Subphenotypes of Sickle Cell Disease: A Guide to Targeting Treatment?

The pathophysiology of sickle cell disease, a phenotype caused by several common and many rarer genotypes, flows “downstream” from the initiating event of sickle hemoglobin (HbS) polymerization [1]. Twenty years ago, building on prior observations, it was suggested that the features of sickle cell disease could be separated into some associated with sickle vasoocclusion (VOE) and some associated with chronic intravascular hemolysis (Figure 1) [2]. Although the acute VOEs that typify sickle cell disease are the major concern of most patients and often a treatment challenge for their physicians, chronic hemolysis is etiologically related to systemic vasculopathy that includes pulmonary, renal, and cerebral vascular disease, which can be silent clinically for many years but is ultimately responsible for morbidity and mortality.

FIGURE 1.

Subphenotypes of sickle cell disease. On the left panels are the common complications most closely associated with sickle vasoocclusion that have been associated with blood viscosity. Acute painful episodes and acute chest syndrome are the most common acute vasoocclusive episodes or VOE. On the right panels are the complications most closely associated with intravascular hemolysis as sickle erythrocytes insufficiently protected from HbS polymer damage lyse in the circulation. An overlap or combined subphenotype has features of both vasoocclusive and hemolytic complications. F‐cells are erythrocytes with HbF detectable by fluorescence activated cell sorting or FACS analysis. Patients with HbS/HbF genotypes have 5%–80% F‐cells. Some F‐cells are protected nearly totally from HbS polymer inflicted damage. If they are present in sufficient numbers, they protect from most disease complications. Other F‐cells and erythrocytes without detectable HbF lyse intravascularly leading to the disease complications associated with hemolysis. [Color figure can be viewed at wileyonlinelibrary.com]

Intravascular hemolysis releases heme and iron driving oxidant stress, nitric oxide depletion, endothelial dysfunction, and sterile inflammation. Nitric oxide activates soluble guanylate cyclase converting GTP to cGMP, relaxing vascular smooth muscle and causing vasodilatation. Along with activation of platelets and endothelium, products of hemolysis represent erythrocyte danger‐associated molecular patterns (eDAMPs), which promote and propagate sterile inflammatory and oxidative stress, further impairing the redox balance [3, 4].

The distinction between vasoocclusive and hemolytic features of sickle cell disease was based on analysis of clinical studies, the largest of which was from the pre‐hydroxyurea era [5]. In this issue of the Journal, Willen. et al., examined subphenotypes of sickle cell disease in a contemporary California administrative database of 7636 patients followed for a mean of 13–18 years, ascertaining a slew of complications using ICD codes. From these patients they extracted a sub‐cohort of 2210 younger individuals aged < 25 years. Using latent class analysis they found four classes of subphenotypes: (1) vasoocclusive (high VOE/acute chest syndrome/avascular necrosis of bone in 23% of males and 15% of females); (2) hemolytic (high chronic kidney disease/pulmonary hypertension/cerebrovascular disease in 8% of males and 13% of females); (3) overlap of both patterns in 8% of males and females; and (4) low complication rate in 62%–63% of all patients. In the younger cohort, classes were: (1) vasoocclusive, 8%; (2) hemolytic, 3%; and (3) acute chest syndrome‐dominant in 12%, and low complications in 78%. When compared with people with few complications, vasoocclusive, hemolytic, and overlap classes had elevated mortality (HR 1.3–2.4, p < 0.05). The younger patients with the hemolytic subphenotype had the worst survival (HR 7.8, p < 0.001). This is the largest clinical study to examine the concept of two major subphenotypes of sickle cell disease, and the results confirm generally the dichotomization of sickle cell disease phenotypes previously described, extending this by defining an overlap subphenotype with the poorest prognosis. An overlap subphenotype is not unexpected given that vasoocclusion and hemolysis have the same instigator. What is unusual is the very large proportion of patients with few complications. Does this really reflect “benign” disease, perhaps a result of better treatment like more widespread use of hydroxyurea and supportive measures like transfusion, or is this a result of the inability to dissect the several common genotypes of sickle cell disease? Patients with intrinsically “milder” compound heterozygous genotypes of sickle cell disease like HbSC disease and HbS‐β⁺ thalassemia have about half the rate of complications as people with the clinically significant HbS/HbF genotypes of sickle cell disease that include homozygosity for the HbS gene (sickle cell anemia) and HbS‐β⁰ thalassemia where only HbS and HbF are present (with a small amount of HbA₂). Although complications of HbSC disease and HbS‐β⁺ thalassemia can be as severe as the HbS/HbF genotypes, on average, they tend to happen later in life, some like stroke are uncommon, and hemolysis is not as severe. Although the authors attempt to deal with this issue by indirectly defining the more severe HbS/HbF genotypes, lumping all sickle cell disease genotypes could account for the stunningly high rate of “benign” disease that most clinicians would be surprised to see, especially in hospitalized patients.

Does this classification of the pathophysiologic features of sickle cell disease have clinical relevance or is it just a nosologic nicety, though one that provides some pathophysiologic intuitions? More than 30 years ago it was noted that patients with sickle cell anemia‐α thalassemia, because of the relative reduction in α‐globin chain synthesis, had a lower mean corpuscular hemoglobin, improved cell hydration, a lower polymerization tendency of HbS—a process dependent on the 50th power of HbS concentration—and that hemolysis had a differential effect on disease complications [1, 6]. α Thalassemia in sickle cell anemia is usually associated with a reduced incidence of stroke, priapism, and leg ulcers but an increased rate of acute painful episodes, acute chest syndrome, and osteonecrosis [7].

A trial of the agent senicapoc, a Gardos channel inhibitor, successfully rehydrated sickle erythrocytes, reduced hemolysis, and improved hemoglobin concentration. The overall incidence of VOE was unaffected; individuals randomized to senicapoc who were not simultaneously taking hydroxyurea had a statistically significant increase in VOE compared with placebo treated controls although an ad hoc reanalysis of the trial data suggested that responders whose hemoglobin was increased did not have more VOE [8, 9]. This drug was not further developed. However, these results raised some concerns about drugs that target hemolysis through mechanisms other than erythrocyte rehydration. The first of these new agents studied in clinical trials, voxelotor, increased hemoglobin‐O₂ affinity (reduced P₅₀) by reversible binding to the amino‐terminus of α globin, reducing the polymerization tendency of HbS, preventing RBC damage, thereby lessening hemolysis and increasing hemoglobin concentration to levels similar to those achieved with senicapoc [10]. There was no statistically significant change in the annualized rate of VOE, although the study design might have made any change difficult to discern. One post hoc analysis suggested that there was a fall in the rate of VOE in patients who responded best in terms of their increment in total hemoglobin. A third class of agents that can reduce hemolysis in sickle cell disease are activators of erythrocyte pyruvate kinase (PKR). These agents also increase hemoglobin‐O₂ affinity but through a mechanism different from that of voxelotor. By increasing glycolytic flux, PKR agonists reduce 2,3 BPG levels thereby increasing hemoglobin‐O₂ affinity and reducing HbS polymerization; they also increase RBC ATP. Two PKR agonists, mitapivat and etavopivat, are in Phase 3 trials. Both increase hemoglobin as much or more than senicapoc and voxelotor. Preliminary data (Phase 3 study results for either PKR agonist have not yet been published in peer reviewed journals) show some differences. Mitapivat treated patients had an annualized VOE rate of 2.29–3.01 that was not statistically different from that of placebo treated subjects that was 2.57–3.64 [11]. A placebo‐controlled Phase 2 study of etavopivat treated patients, most of whom were also on hydroxyurea, showed a ~46% reduction in VOE with nearly a doubling in time to a first VOE after beginning treatment. However, Phase 3 study results reported by the sponsor claimed a 27% VOE reduction [12]. While these three agents decrease HbS polymerization, improve the circulatory competence of the sickle RBC increasing hemoglobin concentration, why is there not a reproducible reduction in VOE? Hydroxyurea, whose main mechanism of action is induction of increased levels of HbF increases hemoglobin level, usually about 0.5–1 g/dL but is associated with a statistically and clinically significant reduction in VOE. Forty to sixty percent of voxelotor and PKR agonist treated patients responded with a hemoglobin increase ≥ 1 g/dL. Unlike senicapoc, voxelotor, and PKR agonists, hydroxyurea directly inhibits HbS polymerization by inducing HbF rather than indirectly decreasing polymerization by reducing the concentration of HbS or increasing hemoglobin‐O₂ affinity. More dramatically, gene therapy can generate > 40% HbF in nearly all sickle cells returning hemoglobin levels to normal while practically eliminating acute VOE [13]. The major lesson from these studies is that therapeutics whose main effect is to reduce hemolysis and increase hemoglobin levels might not always have a major effect on sickle VOE. Over time such agents might reduce the chronic organ damage that is a hallmark of hemolytic anemia (Figure 1). Short term 52‐week studies, the standard in the field, are unlikely to detect improvement or stabilization of the hemolysis‐related complications like pulmonary hypertension and nephropathy. Perhaps if hemolysis‐focused agents had a more potent effect on HbS polymerization indirect polymerization inhibition might lead to a consistent and clinically significant reduction in VOE, but if P₅₀ was decreased enough, might this be at the expense of impaired O₂ transport?

How might codification of sickle cell disease subphenotypes affect treatment? Complications of hemolysis are less common in HbSC disease and HbS‐β⁺ thalassemia, so the use of agents focused on the hemolytic subphenotype is less urgent. Some patients with HbS/HbF genotypes have chronic severe anemia. Voxelotor increased hemoglobin ≥ 5 g/dL in one exceptional patient with severe anemia [13]. In the controlled clinical trial, voxelotor increased hemoglobin ~3–4 g/dL in ~10% of patients [10]. Individuals with this remarkable hemoglobin response might be treated solely with an agent that primarily reduces hemolysis, especially if they responded poorly to hydroxyurea. Unfortunately, if approved by the FDA, PKR activators will be the only class of agents available that reduce Hb‐O₂ affinity because voxelotor has been withdrawn from the market. A Phase 2/3 study of osivelotor that has the same mechanism of action as voxelotor, but improved pharmacokinetics is currently enrolling patients (NCT05431088). Most people treated with voxelotor did not have hemoglobin increases of 3–5 g/dL and the full results of the PKR agonist studies have not been published so we do not know if there are exceptional responders. Based on the voxelotor trials and the early studies of PKR agonists, it is unlikely that monotherapy with drugs whose major effect is on hemolysis will be sufficient for most patients. Nevertheless, PKR agonists should be useful adjuncts to HbF‐inducing drugs as most hydroxyurea treated adults remain anemic and have continued VOE. In addition, reduction of hemolysis is likely to be beneficial over time [14]. Only a single drug is targeted at preventing sickle vasoocclusion. This agent, crizanlizumab, a P‐selectin blocker that inhibits adhesion of leukocytes and platelets to the endothelium, was approved because it reduced VOEs [15]. A follow‐up trial failed to replicate the first results and the drug was withdrawn from the European market [16]. Its current use in the US is probably quite low.

Subphenotypes are clinically useful because they provide a framework for mechanism‐based therapy, but a treatment that primarily reduces hemolysis should not be expected to reproduce the clinical benefits of those that directly suppress HbS polymerization. Defining an individual's dominant subphenotype might permit rational combinations of therapies that target polymerization, hemolysis, or both. Patients with the most severe subphenotypes, regardless of their sickle cell disease genotype, might be the best candidates for the currently available gene therapies that can provide “functional cures” [17].

Funding

The author has nothing to report.

Ethics Statement

The author has nothing to report.

Consent

The author has nothing to report.

Conflicts of Interest

M.H.S. serves on the Steering Committee for the Vertex gene therapy trials in sickle cell disease, is a member of the DMC for Cellarity, and has consulted for Fulcrum, Beam, Editas, and MiNA Therapeutics.

Data Availability Statement

Data sharing not applicable to this article as no datasets were generated or analyzed during the current study.