Haplotypes, sub-haplotypes and geographical distribution in Omani patients with sickle cell disease

Despite the fact that patients homozygous for the sickle cell disease (SCD) mutation have an identical genotype, the severity of the disease can be extremely variable. The hemoglobin (Hb) S mutation has been described on five different haplotypes with different clinical expression. Identifying the genotypes, haplotypes and sub-haplotypes of the(cid:0) β gene cluster in Oman needs to be studied in more details to establish a correlation between the genotype/ haplotype and phenotype diversity observed in SCD patients for prognostic purposes, accurate diagnosis and thus planning for the best tailored treatment. We have investigated 125 HbS homozygotes from different parts of Oman and determined their haplotypes and sub-haplo-types and correlated this to the hematological and clinical expression. We have found 11 haplotype combinations differently distributed in the country, with the Asian/Asian HbS haplotype being the most predominant. Sub-haplo-types was only found among patients with CAR/OmanI haplotype. As expected, the correlation between haplotypes, sub-haplotypes and disease severity was mainly associated with HbF expression. Our study on haplotype/phe-notype correlation has shown which major haplotypes occur in the different regions of Oman. Furthermore, neither the haplotype or sub-haplotype


Introduction
Sickle cell disease (SCD) is one of the most common autosomal recessive disorders in human and was first described by Herrick in 1910. 1 The disease is caused by a single nucleotide transversion at codon 6 GAG>GTG (HBB:c.20A>T) (NM_000518.4) of the β globin gene resulting into the commonest hemoglobin variant worldwide (HbS) characterized by the single Glu®Val amino acid substitution at position β6. 2 Despite the fact that all patients homozygous for the HbS allele have an identical genotype, the severity of the disease can be extremely variable among affected subjects. 3 The disease may manifest with full blow severity, with chronic and acute infarctions in organs and tissues causing excruciating pain episodes (crisis), brain infarctions, splenic infarction, massive hemolytic events and acute chest syndrome with risk of premature death. Other cases however, may present with milder symptoms and the variability is mainly associated with the haplotype, sub-haplotype, a thalassemia 4 and the presence of fetal hemoglobin (HbF) (a2g2), 5 which may be attributed to the coinheritance of Xmn-I polymorphism. This marker is important for early prevention or reduction of morbidity in SCD patients treated with hydroxyurea therapy 6 or to decide whether or not bone marrow transplantation should be considered. 7 The HbS mutation has been described on five distinct haplotypes based on the presence or absence of the 5 different restriction enzyme sites in the β-globin gene cluster located on the 5' and two restriction enzyme sites on the 3' sides of the β gene (Table 1). [8][9][10][11] These haplotypes are known as Benin, Bantu or Central African Republic (CAR), Senegal, Cameroon and Asian. 8 The first four are African haplotypes, named after their origin and ethnic group 9 while the last was described in Central India and Saudi Arabia. 10 It has been previously reported that the CAR haplotype is usually associated with a more severe disease when compared with the intermediate phenotype of the Benin haplotype and to the milder conditions associated with the Senegal and Asian haplotypes. 11 Analysis of the polymorphic sites of the β genes cluster is of genetic, anthropologic and clinical interest, and it can also be used to predict the prognosis of the disease and to plan a tailored treatment.
The work reported here involves the investigation of a serial of polymorphic sites (SNP's) within the β globin gene cluster to identify the HbS haplotype of Omani patients. For this, we have selected 125 Omani SCD patients with homozygous HbS conditions and compared their haplotype with hematological and clinical data. Moreover, to look if there are any subhaplotypes within each known haplotype that might be associated with the clinical differences seen in patients with the same HbS haplotype, extra SNP's, in addition to the common ones have been studied as suggested by Marion et al. 12 Looking for correlations with the clinical phenotypes, we have also charted the distribution of the β gene haplotypes in different regions of the country and characterized the different haplotypes and sub-haplotypes using advanced molecular technologies.

Subjects
We have collected, with signed consent of patients and families, blood samples from a cohort of 125 SCD patients, whether admitted or following up in one of the Ministry of Health Hospitals in Oman. Gender distribution was; 84 males (45%) and 103 females (55%). The ages of the subjects were on average 36±5 years.

Strategy
Cation-exchange high performance liquid chromatography was performed on all samples on either D-10 (short and extended programs) device and/or Variant II (Bio-Rad Laboratories, Hercules, CA, USA) to measure the rate of HbF/HbS 13 in absence or before blood transfusion. DNA was extracted using the Qiagen kit according to the manufacturer instruction as previously described. 14 47 SNP's, covering 13 sites in the β-globin gene cluster that are known to be variable and informative were analysed ( Figure 1). This was carried out by an asymmetric polymerase chain reaction (PCR) 12 and analysed by melting curve analysis (MCA). 15 The common haplotyping of the βglobin gene cluster was determined according to the presence (+) or absence (-) of a composition of SNP's corresponding to the traditional five 5' known polymorphic restriction endonulease sites (F1, F2 (SNP 2), F3 (SNP 2), F4 and F5). 12 The remaining 42 SNP's were used to look for sub-haplotypes to find out if there are nucleotide variations within similar haplotypes that might be associated with the phenotypical differences observed.

Genotyping procedures
Genotypes of PCR products were determined using the Light Scanner (Idaho Technologies Inc., Salt Lake City, UT, USA) after performing asymmetric PCR with LCGreen fluorescent DNA dye and unlabeled oligonucleotide probes. 12 The PCR was performed asymmetrically so that the strand complementary to the probe is produced in excess, allowing probe annealing at the SNP site. The primers were designed to yield a product not larger than 200 bp as it is the optimal length for accurate scanning and genotyping with MCA. A fluorescent dye (LC Green plus) that emits light in the presence of double stranded DNA was added to the reaction and as the temperature increases, the fluorescence decreases as the dsDNA melts out; this produces a characteristic melting curve. 15 Unlabeled oligonucleotide probes were designed and used to genotype targeted sequence variations. These probes increase specificity of the melting reaction as it decreases the size of the product that is melted. Therefore, the probe is designed to anneal to either the wild type or the mutant allele; the characteristic melting curves identify the genotype of each sample. 16 The shape of the PCR amplicon melting curve reveals the presence or absence of the SNP in comparison to the wild type sequence, allowing clear recognition and genotyping. 15 The haplotype/sub-haplotype is then drawn from the obtained genotypes. In case of homozygosity for all markers, the haplotype can be defined. In the case of a single SNP difference, two distinct haplotypes can be defined. In case of two or more SNP differences, the most likely/frequent haplotype in the population is defined. SNP's that were doubtful were sequenced by Sanger sequencing for confirmation.

Phenotype classification
In order to define disease severity, a number of well-defined clinical parameters were analysed: hemoglobin level, frequency of transfusion which is based on episodes of acute hemolysis (e.g. pulmonary hypertension, jaundice, gallstones, splenic crisis), number of annual hospitalization, frequency of vaso-occlusive crisis, splenectomy (indicated by acute splenic sequestration and chronic hypersplenism), acute chest syndrome (ACS), body pain (e.g. abdomen, chest, bones, joints, episode of dactylitis) and records of any major organ damage such as heart and liver ( Table 2). The severity of disease expression was then correlated with the haplotype and hematological parameter readings of HbF and HbS.

Clinical severity
The phenotypes of the patients were classified into mild, intermediate and severe based on the described criteria (Table 2). Data were obtained from patient's medical records anonymously provided by the doctor. Among the Asian/Asian haplotype, mostly were presented with a mild disease and none with a severe form. Conversely, Benin/Benin ranged from severe to intermediate while none presented with the mild form. Among the Asian/OmanI haplotype, the percentage was equal between mild and intermediate. The majority of homozygous CAR and compound heterozygous CAR/OmanI β cluster had a severe clinical profile. Finally, the phenotype of Asian/CAR haplotype ranged from mild to intermediate and even severe cases. Data are summarized in Table 3.

Hematological data
The Hb F value correlates with the different haplotypes, giving a direct indication of the disease severity. The average hemoglobin values, HbF and HbS percentage of each haplotype are summarized in Table 4. The Asian haplotype, being associated with a mild phenotype, presented with the highest expression of HbF% and the lowest HbS% when compared to the other haplotypes. Patients with Asian/OmanI haplotype had a mild to intermediate phenotype. This can be attributed to the elevated expression of HbF% (Table 4). Conversely, the homozygous CAR and compound heterozygous CAR/OmanI haplotypes had the most severe clinical picture with the lowest Hb F values (Table 4).

Geographical distribution
The overall distribution of each haplotype in our SCD patients in the different regions of Oman is presented in percentages in Table 5. The occurrence frequency of each haplotype in a particular region is depicted in Figure 2. In Musandam, CAR/Asian is the only haplotype found. In Batinah, Asian/OmanI is the most abundant. In Muscat, CAR/OmanI is found at the highest frequency and in Sharqiya, the Asian haplotype was the most prevalent. In Dhakhiliya, Benin was the major haplotype found and finally in Dhahira, the CAR haplotype was the most prominent. Data are summarized in Figure 2.

Sub-haplotypes
Based on additional SNP's, a subdivision could be made from the original five HbS haplotypes. Trying to find a molecular explanation for the different phenotypes seen within similar basic haplotypes, sub-haplotypes were determined by looking at a total of 42 SNP's in all the 125 homozygous HbS patients. Out of the 42 SNP's, only 15 SNP's were found modifying the 11 identified haplotypes. However, no sub-haplotypes were found to be associated with a specific haplotype except for the CAR/OmanI that showed nucleotide variations at the G-g (SNP1) (SNP position: 5232979-5232984) located in the G-g promoter region Article Table 3. Summary of the genotypes from which the 11 haplotypes were determined in the Omani HbS/S homozygous patients using the five 5' traditional sites. The Xmn-I genotype and percentages of mild, intermediate and severe cases within each haplotype is also indicated.    from which three different sub-haplotypes were defined (a, b and c) ( Table 6). This site might be associated with the different clinical expression observed in CAR/OmanI patients. The patient with sub haplotype (CAR/OmanIc) presented with a very severe clinical manifestations and no improvement was seen even after doubling the dosage of hydroxyurea. This patient appears to be homozygous for the wild type 6 nucleotides (CTTTAA) at the G-g promoter. On the other hand, another patient with sub haplotype (CAR/OmanI -b) was homozygous for the deletional mutation of the 6 nucleotides at the same position and had a milder clinical presentation and was not taking hydroxyurea consistently as the patient was feeling better after a short period of being on the drug. The remaining patients with subhaplotypes (CAR/OmanI -a) had the compound heterozygous composition of CTTAA/6nt del at the G-g promoter region, needed higher than the average dose of hydroxyurea in order to observe a reduction of the severe phenotype.

Haplotype HcII (5' ε) Xmn-I Hd III (Gg) Hd III (Ag) Hc II (Ψβ) Hc II (3'Ψβ) Ava II (β) Bam HI (3'β) Total Mild Intermediate
Although more case studies are needed to confirm our hypothesis, our data allow us to assume that carrying the 6-nucleotide deletion might be beneficial to CAR/OmanI patients for better response to hydroxyurea.

Discussion
Haplotypes distribution and disease severity The Omani populations are known for their high incidence of hemoglobinopathies, including a and β-thalassemia as well as sickle cell disease. 14 As mentioned above, patients with SCD present with a variable clinical picture ranging from severe to very mild forms, where haplotypes have been found to be associated with the severity of the disease. 17,18 In Daar et al., a study conducted in Oman in 2000 on 52 HbS/S individuals, it was found that the Benin/Benin haplotype was the most prevalent and twice more frequent than the Asian/Asian. 19 In the present study however, the Asian/Asian haplotype was the most prominent while Benin/Benin was the second in rank ( Table 3). The reason for the higher per-centages of Benin haplotype in Daar et al., might be due to a selection among patients attending the Sultan Qaboos University Hospital which are mainly from the Dhakhiliya region and based on our findings, the Benin haplotype has been observed to be present at a high rate in this region (Figure 2). Although the effect of these haplotypes on the phenotypes is clearly correlating with the HbF Article Table 5. The frequency distribution of the 6 most common haplotypes in 6 different regions in Oman.  expression, disease severity remains variable within the same haplotypes and more molecular and external factors need to be taken into consideration. The high frequency of a-thalassemia reported in the Omani population 14 is another modulating factor influencing the clinical outcome of the disease. The effect of a thalassemia on the clinical expression of SCD is under evaluation in our cohort and will be presented in a following paper.

Gene flow
The distribution of the HbS haplotypes in Oman is explained by the historical migrations from Zanzibar and India. The presence of the Asian haplotype can be attributed to ancient migrations and to centuries of trade with India and Pakistan. Contacts with East Africa, Zanzibar and Mombasa, explain the presence of the Benin and CAR haplotypes. 19 Muscat, being the capital, had the widest diversity of different haplotypes. This reflects more recent migrations of native people from the interior to the capital seeking for jobs and better lives. The Asian haplotype was highest in Sharqiya and Batinah and these regions are known to have SCD patients with a mild clinical profile in comparison to Dhakhiliya and Dhahira regions in which they have a more severe manifestation of the disease and this could be explained by the Benin and CAR haplotypes in these regions respectively ( Figure 2). Only one haplotype combination was found among patients from Musandam (CAR/Asian) and this can be due to the isolation of this region from the rest of the country by mountains and the UAE. SCD is absent in Wusta and Dhofar due to low levels of malaria in the past in these two regions. 20 The 4 identified haplotypes, which we have referred to as OmanI, II, III and IV are expected in an admixed population such as Oman and might have been derived differently by recombinant events. 21 OmanI could have probably been derived from CAR with a mutation at F5. It is also possible to say that it is a result from a recombination event between CAR and Cameroon haplotypes but this is less likely as Cameroon haplotype was not found in the Omani population. Oman II could be a result of a recombinant event between Benin and CAR while OmanIII could be the outcome of a recombination between Asian and CAR haplotypes. Oman IV could be a derivative of OmanII with a mutation at F1. However, these are just assumptions and further studies are eventually required to track back the origin of each haplotype.

Early diagnosis, haplotype, hemoglobin F and prognosis
Identifying the disease at an early stage and defining genotype and haplotype allows clinicians to predict to some extent the prognosis and to plan a tailored treatment. Early prediction of the clinical expression will help in preventing or reducing acute painful episodes (crisis) in this cohort, which is the most common traumatic experience in SCD, 22 and ACS, which is the most common cause of death in Omani SCD patients. 23 The Asian haplotype was associated with highest HbF levels, fewer hospitalizations and painful episodes and patients did not develop acute chest syndrome although vaso-occlusive events did occur. Our study also confirms that the CAR haplotype whether homozygous or combined heterozygous is associated with lowest HbF level and the highest incidence of organ damage and renal failure as reported elsewhere. 24,25 Carriers of the HbS gene on the Asian haplotype on one chromosome and OmanI haplotype on the other presented in our cohort high HbF levels (average 14.2%) and a milder clinical course than other compound heterozygous haplotypes. Bakioglu et al. 26 reported mild SCD cases of Asian/Benin haplotype with high levels of Hb F (average 22.2%). This shows that carrying the HbS mutation on an Asian haplotype on one chromosome could still contribute to elevating the HbF expression.
Higher concentrations of HbF in the cell lead to lower concentrations of HbS, 27 better oxygenation and less clinical severity. A potential threshold of 20% HbF has been suggested to effectively prevent recurrent vasoocclusive episodes. 28 This is true in most cases, however, in another study, some patients with HbF levels near 20% had a devastating disease manifestation. 22 The same observation was seen in one of the patients in our cohort, with the Asian/CAR haplotype. Despite the high HbF (19.7%), the patient is frequently admitted to the hospital with vaso-occlusive crisis, dactilitis and severe abdominal pain. This patient has no iron overload and is heterozygous for the a-3.7 deletion. This finding imply that other circumstantial or genetic factors or external transacting determinants such as blood viscosity, elevated packed cell volume, vascular adherence, acidosis and dehydration as well as patient's life style and diet might contend with the beneficial effect of the high HbF level. Our findings support the conclusion by Acquaye et al. 29 and Seltzer et al. 30 that fetal hemoglobin levels are very important but not the only parameters that mitigate the severity of the disease and we are at the moment inquiring which other factors could be associated with the severity of this case.

Sub-haplotype
The sub-haplotype study of 42 SNP's on the 11 haplotype combinations observed in our 125 SCD patients revealed that 15 different positions differentiate the 11 identified haplotypes. No sub-haplotypes were determined except in patients with the CAR/OmanI haplotype, which revealed variation at the G-g promoter SNP-1 giving 3 different sub-haplotypes (Table 6). One patient with sub-haplotype-c who presented with homozygosity for the wild type 6 nucleotide sequence (CTTTAA) at the Gg promoter, a very severe phenotype and not responding to high doses of hydroxyurea treatment. This element could be the cause of low HbF expression and of non-response to hydroxyurea and this hypothesis seems to be sustained by the fact that another patient with sub-haplotype-b with a homozygous mutational deletion of the 6 nucleotides at the same position became better as soon as being on the drug. The presence or absence of the 6nt sequence in at the G-g promoter in the CAR/OmanI patients could be associated with the differences in clinical presentation and with response to hydroxyurea therapy, probably an element is in linkage to the 6nt deletion and is responsible for the positive response to the drug.

Conclusions
Our study on haplotype/phenotype correlation has shown the existence of at least 11 different haplotype combinations in Oman. These are differently distributed among the six main regions of the country. Sub-haplotype was only observed in CAR/OmanI combination and could be associated with the clinical differences observed in patients with the same haplotype. Identifying haplotypes and sub-haplotypes in early life may allow a better prognosis and a more accurate risk predictions and a better-tailored therapy, to match disease-related risks and to facilitate planning of clinical trials to prevent the development of severe complications later in life. Nevertheless, we have shown that when the phenotypes are classified into; mild, intermediate and severe, neither the haplotype or the HbF alone appeared to be fully associable with the clinical phenotypes as also been observed by Alexander et al. 31 External and/or modifying, or epistatic factors, which potentially modulate the phenotype of SCD do occur and more efforts should be done trying to chart them. The implementation of primary prevention with simple cost effective interventions for SCD are essential 32 and are likely to lead to lower incidence, lower costs for public health and improved survival rate of SCD patients in Oman.