Therapeutic challenges in childhood sickle cell disease. Part 1: current and future treatment options.

Abstract

Sickle cell disease (SCD) is one of the commonest inherited diseases in the UK, affecting approximately 1 in 4000 live births every year. For the majority of patients, the mainstays of treatment are preventative and supportive. For those children with severe SCD, three major therapeutic options are currently available: blood transfusion, hydroxyurea and bone marrow transplantation. This review focuses on the relative roles of these therapeutic modalities in severe paediatric SCD and assesses the prospects for new treatment modalities, including non-myeloablative stem cell transplantation, short chain fatty acids, membrane active drugs and gene therapy. After decades with little new to offer patients with sickle cell disease (SCD), there have been several advances in the last 10–15 years which have contributed both to improved quality of life and to an increase in life expectancy, at least for those patients with access to treatment in developed countries. This two-part review outlines the roles of different treatment options in severe paediatric sickle cell disease (SCD). In the first part, we review the three principal current therapeutic modalities [blood transfusion, hydroxyurea and bone marrow transplantation (BMT)] and potential future treatments, and in the second part we describe an evidence-based, problem-orientated approach to some of the major complications of SCD in childhood. Red cell transfusion is widely used in the management of SCD. Approximately 50% of all patients receive a red cell transfusion at some stage in their lives and 5% are on chronic transfusion programmes (Rosse et al, 1990). Nevertheless, few controlled trials have rigorously evaluated either the indications for transfusion in SCD or the best protocols to use. The requirements for red cell transfusion in patients with sickle cell disease have been reviewed elsewhere (Davies & Roberts-Harewood, 1997; Ohene-Frempong, 2001; Vichinsky, 2001). The major indications for transfusion in SCD are summarized in Table I. The main indications for ‘top-up’ red cell transfusion in children are severe anaemia complicating acute splenic or hepatic sequestration (Emond et al, 1985) and aplastic crises due to parvovirus B19 infection (Serjeant et al, 2001). In this situation, the haemoglobin should be raised to the child's steady state (it should never be raised acutely to > 10 g/dl as this may cause an increase in blood viscosity). For most other severe complications of SCD, exchange rather than top-up transfusion has significant advantages. For severe acute chest crises with hypoxia despite continuous positive airway pressure (CPAP) or mechanical ventilation, exchange transfusion is the treatment of choice to reduce sickling and increase oxygen carriage without an increase in viscosity Schmalzer et al, 1987; Emre et al, 1995). Despite a lack of controlled studies, acute exchange transfusion has been shown in observational studies to prevent pulmonary complications, shorten the duration of the acute illness and reduce mortality in children with acute chest syndrome (Lanzkowsky et al, 1978; Emre et al, 1995). Exchange transfusion is also used in the acute management of SCD patients presenting with a new stroke or transient ischaemic attacks. Although widely practised and supported on the theoretical grounds of improving perfusion and oxygenation in the region of the infarct, the value of red cell transfusion in the acute management of stroke is unclear as it may not influence long-term neurological outcome (Ohene-Frempong, 1991). Exchange transfusion appears to improve survival in the acute multi-organ failure which occasionally complicates the course of SCD, although there are few data available for children (Hassell et al, 1994). Exchange transfusion, rather than top-up transfusion, is also the preferred option for hyperhaemolysis secondary to malaria (Newton et al, 1997) as it enables the removal of infected and damaged red cells as well as treating the anaemia. Controversial uses of red cell transfusions for acute episodes in childhood SCD include their role in ameliorating prolonged vaso-occlusive crises and in treating priapism. There is no evidence that red cell transfusion reduces the severity or duration of established painful vaso-occlusive crises. The evidence that red cell transfusion is useful in the management of priapism is anecdotal and mostly derives from studies carried out more than 20 years ago (Ohene-Frempong, 2001). In addition, it is quite frequently ineffective (McCarthy et al, 2000) and may be associated with severe neurological events (Rackoff et al, 1992). On the basis of anecdotal evidence and reports of small series of patients, many centres give pre-operative red cell transfusions with the aim of reducing complications in patients with SCD undergoing anaesthesia and surgical procedures. The largest study to examine the role of transfusion in the preoperative management of SCD was a randomized study comparing exchange transfusion (to achieve Hb of > 10 g/dl and HbS fraction of < 30%) with simple transfusion (to a Hb of > 10 g/dl) (Vichinsky et al, 1995). This showed no difference in the frequency of post-operative sickle chest syndrome, fever, infection or painful crises between the two treatment arms, while allo-immunization and haemolytic transfusion reactions occurred more frequently after exchange transfusion (Vichinsky et al, 1995). This indicates that where peri-operative transfusion is required, ‘top-up’ transfusion to produce a Hb of around 10 g/dl would be the preferred option, exchange transfusion being limited to high-risk surgery (such as organ transplantation, hip/knee replacement and eye surgery) and to patients with a high baseline haemoglobin. The question of which procedures are safe to carry out in SCD children without pre-operative red cell transfusion remains controversial because of a lack of evidence from controlled clinical trials (Riddington & Williamson, 2001); however, there are data from small series which indicate that, depending on the prior clinical course of the individual patient, minor and straightforward procedures (such as tonsillectomy and cholecystectomy) can often be safely undertaken without transfusion (Griffin & Buchanan, 1993; Hatley et al, 1995; Davies & Roberts-Harewood, 1997; Haberkern et al, 1997). The main indications for chronic transfusion in children with SCD are the prevention of recurrent stroke (Pegelow et al, 1995) and, more recently, in the prevention of a first stroke in children with abnormal blood flow identified by transcranial Doppler ultrasonography (Adams et al, 1998). In children with sickle-related stroke, chronic transfusion reduces recurrence by about 90% (Pegelow et al, 1995). Most patients are commenced on a hypertransfusion regimen (aiming to maintain the HbS below 25% and the Hb between 10 and 14·5 g/dl) but exchange transfusion may be used to minimize iron overload (Cohen et al, 1992; Kim et al, 1994). The optimal duration of therapy has not been determined as recurrence has been reported even after 12 years of transfusions (Wang et al, 1991). Chronic transfusion programmes have also been used to prevent recurrence of acute chest syndrome, to reduce the frequency of painful crises and in chronic heart failure or renal failure, although there are no clinical trials to confirm their efficacy in children (Davies & Olatunji, 1995; Ohene-Frempong, 2001). The strongest evidence for the use of chronic transfusion to reduce painful crises is a prospective randomized trial in pregnant patients with sickle cell disease, in which there was a significant reduction in the frequency of vaso-occlusive crises in the patients who were prophylactically transfused (Koshy et al, 1988). Since the introduction of hydroxyurea (see below), chronic transfusion for recurrent crises should be considered only as a last resort in view of the risks associated with long-term transfusions (Castro, 1999). The most common serious complications of transfusion in children with SCD are iron overload, allo-immunization and transfusion-transmitted infections, though problems with vascular access are also an extremely common practical problem especially for children on chronic transfusion programmes. As in thalassaemia major, iron overload causes significant morbidity and mortality in SCD patients on chronic transfusion programmes (Ballas, 2001). In the recent study of 371 adult patients with SCD treated at Thomas Jefferson University Hospital in Philadelphia, 50% of the in-patients were transfused and around 10% of these patients were iron overloaded (Ballas, 2001). Mortality was significantly higher in the iron-overloaded patients (64%vs 5%), as was the proportion of patients with organ failure (71%vs 19%) (Ballas, 2001). There are no specific data for the impact of iron overload on morbidity and mortality in children with SCD. Iron overload can be minimized by limiting red cell transfusions to defined indications, by using manual or automated exchange transfusion rather than hypertransfusion (Kim et al, 1994), by providing intensive support to encourage compliance with desferrioxamine (Treadwell & Weissman, 2001) and by ‘tailored’ iron chelation regimens based on accurate assessment of tissue iron-associated damage (Anderson et al, 2001; Brittenham et al, 2001) and judicious use of the oral iron chelator, deferiprone (Olivieri et al, 1998; Wonke et al, 1998; Pippard & Weatherall, 2000; Ceci et al, 2002; Maggio et al, 2002). The development of red cell allo-antibodies, most frequently anti-K, anti-E or anti-C, was reported in 18–36% of SCD patients (Vichinsky et al, 1990). Red cell allo-immunization makes compatibility testing for future transfusions difficult and increases the risk of life-threatening, delayed haemolytic transfusion reactions (Syed et al, 1996). The rate of allo-immunization is related to the number of units of blood received but is reduced by routine administration of phenotypically matched units at least for K, E and C (Vichinsky et al, 1995, 2001). All SCD patients should, therefore, have extended red cell phenotyping, including ABO, Rh, K, Kidd, Duffy, Lewis and MNSs systems, soon after diagnosis and before transfusions are started so that red cells phenotypically matched for ABO, Rh and K can be selected as required (Davies & Roberts-Harewood, 1997; Olujohungbe et al, 2001; Vichinsky et al, 2001). Children on regular transfusions are also at risk of transfusion-transmitted infections, although the rate of transmission has reduced in recent years, at least in well-resourced countries (Murphy et al, 2001; Vichinsky, 2001). Hepatitis B and C remain the most serious risk: an American study showing that 10% of adults with SCD were infected with hepatitis C (Hasan et al, 1996). The risk of hepatitis B is estimated at 1 in 63 000 units in the USA (Vichinsky, 2001) and 1 in 50 000–170 000 units in the UK (Murphy et al, 2001), making hepatitis B vaccination an important part of management for all children with SCD. Repeated transfusions are also associated with a significant risk of transfusion-related lung injury and post-transfusion hyperhaemolysis characterized by destruction of both autologous and transfused red cells with negative serological findings (Cullis et al, 1995). Hydroxyurea exerts its beneficial effects in SCD via a number of mechanisms, including inhibition of intracellular polymerization of HbS by mixed hybrid molecules (α2βS.γ) with higher solubility than HbS, modification of red cell–endothelial interactions and the rheological properties of HbS-containing red cells, and via its myelosuppressive effects, particularly on neutrophils. Its development, pharmacology and use in SCD have been the subject of a recent comprehensive review (Halsey & Roberts, 2003). Here we focus on the therapeutic aspects in children with SCD. Clinical trials of hydroxyurea in children with SCD. There have now been more than a dozen published studies of hydroxyurea in children with SCD, although most involved small numbers (Ferster et al, 1996; Jayabose et al, 1996; Scott et al, 1996; Kinney et al, 1999; Koren et al, 1999; de Montalembert et al, 1999; Wang et al, 2001; Ware et al, 2002; see also Halsey & Roberts, 2003). The main findings of the four largest series are shown in Table II. Ferster et al (2001) recently reported the long-term follow-up on a cohort of paediatric SCD patients treated with hydroxyurea from the Belgian National Registry. There were 93 patients (87 children) with severe SCD (median age of 7 years) with a median follow-up of 3·5 years. The number of admissions to hospital and the number of days of hospitalization both dropped significantly in the patients on hydroxyurea, so that for those who completed 5 years of therapy over half had no significant vaso-occlusive crises during therapy. There was also a reduced frequency of acute chest syndrome (nine events) and splenic sequestration (one event) compared with that expected from historical series. These clinical benefits were paralleled by significant rises in the haemoglobin levels and HbF levels (% HbF), and no important adverse events were noted. The actual dose of hydroxyurea given in the study was usually less than 25 mg/kg/d and, as found in the multicentre study of hydroxyurea in sickle cell anaemia (MSH) in adults, the clinical benefit of hydroxyurea was seen without escalating the dose of hydroxyurea to the maximum tolerated dose (MTD) (Charache et al, 1995). Other studies in children have given similar results, with reductions in hospital admissions and adverse events (Jayabose et al, 1996; Koren et al, 1999; Maier-Redelsperger et al, 1999). Growth and development also appeared to be unaffected (Ware et al, 2002). Jayabose et al (1996) treated 14 children with a history of three or more painful crises or recurrent chest syndrome in the past year or with a Hb < 7 g/dl with escalating doses of hydroxyurea. The frequency of vaso-occlusive crises fell from 2·5 crises per patient-year prior to hydroxyurea to 0·87 crises per patient-year during treatment (a 65% reduction) with a median rise in Hb of 1·9 g/dl. However, given the open-label nature of both these studies, some of this benefit may have been due to more rigorous supportive care and/or placebo effects. de Montalembert et al (1999) have likewise reported a decrease in the number of painful crises in 27/28 children treated with hydroxyurea. In the larger, multicentre phase I/II trial of hydroxyurea in children with sickle cell anaemia (HUG-KIDS) study (Kinney et al, 1999), 84 children between the ages of 5 and 15 years with a history of three or more painful crises in the previous year or recurrent chest syndrome were treated with hydroxyurea. This study did not address clinical efficacy, instead the major end-points were haematological response and safety. Hydroxyurea significantly increased haemoglobin levels, percentage HbF and percentage F-cells. The rise in percentage HbF was greater in children who had a higher baseline HbF level and Hb concentration, and in those who received a higher MTD (Ware et al, 2002). The most commonly observed toxicity was transient myelosuppression: no severe adverse effects occurred and there was no effect on growth. A number of other small-scale studies have been conducted in children with similar results (reviewed by Maier-Redelsperger et al, 1999). Taken together, these data indicate that hydroxyurea is likely to decrease the frequency of vaso-occlusive crises, acute chest syndrome and transfusion requirements in children at least as effectively as it does in adults. There is no evidence yet that hydroxyurea reduces the recurrence or development of stroke in children or in adults. Similarly, there is no evidence in children that hydroxyurea reduces the SCD-related mortality. This is perhaps not surprising given the relatively low mortality of SCD in childhood and the fact that a significant reduction in mortality in adults was seen only after several years of follow-up in the MSH study (Steinberg et al, 1999). Hydroxyurea in very young children with SCD. The role of hydroxyurea in this age group remains to be defined. Two small studies have been reported. Hoppe et al (2000) treated eight children aged 2–5 years (median 3·7 years) and found no unexpected toxicity, and normal growth and development. However, there were no patients with severe SCD and efficacy could not be assessed in such a small study (Hoppe et al, 2000). A second study (Wang et al, 2001) treated even younger children (aged 6–28 months) with homozygous SCD or S-β0thalassaemia to address whether early use of hydroxyurea might be useful in the primary prevention of organ damage. This study revealed no adverse effects of hydroxyurea in a group of 28 children, and confirmed the preservation of normal growth and development reported in older children. There was also a possible beneficial effect on preservation of splenic function. Unfortunately, two of the 28 children experienced neurological events despite being maintained on hydroxyurea (Wang et al, 2001). Thus, starting hydroxyurea early in life did not prevent major complications of SCD and, until further data from randomized controlled trials are available, there is no indication for the use hydroxyurea in this way in very young children. Factors that predict response to hydroxyurea. The clinical and haematological response to hydroxyurea is variable both in adults and children. In part this may reflect patient compliance, although this can be difficult to assess. In the HUG-KIDS study, baseline HbF level, baseline Hb level, the MTD and compliance were all significantly associated with a higher rise in percentage HbF in hydroxyurea-treated children (Ware et al, 2002), and other studies have shown an association between higher baseline neutrophil and reticulocyte counts and absence of the Bantu (CAR) haplotype with above-average increases in HbF levels. While these studies suggest that, at a population level, selected baseline laboratory parameters, a higher MTD and attention to compliance may be useful in predicting the HbF response to hydroxyurea, the clinical and haematological response to this drug is complex and variable so that these parameters cannot be used to predict a response in an individual patient. Toxicity of hydroxyurea in children. The short-term side-effects of hydroxyurea in children are the same as those in adults. The commonest side-effects are dose-dependent myelosuppression, which is usually transient but occasionally more prolonged (Vichinsky & Lubin, 1994), nausea and vomiting, and skin rashes (Kinney et al, 1999; de Montalembert et al, 1999). In the long term, the most worrying potential risk of hydroxyurea in children is that of leukaemogenicity. While data from hydroxyurea-treated children with cyanotic heart disease are reassuring (Triadou et al, 1994), two cases of haematological malignancy have been reported in children with SCD treated with hydroxyurea: a 10-year-old with Ph-positive ALL (Ferster et al, 2001) and an 8-year-old with Hodgkin's disease (Moschovi et al, 2001), although the duration of treatment (7 weeks and 6 months respectively) makes it unlikely that these malignancies were secondary to hydroxyurea therapy. No cases of haematological malignancy have been reported in the adults in the largest series, the MSH study, suggesting that the risk of leukaemic transformation is low. Nevertheless, continued caution and long-term follow-up is essential (de Montalembert & Davies, 2001), and it may be relevant that recent in vitro studies have shown an increased mutation rate in children treated with hydroxyurea for 7–30 months, as measured by the VDJ recombination assay (Hanft et al, 2000). Bone marrow transplantation (BMT) remains the only curative therapy for SCD. The role of BMT in managing severe SCD is discussed in detail in Part 2 of this review: summarized here are the criteria used to identify the patients most likely to benefit from BMT, the conditioning regimens used and the outcome of BMT. Patient selection is aimed at identifying individuals with SCD who will benefit most from BMT, while excluding those at an unacceptably high risk of transplant-related morbidity and mortality due to pre-existing organ damage. This underlines the importance of a rigorous pretransplant work-up, including formal assessment of pulmonary, cardiac, hepatic and renal function, as well as detailed neurological assessment with magnetic resonance imaging, angiography and neuropsychometric studies. The current British Paediatric Haematology Forum criteria for selection of patients with SCD for BMT (Davies, 1993) are shown in Table III. The Seattle collaborative study have similar inclusion criteria but specify that patients undergoing BMT for recurrent painful vaso-occlusive crises should have had > 2 episodes per year for 3 years consecutively, and also broaden the indications to include sickle nephropathy with a glomerular filtration rate (GFR) of 30–50%, bilateral proliferative retinopathy and major visual impairment, osteonecrosis of multiple joints, and red cell allo-immunization (> 2 antibodies) during long-term transfusion therapy (Walters et al, 1996a). It is estimated that fewer than 10% of children with SCD fulfil these criteria, of whom only one in five will have a matched sibling donor (Davies & Roberts, 1996). Among 4848 patients that were < 16-year-old reported to the Seattle collaborative study, 315 (6·5%) met the entry criteria, of these only 41% had tissue typing performed (24% had no sibling available) and 14% had a human leucocyte antigen (HLA)-identical sibling (Walters et al, 1996a). Thus in this study, the major barrier to BMT for SCD was lack of an HLA-identical donor rather than parental/physician refusal or lack of financial/psychosocial support. Nonetheless, for those patients who are eligible and do have an available donor, the paediatric haematologist needs to be sensitive to cultural issues, particularly the risk of infertility, which clearly do influence the decision on whether to transplant. Approximately 150 patients with SCD, nearly all < 16 years of age, have been transplanted worldwide. The majority of these have been reported in three major series, the results of which are summarized in Table IV. The Seattle collaborative group, which includes the Hammersmith and Birmingham Children's Hospitals in the UK (Walters et al, 1996b, 2000), and the French group (Bernaudin et al, 1997) have reported a total of 76 children transplanted for symptomatic SCD. In contrast, the Belgian group (Vermylen et al, 1998) has reported the results of 36 children transplanted because of previous morbidity, but also 14 asymptomatic patients transplanted at a much younger age because they were to return to countries where medical care was not optimal. All patients received conditioning with busulphan 14–16 mg/kg or 485 mg/m2 and cyclophosphamide 200 mg/kg. In the initial French and Belgian patients, no serotherapy was used but, in view of high early rates of mixed chimaerism and rejection, most groups have now incorporated pretransplant antilymphocyte globulin or Campath into their conditioning regimens (Walters et al, 1996b;Bernaudin et al, 1997). Survival. The results show projected overall survival of 92–94% and event-free survival of 75–84% at 6–11 years in the three series. Deaths were from complications of acute and chronic graft-versus-host disease (GVHD; n = 5), intracranial haemorrhage (n = 1) and sudden death 6 years post BMT (n = 1). All patients with stable, predominantly donor, engraftment became free of the clinical manifestations of SCD. In the Belgian series, the results with the asymptomatic patients were superior to those transplanted for symptomatic disease, with overall survival of 100% vs 88% and event-free survival of 93% vs 76%: in view of this, they have proposed giving parents the option of early transplantation regardless of symptomatology if a matched sibling donor is available. Engraftment, rejection and chimaerism. Primary graft failure was uncommon (3% overall), but the cumulative incidence of graft rejection was 10–18%, which in all but one patient was accompanied by autologous reconstitution. Several of the patients who rejected their grafts have developed increased HbF concentrations, which ameliorated their disease severity for several years (Ferster et al, 1995). No clear risk factors have been associated with rejection but, in the French series, inclusion of serotherapy in the conditioning regimen decreased the incidence of mixed chimaerism/rejection from 25% to 7%. Both the Seattle collaborative and the Belgian studies have reported a number of patients with stable mixed chimaerism, all of whom remained asymptomatic. Toxicity. Patients with SCD are at increased risk of neurological complications after BMT, particularly seizures and intracranial haemorrhage (Walters et al, 1995). In their initial cohort, the Seattle collaborative study reported neurological complications in seven of 21 patients, with seizures in six patients and intracranial haemorrhage in three patients. The overall incidence of neurological complications was not different in patients with or without pretransplant stroke (this may reflect subclinical vasculopathy in those without a prior history of stroke), but mortality (25% vs 0%) and intracranial haemorrhage (38% vs 0%) were higher in the former. However, since implementing prophylactic measures, including a higher platelet transfusion threshold, and rigorous control of blood pressure, magnesium and cyclosporine levels, there have been no further cases of intracranial haemorrhage although the high incidence of seizures persists (21%). With the exception of neurological complications, post-transplant complications are similar to those seen after BMT for β-thalassaemia. Significant acute GVHD (> grade 2) has been reported in approximately 20% of patients but this was rarely severe (Bernaudin et al, 1997; Vermylen et al, 1998). Chronic GVHD occurred in 15–20% of patients, was extensive in 6–8% of patients and is clearly a major cause of post-transplant mortality, accounting for four out of a total of seven deaths in the three published series. Long-term effects and quality of life. The effect of BMT on growth is complicated by growth retardation by SCD itself and transfusion-related iron overload. Based on the experience of patients transplanted with busulphan/cyclophosphamide conditioning regimens for other indications, significant growth impairment is not anticipated (Giorgiani et al, 1995). The Belgian group reported continuing growth along the centiles in all but two patients, who received long-term immunosuppression for chronic GVHD. In their late-effects cohort, the Seattle collaborative group observed little improvement in height compared with age and sex-adjusted norms using standard deviation scores, although in our experience growth has improved in all of our patients post BMT. Thyroid function has been normal in almost all patients. With regard to fertility, as might be expected after high-dose therapy with alkylating agents, in both the Seattle collaborative and Belgian studies, the majority of evaluable females over the age of 13 show primary amenorrhoea, delayed sexual maturation and elevated gonadotrophin levels (Vermylen et al, 1998; Walters et al, 2000). In contrast, the majority of evaluable males have normal sexual development, although some have elevated gonadotrophin and reduced testosterone levels. Fertility has not been assessed, but by analogy with BMT for other indications, is likely to be impaired as a result of the toxicity of busulphan to the germinal epithelium of the testes (Sanders et al, 1996). It is too early to fully assess the risk of secondary malignancy post BMT for SCD, but Vermylen et al (1998) have reported a patient with myelodysplasia evolving into refractory acute myeloid leukaemia, 53 months post BMT, arising in the donor cells after prolonged therapy with azathioprine and thalidomide for chronic GVHD. The quality of life among patients with stable engraftment of donor cells has not been studied in detail. However, for 21/22 patients in the Seattle late effects cohort and 42/45 patients in the Belgian study, Karnofsky or Lansky scores were 100% and those with poorer quality of life measures were the individuals with extensive chronic GVHD. The lack of available donors for the majority of patients has focused attention on the use of alternative sources of haemopoietic stem cells. With the advent of molecular typing for HLA alleles, and more effective GVHD and antiviral agents, the results for VUD transplants for a number of indications in children are now comparable to those transplanted from a matched sibling donor (Oakhill et al, 1996). The only data in SCD refer to the use of unrelated donor cord blood stem cells (see below). The Seattle collaborative study group have recently approved a protocol for VUD BMT in children with SCD with one or more of the following features: renal insufficiency, red cell allo-immunization and recurrent acute chest syndrome. Unfortunately, however, the ethnic groups at risk of SCD are under-represented on unrelated donor panels. Cord blood SCT offers the potential advantages of an apparently lower incidence of GVHD than conventional BMT (Rubinstein et al, 1998), greater tolerance for HLA disparity and the possibility of expansion of the donor pool for patients with SCD by a policy of directed collection from deliveries of mothers from ethnic minorities. Conversely, the risk of graft rejection remains high, particularly when the cell dose is low, and engraftment may be slow (Rubinstein et al, 1998). The series reported by Vermylen et al (1998) included two successful sibling cord blood transplants for SCD and 11 cord blood transplants for S