Enhanced Hemocompatibility via Bivalirudin and Bicarbonate as Alternative to Heparin in a Catheter‐Based Axial‐Flow System

ABSTRACT

Background

Cardiovascular therapeutic devices typically require systemic heparin due to underlying thrombotic risk. The Impella axial flow system further relies on internal perfusion with either a heparin‐containing or sodium bicarbonate purge solution during operation. The combination of systemic and device‐mediated heparin often contributes to varying levels of overall anticoagulation, raising the possibility of increased bleeding risk and complications due to heparin sensitivity. To reduce adverse events, efforts have been ongoing to eliminate heparin, both systemically and via the device. Here, we investigate bivalirudin as an alternative to systemic heparin and sodium bicarbonate as an alternative to local device heparin in the purge solution.

Methods

We examined hemocompatibility in a mock loop environment with circulating whole porcine blood to determine feasibility of heparin‐free Impella system. Anticoagulated (heparin or bivalirudin) blood was loaded into a circulatory loop with a contained Impella 5.5 and run for 4 h with either heparin or sodium bicarbonate‐containing purge solution. Blood samples were collected serially and analyzed for metrics of hemocompatibility: hemolysis, platelet activation, and vWF degradation.

Results

Bivalirudin and sodium bicarbonate adjunctive environment led to similar or improved hemocompatibility as compared to heparin‐containing environment—notably, decreased hemolysis (40 ± 20.8 mg/dL average decrease) and preserved platelet activity measured by P‐selectin exposure (8.0 ± 5.6% average increase).

Conclusions

Our data supports the potential to create an overall heparin‐free approach simplifying anticoagulation management, offering potential for overall reduced bleeding risk and heparin‐related complications.

1. Introduction

Treatment of advanced cardiovascular diseases, such as acute cardiogenic shock or decompensated heart failure, increasingly utilizes cardiovascular therapeutic devices (CTDs), for example, intravascular blood pumps [1, 2, 3]. While effective in restoring hemodynamic function, CTDs remain limited as foreign bodies with intrinsic thrombogenicity [4, 5, 6]. Presently, a wide range of antithrombotic agents are utilized with CTDs to mitigate device‐related thrombosis [7, 8]. Despite agents, significant limitations and adverse events remain, including device‐associated thrombosis, thromboembolic‐related complications, and bleeding [9, 10, 11].

A CTD in widespread use today for hemodynamic support is the catheter‐based micro‐axial blood pump Impella (Abiomed Inc.). Impella pumps are percutaneous, transvalvular mechanical circulatory support devices which propel blood via high‐speed rotating impeller, supporting blood pressure and organ perfusion. The Impella system relies on internal device perfusion with a purge solution to lubricate high‐speed rotating elements, prevent blood ingress, limit purge gap protein denaturation and deposition, and prevent purge gap thrombosis [12, 13]. Standard purge fluid contains heparin (25–50 U/mL) to inhibit clot formation locally in purge gaps. While heparin has demonstrated efficacy in reducing purge gap thrombosis and maintaining purge gap patency, the net intravascular delivery of heparin from purge fluid infusion may differ for a given patient due to device variation in purge gaps [14, 15]. Concurrent with device use, systemic anticoagulation is often required in treatment of a patients underlying cardiac condition, for example, acute coronary syndrome, or as an adjunct to concomitant therapy, for example, percutaneous coronary intervention [16]. This combination of systemic heparin administration combined with variable purge fluid heparin delivery further complicates anticoagulation management, potentially increasing bleeding risk [17]. In recent years, bivalirudin has emerged as an effective alternative to heparin in preventing clot formation [18]. Many clinical trials have shown bivalirudin to be non‐inferior, if not superior, to heparin in terms of ischemic or bleeding outcomes [19, 20, 21, 22]. Bivalirudin for use as a systemic anticoagulant during Impella system operation is therefore a promising solution to replace heparin, however its influence on blood cell function during pump operation remains unexplored. As a replacement to heparin in the purge solution, our prior work and that of others has demonstrated that sodium bicarbonate added to dextrose‐based purge solution exhibits effective anti‐fouling and anti‐thrombotic properties for successful Impella operation [23, 24]. Recently, clinicians have shown that the rate of bleeding and supratherapeutic anticoagulation in patients is lower with sodium bicarbonate added to the purge solution and is effective at ensuring purge patency and limiting device thrombosis, without evidence of systemic consequence [25, 26]. As such, this combination of anti‐thrombotic and anti‐deposition/anti‐fouling properties has made sodium bicarbonate a promising alternative for purge fluid heparin, leading to its approval by the FDA in 2022 for use in Impella patients intolerant to heparin or in whom heparin is contraindicated.

Despite the promising properties and clinical observations related to bivalirudin and sodium bicarbonate separately, limited data exists on their use together in the Impella system. In the present study, we examined the hemocompatibility of a total heparin‐free Impella 5.5 circulation in vitro—one employing bivalirudin for blood anticoagulation, with concomitant sodium bicarbonate for Impella purge. We hypothesized that systemic bivalirudin combined with local sodium bicarbonate purge solutions will have similar or improved blood hemocompatibility metrics as compared to a heparin‐containing environment in the Impella system. We examined alterations over time in three hemocompatibility parameters taken from an Impella 5.5 mock loop system with circulating porcine whole blood. First, we quantified plasma‐free hemoglobin and particulates (2−4 µm) as a metric of hemolysis and circulating cell and cell particulate dynamics in the system. Second, we examined platelet function and microparticle generation via flow cytometric analysis of P‐selectin and Annexin V on platelet or microparticle surfaces. Third, we measured von Willebrand factor (vWF) function and degradation via immunofluorescence spectrometry and collagen binding assays.

2. Methods

2.1. Porcine Blood Collection

The study protocol was approved by the University of Arizona Institutional Animal Care and Use Committee (IACUC; #2022‐0944). Domestic farm swine (Premier BioSource; Yorkshire/Landrace Hybrid; Male 60−70 kg) were sedated and anesthetized according to protocol. Vascular access was established via catheterization (16 Fr) of the femoral artery. Pigs were injected with anticoagulant (100 IU/kg heparin or 0.75 mg/kg bivalirudin) which was allowed to circulate for 10 min. ACT and APTT were measured (Hemochron Signature; Werfen) from blood samples before and after drug administration to ensure proper anticoagulation (Figure S1; Target ACT 150−190 s following heparin or bivalirudin administration, before citrate addition). Fresh porcine blood (2 L) was collected into CPD collection bottles (10% v/v final concentration).

2.2. Circulatory Loop Design

The circulatory loops were assembled as depicted in Figure 1. Loops were constructed with Tygon tubing (ND 100‐65 medical grade); the Impella 5.5 pump was placed in 1” ID tubing to simulate ascending aorta geometry. The Impella 5.5 catheter was accommodated via a modified Y‐connector to allow seamless connection to the Automated Impella Controller (AIC). Two clamps were placed near the inflow and outflow of Impella 5.5 to establish 60 mmHg differential pressure. A total of eight Impella 5.5 pumps were used for experiments; loops were newly assembled before each experiment. The pumps were cleaned between experiments, with multiple washes of Tergazyme solution followed by DI water and then air dried.

Figure 1.

Circulatory loop design with Impella 5.5. Porcine whole blood was loaded into circulatory loop and circulated with Impella 5.5 for 4 h. Blood temperature was maintained at 37°C in a water bath and pressure across pump maintained at 60 mmHg via clamps and pressure sensors at syringe ports. [Color figure can be viewed at wileyonlinelibrary.com]

2.3. In Vitro Experiment Design

Immediately before loading with blood, circulatory loops were run with saline solution to appropriately establish pressure differential (60 mmHg) and assess pump flow rate (4.2−5.0 L/min). If Impella 5.5 pumps were not acceptable, they were exchanged with new Impella 5.5 pumps. Pressure was recorded using sensors at the two syringe ports in the loop (Omega Engineering Inc.) and clamps were adjusted to establish appropriate pressure across the Impella 5.5. Pressure measurement and adjustment (if needed) was performed hourly during the experiment. The Impella 5.5 pump was purged with either heparin solution (25 IU/mL in D5W) or sodium bicarbonate solution (25 mEq/L in D5W). Fresh whole blood (800 mL each) was gently loaded into each loop and air bubbles removed. The ½” ID tubing was submerged in a temperature‐controlled bath maintained at 37°C. Blood was aliquoted into 50mL tubes and incubated in the same bath as “Rest” controls. Blood was circulated in the loops for 4 h (P‐9 Impella setting) with samples collected at 5 min and every 30 min from both circulatory loops; resting samples were collected every 120 min. Samples were immediately processed via centrifugation, according to assay‐specific protocol and re‐calcified with 2 mM CaCl₂ immediately before assay procedure.

2.4. Hemolysis and Particle Testing

Hemolysis of blood samples was determined from plasma‐free hemoglobin levels. Blood was centrifuged at 150×g for 15 min and the top fraction of platelet‐rich plasma (PRP) was collected. Hemoglobin was measured from PRP samples using a Plasma/Low Hb photometer (Hemocue AB). Samples were also quantified for particle count in the range of platelets (2–4 µm) to determine change in platelet count or other circulating particles using a Z1 Particle Counter (Beckman Coulter Inc.).

2.5. Flow Cytometry Analysis

Platelet samples (PRP) were fixed with 2% paraformaldehyde (PFA in phosphate buffered solution (PBS)) at a 1:1 ratio for 30 min at room temperature. PFA‐fixed PRP was diluted (1:5) in a 1% bovine serum albumin (BSA in PBS). Samples were stained with fluorescein‐conjugated anti‐CD62P (PE; 1:100 clone Psel.K02.5) for quantifying P‐selectin exposure or annexin V (FITC; 1:20) for quantifying phosphatidylserine exposure on the platelet surface. Samples were incubated at room temperature, protected from light for 1 h before further diluting (1:10) in PBS for flow cytometry on FACScantoII (BD Biosciences). Platelets were distinguished from other particles (microparticles) according to their forward/side scatter profile and gated to collect 10,000 events within this profile. This gate was kept consistent across all experiments. P‐selectin‐positive or annexin V‐bound events (platelets or microparticles) were quantified according to median fluorescent intensity threshold within their respective fluorescent emission wavelengths (PE or FITC) as compared to unstained controls. Percent positive events were quantified as the number of fluorescent particles in relation to a total number of events in the platelet or microparticle gate. Flow cytometry data were reviewed and analyzed using FCS Express 7 (De Novo Software).

2.6. β‐Thromboglobulin (β‐TG) and vWF Analysis

PRP samples were further centrifuged at 1500×g for 15 min to obtain platelet‐poor plasma (PPP). PPP was aliquoted and frozen at −80°C until batch processed for β‐TG or vWF assays. β‐TG concentration from plasma was analyzed using porcine enzyme‐linked immunosorbent assay (ELISA) kit (MyBioSource). ELISA assays were performed to manufacturer specifications via optical density measurement (450 nm) on VersaMax microplate reader (Molecular Devices Corp).

vWF analysis was performed at the Cornell University Animal Health Diagnostic Center. vWF assays were quantified as a % of vWF human standard. vWF antigen concentration (vWF:Ag) was quantified using an immunoturbidimetric assay configured with polyclonal anti‐human vWF antibody. vWF function was assessed using the collagen binding assay (vWF:CBA), an ELISA configured with collagen bound to a microplate and polyclonal anti‐human vWF antibody. The ratio vWF:CBA to vWF:Ag was calculated to assess vWF functional activity, standardized to concentration.

2.7. Statistical Analysis

Blood was collected from N = 10 pigs and split between multiple circulatory loops. Data were pooled according to circulatory loop purge solution (heparin 25 IU/mL or sodium bicarbonate 25 mEq/L) and systemic anticoagulant injected into the pig (heparin 100IU/kg or bivalirudin 0.75 mg/kg). Data were collected from a minimum of 4 circulatory loops for each experimental condition. All assays from blood samples were run in at least duplicate and data averaged. To account for inter‐donor or inter‐experiment variability, data were offset by parameter value recorded at the initial time point (5 min) and expressed as an average change (Δ) in parameter over time. Data were assessed for statistical significance using a two‐tailed unpaired t‐test with a significance level of α = 0.05 where p < 0.05 is considered statistically significant. All analysis was performed using GraphPad Prism 10 (GraphPad Software Inc.).

3. Results

To investigate the hemocompatibility of the Impella 5.5 in a heparin‐free environment, we circulated porcine whole blood with varying overall or “systemic” anticoagulation (0.75 mg/kg bivalirudin vs. 100 IU/kg heparin [control]) in an in vitro circulatory loop system with varying purge solution (25 mEq/L sodium bicarbonate vs. 25 IU/mL heparin [control]) (Figure 1). We measured change in plasma hemoglobin as a marker of hemolysis and circulating particulates in the blood over time and simultaneously quantified specific markers of platelet and vWF function as an indicator for potential thrombus formation.

3.1. Hemolysis and Circulating Particle Count

Plasma‐free hemoglobin increased over time in all experimental conditions (Figure 2A). Experiments with systemically administered heparin (“Heparin−Heparin” and “Heparin−Sodium Bicarb” showed the greatest progressive increase in hemoglobin concentration following 4 h, corresponding to an average 70.0 ± 20.1 mg/dL (“Heparin−Heparin”; p = 0.006) increase from initial hemoglobin levels with heparin purge and a 57.5 ± 13.3 mg/dL (“Heparin−Sodium Bicarb”; p = 0.008) increase with sodium bicarbonate purge. In contrast, systemic bivalirudin and sodium bicarbonate purge solution (“Bivalirudin−Sodium Bicarb”) resulted in only a 30.0 ± 5.2 mg/dL (p < 0.001) increase, with an increase of 22.1 ± 4.3 mg/dL (p < 0.001) seen in the resting sample. There was no significant difference in hemoglobin change throughout circulation between resting samples and Bivalirudin−Sodium Bicarb (p = 0.26 at 4 h) nor between Bivalirudin−Sodium Bicarb samples compared to Heparin−Heparin (p = 0.08 at 4 h). In contrast, we observed a significant increase in hemoglobin with Heparin−Heparin (p = 0.03) and Heparin−Sodium Bicarb (p = 0.01) as compared to resting samples after 4 h circulation. The significant divergence in hemoglobin level from resting samples began after 210 min with “Heparin−Sodium Bicarb,” but only after 120 min of circulation with Heparin−Heparin indicating a potential contributing factor of heparin to hemolysis in the system.

Figure 2.

Plasma free hemoglobin and particle count. Porcine whole blood was collected from noncirculating (“Rest”) samples and from samples circulated with Impella5.5 using sodium bicarbonate (25 mEq/L in D5W) or heparin (25 IU/mL in D5W) purge solution. Pigs were anticoagulated with either heparin (100 IU/kg; “Heparin−Sodium Bicarb”) or bivalirudin (0.75 mg/kg; “Bivalirudin−Sodium Bicarb”) before blood collection and circulation. Blood samples were collected and analyzed for (A) plasma‐free hemoglobin and processed for (B) platelet‐sized (2–4 µm) particle count. All data are offset from t = 5 min measurements to better compare between experiment conditions at each time point. Values are reported as mean ± standard deviation. Data were collected and pooled from N = 10 pigs; *p < 0.05 between “Rest” and “Heparin−Heparin”; #p < 0.05 between “Rest” and “Heparin−Sodium Bicarb.” [Color figure can be viewed at wileyonlinelibrary.com]

To identify alterations in the concentration of circulating particles, either through a reduction in platelet count via platelet adhesion or an increase in red blood cell particulate via hemolysis, we further quantified particle count (2–4 µm) from PRP samples. Despite an increase in hemolysis, no significant change in particle count was found under any experimental condition tested (Figure 2B). On average, an increase in particles was observed for all experiments over time with the exception of Bivalirudin−Sodium Bicarb (average decrease 26,288 ± 19,953 platelets/µL at 4 h), however, none of these increases were significant due to large deviations between pig donors. This finding indicates that hemolysis may be a contributing factor but not the sole factor to circulating particles (within the 2–4 µm range) as it correlates to our previous finding that Bivalirudin−Sodium Bicarb experiments corresponded to lowest observed hemolysis, but not an increase in overall particle count as seen in other experimental conditions. Factors such as platelet adhesion or destruction, or red blood cell destruction outside the 2–4 µm range with accompanying hemoglobin release could influence these results.

3.2. Platelet Function and Activity

We further examined platelet surface receptors and plasma β‐TG as, an indicator of platelet function throughout the circulation time. We found an average decrease over time in % positive P‐selectin particles for all conditions tested (Figure 3A). This decrease was significant after 4 h with Heparin−Heparin (10.2 ± 4.3% decrease; p = 0.04) and Heparin−Sodium Bicarb conditions (6.0 ± 1.7% decrease; p = 0.03). This is compared to the nonsignificant average decrease seen with Bivalirudin−Sodium Bicarb, corresponding to an average change in positive P‐selectin particles of 2.1 ± 3.7% (p = 0.6) after 4 h circulation time. Despite the difference on average decrease between experimental conditions, these results were variable between runs and therefore there was no statistical difference between the conditions at any time point.

Figure 3.

Platelet surface P‐selectin and phosphatidylserine exposure and plasma β‐thromboglobulin. Porcine blood samples were collected and processed into fixed platelet rich plasma for flow cytometry with fluorescent labeled (A) P‐selectin (CD62P) and (B) phosphatidylserine exposure via Annexin V binding. Platelet poor plasma samples were simultaneously collected and measured for (C) free β‐thromboglobulin using ELISA. All data are offset from t = 5 min measurements to better compare between experiment conditions at each time point. Values are reported as mean ± standard error. Data were collected and pooled from N = 10 pigs. [Color figure can be viewed at wileyonlinelibrary.com]

We similarly quantified phosphatidylserine exposure on platelets, indicated by Annexin V binding, to investigate platelet thrombotic potential (Figure 3B). In contrast to P‐selectin exposure, we did not observe a uniform increase or decrease in phosphatidylserine exposure across experimental conditions tested. Heparin−Heparin experiments led to an average 5.2 ± 4.0% decrease (p = 0.2) in Annexin V‐positive platelets over 4 h, whereas Bivalirudin−Sodium Bicarb experiments resulted in an average 12.6 ± 7.7% increase (p = 0.1) for the same time period; however, these changes were not significantly different. Similar to P‐selectin analysis, the variation observed between runs led to no statistically significant results in measured Annexin V binding between experimental conditions (Figure 3B).

As an indicator of α‐granule release from platelets, β‐TG levels in the plasma were quantified (Figure 3C). We observed no significant difference in β‐TG concentration for all conditions tested after 4 h of circulation. Further, there was no significant difference in β‐TG levels between experimental conditions at any time point examined. Measurable changes of β‐TG concentration were not detected until 4 h of circulation, corresponding to a small and highly variable increase in β‐TG under Rest (83.9 ± 60.4 pg/mL; p = 0.2) and Heparin−Heparin (83.1 ± 70.9 pg/mL; p = 0.4) conditions.

3.3. Platelet‐Derived Microparticle Detection and Function

To elucidate generation of platelet microparticles in the circulating blood, we distinguished between platelets and their microparticles (PDMPs) from forward/side scatter profiles during flow cytometry analysis and quantified them as a percentage of total particle (platelets + PDMPs) population. We found a slight increase in %PDMP over circulation time for all conditions tested (Figure 4A). Notably, there was a statistically significant increase in PDMPs after 4 h circulation conditions in Heparin−Heparin (4.3 ± 1.4% increase; p = 0.01) and Bivalirudin−Sodium Bicarb (3.2 ± 0.78% increase; p = 0.002). While other experimental conditions also led to increases in PDMP generation over time, these alterations were not statistically significant and ranged between 2.0 ± 1.2% and 1.1 ± 0.5% increase observed after 4 h of Bivalirudin−Sodium Bicarb circulation (p = 0.14) and Rest (p = 0.06), respectively. Experimental conditions Heparin−Heparin (p = 0.02), Heparin−Sodium Bicarb (p = 0.04), and Bival−Sodium Bicarb (p = 0.04) were statistically higher than the Rest control sample, but no statistical significance was found between the circulating experiments.

Figure 4.

Platelet‐derived microparticle (PDMP) quantification and surface P‐selectin and phosphatidylserine exposure. Porcine blood samples were collected and processed into fixed platelet‐rich plasma for flow cytometry. Forward and back scatter gates were placed to distinguish platelets from microparticles as a percentage of total particles counted (A). Particles and PDMPs were stained with fluorescent‐labeled (B) P‐selectin (CD62P) and (B) phosphatidylserine exposure via Annexin V binding. All data were offset from t = 5 min measurements to better compare between experiment conditions at each time point. Values are reported as mean ± standard deviation. Data were collected and pooled from N = 10 pigs; *p < 0.05 between “Rest” and “Heparin−Heparin.” [Color figure can be viewed at wileyonlinelibrary.com]

Similar to our flow cytometric analysis of platelets, we repeated analysis of P‐selectin and phosphatidylserine exposure of the PDMP fraction. Interestingly, the decrease of %P‐selectin positive platelets previously observed was not as definitively observed in the PDMP population (Figure 4B). Following 4 h of circulation, the largest decrease in P‐selectin‐positive PDMPs was observed with Heparin−Heparin conditions (5.2 ± 3.8% decrease), however, this was not statistically significant from non‐circulated Rest samples (p = 0.41). In contrast, Bivalirudin−Sodium Bicarb conditions led to an average 1.8 ± 3.3% increase in P‐selectin‐positive PDMPs. Large variation in PDMP markers between pig blood donors led to no statistical significance observed in P‐selectin‐positive PDMP population. Phosphatidylserine exposure, as indicated by Annexin V binding, was more defined on PDMPs (Figure 4C). Resting alone led to an average 8.5 ± 4.8% increase in Annexin V‐positive PDMPs. This is in contrast to the 7.0 ± 3.3% average decrease in Annexin V‐positive PDMPs following 4 h circulation in Heparin−Heparin conditions (p = 0.03). This statistical difference was observed as early as 2 h of circulation (p = 0.01). Other Sodium Bicarb conditions did not exhibit significant changes in Annexin V‐positive PDMP population; Bivalirudin−Sodium Bicarb conditions led to an average 5.4 ± 3.7% increase in Annexin V‐positive PDMPs.

3.4. vWF Function and Activity

We found an overall decrease in vWF functionality via collagen‐binding assay (CBA) over circulation time in all pump conditions tested (Figure 5A). In contrast, our control Rest sample showed a significant increase in vWF activity corresponding to an average 7.1 ± 1.8% increase over 4 h (p = 0.001). Experimental conditions Heparin−Heparin (p = 0.03), Heparin−Sodium Bicarb (p < 0.001), and Bivalirudin−Sodium Bicarb (p < 0.001) were significantly higher than Rest control sample, corresponding to average reduction in collagen binding of 17.7 ± 6.8%, 14.6 ± 2.9%, and 20.4 ± 4.3%, respectively. There was no statistically significant difference in collagen binding activity found between experimental conditions at any time point. The amount of vWF antigen (Ag) was simultaneously measured from a porcine plasma sample and CBA values correspondingly adjusted (Figure 5B). Adjusting for vWF concentration in each pig, the differences between circulating conditions and Rest were less pronounced, however they followed the same general trend as observed in the crude CBA results with overall reduction in vWF functional activity.

Figure 5.

Von Willebrand factor collagen binding activity. Porcine blood samples were collected and processed into platelet poor plasma. Plasma was measured in modified ELISA to quantify (A) collagen binding activity (CBA). Samples were further analyzed via immunoturbidometric assay to quantify vWF antigen (Ag). Collagen binding activity was quantitatively adjusted by vWF antigen concentration and reported as (B) vWF:AG/CBA Ratio. All data are offset from t = 5 min measurements to better compare between experiment conditions at each time point. Values are reported as mean ± standard deviation. Data were collected and pooled from N = 10 pigs; *p < 0.05 between “Rest” and “Heparin−Heparin”; #p < 0.05 between “Rest” and “Heparin−Sodium Bicarb; °p < 0.05 between “Rest” and “Bivalirudin−Sodium Bicarb.” [Color figure can be viewed at wileyonlinelibrary.com]

4. Discussion

The use of heparin as an anticoagulant while mechanistically effective, adds variability to a given patient's overall hemostatic balance and pharmacologic management during Impella operation increasing the risk of potentially avoidable adverse events. In the present study, we explored the hemocompatibility of a heparin‐free Impella 5.5 circulation environment in vitro, replacing heparin with sodium bicarbonate in purge fluid and utilizing bivalirudin as a “systemic” anticoagulant alternative to heparin. Our findings suggest that a bivalirudin and sodium bicarbonate adjunctive environment is similar or superior to a heparin‐containing environment for most metrics of hemocompatibility tested. Notably, the heparin‐free environment revealed a trend toward decreased hemolysis and preserved platelet function as compared to the heparin‐containing environment.

Administration of systemic bivalirudin to blood before circulation led to less plasma free hemoglobin as compared to systemic heparin, regardless of purge solution composition (heparin or sodium bicarbonate; Figure 2). Destruction of red blood cells (“hemolysis”) or membrane pore formation is a well‐known phenomenon in high shear conditions [27, 28]. Such conditions promote release of hemoglobin into the plasma and further destruction of cells can alter the quantity and size of circulating particulates generated. As indicators of hemolysis, these metrics recorded here suggest that a heparin‐free environment resulted in less hemolysis over the circulation time studied in the present investigation.

The observation of a relative reduction in hemolysis in a heparin‐free anticoagulated environment is intriguing, though the exact mechanism remains to be fully defined. It is well‐known that heparin is a highly negatively charged molecule and interacts with the phospholipid bilayer and proteins of the cell membrane. While not fully understood, this heparin effect has been purported to increase red blood cell interaction, aggregation, and potential for lysis. Heparin has been demonstrated to increase the erythrocyte sedimentation rate and overall blood viscosity [29]. This may lead to an increase in local aggregate formation being more prone to shear‐mediated hemolysis. Heparin has also been shown to bind to red cells which may result in cell stiffening thereby increasing their propensity to shear damage [30, 31]. Heparin may alter red blood cell permeability and porogenicity, thereby enhancing plasma free hemoglobin leakage and overall damage and particulation [32, 33]. Heparin may also bind to fibrinogen, fibrin, and other proteins which have been shown to bind to an integrin‐like receptor on red blood cells leading to further micro‐aggregates prone to shear [34, 35, 36, 37].

Contrary to heparin, bivalirudin is a direct thrombin inhibitor which can act on both circulating and clot‐bound thrombin. Therefore, any clots bound in the circulatory loop system can lead to worsening free hemoglobin release and diversely bivalirudin action on developed clots can decrease detectable hemoglobin in circulation. However, it is unlikely that there was significant clot formation in our circulatory loop due to the use of citrated blood without re‐calcification during pump operation. Bivalirudin is a short 20‐residue peptide with both positive and negative termini, affording it the ability to interact with the red blood cell membrane via both termini. Peptides cannot passively enter red blood cell membranes; however, the cationic moiety of bivalirudin can interact with negatively charged elements of the red blood cell membrane for potential partial incorporation [38]. Furthermore, the high shear environment leads to increased porosity of red blood cells, allowing peptides on the range of thousands of Daltons to be delivered into cell membranes [39, 40, 41]. Therefore, it is plausible to expect bivalirudin delivery into cell membranes exposed to high shear. Protein interactions with the cell membrane play an important role in stabilization and deformability of red blood cells [42]. Anticipated bivalirudin interaction with or inclusion into the red blood cell membrane via the bivalirudin peptide termini can transiently bridge the protein structural framework and prevent further pore formation and subsequent hemoglobin release. Therefore, contrary to heparin, bivalirudin may provide a “stabilizing” effect on red blood cells under high shear. In support of this mechanism, clinical reports have suggested that bivalirudin in clinical cases has decreased the need for blood transfusion compared to heparin, which support our findings here [42, 43, 44, 45].

In conjunction with alterations in hemolysis, we found that a heparin‐free circulation environment led to improved preservation of platelet surface P‐selectin exposure (Figure 3). When activated, platelets release a multitude of proteins and inflammatory factors in soluble form or stored in granules. While P‐selectin exposure is a classic indicator of biochemical‐mediated platelet activation and degranulation, our previous work and that of others has shown that shear‐mediated platelet activation typically leads to increased thrombogenicity with a concomitant decrease in platelet surface P‐selectin, as seen in the present study [46, 47]. However, soluble P‐selectin was not detectable (< 0.2 ng/mL; data not shown) for all time points and pump conditions. Simultaneously, only a small nonsignificant increase of soluble β‐TG after 4 h was observed. Therefore, the release of soluble factors via platelet degranulation appears to have been impaired, aligning with our previous reports and other clinical studies. The impairment of granulation can lead to significant risk in device‐related bleeding complications. As such, it is notable that the heparin‐free environment with bivalirudin provided preservation in P‐selectin exposure on the surface of platelets, although these did not correspond to significant changes in measured soluble factors. In addition to directly binding thrombin, both heparin and bivalirudin inhibit thrombin binding to PAR‐mediated receptors for platelet activation and subsequent granule release. A previous study by Lund et al. demonstrated that heparin was a more potent inhibitor of platelet activation via thrombin‐PAR binding, whereas bivalirudin exhibited a biphasic response in which initial inhibition was followed by a significant granule release and overall procoagulant activity [48]. This mechanism of enhanced prothrombotic activity on platelets with bivalirudin may have led to higher platelet function retention observed with reduced impairment of P‐selectin exposure and granule secretion.

The evidence of shear‐mediated platelet activation in the circulatory loop is further evident by the increase in platelet‐derived microparticles over circulation time. As a function of the high shear environment, these platelet‐derived microparticles did not exhibit significant decreased P‐selectin seen in platelets. The high shear environment can promote enhanced dispersion of P‐selectin into microparticles which likely led to no significant change in P‐selectin expression observed on these microparticles over circulation time. vWF quantification and collagen binding activity was similarly quantified over circulation time. As a large multimeric glycoprotein, vWF is susceptible to conformational changes and degradation following exposure to high shear rates such as those employed in our circulatory loop system. As such, it was expected to see a decrease in binding activity over time following circulation in the high shear environment. Our results supported a reduction in vWF binding activity over time with similar or increased activity observed in non‐sheared (“Rest”) blood (Figure 4).

Our study demonstrated that substituting bivalirudin as an alternative to systemic heparin led to decreased hemolysis and preserved platelet function. The alterations in hemocompatibility metrics due to purge solution replacement with sodium bicarbonate were not as significant as those seen with systemic anticoagulation alteration. However, previous investigations have shown the role of sodium bicarbonate as a pH buffer and protein stabilizer is beneficial in the purge gap microenvironment, exerting a local anti‐thrombotic effect in the Impella device. With further investigation Impella circulation in a heparin‐free environment could feasibly be utilized for patients contraindicated for heparin use. This approach offers the potential to create an overall heparin‐free system simplifying anticoagulation management, with overall reduced bleeding risk.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

FIGS. 1.