation (destination therapy [DT] versus bridge-to-transplantation. The cause of bleeding is found in ≈75% of the patients. Of those who rebleed, ≈75% have similar symptoms, and the cause of rebleeding is the same as the index bleeding episode in 50% of the patients. The anatomic location of bleeding is predominantly upper GI in origin, and most of them are from angiodysplastic lesions.⁷

Acquired von Willebrand Syndrome has been shown to develop in patients with CF-LVADs presumably because of high shear forces resulting in destruction of high–molecular weight von Willebrand factor multimers by the LVAD impeller.^8,9 It has been shown that these patients have a higher than normal occurrence of GI angiodysplasia. Whether these arteriovenous malformations are preexisting or result from de novo angiogenesis from the gut mucosa is unknown. It is thought that the loss of pulsatility and potential hypoxia result in upregulation of angiogenic mediators, such as vascular endothelial growth factor and angiopoietin-2, may contribute to the development of angiodysplasia.^10–12 Consequently, appropriate primary and secondary bleeding prophylaxis is essential in improving long-term outcomes after LVAD implantation.

The current treatment for LVAD-associated GI bleeding is primarily focused on identifying the location of bleeding with subsequent endoscopic intervention, along with adjustments to a lower intensity of anticoagulation and antiplatelet therapy. Beyond that, the policies to manage these patients continue to be institution specific. There is a dearth of evidence-based data on how to prevent the risk of recurrent GI bleeds post-LVAD placement. These patients may also receive supportive treatment with blood product transfusion, cryoprecipitate, and desmopressin; however, these approaches do not prevent recurrent hospitalizations and are associated with high costs and morbidity. Importantly, patients receiving multiple transfusions can develop antibodies that make subsequent cardiac transplantation more difficult.

Octreotide acetate, a somatostatin analog, has been thoroughly studied in various conditions and has a known safety profile. It is currently used to control active GI hemorrhage and can also be used for secondary bleeding prophylaxis for vascular ectasias in non-LVAD patients. There are meta-analysis–level data showing decreased need for transfusion in patients with arteriovenous malformations treated with octreotide.¹³ However, the data supporting the use of octreotide for bleeding prophylaxis in LVAD patients are sparse and limited to case reports and single-center experiences.^13–16 A recent prospective study was conducted with octreotide in stable LVAD patients and was noted to be well tolerated with no drug-related side effects or GI bleeding events observed.¹⁷

To further elucidate the potential role of octreotide in preventing recurrent LVAD-associated GI bleeding, we conducted a retrospective, multicenter cohort study to determine the incidence of rebleeding in patients receiving secondary octreotide prophylaxis. Herein, we report the demographics and clinical characteristics associated with rebleeding versus non–rebleeding and further compare these data to a historical control group comprised patients who experienced GI bleeding in the original clinical trial of the HMII (HeartMate II) LVAD.

Methods

Patients and Selection Criteria

This retrospective multicenter study was conducted at Virginia Commonwealth University, The Ohio State University, Columbia University, The University of Chicago, and Allegheny General Hospital. Each site received institutional review board approval before study initiation. The study inclusion criteria included age >18 years, the placement of an HMII LVAD, hospitalization for GI bleeding and subsequent receipt of standard-of-care treatment, and receipt of octreotide acetate given as either monthly depot injection or twice daily subcutaneous injection on hospital discharge. Patients who received octreotide for primary GI bleeding prophylaxis or for another medical indication were excluded.

Data Sources and Collection

Fifty-one patients meeting inclusion criteria at the 5 centers from the years 2009 to 2015 were identified using International Classification of Diseases, Ninth Revision, billing codes and queries of local health system databases. Chart reviews were completed to confirm LVAD placement, GI bleed, and receipt of octreotide treatment for secondary bleeding prophylaxis on hospital discharge. The patients who met these criteria had their anonymized data entered into a secure, Health Insurance Portability and Accountability Act compliant research electronic data capture (REDCap) database. To ensure data entry validity, each center recollected data on 3 patients randomly selected by the coordinating site, and data congruency was analyzed to ensure that the collection strategies were robust. The data, analytic methods, and study materials will not be made available to other researchers for purposes of reproducing the results or replicating the procedure.

In addition to demographics, the clinical characteristics for each patient were collected, which included past and concurrent medical history, medications, LVAD system settings, laboratory data (kidney function, hemoglobin, international normalized ratio [INR]), and interventions performed in response to GI bleed.

These interventions included blood products received (eg, packed red blood cells, fresh frozen plasma, cryoprecipitate) and platelet transfusions. In addition, endoscopy procedures and medication interventions were documented.

Historical Control Group

As a reference for comparing GI rebleeding rates, a historical control group of HMII LVAD patients who had experienced a GI bleed in the original combined HMII bridge-to-transplantation and DT clinical trials (NCT00121472 and NCT00121485, respectively) was identified as a subgroup of the 956 patient cohort previously published by Boyle et al.¹⁸ The clinical bridge-to-transplantation and DT studies enrolled 1302 patients who received an HMII between March 2005 and January 2010. After excluding patients who were implanted for compassionate use or rescue exchange for HeartMate XVE, a total of 956 patients were successfully discharged from the hospital. In this group, there were 240 patients who suffered a GI bleed. Because of the differences in baseline characteristics, each patient in the octreotide study group was matched to a patient experiencing GI bleeding in the HMII clinical trial with the same indication and the closest LVAD support time at the time of the reference GI bleed. Any ties were then broken by matching to the same sex and if needed the closest age at time of the reference GI bleed.

Statistical Analysis

Continuous variables were presented as mean and SD or median and interquartile range. Intergroup differences between rebleeders and non–rebleeders in the current study were assessed using the Student t test for continuous data. For nominal data, a χ² test was performed unless the frequencies in the contingency table were ≤5, in which case the Fisher Exact test was used. Freedom from rebleeding within 6 months after the index event was estimated by the Kaplan–Meier method. The reference time zero in the treatment group was the occurrence of the GI bleeding event first treated with octreotide regardless of prior GI bleeding events on LVAD support, and for the historical control, it was the occurrence of the first GI bleeding event after LVAD implantation. Comparison of freedom from rebleeding was performed with the log-rank test. Statistical significance was defined as P<0.05. All data summaries and analyses were conducted using SPSS statistical software.

Results

Table 1 shows the patient demographics and clinical features of the 51 subjects who received octreotide. Approximately half of the patients were implanted with a CF-LVAD as DT, and 41% had experienced a prior GI bleed on the device. Twelve of the 51 subjects (24%) in this study experienced a rebleed within 6 months of beginning octreotide treatment.

When comparing subject demographics between those who rebled to non–rebleeders, there were no statistically significant differences. There were also few differences in the clinical characteristics between these subgroups. Fewer of the rebleed subjects had atrial fibrillation compared with the nonbleeders (8% versus 44%, respectively; P=0.04); however, the rebleeding subjects had a greater prevalence of a history of GI bleeding before LVAD placement (33% versus 5%, respectively; P=0.02).

Table 2 shows the patients’ clinical characteristics on their initial admission for GI bleed. There were no statistically significant intergroup differences in LVAD parameters (speed, power, pulsatility), nor were there differences in kidney function or hemoglobin values. Most patients (41%) were taking 81 mg aspirin daily. Similarly, 84% of patients were on warfarin therapy with the target INR ≈2.3 in both groups. There was a trend toward significance in the mean INR at the time of bleeding, with the rebleed group having a lower INR compared with the non–rebleeders (INR, 2.0±0.6 versus 3.0±1.9; P=0.08). At the time of GI bleed, very few patients were on dipyridamole, low–molecular weight heparin, or a novel oral anticoagulant, whereas 67% in both groups were receiving proton pump inhibitor therapy.

The clinical interventions performed in the octreotide-treated group are provided in Table 3. The majority of patients (72%) received the monthly octreotide depot formulation after experiencing an LVAD-related GI bleed. There were no statistically significant differences in the amount of packed red blood cells used nor the percentage of patients who received fresh frozen plasma. During the reference bleeding hospitalization, there was a trend toward more patients in the rebleeding group receiving both cryoprecipitate and platelet transfusions compared with non–rebleeders (17% versus 0%, respectively; P=0.05). Nearly all of the population (92%) had an endoscopic procedure performed, with significantly more rebleed patients having confirmed angiodysplasia compared with non–rebleeders (58 versus 23%, respectively; P=0.03). Most of the patients (57%) had a warfarin dose reduction in response to the bleed.

Comparison to Historical Control Group

The octreotide cohort was compared with appropriately matched patients in the HMII clinical trials. There were 240 patients in the clinical trials who experienced a GI bleed on HMII support, and 40% of them had a rebleeding event. However, the majority of these 240 patients (73%) were implanted as DT compared with 49% of the octreotide-treated group, and the median LVAD support duration at the time of the reference GI bleed was also shorter compared with the octreotide-treated patients (91 versus 197 days). To control for differences in baseline characteristics, each octreotide-treated patient was matched (1:1) to a patient experiencing GI bleeding from the historical cohort with the same indication and a similar support duration on device. Comparison of baseline characteristics between the octreotide-treated group and the matched historical control group is shown in Table 4. Patients in the octreotide-treated group have a lower rate of rebleeding compared with matched patients enrolled in the HMII clinical trials (24% versus 43%; P=0.04). Kaplan–Meier curves showing freedom from GI rebleed curves for both groups are shown in Figure. Patients in the octreotide-treated group had a significantly higher freedom from GI rebleed compared with patients in the matched historical control (75±6% versus 52±8%; P<0.01).

• Download figure
• Download PowerPoint

Discussion

Even with more than a decade of clinical experience with CF-LVADs, GI bleeding remains a prevailing complication after device implantation. In contemporary clinical trials, irrespective of device design, the rates of GI bleeding in patients in the short term or bridge-to-transplantation population are 15% at 6 months of follow-up.^19,20 For patients implanted without the intention of transplantation, GI bleeding rates exceed 30% at 1 year, even in less sick ambulatory patients implanted before becoming inotrope dependent.^21,22 To put this further into context, the event rate of GI bleeding in patients with CF-LVADs at 65 per 100 patient-years is much higher compared with 2.60 to 4.6 per 100 patient-years in patients with mechanical valves (on aspirin and warfarin), and 8% in patients receiving triple antithrombotic therapy (aspirin, clopidogrel, and warfarin). In fact, no studies have investigated the optimal anticoagulation management in the cohort of patients with CF-LVADs and GI bleeding. Efforts to understand this adverse event need to extend beyond single-center analyses and focus on therapeutic options targeting different mechanistic causes of bleeding.

Studies such as ours present an important insight into the multifactorial mechanisms that come together to result in GI bleed in patients with CF-LVADs. This is the first multicenter study to evaluate rebleeding from the GI tract in LVAD patients after treatment with octreotide for secondary prophylaxis. After treatment, 24% of the LVAD patients experienced a GI rebleed, which was <43% from the matched historical control patients from the HMII clinical trial. Compared with the bleeding data from the HMII clinical trials,¹⁸ our data appear similar with respect to demographics and clinical characteristics.

Because of the limited number of events in the octreotide-treated group, we can only qualitatively characterize patients who rebled after octreotide. Patients who rebled after treatment with octreotide had clinical interventions suggestive of more severe bleeding during the index bleed, including trends toward a greater number of blood product transfusions and a higher number of endoscopic procedures and were more likely to have a history of GI bleeding before LVAD implantation. Given that nearly all patients with assist devices who experienced GI bleeding had therapeutic or subtherapeutic INR at the time of their bleeding events, it is clear the antithrombotic therapy alone does not account for the higher rate of bleeding in CF-LVADs. Indeed, those that rebled were more likely to have confirmed angiodysplasia on endoscopy, and these patients underwent more endoscopic procedures to identify sources of bleeding, likely indicators of more diffuse disease in this group.

A few previously published retrospective, single-center studies have examined the reoccurrence of GI bleeding in CF-LVAD patients. In HMII patients with GI bleeding, Jabbar et al²³ reported a rebleed rate of 43% (19/44 patients), whereas Morgan et al²⁴ observed a rebleed rate of 21% (4/19 patients). Goldstein et al¹⁹ described a 34% rebleed rate (20/59) in CF-LVAD patients with a history of previous GI bleeding. In this study, the rebleed rate was 40% in the 240 patients enrolled in the HMII clinical trials, which represents the largest studied cohort of patients with a CF-LVAD–related GI bleed.

One inherent limitation of our study is the use of a historical control group. Although the 1:1 matching has resulted in well-balanced patient groups with similar baseline characteristics, it did not control for potential effects of time era associated variables such as changes in clinical practice. However, considering that the rates of GI bleeding have not decreased in more contemporary clinical studies over a variety of CF-LVAD platforms,^19–23 one might argue that any changes in clinical practice between the time eras of the groups have minimal impact on GI bleeding rates. Furthermore, within the 6-year time period of the HMII clinical trials, we found no significant impact of implant era on GI rebleeding (Figures I and II in the Data Supplement). The findings in the present study should be considered to be hypothesis generating and highlight the need for a randomized controlled clinical trial which would more definitively define the risks and benefits of octreotide therapy.

The study has additional limitations related to the retrospective study design that precludes definitive conclusions on the efficacy of octreotide. First, the patients treated with octreotide did not receive a standard formulation, dose or duration of therapy. This is important because octreotide has different formulations including an immediate-release injection given twice daily, and a long-acting release depot injection given monthly. The monthly injection theoretically should increase patient adherence compared with the immediate-release formulation. However, adherence to therapy could not be measured in this study nor could the administered octreotide dose. It is possible that those patients who bled were receiving suboptimal dosing compared with the nonbleeders. Moreover, the medical record in many cases did not capture date of drug discontinuation, or whether patients were unable to reliably obtain the drug because of financial constraints. Second, although the comparison of rebleed rates in the octreotide treatment group to the HMII clinical trials is provocative and further posits the benefits of octreotide, the 2 groups have notable differences. The clinical trial cohort was studied from the time of the first GI bleed after LVAD, whereas the octreotide group was studied at the time of their octreotide-treated bleed. Therefore, the octreotide group selected for patients who survived longer from their LVAD and patients who already suffered a GI bleed on the device. Furthermore, it was not possible to effectively control for all possible confounding variables because the HMII clinical trial patients had limited data collected at the time of the GI bleed.

In conclusion, we observed that the rate of rebleeding in CF-LVAD patients treated with octreotide after a GI bleed was noticeably less than previous clinical trial data. These results are encouraging and support the need for defining the efficacy and safety of octreotide in a prospective, randomized, clinical trial.

Acknowledgments

We thank Jennifer Haas (Ohio State University).

Footnotes