|Home | About | Journals | Submit | Contact Us | Français|
Cutaneous leg ulcers are common in sickle cell anaemia and their risk might be genetically determined. We studied sickle cell anaemia patients to examine the relationship of leg ulcers with haemolysis and with SNPs in candidate genes that could affect sickle vasoocclusion. Leg ulcer patients had lower haemoglobin levels and higher levels of lactate dehydrogenase, bilirubin, aspartate aminotransferase and reticulocytes than did control patients with sickle cell anaemia but without leg ulcers. Age-adjusted comparisons showed that sickle cell anaemia-α thalassaemia was more frequent among controls than cases. These results strongly suggested that the likelihood of having leg ulcers was related to the intensity of haemolysis. Two-hundred fifteen SNPs in more than 100 candidate genes were studied. Associations were found with SNPs in Klotho, TEK and several genes in the TGF-β/BMP signaling pathway by genotypic association analyses. KL directly or indirectly promotes endothelial NO production and the TEK receptor tyrosine kinase is involved in angiogenesis. The TGF-β/BMP signaling pathway modulates wound healing and angiogenesis, among its other functions. Haemolysis-driven phenotypes like leg ulcers could be improved by agents that reduce sickle erythrocyte density or increase NO bioavailability.
Leg ulcers are a common complication of sickle cell anaemia (homozygosity for HBB, glu6val). In the United States, 2.5% of patients with all common genotypes of sickle cell disease had leg ulcers.(Koshy et al, 1989) In Jamaica, more than 40% of patients and in Africa, between 1.5 and 13.5% of sickle cell disease patients were reported to have leg ulcers.(Clare et al, 2002;Durosinmi et al, 1991;Knox-Macaulay, 1983;Akinyanju & Akinsete, 1979)
The cause of leg ulcers in sickle cell disease, and in other types of chronic haemolytic anaemia, is unknown.(Daneshmend & Peachey, 1978;Giraldi et al, 2003;Stevens et al, 1977;Frempong & Steinberg, 2001;Eckman, 1996) Chronic venous ulcers have been associated with thrombophilia, perhaps because of the relationship between thrombophilia and venous thrombosis.(Mackenzie et al, 2002) Whether or not venous pressure abnormalities contribute to the development of sickle cell leg ulcers is inconclusive.(Billett et al, 1991;Clare et al, 2002;Chalchal et al, 2001;Mohan et al, 2000) We hypothesized that leg ulcers, similar to other complications of sickle cell disease, might be related to the severity of haemolytic anaemia and be effected by polymorphic genes that could modulate the subphenotypes of sickle cell disease.(Gladwin et al, 2004;Ohene-Frempong et al, 1998;Adams et al, 1994;Sarnaik & Ballas, 2001) To study the clinical epidemiology of leg ulcer in sickle cell disease and to search for genetic modulators of the leg ulcer phenotype, we re-analyzed the Cooperative Study of Sickle Cell Disease (CSSCD) database that provided the data for a prior report of sickle cell leg ulcers in North America and studied patient DNA samples for single nucleotide polymorphisms (SNPs) in candidate genes that might modulate the course of sickle cell disease.(Koshy et al, 1989)
The CSSCD enrolled and followed more than 4,000 patients with sickle cell disease who had visited one of 23 participating clinical centers across the United States between 1978 and 1998.(Gaston et al, 1987;Gaston & Rosse, 1982) Data were collected on many complications of the disease, including leg ulcers. The diagnosis of an active leg ulcer or history of a previous leg ulcer was recorded and patients’ medical records were confirmed by trained clinicians. Some CSSCD data on leg ulcers was previously reported.(Koshy et al, 1989)
A case is defined as CSSCD participant who had an ulceration of the skin of the lower legs, especially the medial or lateral surfaces, which failed to heal in a period of two weeks. Controls were CSSCD participants who had never experienced a leg ulcer. Controls were restricted to patients aged 20 years and older to avoid over representation of very young patients. These studies were approved by the IRB of Boston University School of Medicine.
Blood samples for the detection of α thalassaemia and the β-globin gene cluster haplotype were stored at a National Institute of Health repository. We obtained DNA from the repository for individuals with sickle cell anaemia, with or without coincident α thalassaemia. From this sample, SNP genotyping was done on 243 leg ulcer cases and 516 controls. The laboratory methods used to define the haemoglobin phenotype and the haplotypes of the β-like and α-globin gene cluster were described previously.(Dozy et al, 1979;Steinberg et al, 1984;Steinberg et al, 1997)
Candidate genes were picked for their potential role in the pathophysiology of sickle cell disease. They included: mediators of inflammation; oxidant injury; NO biology; vasoregulation; cell-cell interaction; blood coagulation; haemostasis; growth factors. Validated SNPs with population frequency information and heterozygosity values of >0.1 in the candidate genes were first selected from public databases (http://www.ncbi.nlm.nih.gov/) and SNP analysis was first done by mass spectrometry using the Sequenom mass spectrometry SNP genotyping system, as reported.(Baldwin et al, 2005;Nolan et al, 2004) Following findings of significant associations among SNPs of the TGF-β/BMP pathway with many of the subphenotypes of sickle cell disease(Sebastiani et al, 2005;Baldwin et al, 2005), a second phase of genotyping was done to further explore this particular pathway. Follow-up SNPs in the TGF-β/BMP pathway were selected using SNPBROWSER v3.0 software (Applied Biosystems, Foster City CA). Each haplotype tagging (ht) SNP selected met the assay design requirements for genotyping using the ABI SNPLEX genotyping system. SNPLEX reactions were analyzed on an Applied Biosystems 3730 DNA Sequencer. The resulting data were analyzed to determine the association between SNPs of the TGF-β/BMP pathway and leg ulcers. A summary of all SNPs and genes studied can be found at http://ww.bu.edu/sicklecell/downloads/Publications_SupplementaryInformation/TGFB_SNPs.xls
For quality control purposes, about 3% of the DNA samples were re-genotyped, and Hardy-Weinberg equilibrium was assessed for each SNP. Hardy-Weinberg equilibrium was determined prior to analysis and was done for quality control purposes rather than to evaluate if the genotypes met Mendelian expectation because all members of our study population have sickle cell disease.
An extensive data file containing medical history and laboratory findings was obtained from the CSSCD study database. A tracking database was developed to record data set characteristics including file name, date, record count, and contents of each data file. Identities were verified against those in the central sample tracking database and data formatting was checked for compatibility with the cumulative genotype data set. Data were forwarded to a SQL-server database that housed the genotype data set used in this study.
Genotypic counts were compared between sickle cell patients with leg ulcers and patients without leg ulcers using multiple logistic regression. For each SNP, odds ratios (OR) and their 95% confidence intervals (CI) were calculated by assigning either one of the homozygote genotype classes or a pooled group of heterozygotes and one homozygote class as the referent. In our initial screen, we considered a SNP to have an association with a phenotype when the p-value was less than or equal to 0.01, or if one or more SNPs in the same gene were significant at the 0.05 level. If a SNP met these criteria, a second phase of genotyping was done to study additional SNPs in the gene. The individual associations between the additional SNPs and leg ulcer were ascertained using multiple logistic regression analysis.
Because the large number of significance tests conducted could inflate the type one error rate, we controlled the false discovery rate (FDR) using the Benjamini and Hochberg algorithm.(Benjamini & Hochberg, 1995) The two sided p-values were sorted in increasing order (p1 < p2 <…<pi) and we identified the largest index j such that p < jα/k, where α is the FDR and k is the overall number of significance tests that we conducted. We repeated the search for different α-values and chose to optimize the trade-off between selected significant associations and the rate of falsely significant associations as previously described.(Baldwin et al, 2005) The selection of significant SNPs was based on 215 tests therefore the largest p-value to accept a significant association with 10% FDR was 0.0008. Increasing the FDR has little effect on the largest p-value to accept as significant until we reach a FDR of 30% where the p-value to accept significance is 0.0263.
Among the patients enrolled in the CSSCD, we found information on 1,307 individuals with sickle cell anaemia and sickle cell anaemia-α thalassaemia, 830 patients with HbSC disease (compound heterozygosity for HbS and HbC [HBB glu6lys]) and 202 patients with HbS-β+ thalassaemia. Three-hundred eighty-seven patients with sickle cell anaemia had a confirmed history of leg ulcers or had leg ulcers at the time of examination (Table 1). Nine-hundred twenty patients without a leg ulcer served as controls. Cases were on average 4 years older than controls (p<0.001), therefore, all the comparisons reported in Tables 1 and and22 were age-adjusted. Selected laboratory data are shown in Table 2. Leg ulcer patients had more severe haemolytic anaemia than did controls as shown by their higher reticulocyte counts and higher levels of lactate dehydrogenase (LDH), bilirubin and aspartate aminotransferase (AST). Cases also had higher white blood cell counts than controls. As reported previously, leg ulcer patients had lower total haemoglobin and fetal haemoglobin (HbF) levels than controls.(Koshy et al, 1989) Males were more likely to have leg ulcers than females, as noted in other studies and thus the genotype association analyses were adjusted for sex as well as age. Mean systolic and diastolic blood pressure, creatinine and alanine aminotransferase (ALT) were not associated with leg ulcers.
The haplotype of the β-like globin gene cluster was not associated with leg ulcers. As reported previously, sickle cell anaemia-α thalassaemia was more frequent among controls (cases: 22%, controls: 25%, OR: 0.7, 95% CI: 0.5-0.9), indicating a protective effect of this genotype.(Koshy et al, 1989) Among the 830 patients with HbSC disease and 200 cases of HbS-β+ thalassaemia, leg ulcers were reported in only 31 and 7 patients, respectively. Leg ulcers were associated with other clinical manifestations of sickle cell disease such as ischaemic stroke (OR: 1.4, 95% CI: 0.9-2.0) and acute chest syndrome (OR: 1.52, 95% CI: 1.1-1.9).
When analysis of laboratory data was restricted to the 759 sickle cell anaemia and sickle cell anaemia-α thalassaemia patients (243 cases vs. 516 controls) that were also genotyped for SNPs in candidate genes, the results were similar except that only priapism was associated with leg ulcers and neither the reticulocyte count nor leukocyte count were significantly different. The dissimilar results may be due to a lack of power and not a lack of association.
In our initial studies, 129 SNPs in 47 candidate genes were genotyped. Three SNPs in and near KL were associated with leg ulcers, as well as two SNPs in TEK, and one SNP each in TGFBR2 and TGFBR3 (Table 3a). Following this initial genotyping, another 86 SNPs in over 20 genes of the TGF-β/BMP pathway and KL were typed. Another SNP in KL was found to be associated with leg ulcers along with seven SNPs in the TGF-β/BMP pathway in the genes BMPR1B, MAP2K, MAP3K7, SMAD7 (rs736839), SMAD9 and SMURF1 (Table 3).Since it is likely that the control group contains patients who might ultimately develop leg ulcers the reported odds ratios might be underestimates of the true association.
While a unique HBB mutation leads to the production of sickle haemoglobin (HbS), individuals homozygous for the HbS mutation display uncommonly heterogeneous phenotypes.(Steinberg, 2005) Leg ulcers are a common subphenotype of sickle cell disease. Their cause is unknown, their prevention is impractical and their management, once present, is often difficult.(Steinberg et al, 2001) We have hypothesized that the likelihood of developing leg ulcers, like other subphenotypes of sickle cell disease, is genetically modified, although the higher prevalence of leg ulcers in the tropics suggests that environmental factors are also important.(Christakis et al, 1990)
Recent studies of the effects of plasma haemoglobin on NO bioavailability suggested that some subphenotypes of sickle cell disease were related to the haemolytic rate.(Gladwin et al, 2004;Nolan et al, 2005) In a previous report from the CSSCD, coexistent α thalassaemia and elevated HbF afforded protection from the development of leg ulcers while reduced steady-state haemoglobin concentration was associated with a higher incidence of leg ulcers.(Koshy et al, 1989) Our analysis of this same patient group confirmed and extended these findings showing that leg ulceration in sickle cell anaemia were associated with higher reticulocyte counts, LDH, bilirubin and AST which are all markers of haemolysis.(Rother et al, 2005) Haemolytic anaemia varies in intensity among the different genotypes of sickle cell disease; it is most pronounced in sickle cell anaemia and progressively less severe in sickle cell anaemia-α thalassaemia, HbSC disease and HbS-β+ thalassaemia.(Steinberg, 2001) Consistent with this, leg ulcers are rare in the latter two genotypes. The higher leukocyte count in leg ulcer cases compared with controls could reflect an increased haematopoiesis associated with haemolysis although leukocytosis might also be a marker of more severe vasculopathy.
The severity of haemolytic anaemia determines the development of some complications of sickle cell disease because of its effects on NO bioavailability. Plasma haemoglobin liberated from intravascularly destroyed sickle erythrocytes consumes NO producing methaemoglobin and nitrate while catalyzing the production of oxidant radicals and certain adhesion molecules. NO, by binding soluble guanylate cyclase, converts GTP to cGMP relaxing vascular smooth muscle and causing vasodilation. With NO depletion, the normal balance of vasoconstriction:vasodilation is tipped toward vasoconstriction. NO availability is further diminished by the liberation of erythrocyte arginase that destroys arginine, the substrate for NOS and by reactive oxygen radicals that are generated in sickle cell disease.(Hebbel et al, 1982;Morris et al, 2005;Aslan & Freeman, 2004;Wood et al, 2005)
Haemolysis-related subphenotypes in sickle cell disease include stroke in children, priapism and pulmonary hypertension; leg ulcers can now be added to this list. Priapism, leg ulcers and pulmonary hypertension are seen in other types of haemolytic anaemia, albeit less frequently than in sickle cell disease, suggesting the ubiquitous role of haemolysis and the liberation of haemoglobin into plasma as a mediator of some complications of haemolysis.(Andrieu et al, 2003;Gyan et al, 2001;Dore et al, 1991;Macchia et al, 1990;Thuret et al, 1996;Daneshmend & Peachey, 1978;Giraldi et al, 2003;Stevens et al, 1977;Ohene-Frempong et al, 1998;Adams et al, 1994;Sarnaik & Ballas, 2001;Murray et al, 1988;Wu et al, 2004;Verresen et al, 1991;Derchi et al, 2005)
Studies of the genetic modulation of leg ulcers have been limited. The V34L G/T SNP (rs5985 reported in dbSNP as amino acid position 35) in the factor XIII gene (F13A1) was associated with leg ulcer size in 91 patients with ulcers of mixed etiology and the genotype correlated with FXIIA activity. Nevertheless, genotype distribution was similar in cases and controls.(Gemmati et al, 2004) In this same study, the factor V (F5) Leiden gene but not the prothrombin (F2) G20210A SNP was associated with leg ulcer. The common haemochromatosis mutation was associated with the likelihood of having chronic venous leg ulcers.(Zamboni et al, 2005) One of five SNPs in the 3′ untranslated region of the fibroblast growth factor receptor (FGFR2) was associated with leg ulcer in patients with chronic venous insufficiency.(Nagy et al, 2005) Sickle cell leg ulcers have also been related to the HLA-B3525 antigen.(Ofosu et al, 1987)
In chronic venous ulcer tissue compared with normal skin, there was an increased distribution, level and activity of NOS and arginase originating in inflammatory cells and the endothelium leading to the hypothesis that NO over-expression could, indirectly through the production of peroxynitrite, cause tissue damage and that arginase could enhance perivascular matrix deposition.(Abd-El-Aleem et al, 2000) However, increased arginase activity should reduce the concentration of arginine needed for optimal NO production.(Morris et al, 2005)
In this study we found significant associations (p-values <0.05) with 16 SNPs; 10 SNPs in 9 genes of the TGF-β/BMP signaling pathway, 4 in or near KL and 2 in TEK. Using the Benjamini and Hochberg algorithm to control the FDR yielded 1 significant SNP (rs736839) at an FDR or 10% and 10 significant SNPs at an FDR of 30% (Table 3). Though a few SNPs are identified as being significant by this method, genetic modulation of this phenotype is likely to be a multigenic trait so that each gene contributes but a small amount to the phenotype. Therefore, using FDR or the Bonferroni correction for multiple comparisons might be unduly stringent.
Four SNPs in or near KL were associated with an increased risk of leg ulcer by genotypic association analyses. We previously found association of KL SNPs and haplotypes with sickle cell stroke, priapism and osteonecrosis.(Nolan et al, 2004;Baldwin et al, 2005;Sebastiani et al, 2005) Klotho (13q12), a β-glucosidase-like protein, has both membrane-bound and secreted spliceoforms and directly or indirectly promotes endothelial NO production. In kl knockout mice, the vasodilator response of arterioles to acetylcholine was attenuated while levels of NO metabolites in urine were reduced. Recently, murine klotho protein was found to circulate as a hormone repressing signaling of insulin and insulin-like growth factor and suppress aging and increase resistance to oxidative stress. (Yamamoto et al, 2005;Kurosu et al, 2005) How KL might regulate the occurrence of leg ulceration and other complications of sickle cell disease is unknown but disordered NO and oxidative biology are features of the pathobiology of this disorder.(Jison et al, 2004)
The TEK receptor tyrosine kinase (also referred to as TIE2) has been linked to the phenotype of venous malformations as well as sickle cell anaemia associated stroke. (Boon et al, 1994;Vikkula et al, 1996, 2005;Sebastiani et al, 2005) Expressed almost exclusively in endothelial cells and involved in angiogenesis, it is the ligand for angiopoietin-1 (ANG1).(Hashimoto et al, 2004) The TEK signaling pathway appears to be involved in endothelial-smooth muscle cell communication in venous morphogenesis. Tek was upregulated and activated in the endothelium of healing rat skin wounds. Ptpn11, a protein tyrosine phosphatase and grb2, a growth factor receptor-bound protein, are part of the pathway upstream of mitogen-activated protein kinase activation, a pathway that possibly responsible for morphogenetic effects of tek on endothelial cells. Akt/phosphatidylinositol 3 kinase is a pathway downstream of tek necessary for cell survival, chaemotaxis, activation of endothelial nitric oxide synthase and perhaps for anti-inflammatory effects of tek activation.(Peters et al, 2004)
TGF-β binds its several receptors and signals via Smads, activating MAP kinases. This pathway, present in most tissues, has, among its many roles, effects on cell proliferation, apoptosis, response to tissue injury, infection, bone homeostasis, endothelial growth, diabetic nephropathy, pulmonary fibrosis, inflammation, immune regulation and extracellular matrix synthesis and is susceptible to pharmacologic modulation.(Attisano & Wrana, 2002;Miyazono et al, 2001;Ziyadeh, 2004;Moustakas et al, 2002)
Leg ulcers in sickle cell anaemia are a disease complication dependent on the intensity of haemolysis that is likely to be mediated by a reduction in bioavailable NO. SNPs in KL, a gene with a role in NO and oxidative biology, TEK, an endothelial-expressed gene and with genes in the TGF-β/BMP signaling pathway were associated with this phenotype. Haemolytic anaemia-induced phenotypes in sickle cell anaemia are likely to be improved by agents that increase NO bioavailability or reduce sickle erythrocyte density.(De Franceschi et al, 2000;De Franceschi et al, 1994) For example, arginine, the substrate of NO synthase, has been associated with a reduction in pulmonary artery pressure in sickle cell anaemia and sildenafil, an inhibitor of phosphodiesterase 5, might be effective in pulmonary hypertension and priapism.(Machado et al, 2005;Bialecki & Bridges, 2002;Morris et al, 2000;Chockalingam et al, 2005;Derchi et al, 2005)
This study may open new avenues for investigation of the pathophysiology of the microvascular pathology of chronic anaemias. Our findings could lead to novel, directed therapies for leg ulcers, which are at the present time, a most difficult therapeutic problem.
We thank the investigators of the Cooperative Study of Sickle Cell Disease who obtained blood samples for DNA-based studies and Stephen H. Embury, MD, who genotyped the α- and β-globin genes.
Supported by NHLBI grant HL R01 68970 (MHS). VGN was supported by T32 HL007501