Measuring success: utility of biomarkers in sickle cell disease clinical trials and care

Biomarkers have been defined as characteristics that are objectively measured and evaluated as indicators of normal biologic processes, pathological states, or pharmacologic responses to a therapeutic intervention.¹  As such, biomarkers can be useful at multiple stages of disease processes, from the detection of risk before a disease or a complication has developed to the screening, diagnosis, and monitoring of the course of an established disease. In this latter setting, biomarkers can predict outcomes (eg, death) and response to treatments. Beyond clinical use, biomarkers are also useful in clinical research where they help in the determination of dose response, provide surrogate outcomes for intervention trials, and help unravel mechanistic pathways. While there are many classifications of biomarkers based on their characteristics and use, a pragmatic classification is based on their method of detection, which may rely on imaging vs molecular or biochemical testing of a body fluid or an explanted tissue or organ (Tables 1 and 2). There has been enormous progress in the development and clinical applications of biomarkers in many conditions such as cancer and neurological diseases; however, biomarker development has been limited in sickle cell disease (SCD), and very few biomarkers are currently employed in the routine care of patients. In the following text, we will review the established and recently developed biomarkers in SCD.

SCD defines a group of syndromes caused by the inheritance of a mutated β chain of hemoglobin (hemoglobin S [HbS]) either as a homozygous allele (HbSS) or in combination with another β-globin abnormality. The SCD syndromes have phenotypes of varying severity; HbSS and HbSβ⁰ thalassemia (together called sickle cell anemia [SCA]) are characterized by a severe phenotype, while HbSC and HbSβ⁺ thalassemia have a generally milder course. Mutated HbS polymerizes under low-oxygen states and leads to the “sickling” deformation of the red blood cells (RBCs). Sickling of the RBC is the fundamental primary lesion of SCD and results in multiple downstream rheological and cellular abnormalities that, as a whole, lead to vaso-occlusion.²  Vaso-occlusion is a reversible, unpredictable occlusion of specific vascular beds and is responsible for acute and chronic ischemia and organ failure. Acute, painful vaso-occlusive events (vaso-occlusive crises [VOCs]) affecting the musculoskeletal system are a hallmark feature of SCD, but their risk factors and pathophysiology remain elusive.³  Murine models of SCD have dissected the complex processes that mediate vaso-occlusion in the microcirculature,⁴  but the relative contribution of each process to clinical VOCs is unclear. Hemolysis and endothelial dysfunction, sterile inflammation, hemostatic activation, oxidant stress, blood hyperviscosity, and cellular hyperadhesion (Figure 1) have all been described as necessary components of vaso-occlusion and individually may be the predominant mechanism in certain vascular beds or in response to specific triggers. These processes are not unique to SCD, yet no other disease presents with a syndrome resembling the VOC of SCD, underscoring the primacy of upstream pathognomonic RBC sickling in its pathogenesis. The phenotype of patients with SCD is extremely variable, ranging from a mildly symptomatic course characterized by rare acute events and hospitalizations to a rapidly progressive and invalidating disease with median mortality in the fourth decade of life.⁵  The phenotypic heterogeneity of SCD is the result of genetic and epigenetic factors that modulate the pathogenic processes of SCD, from the degree of HbS polymerization to the intensity and the relative contribution of the interlinked pathways (Figure 1). Thus, it is possible for a patient with a high baseline rate of hemolysis, severe anemia, and endothelial dysfunction to experience frequent complications linked to hemolysis but rare VOCs and other complications linked to viscosity. On the other end of the spectrum, HbS polymerization may be more limited and hyperviscosity from a relatively high hematocrit may be present and lead to frequent VOCs. These 2 major, distinct subphenotypes of SCD have been identified in epidemiological studies,⁶  but they do not fully account for the variability of the SCD phenotype. Biomarkers, therefore, may be critical in identifying the relative contribution of each pathway to vaso-occlusion and, in turn, the probability of an individual patient to develop a specific set of complications. While an in-depth description of each pathogenic process in SCD is beyond the scope of this chapter, we will discuss its importance in relation to biomarker development in SCD (Table 1).

Sickle RBC have a decreased lifespan of 15 days⁷  due to both intra- and extravascular hemolysis in SCD, resulting in a chronic hemolytic anemia with acute flares. Both chronic and acute anemia are responsible for pathology by reducing the functional capacity of highly metabolically active organs (eg, the developing brain in children)⁸  and inducing compensatory, partially maladaptive changes such as cardiac hypertrophy and diastolic dysfunction.⁹  Thus, anemia per se is a biomarker of important complications and overall mortality, as discussed later. Compounding the detrimental effects of anemia is the pathology deriving from the byproducts of hemolysis (ie, free plasma hemoglobin and its redox products, free heme and heme iron, and RBC-derived arginase and adenosine diphosphate). Both free plasma hemoglobin and its byproducts lead to endothelial dysfunction by inhibiting nitric oxide metabolism¹⁰  and act as erythrocyte damage-associated molecular proteins (DAMPS) to promote sterile inflammation.¹¹  Furthermore, free hemoglobin and free adenosine diphosphate released by RBCs activate platelets and may contribute to thrombosis in SCD.^12-14  Finally, erythrocytic arginase released by hemolysis also contributes to nitric oxide depletion by limiting the bioavailability of arginine, a biomarker linked to pulmonary hypertension and decreased survival.¹⁵  The gold-standard nuclear medicine methods of RBC labeling to determine hemolysis are no longer widely available because of cost and unfavorable logistics, and the identification of dense and irreversibly sickled cells both by sophisticated methods (ie, ektacytometry and flow cytometry) or the peripheral smear examination only carry historical value, as they have not been conclusively associated with disease states. Among the clinically available markers of hemolysis, haptoglobin has been unhelpful, as it is constantly saturated by free hemoglobin released in the setting of chronic hemolysis and is either decreased or undetectable at baseline. However, other markers of hemolysis, such as lactate dehydrogenase (LDH), bilirubin, and reticulocyte count, either alone or in combination (ie, hemolytic index) have been useful to identify a subset of patients at high risk of hemolysis-related complications, including pulmonary hypertension, leg ulcers, priapism, and chronic kidney disease.¹⁶  In particular, the hemolytic index, a principal-component analysis product of aspartate transaminase, bilirubin, and LDH has been particularly useful, as it circumvents the poor specificity of each marker of hemolysis when assessed individually and has been found to predict increased risk of death at 2 years (hazard ratio [HR], 3.44; 95% confidence interval [CI], 1.2-9.5; P = .02).¹⁷  One cautionary tale on the use of markers of hemolysis for clinical outcomes in SCD stems from the senicapoc clinical trial. Senicapoc, a Gardos channel blocker, inhibited hemolysis and improved hemoglobin yet did not prevent vaso-occlusive complications,¹⁸  underscoring the importance of the other SCD mediators in vaso-occlusive complications and the need to adopt a pleiotropic preventive approach to VOCs that targets multiple pathways.

Tissue- and cell-derived DAMPs are a heterogeneous group of compounds that lead to the activation of inflammasome complexes and the promotion of sterile inflammation in SCD. While the notion of chronic inflammation as a component of SCD is not new (eg, erythrocyte sedimentation rate and C-reactive protein levels have historically been linked to the evolution of pain crises in SCD¹⁹ ), new insight into inflammasome pathways has emphasized their role in the pathogenesis of SCD. Multiple molecules produced during hemolysis, oxidant stress, and vaso-occlusion, including heme, chromatin-binding protein high mobility group box 1 (HMGB1), adenosine triphosphate, and cardiolipin act as DAMPs to activate innate immune responses by binding to the immune cell receptor toll-like receptor 4 (TLR4) and signaling via mitogen-activated protein kinase (MAPK) and nuclear factor κ-light-chain-enhancer of activated B cells (nuclear factor κB).^20-22  Recent studies have detected high levels of components of inflammatory pathways in patients with SCD as compared with controls. For instance, HMGB1 was shown to be significantly elevated in the plasma of SCD patients, and the levels further increased following VOCs.²³  The elevated levels of HMGB1 were also shown to promote TLR4 activity in plasma of patients with SCD.²³  NOD-like receptor family, pyrin domain containing 3 (NLRP3) is one of the most-studied components of inflammasome complexes; NLRP3 recognizes DAMPs and leads to the cleavage of procaspase-1, which is required for activation of inflammatory cytokines. Levels of NLRP3²⁴  and activated cytokines interleukin-1β (IL-1β), IL-6, and IL-8 were significantly elevated in peripheral blood mononuclear cells of patients with SCD.²⁵  While the role of components of inflammasomes as biomarkers is not known, these molecules could become therapeutic targets. Among the clinically available inflammatory markers, high-sensitivity C-reactive protein has been independently associated with VOCs in children with SCD, predicting a 6.6% increase in frequency of pain crises for every 1-mg/L increase in high-sensitivity C-reactive protein.²⁶ 

Multiple lines of evidence have shown that hemostatic activation is present at baseline in SCD and worsens in VOCs. Increased thrombin generation, abnormal activation of fibrinolysis, decreased levels of anticoagulant proteins, activation of platelets, increased levels of endothelial cells with tissue factor phenotype and tissue factor procoagulant activity, and externalization of RBC phospholipids have all been reported in SCD.²⁷  Thrombophilia has also been amply documented in SCD²⁸  and affects both the venous and arterial circulation; large vessel stenosis with thrombosis is present in children with stroke,²⁹  old and new thrombi have been documented in autopsies of patients with pulmonary disease,³⁰  and the incidence and prevalence of pulmonary embolism are increased.³¹  However, the link between biomarkers of hemostatic activation and thrombotic complications has not been fully established. Elevated levels of von Willebrand factor (VWF)³²  and platelet activation¹⁴  have been associated with a higher rate of hemolysis and elevated levels of the platelet α granule protein thrombospondin 1³³  and undetectable activity of the metalloproteinase that cleaves large VWF multimers (ADAM metallopeptidase with thrombospondin type 1 motif 13) with acute chest syndrome (ACS),³⁴  but causality between hemostatic alterations and complications has not been determined. Conventional biomarkers of pulmonary embolism, including d-dimer and presence of thrombosis in the lower extremities by Doppler, have been unhelpful in predicting pulmonary thromboembolism in SCD.³⁵  Elevated d-dimer levels have, however, been independently associated with an increased prevalence of stroke.³⁶  There are many published studies of anticoagulants and antiplatelet agents in SCD, but most have not correlated clinical end points with markers of hemostatic and platelet activation, have low-level evidence, and are plagued by small sample sizes. The largest study to date on the effects of antiplatelet therapy in SCD has shown decreased platelet reactivity with the thienopyridine prasugrel but no effect on VOCs.³⁷  Controlled studies to determine the safety and efficacy of new anticoagulants in specific SCD-related complications are underway, and antiplatelet agents may have a therapeutic role as part of a multidrug approach or in specific subpopulations.

Increased oxidant stress has long been appreciated in SCD³⁸  and promotes disease indirectly by peroxidation of substrates or via modulation of intracellular signaling pathways (eg, nuclear factor κB). Ischemia-reperfusion injury, the process of restoration of circulation after vascular occlusion, is a critical step in the generation of reactive oxygen species (ROS), predominantly via xanthine oxidase generation of superoxide radicals.³⁹  It is likely that ischemia-reperfusion occurs incessantly at a basal level in SCD in accordance with the reversible nature of RBC sickling and that it may be worsened by acute events. Other important sources of ROS include uncoupling of endothelial nitric oxide synthase, free hemoglobin reaction with hydrogen peroxide (Fenton reaction),¹⁰  and cellular sources, including RBCs⁴⁰  and platelet mitochondrial ROS generation.¹³  Among the biomarkers of redox status, oxidized glutathione (GSSH) has been the most widely studied in SCD. The ratio of reduced glutathione (GSH), an antioxidant, and GSSH is decreased as a result of oxidant stress in patients with SCD.⁴¹  Additionally, glutamine, which is essential in the generation of reduced NAD phosphate, the enzyme that reduces GSSH, is also depleted. Reduction of both GSH and glutathione have been associated with worse rate of hemolysis and pulmonary hypertension in SCD.⁴²  While antioxidant therapies have generally held promise in SCD, only l-glutamine supplementation has been proven beneficial in reducing VOCs in a phase 3 clinical trial,⁴³  a finding that has led to the US Food and Drug Administration approval of l-glutamine (Endari) for the prevention of acute complications of SCD such as VOC and ACS.

Pioneering research dating back to the 1970s has shown that blood viscosity is increased in patients with HbSS as compared with HbAA controls at similar hematocrit levels⁴⁴  and rises further when deoxygenated HbS polymerizes and the proportion of sickled, dense, dehydrated RBC increases.⁴⁵  The role of deoxygenated hemoglobin on blood viscosity has been recently confirmed by studies of the rheological effects of the small-molecule hemoglobin modifier GBT440 (voxelotor), which increases hemoglobin oxygen affinity, thereby preventing HbS polymerization and sickling; the effects of GBT440 were accompanied by a reduction in blood viscosity.⁴⁶  Under homeostatic conditions, a low hematocrit opposes the effect of polymerized HbS and prevents patients with HbSS from developing symptoms of hyperviscosity. Blood viscosity is not routinely measured in SCD, but its importance is underscored by several clinical observations. First, hypertransfusion has been associated with neurological symptoms in SCD.⁴⁷  Second, patients with HbSC disease or HbSS in combination with α-thalassemia also have a high rate of VOCs,^48,49  despite lower degrees of anemia and hemolysis.⁵  In these patients, hyperviscosity has been implicated as the predominant pathway to vaso-occlusion. The observation that phlebotomy improves headaches and the frequency of pain crises in patients with HbSC lends support to this hypothesis.⁵⁰  Third, while the development of ocular microangiopathy is linked with the rate of hemolysis in patients with HbSS, this complication has a higher incidence in patients with HbSC who have a lower hemolysis rate.⁵¹  In this latter group, ocular microangiopathy is not linked to the rate of hemolysis,⁵²  suggesting that increased blood viscosity may be its primary underlying cause.

The notion that increased cellular adhesion is a pathogenic process in SCD has evolved, over the decades, from an ancillary hypothesis to RBC sickling to its recognition as a critical component of vaso-occlusion being successfully harnessed for therapeutic strategies. Hyperadhesion is the result of the following major mechanisms: (1) membrane damage from HbS polymerization leading to externalization of phosphatidylserine and other adhesion molecules and microparticle shedding, (2) hemolysis-induced platelet activation and neutrophil adhesion (from heme-induced endothelial activation), (3) anemia-induced stress reticulocytosis with release of erythroid precursors rich in adhesion molecules such as CD36, and (4) inflammation-induced overexpression of adhesion molecules and cellular activation. The complex interactions between plasma and cellular adhesion molecules have been extensively reviewed elsewhere⁵³  and offer multiple potential therapeutic targets. However, clinical studies have shown that some molecules are likely to be central or upstream to vaso-occlusion, while others may represent epiphenomena with only marginal clinical relevance. For instance, genetic absence of CD36 did not protect against vaso-occlusion in humans.⁵⁴  On the contrary, targeting of P-selectin⁵⁵  and E-selectin⁵⁶  with monoclonal antibodies has led to reduction of VOC and represents a major breakthrough in SCD therapy, suggesting that molecules implicated in the early processes of neutrophil adhesion to the endothelium and rolling are paramount to vaso-occlusion. The role of soluble adhesion molecules as biomarkers has not yet been established but in vivo assessment of overexpression or activation of adhesion molecules holds promise to predict response to anti-adhesive strategies. Reduced phosphatidylserine⁵⁷  and integrin α4β1 (VLA4)⁵⁸  expression on RBC and reduced white blood cell count have all been associated with hydroxyurea therapy and protection from VOC, and markers of adhesion resulting from endothelial activation (eg, soluble E-selectin and P-selectin), and leukocyte activation (eg, macrophage-1 antigen) have decreased in response to blockade with the pan-selectin antagonist GMI1070.⁵⁹ 

The following paragraphs examine the role of biomarkers in the major acute and chronic complications of SCD (Table 2).

Pain crises or VOCs are the most common clinical manifestation of SCD. VOC is also the most common cause of hospitalization, posing a major health burden in this population. In addition, frequent painful episodes and chronic pain significantly decrease the quality of life (QOL) in patients with SCD. Historically, it has been shown that advancing age, high hemoglobin, low fetal hemoglobin (HbF)⁶⁰  and coinheritance of α-thalassemia⁶¹  are associated with increased number of VOCs in children and adults with SCD. Among the experimental biomarkers, elevated plasma thrombospondin 1, a secretory product of activated platelets that promotes sickle vascular adhesion, plasma levels^33,62  and imbalance of apelin (vasodilator) and endothelin 1 (vasoconstrictor)⁶³  are associated with a vaso-occlusive phenotype. Clinical management of SCD and clinical trials evaluating the success of various drugs on VOCs are mainly based on patient-reported pain history. There is no reliable marker or physical examination finding to objectively diagnose and measure pain in SCD, with the exception of the finding of dactylitis in infants and young children, significantly hampering the progress of this field until recently. The pharmacotherapy clinical trials targeting a variety of biomarkers related to the pathogenic pathways of SCD and aimed at minimizing or preventing VOCs are extensively reviewed elsewhere⁶⁴  and will be summarized herein. Many phase 3 multicenter and multinational studies (such as the senicapoc trial¹⁸ ; the MAGiC trial of magnesium sulfate as an analgesic, antiinflammatory agent, and vasodilator⁶⁵ ; and the DOVE study of prasugrel³⁷ ) failed to show improvement in the rate and management of VOC, despite mounting preclinical and clinical evidence on the role of these agents in the pathogenesis of VOCs. Thus, hydroxyurea has remained the only available drug to decrease VOCs in children and adult patients until recently. Fortunately, 3 major recent clinical trials utilizing antioxidants (l-glutamine and ω-3 fatty acids) and selectin inhibitors (crizanlizumab and rivipansel, see the Cellular hyperadhesion section) have shown promising results in decreasing VOCs and their related hospitalizations in children and adults with SCD. The main advantage of antioxidants is that they are orally administered, well tolerated, and without major side effects. Besides l-glutamine (see the Oxidant stress section), supplementation with ω-3 fatty acids that also have antiadhesive, antiinflammatory, antiplatelet, and antithrombotic activities significantly decreased VOCs and related school absenteeism compared with placebo in a randomized double-blind, placebo-controlled, single-center study.⁶⁶  Two new phase 3 multicenter clinical trials of ω-3 fatty acids supplementation in SCD are underway and expected to start enrollment soon (www.clinicaltrials.gov identifiers NCT02604368 and NCT02525107). Vitamin D deficiency has been implicated as a risk factor for both acute and chronic pain in patients with SCD^67,68 ; it is positively correlated with biomarkers of hemolysis⁶⁹  and associated with increased levels of proinflammatory cytokines in patients with SCD.⁷⁰  Vitamin D supplementation was associated with decreased daily pain in one small study of pediatric and adult patients with SCD and chronic pain.⁶⁸  However, the prevalence of vitamin D deficiency is very high (56% to 94%) in individuals with SCD,⁷¹  which limits its role as a useful biomarker for all but possibly those with chronic pain. In a single-center, randomized, triple-blind, placebo-controlled trial, inhaled mometasone was associated with reduction in soluble vascular cell adhesion molecule (_SVCAM) that correlated with reduction in daily pain scores in nonasthmatic patients with SCD.⁷²  Finally, initial results of a phase 2a study of once-daily oral GBT440 (now voxelotor) showed improvement in hemoglobin and reduction in total daily symptoms (pain) in adolescents and adults with SCD by increasing hemoglobin oxygen affinity and decreasing HbS polymerization,⁷³  and a phase 3 study is currently enrolling patients (www. clinicaltrials.gov identifier NCT03036813). The recent flurry of novel therapies is likely to herald a new age of precision medicine in SCD where individual patients may be directed toward specific therapies based on molecular biomarker profiles.

ACS is a major cause of morbidity and mortality in children and adults with SCD.^74,75  It is associated with prolonged hospitalization, intensive care unit care, the need for mechanical ventilation, increased transfusion requirements, and multiorgan failure. In addition, recurrent episodes of ACS can lead to the development of debilitating chronic lung disease.⁷⁶  Several risk factors such as young age, low HbF, high baseline hemoglobin, steady-state leukocytosis,⁷⁴  asthma/airway hyperresponsiveness,⁷⁷  and aeroallergen sensitization⁷⁸  have been associated with the development ACS. ACS is an ideal target for biomarker development, as two-thirds of ACS episodes occur in patients who are hospitalized for other reasons (primarily VOCs). Unfortunately, secretory phospholipase A2 (_SPLA2), a potent inflammatory mediator, is the only extensively studied laboratory biomarker in ACS thus far.⁷⁹ _SPLA2 was found to be elevated in the majority of patients with ACS, particularly 24 to 48 hours before the onset of symptoms.⁷⁹  In a small study of children with SCD who were hospitalized for VOCs, blood transfusion was shown to prevent ACS in those with elevated _SPLA2.⁸⁰  However, the positive predictive value of _SPLA2 for the development of ACS was only 24% in a large cohort of hospitalized patients with VOC, a finding that limits its clinical utility.⁸¹  Acute-phase reactants such as C-reactive protein,⁸²  pentraxin-3,⁸³  platelet-derived proteins (CD40L and TSP-1),^33,84  and circulating plasma exosomes⁸⁵  were shown to be potential markers of susceptibility to ACS, but none have been translated into clinical practice yet. There exists an urgent need for biomarker development for ACS to prevent this potentially fatal complication.

MOFS is a catastrophic complication that typically follows a severe VOC.⁸⁶  It is characterized by sudden onset of fever, altered sensorium, a rapid drop in hemoglobin levels and platelet counts, and the development of respiratory (eg, severe ACS and acute respiratory distress syndrome), hepatic, and/or renal failure. Interestingly, most patients have mild symptomatic SCD without organ failure prior to the development of MOFS, which complicates identification of at-risk patients. Aggressive blood transfusion may reverse MOFS early in its course; however, hemodynamic instability heralds high mortality at any time during the illness.⁸⁷  Development of biomarkers that are specific to MOFS is severely hampered, as many patients have clinical, laboratory, and autopsy characteristics that overlap with other syndromes, such as fat embolism and thrombotic microangiopathy.

Central nervous system (CNS) injury arguably represents the most feared and debilitating complication of SCD and is particularly common in patients with HbSS. The major CNS complications of SCD include stroke, silent cerebral infarct (SCI), and neurocognitive dysfunction.

Stroke is one of the most devastating CNS complications of SCD. Historically, 10% of children developed stroke before the age of 20 years, with a peak incidence between age 2 to 9 years.⁶⁰  Chronic blood transfusion remains the standard of care for individuals with SCA who develop stroke. This practice significantly decreases the risk of secondary stroke but does not eliminate it completely. To date, no clinically significant biomarkers have been identified to predict the risk of secondary stroke, while systolic hypertension, presence of SCI (ie, a magnetic resonance imaging [MRI] lesion in the absence of symptoms of stroke), low baseline hemoglobin and HbF, leukocytosis, and absence of α-thalassemia have all been implicated as risk factors in the development of primary overt stroke.⁸⁸  The combined effects of severe anemia and hemodynamically significant cerebral large vessel stenosis have been harnessed to develop the most important clinical biomarker of stroke in pediatric patients with SCD; prevention of primary stroke by identifying an at risk population by transcranial Doppler ultrasound (TCD) and initiation of chronic blood transfusion is a major success story of biomarker development in SCD. Because of TCD screening and prophylactic transfusions, the incidence of first stroke has dropped dramatically (by 92%), as demonstrated in The Stroke Prevention Trial in Sickle Cell Anemia (STOP).⁸⁹  One limitation of this strategy, however, is that many children who would not go on to develop clinical stroke despite having abnormal TCD still receive chronic transfusions. Thus, refinement of stroke prediction with additional biomarkers is urgently needed. The follow-up STOP 2 trial showed that chronic blood transfusions need to be continued indefinitely to prevent stroke in individuals who had abnormal TCD.⁹⁰  As a result of clinical guidelines based on STOP and STOP 2, long-term complications of transfusional therapy, including iron overload and RBC alloimmunization, are on the rise. More recently, the TCD With Transfusions Changing to Hydroxyurea (TWiTCH) trial offered an alternative to transfusion for children at risk by showing that hydroxyurea with phlebotomy is noninferior to chronic blood transfusion in preventing primary stroke in those with abnormal TCD, a history of blood transfusions (at least 1 year), and no severe cerebral vasculopathy.⁹¹  While elevated TCD presumably detects vasculopathy once it has already developed, a few retrospective studies have shown that reticulocytosis and worse anemia in early infancy are associated with development of abnormal TCD and cerebrovascular disease in children with SCA,^92,93  offering hope of even earlier prevention. These findings, however, need to be confirmed in larger prospective studies.

SCI occurs in ∼33% of children with SCA,⁹⁴  particularly at younger age. SCI is defined as an infarct-like lesion, an MRI signal abnormality that is ≥3 mm in 1 dimension and visible in 2 planes on fluid-attenuated inversion recovery T₂-weighted images with normal neurological examination findings.⁹⁵  The presence of SCI is associated with significant neurological morbidities such as increased risk of overt stroke, new or progressive SCI, cognitive impairment, and poor academic achievement.^96-98  In the Silent Cerebral Infarct clinical trial (SIT), regular blood transfusion has been shown to decrease the recurrence of new infarcts in children with SCI as compared with standard care.⁹⁵  There is an urgent need for the development of biomarkers that can predict the development of SCI and its recurrence, and that may obviate for the need of an MRI in younger children, as this test often requires sedation. In the SIT trial, worse anemia and high relative systolic blood pressure were found to be independent risk factors for SCI in children with SCA.^95,99  These findings need to be confirmed in a larger prospective trial to identify children with SCA at risk of SCI.

Among the CNS complications, neurocognitive dysfunction is the most prevalent but the least studied.^60,100  Neurocognitive dysfunction, particularly in the functional domains of executive function (processing speed, memory, attention and concentration, visual processing, and visual-motor integration), can also occur in the absence of overt stroke or SCI as reported in The Cooperative Study of Sickle Cell Disease (CSSCD).⁹⁸  In addition, a recent study of adults demonstrated significant neurocognitive dysfunction even in otherwise neurologically intact participants with seemingly uncomplicated HbSS disease.¹⁰¹  The cerebrovascular pathology responsible for neurocognitive dysfunction in SCD is largely unknown. Neurocognitive deficits are seen in a significant number of children and adults with SCD who have had no evidence of overt stroke and/or SCI, and risk factors in these cases are not well understood.¹⁰²  Besides the age-related decline in neurocognitive function, worse anemia has been shown to be associated with lower executive functioning and performance IQ scores in children⁶⁰  and adults,¹⁰¹  respectively, and severe SCD genotype (HbSS or HbSβ⁰ thalassemia) is associated with slower psychomotor speed in adults.¹⁰³ 

The finding that elevated tricuspid regurgitant jet velocity (TRV), an echocardiographic biomarker of pulmonary hypertension, is associated with increased mortality risk in SCD¹⁰⁴  has spurred an intense research activity aimed at delineating its role in clinical care. Not unlike TCD velocity, TRV is an attractive biomarker, since it can be measured with inexpensive, widely available echocardiographic methods and predicts a complication with high morbidity and mortality.¹⁰⁵  Unfortunately, TRV also shares with TCD an unclear underlying mechanism. While a TRV value ≥2.9 m/s has a 64% positive predictive value for pulmonary hypertension,¹⁰⁶  the implications of values ranging from 2.5 to 2.8 m/s are less clear. Thus, consensus has emerged over the importance of treating a TRV ≥2.9 m/s as a surrogate marker of pulmonary hypertension in SCD that should prompt referral to a pulmonologist, whereas there is no wide consensus yet on a care pathway for patients with more modest TRV elevations. However, TRV values between 2.5 and 2.8 m/s are not usually a benign finding. They are associated with proteinuria¹⁰⁷  and leg ulcers¹⁰⁸  and therefore appear to reflect a systemic vasculopathy characterized by hemolysis and endothelial dysfunction. TRV elevations in this range may also be useful as a screening test of pulmonary hypertension when coupled with a 6-minute walk test distance of <333 m or elevation in the serum N-terminal pro–brain natriuretic peptide (NT-pro-BNP) of >166 pg/mL (positive predictive value of 62%).¹⁰⁶  In any case, elevated TRV is a poor prognostic finding that should prompt closer surveillance, a correction of reversible causes of pulmonary disease (eg, smoking, airway hyperreactivity, or obstructive sleep apnea), and an intensification of disease-modifying therapy for SCD. In clinical research, elevated TRV has been used as a biomarker of endothelial dysfunction in the Sildenafil Therapy for Pulmonary Hypertension and Sickle Cell Disease (walk-PHaSST) clinical trial¹⁰⁹  and in the ongoing Multi-center Study of Riociguat in Patients with Sickle Cell Disease (STERIO-SCD; www.clinicaltrials.gov identifier NCT02633397) and is likely to continue to be included as an important surrogate marker of mortality and systemic vasculopathy.

Kidney dysfunction is among the earliest evidence of organ damage in SCD. Children with SCD develop glomerular hyperfiltration, increased excretion of creatinine, microalbuminuria, and hyposthenuria early on in the course of the disease,¹¹⁰  and the prevalence of these abnormalities increases in relation to age. The whole nephron is affected in SCD, with focal and segmental golmerulosclerosis being the most common histological finding.¹¹¹  Because kidney disease is associated with high morbidity and mortality in SCD,⁵  there has been major interest in identifying biomarkers to predict which patients will develop advanced nephropathy and identify early stages amenable to secondary prevention strategies. Together with TCD and TRV, microalbuminuria is a third biomarker that has been incorporated into clinical guidelines.¹¹²  Microalbuminuria is likely to be a relatively late marker of glomerular dysfunction, and its detection should lead to further testing, including a 24-hour urine collection for protein and initiation of angiotensin-converting enzyme inhibitor therapy in appropriate cases.¹¹²  It is not known, however, whether these interventions prevent the development of end-stage renal disease. Thus, there is an ongoing effort to identify other biomarkers with high predictive value of further complications and that could shed light on the mechanisms of kidney dysfunction. Among these, hemoglobinuria holds promise as it is clinically available for a minor cost, has been associated with progression of chronic kidney disease (HR, 13.9; 95% CI, 1.7-113.2; P = .0012) and albuminuria (HR, 3.1; 95% CI, 1.3–7.7; P = .0035) and is linked to higher hemolysis rate.¹¹³  Among experimental markers of early kidney disease the lysosomal enzyme N-acetyl-b-d-glucosaminidase and proximal tubular trans-membrane protein urinary kidney injury molecule 1 have also been strongly associated with albuminuria cross-sectionally in a relatively young cohort of patients.¹¹⁴  Additionally, high ambulatory systemic blood pressure¹¹⁵  and impaired nocturnal systolic blood pressure dipping detected on 24-hour ambulatory blood pressure monitoring¹¹⁶  were associated with microalbuminuria cross-sectionally in children and adolescents with SCD. Longitudinal studies across the lifespan are urgently needed to fully explore the role of kidney biomarkers in SCD.

Although SCD is a monogenic disorder, the clinical manifestations, disease severity, and mortality rates vary significantly among different phenotypes. In general, homozygous HbSS and double-heterozygote HbSβ⁰ thalassemia are the most severe phenotypes, whereas HbSC and HbSβ⁺ thalassemia have a milder clinical course. However, there exists a significant heterogeneity of disease severity, even among patients with the most severe phenotypes. There are several genetic and nongenetic modifiers that have been implicated in modulating disease severity. Genetic modifiers such as Senegal and Arab-Indian haplotypes and coinheritance of genes that lead to persistence of HbF are typically associated with mild disease.^117,118  In contrast, coinheritance of glucose-6-phosphate dehydrogenase deficiency is associated with severe anemia and increased blood transfusion requirement.^119,120  Coinheritance of α-thalassemia has mixed effects on disease severity, as it decreases hemolysis^121,122  and risk of stroke^60,123  but increases the frequency of VOCs and avascular necrosis of bone.⁶¹  Nongenetic factors such as environmental factors (eg, high wind speed, humidity, cold temperatures, and air pollution), infection, asthma, smoking, socioeconomic status, and access to care can adversely affect the disease course.^117,124,125  Unfortunately, there is no systematic way of accurately predicting which patient will have worse disease and die early preemptively by taking into account the factors listed above. This makes it difficult for practitioners to select patients for disease-modifying therapy with small therapeutic indices or high-risk profile such as bone marrow transplantation. Previous studies on patients with HbSS disease have attempted to identify clinical and laboratory markers that can predict disease severity and early mortality.^126,127  In the analysis of the infant cohort in the CSSCD, Miller et al reported that having dactylitis in the first year of age, mean baseline hemoglobin <7 g/dL, and leukocytosis in the absence of infection in the second year of age were predictive of severe disease (ie, stroke, frequent pain, recurrent ACS, or death) in the first decade of life.¹²⁶  However, Quinn et al found that the same risk factors were not predictive of severe disease, either alone or in combination, in the large, independent Dallas Newborn SCD Cohort.¹²⁷  These conflicting findings are probably due to the era in which these studies were conducted; the CSSCD study included participants who were enrolled prior to the implementation of penicillin prophylaxis and TCD screening, whereas the Dallas Newborn Cohort analysis occurred after those interventions. In a further effort to define pediatric disease risk, van den Tweel et al developed the first pediatric severity index in SCD (using 12 clinical complications and common laboratory test results), which clearly differentiated patients by genotypes and α-gene deletions but was only weakly associated with risk of death.¹²⁸  This score was validated in a subsequent retrospective pediatric cohort that showed that presence of α-gene deletions was not associated with lower pediatric severity index, questioning the role of α-thalassemia in decreasing global SCD severity.¹²⁹  Similarly, Sebastiani et al developed a Bayesian network model of the entire CSSCD cohort by including clinical, laboratory, and treatment data to predict the risk factors of near-term mortality in individuals with SCD.¹³⁰  In that model, age, male gender, sepsis, renal failure, stroke, and markers of hemolysis were associated with increased risk of near-term death. In a recent systematic review, Meier et al found that TCD velocities and reticulocytosis were predictive of serious complications of SCA.¹³¹  In particular, increasing reticulocytosis at a younger age was associated with increased hospitalizations, stroke, increased TCD velocities, and death. Interestingly, HbF was found to yield mixed results on disease severity, likely due to methodological issues of the studies included in the systematic review. In a meta-analysis of adults with SCD, Maitra et al found that advancing age, TRV ≥2.5 m/s, reticulocytosis, increased NT-pro-BNP, and decreased HbF were associated with increased mortality, while male gender, history of ACS, white blood cell count, hemoglobin, platelet count, LDH, ferritin, or creatinine were not.¹³²  Taken together, these studies show there is still controversy on what are the most important biomarkers of disease severity and mortality in SCD. Recently, gene expression profiling has harbored promise as a novel tool to risk-stratify patients with SCD. A composite risk score developed by combining established clinical risk factors along with circulating blood mononuclear transcriptomes has been shown to be predictive of mortality in 2 prospective cohorts of adults with SCD compared with either clinical risk factors or gene expression profiling alone.¹³³  Similarly, Du et al used cluster analysis to test the signatures of 17 biomarkers with various clinical complications in the CSSCD cohort and correlated them with markers of pulmonary vascular disease in 2 recent pulmonary hypertension clinical trials in children and adults with SCD.¹³⁴ 

SCD is a heterogeneous disorder with a wide range of clinical complications and chronic end-organ damage. Although SCD was described >100 years ago, significant limitations remain in the understanding of the pathophysiology of its complications and consequently in the development of biomarkers and new therapeutic modalities. Thus far, TCD, the only biomarker that has been prospectively studied in clinical trials, along with chronic blood transfusions have made a significant impact on the lives of patients with SCD by preventing primary stroke. Steady-state hemoglobin, HbF, reticulocytosis, microalbuminuria, and TRV are among the other laboratory tools available for the management of patients with SCD, but their utility needs to be validated in clinical trials. Clinically, there is not a wide range of treatment options other than hydroxyurea, blood transfusion, bone marrow transplantation, and recently l-glutamine (Endari). Ongoing studies of novel agents such as rivipansel, crizanlizumab, and voxelotor may provide additional therapeutic options to manage certain complications of SCD. Prediction modeling utilizing clinical variables, laboratory markers, and gene expression profiling may enhance the prediction of disease complications and global severity and in turn result in better identification of patients eligible for old and new treatments, potentially used in combination, and curative treatment options that may have an uncertain risk/benefit ratio. Thus, it is likely that clinical, laboratory, imaging, and genetic biomarkers will need to be combined to provide composite scores to predict specific complications and overall disease severity and mortality.

This work was supported by the National Institutes of Health National Heart, Lung and Blood Institute (grants 1K23HL112848-01A1 and 5RO1HL127107-02) (E.M.N.).

Enrico M. Novelli, Division of Hematology/Oncology, UPMC Heart, Lung and Blood Vascular Medicine Institute, University of Pittsburgh School of Medicine, E1240 Biomedical Science Tower, 200 Lothrop St, Pittsburgh, PA 15261; e-mail: noveex@upmc.edu.