Portal de Periódicos da CAPES

MANAGEMENT OF SICKLE CELL DISEASE: RECENT ADVANCES AND CONTROVERSIES

1999; Wiley; Volume: 107; Issue: 1 Linguagem: Inglês

10.1046/j.1365-2141.1999.01565.x

1365-2141

Oswaldo Castro,

Iron Metabolism and Disorders

Sickle cell disease was first recognized as a haematological disorder almost 90 years ago ( Herrick, 1910). Progress in understanding of the disease was initially slow but by the mid 1950s the genetics and main features of its molecular pathology had been established ( Ranney, 1994). The concept of sickle cell anaemia as a 'molecular disease' eventually led to an explosion of knowledge fuelled by molecular genetics research, and many other molecular-genetic disorders were identified or better understood. Molecular research, which traces its origin to the study of an abnormal haemoglobin, also has led to advances in diverse disciplines such as genetics ( Fink & Collins, 1997), evolution ( Kumar & Hedges, 1998; Hoelzer et al, 1998 ) and history ( Schroeder et al, 1990 ; Adekile, 1997). Unfortunately, the benefit of much of this research to sickle cell patients has not yet been fully realized. In part this is because the precise pathogenesis of the most common complication of sickle cell disease, the painful episode (painful or 'vaso-occlusive' crisis), remains elusive and its management largely limited to the control of pain. In the last decade there have been clinical trials showing that painful episodes, which severely affect the patient's quality of life, can be prevented or their frequency diminished. Clinical trials also have shown that life-threatening complications of sickle cell disease such as stroke and overwhelming infection now can be avoided as well. Indeed, bone marrow transplantation demonstrates that sickle cell disease can be cured in some patients. The following is a summary of these management advances. Most of them naturally lead to new questions and controversies, some of which are discussed also. Some of the new mamagement strategies reviewed are not strictly 'evidence-based'. Their inclusion here reflects my own view of their value, a view that may not be shared by others. Hydroxyurea treatment is reviewed in greater detail in consideration of the rapidly spreading use of the drug for sickle cell treatment. Optimal management of sickle cell pain, though an important issue ( Ballas, 1997), has not been examined in large randomized trials, and is not a subject of this review. I will concentrate on clinical experiences in the United States, which are likely to be similar to those in the United Kingdom. Sickle cell disease is a world-wide disorder, but due to diverse genetic and environmental factors some of its manifestations are different in other areas of the globe such as Africa ( Konotey-Ahulu, 1991), Brazil ( Zago et al, 1980 ), the Caribbean ( Serjeant, 1985), India ( Mukherjee et al, 1998 ) and the Middle East ( Perrine et al, 1978 ; el-Hazmi, 1991). Medical advances based on data from American patients may not always be applicable to other sickle cell populations. Because of early loss of splenic function and maternal antibody, the very first manifestation of sickle cell disease in infants can be death from overwhelming infection due to encapsulated bacteria, mainly pneumococci ( Rogers et al, 1978 ). The risk of overwhelming sepsis remains quite high until about the age of 6 years. Prevention of these early deaths requires all three of the following interventions: (a) neonatal diagnosis followed by family education and penicillin prophylaxis ( Gaston et al, 1986 ), (b) management of febrile episodes as potential sepsis events ( McIntosh et al, 1980 ), and (3) vaccination against pneumococcal and Haemophilus influenzae infections at an appropriate age ( Smith & Wethers, 1994). Penicillin prophylaxis should be continued until 5 years of age, as beyond this age there are no clear-cut benefits for this intervention ( Falletta et al, 1995 ). Prescription of prophylactic penicillin, although necessary, is by itself insufficient to prevent overwhelming infections. This is because of non-compliance ( Berkovitch et al, 1998 ), emerging penicillin-resistant pneumococci ( Kaplan et al, 1998 ), and, of course, the possibility of infection with other encapsulated microorganisms. Neonatal diagnosis coupled with aggressive management of febrile episodes in infants and young children was effective in preventing septicaemia deaths before the era of penicillin prophylaxis ( McIntosh et al, 1980 ). Newborn diagnosis and an aggressive approach to febrile episodes are still essential. Approximately 5% of children with sickle cell anaemia develop a stroke due to cerebrovascular occlusion ( Balkaran et al, 1992 ; Ohene-Frempong et al, 1998 ). Once an ischaemic stroke has occurred, subsequent strokes are likely but most can be prevented by a monthly transfusion programme which lowers the proportion of sickling cells to < 30% ( Russell et al, 1984 ; Pegelow et al, 1995 ). There is no evidence that transient ischaemic attacks ( Pegelow et al, 1995 ) or intra-cerebral haemorrhage, a long-term consequence of the cerebral vasculopathy in stroke patients ( Powars et al, 1990 ), are preventable by a transfusion programme. It is not known how long transfusion programmes for stroke should be continued. In view of the high recurrence risk, most paediatricians recommend transfusions of stroke patients for an indefinite period of time ( Wang et al, 1991 ). Others, the author included, tend to stop transfusions when the patient reaches 18 years of age, provided that at least 3 years of transfusions have been given since the last stroke ( Rana et al, 1997 ). Primary prevention of stroke is now possible. Children at risk can be identified before their first stroke by detection of increased flow velocity in their cerebral arteries via trans-cranial Doppler ( Adams et al, 1992 ). A clinical trial randomized 130 such children (aged 2–16 years) to standard medical care plus a monthly transfusion programme or to standard medical care alone (control). When compared to the control arm, there was a 92% reduction in the incidence of stroke in the transfusion group. There was one stroke in 63 transfused patients versus 11 strokes in 67 control patients during a 21-month median follow-up period ( Adams et al, 1998 ). Certainly this advance has led to a number of important but unanswered questions: (a) How often do children need to be screened for stroke risk?, (b) is another method such as MRA just as good or better than trans-cranial Doppler in screening for stroke risk?, (c) if a marrow donor is available, should marrow transplantation instead of long-term transfusions be considered?, (d) should the duration of the transfusion programme be 'indefinite' as recommended by some for secondary stroke prevention? Additional clinical trials are needed to answer these questions, and without their results decisions regarding primary stroke prevention will remain difficult. On the one hand the catastrophic nature of stroke and the likelihood that transfusions will also prevent events such as pain episodes, chest syndromes and priapism must be considered. On the other, the problems inherent to a lifetime of transfusion and chelation therapy are notorious, and the possibility that not all 'at risk' children will go on to develop a stroke is a real one ( Cohen, 1998). The decision to start long-term transfusions in adult patients presenting with cerebrovascular occlusion is also difficult since there are no published series showing that transfusions prevent recurrent ischaemic strokes in this age group. Because of the experience in SS children some haematologists, such as this author, recommend a 3-year transfusion programme for adult sickle cell patients who develop cerebrovascular occlusion. Proliferative sickle retinopathy is most common in Hb SC disease (33%) but can occur also in about 14% of patients with Hb S-β-thalassaemia and in 3% of SS patients ( Lutty & Goldberg, 1994). Infarction of the peripheral retina results in the proliferation of fragile thin-walled blood vessels ('sea fans') at high risk of bleeding. Untreated retinopathy can lead to loss of vision through vitreous haemorrhage and/or retinal detachment. Two randomized trials of photocoagulation treatment of retinopathy have been reported. In one trial argon laser photocoagulation was used to treat sea fan feeder vessels in 25 eyes; 20 untreated eyes served as controls. After about 10 years of follow-up there was only one haemorrhagic event in the eyes assigned to photocoagulation, compared to nine vitreous haemorrhages (45%) in the control eyes ( Jacobson et al, 1991 ). Complications with this treatment method include rupture of Bruch's membrane and choroidal neovascularization ( Lutty & Goldberg, 1994). The second randomized trial used scatter photocoagulation around neovascular formations, avoiding sea fans and their feeder vessels ( Farber et al, 1991 ). 99 eyes were so treated and compared to 75 control eyes. This treatment also significantly reduced vitreous haemorrhage and loss of visual acuity but without any treatment-related complications. However, because of the relatively small number of patients enrolled, the above clinical trials probably lacked the power to answer all efficacy/safety issues. At present scatter photocoagulation is considered the preferred treatment of proliferative sickle retinopathy. It is essential to refer sickle cell patients, particularly those with Hb SC disease, to an ophthalmologist for detection of retinopathy at an early stage. Laser photocoagulation is most effective when peripheral retinal lesions are diagnosed early by fluorescin angiography, before they involve the central retina and threaten vision. The U.S. Food and Drug Administration (FDA) recently approved hydroxyurea as effective palliative treatment of adults with sickle cell anaemia. The development of hydroxyurea therapy is an excellent example of the application of basic research principles [inhibition of Hb S polymerization by Hb F ( Nagel et al, 1979 ), increased in Hb F synthesis in animals treated with myelosuppressive agents ( DeSimone et al, 1982 )] to clinical science ( Charache et al, 1992 ) and eventually to the benefit of patients. FDA approval was based on the results of a randomized placebo-controlled clinical trial, the Multicenter Study of Hydroxyurea in Sickle Cell Anemia (MSH). The trial involved SS patients >18 years of age with normal renal and liver function and who had at least three vaso-occlusive events in the year before enrolment ( Charache et al, 1995 , 1996). In this trial 152 patients were assigned to hydroxyurea and 147 to placebo and were followed for a mean of 28 months. Commonly accepted criteria for defining painful crises and other vaso-occlusive events were used and clinical records were reviewed centrally to ensure correct and uniform classification of the study endpoints. The incidence of painful episodes was significantly lower in hydroxyurea patients (median 2.5/year v 4.6/year, P = 001), as were also hospitalizations for these episodes (median 1.0/year v 2.5/year, P = 0.0027). Chest syndrome developed in 25 patients in the hydroxyurea group, compared to 51 patients taking placebo (P = 0.001). Similarly, both the number of patients needing transfusions (48 v 73, P = 0.002), and the number of red cell units transfused (336 v 586, P = 0.003) were lower in the hydroxyurea treatment arm. Haematological changes seen in hydroxyurea-treated MSH patients included a higher haemoglobin level (9.1 v 8.5 g/dl, P = 0.0009) and red cell MCV, and lower neutrophil (4.9 v 6.4 × 109/l, P = 0.0001) and reticulocyte counts (231 v 300 × 109/l, P = 0.0001) ( Charache et al, 1996 ). Improvement of the anaemia and the lower reticulocyte count were probably due to longer red cell lifespan in hydroxyurea-treated patients ( Ballas et al, 1999 ). No MSH patient developed serious myelotoxicity, though there is a report of prolonged pancytopenia in one hydroxyurea-treated patient not participating in the trial ( Vichinsky & Lubin, 1994), and the author has observed another non-MSH patient with a similar complication. Avoidance of serious myelotoxicity requires adherence to the safety features of the MSH protocol ( Charache et al, 1995 , 1996): (a) starting at 15 mg/kg/d, (b) monitoring complete blood cell counts, and serum creatinine levels every 2 weeks, (c) increasing hydroxyurea dose by about 5 mg/kg/d at intervals of no less than 8 weeks until a haematologically safe stable dose has been established for each patient, with monthly monitoring thereafter, (d) not exceeding 35 mg/kg/d, (e) stopping hydroxyurea in case of 'myelotoxic' blood cell counts (see Table I), or if renal or liver function decrease (creatinine >150.3 μmol/l, ALT > 2× normal, respectively). The drug should be stopped for 2 weeks or until blood cell counts have risen above 'toxic' levels ( Bristol-Myers Squibb Co., 1998), and (f) prescribing only that number of hydroxyurea capsules needed until the next scheduled monitoring visit. Hydroxyurea was reported to be effective also at minimal doses (15 mg/kg/d) and without dose escalation ( Lima et al, 1997 ) but the safety and efficacy of this approach have not been examined in controlled trials. Not all MSH patients responded to hydroxyurea. An above-average response in terms of Hb F increase was more likely in patients with higher baseline neutrophil or reticulocyte counts (presumably those patients with high marrow 'reserve'), and in those lacking the Bantu (CAR) haplotype ( Steinberg et al, 1997a ). Hydroxyurea administration results in only transient blood levels of the drug so that compliance could not be monitored reliably by blood levels. Poor compliance may well have been a cause for lack of response in some patients ( Steinberg et al, 1997a ). From the preceding it is clear that hydroxyurea, though effective, is not an ideal agent for treating sickle cell disease. The drug has three main disadvantages. First, the frequent blood tests required for toxicity prevention affect the quality of the patients' lives and furthermore it is possible that an occasional patient may still develop serious myelotoxicity despite these measures. Second, most young adults need to consider the mutagenicity ( Ziegler-Skylakakis et al, 1985 ) and potential teratogenicity ( DePass & Weaver, 1982) of hydroxyurea so that this treatment may affect their decision to have children. Third, there is the potential oncogenicity of anti-neoplastic drugs so that long-term hydroxyurea treatment could predispose to cancer or leukaemia. Polycythaemia vera (PV) is a disorder with a high risk of leukaemic transformation. A randomized prospective trial showed that the leukaemogenic risk of PV was significantly increased by treatment with radiation or with chlorambucil, an alkylating anti-neoplastic agent ( Berk et al, 1995 ). The leukaemogenic effects of radiation and chlorambucil in PV were demonstrable after 8 years of exposure to these agents. Currently a cohort of hydroxyurea-treated PV patients is being followed to determine the drug's benefits as compared to a historical control group (the group that had been randomized to treatment without radiation or chlorambucil in the earlier PV trial). Though not designed specifically for this purpose, the incidence of leukaemia in the hydroxyurea-treated PV patients is also being examined ( Fruchtman et al, 1997 ). After 8 years of hydroxyurea treatment, a period of time by which radiation and chlorambucil had clearly shown a leukaemogenic effect, there was no statistical evidence of increased leukaemia incidence in hydroxyurea-treated PV patients. However, there may be a trend for more cases of leukaemia in the hydroxyurea cohort (9.8% v 3.7%, P = 0.097) and thus it is possible that increased leukaemia risk could be demonstrated with longer follow-up. The MSH investigators are monitoring the initial sickle cell patient cohort exposed to the drug in order to ascertain mortality, as well as cases of birth defects or neoplasia. Alternative, safer approaches to pharmacologic induction of Hb F production are currently under investigation ( Liakopoulou et al, 1995 ; Perrine et al, 1994 ; Wood & Davies, 1995; Selby et al, 1997 ). Interestingly, the MSH data suggested that increasing fetal haemoglobin may not be the only mechanism by which hydroxyurea benefits sickle cell patients: (a) painful events began to decrease before F cells increased ( Charache et al, 1996 ), (b) the mean Hb F level after 2 years of hydroxyurea treatment was only 8.6% (it was 4.7% in the placebo group), and (c) there was a positive statistical association between painful event frequency and neutrophil counts ( Charache et al, 1996 ). In the U.S. Cooperative Study of Sickle Cell Disease high steady-state leucocyte counts were associated with higher frequency of painful events ( Platt et al, 1991 ) and chest syndrome ( Castro et al, 1994 ), and with mortality ( Platt et al, 1994 ). Thus leucocytosis in sickle cell anaemia may not be merely a response to on-going tissue injury, but could itself precipitate or contribute to sickle-cell-related tissue damage. Leucocyte reduction by hydroxyurea theoretically could be a mechanism for lowering crisis rate. Hydroxyurea treatment lowers the number of adhesive red cell receptors in sickle cell patients ( Styles et al, 1997 ) and animal studies suggest that it also could increase Hb–nitric oxide association ( Jiang et al, 1997 ). The MSH raised several management-related questions and controversies, the most important of which are: 1. Should hydroxyurea be given to SS children, and if so, at what age? This issue was reviewed by Vichinsky (1997) and by Ohene-Frempong & Smith-Whitley (1997). It seems clear that, at least in the United States, the MSH will not be repeated in children. Several groups are already using hydroxyurea for paediatric patients who have frequent vaso-occlusive events ( Scott et al, 1996 ; Vichinsky, 1997; Ferster et al, 1996 ; Meier-Redelsperger et al, 1998 ). A consensus seems to have emerged that, in children, hydroxyurea will reduce the frequency of vaso-occlusive crises, chest syndromes and transfusions at least as effectively as in adults. If a randomized paediatric clinical trial is implemented in the United States it is likely to focus on prevention of damage to organs such lungs, kidneys and spleen, on stroke prevention, and/or on long-term effects (growth and development, and potential carcinogenicity). In the design of such a trial, patient age will be the most controversial issue ( Ohene-Frempong & Smith-Whitley, 1997). Treating younger patients with hydroxyurea is likely to result in better outcomes but will also expose them to potential long-term adverse effects. In this context it should be emphasized that hydroxyurea does not cure sickle cell disease. Therefore whether its potential long-term problems in children will be viewed as an acceptable risk will depend on whether clinical trials eventually show (or pilot data suggest) that the drug also lowers mortality or prevents catastrophic events such as stroke. 2. Does hydroxyurea benefit patients with Hb SC disease or sickle-β-thalassaemia? Clinical trials of hydroxyurea in Hb SC disease or in sickle cell-β+-thalassaemia are not available. Steinberg et al (1997b ) treated six adult Hb SC patients with 1000 mg hydroxyurea/d. After 12 months there were laboratory changes that suggested a lower haemolytic rate: increased haematocrit (and red cell MCV) with decreases in reticulocyte counts, bilirubin and dense cells. Fetal haemoglobin did not increase. Whether these cellular changes can produce a favourable clinical effect is speculative. Because the erythroid marrow in Hb SC disease is less hyperplastic than in sickle cell anaemia, tolerance to hydroxyurea myelotoxicity could be lower. Voskaridou et al (1995 ) reported that hydroxyurea treatment of Greek patients with sickle-β-thalassaemia improves their anaemia and lowers the frequency of vaso-occlusive crises just as in American SS patients. Most Greek patients with sickle-β-thalassaemia have the β° thalassaemic type and those with sickle-β+-thalassaemia produce very little Hb A. In contrast, African and American sickle-β+-thalassaemia patients have Hb A levels of 10–30% ( Adams, 1994). In these patients bone marrow hyperplasia is less pronounced than in SS disease. Theoretically they could tolerate only low hydroxyurea doses, as stated above for Hb SC disease. 3. Is hydroxyurea indicated for sickle cell ankle ulcers? On enrolment in the MSH about 32% of patients had ankle ulcers ( Charache et al, 1992 ). Although it was not a study endpoint, follow-up data showed that new leg ulcers developed with equal frequency in hydroxyurea and in placebo-treated patients ( Charache et al, 1996 ). These MSH data may need additional analysis in light of reports of hydroxyurea causing ankle ulcers in non-sickle cell patients with chronic myeloproliferative disorders ( Best et al, 1998 ). Ankle ulcers in sickle cell disease are probably related to chronic haemolysis rather than to vaso-occlusion per se, since very similar ulcers occur in non-sickling haemolytic conditions such as thalassaemia intermedia ( Gimmon et al, 1982 ) and hereditary spherocytosis ( Lawrence et al, 1991 ). Hydroxyurea decreases haemolysis in sickle cell disease and it would seem reasonable to try hydroxyurea in SS patients who have non-healing leg ulcers ( Orringer et al, 1992 ). 4. Should hydroxyurea be used instead of transfusions for stroke prevention? Iron overload is a predictable complication of long-term transfusion programmes recommended for stroke prevention. To date there is no demonstrably effective iron chelator other than deferoxamine, which needs to be administered as daily subcutaneous or intravenous infusions. Deferiprone, an iron chelator that is given orally, was reported to be as effective as deferoxamine in reducing tissue iron ( Collins et al, 1994 ), but more recent data suggest otherwise ( Olivieri et al, 1998 ). Compliance with deferoxamine infusions remains a major problem, so some transfused stroke patients will succumb to iron overload unless transfusions are stopped or an effective oral iron chelator becomes available. Another, less frequent, problem with long-term transfusion is the development of multiple red cell alloantibodies that can make the procurement of compatible units very difficult in the long run. In some areas of the world transfusions are not only less available but they could carry a greater risk of transmission of infections. For these reasons, hydroxyurea might be viewed as an alternative for blood transfusions in stroke patients who are iron overloaded, or who have difficulty obtaining blood because of problems related to alloimmunization ( Ware et al, 1995 ) or to unavailability of a safe blood supply. It should be emphasized, however, that in the United States a long-term transfusion programme coupled with iron chelation therapy is the 'standard of care' (a medico-legal term) for prevention of stroke recurrence. Furthermore, some children on hydroxyurea have developed primary or recurrent stroke despite this therapy ( Vichinsky, 1997). Although less desirable than hydroxyurea treatment, transfusion or exchange transfusion programmes are sometimes used to control the frequency of vaso-occlusive events ( Leitman et al, 1991 ; Wayne et al, 1993 ; Davies & Olatunji, 1995). The rationale for the transfusion approach is based on the following laboratory and clinical observations: (a) adding even small quantities of Hb AA red cells to Hb SS red cells in vitro improves the rheologic characteristics of the mixture ( Lessin et al, 1978 ); (b) in SS children with hyposthenuria or functional asplenia transfusions can temporarily reverse these abnormalities ( Statius van Eps et al, 1970 ; Pearson et al, 1970 ; Wethers & Grover, 1987); (c) exchange transfusions improve exercise capacity in SS patients without substantially raising their haematocrit ( Miller et al, 1980 ), suggesting that addition of non-sickling RBCs improves blood rheology and hence tissue perfusion; (d) vaso-occlusive complications such as pain crises, chest syndrome and priapism are considered rare in children with sickle cell anaemia who are on transfusion programmes for stroke; (e) a prospective randomized trial showed that prophylactic transfusions reduced the incidence of vaso-occlusive events in pregnant sickle cell patients ( Koshy et al, 1988 ). To prevent doing more harm than good, transfusion programmes should be considered only as a last resort for patients in whom vaso-occlusive events are so frequent that they seriously interfere with normal life activities. Patients with frequent vaso-occlusive events are also at greater risk of early death. A decision on recommending transfusions to prevent these patients' frequent and severe painful episodes must weigh the expected improvement in their quality of life against the known risks of long-term transfusions (haemolytic reactions, transmission of infections and iron overload). Transfusional iron overload can be delayed by the use of automated red cell exchanges ( Hilliard et al, 1998 ) or of manual exchange transfusions techniques. Unfortunately, in many patients exchange transfusions, and particularly erythrocytapheresis, require insertion of permanent high-flow vascular catheters, which, in turn, lead to a substantial risk of sepsis. The first SS patient who underwent bone marrow transplantation also had acute leukaemia ( Johnson et al, 1984 ). 14 years after this intervention the patient remains cured of both the leukaemia and sickle cell disease. Encouraging experience with marrow transplantation in SS patients (without leukaemia) was reported from Belgium ( Vermylen et al, 1988 ). Currently a collaborative protocol is enrolling sickle cell patients in Brazil, Canada, France, Germany, the U.K. and the U.S.A. to evaluate the role of bone marrow transplantation as definitive treatment of sickle cell disease ( Walters et al, 1996a ). To date, about 100 children ( Bernaudin et al, 1993 ; Ferster et al, 1995 ; Vermylen et al, 1998 ; Walters et al, 1996b ), with an average age of 7–10 years, have received allogeneic bone marrow grafts from HLA-matched relatives as curative treatment of their sickle cell disease. Ablation of recipient bone marrow is accomplished with busulphan and cyclophosphamide, without total body irradiation. Following early experience with frequent neurologic post-transplant events careful attention is given to prevention of hypertension, seizures and thrombocytopenia ( Walters et al, 1995 ). Early, procedure-related mortality ranges between 5% and 10% and will probably become lower in the future. Cure rates range from 71% to 82% and chronic graft-versus-host disease develops in < 15% of patients. The median periods of follow-up reported vary between 23 and 60 months. These are very positive results, at least for those sickle cell patients who have an HLA-compatible and healthy related donor. Still, issues dealing with patient selection and with long-term outcomes need to be considered. Because there is no simple reliable index of severity in sickle cell disease, a wide range of complications is used to define patient eligibility in the collaborative transplantation protocol mentioned above ( Walters et al, 1996b ). Qualifying complications include both intermittent events (stroke, frequent painful crises) and evidence of chronic organ damage (eyes, lungs, kidneys, bones). Parent and physician opinions on the validity of some of these inclusion criteria are likely to vary widely. While some would consider the procedure for all sickle cell patients who have a marrow donor ( Piomelli, 1991), others might take a position that transplantation will help only very few SS patients ( Serjeant, 1996). It is not unreasonable to speculate that parents, patients and doctors face this dilemma. The best transplantation outcomes are expected early in the life of the sickle patient, before significant organ damage has occurred, and also in those young patients with the mildest disease. Yet it is precisely these patients who have the highest probability of 'event-free' long-term survival without transplantation. Such patients have the most to lose if, despite being at the lowest risk for transplant complications, they are still unfortunate enough to die early of a transplant-related event or to develop chronic graft-versus-host disease (GVH). Transplantation with umbilical cord haemopoietic stem cells is associated with a lower incidence of GVH. Preliminary data in three children with this type of haemopoietic graft suggest that the procedure will become a useful alternative to bone marrow transplant ( Miniero et al, 1998 ). Long-term adverse effects of bone marrow transplantation also are significant but need to be seen in the light of the long-term effects of sickle cell disease itself, or of those of its alternative treatments. Long-term transplant complications include sterility and acute leukaemia from the high-dose anti-neoplastic drugs required for the procedure ( Deeg et al, 1998 ; Vermylen et al, 1998 ). Some of the recent advances in sickle cell disease treatment reviewed above are based on red cell transfusions ( Table II). The randomized trial for primary stroke prevention was described earlier. Koshy et al (1988 ) reported a study in which 72 pregnant SS patients receiving standard prenatal care were randomized to either a 'prophylactic' red cell transfusion arm (36 patients) or to have transfusions only on an 'as needed' basis, i.e. for medical or obstetric emergencies (36 patients). Data analysis included a number of perinatal outcomes (birth weight, fetal distress, stillbirth, neonatal death), obstetrical outcomes (premature delivery, toxaemia and caesarean section) and sickle-cell-related complications (pain crisis, chest syndrome and severe anaemia). With the exception of painful crises, there were no statistically significant differences between any of the outcomes examined. However, the study probably did not have statistical power to detect differences in serious but uncommon events such as maternal death or perinatal death ( Powars et al, 1986 ). Half of the patients assigned to the 'as needed' transfusion group had vaso-occlusive events, compared to 14% of patients who developed this complication in the prophylactic transfusions arm (P < 0.01). There was no significant difference in alloimmunization rate between the two groups. This first randomized trial of transfusions in sickle cell disease (probably also the first such trial in any disease) clearly demonstrated that transfusions prevent vaso-occlusive events, at least in pregnant sickle cell patients. A more recent randomized trial investigated whether preoperative management in SS patients requires 'aggressive' (essentially exchange transfusions), as opposed to 'conservative' (simple) transfusions ( Vichinsky et al, 1995 ). A total of 551 Hb SS patients were randomized. The aim of the aggressive regimen was to achieve a preoperative Hb level of at least 10 g/dl as well as a Hb S fraction of 30% or less. The conservative regimen attempted to reach an Hb level of 10 g/dl regardless of the percent Hb S. Both children and adults participated in the trial. The most common operations were cholecystectomy, ENT and orthopaedic procedures. Over 80% of the procedures were carried out using a combination of inhalation and intravenous anaesthesia. The study showed that the aggressive regimen was no better than the conservative regimen in preventing postoperative sickling complications, the most common of which was sickle cell chest syndrome (10% of the cases in both treatment arms). Similarly postoperative fever, infection and pain crises were not affected by the preoperative transfusion regimen. In contrast, new RBC alloantibodies were twice as common in the aggressive transfusion arm (10% v 5% in the conservative regimen, P = 0.01). More important, haemolytic transfusion reactions occurred in 6% of cases managed with aggressive transfusions, whereas only 1% of patients in the conservative transfusion arm developed this problem. Based on these findings, simple transfusions are at present preferable in the preoperative management of SS patients. Limited exchange transfusions might still be required for those few SS patients with higher haematocrits or for patients with Hb SC disease ( Neumayr et al, 1998 ). This randomized trial demonstrated once again the well-known alloimmunization risk in transfused sickle cell patients ( Rosse et al, 1990 ). Alloimmunization to minor red cell antigens not only causes difficulty in finding compatible red cell units for future transfusions. It is also responsible for serious, even lethal, delayed haemolytic transfusion reactions. Such reactions occur because up to 40% of the previously formed red cell antibodies become undetectable with time ( Rosse et al, 1990 ). When patients with these in vitro undetectable antibodies are subsequently transfused with in vitro'compatible' RBCs, an anamnestic antibody response ensues in which the transfused red cells are rapidly removed from circulation by phagocytosis or by intravascular haemolysis ( Anderson et al, 1997 ). The reason for the high alloimmunization rate in sickle cell patients is not known. Differences between red cell antigen patterns in blood donors (most U.S. donors are Caucasian) and sickle cell recipients (in the U.S. almost exclusively of African ancestry) undoubtedly play a role. Whether this difference is the only or even the most important factor in alloimmunization is unclear. Greek thalassaemic patients, for example, have a lower alloimmunization rate (21%) than Greek sickle cell patients (37%), even though the former receive many more transfusions and even though donor–recipient antigen red cell patterns are likely to be similar ( Spanos et al, 1990 ). This could suggest that there is something about sickle cell disease that makes patients more likely to respond to red cell alloantigens. Sickle cell patients who form multiple and uncommon red cell antibodies are reminiscent of some autoimmune disease patients who have a similar red cell antibody response pattern ( Ramsey & Smietana, 1995). The seriousness of the alloimmunization problem has led to recommendations that all sickle cell patients be transfused with blood of donors whose RBC antigens are more closely matched to those of the sickle cell recipients. Examples of such donors include: (1) racially matched donors ( Vichinsky et al, 1990 ); (2) donors matched for the following antigens in addition to ABO and D: K, C, E, S, Fya and Fyb ( Tahhan et al, 1994 ), K, Rh, Jka, Jkb and Fya ( Ambruso et al, 1987 ) and C, D, E and K ( Adams et al, 1998 ). The above recommendations are controversial ( Ness, 1994) and are at variance with standard blood bank practices. Their adoption, at least in the United States could result in depletion of rare blood resources urgently needed for sickle cell patients who have formed multiple alloantibodies ( Mallory et al, 1992 ). Extended antigen match for all patients also has the potential for delaying transfusions in critical situations. In view of this an attempt was made to estimate the expected benefit of antigen matching based on a review of the alloimmunization experience at the author's institution ( Castro & Sandler, 1995). 137 patients at our institution became alloimmunized after transfusion with standard ABO and D antigen matched units. If all transfused patients (since it is not known beforehand which patient will form antibodies) always had been given red cell units that were matched, in addition to ABO and D, also for C, c, E, e, K, S, Fya, Fyb, Jka and Jkb antigens (an even better extended match than those recommended above), 37 of these patients would still have remained who had uncommon antibodies not included in the extended match (anti-Jsa, -Cw, -s, -V, -U, -Lua, -Jsb). In other words, even this extended match pattern would fail to prevent alloimmunization in 27% of the patients. Randomized trials need to be conducted before definitive recommendations on extended RBC matching are implemented. Trials should assess not only endpoints such as decreases in alloimmunization and transfusion reactions, but should also measure the potential impact of delays in obtaining matched red cell units. Identification of rare antigen-negative units could be enhanced if mass screening of all red cell donor units with monoclonal anti-Fy3 is implemented ( Sandler et al, 1997 ). In the meantime it would seem prudent to limit the use of antigen-matched RBCs to patients needing scheduled, rather than emergency, transfusions. If, as it appears to be the case, regularly scheduled transfusions decrease sickle cell complications, then the need for emergency transfusions in these patients could be eliminated. Patients at centres with blood bank programmes that identify and recruit matched donors would be logical candidates for transfusions with antigen-matched units. The management advances reviewed above are examples of evidence-based medicine applied for the benefit of sickle cell patients. Hopefully the questions and controversies raised by these advances will lead to additional clinical research and to more effective patient care.