Current therapy for sickle cell disease (SCD) is limited to supportive treatment of complications, red blood cell transfusions, hydroxyurea, and stem cell transplantation. Difficulty in the translation of mechanistically based therapies may be the result of a reductionist approach focused on individual pathways, without having demonstrated their relative contribution to SCD complications. Many pathophysiologic processes in SCD are likely to interact simultaneously to contribute to acute vaso-occlusion or chronic vasculopathy. Applying concepts of systems biology and network medicine, models were developed to show relationships between the primary defect of sickle hemoglobin (Hb S) polymerization and the outcomes of acute pain and chronic vasculopathy. Pathophysiologic processes such as inflammation and oxidative stress are downstream by-products of Hb S polymerization, transduced through secondary pathways of hemolysis and vaso-occlusion. Pain, a common clinical trials endpoint, is also complex and may be influenced by factors outside of sickle cell polymerization and vascular occlusion. Future sickle cell research needs to better address the biologic complexity of both sickle cell disease and pain. The relevance of individual pathways to important sickle cell outcomes needs to be demonstrated
Sickle cell disease (SCD) is a group of disorders caused by a mutation in the sequence of beta globin, leading to polymerized hemoglobin (sickle hemoglobin, hemoglobin S), hemolytic anemia, painful vaso-occlusive events, vascular remodeling, acute and chronic organ injury, and shortened lifespan. Sickle cell disease affects over 70,000 individuals in the United States, and there are at least 75,000 hospitalizations costing over $500 million annually for treatment of SCD complications [
Since the passing of the National Sickle Cell Control Act in 1972, over one billion dollars have been allocated from the National Heart, Lung and Blood Institute of the National Institutes of Health (NIH) for SCD research [
Major types of sickle cell intervention studies registered on the website
Pathway/mechanism | Number of studies |
---|---|
Bone marrow transplantation | 21 |
Hemoglobin F induction | 9 |
Nitric oxide related | 8 |
Analgesic regimens | 6 |
Nutritional supplements | 4 |
Adhesion inhibition | 3 |
Transfusion therapy | 3 |
Red cell hydration | 3 |
Noninvasive ventilation | 3 |
Statins | 2 |
Renin-angiotensin pathway in nephropathy | 2 |
Iron chelation | 2 |
Educational tools | 2 |
Antiinflammation | 2 |
Gene transfer | 1 |
Carbon monoxide donation | 1 |
Anti-coagulation | 1 |
Director of the Division of Blood Diseases and Resources at the NIH, W. Keith Hoots, recently wrote, “Research over the decades indicates that the primary defect in hemoglobin that results in polymerization of the protein under low oxygen conditions and resultant cellular deformity of the red blood cell initiates a complex downstream pathogenesis associated with vascular injury, and organ ischemia. Deciphering this in a manner that informs successful therapies that improve all target organs continues to challenge hematologists” [
Preclinical studies and clinical trials targeting three different sickle cell pathways will be reviewed, including inhibition of adhesion by poloxamer 188, inhibition of Gardos channel-induced erythrocyte dehydration by senicapoc (ICA-17043), and treatment of acute pain episodes with inhaled nitric oxide. Comprehensive reviews of approaches to sickle cell treatment have been published elsewhere [
Fluorocarbon emulsions, including identical but variously named compounds Pluronic F-68, Flocor, RheoThRx, and poloxamer 188, have been studied in SCD since 1975 [
A phase II randomized, double-blinded placebo-controlled trial (RCT) testing poloxamer 188 was conducted with fifty subjects with SCD who presented within 4–18 hours of onset of acute pain. They were treated with either placebo or poloxamer 188 infusion of 300 mg/kg for 60 minutes, followed by 47-hour maintenance infusion of 30 mg/kg/hr. In the 31 subjects who completed the 48 hour infusion, there was a statistically significant reduction in pain duration by 36 hours and a 3–5-fold reduction in analgesic use.
This was followed by a phase III RCT using poloxamer 188, with 255 individuals enrolled at 40 study sites [
Close comparison shows differences between the Phase II and III trials. There was a shorter duration of pain prior to beginning study drug in the Phase II trial. The inclusion criteria for the Phase II trial included 4–18 hours of moderate pain on presentation, and there was a median of 17 hours between onset of pain and start of study infusion. In contrast, subjects in the phase III trial had 1.87–2.25 days of pain from onset of crisis to randomization, with an additional 2.3 hours between randomization to start of study infusion, or almost 3 times longer pain duration before study drug infusion was begun. In addition, the loading dose of poloxamer 188 was 3-fold higher in the phase II trial compared to phase III. Lastly, the phase II trial allowed nonsteroidal anti-inflammatory drugs (NSAIDs) for analgesia, but they were not allowed during the study drug infusion and for 12 hours following discontinuation in the phase III trial. The absence of opioid-sparing effect of NSAIDs may have contributed to the lack of difference in opioid usage.
Erythrocyte water content is an important determinant of sickle hemoglobin concentration and polymer formation. Erythrocyte hydration status is controlled by KCl and water loss through two transport systems, K-Cl cotransport and the calcium-dependent potassium, or Gardos channel. Gardos channel inhibition in sickle cells was found to improve erythrocyte hydration status [
Based on the structure of clotrimazole, alternative agents were developed to more potently and specifically inhibit the erythrocyte Gardos channel. Senicapoc (ICA-17043) was found to inhibit SAD mouse red cell dehydration
This was followed by a phase III RCT with 297 adult subjects randomized at 75 study centers to receive senicapoc for 52 weeks [
Of note, the Gardos channel, also known as the intermediate conductance Ca2+-activated K+ channel, KCa3.1, KCNN4, or SK4, is expressed in other cell types, such as T- and B-lymphocytes, macrophages, endothelial cells, fibroblasts, vascular smooth muscle cells, and neurons. In blood vessels, the KCa3.1 channel plays a role in endothelium-derived hyperpolarizing factor- (EDHF-) induced vasodilation, which may play a major role in the microcirculation [
Nitric oxide (NO) is an important vasoactive molecule, with effects on vascular smooth muscle dilation and modulating leukocyte and platelet activation. The use of nitric oxide in treatment sickle cell disease acute chest syndrome (ACS) was first reported in 1997 [
In SAD mice, inhaled NO at 20 ppm improved survival rates during exposure to hypoxia when it was given 30 minutes prior to hypoxic exposure and continued during the entire exposure [
There have been three published studies of inhaled NO therapy for acute pain episodes in SCD. In the first study, 20 pediatric subjects with SCD were enrolled, 10 who were treated with inhaled NO 80 ppm for 4 hours and 10 with placebo [
With the strength of the preclinical findings, why were these therapeutic strategies not effective in reducing the impact of acute pain, the most common complication of sickle cell disease? Sickle cell clinical research probably suffers from the same challenges as other clinical trials in the United States. In 2012, the Institute of Medicine published a report that discusses the major challenges with current clinical trials, including high costs due to elaborate administrative procedures, failure to enroll sufficient numbers of participants, regulatory issues, and failure to publish negative results [
In the studies reviewed in this paper, the outcome measures for acute pain were different in many of the studies conducted. The need to use multiple study sites to accommodate the number of subjects necessary for large clinical trials makes consistency of study treatments and outcome measures extremely important. For example, in the phase III inhaled NO RCT, there were institutional differences between study sites, such that two sites had significantly different outcomes than the others. In the poloxamer 188 studies, there were several methodologic differences between the phase II and III trials which may account for the disparity in efficacy results.
Beyond challenges inherent to clinical trials, there are some recurring themes in these studies. First, preclinical studies were surrogates, but not sufficiently good model systems, for actual pain episodes. In general, preclinical
At the heart of the matter, it remains an article of faith that sickle cell polymerization and microvascular occlusion is the actual cause of acute pain in SCD. There is currently no way to visualize vaso-occlusion to corroborate that sites of sickle erythrocyte microvascular occlusion correspond to locations where patients feel pain or to show that therapies that restore or maintain normal blood flow will relieve vaso-occlusion and pain. Intravital microscopy of various vascular beds in rodents has been used to study agents that block adhesion of sickle erythrocytes and leukocytes, including
Even if a therapy is effective in the person with SCD, there may be difficulty in maintaining effective pharmacokinetics, timing of therapy, and/or drug delivery to the site of vaso-occlusion. A treatment may work well at certain concentrations
As in stroke therapy with neuroprotectants and thrombolytics [
In the hemoglobin SAD mouse hypoxia studies, inhaled NO was only effective in preventing mortality when the animals were pretreated for 30 minutes prior to hypoxic exposure, which would not be feasible in people. Nitric oxide may be effective if therapy is begun at the very outset of vaso-occlusion
Lastly, it is likely that therapies which target specific components of sickle cell pathophysiology do not sufficiently inhibit the entire process that makes up acute pain episodes. Poloxamer 188 has the potential to inhibit multiple cell-cell interactions, but adhesion has not yet been demonstrated to be a major mechanism contributing to acute painful episodes. While the effects of senicapoc on sickle erythrocyte hydration and total hemoglobin concentration has been consistent between
Early reports discussing complexity in SCD date back to 1974 and 1983 [
To date, systems biology high-throughput and large dataset methodologies, or “omics” studies, of SCD have included transcriptome analysis of blood outgrowth endothelial cells, monocytes and reticulocytes [
Reductionism in science dates back to René Descartes in the 1600s [
Systems theory was first coined by Ludwig von Bertalanffy in the 1940s, defining it as the transdisciplinary study of the abstract organization of phenomena, independent of their substance, type, or spatial or temporal scale of existence. It investigates both the principles common to all complex entities and the (usually mathematical) models which can be used to describe them [
Donella Meadows defines a system as “a set of things interconnected in such a way that they produce their own pattern of behavior over time. The system may be buffeted, constricted, triggered, or driven by outside forces. But the system’s response to these forces is characteristic of itself, and that response is seldom simple in the real world… Our own bodies are magnificent examples of integrated, interconnected, self-maintaining complexity [
Systems biology is the application of systems thinking to the study of biologic processes. Kitano wrote that systems biology is a ““holistic” approach to interconnect different cellular processes, such as metabolism and genetic regulation, instead of traditional reductionist methods [
Network thinking was defined by Mitchell as “focusing on relationships between entities rather than the entities themselves [
These concepts of systems thinking and network analysis were used to develop diagrammatic models that incorporate and which show potential interactions between multiple mechanisms putatively involved in acute pain episodes (Figure
Acute pain model. This diagram shows proposed interactions between pathophysiologic mechanisms in sickle cell disease that lead to acute painful episodes. Key to mechanisms: green: balancing feedback loops, red: erythrocytes, orange: hemolysis, light blue: inflammation and oxidant stress, dark blue: ischemia and reperfusion, lavender: vasomotor, brown: coagulation, gray: angiogenesis, and light gray: pain modifiers. See text for details.
Chronic vasculopathy model. This diagram shows proposed interactions between pathophysiologic mechanisms in sickle cell disease that lead to chronic vasculopathy and organ injury. Key to mechanisms: Green arrows: balancing feedback loops, Red: abnormal vascular structure and function, orange: hemolysis, light blue: inflammation and oxidant stress, dark blue: ischemia and reperfusion, lavender: vasomotor, brown: coagulation, and gray: angiogenesis. See text for details.
Figure
Hemolysis can also contribute to vaso-occlusion, through the release of free heme, reactive iron species, and membrane microparticles. Free heme can bind NO and reduce its bioavailability, which promotes vaso-constriction, inflammation, and platelet aggregation. Heme and reactive iron species can directly cause injury and oxidative damage to endothelium. Erythrocyte membrane microparticles with exposed phosphatidylserine may active platelets and promote coagulation. Hemolytic anemia reduces oxygen delivery to tissues and contributes to reduced tissue and organ perfusion chronically, likely leaving them vulnerable to the effects of acute vaso-occlusion.
After vaso-occlusion has occurred, there is local tissue ischemia and reperfusion, which includes inflammatory and oxidant responses. Our group has demonstrated elevated levels of hypoxia-inducible factor- (HIF-) associated angiogenic growth factors in children with hemoglobin SS during steady state [
Pain is itself a complex process and is not diagrammed in detail in this model. The pain experience is the product of the nociceptive input from injured tissue but also involves cognitive, contextual, mood, and individual differences, such as sex, age, and genetics. Stress related to acute pain can induce neuro-endocrine responses, such as stress steroids, catecholamines, and pain peptides (substance P, neurokinins). Catecholamines promote vaso-constriction, and the pain peptides can be proinflammatory. The current model also does not attempt to delineate all of the external factors that may also influence a pain episode, such as environment (ambient temperature, second-hand smoke) or psychosocial factors. Neuropathic pain resulting from ischemic injury directly to nerves may account for some of the pain experienced in SCD and would not necessarily respond to antisickling therapies or those targeting inhibition of vaso-occlusion. Chronic pain may involve “imprinting” of the nervous system by epigenetic modifications (DNA methylation, histone modification) that regulate gene expression [
A model with similar features describes the development of chronic vasculopathy in SCD (Figure
Hemolysis will cause disequilibrium in favor of vasoconstriction, and longterm exposure can eventually cause sustained impairment in vasodilation and reduced vessel compliance (stiffness). Early stage remodeling may cause arterial wall stiffness and reduced compliance before arterial stenosis is apparent in imaging studies. This may be the case in those children with elevated transcranial doppler (TCD) velocities in the cerebral arteries that is associated with high stroke risk, but whose brain magnetic resonance arteriography (MRA) shows no arterial stenoses.
Anemia reduces oxygen delivery and organ perfusion. Ischemia and reperfusion due to the combination of anemia and repetitive vaso-occlusion stimulate HIF-associated angiogenic growth factors. Anemia-associated compensatory circulatory changes are likely to cause disturbed blood flow, which can promote endothelial injury, and potentially contribute to vessel wall remodeling (intimal proliferation and arterial stenosis). Our group has demonstrated elevated serum concentrations of platelet-derived growth factor- (PDGF-) AA, a mediator of vascular remodeling, and brain derived neurotrophic factor (BDNF), a biomarker of cerebral ischemia, in children with hemoglobin SS and high TCD velocities [
Considering sickle cell disease through the lenses of these models helps explain the potential limitations of therapies that are targeted to single pathways. While the therapies discussed earlier may have some multimodal effects, there are many additional contributors to acute pain that were not modulated by individual therapy. These models contrast conceptually to the bimodal model proposed by Kato, in which certain SCD complications are primarily related to either hemolysis or vaso-occlusion, and that individuals exhibit one or the other subphenotype [
Complexity in SCD does not necessarily mean that the disease is hopelessly complicated and cannot be successfully treated. In these models, sickle hemoglobin polymerization is the network hub, and therefore most vulnerable to attack (correction). This is consistent with the clinical observation that red cell transfusions and hydroxyurea, which correct hemoglobin S polymerization and/or sickled erythrocytes, are often effective SCD treatments and also reduce downstream mediators. However, while hydroxyurea is an alternative to transfusions for certain SCD indications, it is not an equivalent therapy. In the setting of established structural vascular disease, such as significant cerebral vasculopathy in individuals with stroke, hydroxyurea therapy in combination with phlebotomy to relieve iron overload was not as effective as transfusions in preventing stroke [
Viewed from this perspective, replacement of sickle erythrocytes by stem cell transplantation, gene therapy correction of the hemoglobin S mutation, or very effective fetal hemoglobin induction are likely to be the most effective SCD treatments in the long run. However, until these are widely available, should severe sickle cell disease be treated like thalassemia, with lifelong chronic red cell transfusions? How do we identify those at highest risk who would benefit from life-long transfusions begun early in life?
This paper of selected clinical trials and discussion of complexity in SCD has identified some challenges in the search for alternative effective therapies. There are logistical and methodological issues related to clinical trials, such as consistent study endpoint definition and effective timing and delivery of therapeutics; lack of good model systems in which to test the effect of therapeutics; and the larger problem posed by complexity—it is difficult to shut down all aspects of the system at once without using transfusions or stem cell transplantation. Acknowledgment of complexity does not imply that therapies targeting individual intermediary mechanisms in SCD should be abandoned but necessitates that their effectiveness needs to be tested in highly predictive model systems prior to embarking on large-scale clinical trials. In this section, some possible approaches to these challenges are suggested.
There is a need to develop preclinical model systems for complications such as acute pain or chronic vasculopathy that are highly predictive of those processes in people with SCD. There are currently transgenic sickle cell mouse models of acute chest syndrome and pulmonary hypertension, but none yet for SCD-associated stroke or acute painful crisis. The most extreme model used in transgenic sickle cell mice is hypoxia-induced death, which presumably occurs from sickle cell-induced vaso-occlusion in the entire animal. While not similar to human acute pain episodes in severity, drugs that are potent enough to prevent this degree of sickling or vaso-occlusion should be effective in less extreme situations. For example, the lack of beneficial effect of inhaled NO in the phase III RCT was accurately predicted by the lack of rescue in the hypoxia-exposed SAD mice who were treated with only posthypoxia inhaled NO. If there was a way to image sites of pain while simultaneously measuring pain behaviors in the sickle cell mouse, it might be useful to titrate the severity of the hypoxic exposure to a sublethal dose that might simulate acute painful events.
Transplantation of sickle cell mouse bone marrow or injection of sickle mouse erythrocytes into a transgenic strain with a desired gene knockout may be used to prove the essential role of an individual molecule in sickle cell pathogenesis. These approaches have been used to demonstrate the role of P-selectin in sickle cell microvascular occlusion [
One of the most basic tenets of the field is that acute pain episodes are caused by the occlusion of microvessels by sickled erythrocytes. While this makes theoretic sense on the basis of hemoglobin polymerization, rheologic and microvascular studies, this has never been proven in people with SCD. When does acute pain begin relative to vaso-occlusion? Do individuals with full body pain really have sickling everywhere? Does vaso-occlusion in one part of the body induce pain at distant sites through crosstalk between neurons in the sensory pathway or central nervous system? The field needs ways to visualize both sites of vaso-occlusion and pain pathways to demonstrate that they are actually related and to have an outcome measure for testing pain treatments. Vaso-occlusive sites could potentially be visualized by radionuclide-tagged particles that home to ischemic tissue markers, such as SDF-1
The pain research field apparently has similar challenges in developing chronic pain treatments as has been described here for SCD. Borsook wrote “Drug development for pain often fails, paralleling many other CNS areas, because preclinical and experimental clinical proof-of-concept (POC) studies do not translate well to clinical conditions and patient populations [
For agents that prevent pain, the number and duration of acute painful episodes, including both facility- and home-treated events, and quality of life measures related to pain may be a more accurate assessment of the true effectiveness of a pain prevention therapy. In the treatment of acute pain, there needs to be agreement on the definition of who should be hospitalized for therapy to assure comparability of groups and on the definition of what constitutes “resolution of pain.” It has also been recommended that study endpoints better match the mechanism of action of the therapeutic [
To better understand the role of therapeutic mechanisms, biomarkers or other functional outcome measures should be included as part of early phase clinical trials. As mentioned in the earlier discussion of inhaled NO, none of the published clinical trials reported on vasodilation, platelet aggregation, or inflammatory biomarkers, such as sVCAM. Senicapoc’s red cell effects were consistently measured in each clinical trial, so that lack of efficacy in reducing pain frequency appears to be unrelated to improvement in cell hydration status. In this situation, the data helps guide investigators in the analysis of therapeutic and unwanted side effects. Measurement of biomarkers makes clinical trials more labor intensive and is unlikely to be feasible at every study site. However, it provides important evidence to explain drug efficacy or lack thereof.
In combination with biomarker and functional assays, early phase clinical trials should include well-described pharmacokinetics to establish that the administered doses result in plasma concentrations that are comparable to those used in preclinical studies. However, systemic concentrations of a drug may not be the same as the amount delivered to the affected tissues, especially in areas of reduced perfusion. There is an opportunity to apply nanotechnology to selectively deliver analgesics, antisickling, other vaso-occlusion disrupting agents, or drugs that improve tissue oxygenation to ischemic regions (presumably corresponding to vaso-occlusion) by targeting ischemia markers, such as SDF-1
Systems biology and “omics” technologies may be useful in understanding complex biologic systems. This approach is in its infancy in SCD and could be applied to dissecting specific problems, such as identifying gene variants or microRNAs that predispose to higher risk of well-defined complications such as stroke, chronic pain, changes in gene expression, or epigenetic modifications that occur with therapies. Another goal for the application of “omics” to SCD would be to identify master regulators that control multiple pathophysiologic mechanisms, so that these could be targeted for inhibition.
Predictive computational models can potentially be used to integrate multiple types of patient data, such as laboratory values, radiographic findings, circulating biomarkers, and genetic and epigenetic data, with disease phenotype to define risk categories. The ability to identify individuals at highest risk for severe complications would allow the option of early treatment with high-risk therapies, currently red blood cell transfusion or stem cell transplantation, well before the onset of complications. The predictive strength of such modeling approaches would be enhanced by including as many individuals with SCD as possible, perhaps through a collaborative national data registry and biorepository system.
It has been well described that stresses related to poverty and racism affect cardiovascular risk and disease and disproportionately affect African Americans. It is very likely that such gene-environment interactions are additional factors influencing SCD complexity and outcomes and are currently not adequately understood. For example, how do chronic undernutrition, lack of utilities, poverty, or personal or familial mental illness affect the frequency and severity of acute illness or the response to therapy in a person with SCD? Epidemiologic methodologies including geocoding could be applied to studying some of these factors in SCD. Such variables can be added to computational predictive models to help us begin to understand the relative contributions of factors in SCD complications.
SCD is caused by a single mutation in beta globin but triggers several pathophysiologic pathways and results in a highly complex disease. This complexity is likely to be one of the major barriers to the development of successful new treatments which, to date, has largely concentrated on individual mechanistic pathways. Future development of therapeutics needs to continue to focus on correcting the underlying problem of sickle hemoglobin polymerization but should also include development of better model systems for acute and chronic SCD complications, methods for visualizing and measuring vaso-occlusion and associated pain, directed delivery of therapies to sites of vaso-occlusion, systems biology approaches to identify master regulators of the multiple downstream effectors of hemolysis and vaso-occlusion, and better understanding of the contribution of gene-environment interactions on sickle cell disease complications. Considering the number of pathophysiologic processes caused by SCD, it is astonishing how well the body maintains homeostasis sufficient for growth, development, and general health for periods between acute illnesses. The approach to this disease should also include an effort to identify mechanisms that are crucial to maintaining homeostasis and wellness. While there have been many life-saving advances in the treatment of SCD, much work remains to achieve the goal of curing the disease and developing safe and effective therapies to improve health and well-being.
The author has no financial interests in any of the commercial products mentioned.
The author receives research funding from the Atlanta Clinical Translational Science Institute (UL1 TR000454), Morehouse School of Medicine RCENTER (U54 RR026137), and Grants 5R01HL095647 and P20 MD006881. The author would like to thank her family, teachers, mentors, colleagues, and patients for teaching her much about the art and science of medicine.