Genomic polymorphisms in sickle cell disease: implications for clinical diversity and treatment

Abstract

Sickle cell disease (SCD) is one of the best characterized human monogenic disorders. The development of molecular biology allowed the identification of several genomic polymorphisms responsible for its clinical diversity. Research on the first genetic modulators of SCD, such as coinheritance of α-thalassemia and haplotypes in the β-globin gene cluster, have been followed by studies associating single nucleotide polymorphisms (SNPs) with variable risks for stroke, leg ulceration, pulmonary hypertension, priapism and osteonecrosis, with differences in the response to hydroxyurea, and with variability in the management of pain. Furthermore, multigenic analyses based on genome-wide association studies have shed light on the importance of the TGF-β superfamily and oxidative stress to the pathogenesis of complex traits in SCD, and may guide future therapeutic interventions on a genetically oriented basis.

Keywords::

• α-thalassemia
• β-globin gene cluster haplotype
• fetal hemoglobin
• genome-wide association study
• klotho
• nitric oxide
• sickle cell disease
• single nucleotide polymorphism
• TGF-β/BMP pathway

Figure 1. Timeline of major scientific achievements relevant to the study of sickle cell disease.

HbF: Fetal hemoglobin; SCD: Sickle cell disease; SNP: Single nucelotide polymorphism.

More than 50 years after the first description of the chemical difference between normal and sickle cell hemoglobins [1], sickle cell disease (SCD) has become one of the best characterized monogenic human disorders. However, while its clinical heterogeneity has been long recognized, scientific research over the last 20 years has tried to elucidate the role of several factors responsible for its clinical variability. The first single nucleotide polymorphism (SNP) was described in SCD. Since the first identification of haplotypes in the β-globin gene cluster to the present day [2,3], over 19 million SNPs have been described in the human genome [201], and most of these are still lacking a possible clinical correlation or function description. In SCD, previous reviews have reported that several of these polymorphisms are associated not only with the degree of anemia, but also with pain rate, prevalence of stroke, leg ulcers, pulmonary hypertension, osteonecrosis, hepatobiliary complications and priapism, among other several clinical aspects [4–6]. Moreover, treatment with hydroxyurea (HU), the main pharmacological agent known to improve SCD complications, has variable rates of effectiveness and many studies have aimed to investigate genetic variations that could explain why only some patients tolerate and respond to this treatment, while the rest still need to be treated with blood transfusion-based strategies [7]. This article aims to summarize the current knowledge on genetic polymorphisms that are related to the modulation factors of SCD clinical diversity and influence its treatment.

Study types in genomic research

Genomic research has rapidly evolved during the past decades, and the scientific approach to genetic-based understanding of human disease has changed along with technological development. Figure 1 shows a timeline of scientific discoveries important for the study of SCD, and it has become evident that knowledge in this field has been growing faster in time and in quantity alike [8–13]. Table 1 summarizes the basic characteristics of the main approaches used in genomic research, their advantages and some of their limitations [14–21]. On the one hand, a reasonably great amount of what is known about genetic modulation leading to phenotype variability in SCD stills derives from genetic epidemiological- and candidate gene-based studies (see later). On the other hand, recent technological developments strive to accomplish more extensive and refined, yet progressively less time-consuming, strategies in the study of the human genome. As a matter of fact, the importance of the first, smaller family-based studies that managed to characterize SCD and its Mendelian heritability, and of subsequent studies (e.g., on haplotypes and interaction with α-thalassemia and candidate gene SNPs) should not be underestimated. The completion of the human genome sequencing led to growing knowledge of human genetic variability, enabling the construction of whole electronic libraries containing the SNPs data, such as the International HapMap Project, launched in 2003 [11]. After this, the recognition of the inherent variation in the human genome as a possible explanation to diverse phenotypes, along with the feasibility of DNA sequencing in a large scale, initiated what could be called the ‘genome-wide study era’. Genome-wide association studies (GWASs) have become a new standard in genomic research in several medical fields for the search for genetic risk factors for common diseases (e.g., studies by the Wellcome Trust Case Control Consortium), focusing on coronary artery disease, hypertension, bipolar disorder, breast cancer, rheumatoid arthritis, Crohn’s disease and diabetes [22]. GWASs use microarrays currently able to identify millions of SNPs in a single run, and the further discovery that copy number variants (CNVs; DNA sequences undergoing deletions, duplications, inversions and translocations) may be more frequent in the human genome than previously expected, has been followed by the development of detection and sequencing technologies enabling scientists to look into these variations and their potential biological importance, although a recent study suggested that CNV genome-wide detection not necessarily enriches previous SNP-based GWASs [23]. This warrants further investigation before drawing definitive conclusions on the use of each approach. Deep DNA sequencing should play a major role in confirming genetic variants in a large number of samples from different populations, and should further extend our understanding of differential expression of genes in normal and pathological states. Regarding hematological disorders, GWASs have recently started a new chapter in genetic characterization of SCD, and these first published results have been included in this review. These newer data might give researchers a glimpse of what future steps should be taken in SCD genomics.

The following sections present brief summaries of what is already known about genetic modulation in SCD according to common subphenotypes encountered by the clinician in the care for these patients.

Anemia & hemolysis

Hemoglobin (Hb) and hematocrit levels are highly variable among patients with SCD, usually being lower in homozygous sickle cell anemia and progressively higher in HbS-β thalassemia and HbSC disease [24]. As variability is broad, even among homozygous sickle cell patients alone, one of the first identified genetic modulators was the coinheritance of α-thalassemia (Box 1). Approximately 30–40% of African–American and African–Brazilian patients with sickle cell anemia are heterozygous, and up to 3% are homozygous for the most common deletion-type α-thalassemia mutation – α^3,7kb[25–31]. Patients with one or two deleted α-globin genes have reduced hemolysis, lower reticulocyte counts and lower plasmatic lactate dehydrogenase (LDH) levels, owing to a lower mean corpuscular hemoglobin content (MCHC), less deoxygenated HbS polymerization, fewer denser dehydrated cells and fewer irreversibly sickled cells [32,33]. The averages of some studies report that, comparing genotypes αα/αα, α-/αα and α-/α-, mean hemoglobin levels are 8.1, 8.6 and 9.2 g/dl, respectively; mean corpuscular volume (MCV) is 92, 83 and 72 fl, respectively, and reticulocyte percentage is 11.4, 9 and 6.7%, respectively [33–36]. As far as the reduction of denser cells is concerned, studies in the future may reveal whether noncarriers of α-thalassemia have a better response to treatment with newly developed drugs aiming at reversing hemolysis, such as senicapoc, a Gardos channel blocker that preserves sickle cell hydration and increases red blood cell survival [37]. More recently, the coinheritance of the sickle cell gene and α-thalassemia genes has been studied at the populational level, demonstrating the negative epistatic effect between these genes based on how each of them affects intracellular globin production and on protection against malaria. This study shed light on why high frequencies of α-thalassemia are found among sickle cell patients in certain areas, and presented a possible explanation on the exclusion of the sickle cell alelle from thalassemia-containing populations, thus accounting for the different distributions of these mutations across Africa and the Mediterranean [38], and giving a geographic perspective of the SCD/α-thalassemia phenotype.

Fetal Hb & response to HU

The first study on genetic differences accounting for clinical variability in SCD reported the discovery of a site in linkage disequilibrium with the β^S gene (HBB Glu6Val) by use of restriction endonucleases [13]. Subsequent studies characterized the presence of at least five different haplotypes of the β-like-globin gene cluster, suggesting distinct geographic origins of the same β^S gene (Senegal, Benin, Bantu, Arab–Indian and Cameroon) and documented that haplotypes differed in fetal Hb (HbF) levels [2,3]. HbF (α₂γ₂) has the ability to inhibit the polymerization of deoxygenated HbS, making its levels an important modulator of the disease [39–42]. Senegal and Arab–Indian haplotypes are strongly associated with higher HbF levels, especially in carriers of the SNP C-T 158 bp 5´ to the γ-globin gene HBG2 (rs7482144), which creates a restriction site for the enzyme XmnI [43–45]. The XmnI-HBG2 polymorphism was the first quantitative-trait locus (QTL) described in HbF heritability, and elevates the relative production of the Gγ-globin. A high Gγ:Aγ ratio is also found in the Cameroon haplotype, but lacks the XmnI-HBG2 SNP [46]. Instead, this haplotype is characterized by an (AT)₈(T)₅ sequence in the -500 region of the HBB gene, and is associated with an AGCA deletion 5´ to the β-globin gene linked to the ^Aγ^T allele, possibly increasing the expression of γ-globin [47]. Absence of a Bantu haplotype is apparently associated with higher responses to HU [48]. HU is an antineoplastic agent known to increase HbF levels and is currently a useful pharmacological treatment for SCD. Nonetheless, not all patients in one specific haplotype demonstrate enhanced HbF production with HU, reinforcing the multigenic character of the HbF expression. Although there is evidence that the 5´ hypersensitive site (5´ HS) is highly polymorphic among sickle cell anemia patients [49,50], definite polymorphisms responsible for modulating HbF levels remain to be described.

Since the discovery that haplotypes with higher HbF levels and the association of SCD with hereditary persistence of HbF are associated with decreased comorbidity, there is an increasing interest in unraveling how HbF production is genetically determined, providing sites for modulation and treatment in SCD. Cis-acting regulatory elements are estimated to account for less than 25% of the variability, which has led to new insights into HbF regulation by trans-acting elements. Four trans-acting QTLs have already been linked to F cell number. The chromosomic region Xp22 maps to the FCP locus [51,52], but has not been validated in subsequent studies. At 6q22.3–24, a QTL was described in an Asian–Indian subject [53–55], and higher resolution studies identified the HBS1L-MYB intergenic polymorphism (HMIP) locus blocks 1–3, with HMIP-2 responsible for the most important effect [56], accounting for 3–7% of HbF levels variance. An interaction between a QTL at 8q and the C–T 158 bp SNP appeared to occur in this same family, and such findings were replicated by a study with 874 twin pairs [57]. Six SNPs in a gene in chromosome 8 called TOX have been described to be associated with HbF levels in more than 1500 SCD patients [58]. At 2p16.1, a region mapping to the BCL11A gene was defined by a GWAS in two different cohorts, African–American and African–Brazilian patients [59]. The QTL in intron 2 of this gene, formerly described as an oncogene, accounted for 15% of F cell number variance, and simultaneous analysis of BCL11A rs11886868, XmnI-HBG2 rs7482144 and HMIP loci yielded more than 20% of HbF variation in SCD patients. More recently, another GWAS described a novel region on chromosome 11 containing the ORF gene cluster, which might be important in the regulation of g-globin gene’ expression and HbF production, but it remains to be determined if BCL11A and ORF are involved in the response to HU [60]. SNPs in genes within the 6q22.3–23.2 and 8q11–q12 linkage peaks, and also the ARG2, FLT1, HAO2 and NOS1 genes, have been associated with the response of HbF to HU [61]. SAR1A is a gene encoding the small GTP-binding protein, secretion-associated and RAS-related (SAR) protein. SNPs have been identified in the upstream 5´ untranslated region (-809 C>T, -502 G>T and -385 C>A) of this gene, possibly associated with the HbF response to HU [62]. GWASs and CNV studies should also yield further knowledge about other HbF-related genes and provide in-depth understanding of the molecular mechanism involved in globin switching, offering the possibility of targeted activation of HbF production in the treatment of SCD [63–67].

Vaso-occlusive events & pain treatment

Either in the form of painful episodes, or presenting as a chronic disorder, pain remains an absolute hallmark of SCD, and clinical heterogeneity has always been a challenge to physicians in charge of the care for sickle cell patients. The first genetic trait described to influence the clinical severity of sickle cell painful episodes was HbF level, since high levels of HbF were recognized to be associated with a lower incidence of crises and, therefore, lower mortality [42]. Sickle cell anemia α-thalassemia is not associated with higher HbF levels and by raising the hemoglobin concentration, consequently elevating the blood viscosity, there seems to be an increase in the likelihood of some types of vaso-occlusive events, although there are conflicting results among studies [68–70]. Pharmacogenomics of opioid metabolism has tried to shed some light on the challenging management of sickle cell pain. Some polymorphisms of the CYP2D6 gene cause decreased metabolization of codeine to morphine, having been associated with failure of codeine treatment for pain crisis in children on HU [71], and possibly with a higher likelihood of adults being admitted to a hospital [72]. Differential morphine metabolism by UDP-glucuronosyltransferase 2B7 (UGT2B7) is associated with the SNP -840G→A, contributing to the variability of hepatic clearance of opioids, and represents a suitable candidate for future genotype–phenotype studies in pain management [73]. Tetrahydrobiopterin is a cofactor required for nitric oxide production, and its synthesis rate-limiting enzyme is GTP cyclohydrolase (GCH1). A GCH1 haplotype defined by three linked SNPs, for the first time, has been described to be possibly associated with patients with more frequent pain. Its relevance needs to be confirmed in other samples, as it might be relevant to future studies in pain management [74]. More recent work has correlated polymorphisms in the MBL2 gene to vaso-occlusive events in children with sickle cell anemia. Genotypes related to lower/intermediate levels of serum mannose-binding lectin (MBL) have been associated with a higher frequency of painful episodes, but how MBL variation and the activation of the complement system by the lectin pathway influence the sickling process remains to be better characterized [75,76]. Acute chest syndrome (ACS) is a type of severe vaso-occlusive episode, also thought to be multifactorial. Preliminary data showed that polymorphisms in genes, such as TGFBR3 and SMAD7, are associated with its occurrence, and corroborate the importance of the TGF-β pathway in the modulation of sickle cell anemia [77]. Pediatric ACS was also associated with SNPs in the gene encoding a member of the PI3K/PI4K family involved in cell–cell adhesion, PIK3CG. Older children and adults presenting with ACS were described, showing association with polymorphisms in other genes, such as SMAD1, NRCAM, SMAD3, klotho (KL) and STARD13 (near KL). The klotho protein is encoded by the KL gene in the chromosomal region 13q12. This β-glycosidase-like protein can be membrane bound and secreted, and it may have a role in nitric oxide metabolism, by stimulating nitric oxide production by endothelial cells. Polymorphisms affecting its production could modulate the intensity of oxidative stress in the microcirculation, alter endothelial-dependent vasodilation and, therefore, affect how the vaso-occlusive process takes place in SCD patients [78]. One study has shown the association of the T-786C endothelial nitric oxide synthase (eNOS) gene polymorphism with increased susceptibility to ACS in female patients [79].

Femoral & humeral avascular necrosis

The pathogenesis of femoral and humeral head osteonecrosis in sickle cell patients is not completely understood, but the impression that the coagulation mechanisms play some role in idiopathic osteonecrosis has led to several studies looking at polymorphisms involved in inherited disorders of thrombosis that would justify a putative hypercoagulable state. Controversial published results have failed to clearly demonstrate that known thrombophilic mutations are implicated in SCD osteonecrosis, even though a potential increased risk for avascular necrosis may involve the 5,10-methylene tetrahydrofolate reductase (MTHFR) C677T mutation [80–84]. Platelet adhesion may also play an important role in sickle cell vaso-occlusion. Inherited polymorphisms of platelet glycoproteins have been associated with occlusive arterial disease. One study described the association of the HPA-5b allele with vascular complications in SCD, such as osteonecrosis, stroke, priapism and ACS [85]. Another study failed to confirm this association in a multivariate analysis, but correlated more frequent and more severe vaso-occlusive crises to the HPA-3b/3b genotype [86]. Deletion of one or more α-globin genes remains a well-known genetic risk factor for osteonecrosis in sickle cell anemia [87], as do other forms of SCD with higher hematocrits, such as HbS-β⁺ thalassemia and HbSC disease, in which avascular necrosis is more frequent and appears earlier [88]. One study described the association of osteonecrosis with SNPs in several genes related to bone metabolism [89]. Ten SNPs in the KL gene could relate to a deficiency of klotho protein function, generating a dysregulation of the vitamin D metabolism, which may trigger cellular damage by the toxic action of increased levels of calcium, phosphorus and 1,25(OH)₂D₃[90]. Six SNPs in the ANXA2 gene encoding annexin A2, which is involved in osteoblastic mineralization, were also associated with osteonecrosis. Genes encoding proteins in the TGF-β/SMAD/BMP pathway have also been studied. SNPs in the TGF-β receptors type 2 and type 3 genes (TGFBR2 and TGFBR3) have been associated with a higher risk for avascular necrosis. Five SNPs in the BMP6 may be related to the role of BMP6 in bone formation and in inflammation, and the association with BMP6-3 SNP (rs3812163) was apparently confirmed by another study [91,92].

Infection & bacteremia

Infections are one of the leading causes of morbidity among SCD patients, and much effort has been made at discovering genetic predispositions to certain organisms. Children bearing the H/H131 Fcγ RIIA genotype have more frequently shown a positive history of Haemophilus influenzae type B infection [93]. Several studies have investigated the HLA locus and, if on one hand HLA class II DRB1*15 has a potential protective effect, on the other hand, HLA class II DQB1*03 and HLA-E*0101, in homozygosity, seem to increase infection susceptibility [94,95]. HLA DRB1*100101 was more frequently found in children with osteomyelitis [96], and the HLA-G +3142 polymorphism was associated with a possible protective effect against hepatitis C virus (HCV) [97]. Children with SCD were less likely to have infections if bearing genetic variants in the promoter region of the MBL2 gene, which encodes the mannose-binding protein (MBP), although the relationship between protection against infection and MBP low-producing variants has not yet been explained [98]. In sickle cell anemia patients, the G463A polymorphism in the gene encoding myeloperoxidase (MPO) may be a significant genetic modulator that renders patients more susceptible to infection [99]. Furthermore, a case–control study on bacteremia in SCD showed an association of several SNPs in IGF1R, as well as genes of the TGF-β/SMAD/BMP pathway (BMP6, TGFBR3, BMPR1A, SMAD3 and SMAD6), suggesting the importance of this pathway in the immune function [100]. The gene encoding RANTES (CCL5), a chemokine responsible for bringing immune cells to areas of infection or injury, has a genetic variant associated with protection against infection in sickle cell patients [101], while a study on polymorphisms in the gene encoding its receptor, CCR5, has suggested this gene as a potential candidate gene influencing clinical severity in SCD patients, although there was no statistical significance in this study [102].

Stroke

Sickle cell disease is the most common cause of stroke in childhood, and as one of the most important and potentially devastating complications in SCD patients, its genetic susceptibility has been focused on a large number of polymorphism-association studies [103]. Classic modulators of SCD severity, such as β^S cluster haplotype and α-thalassemia, have controversial effects. Decreased hemolysis and less dense cells seem to protect carriers of α-thalassemia [24,104–106], but this is still not a scientific consensus [107–109]. The first studies on HLA genes showed that apparently both HLA class I and II are involved in genetic modulation of the stroke phenotype. HLA B*5301, DQB1*0201, DRB1*0301 and *0302 increased the risk for stroke, while HLA DRB1*1501 protected against it [110]. Further research showed that the genetic contribution to stroke is distinct between large-vessel and small-vessel disease, suggesting different pathological mechanisms. With regard to small-vessel disease, adhesion molecules, such as VCAM-1, have been a preferential target of studies. The G1238C allele of the VCAM1 gene was associated with protection against small-vessel stroke, while carriers of the T-1594C allele were reported as more predisposed to cerebrovascular disease [111,112]. A SNP in the low-density lipoprotein receptor, LDLR, has also been shown to be protective against this type of stroke in children [112]. In addition, HLA DPB1*0401 was associated with small-vessel stroke, while HLA DPB1*1701 had a small protective effect against it [113]. Large-vessel stroke seems to have a different genetic predisposition and SNPs in the IL4R and TNFA genes seem to increase risk for this type of stroke more than fivefold [112,114]. A recently published abstract reported some TNFA polymorphisms associated with CNS events other than stroke, such as seizures and transient ischemic attacks [115]. A genetic variant that upregulates transcription of the gene encoding leukotriene C4 synthase, LTC4S (-444)C, is associated with protection against large-vessel disease, even though it increases proinflammatory mediators [114]. HLA A*1012, A*2612 and A*3301 were described to confer susceptibility to large-vesel stroke [113]. A Bayesian network has been developed to analyze SNPs and their complex interaction in the modulation of risk for stroke in SCD. This kind of approach seems to have a better accuracy in predicting multigenic traits, and the findings showed involvement of proteins from the TGF-β pathway, such as BMP6, TGFBR2 and TGFBR3, and P-selectin. GWASs should also lead to functional research of genes involved in angiogenesis in the modulation of stroke risk [116–118].

Gallstones & bilirubin levels

Biliary complications with higher incidence of gallstones are frequent in hemolytic anemias [119,120]. In SCD, studies have shown the influence of a polymorphism in the promoter region of the uridine diphosphate glucuronosyltransferase 1A (UGT1A) gene, which codifies the enzyme uridine diphosphate glucuronosyltransferase 1A, responsible for bilirubin conjugation. The promoter region contains a sequence of TA repetitions, with the wild type containing six, and polymorphic alleles containing five, seven or eight repetitions. Longer repetitions lead to a decrease in the expression of the gene, thus decreasing effectiveness of bilirubin glucuronidation [121], and a reduction in UGT1A transcription associated with the 7/7 genotype accounts for the genetic basis of Gilbert syndrome. In sickle cell patients, the number of TA repetitions has been shown to exert an important influence on serum bilirubin levels [122]. Homozygosity for the 7 alelle has been associated with earlier and increased incidence of gallstones in children, and a higher need for cholecystectomy in adults with symptomatic biliary disease [123–127]. Coinheritance of α-thalassemia decreases hemolysis and, therefore, can decrease bilirubin levels, but data are contradictory in defining whether it is sufficient to compensate for the effect of the 7/7 genotype [128,129].

Renal impairment

Many renal complications are observed in patients with SCD and kidney failure has emerged as an important comorbidity, as survival of these patients has improved over the last decades but, still, little is known about the genetic modulation of renal function in these patients [130–132]. There are insufficient data on the influence of α-thalassemia on sickle cell renal disease, except in sickle cell-trait bearers, where coinheritance seems to have a protective effect against loss of urinary-concentrating ability [133]. The inheritance of the Bantu haplotype has been associated with worse renal function, supposedly because of the lower HbF levels found in this haplotype [134]. With growing evidence of the importance of the TGF-β/BMP pathway in sickle cell-associated complications, one study managed to demonstrate the association of haplotypes in the TGFBR1 gene harboring three SNPs with glomerular filtration rate. The common haplotype I A-A-G was inversely associated with renal function, while higher filtration rates were found in association with the haplotype VI G-G-A [135]. A better knowledge of how genetic modulation may account for renal impairment should develop increasing importance, as late complications, such as heart failure and transfusion-related iron overload, invariably indicate the use of drugs with potential renal adverse effects (e.g., hyperpotassemia with use of angiotensin-converting enzyme inhibitors or proteinuria with use of deferasirox).

Leg ulceration

Leg ulceration is a complication that varies widely among SCD patients, although it is also common to other hemolytic anemias. They affect between 2 and 40% of patients, depending on the origin of the cohort analyzed. Furthermore, this symptom seems to be closely related to lower Hb levels, lower socioeconomical status, and is more frequent in males and virtually inexistent in children. From the molecular point of view, several genes have been linked recently to the development of chronic leg ulcers in SCD. Individuals with coinheritance of α-thalassemia, as well as compound genotypes in SCD (e.g., HbSC and HbS-β) present with a lower degree of hemolysis, higher Hb levels and, consequently, have been shown to protect against the development of chronic cutaneous ulceration [136]. The polymorphisms rs685417 and rs516306 in the KL gene have been associated with leg ulceration, but they seem important in other SCD complications as well, such as ACS, stroke, osteonecrosis and priapism [89,117,137]. TEK tyrosine kinase (or TIE2) is a receptor for angiopoietins implicated in hypoxia-induced and normal angiogenesis [138]. This could account for the importance of the SNPs rs603085 and rs671084 in impairing the healing process of leg ulcers. Other polymorphic genes possibly implicated in the pathogenesis of leg ulcers belong to the TGF-β/BMP pathway. Members of the TGF-β superfamily activate MAPKs through SMAD signaling, affecting cell proliferation, inflammation, immune regulation and response to tissue injury, among others. SNPs increasing the risk for leg ulcers were described in several genes, such as those encoding receptors TGFBR2, TGFBR3 and BMPR1B, BMP6, signaling proteins SMAD7 and SMAD9, SMURF1 (a ligase specific for SMAD), and downstream enzymes, such as MAP2K1 and MAP3K7[137].

Pulmonary hypertension

Pulmonary hypertension (PH) has increasingly become a major issue in SCD care since it was recognized as a relatively frequent complication and strongly related to early mortality, as predicted by the measurement of tricuspid regurgitant jet velocities [139]. Compound heterozygotes, such as HbSC disease and HbS-β-thalassemia, present with a lower risk for PH, probably owing to a lower rate of hemolysis [140]. Thus, genetic susceptibility to the development of PH seems to be related to factors affecting the hemolytic process itself. Once again, the possible association of polymorphisms has been described in KL, TEK and in several genes from the TGF-β/BMP signaling pathway (ACVRL1 and BMPR2 – previously associated with primary idiopathic PH [141,142], along with BMP6 and ADRB1), and may influence how impaired nitric oxide metabolism secondary to hemolysis leads to PH [143]. Preliminary results showed that SNPs in intron 1 of the NEDD4L gene may be associated with an elevated serum level of N-terminal pro-brain natriuretic peptide (NT-pro-BNP) in sickle cell anemia patients enrolled in a GWAS. Since high levels of NT-pro-BNP are related to high mortality and PH, NEDD4L should be a candidate gene in PH pathogenesis [144]. Another abstract reported a protective effect against pulmonary hypertension attributed to CYBR5 T116S polymorphism, which was associated with lower tricuspid regurgitation velocities, presumably due to lower hemolysis rates in carriers of the SNP [145].

Priapism

Priapism is classically associated with sickle cell hemoglobinopathies [146,147] and, as up to 30% of male patients experience this manifestation, some studies have identified genetic polymorphisms associated with priapism as a dominant feature. An association between SNPs in the KL gene and priapism has been established, corroborating a common phenotype in which severe hemolysis, pulmonary hypertension, leg ulceration and priapism are associated with each other [148]. Carrying a TT genotype for rs2249358 yielded an odds ratio of 2.6 compared to a C_ genotype. The odds ratio for the G–G haplotype associating two other SNPs (rs211239 and rs211234) was 2.3 [149]. Other genes whose polymorphisms have been identified as associated with priapism are TEK, TGFBR3, ITGAV, AQP1 and F13A1. The TGFBR3 gene has already been implicated in stroke and avascular necrosis. The type III receptor for TGF-β is strongly expressed by endothelial cells and is required for endothelial cell transformation [150]. AQP1 encodes acquaporin-1, a water channel expressed by both erythrocytes and endothelial cells. AQP1 may influence the sickling process because it regulates cell volume and controls carbon dioxide permeability through the erythrocyte membrane [151,152]. ITGAV encodes the α_v subunit of an integrin expressed by the endothelium, and with a key role in angiogenesis [153]. Some polymorphisms in the α-subunit of Factor XIII (F13A1) have been correlated to priapism, and there is evidence of its importance in coagulation, based on a higher risk of ischemic stroke in the general population involving this same gene [154]. Priapism has also been associated with polymorphism of the human platelet antigen-5 gene (ITGA2) [85].

Transfusion therapy in SCD patients

Patients with SCD may need several blood transfusions throughout their lifetime and may be placed on a chronic transfusion program, with increased risks for alloimmunization and iron overload. Polymorphic mutations of the human hemocromatosis HFE gene have been studied in SCD, but no association with higher degrees of iron overload has been established [155]. Alloantibody development after blood transfusion is relatively common in SCD, reaching up to 47% of patients, depending on the cohort [156]. The presence of a polymorphism in the TRIM21 gene has been associated with the rate and time until alloimmunization in children with SCD. Carriers of the T/T genotype were possibly more tolerant to allogenic red blood cells, causing lower rates of alloimmunization and later development of alloantibodies [157].

Clinical severity of SCD

A fine GWAS applied to SCD was conducted by Sebastiani et al.[158], which managed to correlate clinical severity of SCD based on a network model (composed of 25 different clinical parameters) to 40 different SNPs, including genes not previously identified as pathogenetically important in SCD. A microarray containing probes able to identify more than 600,000 SNPs was used, and the need for better statistical analysis was supplied by the development of an additional subanalysis method called SNP set enrichment analysis (SSEA), designed to further evaluate for sets of SNPs that could define regions of the genome where these genetic variations occur with statistical significance; therefore, identifying more candidate genes associated with a certain subphenotype. New genes identified as probably important to the pathogenesis of SCD included KCNK6 (a potassium channel protein) and TNKS (which encodes a protein called tankyrase1, a polymerase that may influence telomere shortening), and also several other genes that are not directly related to SCD pathophysiology or have unknown functions.

This first study of SCD heterogeneity through a GWAS still needs additional validation and functional studies, but it represents a major scientific breakthrough regarding the possibilities to identify novel genes and signaling pathways possibly involved in the pathophysiology of different SCD complications. Furthermore, GWAS studies should yield more data about non-protein-encoding regions of the human genome, whose functions remain obscure. It is also an example of how important developments in the statistics field have become to the study in populational and genome-wide genetic studies.

Expert commentary

The unexpected clinical diversity in a monogenic disease such as SCD has led to countless genetic studies and current knowledge has evolved, together with technological development in molecular biology. Evolution from the basic identification of polymorphic sites with restriction endonucleases to robust GWASs has provided the tools to discover the genetic complexity that affects genotype–phenotype correlation. The main associations between polymorphic genes, their importance and subphenotypes in SCD are summarized in Table 2, with special attention to the KL gene and genes involved in the TGF-β/SMAD/BMP pathway in Table 3.

The number of SNPs and genes possibly associated with clinical variability of SCD are increasing very rapidly, but great amounts of data need to be confirmed in larger multicentric studies before they may be used as an effective tool in the management of the patients. Possibly, in the near future, a number of genetic predictors of the variability will be relevant for clinical purposes.

The possibility of working with a whole-genome perspective has turned sample size into an important issue, since the pursuit for genetic variants with smaller effects meant a larger number of samples, which has led to an increasing risk of discovering ‘false-positive’ associations. This, along with the difficulty to validate and replicate results, might be the main challenges researchers will have to face in GWASs, so multicentric studies and multiprofessional work (particularly including bioinformatic experts) should progressively become a standard in this field.

Five-year view

In 5 years, GWASs, Bayesian networks and CNV studies will have yielded much more information about new genes, polymorphisms and their correlation with most subphenotypes in SCD. Individual studies of each gene shall lead to a better interpretation of associations with SNPs lacking an obvious pathophysiological link to clinical diversity, as well as a better understanding of the underlying mechanisms of gene expression and regulation and, possibly, lead to the characterization of new signaling pathways that could provide novel therapeutics aims. The probable evolution of GWASs should be genome-wide gene-interaction studies, in which characterization of epistatic phenomena might uncover how genetic variations in noncoding regions could be responsible for transcriptional, translational or post-translational regulation, and could affect phenotype and gene–environment interactions as well. Further studies should also strive for a more genetically oriented therapeutic approach to this heterogenous disease, with special interest in the influence of pharmacogenomics on pain management and HbF induction, and the development of new drugs based on updated knowledge on vascular biology in the sickle cell vaso-occlusive process.

Table 1. Comparison among different approaches to genomic study in human diseases.

Type of study	Main characteristics	Advantages	Limitations	Examples in sickle cell disease
Genetic-epidemiological studies (Linkage analysis: family-based - ‘pedigree’, sibling and twin-pair studies)	Defined by a heritable phenotype, uses a low number of genetic markers to define linkage disequilibrium	• Lower number of samples needed (tens)	• Low statistical power if involvement of multiple genes, better suited to monogenic diseases • Familial clustering due to environmental factors • Dependent on high penetrance of genetic mutation/polymorphism	Sickle cell disease and α-thalassemia
Candidate-gene based studies	Screening cases and controls for new mutations and variable SNPs in genes possibly influencing phenotype	• Possible more direct clinical application • Low-complexity technology needed (regular PCR, restriction enzymes or direct sequencing of low number of samples)	• Depends on previous knowledge of the pathogenesis of the disease • Multiple testing needed if several genes/proteins known to be involved in the disease pathogenesis • Multiple tests increase false-positive results • False-negative results in small samples and low relative risk • Large sample size needed depending on minor allele frequency and statistical power (hundreds to thousands)	UGT1A polymorphism and cholelithiasis in sickle cell disease
Genome-wide association studies (SNPs and copy-number variation arrays and Bayesian networks)	Matching and comparing DNA sequences of populational samples of cases and controls to millions of genetic variants previously described in a high-resolution genetic map	• Detection of variants in genes with lower genetic relative risk for disease • Detection of genes not pathogenetically related to a certain disease • Identification of noncoding regions associated with disease • Detection of dependency across different genes and chromosomes	• Need for specific data management of large quantities of information to avoid false-positive results • Larger sample sizes (than candidate gene-based studies) needed to detect genome-wide low relative risks, to account for high number of variants and to validate data (thousands to tens of thousands) • False-negative results if SNPs analyzed separately instead of simultaneously • Replication needed in populations with different ancestries (e.g., European vs African)	Fetal hemoglobin levels SNPs associated with sickle cell disease severity
Deep DNA sequencing (ChIP assays followed by massive sequencing)	Genome-scale view of protein–DNA interactions by hybridizing fragmented immunoprecipitated chromatin to a microarray (ChIP-chip) followed by direct sequencing (ChIP-seq)	• Characterization of whole genomes with resolution to detect single nucleotide differences with low amounts of DNA (10–50 ng in ChIP-seq)	• High cost • Lack of easy access to user-friendly computational platforms	Identification of transcription factors able to activate γ-globin gene expression in sickle cell disease

ChIP: Chromatin immunoprecipitation; SNP: Single nucleotide polymorphism.

Table 2. Main associations between genes and complications in sickle cell disease.

Main importance	Gene	Pain	ACS	Infection	Stroke	Leg ulcer	Priapism	AVN	PH	Bil	RF	Ref.
Cell adhesion	SELP				×							[116–118]
	ITGAV											[148]
	NRCAM		×									[77]
	PIK3CG		×									[77]
	VCAM1				×							[111,112]
Angiogenesis	TEK					×	×					[138,148]
Cell growth	ANXA2							×				[89]
	IGF1R			×								[100]
Inflammation	IL4R				×							[112]
	TNFA				×							[114,115]
	CCL5			×								[101]
Cholesterol metabolism	LDLR				×							[112]
Coagulation	F13A1						×					[148]
	HPA3	×										[86]
	ITGA2		×		×		×	×				[85]
	MTHFR							×^†				[80–84]
Immunity	HLA (several)			×	×							[94–97,110,113]
	MBL2	×		×								[75,76,98]
	MPO			×								[99]
Nitric oxide metabolism	GCH1	×										[74]
	NOS3		×									[79]
	KL		×			×	×	×	×			[77,89,137,143,148]
TGF-β/SMAD pathway	ACVRL1								×			[143]
	BMP6			×	×	×		×	×			[89,91,92,100,118,137,143]
	BMPR1A			×								[100]
	BMPR1B					×						[137]
	BMPR2								×			[143]
	MAP2K1					×						[137]
	MAP3K7					×						[137]
	SMAD1		×									[77]
	SMAD3		×	×								[1,77]
	SMAD6			×								[100]
	SMAD7		×			×						[77,137]
	SMAD9					×						[137]
	SMURF1					×						[137]
	TGFBR1										×	[135]
	TGFBR2				×	×		×				[92,116,137,148]
	TGFBR3		×	×	×	×	×	×				[92,100,116,118,137]
Bilirubin metabolism	UGT1A									×		[122–127]
Unknown	STARD13		×									[77]

^†Controversial reports.

ACS: Acute chest syndrome; AVN: Avascular necrosis; Bil: Biliary disease and bilirubin elevation; PH: Pulmonary hypertension; RF: Renal failure.

Table 3. Polymorphisms in genes involved in the TGF-β/SMAD/BMP pathway and in the klotho gene, and genotypes associated with subphenotypes in sickle cell disease.

Subphenotype	Gene	Polymorphism	Genotypes (higher→lower risk)	Patients (n)	Ref.
Lower glomerular filtration rate	TGFBR1	rs2240036	CC→CT→TT	1140	[105]
		rs4145993	TT→CT→CC
		rs17022863	GG→AG→AA
		rs1434549	TT→CT→CC
Leg ulcers	TGFBR2	rs1019856	G_→AA	759	[107]
	TGFBR3	rs2038931	TT→C_
	BMPR1B	rs1560909 rs7661539	AA→G_ T_→CC
	BMP6	rs270393	GG→C_
	KL	rs685417 rs516306	AG→GG CT→TT
	SMAD9	rs9576135	A_→GG
	SMAD7	rs736839^†	CC→T_
Pulmonary hypertension	BMP6	rs267192	CT→CC	111	[113]
	ACVRL1	rs3847859 rs706814	AA→AG→GG TT→AT→AA
	TGFBR3	rs10874940	TT→CT→CC
	BMPR2	rs35711585 rs17199249	NR
Infection	BMPR1A	rs6586039 hCV1663921	GG→A_ CC→T_	145	[73]
	BMP6	rs270387 rs267188 rs408505 rs449853	GG→AA AA→T_ TT→C_ TT→C_
	SMAD3	rs10518707	A_→GG
	SMAD6	rs5014202	C_→ TT
	TGFBR3	rs2765888 rs6662385^†	TT→CT TT→C_
Stroke	TGFBR2	rs1019856 rs876687 rs3773658	^‡	1398	[88]
	TGFBR3	rs1555889 rs284875 rs2038931 rs901917 rs2148322 rs2765888 rs2007686 rs284874	^‡
	BMP6	rs199205 rs267196 rs267201 rs408505 rs449853 rs1225934 rs270387 rs270393 rs267192	^‡
Priapism	TGFBR3	rs7526590	TT→AT→AA	199	[118]
	KL	rs2249358 rs211239 rs211234/rs211239	C_→TT G_→AA Haplotype G-G	768	[119]^§
Acute chest syndrome	TGFBR3	rs284157	NR	1442	[51]^¶
	SMAD3	NR	NR
	SMAD7	rs736839^†	NR
Acute chest syndrome in patients aged over 5 years	SMAD1	rs2068991	NR	1442	[51]^¶
	KL	rs2149860 rs656525	NR
Avascular necrosis	TGFBR2	rs934328	CC→T_	514	[63]
	TGFBR3	rs284157	A_→GG	792	[63]
	BMP6	rs3812163 rs270393 rs267196 rs449853 rs1225934	TT→A_ GG→C_ TT→AT TT→C_ CC→AC	244 539 673 764 766	[65,63]
	KL	rs480780 rs211235 rs2149860 rs685417 rs516306 rs565587 rs211239 rs211234 rs499091 rs576404	AC→CC C_→AA G_→AA AG→GG T_→CC AG→AA A_→G G _→AA A_→GG C_→AA	124 399 803 651 803 557 782 767 730 804	[63]

Single nucelotide polymorphisms occurring more than once in the table are bold.

^†Single nucleotide polymorphism is adjacent to the gene.

^‡Genotype effect requires simultaneous consideration of other single nucleotide polymorphisms (see [88]).

^§Association not replicated in [118].

^¶Data reported in abstract form only.

NR: Not reported in the original publication.

Box 1. Effects of α-thalassemia on sickle cell disease.

• • Lower mean corpuscular hemoglobin content

• • Fewer denser dehydrated cells and irreversibly sickled cells

• • Reduced hemolysis

• • Lower reticulocyte counts

• • Lower plasmatic lactate dehydrogenase levels

• • Increased frequency of vaso-occlusive events^†

• • Increased risk for osteonecrosis

• • Decreased risk for stroke^†

• • Decreased risk for leg ulceration

Key issues

• • Sickle cell disease is a monogenic, yet clinically heterogenous entity, and several genetic polymorphisms are markers of genes that may account for this variation.

• • Higher fetal hemoglobin (HbF) levels associated with the XmnI-HBG2 single nucleotide polymorphism (SNP) and coinheritance of α-thalassemia were among the first genetic modulators to be described.

• • Current knowledge on BCL11A and other genes may explain how HbF levels are regulated, and may lead to novel therapeutic approaches in HbF induction in sickle cell disease.

• • Further functional studies of genetic polymorphisms involving KL, TEK and genes in the TGF-β/SMAD/BMP pathway may explain different clinical presentations of several complications, such as osteonecrosis, pulmonary hypertension, priapism, leg ulcers and renal failure, and may allow the development of novel therapies for these complications.

• • Genome-wide association studies and deep DNA sequencing studies may lead to Bayesian network models able to explain multigenic traits in sickle cell disease, identify novel genes and subsequent variations in signaling pathways involved in its pathophysiology, and guide future therapeutic interventions on a genetically oriented basis.

• • Further multicentric research and development of more complex bioinformatic analysis are warranted to characterize genetic variants in sickle cell disease among different populations across the world.

Acknowledgements

We would like to thank Nicola Conran for the English review of this manuscript.

Financial & competing interests disclosure

The authors are supported by FAPESP and CNPq, Brazil. The Hematology and Hemotherapy Center, UNICAMP, forms part of the National Institute of Science and Technology of Blood (INCT do Sangue - CNPq/MCT). The authors have no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed.

No writing assistance was utilized in the production of this manuscript.

Notes

^†Indicates studies with controversial results.