Continuous-Flow LVAD Support Causes a Distinct Form of Intestinal Angiodysplasia.

RATIONALE: The objective of this autopsy study was to determine whether gastrointestinal angiodysplasia develops during continuous-flow left ventricular assist device (LVAD) support. OBJECTIVE: LVAD support causes pathologic degradation of von Willebrand factor (vWF) and bleeding from gastrointestinal angiodysplasia at an alarming rate. It has been speculated that LVAD support itself may cause angiodysplasia. The relationship to abnormal vWF metabolism is unknown. We tested the hypothesis that abnormal gastrointestinal vascularity develops during continuous-flow LVAD support. METHODS AND RESULTS: Small bowel was obtained from deceased humans, cows, and sheep supported with a continuous-flow LVAD (n=9 LVAD, n=11 control). Transmural sections of jejunum were stained with fluorescein isothiocyanate-conjugated isolectin-B4 for endothelium to demarcate vascular structures and quantify intestinal vascularity. Paired plasma samples were obtained from humans before LVAD implantation and during LVAD support (n=41). vWF multimers and degradation fragments were quantified with agarose and polyacrylamide gel electrophoresis and immunoblotting. Abnormal vascular architecture was observed in the submucosa of the jejunum of human patients, cows, and sheep supported with a continuous-flow LVAD. Intestinal vascularity was significantly higher after LVAD support versus controls (5.2±1.0% versus 2.1±0.4%, P=0.004). LVAD support caused significant degradation of high-molecular-weight vWF multimers (-9±1%, P<0.0001) and accumulation of low-molecular-weight vWF multimers (+40±5%, P<0.0001) and vWF degradation fragments (+53±6%, P<0.0001). CONCLUSIONS: Abnormal intestinal vascular architecture and LVAD-associated vWF degradation were consistent findings in multiple species supported with a continuous-flow LVAD. These are the first direct evidence that LVAD support causes gastrointestinal angiodysplasia. Pathologic vWF metabolism may be a mechanistic link between LVAD support, abnormal angiogenesis, gastrointestinal angiodysplasia, and bleeding.

Patient-specific severity of von Willebrand factor degradation identifies patients with a left ventricular assist device at high risk for bleeding.

BACKGROUND: Continuous-flow left ventricular assist devices (LVADs) cause an acquired von Willebrand factor (VWF) deficiency and bleeding. Models to risk-stratify for bleeding are urgently needed. We developed a model of continuous-flow LVAD bleeding risk from patient-specific severity of VWF degradation. METHODS: In a prospective, longitudinal cohort study, paired blood samples were obtained from patients (n = 67) with a continuous-flow LVAD before and during support. After 640 ± 395 days, patients were categorized as all-cause bleeders, gastrointestinal (GI) bleeders, or nonbleeders. VWF multimers and VWF clotting function were evaluated to determine bleeding risk. RESULTS: Of 67 patients, 34 (51%) experienced bleeding, 26 (39%) experienced GI bleeding, and 33 (49%) did not bleed. In all patients, LVAD support significantly reduced high-molecular-weight VWF multimers (P < .001). Bleeders exhibited greater loss of high-molecular-weight VWF multimers (mean ± standard deviation, -10 ± 5% vs -7 ± 4%, P = .008) and reduced VWF clotting function versus nonbleeders (median [interquartile range], -12% [-31% to 4%] vs 0% [-9 to 26%], P = .01). A combined metric of VWF multimers and VWF function generated the All-Cause Bleeding Risk Score, which stratified bleeders versus nonbleeders (86 ± 56% vs 41 ± 48%, P < .001) with a positive predictive value of 86% (95% confidence interval, 66%-95%) and diagnostic odds ratio of 11 (95% confidence interval, 2.9-44). A separate GI Bleeding Risk Score stratified GI bleeders versus nonbleeders (202 ± 114 vs 120 ± 86, P = .003) with a positive predictive value of 88% (64%-97%) and diagnostic odds ratio of 18 (3.1-140). CONCLUSIONS: The severity of loss of VWF multimers and VWF clotting function generated Bleeding Risk Scores with high predictive value for LVAD-associated bleeding. This model may guide personalized antithrombotic therapy and patient surveillance.

Toward a Standard Practice to Quantify von Willebrand Factor Degradation During Left Ventricular Assist Device Support.

BACKGROUND: Continuous-flow left ventricular assist devices (LVADs) cause degradation of von Willebrand factor (VWF) multimers and bleeding. Multiple techniques exist to characterize VWF deficiency. However, a standard methodology has not been established in LVAD patients. Toward this goal, we evaluated 4 methods to quantify VWF multimers. METHODS: We collected paired blood samples from patients (n = 48) before and after 1 week of LVAD support. After 652 ± 59 days of support, patients were classified as bleeders (≥1 bleeding episode) or nonbleeders. VWF multimers were resolved with electrophoresis and immunoblotting, the gold-standard to evaluate VWF multimers. We evaluated 4 quantification methods. RESULTS: Each method demonstrated significant VWF degradation during LVAD support vs a paired, pre-LVAD sample (method 1, VWF length: 48 of 48 patients, -10% ± 1%, P < .0001; method 2, VWF density: 40 of 48, -34% (interquartile range, -46% to -8%), P < .0001; method 3, pre-LVAD to LVAD ratio: 46 of 48, 17 ± 5: 10 ± 1, P < .0001; method 4, LVAD/pre-LVAD index: 46 of 48, 57% (interquartile range, 50% to 73%), P < .0001). Bleeding occurred in 27 of 48 patients. Method 1 demonstrated significantly fewer VWF multimers in bleeders compared with nonbleeders (-11% ± 1% vs -8% ± 1%; P = .01). Other methods did not demonstrate this potentially important clinical relationship. CONCLUSIONS: A standardized methodology is needed to quantify VWF multimer degradation with mechanical circulatory support devices. Novel method 1 successfully quantified the patient-specific change in VWF multimer length during LVAD support and demonstrated a difference in VWF multimers between bleeders and nonbleeders. Adoption of consensus methodology will assist to standardize patient-specific bleeding risk, inform anticoagulation and antiplatelet therapy, and evaluate LVAD hemocompatibility.

Abnormalities in the Von Willebrand-Angiopoietin Axis Contribute to Dysregulated Angiogenesis and Angiodysplasia in Children With a Glenn Circulation.

Children with a bidirectional superior cavopulmonary (Glenn) circulation develop angiodysplasia and pulmonary arteriovenous malformations (AVMs). The von Willebrand factor (vWF)-angiopoietin axis plays a major role in AVM formation in multiple diseases. We observed derangements in global angiogenic signaling, vWF metabolism, angiopoietins, and in vitro angiogenesis in children with a Glenn circulation versus controls and within Glenn pulmonary versus systemic circulations. These findings support the novel hypothesis that abnormalities in the vWF-angiopoietin axis may dysregulate angiogenesis and contribute to Glenn pulmonary AVMs. The vWF-angiopoietin axis may be a target to correct angiogenic imbalance in Glenn patients, for whom no targeted therapy exists.

A Novel Toroidal-Flow Left Ventricular Assist Device Minimizes Blood Trauma: Implications of Improved Ventricular Assist Device Hemocompatibility.

BACKGROUND: Continuous-flow left ventricular assist devices (LVADs) cause blood trauma that includes von Willebrand factor degradation, platelet activation, and subclinical hemolysis. Blood trauma contributes to bleeding, thrombosis, and stroke, which cause significant morbidity and mortality. The TORVAD (Windmill Cardiovascular Systems, Inc, Austin, TX) is a first-of-its kind, toroidal-flow LVAD designed to minimize blood trauma. We tested the hypothesis that the TORVAD causes less blood trauma than the HeartMate II (Abbott Laboratories, Pleasanton, CA) LVAD. METHODS: Whole human blood was circulated for 6 hours in ex vivo circulatory loops with a HeartMate II (n = 8; 10,000 rpm, 70 ± 6 mm Hg, 4.0 ± 0.1 L/min) or TORVAD (n = 6; 144 rpm, 72 ± 0.0 mm Hg, 4.3 ± 0.0 L/min). von Willebrand factor degradation was quantified with electrophoresis and immunoblotting. Platelet activation was quantified by cluster of differentiation (CD) 41/61 enzyme-linked immunosorbent assay (ELISA). Hemolysis was quantified by plasma free hemoglobin ELISA. RESULTS: The TORVAD caused significantly less degradation of high-molecular-weight von Willebrand factor multimers (-10% ± 1% vs -21% ± 1%, p < 0.0001), accumulation of low-molecular-weight von Willebrand factor multimers (22% ± 2% vs 45% ± 2%, p < 0.0001), and accumulation of von Willebrand factor degradation fragments (7% ± 1% vs 25% ± 6%, p < 0.05) than the HeartMate II. The TORVAD did not activate platelets, whereas the HeartMate II caused significant platelet activation (CD 41/61: 645 ± 20 ng/mL vs 1,581 ± 150 ng/mL, p < 0.001; normal human CD 41/61, 593 ng/mL; range, 400 to 800 ng/mL). Similarly, the TORVAD caused minimal hemolysis, whereas the HeartMate II caused significant hemolysis (plasma free hemoglobin: 11 ± 2 vs 109 ± 10 mg/dL, p < 0.0001; normal human plasma free hemoglobin <4 mg/dL). CONCLUSIONS: The TORVAD design, with markedly lower shear stress and pulsatile flow, caused significantly less blood trauma than the HeartMate II. LVADs with reduced blood trauma are likely to improve clinical outcomes and expand LVAD therapy into patients with less advanced heart failure.

Clinical and In Vitro Evidence That Left Ventricular Assist Device-Induced von Willebrand Factor Degradation Alters Angiogenesis.

Background Gastrointestinal bleeding from angiodysplasia is a major problem in continuous-flow left ventricular assist device (LVAD) patients. LVAD shear stress causes pathologic degradation of VWF (von Willebrand factor). A mechanistic relationship between VWF degradation and angiodysplasia has not been explored. We tested 2 novel hypotheses: (1) clinical hypothesis: VWF fragments are elevated in LVAD patients that develop angiodysplasia and (2) in vitro hypothesis: VWF fragments generated during LVAD support alter angiogenesis, which may contribute to angiodysplasia. Methods and Results Clinical study: Paired blood samples were collected from continuous-flow LVAD patients (n=35). VWF was quantified with immunoblotting. In vitro experiments: (1) To investigate whether LVAD support alters angiogenesis, human endothelial cells were cultured with LVAD patient plasma (n=11). To investigate mechanism, endothelial cells were cultured with VWF fragments produced by exposing human VWF and ADAMTS-13 (VWF protease) to LVAD-like shear stress (175 dyne/cm2, n=8). Clinical study results: in all patients (n=35, mean support 666±430 days), LVAD support degraded high-molecular-weight VWF multimers ( P<0.0001) into low-molecular-weight VWF multimers ( P<0.0001) and VWF fragments ( P<0.0001). In patients with gastrointestinal bleeding from angiodysplasia (n=7), VWF fragments were elevated ( P=0.02) versus nonbleeders. In contrast, in patients with gastrointestinal bleeding without angiodysplasia, VWF fragments were not elevated versus nonbleeders ( P=0.96). In vitro experiments results: LVAD patient plasma caused abnormal angiogenesis with reduced tubule length ( P=0.04) and migration ( P=0.05). Similarly, endothelial cells grown with VWF degradation fragments exhibited reduced tubule length ( P<0.001) and migration ( P=0.01). Conclusions LVAD patients who bled from angiodysplasia had higher levels of VWF fragments than nonbleeders and gastrointestinal bleeders without angiodysplasia. VWF fragments caused abnormal angiogenesis in vitro. These findings suggest that VWF fragments may be a mechanistic link between LVAD support, abnormal angiogenesis, angiodysplasia, and gastrointestinal bleeding.

Clinical and In Vitro Evidence That Subclinical Hemolysis Contributes to LVAD Thrombosis.

BACKGROUND: Recent data suggest that hemolysis contributes to left ventricular assist device (LVAD) thrombosis, but the mechanism is unknown. In a clinical study, we measured plasma free hemoglobin (pfHgb) and the incidence of LVAD thrombosis. In an in vitro study, we examined biophysical relationships between shear stress, pfHgb and von Willebrand factor (vWF) metabolism toward understanding mechanisms of LVAD thrombosis. METHODS: In the clinical study, blood samples were obtained from continuous-flow LVAD patients (n = 30). Plasma free hemoglobin was measured via enzyme-linked immunosorbent assay. Plasma lactate dehydrogenase (LDH) was measured with a fluorimetric assay. In the in vitro study, to investigate mechanism, human plasma (n = 10) was exposed to LVAD-like shear stress (175 dyne/cm2) with and without free hemoglobin (30 mg/dL). ADAMTS-13 (the vWF protease) activity was quantified with Förster resonance energy transfer. vWF size was quantified with immunoblotting. vWF clotting function was quantified with an enzyme-linked immunosorbent assay. RESULTS: In the clinical study, LVAD support caused subclinical hemolysis. In all patients, LDH increased significantly from 213 ± 9 U/L to 366 ± 31 U/L at 10 days of support (p < 0.0001) and remained significantly elevated at 280 ± 18 U/L at 1 month of support (p < 0.01). In 21 patients that did not develop LVAD thrombosis, pfHgb increased early but decreased over time (pre-LVAD: 5.2 ± 0.8 mg/dL; 1 week: 19.8 ± 4.4 mg/dL, p < 0.01; 3 months: 9.3 ± 2.2 mg/dL, p = 0.07). In 9 patients that developed LVAD thrombosis, pfHgb was significantly elevated versus patients without thrombosis before (p < 0.001) and after 3 months (p < 0.05) of support (pre-LVAD: 20.2 ± 6.3 mg/dL; 1 week: 17.3 ± 3.7 mg/dL; 3 months: 21.5 ± 7.8 mg/dL). Similarly, after 3 months, patients that did not develop LVAD thrombosis had an LDH of 271 ± 28 U/L, whereas patients that later developed LVAD thrombosis had a significantly higher LDH of 625 ± 210 U/L (p = 0.02). In the in vitro study, shear stress degraded vWF similarly to an LVAD. Free hemoglobin inhibited ADAMTS-13 activity during shear stress (633 ± 27 ng/mL to 565 ± 24 ng/mL; p < 0.001). vWF was thereby protected from degradation, 4 vWF fragments decreased significantly (p ≤ 0.05), and vWF clotting function increased (1.15 ± 0.09 U/mL to 1.29 ± 0.09 U/mL, p = 0.06). CONCLUSIONS: These are the first data to demonstrate mechanistic relationships between subclinical hemolysis and a procoagulant state during continuous-flow LVAD support. Patients with high pfHgb and LDH were more likely to develop LVAD thrombosis. In vitro experiments demonstrated that free hemoglobin inhibited ADAMTS-13, protected vWF from degradation, increased vWF clotting function, and created a procoagulant state. As such, pfHgb may be a clinical target to prevent LVAD thrombosis.

Fetal hypoxemia causes abnormal myocardial development in a preterm ex utero fetal ovine model.

In utero hypoxia is a major cause of neonatal morbidity and mortality and predisposes to adult cardiovascular disease. No therapies exist to correct fetal hypoxia. In a new ex utero fetal support system, we tested the hypothesis that hypoxemic support of the fetus impairs myocardial development, whereas normoxic support allows normal myocardial development. Preterm fetal lambs were connected via umbilical vessels to a low-resistance oxygenator and placed in a sterile-fluid environment. Control normoxic fetuses received normal fetal oxygenation, and hypoxemic fetuses received subphysiologic oxygenation. Fetuses with normal in utero development served as normal controls. Hypoxemic fetuses exhibited decreased maximum cardiac output in both ventricles, diastolic function, myocyte and myocyte nuclear size, and increased myocardial capillary density versus control normoxic fetuses. There were no differences between control normoxic fetuses in the fetal support system and normal in utero controls. Chronic fetal hypoxemia resulted in significant abnormalities in myocyte architecture and myocardial capillary density as well as systolic and diastolic cardiac function, whereas control fetuses showed no differences. This ex utero fetal support system has potential to become a significant research tool and novel therapy to correct fetal hypoxia.