10 Autologous bone marrow transplant in the treatment of acute leukaemia ALAN K. BURNETT
In recent years there has been considerable improvement in the outcome of treatment for acute leukaemia but several obstacles remain. Acute myeloid leukaemia (AML) has a high remission rate in young patients (70-80%), significantly attributable to improved supportive care. Despite various permutations of treatment and strategy in remission, prevention of relapse remains elusive for most patients. There is a possibility of preventing relapse after high-dose chemoradiotherapy followed by marrow transplantation from an HLA-matched sibling donor (Thomas et al, 1979; Blume et al, 1980; Powles et al, 1980; Santos et al, 1983; Zwaan et al, 1984). Several problems relating to this have limited its potential success, and been the subject of substantial investigation in recent years. A major limitation is that allograft can only be safely offered to younger patients. The age limit remains unclear and varies with the policy of a transplant unit, but it is usually around 40 years. This excludes most patients with AML. Nevertheless the allograft experience suggested that myeloablative chemoradiotherapy could eradicate leukaemia and encouraged investigation of ways of circumventing these restrictions. To date there is little evidence to suggest that family donors who are not fully HLA-matched are suitable (Powles et al, 1983; Beatty et al, 1985) and the use of phenotypically matched unrelated donors is still in development. In neither case is this an alternative for the older patient. Autologous transplantation is not new, but many of the early efforts failed either because of supportive care failure or inability to have a lasting effect on the underlying disease. A fresh initiative was made by the Houston group in the late 1970s (Dicke et al, 1979) when marrow stored in first remission was used to support high-dose therapy at relapse. Although further remissions were achieved these were not durable. This outcome was not surprising because even an allogeneic transplant (Thomas et al, 1977), which may have the advantages of no risk of contamination with leukaemic cells and a possible additional graft-versus-leukaemia effect, failed to eradicate disease in most patients at that stage. The key to success-as it had been in allogeneic transplantation-was to undertake the autograft in remission, where the bulk disease was less and resistance less likely to have evolved. Initial fears about the ability of heavily Bailliere's ClinicalHaematologyVol. 4, No.3, July 1991 ISBN 0--7020--1546-6
751 Copyright© 1991, by Bailliere Tindall All rights of reproduction in any form reserved
752
A. K. BURNETT
treated marrow to regenerate after total body irradiation were allayed by the preliminary clinical results. Indeed it was possible to undertake myeloablation with marrow stored at 4°C for up to 54 h (Burnett et al, 1983). The decade of experience which has followed has witnessed the development of an enthusiasm for autologous bone marrow transplantation, particularly in AML. Most series suggest that this treatment does have a role in acute leukaemia but the data are mostly anecdotal. Substantial collaborative studies, some of which are more practical than others, are needed; this review delineates these areas.
RATIONALE OF AUTOLOGOUS BONE MARROW TRANSPLANTATION IN ACUTE LEUKAEMIA It seems illogical to reinfuse autologous bone marrow to a patient because of the high probability ofrecurrence ofleukaemia from that marrow. However, the regenerative stress associated with the autograft may permit normal haemopoiesis (possibly including immune recovery) to preferentially repopulate, to the disadvantage of the leukaemic clone. The responsiveness of minimal residual disease in AML to high-dose chemoradiotherapy is well established by the allograft experience, with relapse of about 20%. This reduction in relapse rate, as compared with chemotherapy alone (75% ), is attributed to the combination of myeloablative chemoradiotherapy and immunological mechanisms mediated by a population of T lymphocytes of donor origin (Butturini et al, 1987). How much is contributed by each component is unclear. It appears that the immune-mediated graft-versus-leukaemia (OVL) effect in man is associated with graft-versus host disease (GVHD), which would not be operative in the autograft setting (Weiden et ai, 1981; Bacigalupo et al, 1985; Sullivan et al, 1989). Syngeneic transplants in first remission, and allografts without (GVHD) are equivalent situations and are associated with an intermediate relapse rate of about 50-60% suggesting that the antileukaemic potential of the myeloablative protocol itself is limited (Gale and Champlin, 1984; Gale et aI, 1990). Whether these situations are equivalent to an autograft is unclear, since the contribution of post-remission. conventional chemotherapy pretransplant may have varied considerably between these groups. The autograft has the additional possibility of leukaemic cell contamination. Whether, and in what context, purging of occult leukaemia might be successful will be discussed later, but the twin data and allografts without OVHD can help to predict the outcome of autograft. Any increase in the relapse rate beyond what is seen in these groups will give a crude indication of the contribution of relapse attributable to leukaemic contamination of the autograft. The lack of toxicity and immunosuppression normally associated with an allogeneic transplant, together with other syngeneic experience with a total body irradiation approach (Appelbaum et al, 1982a), suggests that while the relapse rate might be greater this may be offset by reduced morbidity and
AUTOLOGOUS BONE MARROW TRANSPLANT
753
mortality of an autograft. This will be available to young patients who lack a matched donor and also to older patients up till the sixth decade, thus including many more patients with AML. Although most patients with AML will still be excluded on age grounds, the remission rate is importantly higher in the under 60sso that this strategy is applicable to most patients with the disease who enter remission. DEFINITION OF COMPLETE REMISSION
The generally accepted concept of remission of leukaemia is derived from the rodent models which indicate that a substantial residual bulk of tumour (e.g. up to 109 cells) may still remain at the point clinically defined as remission. While alternative explanations, such as a temporary reversion to a clinically benign preleukaemic state with restoration of normal differentiation potential, should not be excluded (Powles et aI, 1972), it is useful to work with the concept that there is a level of residual leukaemia which requires more sophisticated means of detection than generally used at present. In first remission this concept is relevant in two ways to autologous bone marrow transplantation. First, when is it appropriate to harvest the marrow for subsequent autograft? The potential danger to the graft of chemotherapy given to the patient-both because of the amount of treatment and agents used-must be balanced against the benefit of further cytoreduction in the patient, which will also reduce the potential contamination of the autograft. It would seem illogical to harvest marrow immediately upon entering remission (with a conceptual residual burden of 109 cells) when another course or two of post-remission intensive treatment could reduce this starting burden to 107 cells. Assuming that 1% of marrow cells are removed at harvest (1 x 108 nucleated cells/kg) then the autograft would contain from the in vivo starting burden of 107 cells, a leukaemic population of 105 cells. Of these 1-2% may be clonogenic precursors, and half may be lost in the freezing and thawing process. Of the potential clonogenic leukaemia burden of up to 103 cells remaining, it is not known how many may be successful in re-seeding into the marrow microenvironment. The starting burden of leukaemic cells in the harvest is variable but theoretically can be altered by effective induction therapy, as illustrated by high remission rate and postremission intensification or purging of the marrow carried out in vitro. Although these figures are speculative they are consistent with preclinical models (Hagenbeek et aI, 1989). (See Chapter 3.) The second issue is whether autograft permits sufficient cytoreductive potential to bring about cure at any level of minimal residual disease, or whether it can only be effective as an adjunct to additional post-remission treatment. The choice in first remission is whether to use the autograft to induce minimal residual disease, or as a treatment of minimal residual disease. From the model, post-remission cytoreduction chemotherapy preceding the autograft itself seems likely to have the best chance of leukaemia eradication. However, there are the potential drawbacks of
754
A. K. BURNETI
losing patients either because of relapse, such that those autografted are in one way or another selected, or from haematological toxicity which reduces the repopulative ability of the autografted marrow. Current morphological techniques which conventionally define remission are inaccurate. The clinical value of new approaches still needs to be put to the test. More sophisticated immunophenotyping with inappropriate coexpression of normal antigens can detect low levels of leukaemic cells (Campana et aI, 1990--review). In AML, where combinations may be available in about half the cases, the predictive value is unknown, but is under evaluation. In ALL, detection of cells of unique phenotype predates clinical relapse by about 6 months. There is the additional exciting potential of devising a unique gene probe recognizing individual sequences in the progeny of the leukaemic clone (Bartram et ai, 1990). Such techniques will be equally valuable in evaluating purging techniques, or in assessing the relevance of reinfusing cells defined to be of malignant origin by these techniques (Gribben et al, 1990). Extinction of the leukaemic clone is generally assumed to be associated with re-establishment of polyclonal haemopoiesis in remission. Some patients, who by all conventional criteria are in remission, have clonal haemopoiesis (Fearon et al, 1986; Fialkow et al, 1987); often this appears to originate from the leukaemic clone. The long-term clinical significance of clonal remission is not yet known, but it is a further indication that conventional definitions of complete remission fall short of what is now required. AUTOLOGOUS TRANSPLANT IN ACUTE MYELOID LEUKAEMIA Clinical results in first remission Several single-centre studies started in the early 1980s in AML now have worthwhile patient follow-up (Burnett et al, 1984; Stewart et al, 1985; Goldstone et al, 1986; Linch and Burnett, 1986; Lowenberg et al, 1987; Maraninchi et al, 1987; Carella et al, 1988; Meloni et al, 1987). In general the survival is 45-55%. The predominant cause of failure is relapse with procedural related mortality of 5-8%. The level of morbidity seems to be acceptable in the short term. Prolonged thrombocytopenia is not unusual and remains largely unexplained (Pendry et al, 1990); although late effects such as infertility and cataract are not likely to differ from allogeneic experience, they are not yet well documented. Most of the early studies made no attempt to purge the marrow in vitro but the possibility of occult contamination was not necessarily ignored because most administered post-remission chemotherapy before the autograft. Auto transplant preconditioning therapy Several groups simply transferred the standard allogeneic preparative protocol of cyclophosphamide and total body irradiation (TBI). There
AUTOLOGOUS BONE MARROW TRANSPLANT
755
appears to be little to choose between fractionated or unfractionated protocols. Higher dose TBI may have a better antileukaemic effect, and this requires further study. Heterogeneity between patients in several variables, makes demonstration of differences between myeloablative protocols difficult (Gale and Horowitz, 1990). Total body irradiation is not a prerequisite. Foremost amongst the highdose chemotherapy is busulphan/cyclophosphamide schedule in its original or modified form (Santos et al, 1983) but data on this protocol in first remission is lacking. Early series using the TACC protocol (6 thioguanine 400 mg/m- days -6 to -2; ara-C 400 mg/rrr' continuous infusion days -6 to -2; CCNU 400 mg/rrr' day -5; cyclophosphamide 45 mg/kg/daydays -5 to - 2; ABMT infusion day 0) (Cahn et al, 1986)or high-dose melphalan either singly or as a double procedure were ineffective (Maraninchi et al, 1987). In recent years, BAVC protocol (BCND 800 mg/rrr'day -6; AMSA 150 mg/rrr' day -5, -4, -3; VP-16150mg/m2 days -5, -4, -3; ara-C 300mg/rrr' by continuous infusion days -5, -4, -3; ABMT on day 0) has produced good results in second remission, but in first remission does not appear to be superior to other approaches (Meloni et al, 1987). The double auto transplant
The concept of double autograft was devised by the University College Hospital, London, group (Goldstone et al, 1986). Following chemoablation with a BACT protocol (BCND 300mg/m 2 day -5; cyclophosphamide 1.5 g/rn? days -5, -4, -3; doxorubicin 50 mg/nf day -5; ara-C 100 mg/m? b.d. days -5 to -1; 6-thioguanine 100mg/rrr' b.d. days -5 to -1) supported by autograft, the marrow is reharvested and a second graft with the same high-dose treatment undertaken. The first autograft effects an
In recent years there has been considerable improvement in the outcome of treatment for acute leukaemia but several obstacles remain. Acute myeloid leukaemia (AML) has a high remission rate in young patients (70-80%), significantly attributable to improved supportive care. Despite various permutations of treatment and strategy in remission, prevention of relapse remains elusive for most patients. There is a possibility of preventing relapse after high-dose chemoradiotherapy followed by marrow transplantation from an HLA-matched sibling donor (Thomas et al, 1979; Blume et al, 1980; Powles et al, 1980; Santos et al, 1983; Zwaan et al, 1984). Several problems relating to this have limited its potential success, and been the subject of substantial investigation in recent years. A major limitation is that allograft can only be safely offered to younger patients. The age limit remains unclear and varies with the policy of a transplant unit, but it is usually around 40 years. This excludes most patients with AML. Nevertheless the allograft experience suggested that myeloablative chemoradiotherapy could eradicate leukaemia and encouraged investigation of ways of circumventing these restrictions. To date there is little evidence to suggest that family donors who are not fully HLA-matched are suitable (Powles et al, 1983; Beatty et al, 1985) and the use of phenotypically matched unrelated donors is still in development. In neither case is this an alternative for the older patient. Autologous transplantation is not new, but many of the early efforts failed either because of supportive care failure or inability to have a lasting effect on the underlying disease. A fresh initiative was made by the Houston group in the late 1970s (Dicke et al, 1979) when marrow stored in first remission was used to support high-dose therapy at relapse. Although further remissions were achieved these were not durable. This outcome was not surprising because even an allogeneic transplant (Thomas et al, 1977), which may have the advantages of no risk of contamination with leukaemic cells and a possible additional graft-versus-leukaemia effect, failed to eradicate disease in most patients at that stage. The key to success-as it had been in allogeneic transplantation-was to undertake the autograft in remission, where the bulk disease was less and resistance less likely to have evolved. Initial fears about the ability of heavily Bailliere's ClinicalHaematologyVol. 4, No.3, July 1991 ISBN 0--7020--1546-6
751 Copyright© 1991, by Bailliere Tindall All rights of reproduction in any form reserved
752
A. K. BURNETT
treated marrow to regenerate after total body irradiation were allayed by the preliminary clinical results. Indeed it was possible to undertake myeloablation with marrow stored at 4°C for up to 54 h (Burnett et al, 1983). The decade of experience which has followed has witnessed the development of an enthusiasm for autologous bone marrow transplantation, particularly in AML. Most series suggest that this treatment does have a role in acute leukaemia but the data are mostly anecdotal. Substantial collaborative studies, some of which are more practical than others, are needed; this review delineates these areas.
RATIONALE OF AUTOLOGOUS BONE MARROW TRANSPLANTATION IN ACUTE LEUKAEMIA It seems illogical to reinfuse autologous bone marrow to a patient because of the high probability ofrecurrence ofleukaemia from that marrow. However, the regenerative stress associated with the autograft may permit normal haemopoiesis (possibly including immune recovery) to preferentially repopulate, to the disadvantage of the leukaemic clone. The responsiveness of minimal residual disease in AML to high-dose chemoradiotherapy is well established by the allograft experience, with relapse of about 20%. This reduction in relapse rate, as compared with chemotherapy alone (75% ), is attributed to the combination of myeloablative chemoradiotherapy and immunological mechanisms mediated by a population of T lymphocytes of donor origin (Butturini et al, 1987). How much is contributed by each component is unclear. It appears that the immune-mediated graft-versus-leukaemia (OVL) effect in man is associated with graft-versus host disease (GVHD), which would not be operative in the autograft setting (Weiden et ai, 1981; Bacigalupo et al, 1985; Sullivan et al, 1989). Syngeneic transplants in first remission, and allografts without (GVHD) are equivalent situations and are associated with an intermediate relapse rate of about 50-60% suggesting that the antileukaemic potential of the myeloablative protocol itself is limited (Gale and Champlin, 1984; Gale et aI, 1990). Whether these situations are equivalent to an autograft is unclear, since the contribution of post-remission. conventional chemotherapy pretransplant may have varied considerably between these groups. The autograft has the additional possibility of leukaemic cell contamination. Whether, and in what context, purging of occult leukaemia might be successful will be discussed later, but the twin data and allografts without OVHD can help to predict the outcome of autograft. Any increase in the relapse rate beyond what is seen in these groups will give a crude indication of the contribution of relapse attributable to leukaemic contamination of the autograft. The lack of toxicity and immunosuppression normally associated with an allogeneic transplant, together with other syngeneic experience with a total body irradiation approach (Appelbaum et al, 1982a), suggests that while the relapse rate might be greater this may be offset by reduced morbidity and
AUTOLOGOUS BONE MARROW TRANSPLANT
753
mortality of an autograft. This will be available to young patients who lack a matched donor and also to older patients up till the sixth decade, thus including many more patients with AML. Although most patients with AML will still be excluded on age grounds, the remission rate is importantly higher in the under 60sso that this strategy is applicable to most patients with the disease who enter remission. DEFINITION OF COMPLETE REMISSION
The generally accepted concept of remission of leukaemia is derived from the rodent models which indicate that a substantial residual bulk of tumour (e.g. up to 109 cells) may still remain at the point clinically defined as remission. While alternative explanations, such as a temporary reversion to a clinically benign preleukaemic state with restoration of normal differentiation potential, should not be excluded (Powles et aI, 1972), it is useful to work with the concept that there is a level of residual leukaemia which requires more sophisticated means of detection than generally used at present. In first remission this concept is relevant in two ways to autologous bone marrow transplantation. First, when is it appropriate to harvest the marrow for subsequent autograft? The potential danger to the graft of chemotherapy given to the patient-both because of the amount of treatment and agents used-must be balanced against the benefit of further cytoreduction in the patient, which will also reduce the potential contamination of the autograft. It would seem illogical to harvest marrow immediately upon entering remission (with a conceptual residual burden of 109 cells) when another course or two of post-remission intensive treatment could reduce this starting burden to 107 cells. Assuming that 1% of marrow cells are removed at harvest (1 x 108 nucleated cells/kg) then the autograft would contain from the in vivo starting burden of 107 cells, a leukaemic population of 105 cells. Of these 1-2% may be clonogenic precursors, and half may be lost in the freezing and thawing process. Of the potential clonogenic leukaemia burden of up to 103 cells remaining, it is not known how many may be successful in re-seeding into the marrow microenvironment. The starting burden of leukaemic cells in the harvest is variable but theoretically can be altered by effective induction therapy, as illustrated by high remission rate and postremission intensification or purging of the marrow carried out in vitro. Although these figures are speculative they are consistent with preclinical models (Hagenbeek et aI, 1989). (See Chapter 3.) The second issue is whether autograft permits sufficient cytoreductive potential to bring about cure at any level of minimal residual disease, or whether it can only be effective as an adjunct to additional post-remission treatment. The choice in first remission is whether to use the autograft to induce minimal residual disease, or as a treatment of minimal residual disease. From the model, post-remission cytoreduction chemotherapy preceding the autograft itself seems likely to have the best chance of leukaemia eradication. However, there are the potential drawbacks of
754
A. K. BURNETI
losing patients either because of relapse, such that those autografted are in one way or another selected, or from haematological toxicity which reduces the repopulative ability of the autografted marrow. Current morphological techniques which conventionally define remission are inaccurate. The clinical value of new approaches still needs to be put to the test. More sophisticated immunophenotyping with inappropriate coexpression of normal antigens can detect low levels of leukaemic cells (Campana et aI, 1990--review). In AML, where combinations may be available in about half the cases, the predictive value is unknown, but is under evaluation. In ALL, detection of cells of unique phenotype predates clinical relapse by about 6 months. There is the additional exciting potential of devising a unique gene probe recognizing individual sequences in the progeny of the leukaemic clone (Bartram et ai, 1990). Such techniques will be equally valuable in evaluating purging techniques, or in assessing the relevance of reinfusing cells defined to be of malignant origin by these techniques (Gribben et al, 1990). Extinction of the leukaemic clone is generally assumed to be associated with re-establishment of polyclonal haemopoiesis in remission. Some patients, who by all conventional criteria are in remission, have clonal haemopoiesis (Fearon et al, 1986; Fialkow et al, 1987); often this appears to originate from the leukaemic clone. The long-term clinical significance of clonal remission is not yet known, but it is a further indication that conventional definitions of complete remission fall short of what is now required. AUTOLOGOUS TRANSPLANT IN ACUTE MYELOID LEUKAEMIA Clinical results in first remission Several single-centre studies started in the early 1980s in AML now have worthwhile patient follow-up (Burnett et al, 1984; Stewart et al, 1985; Goldstone et al, 1986; Linch and Burnett, 1986; Lowenberg et al, 1987; Maraninchi et al, 1987; Carella et al, 1988; Meloni et al, 1987). In general the survival is 45-55%. The predominant cause of failure is relapse with procedural related mortality of 5-8%. The level of morbidity seems to be acceptable in the short term. Prolonged thrombocytopenia is not unusual and remains largely unexplained (Pendry et al, 1990); although late effects such as infertility and cataract are not likely to differ from allogeneic experience, they are not yet well documented. Most of the early studies made no attempt to purge the marrow in vitro but the possibility of occult contamination was not necessarily ignored because most administered post-remission chemotherapy before the autograft. Auto transplant preconditioning therapy Several groups simply transferred the standard allogeneic preparative protocol of cyclophosphamide and total body irradiation (TBI). There
AUTOLOGOUS BONE MARROW TRANSPLANT
755
appears to be little to choose between fractionated or unfractionated protocols. Higher dose TBI may have a better antileukaemic effect, and this requires further study. Heterogeneity between patients in several variables, makes demonstration of differences between myeloablative protocols difficult (Gale and Horowitz, 1990). Total body irradiation is not a prerequisite. Foremost amongst the highdose chemotherapy is busulphan/cyclophosphamide schedule in its original or modified form (Santos et al, 1983) but data on this protocol in first remission is lacking. Early series using the TACC protocol (6 thioguanine 400 mg/m- days -6 to -2; ara-C 400 mg/rrr' continuous infusion days -6 to -2; CCNU 400 mg/rrr' day -5; cyclophosphamide 45 mg/kg/daydays -5 to - 2; ABMT infusion day 0) (Cahn et al, 1986)or high-dose melphalan either singly or as a double procedure were ineffective (Maraninchi et al, 1987). In recent years, BAVC protocol (BCND 800 mg/rrr'day -6; AMSA 150 mg/rrr' day -5, -4, -3; VP-16150mg/m2 days -5, -4, -3; ara-C 300mg/rrr' by continuous infusion days -5, -4, -3; ABMT on day 0) has produced good results in second remission, but in first remission does not appear to be superior to other approaches (Meloni et al, 1987). The double auto transplant
The concept of double autograft was devised by the University College Hospital, London, group (Goldstone et al, 1986). Following chemoablation with a BACT protocol (BCND 300mg/m 2 day -5; cyclophosphamide 1.5 g/rn? days -5, -4, -3; doxorubicin 50 mg/nf day -5; ara-C 100 mg/m? b.d. days -5 to -1; 6-thioguanine 100mg/rrr' b.d. days -5 to -1) supported by autograft, the marrow is reharvested and a second graft with the same high-dose treatment undertaken. The first autograft effects an
