Gap analysis of assays for evaluating blood-handling devices.
Blood handling devices such as left ventricular assist devices and total artificial hearts offer life-saving treatments for patients suffering from severe heart failure. Current devices have clinically proven that heart assist pumps are a safe and effective therapy, and indeed in many cases they are the only available method of treatment. However, current devices cause side effects including stroke, bleeding, infection, and thrombosis, preventing the technology from reaching its full potential. If the side effects could be reduced, then more patients could benefit from these devices. The complications are related to damage to blood cells and proteins as a result of contact with foreign materials and mechanical stress. There is a need for better devices with minimal blood impact to enable more patients to be safely treated; better tools, especially flow cytometry, could support the device development life cycle. In this chapter we review the clinical, in vivo, and in vitro flow cytometry data available for ventricular assist devices, conduct a gap analysis, and identify areas of future possibilities for device developers to establish new flow cytometry-based methodologies.
- multidimensional flow cytometry
- translational research
- ventricular assist
Blood handling devices such as left ventricular assist devices and total artificial hearts offer life-saving treatments for patients suffering from severe heart failure. Current devices have clinically proven that heart assist pumps are a safe and effective therapy, and indeed in many cases they are the only available method of treatment. However, current devices cause side effects including stroke, bleeding, infection, and thrombosis, preventing the technology from reaching its full potential . If the side effects could be reduced, then more patients could benefit from these devices. The complications are related to damage to blood cells and circulating proteins as a result of contact with foreign materials and mechanical stress. There is a need for better devices with minimal negative effect on blood to enable more patients to be treated safely; better tools, especially flow cytometry, could support the device development life cycle.
The use of multidimensional flow cytometry during pre-clinical development of blood handling devices offers a powerful tool to monitor changes to erythrocytes, leukocytes, and platelets, as well as circulating mediators such as von Willebrand factor. A key challenge is the need to study these in cows, sheep and pigs which are used for pre-clinical studies. This is associated with markedly reduced reagent options compared to studies using human blood. While there are some species-specific antibodies suitable for flow cytometry, the preferential use of cross-reactive reagents and species non-specific tools enables multicolor panels to be developed that can be used with blood from multiple species. Such an approach also allows for comparisons at all stages of device development and implementation: in vitro, in vivo pre-clinical, and ex vivo clinical settings. Flow cytometry methods could also support personalised treatment strategies to potentially predict patients at risk of complications . This could be prior to or following implantation of a device. Here, we will provide an overview of the development of flow cytometry tools to address this need including a review of work performed to date, as well as future possibilities for this technology platform.
2. Current use of flow cytometry for ventricular assist and total artificial heart research
The majority of the flow cytometry research performed in the area of blood-handling medical devices is to further the understanding of the complications that arise upon ventricular assist device implantation into heart failure patients. Since the aim of using flow cytometry during
2.1. Clinical flow cytometry data from ventricular assist devices
Flow cytometry has been used to analyse all major cellular components of the blood—i.e. erythrocytes, leukocytes, and thrombocytes—in VAD-patients.
Sansone et al. were the first group to use flow cytometry in the clinic to evidence VAD-related damage to erythrocytes, in addition to the standard method of measuring plasma free haemoglobin . Patients implanted with continuous flow (CF) VADs (HVAD, HeartWare), showed significantly greater levels of CD235+ erythrocyte MPs compared to controls (both age-matched healthy and patients with stable coronary artery disease, CAD). Their erythrocyte counts were not described, but the VAD-patients had significantly greater levels of free haemoglobin compared to controls. Increased levels of erythrocyte MPs have been found in patients suffering sickle cell disease and β-thalassemia major, which are diseases also characterised by haemolysis .
From a flow cytometry perspective leukocytes have received more attention than erythrocytes. VAD-related leukocyte damage has been demonstrated using flow cytometry in all major leukocyte subsets. Using the pan-leukocyte marker CD45, leukocyte microparticles (CD45+) were shown to be elevated in CF VAD patients compared to healthy and CAD controls . This is indicative of overall leukocyte destruction and is supported by Woolley et al. who observed decreased total leukocyte counts in CF VAD patients . In the same study, CD15+ neutrophils were found to become activated as measured by an increase in MAC-1 (CD11b) expression. The level of activated neutrophils was dependent on pump type: HeartMate II causing greater levels than HVAD and PVAD . Neutrophil activation status might also influences the patient’s susceptibility to infection as more HeartMate II patients than HVAD patients suffered from infection. The PVAD has a larger driveline exit area which contributes to infection rates, hence it cannot be directly compared to the other two pumps .
Monocytes also become activated in VAD-patients with the expression of tissue factor (TF) increased significantly within the first month of pulsatile Novacor or HeartMate XVE support versus healthy controls [10, 11]. TF is a key element of the extrinsic coagulation cascade, and it is able to trigger coagulation, even with endothelial integrity virtually preserved. The major source of TF in blood is monocytes, and the expression is upregulated by for example lipopolysaccharides (LPS) . As summarised by Angelillo-Scherrer: volunteers exposed to endotoxin, patients with meningococcal sepsis, and primates with Ebola fever, all show increased levels of CD14+/TF+ MPs, indicating a potential role for these MPs in disseminated intravascular coagulation associated with severe infections . As driveline infections is a common problem in VAD-patients, there is a possibility that CD14+ monocytes expressing TF and/or CD14+TF+ MPs, could be a thrombosis risk marker in patients with ongoing infection.
Lymphocytes are affected by both pulsatile and CF-VADs [13, 14, 15]. A general lymphopenia occurred in patients, implanted with the early pulsatile HeartMate XVE . This was accompanied by a significant reduction in the mean CD4/CD8 T cell ratios, and in the mean number of circulating CD4 T cells. The CD4/CD8 T cell ratio decreased rapidly within the first month and remained low at the 2 month follow-up assessment . The decline in CD4 T cells was attributed to a heightened susceptibility to apoptosis as measured by surface expression of phosphatidylserine through annexin V binding . These results were confirmed in a group of patients implanted with either the pulsatile HeartMate XVE or the Novacor, where the mean number of circulating CD4 T cells was significantly lower compared to medically managed heart failure patient controls. While the levels of CD8 T cells remained unaffected , both CD4 and CD8 T cells had increased CD95(Fas) expression and annexin V binding versus controls, indicating apoptosis. Furthermore, the LVAD-patients had a significantly greater risk of developing candidal infection compared to the controls or other patients undergoing cardiac surgery. Altogether, this suggests that the pulsatile LVADs cause T-cell defects, most notably CD4 T-cell defects, and that some of these defects are measurable through flow cytometry [14, 15]. How this translates to an effect on the function of T cells remains to be determined.
Patients with CF-VADs also have changes in their T cell levels. However, contrary to pulsatile LVAD-patients, those implanted with CF-VADs (specific device not published) and who suffered from infection had significantly higher levels of CD4+/CD25+ Tregs and increased lymphocyte reactive oxygen species (ROS) compared to VAD-patients without infection . Whether these differences between pulsatile and CF VADs relates to pulsatility would be an interesting topic for further studies.
As far as we are aware, circulating B cells and NK cells have not been studied by flow cytometry in LVAD-patients. Nor have other minor populations such as dendritic cells or innate lymphoid cells. There are some data of the effect of VADs on B cells from in vitro studies (Schuster 2002) and these are discussed below.
Platelets have been studied on their own and in microaggregates with leukocytes. Wilhelm et al. found that platelets were activated in patients with pulsatile VADs (Novacor and HeartMate XVE) compared to healthy controls . This was measured as significantly increased CD62P expression. However, Dewald showed that increased platelet activation might not be due to the VAD as platelets in heart failure patients are already activated prior to implantation. This was shown using antibodies against CD62P, CD63, and antithrombospondin . Similarly, Matsubayashi showed that CD62P and CD63 expression are elevated on platelets in Novacor-patients compared to healthy controls, but preoperative values were already high with no clear increase or decrease during implantation . Further highlighting the impact of heart failure rather than VAD use on platelet activation
Platelet MPs also have been detected using flow cytometry in CF-LVAD patients [3, 9]. In 2010, Diehl P et al. showed that LVAD-patients, the majority of whom were implanted with the CF-LVAD HeartMate II, had significantly increased levels of CD31+/CD61+ platelet MPs compared to healthy controls . Five years later, Sansone et al. showed that CD31+/CD41+ platelet MPs were elevated in patients 3 months post-implantation of a HVAD (a CF-LVAD) in comparison to both age-matched healthy controls and patients suffering from coronary artery disease (CAD) . Hence, although the HeartMate II and the HVAD differ drastically in design, both were associated with platelet microparticle formation.
The focus has been on platelet activation and platelet MPs but these have so far not shown a utility as predictors of adverse events or stratifiers. However, there are other platelet parameters, all measurable by flow cytometry, that have potential as patient risk stratifiers, or even predictors, for bleeding complications in patients with CF-VADs [18, 19, 20]. These include significantly greater levels of reactive oxygen species, mitochondrial damage, surface phosphatidylserine (PS) expression/apoptosis, and significantly decreased expression of α2bβ3 on the platelet surface, in bleeders compared to non-bleeders [18, 19, 20]. The CF-VADs studied included the HeartMate II, Jarvik 2000, and the HVAD and no differences were found between the devices.
2.1.4. Endothelial cells and microparticles of unknown origin
CD62E+ endothelial cell microparticles (EMPs) are increased in VAD-patients compared to healthy controls [3, 9] and CAD patients . EMPs phenotyped as CD31+/CD41−, and CD144+ are also elevated levels in VAD-patients .
Additional to the MPs of specific lineages described above—erythrocytes, leukocyte, platelet and endothelial—PS-expressing MPs of unknown lineage have been suggested as a potential biomarker of adverse events in VAD-patients implanted with a HeartMate II . Patients who developed an adverse event, including ventricular tachycardia storm, non-ST elevation myocardial infarction, arterial thrombosis, gastrointestinal bleeding, and stroke had significantly higher levels of PS+ MPs than patients with no adverse events .
2.2. Pre-clinical in vivo haemocompatibility of blood-handling devices
The flow cytometry data from published pre-clinical
Snyder et al. have published the only
Bovine platelet activation and microaggregates have been studied using various markers in calves implanted with CF-VADs. Baker et al. developed a method using the antibodies BAQ56, BAQ125 and GC5A (platelet antigen, CD equivalent unknown, available from Washington State University) to measure platelet activation, and CAPP2A (anti-CD41/CD61) to measure platelet microaggregates in calves implanted with CF-LVADS (the Sun Medical centrifugal pump or the HeartMate II) . Platelet microaggregates, i.e. platelet positive events (binding anti-bovine CD41/61 antibody) with forward scatter larger than single platelets, increased post-operatively and then showed some decline in calves implanted with Sun Medical whereas they remained elevated in the calves implanted with HeartMate II. Platelet activation increased post-operatively for both implants and remained elevated. BAQ56 provided the strongest signal and GC5 the weakest signal of the three antibodies, but BAQ125 and GC5 exhibited the strongest agreement with one another, and with the circulating microaggregates .
Snyder et al. used a modified version of Baker’s methodology in two studies of calves implanted with the CF-LVADs—HeartMate II  and the EVAHEART . Snyder showed that the microaggregate levels in surgical sham controls remained at pre-operative levels, confirming that the CF-LVAD and not the cardiac surgery cause elevated levels . Microaggregate levels increased post-implantation, and decreased within the first week in all animals implanted with the EVAHEART , and in those animals implanted with the HeartMate II who had an uneventful post-operative course . In calves who suffered adherent thrombi in the outflow region of the pumps, the microaggregate levels either remained elevated or increased before conclusion of the study .
Part of Snyder’s modification was the inclusion of additional markers for platelet activation to address the limitations BAQ56, BAQ125, and GC5 with their unknown target antigens, namely Annexin V binding , anti-CD62P and anti-CD63 . All platelet activation/apoptosis markers tested (BAQ125, GC5, Annexin V, CD62P, CD63) increased immediately after implantation and remained significantly elevated in CF-LVAD animals versus the pre-operative control [24, 25]. Those tested in sham animals (BAQ125, GC5, Annexin-V) decreased around 2 weeks post-operatively . Although CD62P and CD63 successfully identified activated platelets in CF-LVAD calves, the signal was weaker than that of the BAQ125 and GC5 antibodies .
CAPP2A also binds to ovine platelets whereas anti-bovine platelet activation antibodies BAQ125 and GC5 do not. Hence, Johnson et al. used CAPP2A as a platelet lineage marker along with cross-reactive anti-human CD62P antibodies (clones Psel.KO.2.7 and Psel.KO.2.12) to assess platelet activation in sheep implanted with paediatric CF-LVADs [26, 27, 28]. Johnson showed that, similarly to calves, platelet activation increases post-operatively in surgical sham control sheep and returns to pre-operative levels at around 2 weeks. The implanted sheep did not demonstrate a common pattern . However, a finding also consistent with previous work was that platelet activation spiked in animals that suffered complications [26, 27, 28].
2.3. In vitro haemocompatibility of blood-handling devices
Similarly to the
Work from our group has shown that bovine leukocytes shed microparticles, measured as increasing levels of CD45+ MPs (Figure 1), during
The first flow cytometry assessments of platelet activation and microaggregates during
3. Future possibilities for device developers
Although flow cytometry has been used clinically to study phenotype, activation status and MPs of the main circulating cell types in patients implanted with mechanical circulatory support devices, the use of flow cytometry for pre-clinical
|Cell type||Assay||In vitro||In vivo||Clinical|
The VADs in current use are designed to minimise high shears to blood for prolonged durations to avoid haemolysis. However, erythrocytes demonstrate impaired capacity to deform when exposed to shear stress well below the “haemolytic threshold” . Thus, there is a need for assays that can pick up more subtle damage to erythrocytes. We propose that quantifying RBC MPs as a possible means of assessing sub-haemolytic damage should be investigated to see whether or not it can fill this need.
Activated CD15+CD11b+ neutrophils, have been detected in VAD-patients and clinical data shows that the activation status might influence the patient’s susceptibility to infection . With additional evidence that CD11b+ MPs are increased in VAD-patients, and generated during
VAD-related increases in monocytes expressing TF have been demonstrated both clinically [10, 11] and
The lack of pulsatility has been suggested as a factor that could be contributing to the complications related to aortic valve insufficiency, gastrointestinal bleeding, stroke, pump thrombosis, and haemolysis . A link between pulsatility and blood damage has so far only been described for the blood coagulation protein von Willebrand Factor (vWF), which appears to degrade more in patients with pulsatile compared to non-pulsatile VADs . As our group has shown, vWF degradation caused by shear stress in vitro can be assessed using flow cytometry a flow cytometry-based ristocetin assay . However, it appears as if T cells could be another missing link needed to describe the effects of non-pulsatile  and pulsatile [14, 15] flow on the blood. Therefore, we propose that assays for T-cells are established for sheep and cow blood in order to study the effects of pulsatility, in order to gather more scientific evidence that could be used to interpret the clinical data.
Other leukocyte subsets of interest to develop methods for studying would be B-cells. Schuster et al. described that the T-cell apoptosis observed in patients with pulsatile VADs, was induced by a B-cell response to polyurethane, a material commonly used for the membranes in pulsatile devices [38, 39]. Continuous flow devices are typically made from metals and ceramics, and do not contain polyurethane. Hence, pulsatility studies should be made in combination with various biomaterial controls, including for example titanium alloys commonly used in VADs [40, 41].
Cytokines have not been assessed using flow cytometry in a VAD-setting, but recent clinical data on TNFα-levels in VAD-patients might spur a change. High levels of TNF-α were associated with increased risk of bleeding complications in patients with CF-VADs (HeartMate II or HVAD) . Measurement of TNF-α by flow cytometry in whole blood would offer a much more rapid method, a lower sample volume required, and could be combined with other cellular markers of interest into a multidimensional approach .
3.4.1. Overcoming practical hurdles
A reason for hesitating in this approach might be the limited availability of reagents for animal blood. However, progress has been made recently in this field by exploiting the cross-reactivity of some antibody clones. We identified the utility of an antibody clone targeting CD62P, clone Psel.KO.2.5, that could measure shear-induced platelet activation in human, bovine, ovine, and porcine blood . Johnson et al. also made significant contributions in identifying cross-reactive platelet antibodies against bovine and ovine platelets . We have used antibodies cross-reactive against bovine and human targets in developing a multicolour flow cytometry panel for sheep leukocyte MPs released during
Assessment of red blood cell damage through measurement of plasma free haemoglobin is a key assay required by regulatory bodies in the development of new blood-handling devices. However, more sophisticated flow cytometry screening methods that enable evaluation of a wider array of blood cell types and blood borne proteins could be used in combination with other design tools to inform iterative design improvements. Such an approach would enable device developers to progress new designs that have minimal total blood trauma, and ultimately put safer devices on the market. The use of multiparametric flow cytometry in the design cycle and clinical use of medical devices that contact blood is in its infancy. Clearly, there are many advantages to such an approach. The key advantage is the ability to analyse the phenotype of multiple haematopoietic cell populations simultaneously in small volumes of blood. Such methodology would enable the same analysis protocols to be used at all stages of design and implementation: in vitro studies, pre-clinical studies in different animal species, and then in clinical studies once devices are implanted in humans. Multiparametric flow cytometry also enables the incorporation of functional outputs such as cell viability, apoptosis, reactive oxygen species (ROS) production and cytokine production. These are critical if the impact of device-associated biomaterials and shear stress on haematopoietic cells is to be determined; there might not be phenotypic changes but cell function might be affected dramatically. The main challenges to implementing multiparametric flow cytometry in this setting relate to the availability of reagents that are truly multi-species and enable a single methodological approach to be used across the device development life cycle. This is more challenging when it comes to cell phenotyping that typically requires antibodies to specific cell surface markers but easier to implement for functional assays related to cell viability/apoptosis and measures such as ROS. Other challenges relate to flow cytometer availability to the research and development team as that will dictate fluorochrome choice and assay design. Also, data analysis becomes demanding as the number of parameters studied simultaneously increases and machine learning approaches need to be developed alongside development of multiparameter strategies. Rapid advances in fluorochrome technology and the development of mass cytometry facilitate a multiparametric approach but there remain limitations in fluorochrome choices for reagents targeting surface markers and cytokines of cells from animal species such as cows, pigs and sheep. Demand for these reagents will drive their availability but this demand will only occur if the device development community adopt standardised approaches. Underpinning this would be a requirement from regulatory bodies for flow cytometry-based analysis of cell health and function as part of the development and evaluation of blood handling medical devices.
Conflict of interest
Ina Laura Pieper is a past employee of Calon Cardio-Technology Ltd., and a current employee and shareholder of Scandinavian Real Heart AB. Gemma Radley is an employee of Calon Cardio-Technology Ltd. Catherine A Thornton has no conflicts of interest.