The endothelium is a diaphanous film of tissue that invests the lumen of all blood vessels. This single-cell monolayer senses and is responsive to the hemodynamic forces of shear stress and cyclic strain to maintain normal vessel health. Endothelial cells release factors that regulate vessel tone, vessel structure, and controls the interaction of the vessel wall with circulating blood elements.1–3 Due to a relative reduction in pulsatility in the setting of a Continuous Flow-Left Ventricular Assist Device (CF-LVAD) support, the conventional hemodynamic forces acting on the endothelium are altered and possibly disrupt normal endothelial signaling. Patients with CF-LVADs have a higher risk of bleeding events secondary to arteriovenous malformation and impaired platelet aggregation.4–8 Mucosal bleeding (predominantly presenting as gastrointestinal bleeding secondary to angiodysplasia) is the most common adverse event after CF-LVAD and contributes to morbidity and hospitalizations.9,10 While high blood pressure has been associated with cerebral bleeding in CF-LVAD patients and shear stress-induced degradation of von Willebrand Factor (vWF)11 contributes to GI bleeding, the role of endothelial function in these adverse events has not been studied.
Flow-mediated dilatation of the brachial artery by high-resolution ultrasound imaging has been used to assess endothelial function. Subjects with CF-LVADs have more significant impairment of flow-mediated dilatation compared with subjects with older generation pulsatile flow devices.12 Newer noninvasive tests that rely on fingertip small vessel arterial circulation have been developed that are simple to implement in an office setting and have been validated to assess endothelial dysfunction as a marker of future risk of cardiovascular events.13,14 Utilizing 1 such technology that relies on digital thermal changes, we aimed to test the hypothesis that vasodilatory endothelial dysfunction is associated with bleeding events in CF-LVAD patients.
METHODS
We enrolled stable outpatients supported with a durable CF-LVAD from the clinic at our institution between July 2014 and September 2016. We conducted a cross-sectional study with the approval from our institutional review board and all patients consented.
Digital Thermal Monitoring
We used a noninvasive, operator-independent device called VENDYS® to measure endothelial vasodilatory function. This technology is based on digital thermal measurements in the peripheral circulation at the acral segment of the finger before and after the release of a brachial artery occlusion to assess vasodilatory response. The VENDYS® test is commercially available and has documented reproducibility in measuring vasodilatory response. Endothelial dysfunction by this test is defined by a reduced vasodilatory response and predicts cardiovascular events.15,16 The test takes approximately 15 minutes to perform using the manufacturer’s guidelines. We performed the test in a quiet, temperature-controlled room in the outpatient setting (with ambient relaxing music). Temperature sensors were attached to both index fingers and blood pressure cuffs placed on both arms with the left arm being the control. There is an initial stabilization phase for 5 minutes where the blood pressure cuff autoinflates to record the baseline blood pressure. Baseline temperature of both the fingers is measured during this initial phase. The right-arm cuff then inflates to 250–300 mm Hg in the next 5 minutes to achieve full brachial artery occlusion followed by cuff release in the final 5 minutes. Reestablishment of blood flow after cuff release results in a reactive hyperemic response, as manifested by a rebound in the temperature, generally to higher levels than the baseline temperature of the test arm. Thermal changes before, during and after arm occlusion from each patient are computed, and the software generates a vascular reactivity index (VRI) with predefined categories based on observed and expected temperature curves.17 (VRI < 1: poor endothelial function; VRI 1–2: intermediate endothelial function; and VRI >2: good endothelial function) (Figure 1, A and B). These categories have been previously established for prediction of adverse cardiovascular events in non-LVAD populations. For the purpose of our study, we decided to compare those with VRI < 1 to patients with VRI>1.The software accounts for manufacturer built-in quality checks during the test performance. Resting fingertip temperature ranges from 31° to 33° and is affected by sympathetic activation. The zero-reactivity curve reported in the VENDYS® test report is automatically adjusted for these normal variations. However, if the variations exceed the normal range, the software flags the test and alerts the operator that the test conditions are suboptimal. The cold finger is defined as a baseline fingertip temperature below 27° per manufacturer guidelines and is indicative of increased sympathetic activity, which can confound the results. VRI was measured at the time of clinic visit. Patients were followed for outcomes as defined earlier. No follow-up endothelial function testing was performed. In the patients with a cold finger, we repeated the test 30 minutes apart, and the conditions were optimized including adjusting the room temperature, using a VENDYS® blanket to warm up if needed and instructing the patients to relax. The reproducibility of this test is yet to be defined in CF-LVAD patients.
;)
Figure 1.: VENDYS® SYSTEM of DTM. A: Graphical representation of a patient connected to a digital thermal monitoring system. B: Representation of the three phases in performing the DTM testing. DTM, digital thermal monitoring.
Aortic Valve Opening Assessment
For patients with a CF-LVAD, a GE V-scan® portable hand-held echocardiogram machine was utilized for assessment of aortic valve (AV) opening right before the performance of the digital thermal monitoring test. A parasternal long axis view was utilized to visualize the AV. Five cardiac cycles were observed, and the opening was classified as 0—no opening for all cardiac cycles; 1—opening intermittently; and 2—opening for all cardiac cycles.
Outcomes
Clinical variables were documented at the time of the testing for patients with CF-LVADs (Table 2). Predefined bleeding outcomes were abstracted from chart review prospectively for 1 year from the time of the VENDYS® test bleeding events included hospital admission for any gastrointestinal bleeding (hematemesis, coffee-ground emesis, melena, and heme-positive stools), intracranial hemorrhage (evidence on CT scan), epistaxis, hemoptysis, and any other suspected bleeding without identifiable source (needing blood transfusions). Survival data were collected from internal quality data.
Statistical Analysis
Demographic and clinical data were reported as frequencies and proportions for categorical variables and as median and interquartile range or mean (standard deviation, SD) for continuous variables as appropriate. Differences between groups were determined by χ2 or Fisher’s exact tests for categorical variables and unpaired t-test or Kruskal Wallis test for continuous variables as appropriate. Kaplan-Meier curves were used to depict patient survival and bleeding-free survival within 1 year from VENDYS® testing, stratified by the VRI.
Cox proportional hazards modeling was used to determine the contribution of potential prognostic variables to the patient outcome. Significant variables in the univariate analysis or variables considered as clinically relevant were investigated further by the multivariate modeling. The Likelihood Ratio test was used to reduce the model subsets further. The best model was selected based on the smallest Bayesian information criterion (BIC). The model discrimination was determined by the C-statistic. All the analyses were performed using Stata version 15.1 (StataCorp LLC; College Station, TX). A P value of <0.05 was considered statistically significant.
RESULTS
A total of 56 patients with a HeartMate-II CF-LVAD underwent the VENDYS® test successfully (11 patients were excluded for the quality control alerts described in the methods section). Median days of CF-LVAD support was 438 (117, 849) days at the time of the. Majority of the patients had a CF-LVAD implant for more than 6 months. Sixty-six percent were implanted as a destination therapy strategy. While 57% had a persistent AV opening, only 21% had a persistently closed AV (Table 1). Demographics and clinical variables were compared between patients with poor endothelial function (VRI < 1) and the rest of the group (VRI ≥ 1) (Table 2). The number of males and international normalized ratio were significantly higher in the VRI ≥ 1 group, and there was no difference in blood pressure, laboratory tests, and neuro-hormonal modulating medication use between the groups (Table 2).
Table 1. -
CF-LVAD Parameters
|
Total (N = 56) |
VRI < 1 (n = 11) |
VRI ≥ 1 (n = 45) |
P
|
| Days of LVAD support at the time of VENDYS®, median (IQR) |
437.5 (117.0, 849.0) |
574.0 (123.0, 1,117.0) |
435.0 (111.0, 730.0) |
0.64 |
| Therapy goal of LVAD |
|
|
|
0.69 |
| LVAD as a BTD |
9 (16.1) |
1 (9.1) |
8 (17.8) |
|
| LVAD as a BTT |
10 (17.9) |
1 (9.1) |
9 (20.0) |
|
| LVAD as a DT |
37 (66.1) |
9 (81.8) |
28 (62.2) |
|
| Flow, median (IQR) |
5.1 (4.6, 5.9) |
5.2 (4.4, 6.6) |
5.0 (4.6, 5.8) |
0.55 |
| Speed, median (IQR) |
9,200 (8,800, 9,400) |
9,200 (8,990, 9,580) |
9,190 (8,800, 9,400) |
0.42 |
| Pulse index, median (IQR) |
5.6 (4.5, 6.4) |
5.7 (4.5, 6.3) |
5.5 (4.5, 6.5) |
0.90 |
| Power, median (IQR) |
5.8 (4.9, 6.5) |
6.2 (4.6, 6.7) |
5.7 (4.9, 6.4) |
0.60 |
| Atrioventricular opening |
|
|
|
0.72 |
| Opened |
32 (57.1) |
5 (45.5) |
27 (60.0) |
|
| Closed |
12 (21.4) |
3 (27.3) |
9 (20.0) |
|
| Intermittent opened |
12 (21.4) |
3 (27.3) |
9 (20.0) |
|
Values are in number and % unless otherwise specified.
BTD, bridge to decision; BTT, bridge to transplant; DT, destination therapy; CF-VAD, continuous flow ventricular assist device; IQR, interquartile range; LVAD, left ventricular assist device; VRI, vascular reactivity index.
Table 2. -
Demographics and Clinical Parameters of Patients With CF-LVAD
|
Total (N = 56) |
VRI < 1 (n = 11) |
VRI ≥ 1 (n = 45) |
P
|
| Age (years), median (IQR) |
61.7 (51.2, 68.8) |
63.8 (50.0, 70.2) |
61.2 (51.5, 68.4) |
0.64 |
| Males (%) |
44 (78.6) |
5 (45.5) |
39 (86.7) |
0.003 |
| Whites (%) |
25 (44.6) |
6 (54.5) |
19 (42.2) |
0.51 |
| Body mass index ≥30 (%) |
29 (51.8) |
6 (54.5) |
23 (51.1) |
0.84 |
| Diabetes (%) |
36 (64.3) |
8 (72.7) |
28 (62.2) |
0.51 |
| Hypertension (%) |
41 (73.2) |
7 (63.6) |
34 (75.6) |
0.42 |
| COPD (%)†
|
16 (28.6) |
5 (45.5) |
11 (24.4) |
0.17 |
| Chronic kidney disease (%) |
17 (30.4) |
5 (45.5) |
12 (26.7) |
0.22 |
| Hyperlipidemia (%) |
20 (35.7) |
3 (27.3) |
17 (37.8) |
0.51 |
| Ischemic cardiomyopathy (%) |
31 (55.4) |
5 (45.5) |
26 (57.8) |
0.46 |
| Mean blood pressure, mean (±SD) |
87.8 (±15.8) |
90.7 (±16.0) |
87.1 (±15.8) |
0.50 |
| Systolic blood pressure, mean (±SD) |
110.7 (±18.2) |
118.0 (±25.0) |
108.9 (±16.0) |
0.14 |
| Diastolic blood pressure, mean (±SD) |
72.4 (±15.0) |
74.0 (±17.7) |
72.0 (±14.4) |
0.69 |
| Pulse pressure, median (IQR) |
38.3 (±16.7) |
44.0 (±20.5) |
36.9 (±15.6) |
0.21 |
| INTERMACS 1 (%) |
11 (20.8) |
3 (27.3) |
8 (19.0) |
0.55 |
| INTERMACS 2 (%) |
12 (22.6) |
4 (36.4) |
8 (19.0) |
0.22 |
| Ejection fraction, median (IQR) |
22.0 (20.0, 27.0) |
21.0 (20.0, 27.0) |
22.0 (20.0, 27.0) |
0.41 |
| GFR‡, mean (±SD) |
52.3 (±18.5) |
52.4 (±21.5) |
52.3 (±18.0) |
0.99 |
| Creatinine, median (IQR) |
1.4 (1.1, 1.8) |
1.3 (1.0, 1.8) |
1.5 (1.2, 1.8) |
0.41 |
| Bilirubin, median (IQR) |
0.5 (0.4, 0.7) |
0.4 (0.2, 0.7) |
0.5 (0.4, 0.7) |
0.40 |
| Hemoglobin, median (IQR) |
10.3 (9.5, 12.3) |
9.8 (9.7, 11.7) |
10.7 (9.4, 12.4) |
0.44 |
| Platelets, median (IQR) |
197.0 (162.0, 233.0) |
198.0 (185.0, 234.0) |
197.0 (140.0, 229.5) |
0.42 |
| International normalized ratio, median (IQR) |
1.9 (1.6, 2.2) |
1.6 (1.3, 2.0) |
2.0 (1.6, 2.3) |
0.04 |
| Partial thromboplastin time, median (IQR) |
41.0 (36.8, 45.8) |
43.2 (40.4, 45.3) |
40.7 (35.6, 46.6) |
0.54 |
| Thromboelastography, maximal amplitude (55-73 mm), median (IQR) |
66.6 (61.8, 68.9) |
67.5 (64.9, 68.2) |
66.3 (60.8, 69.1) |
0.36 |
| Lactate dehydrogenase, median (IQR) |
304.5 (253.0, 353.0) |
303.0 (265.0, 340.0) |
305.0 (251.0, 372.5) |
0.89 |
| Warfarin (%) |
53 (94.6) |
10 (90.9) |
43 (95.6) |
0.54 |
| Aspirin (%) |
49 (87.5) |
8 (72.7) |
41 (91.1) |
0.10 |
| ACE* inhibitors (%) |
22 (39.3) |
4 (36.4) |
18 (40.0) |
0.82 |
| Furosemide (%) |
35 (62.5) |
6 (54.5) |
29 (64.4) |
0.54 |
| Spironolactone (%) |
20 (36.4) |
5 (45.5) |
15 (34.1) |
0.48 |
Values are in number and % unless otherwise specified.
*Angiotensin converting enzyme.
†Chronic obstructive pulmonary disease.
‡Glomerular filtration rate.
CF-VAD, continuous flow ventricular assist device; IQR, interquartile range; VRI, vascular reactivity index.
Patients with VRI < 1 had poorer survival at 1 year compared with patients having VRI ≥ 1 [72.7% (95% confidence interval [CI]: 37.1–90.3) vs. 88.6% (95% CI: 74.6–95.1), respectively]. (Figure 2A). Compared with patients having VRI ≥ 1, patients having VRI < 1 had significantly lower bleeding-free survival [45.7% (95% CI: 14.3-73.0) vs. 73.7% (95% CI: 57.4-84.5), P = 0.04] (Figure 2B). Table 3 lists the distribution of the various kinds of bleeding events. There was a significant increase in the proportion of hemorrhagic strokes in patients with VRI < 1 compared with patients having VRI ≥ 1 (3/11, 27% vs. 3/45, 7%; P = 0.048). The international normalized ratio of the patients at the time of bleeding events was 1.87 ± 0.64. In univariate analysis, VRI < 1 was significantly associated with bleeding. Table, Supplemental Digital Content 1, https://links.lww.com/ASAIO/A546 shows the variables included in the univariate analysis. In adjusted Cox proportional hazards model, VRI < 1 [HR: 5.56 (1.56–19.8); P = 0.01] and Hemoglobin [0.62 (0.4–0.96); P = 0.03] were significant variables.
Table 3. -
Breakdown of the Various Kinds of Bleeding Events (Within 1-year Post-VENDYS®) and Comparison of Patients With VRI < 1 and those ≥1
|
TotalN = 56 (%) |
VRI < 1n = 11 (%) |
VRI ≥ 1n = 45 (%) |
P
|
| Gastrointestinal bleeding |
9 (16) |
1 (9) |
8 (18) |
0.14 |
| Epistaxis |
4 (7) |
2 (18) |
2 (4) |
0.11 |
| Hemorrhagic stroke |
6 (11) |
3 (27) |
3 (7) |
0.048 |
| Other bleeding (hemoptysis, transfusion) |
3 (5) |
0 |
3 (7) |
0.38 |
Represented as number of patients with percentage of patients within the group. Specific number of events for these patients are reported in Table, Supplemental Digital Content 2,
https://links.lww.com/ASAIO/A567.
VRI, vascular reactivity index.
Figure 2.: A: Kaplan-Meier curves for survival of patients with VRI < 1 vs. VRI ≥ 1 for a follow-up of 1 year; B: Kaplan-Meier curves for bleeding-free survival of patients with VRI < 1 vs. VRI ≥ 1 for a follow-up of 1 year. VRI, vascular reactivity index.
DISCUSSION
Initial fears of adverse physiologic consequences of loss of pulsatility had been alleviated by the superior survival of continuous flow pumps in the current era.18,19 Despite such assurance, morbidity remained high after CF-LVAD placement.20–22 Many studies have documented the aberration in endothelial function in patients with continuous flow physiology20 when compared with pulsatile pumps. While the reason for endothelial dysfunction can be a direct loss of the normal cyclic fluctuations of endothelial shear stress or other alterations in the immune system that interact adversely with the endothelium,23 the clinical relevance of this dysfunction has not been established.
Our study is the first to show that a simple clinic-based test assessing endothelial function is associated with bleeding. The contributors for bleeding events in the setting of a CF-LVAD has been presumed to be multifactorial and may include acquired vWF deficiency, acquired impaired platelet aggregation, and angiodysplasia.12,24,25 Endothelial dysfunction has a biologic plausibility and could be the missing contributing factor in these patients. Individuals with a poor endothelial vasodilatory function defined as a VRI < 1 measured using a digital thermal monitoring machine had a 5 times higher likelihood of a bleeding event when compared with those with a VRI ≥ 1. This adverse outcome was driven by hemorrhagic strokes. A recent analysis of Interagency Registry for Mechanically Assisted Circulatory Support (INTERMACS) data showed that the incidence of stroke was 0.123 per patient-year with an almost equal representation of ischemic and hemorrhagic etiologies.26 While any stroke portended a poor prognosis, hemorrhagic stroke was particularly devastating with a 30-day survival of 45% compared with 80% survival after ischemic stroke. With our study revealing an association of endothelial dysfunction with future hemorrhagic stroke, further studies are warranted to see if we can use vascular reactivity testing to prospectively stratify individuals at risk for this devastating outcome and device tailored interventions to modulate the risk. An interesting observation to note was higher number of GI bleeds in individuals with VRI ≥ 1 compared with VRI < 1 group with no statistical significance (Table 3). To explain this trend, we wonder if endothelial dysfunction is playing a more relevant role in hemorrhagic stroke while GI bleeding might be related to rheological changes. Further studies are needed to understand the biologic mechanisms of various bleeding sites.
There are many limitations of our study including small sample size and single center experience. The VENDYS® test has also never been validated in the context of reduced pulsatility (nor have other measures of endothelial vasodilator function). We also have no VRI measurements at time points before CF-LVAD implant to look at the serial changes. Variability in time on CF-LVAD support is another limitation. Thus, it remains to be answered if the reduction of the reactivity is a mere function of reduced cyclic shear stress or is a true reflection of endothelial function. Yet, our findings successfully established a relationship between reduced VRI by this test and bleeding complications and hence possibly reflect actual changes in endothelial biology.
All the patients in our study had an axial flow Heartmate-II LVAD. While the anatomy of centrifugal pumps is different compared to Heartmate-II, no physiologic differences in end-organ function have been shown. Hence, our findings could also be relevant to centrifugal pumps. While the latest generation of LVADs simulate a pulse and has shown a promise of lower pump thrombosis and lower bleeding events when compared with Heartmate-II, in the MOMENTUM trial, the prevalence was still clinically significant even in the HeartMate-II arm. Hence, our findings might still be relevant in the latest generation pumps and a recurrent change in pulse volume in Heartmate-III might not be entirely sufficient to improve endothelial function. There are no reports of comparison of endothelial function between these 2 pumps.
CONCLUSION
Our study reveals that endothelial vasodilatory function is abnormal in many patients with axial flow continuous flow mechanical devices, and this abnormality is associated with bleeding complications, primarily neurologic. With the availability of simple office-based tests of vasodilatory function, more extensive studies are needed to assess the clinical utility of such tests in the current generation CF-LVAD patients.
ACKNOWLEDGMENTS
Schematic illustration of VENDYS®-I device obtained with permission from Endothelix Inc., Palo Alto, California.
REFERENCES
1. Cooke JP, Rossitch E Jr, Andon NA, Loscalzo J, Dzau VJ. Flow activates an endothelial potassium channel to release an endogenous nitrovasodilator. J Clin Invest. 1991; 88:1663–1671
2. Ohno M, Gibbons GH, Dzau VJ, Cooke JP. Shear stress elevates endothelial cGMP. Role of a potassium channel and G protein coupling. Circulation. 1993; 88:193–197
3. Cooke JP. Flow, NO, and atherogenesis. Proc Natl Acad Sci USA. 2003; 100:768–770
4. Shrode CW, Draper KV, Huang RJ, et al. Significantly higher rates of gastrointestinal bleeding and thromboembolic events with left ventricular assist devices. Clin Gastroenterol Hepatol. 2014; 12:1461–1467
5. Harvey L, Holley CT, John R. Gastrointestinal bleed after left ventricular assist device implantation: Incidence, management, and prevention. Ann Cardiothorac Surg. 2014; 3:475–479
6. Stulak JM, Lee D, Haft JW, et al. Gastrointestinal bleeding and subsequent risk of thromboembolic events during support with a left ventricular assist device. J Heart Lung Transplant. 2014; 33:60–64
7. Crow S, John R, Boyle A, et al. Gastrointestinal bleeding rates in recipients of nonpulsatile and pulsatile left ventricular assist devices. J Thorac Cardiovasc Surg. 2009; 137:208–215
8. Klovaite J, Gustafsson F, Mortensen SA, Sander K, Nielsen LB. Severely impaired von Willebrand factor-dependent platelet aggregation in patients with a continuous-flow left ventricular assist device (HeartMate II). J Am Coll Cardiol. 2009; 53:2162–2167
9. Patel SR, Vukelic S, Jorde UP. Bleeding in continuous flow left ventricular assist device recipients: An acquired vasculopathy? J Thorac Dis. 2016; 8:E1321–E1327
10. Kang J, Hennessy-Strahs S, Kwiatkowski P, et al. Continuous-flow LVAD support causes a distinct form of intestinal angiodysplasia. Circ Res. 2017; 121:963–969
11. Bartoli CR, Zhang DM, Hennessy-Strahs S, et al. Clinical and
in vitro evidence that left ventricular assist device-induced von Willebrand factor degradation alters angiogenesis. Circ Heart Fail. 2018; 11:e004638
12. Amir O, Radovancevic B, Delgado RM 3rd, et al. Peripheral
vascular reactivity in patients with pulsatile vs axial flow left ventricular assist device support. J Heart Lung Transplant. 2006; 25:391–394
13. Bonetti PO, Pumper GM, Higano ST, Holmes DR Jr, Kuvin JT, Lerman A. Noninvasive identification of patients with early coronary atherosclerosis by assessment of digital reactive hyperemia. J Am Coll Cardiol. 2004; 44:2137–2141
14. Schier R, Schick V, Amsbaugh A, et al. Brachial artery reactivity and vascular reactive hyperemia for preoperative anaesthesia risk assessment - an observational study. BMC Anesthesiol. 2014; 14:47
15. Endothelix. 510(k) Summary for Vendys Model 5000 B/BC Houston, Texas. Available at
https://www.accessdata.fda.gov/cdrh_docs/pdf6/K060980.pdf
16. Ahmadi N, McQuilkin GL, Akhtar MW, et al. Reproducibility and variability of digital thermal monitoring of
vascular reactivity. Clin Physiol Funct Imaging. 2011; 31:422–428
17. Naghavi M, Yen AA, Lin AW, Tanaka H, Kleis S. New indices of endothelial function measured by digital thermal monitoring of
vascular reactivity: Data from 6084 patients registry. Int J Vasc Med. 2016; 2016:1348028
18. Bhimaraj A, Loebe M. Editorial: An interview with Dr. George P. Noon. Methodist Debakey Cardiovasc J. 2015; 11:45–47
19. Kirklin JK, Naftel DC, Pagani FD, et al. Seventh INTERMACS annual report: 15,000 patients and counting. J Heart Lung Transplant. 2015; 34:1495–1504
20. Witman MA, Garten RS, Gifford JR, et al. Further peripheral vascular dysfunction in heart failure patients with a continuous-flow left ventricular assist device: The role of pulsatility. JACC Heart Fail. 2015; 3:703–711
21. Kirklin JK, Naftel DC, Kormos RL, et al. Fifth INTERMACS annual report: Risk factor analysis from more than 6,000 mechanical circulatory support patients. J Heart Lung Transplant. 2013; 32:141–156
22. Hasin T, Matsuzawa Y, Guddeti RR, et al. Attenuation in peripheral endothelial function after continuous flow left ventricular assist device therapy is associated with cardiovascular adverse events. Circ J. 2015; 79:770–777
23. Grosman-Rimon L, McDonald MA, Jacobs I, et al. Markers of inflammation in recipients of continuous-flow left ventricular assist devices. ASAIO J. 2014; 60:657–663
24. Gurvits GE, Fradkov E. Bleeding with the artificial heart: Gastrointestinal hemorrhage in CF-LVAD patients. World J Gastroenterol. 2017; 23:3945–3953
25. Devore AD, Mentz RJ, Patel CB. Medical management of patients with continuous-flow left ventricular assist devices. Curr Treat Options Cardiovasc Med. 2014; 16:283
26. Acharya D, Loyaga-Rendon R, Morgan CJ, et al. INTERMACS analysis of stroke during support with continuous-flow left ventricular assist devices: Risk factors and outcomes. JACC Heart Fail. 2017; 5:703–711
