Microfluidic Methods to Advance Mechanistic Understanding and Translational Research in Sickle Cell Disease

Abstract

Sickle cell disease (SCD) is caused by a single point mutation in the β-globin gene of hemoglobin, which produces an altered sickle hemoglobin (HbS). The ability of HbS to polymerize under deoxygenated conditions gives rise to chronic hemolysis, oxidative stress, inflammation, and vaso-occlusion. Herein, we review recent findings using microfluidic technologies that have elucidated mechanisms of oxygen-dependent and -independent induction of HbS polymerization and how these mechanisms elicit the biophysical and inflammatory consequences in SCD pathophysiology. We also discuss how validation and use of microfluidics in SCD provides the opportunity to advance development of numerous therapeutic strategies, including curative gene therapies.

Introduction

Sickle cell disease (SCD) is an autosomal recessive disorder caused by a single point mutation in the β-globin gene of hemoglobin. This mutation results in altered sickle hemoglobin (HbS). Under deoxygenation, polymerization of HbS fibers occur within the red blood cell (RBC) and deforms the shape of the sickle RBC (sRBC). Additionally, a spectrum of molecular and biophysical consequences within the sRBC, on the sRBC’s surface, and in the local vascular environment simultaneously contribute to the altered hemorheology seen in SCD.¹

Since the discovery of the single point mutation in Hb more than 60 years ago, our understanding of SCD pathophysiology has increased dramatically. However, the clinical phenotype varies widely, with marked heterogeneity in symptoms and severity. The heterogeneity in clinical phenotypes is poorly understood, hence there is no standard protocol to risk stratify according to disease severity. This lack of risk stratification contributes to difficulties in disease management by healthcare providers.² For example, clinical symptoms can manifest as early as infancy and can continue to adulthood. Chronic disease burden often leads to end-organ damage and early mortality in those with SCD, who have a median life expectancy of 54 years.³ Among the many potential acute clinical manifestations of SCD, extreme episodes of pain known as vaso-occlusive episodes (VOEs) remain the clinical hallmark of the disease. VOEs are the most common reason for hospitalizations in SCD and can significantly reduce an individual’s quality of life.^4–6

Development of new therapies for SCD has languished for many non-biologic reasons,⁷ however lack of sufficient and validated biomarkers to stratify patient phenotypes and serve as clinical trial endpoints has also hindered progress. Additionally, lack of biomarkers also has limited development of standardized treatments for SCD individuals. Despite these barriers, three new SCD-modifying therapies have been brought to market since 2017, with more under investigation — including potential curative gene therapies.^8,9 With more therapies clinically available and many more under various stages of clinical development,⁸ now more than ever there is a need to fundamentally define how HbS and sRBCs drive SCD pathophysiology in order to accelerate development of new SCD therapies and provide evidence-based recommendations for personalized SCD management.

In the past decade, developments in microfluidic technologies have advanced the scientific understanding of sRBC biophysical interactions in SCD pathophysiology. Furthermore, new therapeutics targeting the sRBC and its biophysical composition are currently under development. Herein, we highlight how these technologies have advanced understanding and treatment of SCD and discuss how the development and use of these technologies can be incorporated into the validation pipelines of new SCD therapeutic approaches, including gene therapies.

Contributions of Oxygen-Dependent Microfluidics to Assess RBC Function

Intravital microscopy in SCD mouse models has been and remains a mainstay experimental tool for evaluating sRBC behavior.¹⁰ However, in vivo methods often lack quantitative control of vital biophysical parameters such as shear stress, blood vessel geometry, and mass transport, all of which may contribute to vaso-occlusive physiology. Microfluidics advancements allow for researchers to capture the dynamic effects of the deoxygenated-induced polymerization on hemorheology in a controlled, physiologically relevant system. In microfluidic devices that lack an endothelial cell layer, but otherwise recapitulate the in vitro vasculature anatomy and shear rates, blood can be perfused through the device at a constant pressure and exposed to decreased oxygen tensions while sRBC rheological flow response can be measured (Fig 2A). These microfluidic devices can be used to demonstrate that deoxygenation initiates alterations in sRBC rheology that are sufficient to increase viscosity, slow sRBC velocity, and impede sickle cell blood flow; these data provide additional insight into the dynamics of vaso-occlusive events (Fig 2B).^11–20 Furthermore, microfluidic studies show that sRBC velocity is reduced when oxygen tension is reduced and that there is an oxygen tension threshold where blood flow transitions from oxygen independent flow to oxygen-dependent flow (Fig 2C). Using these devices, oxygen tension can be decreased in a step-wise fashion to determine the lowest oxygen tension that can be tolerated without a reduction in sRBC velocity. For example, in a set of experiments using SCD patient blood, increased viscosity was observed at oxygen tensions as high as 80 mmHg, similar to that of arterial circulation.²¹ These findings support the arterial clinical manifestations seen in SCD such as stroke, moyamoya, and pulmonary hypertension.^22,23 These studies suggest that arterial oxygen tensions may not be protective against stasis-mediated physiology.

Figure 2: Microfluidics devices to interrogate oxygen-dependent sickle RBC rheology.

A. Photo of a microfluidic device for scale, top-down view. B. Still image of red blood cells moving through microfluidic device. C-D: Raw representative RBC velocity data from a SCD subject. C. RBC velocity is tracked through the device while environmental oxygen tension within the device is varied between 0 % and 21%. In the untreated sample, RBC velocity is dependent upon oxygen tension as RBC flow velocity is reduced with hypoxia. D. RBCs from the same SCD subject after 1-hour incubation with 500 μM of voxelotor at 37C ex vivo. Identical oxygen tension protocol as in B; however, voxelotor-treated RBCs experience smaller velocity reductions with hypoxia and reach oxygen-independent flow at lower oxygen tensions than untreated samples.

As microfluidics technologies have been useful in evaluating the biophysical consequences of HbS polymerization under deoxygenated conditions, they also have revealed several other mechanisms known to be fundamental to SCD pathologic rheology. By first observing the pathology in these devices, microfluidic studies can serve as an indicator of efficacy in current and potential treatments involved in the management of SCD. As many therapies target hypoxia-induced polymerization, oxygen-dependent devices particularly can serve as a marker of drug efficacy. For example, there has been a long-standing observation that the presence of fetal hemoglobin (HbF) reduces the concentration of HbS and thereby also reduces polymerization rates.^24–28 Therefore, the co-inheritance of additional genetic mutations that increase HbF concentrations such as hereditary persistence of HbF, or reduce HbS concentration, such as α-thalassemia or β^C allele with β^S may reduce disease severity.³ Based on this premise, induction of HbF has been a long-standing treatment for SCD, and hydroxyurea, which induces HbF in patients, has been approved for use in adults SCD since 1998.^29,30 Indeed, in microfluidics models, sRBC treated with hydroxyurea demonstrate decreased oxygen-dependent viscosity changes compared to untreated blood.^13,15 Collectively, these data offer a benchmark by which to compare efficacy and benefits of other HbF induction strategies, including gene therapies.

In addition to evaluating the effects of reducing HbS concentration by increasing HbF, oxygen-dependent microfluidics can also be used to screen for treatment effects of oxygen-dissociation agents on rheology. Approved since 2018 to prevent hemolysis in SCD, voxelotor (GBT 440) is thought to stabilize HbS in the oxygenated state, thereby reducing HbS polymerization. In a phase 3 randomized, double-blind, placebo-controlled trial, voxelotor demonstrated that patients’ hemoglobin improved by 1g/dL after 6 months of use along with a reduction in hemolysis markers.³¹ This clinical finding was confirmed in oxygen-dependent fluidic devices (Fig 2A), where sRBCs incubated ex vivo with voxelotor were resistant to a decrease in hypoxia-induced velocity slowing (Fig 2 B, D); this finding indicates less oxygen-dependent rheology. In a representative patient sample (Fig 2C-D), it appears that oxygen-independent flow is achieved around 6% O₂ (45mmHg) in the voxelotor-treated sample. However, in the untreated sample (Fig 2C), there continues to be oxygen-dependent flow at the same oxygen tension. Collectively, these data suggest that the development, validation, and use of oxygen-sensitive microfluidic technologies may identify strategies to prevent oxygen-dependent sRBC rheologic changes and inform strategies to better treat and prevent SCD complications.

sRBC Dehydration through Activation of Cation Transport Channels during Deoxygenation

The ability to evaluate sRBC oxygen-sensitive rheology also permits for concurrent assessment of the interactions between oxygen-dependent HbS polymerization and sRBC cellular dehydration. HbS polymerization and decreased oxygen tensions lead to changes in RBC membrane permeability and permits the passage of cations and other small molecules to efflux through several channels (Fig 3A). Altered permeability ultimately leads to dehydration and increases in mean corpuscular hemoglobin concentration (MCHC). Increased MCHC in a sRBC increases HbS concentration, making cells at increased risk of polymerization. For example, under deoxygenation and polymerization, the P_sickle channel, potentially identified as the PIEZO1 channel, becomes activated and allows an intracellular influx of sodium (Na⁺) and calcium (Ca²⁺) and an efflux of potassium (K⁺). Subsequent activation of the RBC Gardos (KCCN4) channel by the influx of Ca²⁺ causes a net loss of K⁺ and dehydration.³² Further dehydration also occurs due to activation of the K-Cl cotransporter. Additionally, Ca²⁺ efflux leads to increased exposure of pro-coagulant phospholipid phosphatidylserine (PS) and increased vascular adhesion.^33,34 Using several methods to assess sRBC biophysical changes, several groups have noted that cyclic deoxygenation augments sRBC channel permeability leading to decreased deformability and significant shifts in RBC rheology (Table 1).^{16,17,20,35–38} Overall, deoxygenation triggers HbS polymerization, which increases sRBC dehydration, which compounds polymerization and ultimately drives development of a rigid, dehydrated sRBC containing polymerized HbS.^16,17

Figure 3: Implications of sickle red blood cell dehydration.

A. Channels involved in sickle cell red cell dehydration. The Na-K pump (Na, K-ATPase) is more active in sickle RBCs. Monensin and other drugs can target this pump. The P-sickle channel, which may be PIEZO1, leads to Ca2+ influx. Furthermore, activation of GPCR and other cytokine-mediated channels leads to Ca²⁺ influx, which leads to Gardos channel activation. This leads to K⁺ and water efflux by aquaporins (AQ1). In SCD, the K-Cl cotransporter is also activated at low oxygen tensions, causing efflux of K⁺ and subsequently loss of water by aquaporins (AQ1).. Image created in Biorender. Fluid tonicity effects sRBC adherence to endothelium. B. Fluorescent-stained sRBC (red) adhered to human umblical vein endothelial cells. C. Bar graph illustrating how sRBC from 6 indivudals were less adherent to endothelium after exposure to hypotonic fluids. Figure modified and used with permission from Carden et. al. 2017.

Table 1:

Examples of microfluidics models applied for investigation of sickle RBC biophysical properties

Model Type	Measurements	Findings	References
Microvasculature-on-a-chip	Transit time Time to occlusion Adherent sickle cells Platelet and fibrin deposition	Endothelial-sRBC interactions and geometry contribute to vaso-occlusion	Carden et al. 2017. Microcirculation Carden et al. 2017 Blood Mathur et al. 2021. Bioeng Transl Med
Oxygen-Tension modulation Microvasculature model Shear-thinning	Oxygen-dependent velocity Oxygen-dependent rheology Oxygen-gradient and vascular size Oxygen-dependent viscosity changes	Sickle Hb concentration increases blood viscosity with decreasing oxygen tension	Wood et al. 2012. Sci Transl Med Lu et al. 2018. Am J Hematol Valdez et al. 2019. APL Bioeng
Single-RBC Hemoglobin polymer content	Measurement of single-RBC Hb polymer	Semi-quantitative measurement of hemoglobin polymer in single RBCs as function of oxygen	DiCaprio et al. 2019. PNAS
Red cell biomechanics and rheology	Sickling kinetics Image-based quantification of SRBC deformity RBC cortical tension Splenic-like slit device pH and oxygen associated biomechanics sRBC behavior in cycles of hypoxia	Cortical tension in sRBC and pH increase stiffness of RBCs Shape-dependent velocities are important.	Abbyad et al. 2010. Lab Chip Guraprasad et al. 2019. Am J Hematol. Li et al. 2017. PLoS Comput. Biol. Gambhire, P et al. 2017. Small Guo et al. 2014. J Biomech. Du E. and Dao M. 2019. Exp Mech.
RBC adhesion	Flow dependent RBC adhesion to laminin and fibronectin RBC deformability	Sickle RBC adhesion to endothelial ligands or activated endothelium	Alapan et al. 2014. Scientific Rep Alapan et al. 2016. Transl Res Kucukal et al. 2018. Am J Hem
OcclusionChip Microfluidic electrical impedance	RBC-mediated microvascular occlusion	Percent occlusions and percent impedance change across device correlates with HbS phenotype	Man et al. 2021. Microcirculation Man et al 2021. Lab Chip Liu et al. 2018. Phys Fluids.
Pulsatile parallel plate flow adhesion	Sickle RBC adhesion to VCAM-1, Laminin	Sickle RBC adhesion to endothelial ligands or activated endothelium can be blocked by antibodies	White J et al. 2015. Clin Hemorheol Microcirc

Clinically, the interplay between sRBC dehydration and permeability is important. For example, previous research has evaluated if systemic fluid resuscitation in individuals with SCD, particularly during times of VOE, can mitigate some of the effects of sRBC dehydration and reduce VOE. However, the most recent American Society of Hematology (ASH) Clinical Guidelines for SCD on managing acute or chronic pain does not offer recommendations regarding management of supplemental fluids to aid patients with SCD.³⁹ This is in part due to the known pathologic permeability of sRBCs and the shift in cellular osmolality that it presents. By administering hyperosmolar fluids, there is a theoretical potential to increase cellular dehydration by increasing plasma osmolality and changing the efflux of potassium and calcium through transport channels (Fig 3B-C).^40–42 In fact, some suggest using hypo-osmolar solutions for hydration for the potential reverse effect.⁴² For example, in Carden et al. microvasculature-on-a-chip model, when comparing sRBCs exposed to hypo-osmolar solutions, sRBCs exposed to normal saline in vitro were stiffer, had longer transit times, and were more likely to occlude a microfluidic channel under normoxic and hypoxic conditions (Fig 3C).⁴³ Interestingly, in a retrospective cohort study of SCD patients receiving intravenous fluid and pain medication, administration of normal saline was associated with poorer pain control.⁴⁰ Collectively, these studies justify further work to identify how sRBC hydration, patient fluid resuscitation, and volume homeostasis affect HbS polymerization and VOE outcomes. Furthermore, this work highlights that incorporation of microfluidics into clinical studies investigating the effects of various osmolar fluids on sRBC rheology may provide better insight into our understanding of sRBC hydration on a cellular level and can help answer important clinical questions regarding how to best use fluids to manage patients.

Microfluidics to Model Sickle Red Cell Interactions with Endothelium

Decades of evidence demonstrates that sRBC adhesion to the endothelium contributes to disease pathophysiology.^44,45 In addition to heme-mediated activation of endothelial cells⁴⁶, the increased expression of surface molecules on the sRBC are known to contribute to endothelial adhesion. Increased expression of glycoprotein IV, integrin VLA-4, Landsteiner-Wiener intercellular adhesion molecule 4 (ICAM-4), and endothelial Lutheran/basal-cell adhesion molecule (Lu/BCAM) have all been studied and reported in the literature to play a role in increased sRBC adhesion.^46–51 Furthermore, as HbS polymerization increases sRBC density, early adhesion studies conducted under static conditions demonstrated that dense sRBCs exhibited higher amounts of endothelial adhesion when compared to light-density sRBCs.^18,45 Therefore, sRBC interactions with the endothelium or endothelial activation ligands is an important consideration in modeling.

Recently, a series of microfluidic systems have served as a tool to recapitulate RBC-endothelial adhesion occurring in SCD (Table 1). Collectively, results from these studies demonstrate that increased sRBC adhesion molecule expression contributes to direct vaso-occlusion or increased capillary transit time, leading to longer exposure to deoxygenation and increased polymerization rates.^14,52–57 For example, in Papageorgiou et al.’s microfluidic device, which recapitulates the molecular and dimensional environment of post-capillary venules and is coated with the endothelial matrix protein fibronectin, hypoxia enhances sRBC adhesion four-fold. Furthermore, during HbS polymerization, the growing HbS fiber bundles also contribute to increased adhesion.⁵⁷ Other groups have used microfluidic devices coated with endothelial matrix proteins laminin and VCAM, and demonstrated increased endothelial adhesion strength among the most non-deformable sRBCs as compared to more deformable sRBCs due to increased glycoprotein IV, integrin VLA-4, ICAM-4, and Lu/BCAM expression.^58–63 Likewise, White et al. has also modeled the microvasculature with a VCAM-1-coated device and has shown that pulsatile flow similar to that in vivo may lead to increased sRBC adhesion.^64,65 Furthermore, Mathur et al. have also cultured SCD patient-derived blood outgrowth endothelial cells inside microfluidics to assess flow-mediated endothelial changes and sRBC adhesion.^66,67 These model devices provide in-depth insight into the complex interactions occurring at the endothelium and how they might be contributing to vaso-occlusive events.

Red cell heterogeneity and implications for use of ektacytometry and microfluidics

sRBC Heterogeneity

While the various mechanical characteristics of sRBCs as a whole are important to SCD pathophysiology, RBC deformability and adhesion can vary across the sRBC lifespan.¹⁶ At any given time, RBCs of different ages co-exist and add to the complexity and heterogeneity of SCD pathophysiology. Given that RBCs must pass through the narrowest of vessels to deliver oxygen in the microcirculation, deformability plays a critical role in the survival of an RBC and SCD pathology.^16,18,36,38 For example, as previously mentioned, HbS polymerization amplifies other circulation-dependent changes in RBC deformability and rigidity. Within the circulation, all RBCs undergo dynamic, transient deformations in order to migrate through various capillary diameters or the spleen.^37,68,69 Over time, these deformations lead to alterations in membrane structure, generation of microvesicles (MV), and increased exposure of negatively-charged PS.⁷⁰ Indeed, in computational studies of splenic function, RBC surface-to-volume ratio is a critical parameter of splenic filtration. ^37,69 Additionally, using models of splenic slits, sRBCs exhibit cytoskeletal changes that result in a threefold increase in RBC cytoplasmic viscosity.³⁷ Compared to healthy controls, return to equipoise after splenic transition is delayed. Collectively, these data suggest that even at steady state and under fully oxygenated conditions compared to non-sickle RBCs, sRBC undergo alterations in cell membrane plus cycles of polymerization throughout circulation that contribute to an overall decrease in deformability in RBCs from patients with SCD.⁷¹ The increase in sRBC rigidity can cause RBCs to be retained in the spleen, leading to extravascular hemolysis and contributing to chronic anemia in SCD.^37,69 In early studies investigating cellular densities in sRBCs, 4 different subpopulations of cells based on density were identified: reticulocytes, discocytes, dense discocytes, and irreversibly sickled cells (ISCs). These subpopulations demonstrated corresponding increases in density with MCHC and increases in bulk viscosity measured in a viscosimeter.⁷² Furthermore, the extent to which these subpopulations exist within a patient varies and may contribute to an individuals’ clinical disease.^36,72 The critical level at which each subpopulation causes clinical complications or severe disease is unknown. However, with the ability to measure adhesion and deformability, investigators may begin to understand the impact of each sRBC subpopulation and how its adhesive and deformable properties can drive SCD pathology, providing for targeted therapeutic development. Overall, the best technique to accurately measure deformability or density remains under debate; however, several techniques are being used, such as ektacytometry and single-cell measurements (Table 1)

Deformability by Ektacytometry

Ektacytometry is useful in determining RBC deformability, particularly while measuring deformability in red cell populations in flow. Because of this, ektacytometry is often used to study SCD and the effects on blood flow of sRBCs’ decreased deformability and increased rigidity.⁷³ In one study using ektacytometry in SCD, a correlation of decreased deformability and increased rates of hemolysis was observed.⁷⁴ Likewise, compared to HbSS individuals with high RBC deformability by ektacytometry, HbSS individuals with lower RBC deformability had significantly more leg ulcers and increased VOE events.^75,76

Unfortunately, ektacytometry does not measure single-cell deformability and often provides unclear data on specific rigid, or dense RBC subpopulations that are often implicated in SCD pathology. This is, in part, because ektacytometry methodology is dependent on diffraction patterns that lack universal definitions.⁷³ When looking at the two most common assays to measure deformability, ektacytometry and micro-pore filtration, neither technique can accurately predict the ability of RBCs to perfuse a microvascular network.⁷⁷ One reason for this inaccuracy is that the distribution of polymer is bimodal, making bulk flow deformability measurements incomplete and inconclusive, as has been shown that under finite oxygen tensions.⁷⁸ Therefore, as microfluidics models reveal the importance of sRBC subpopulations and cellular heterogeneity, assessing general deformability across all cells may not provide a specific nor accurate reflection of sRBC deformability and lead to inaccurate data interpretation. Hence, development of strategies to evaluate single-cell deformability and behavior may improve upon ektacytometry.

Deformability and Single-Cell Measurements by Microfluidics

More recently, with microfluidic devices, the ability to probe single-cell behavior provides better insight into the variable behaviors of individual sRBC subpopulations and their effect on blood flow. Man et al. have developed microfluidic models of progressively narrowing channels from 20 to 4 μm to recapitulate a capillary bed. To assess deformability, the percent of occlusions after a fixed time of blood flow may be measured using the occlusion index (OI). Lower OI has correlated with patients with more severe phenotypes, providing potential for a tool to assist in clinical risk stratification or predictive outcomes.^79,80 Further studies by Li et al. also found correlation between deformability and sRBC flow.⁸¹ ISCs are not thought to contain polymerized hemoglobin, but are thought to be a product of membrane damage (i.e., MV and PS exposure) during repeated cycles of polymerization. ISCs are poorly deformable, have a short lifespan, and are highly correlated with hemolytic anemia.⁸² In a microfluidic chip designed with narrow channels as small as 5 μm to mimic the mechanical stress exerted on sRBCs in microcirculation, a significant reduction in ISCs were observed at the outlet of the device, suggesting the ISCs hemolyzed in transit through the device as a result of mechanical stress.⁸³ This finding is not surprising given that ISCs are prone to hemolysis compared to other RBC populations.⁵⁴ Within patients with identical genotypes of HbSS disease, it has been reported that ISCs comprise a wide, variable range between 5–50% of the total red cell population. The effects of a higher proportion of ISCs on whole blood rheology and its manifestations are highlighted in a study that correlated ISC proportion to spleen function, an organ that filters RBCs through endothelial slits 1–3 μm wide. In this study, pediatric patients that had higher percentages of ISCs experienced more episodes of acute splenic sequestration and splenic dysfunction.⁸⁴

At present, defining and assessing the contribution of sRBC heterogeneity to SCD pathogenesis has been a major barrier to predicting success of many therapeutics, including gene therapy. As discussed earlier, induction of HbF has been a long-standing treatment for SCD. HbF is heterocellularly expressed only in a subset of sRBCs, F-cells, which vary in concentration in individuals ranging from 2 to 80% of sRBCs.⁸⁵ However, the HbF concentration within each F-cell has a variable distribution. These individual differences in HbF expression and varied distribution among F-cells are likely controlled genetically and continue to be poorly understood.⁸⁶ Although high levels of HbF are strongly associated with fewer vaso-occlusive complications, such as VOEs and acute chest syndrome,⁸⁷ as polymerization inhibition requires a threshold intracellular HbF concentration, it is the distribution of HbF concentrations among F-cells that has a greater impact on disease.⁸⁵ In recent work, Di Caprio et al. developed a single-RBC hemoglobin polymer microfluidic model designed to assess oxygen-dependent Hb polymers.⁷⁸ This systems demonstrates that sRBC morphology and HbS polymer distribution is oxygen sensitive. Additionally, at a single-cell level this system identified patient-specific variable oxygen saturation distribution that is correlated to Hb polymer fraction.⁷⁸ Therefore, moving beyond ektacytometry, microfluidics and single-RBC analysis may offer the ability to assess single RBC HbS polymers and HbF concentration as it pertains to the effects of hemoglobin modification.⁷⁸

Application of Microfluidic Technologies to Assess New SCD-Directed Therapies

A persistent challenge in understanding the pathophysiology of SCD is the broad clinical phenotypic heterogeneity and lack of correlative biomarkers to stratify phenotypes. These factors significantly complicate the diagnosis and treatment process for patients. A recent hallmark study, the ELIPSIS study, monitored patients with SCD at home with an electronic patient-reported outcome tool and actigraphy over the course of 6 months. Additionally, a mobile-based blood collection team obtained blood samples when VOE was reported-permitting investigators to analyze trends over the course of a patient-reported VOE.⁶⁵ During times of patient-reported VOE at home, whole blood adhesion to P-selectin was evaluated in a fluidic-based assay and found to be significantly elevated from individual patient’s baseline.⁶⁵ These studies are notable as they provide insight into pathophysiology during VOE and substantiate previous clinical trial findings demonstrating that patients receiving crizanlizumab, a human monoclonal antibody against P-selectin, experienced a reduction in VOE.⁸⁸

The ELIPIS study elegantly highlights that microfluidics assessment has the potential to identify and validate treatment strategies. As recently extensively reviewed by Telen et al., multiple strategies directed toward sRBC rheology and interactions continue to be explored.⁸ With expanding categories of interventions (anti-adhesion, anti-dehydration, oxygen-dissociation), when and how to proceed with treatment in a phenotypically heterogenous patient population will need to be identified. Going forward, combining microfluidic measurements, along with other tools such as computational modeling, may allow for non-invasive, upfront strategies to predict ability of candidate molecules to meaningfully alter patient-specific SCD parameters. An example of how this could be integrated is illustrated in Figure 4. First, patient-specific pre-intervention studies can be correlated to patient-reported outcomes (ie pain), and standard biometrics (blood work, vitals). If, with integration of these parameters, a subject is found to have increased adhesive properties on oxygen-dependent sRBC adhesion or occlusion models, one may select for anti-adhesive based therapies (ie crizanulizumab). However, if a subject has increased single-RBC Hb polymer content or oxygen-dependent sRBC velocity changes, one may instead chose oxygen-dissociation agents (ie voxeletor). After intervention, use of these technologies also will permit for monitoring of response, which along with patient-reported outcomes related to pain and clinical data may permit for increased standardization between trials. Finally, as multiple agents are in the developmental therapeutic pipeline-leveraging computational methods to quantify and predict patient-specific rheological patterns may inform studies combining therapies to maximize benefit and minimize toxicity. For example, if cohort of subject has rheologic patterns with increased single RBC polymer content and increased adhesion, combining anti-adhesive agent (crizanulizumab) with hemoglobin F induction (hydroxyurea) may be beneficial.^89,90 As microfluidic models are non-invasive and can be standardized, incorporation of microfluidic measurements into the clinical development pipeline for SCD therapies is justified.⁹¹

Figure 4: Potential applications of microfluidic technologies in the development of SCD therapies.

1. Samples can be collected before or after treatment. 2. Ex vivo studies from individuals with SCD can be used. These would include oxygen-dependent sRBC velocity, oxygen-dependent sRBC adhesion studies, and oxygen-dependent occlusion studies to evaluate RBC behavior under physiologic relevant conditions. Furthermore, single-RBC polymer content can be established. 3. These findings can be integrated with clinical parameters, including patient-based assessments, to assess response. Furthermore, combined with computational modeling, cycles of further interventions to optimize combination therapies may be completed. Image created in Biorender.com.

Strategies to Leverage Microfluidics Technologies in Sickle Cell Gene Therapy

Gene therapy to prevent HbS polymerization is the ultimate panacea in SCD treatment.⁹² At present, there are several ongoing clinical trials evaluating gene addition and gene editing strategies to cure SCD (Table 2). In general, gene addition approaches use lentiviral vectors to carry therapeutic genes into hematopoietic stem cells (HSCs). Gene editing using CRISPR-Cas is also underway to edit hemoglobin S translation.^93–95 Early clinical trials using gene additive strategies to introduce alternative γ-globin genes have shown clinical efficacy in patients with SCD and β-thalassemia.^9,94–99 Since then, several clinical trials have emerged to introduce anti-sickling globins (Table 2). Recently, two of the largest trials did report serious adverse outcomes prompting temporary cessation of enrollment.^100,101 However, for those patients with SCD who have received lentiviral therapy, overall survival (OS) and event-free survival (EFS) are both 96% compared to historical allogeneic matched sibling HSCT OS and EFS of 95% and 92%, respectively. Importantly, after treatment, SCD subjects had no episodes of acute chest syndrome or VOE. Moreover, participants reported clinically meaningful improvements in pain reduction at 12 months post-treatment.¹⁰⁰

Table 2:

Clinical Trials for Gene therapy and editing in sickle cell disease

Strategy	NCT Number	Phase/Enrollment/Status	Results	Citations
Lentiviral transfer of γ-globinG16D – ARU-1801	NCT02186418	Phase 1/2 Enrollment- 10 Status: open	Increased F-cells, no VOE	Grimley M, et al. 2020. Blood
Lentiviral transfer of modified HBB encoding antisickling variant βA87Thr:Gln[βA-T87Q] LentiGlobin BB305	NCT02151526	Phase 1/2 Enrollment- 7 Status: closed	Stable CBC, Decreased reticulocyte, decreased RBC sickling under hypoxia and improved RBC deformability	Ribel JA et al. 2017. NEJM
	NCT02140554	Phase 1/2 Enrollment- 50 Status: not recruiting*
	NCT04293185	Phase 3 Enrollment- 35 Status: recruiting*
AS3-FB vector transduced peripheral blood CD34+ cells	NCT02247843	Phase 1/2 Enrollment- 6 Status: not recruiting	None reported Pre-clinical: decreased sickling, engraftment of mice	Romero Z. et al. 2013. JCI
Lentiviral targeting of γ-globin repressor BCL11A with BCH-BB694 BCL11A shmiR vector	NCT03282656	Phase 1 Enrollment- 15 Status: SUSPENDED*	Increased HbF, increased % F-cells, increased HbF per F cells	Esrick, EB. Et al. 2021. NEJM
Reactivation of fetal hemoglobin by using Crisper-Cas (CTX001) editing to suppress BCL11A β-globin switch	NCT03745287	Phase 1/2 Enrollment- 45 Status: recruiting	Increased HbF, reduced transfusion needs and no VOE	Frangoul, H. et al. 2021. NEJM

^*Studies were paused in March 2021 due to the development of myelodysplastic syndrome and acute myeloid leukemia in patients in NCT02140554 and NCT04293185 prompting suspension of NCT03282656 out of caution. NCT02140554 and NCT04293185 currently active.

Overall, for gene therapy and other potentially disease-modifying therapies, use of microfluidic strategies, such as oxygen-dependent velocity, endothelial interaction, and single-RBC Hb polymerization assays will be valuable in identifying how HbF concentration distribution among F-cells correlates to SCD clinical phenotype. Thus far, the largest trial, by Rivel et al. used ektacytometry and morphologic assessment to demonstrate sRBC benefits of lentiglobin modification.⁹⁷ More recent work by Frangoul et al. and Esrick et al. demonstrated increased F-cell number, with further studies evaluating sRBC biophysical characteristics anticipated as more subjects enroll in the trials.^95,98 Collectively, these studies highlight the importance of incorporating measurements of sRBC biophysical behavior to obtain a robust characterization of the myriad of effects driven by the sRBC.

Conclusions

In conclusion, microfluidics models have advanced our fundamental understanding of how sRBCs drive SCD pathogenesis. Thus far, the importance of HbS polymerization has driven much of the evaluation, with ektacytometry studies increasingly being incorporated into clinical trials. Moving forward, adoption of single-RBC polymerization models may permit for further insights into oxygen-sensitive RBC characteristics. Additionally, incorporation of patient samples in microvascular occlusion models offers the ability to rapidly screen candidate therapeutics for further pre-clinical (i.e., mouse) studies. During clinical studies, use of microfluidic assays allows for non-invasive measurement of candidate interventions on hemorheology. For far too long, the SCD therapeutic pipeline has been stagnant. However, with multiple therapeutics and gene therapy under rapid develop, the addition of microfluidic technologies offers the exciting opportunity to accelerate the investigational pipeline for sickle cell therapies.

Figure 1: Sickle hemoglobin fiber polymerization is related to hemoglobin S concentration and is highly inefficient.

A. Hemoglobin S polymerization (C) as a function of HbS molecules per cell as calculated by Lu, L et al. 2017. Figure moodified and used with permission. B. Differential interference contrast images showing an example of heterogeneous nucleation (left) and the intersection of two separate HbS fibers (right). Red and yellow arrowheads indicate the region where the intensity value was calculated for the multifiber region (nucleation or intersection) or the adjacent region, respectively. Scale bars, 2 μm. (Figure modified and used with permission from Castle et al. 2019).