The modern management of severe aplastic anaemia

The management of severe aplastic anaemia (SAA) has evolved rapidly during the last quarter of the century. Twenty years ago, a prospective multicentre study reported a 57% chance of survival for severe aplastic anaemia following allogeneic bone marrow transplantation (BMT) from an HLA-identical sibling donor ( Camitta et al, 1979) , in contrast with recent reports showing that the success of matched sibling donor BMT may now be as high as 90% ( Storb et al, 1994; Pitcher et al, 1999 ) . The 25% survival rate reported in the 1970s for the non-transplanted group ( Camitta et al, 1979 ) contrasts similarly with the 87% probability of 4-year survival reported 20 years later for a cohort of children treated with antithymocyte globulin (ATG) and cyclosporin ( Führer et al, 1998 ). However, the legacy of success with ATG is the recognition that long-term survivors are at significant risk of clonal haemopoietic disorders, and that it is no longer possible to measure success simply in terms of improved 5-year survival figures. This review gives an overview of the current management of severe acquired aplastic anaemia with this in mind, including the choice of definitive therapy and aspects of supportive care. It is inevitably not exhaustive, concentrating mostly on the factors influencing choice of therapy and outcome, both short-and long-term. For a fuller discussion, readers are referred to the excellent recent text on aplastic anaemia by a series of international experts ( Schrezenmeier & Bacigalupo, 2000). The classification of severe aplastic anaemia (SAA), as defined by a neutrophil count below 0·5 × 109/l, was based on early studies of response to treatment ( Camitta et al, 1975 ). A further subclassification followed the recognition that individuals with neutrophils below 0·2 × 109/l constituted a very severe group (VSAA) ( Gluckman et al, 1982 ; Bacigalupo et al, 1988 ); the distinction has remained valid despite advances in treatment. The diagnosis of familial hypoplastic anaemia may be obvious in the presence of associated physical anomalies or a positive family history, but should always be actively considered. Thus, testing for Fanconi anaemia (FA) is mandatory, even in adults presenting with apparently acquired aplastic anaemia (AA) ( Zatterale et al, 1995 ). The presenting features and management of FA are not discussed further here; see Alter & Young (1998) and Gluckman et al (2000) for review articles, and Guardiola et al (1998) for an overview of BMT in FA. Dyskeratosis congenita (DKC) may also present as SAA ( Alter & Young, 1998), but in the absence of a screening test it is not possible to determine how often this occurs. However, it does appear that patients transplanted for bone marrow failure in association with DKC have a different pattern of complications after BMT ( Rocha et al, 1998 ) and it may be that pre-BMT conditioning regimens in the future should be tailored for DKC as they are at the present time for FA. Similarly, the diagnosis of DKC may prompt the search for an alternative donor in a child without an HLA-identical sibling. It will be interesting to see whether the findings of the recent UK aplastic anaemia survey cast further light on the factors which cause bone marrow failure, but, in general, the aetiology (if it can be identified) of acquired AA does not seem to influence the short-term outcome ( Bacigalupo, 1996). A drug history is essential, both to ensure that adverse drug events are notified to the Committee on Safety of Medicines as well as to avoid the future use of potential culprits in that individual. Patients with SAA should also be questioned about recreational drug use. Interestingly, spontaneous recovery may occur with SAA following exposure to Ecstasy ( Marsh et al, 1994 ), but it may not be advisable to wait for this to occur ( Clark & Butt, 1997) . This may be viewed as either a chance coincidence in a young population ( Oosterkamp et al, 1998 ) or as an aetiological factor ( Aichison et al, 1989 ), leading to the extremely difficult decision among termination, supportive care only during pregnancy or treatments of untested safety to the fetus. The possible association between pregnancy and AA also has very important implications with respect to future pregnancies in women of child-bearing age following successful treatment of AA. The potential risk of pregnancy, with respect to both fetus and AA relapse, is the subject of an ongoing study by the European group for Blood and Marrow Transplantation (EBMT) SAA working party. On rare occasions, acute lymphoblastic leukaemia presents as spontaneously remitting SAA ( Camitta et al, 1975 ; Breatnach et al, 1981 ), but it is clearly unwise to wait for recovery in every case of childhood SAA, as delaying appropriate therapy may compromise the outcome. The differential diagnosis of SAA in adults should include hairy cell leukaemia ( Dixey & Gazzard, 1983). There is of course a long-standing recognition of the close links between paroxysmal nocturnal haemoglobinuria (PNH) and AA ( Dameshek, 1967; Lewis & Dacie, 1967). The identification of the defect in PNH cells as a failure to express the phosphatidylinositol glycan (GPI) anchor because of somatic mutations of the PIG-A gene ( Takeda et al, 1993 ; Bessler et al, 1994 ) has allowed the development of flow cytometric assays of GPI-anchored proteins to screen for PNH. These are more sensitive than the traditional Ham's test and are not influenced by transfusion ( van der Schoot et al, 1990 ). Estimates of a GPI-deficient population affecting at least one lineage at presentation now range from 15% ( De Lord et al, 1998 ) to as high as 52% ( Schrezenmeier et al, 1995a ), providing an important baseline for future studies on the development of secondary PNH, as discussed further below. The high proportion of patients with a GPI-deficient population at presentation raises many questions about issues related to management and prognosis. Classic haemolytic PNH, especially when associated with thrombosis, has a poor prognosis ( Hillmen et al, 1995 ; Sociéet al, 1996 ). Data from the international BMT registry (IBMTR) confirm that BMT is potentially curative in PNH ( Saso et al, 1999 ), whereas the AA component of classic haemolytic PNH may respond to ATG ( Paquette et al, 1997 ). There are, however, conflicting data as to whether the presence of a GPI-deficient population in otherwise typical AA affects the response to immunosuppressive therapy ( Schrezenmeier et al, 1995a ; De Lord et al, 1998 ). A larger prospective study is clearly needed to control for other prognostic factors. The presence of a GPI-deficient population is also of potential relevance when CAMPATH is used in pre-BMT conditioning regimens, as the CAMPATH antigen (CD52) is GPI anchored ( Xia et al, 1991 ). Despite claims that hypoplastic MDS (hypo-MDS) and AA can be differentiated on the basis of the peripheral blood picture ( Elghetany et al, 1997 ) or by immunostaining ( Orazi et al, 1997 ), in practice the boundary remains blurred. Cytogenetics at presentation may not be discriminatory; abnormalities have been detected at diagnosis in 4% of cases of SAA ( Appelbaum et al, 1987 ) and may in fact disappear with immunosuppressive therapy ( Geary et al, 1999 ). In another report, 6 out of 69 patients with acquired AA had abnormal cytogenetics at diagnosis. One with monosomy 7 progressed to acute myelogenous leukaemia (AML) and two died early from causes related to cytopenia, but the abnormalities in the remaining three disappeared after immunosuppression ( Mikhailova et al, 1996 ). It is clearly important to determine which cytogenetic abnormalities may be associated with a high risk of progression to leukaemia, especially as standard AA pre-BMT conditioning may not be appropriate for hypo-MDS with clonal abnormalities on presentation cytogenetics ( Appelbaum et al, 1987 ). In a study reported by the German–Austrian paediatric aplastic anaemia working group, two children with apparent AA who developed morphological evidence of MDS or AML within 19 months of presentation had chromosomal abnormalities (−7 or 5q–) detected retrospectively in diagnostic marrow samples ( Führer et al, 1998 ). A pre-existing abnormal clone may also provide an explanation for the high risk of MDS/AML in Japanese patients treated with granulocyte colony-stimulating factor (G-CSF) ( Kojima et al, 1992 ; Ohara et al, 1997 ; Kaito et al, 1999 ), with potential implications for the use of cytokine therapy, as discussed below. The overlap between hypo-MDS and AA, as syndromes of the bone marrow which are responsive to immunosuppression ( Jonasova et al, 1998 ), may reflect common pathogenetic mechanisms for the marrow failure in the two conditions, notably apoptosis ( Maciejewski et al, 1995 ; Philpott et al, 1995 ; Horikawa et al, 1997; Greenberg, 1998). Interestingly, Goyal et al (1999) noted a higher apoptosis rate in the subgroup of hypo-MDS with abnormalities of chromosomes 5 or 7. ATG may induce transfusion independence in typical MDS with a hypercellular marrow, not just in hypo-MDS ( Molldrem et al, 1997 ), and, interestingly, the presence of cells deficient in GPI-anchored proteins in MDS may predict a response to immunosuppression ( Dunn et al, 1999 ). Thus, in (partial) answer to the riddle posed by Dameshek (1967), one feature in common among AA, PNH and hypoplastic MDS/AML may well be immune-mediated apoptosis in the pathogenesis of bone marrow failure. Mortality is high in the first 6 months in all series, both ancient and modern, with severity as a major prognostic indicator, underscoring the importance of supportive care, which may be required for months or years. Thus, although all patients with SAA should be under the care of a centre experienced in its management, participating in appropriate trials and contributing data to SAA, BMT and immunosuppression data bases, a local base with good channels of communication to the centre is essential for ongoing supportive care. Infection is a major cause of death, and the management of neutropenia in this patient group may differ from that of the short-term marrow suppression more familiar to haemato-oncologists. For example, an individual who fails to respond to two courses of ATG is likely to be neutropenic for 8 months or longer. Life cannot be placed on hold and advice has to be given with respect to returning to work or school, to diet, to leisure activities and regarding when to seek medical attention for fever. Fungal infections, especially when due to aspergillus, represent a particular risk in aplastic anaemia ( Weinberger et al, 1992 ), but it is difficult to prevent exposure once the patient leaves hospital. Indwelling central lines, although essential for ATG and BMT, pose an additional infection risk for longer-term supportive care. Long-term antibiotic prophylaxis carries the risk of infection with resistant organisms. Empirical antibiotics with early use of antifungals thus provide the mainstay of treatment, although in the absence of other evidence of sepsis, fever occurring during ATG may not warrant the usual progression through broad spectrum antibiotics to amphotericin ( Dearden et al, 1998 ) . Cytokine support with G-CSF may be used to increase neutrophils, as well as potentially enhancing the response to immunosuppressive therapy, whereas an increase in monocytes in response to granulocyte–macrophage colony-stimulating factor (GM-CSF) may be of value in fungal infection. The long-term use of cytokines carries both the very real problem of significant expense and the potential problem of enhancing the long-term risk of secondary malignancy (see below). Historically, there has been a high risk of sensitization to blood products, as illustrated by the 34% of patients in one series who were refractory to random donor platelets, with 7% being refractory to HLA-matched platelets ( Klingemann et al, 1987 ). Sensitization is caused by leucocytes in a canine model ( Deeg, 1986). In keeping with this, leucodepletion of blood products, which should now be considered mandatory, significantly reduces the risk ( Killick et al, 1997 ). Irradiation of non-leucodepleted blood products also reduces sensitization in dogs ( Bean et al, 1994 ); although it may not add to the protection of leucodepletion, it is of course essential after BMT for HLA-matched platelets and for buffy coat infusions. The historic viewpoint of attempting to withhold platelets at all costs has largely been superseded, but it is reasonable to restrict platelet support to a level of 10 × 109/l ( Sagmeister et al, 1999 ), unless there is active bleeding or infection or in order to cover ATG infusions. Cytomegalovirus (CMV) antibody status should be determined before exposure to blood products and CMV-screened products used for potential BMT recipients, although leucodepletion also reduces the risk of CMV transmission ( Pamphilon et al, 1999 ). Unfortunately, despite the early promise of thrombopoietin, the potential usefulness of current preparations has been limited by antibody formation (reviewed by von dem Borne et al, 1998 ). Interleukin 6 (IL-6) also promotes megakaryocyte maturation in vitro, but a phase I/II study found that IL-6 failed to improve the platelet count in 11 refractory or relapsed patients, and in fact was discontinued early because of toxicity ( Schrezenmeier et al, 1995c ). Red cells should also be leucodepleted from the outset. Transfusion-related iron overload is a consequence of the improved short-term survival in aplastic anaemia and chelation therapy may need to be introduced. This undoubtedly detracts from the quality of life and may increase the risk of infection, but is essential especially if alternative donor transplantation after several months of transfusion dependence is considered. Erythropoietin has been shown to increase the erythroid response in patients treated with epo/G-CSF compared with G-CSF alone, but mostly in patients with non-severe AA ( Bessho et al, 1997 ). The rapidly evolving field limits the value of historical comparisons in the assessment of new therapeutic approaches, and the rarity of aplastic anaemia limits the usefulness of single-centre studies. Advances in the treatment of aplastic anaemia have largely been the result of prospective collaborative studies, notably those under the auspices of the EBMT SAA working party. A landmark randomized study showed the benefit of ATG in comparison with supportive care alone ( Champlin et al, 1983 ), with 11 out of the 21 patients treated with ATG responding, in contrast to 0 out of the 21 controls. The benefit of ATG was endorsed by an EBMT SAA working party retrospective analysis of 291 patients treated with ATG from 1981 to 1986; there was an overall 61% survival rate at 5 years after ATG, allowing the identification of age and severity as prognostic factors and leading to the definition of VSAA. The 38% response rate seen in patients under the age of 20 with VSAA was in contrast to the 82% response in the over-20s with non-severe AA. The major cause of death was infection or haemorrhage ( Bacigalupo et al, 1988 ). The results obtained with cyclosporin (CSA) used alone as first-line therapy are no better than ATG ( Gluckman et al, 1992 ). However, the combination of ATG followed by cyclosporin has been shown to be superior, especially for SAA or VSAA. In 1991, the German AA study group, in a randomized study of patients not eligible for BMT, reported a 65% response at 6 months for ATG/CSA in comparison with 31% for ATG alone ( Frickhofen et al, 1991 ). This was supported by the results of a prospective non-randomized study of ATG/CSA in SAA, showing a 67% response at 3 months and 78% response at 12 months ( Rosenfeld et al, 1995 ). Older patients also respond to ATG/CSA, the outcome being dependent more on their general health than on chronological age ( Tichelli et al, 1999 ) and making it inappropriate to apply an arbitrary upper age limit for treatment. There have been conflicting results on the potential value of androgens as an adjunct to ATG therapy. In one prospective randomized study of ATG with or without androgens, there was no difference between the treatment arms ( Champlin et al, 1985 ). However, two subsequent studies showed a significantly increased response rate with ATG and androgens, although with similar survival to ATG alone ( Kaltwasser et al, 1988 ; Bacigalupo et al, 1993a ). The increased response was most apparent in females with SAA ( Bacigalupo et al, 1993a ) . The role of androgen therapy in the modern management of SAA is thus not clearly defined, but can be considered for patients who either fail to respond to or relapse after treatment with ATG and cyclosporin. There is little indication for androgens given alone as first-line therapy, except as a holding measure in constitutional anaemia pending identification of a suitable donor. Granulocyte colony-stimulating factor (G-CSF) In a pilot study by the EBMT SAA working party, the combination of ATG/CSA followed by G-CSF 5 µg/kg/d during days 1–90 produced encouraging results with respect to trilineage response (82%) and reduced infection rate in 40 patients with SAA ( Bacigalupo et al, 1995 ). Results are awaited from the randomized study, now closed, which followed the pilot. A different randomized study showed a better response to ATG/CSA than to CSA/G-CSF, but unfortunately there was no ATG/CSA/G-CSF arm for comparison ( Raghavachar et al, 1998 ). Interleukin 3 (IL-3) Ganser et al (1990 ) in a phase I/II study of IL-3 in nine patients, including two who were previously untreated, reported an increase in granulocytes and monocytes, which was not sustained. However, there was evidence for an IL-3-dependent response in three out of eight refractory and four out of four relapsed patients treated with ATG/CSA in combination with IL-3 daily to day 90, although two patients subsequently developed AML ( Raghavachar et al, 1996 ). Granulocyte–monocyte colony-stimulating factor (GM-CSF) alone, may increase granulocytes and monocytes, with evidence of trilineage response ( et al, 1988 ; Champlin et al, 1989 ; et al, 1990 ), but there are no data on the combination of with cell factor In a study of patients who had failed ATG, trilineage response to the combination of with G-CSF was seen in of patients ( et al, 1998 ). there are as no data on the combination of which represent a future but with the of potentially the risk of secondary In a study of patients in the EBMT data there was a chance of response to a course of ATG after failure to survival at years were for the who only after two in comparison with the for to a first survival was only for the patients who supportive care only after to respond to a first course of ATG ( Schrezenmeier et al, ). Thus, with ATG is clearly in patients who fail to respond or who an response to therapy. A significant proportion of to immunosuppression may subsequently Rosenfeld et al , for example, reported a relapse rate after Patients who relapse have a good chance of to courses of ATG, but it clearly be to the first In one study, relapse was at a of 12 months from the of and in out of 11 cases cyclosporin was being or had been children subsequently to ( Führer et al, 1998 ). Thus, simply the of cyclosporin therapy may be in Interestingly, the may be associated with ( et al, 1994 ). is as a supportive measure to transfusion in the presence of cell its potential benefit as an adjunct to immunosuppressive therapy in SAA is more In a single-centre study, et al reported patients with SAA treated from to of had been The for or response to ATG, as well as transfusion or refractory to platelet The group showed evidence of an increase in blood and there was no significant difference in comparison with with respect to long-term survival or secondary clonal However, only in the of the study was cyclosporin used in to ATG ( et al, 1996 ), the of results from historical studies to current On the basis that bone marrow recovery may occur after in BMT for aplastic previously patients with SAA without a sibling donor were treated with with or without cyclosporin. with of was seen in 7 out of the 10 The time to platelet independence was ( et al, 1996 ). The results have been reported from a study in which patients were randomized between ATG and both arms cyclosporin. treatments had similar response although patients in the arm required significantly more supportive care ( et al, 1999 ). cell The of peripheral blood cells after may cells for cell in SAA ( Bacigalupo et al, ), and a successful in a patient with SAA has been reported ( et al, 1998 ). ATG/CSA the current standard as first-line immunosuppressive therapy to the of the that it even sibling donor BMT as first-line therapy in children with SAA ( et al, 1997 ). However, it is that factors other than short-term survival now be not least the risk of relapse and clonal remains the treatment of choice in children and young adults with severe aplastic survival for in SAA, as reported to the an from for patients transplanted during to for the The for the are with the to the use of cyclosporin in prophylaxis and to improved supportive care, including and prophylaxis ( Bacigalupo, the improved outcome in a single-centre cohort was related to than in years ( Pitcher et al, 1999 ). The risk of in SAA has been being by sensitization and a cell ( Storb et al, ), but may also even many years later ( et al, 1999 ). This may in be to the of aplastic anaemia BMT is more likely to result in recovery with the use of immunosuppressive conditioning than without ( et al, 1997 ). The of ATG to conditioning has also been shown to and improve survival in ( Storb et al, 1994 ), whereas cyclosporin also reduces the risk of For patients transplanted during the of cyclosporin in the conditioning was associated with a in rate from to ( et al, 1994 ). was also as a significant factor by the group, survival with a BMT after as pre-BMT conditioning ( et al, 1998 ). conditioning with alone, was apparent in of patients after BMT for aplastic anaemia ( et al, 1996 ), and it has been that the of ( et al, 1995 ). This is of particular relevance in the of cyclosporin therapy after treatment with cyclosporin is in after acute ( Storb et al, 1997 ), but the role of cyclosporin after BMT for aplastic anaemia may be in particular to the risk of that is of AA. before BMT has been which has in that long-term is ( et al, 1991 ), is ( et al, 1999 ), the risk of secondary malignancy is reduced ( et al, 1994 ) and the potential for marrow recovery is In is associated with an increased risk of ( et al, 1989 ). However, et al found that reduced the risk of and improved for SAA in have been to the use of with and cyclosporin ( et al, 1999 ). Unfortunately, the historical comparison used in the last study was alone before BMT, with but no as than the now standard combination of which have been more The of a group of international experts is that HLA-matched sibling bone marrow transplantation remains the treatment of choice for patients under the age of years with severe aplastic may also be considered in patients if there is no response to first-line therapy with ATG and cyclosporin. The use of in for SAA is not ( Schrezenmeier et al ). The of comparison with historical is even more apparent in the of alternative donor BMT in the modern management of severe aplastic This to in conditioning and as in cases of but more to advances in et al (1999) a retrospective analysis of patients with SAA who had been transplanted during from the Marrow there were with a of from matched had a better outcome than from but only out of matched were also matched at on at the of the data that the better than patients 15% survivors at 3 be to further In a study from patients and potential for and by for were of and showed that only of the out of were out of with the level of at et al, 1999 ). et al (1998) , also at and and by at found that only of were matched on including to the donor reported by et al (1999) , it is possible that of the survival even between matched and with this are recent results from for eight patients with acquired SAA to immunosuppressive therapy, transplanted from marrow has been out of patients were transplanted from a matched whereas three out of three died after BMT from