¹Department of Medicine, University of California San Diego, La Jolla, CA, USA

²Department of Graduate School of Public Health, University of Pittsburgh, and the University of Pittsburgh Cancer Institute, Pittsburgh, PA, USA

³Department of Medicine, University of Pittsburgh School of Medicine, Pittsburgh, PA, USA

⁴Department of Pediatrics, University of Pittsburgh School of Medicine, Pittsburgh, PA, USA

Correspondence to: Dr E D Ball, University of California San Diego, 9500 Gilman Drive, La Jolla, CA 92093-0960, USA

One hundred and thirty-eight patients with AML underwent ABMT with monoclonal antibody plus complement-purged marrow between August 1984 and March 1997. One hundred and ten patients were in CR (CR1: 23; CR2/3: 87) and 28 were in first relapse (R1) at ABMT. Preparative regimens included busulfan (16 mg/kg) and CY (120 mg/kg) (n = 93), CY (120 mg/kg over 2 days) with TBI (1200 cGy) (n = 35), and busulfan (16 mg/kg) plus etoposide (60 mg/kg) (n = 10). CR1 patients treated with CY/TBI (n = 7) had 3- and 5-year disease-free survival (DFS) rates of 71% and 57%. CR1 patients treated with BU/CY (n = 12), had 3- and 5-year DFS rates of 45%. Three and 5-year DFS for CR2/3 patients treated with CY/TBI (n = 26) was 23%. Three- and 5-year DFS for patients in CR2/3 treated with BU/CY (n = 55) was 31 and 28%. Three- and 5-year DFS for patients in R1 treated with BU/CY (n = 26) was 37%. In multivariate analysis, increased age was associated with greater risk of death and relapse. For CR2/3 patients, the length of CR1 was a significant predictor of DFS. ABMT performed in CR or R1 results in excellent 5-year DFS and OS. The contribution of purging may require a randomized trial comparing purged vs unpurged stem cell infusions. Bone Marrow Transplantation (2000) 25, 823-829.

monoclonal antibodies; acute myeloid leukemia

The treatment of AML in children and adults has steadily improved over the past two decades.^1,2,3 However, chemotherapy as the sole therapy of AML has limits. Combination chemotherapy can induce a CR in 50-80% of patients. However, at least 50% of patients subsequently relapse and ultimately die of their disease.

Allogeneic BMT has been shown to reduce relapse rates significantly in patients with AML in first CR.⁴ However, due to significant treatment-related mortality, overall disease-free survival (DFS) is approximately 50% at 5 years.⁴ In second and third CR, allogeneic BMT results in relapse-free survival of 20-30% at 5 years. The results in these more advanced patients are worse due to a higher relapse rate and to higher treatment-related toxicity.

A major limitation of allogeneic BMT is that it can be applied to only a minority of patients with AML. Only about 40% of patients with AML have histocompatibility antigen (HLA)-matched donors, and most patients over 55 years are considered too old to tolerate this procedure. Studies using allogeneic BMT with matched unrelated donors have demonstrated that advanced disease, older age and higher degrees of HLA disparity are associated with a poor outcome due to excessive morbidity and mortality of GVHD.⁴ Therefore, other treatment strategies are necessary for the majority of patients with AML.

ABMT is a promising therapy for the treatment of AML with several advantages over allogeneic BMT: the lack of a bone marrow donor does not preclude treatment, it can be used in patients as old as 65 years and there is no problem with GVHD. There has been concern that the relapse rate with ABMT will be higher than that seen with allogeneic BMT, due to the potential reinfusion of marrow contaminated with clonogenic leukemia cells and the absence of a graft-versus-leukemia effect. However, numerous studies have shown that ABMT can result in outcomes that compare very favorably to those of allogeneic BMT.^{5,6,7,8,9,10,11,12} To increase the efficacy of this treatment, methods of purging autologous marrow using moAb or cytotoxic drugs have been evaluated.^{13,14,15,16,17,18}

We have been using cytotoxic moAb originally created by one of us (EDB) and that react specifically with myeloid cells and recognize antigens expressed on AML blast cells.^19,20 Of these moAb, PM-81 (anti-CD15) and AML-2-23 (anti-CD14) are the most reactive, binding with leukemia cells from greater than 95% of AML patients.^19,20 These moAb are cytotoxic to cells bearing the respective cell surface antigens in the presence of complement (C'), and thus can lyse leukemia cells from almost all patients with AML, including their progenitor cells.²¹ In addition, PM-81 and AML-2-23 do not recognize antigens on multi-lineage progenitor cells necessary for successful engraftment of bone marrow.²¹

From 1 August 1984 until 10 February 1997, 140 purged autologous transplants were performed on 138 patients who were in CR or first relapse (R1) at the time of transplant. Sixty of these patients were described previously.¹⁷ We now report on the long-term results of these phase II studies.

Patients and methods

Patients

Patients less than 65 years of age with a Karnofsky performance status of 80-100% and an expected survival time of greater than 2 months were eligible for this protocol.

At the time of bone marrow harvest, remission was required to be documented by bone marrow aspirate and biopsy. Patients underwent harvest in either first, second or third CR. Transplants were performed in either CR1, CR2, CR3, or first relapse. All patients had a left ventricular ejection fraction 50%, a DLCO of 60% or an FEV 75% predicted as well as adequate renal and liver function as determined by a serum creatinine 2 times normal, and a bilirubin, SGOT (AST), and alkaline phosphatase 3 times normal, respectively. Leukemia blast cells obtained at diagnosis or at relapse, when available, were required to express the antigens reactive with PM-81 and/or AML-2-23 on >20% of cells. The study was approved by the Institutional Review Board of the respective institutions and a signed informed consent was obtained from each patient prior to study entry.

One hundred and thirty-eight AML patients (with two patients undergoing retransplantation) ranging in age from 2 to 57 and who were in CR or R1 were transplanted between August 1984, and 10 February 1997 (Table 1). All but 11 patients had de novo AML at the time of initial diagnosis. Eleven patients had a myelodysplastic syndrome (MDS) before the diagnosis of AML. Twenty patients were treated on CALGB protocols 8882 or 8781. Four patients were transplanted at the Scripps Clinic (La Jolla, CA), 59 patients at the Dartmouth-Hitchcock Medical Center (Hanover, NH), 14 patients at Bowman-Gray School of Medicine (Winston-Salem, NC), four patients at the Medical Center of Delaware (Newark, DE), six patients at the University of Iowa Hospitals, 23 patients at the University of Pittsburgh, four patients at the University of Alabama, two patients at Children's Hospital of Pittsburgh, one patient at the University of Colorado, four patients at Miami Childrens's Hospital, four patients at Shands Hospital (Gainsville, FL), 10 patients at Standford University Medical Center, and one patient at the Harris Methodist Hospital (Ft Worth, TX).

The FAB subclasses of the cases were as follows: M1/M2, 61; M3, 16; M4/M5, 51; M6/M7, three; biphenotypic, one; unknown, seven. Cytogenetic analysis on the leukemia cells at presentation were available from the patient records of 103 patients. Of these, 52 patients had a normal karyotype, nine had abnormalities of 16q, five had 7q-, four had 8+, nine had t(15;17), eight had t(8;21). The distribution of karyotypes in patient groups defined by CR status at transplant was not significantly different (P = 0.895).

The median time between the current remission or relapse and ABMT (n = 137) was 53 days (range 3-420 days). The median time between the current remission or relapse and ABMT for patients in CR1 was 138 days (24-421 days). The median time between the current remission or relapse and ABMT for patients in CR2 was 64 days (3-391 days). The median time between the current remission or relapse and ABMT for patients in CR3 was 36 days (24-83 days). The median time between the current remission or relapse and ABMT for patients transplanted in R1 was 15 days (4-374 days). Sixty-eight patients were harvested in CR1, 64 in CR2, and seven in CR3. One marrow harvest was performed in a patient with an extramedullary relapse (breast chloroma, marrow was normal).

Marrow harvesting and purging

Bone marrow was harvested from the posterior and anterior iliac crests under general anesthesia and passed through a series of filters.²¹ Efforts were made to harvest 6 ´ 10⁸ cells/kg from each patient. A mean of 5.37 ´ 10⁸ cells/kg were actually harvested. Bone marrow mononuclear cells were prepared first by buffy coat concentration by apheresis of the marrow. The buffy coat preparation was further treated in some cases on a Ficoll-Hypaque gradient centrifugation on the Haemonetics automated cell processor to obtain a mononuclear cell preparation to be treated with moAb + C'. A mean of 1.4 ´ 10⁸ cells/kg were treated; from these, there was a mean recovery of 47%. An average of 6.59 ´ 10⁷ cells/kg was used for the transplant.

Saturating amounts of purified moAb were preincubated with these cells as previously described.^22,23 To ensure saturation of all antigenic sites, the amount of each moAb used was 10 U per 10⁶ cells. Treatment on the Haemonetics cell processor was performed for 1 h with continuous exposure to fresh C' and simultaneous removal of spent C', while centrifuging at room temperature.²² The moAb treatment was performed in the presence of the enzyme deoxyribonuclease (10 U/ml) to decrease cell clumping. This treatment was performed on the Haemonetics cell processor for patients treated after May 1987 at the Dartmouth-Hitchcock Medical Center, Bowman Gray Medical Center, and at the University of Pittsburgh. Before that date at the DHMC, Scripps Clinic, Children's Hospital and Health Center (San Diego), and the Medical Center of Delaware, the marrow cells were treated in plastic or Teflon vessels (Savillex, Minnitonka, MN, USA) by gentle shaking. For these treatments, two separate incubations with moAb and C' were performed as previously described.²³ As of March 1996, the AML-2-23 moAb was eliminated from the purging regimen based on in vitro experiments demonstrating that AML-2-23 did not add to the cytotoxicity mediated by PM-81 alone.

Cells were then washed and resuspended in a mixture of medium 199 containing 10% dimethyl sulfoxide (Tera Pharmaceuticals, Buena Park, CA, USA) and 5% irradiated autologous plasma, frozen at 1°C/min in a controlled-rate freezer and stored in the vapor phase of liquid nitrogen.

Preparative regimens

Thirty-five patients were treated with the following preparative regimen: cyclophosphamide (CY) (60 mg/kg i.v. for 2 days) (days -5 to -3) and fractionated total body irradiation (TBI) (200 cGy twice daily for 3 days, total dose of 1200 cGy, dose rate = 5-10 cGy/min) (days -2 to 0). In 1988, the preparative regimen was changed from CY/TBI to busulfan (BU)/CY2. Ninety-three patients were treated with BU (4 mg/kg/day orally for days -8 to -5) and CY2 (60 mg/kg/day intravenously for 2 days) (days -4 and -3). One patient in CR2 was treated with BU (4 mg/kg/day orally for 4 days (days -9 to -6) and CY (50 mg/kg/day intravenously for 4 days) (days -5 to -2). Ten patients (Stanford) were conditioned with BU (4 mg/kg/day orally days -7 to -4) and VP-16 (60 mg/kg day -3).

Statistical methods

Two patient sets were included in this analysis: The first was the total set of 138 patients and the second was the set of 55 patients who underwent ABMT in CR2/3 and received BU/CY as their conditioning regimen. For both analyses, the last transplant represented was performed on 10 February 1997 and the last follow-up contact was 25 September 1997.

We estimated rates of relapse, OS, and DFS following ABMT (and 95% confidence intervals) using the product-limit or Kaplan-Meier method.²⁵

In addition, proportional hazards regression (Cox regression, 'relative risk' regression) was used to investigate the association of factors of interest with OS and DFS in these patients. This method assumes that although the survival rates may differ between groups of patients, the probability of death or relapse at a given time maintains a constant ratio throughout follow-up. For the factors examined in this analysis, the assumption appears to be satisfied. The factors listed in Table 2 were allowed to enter the OS and DFS models in a 'forward stepwise' fashion (ie factors were evaluated first individually to see whether they contributed substantially to predicting outcome). The most significant univariate factor was then 'forced into' the model, other factors being allowed to enter if they contributed significantly given the presence of the first. Additional factors were allowed to enter the models with interactions as appropriate.

Some factors were transformed before analysis using logarithms to make their distributions more symmetric. Body mass index was calculated as:

weight (kg)/(height in m)²

Results

Engraftment

A median number of 4.0 ´ 10⁷ cells/kg body weight (range 2.30-8.23 ´ 10⁷) were infused into each CR1 patient. The median number of cells transfused into the CR2/3 group was 2.80 ´ 10⁷ (range 0.075-1.16 ´ 10⁸). A median number of 4.10 ´ 10⁷ cells/kg body weight (range 2.38-59.6 ´ 10⁷) were infused into each R1 patient.

Median observed recovery times for neutrophils to 500 cells/l was 33 days. Median times to reach platelet counts of greater than 20 000 and greater than 50 000/l independent of platelet transfusions were 51 and 79 days, respectively.

CD34⁺ cell analysis was available from a subset of 18 patients transplanted at the University of Pittsburgh. There was no significant association between CD34⁺ cells/kg body weight and relapse or OS (P = 0.17 and 0.71, respectively). There was a significant association (P = 0.036) between CD34⁺ cells/kg and neutrophil recovery to 500/l as well as platelet recovery to >20 000 platelets/kg (P = 0.047).

Disease-free and overall survival

The DFS from transplant of all patients as of 25 September 1997 are shown in Figures 1,2 and 3 by status at transplant (CR1, CR2/3, and R1).

CR1: For patients in CR1 treated with CY/TBI (n = 7) 3- and 5-year DFS rates were 71% and 57% (95% CI: 38-100% and 20-94%) (Tables 3 and 4). For CR1 patients treated with BU/CY (n = 12) 3- and 5-year DFS rates were both 45% (95% CI: 16-75%). Overall survival at 3 and 5 years for the patients treated with CY/TBI was 71% and 57% (95% CI: 38-100% and 28-94%). OS for patients treated with BU/CY at 3 and 5 years was 46% (95% CI: 16-75%).

CR2/3: Three- and 5-year DFS for patients in CR2/3 treated with CY/TBI (n = 26) were both 23% (95% CI: 7-40%) (Tables 3 and 4). Three- and 5-year DFS for patients in CR2/3 treated with BU/CY (n = 55) was 31% and 28%, respectively (95% CI: 17-44% and 14-41%). OS for patients treated with CY/TBI at 3 and 5 years was 27% (CI: 10-44%). OS for the patients treated with BU/CY at 3 and 5 years was 38% and 30% (CI: 24-52% and 16-45%).

R1: Three- and 5-year DFS for patients in R1 treated with BU/CY (n = 26) were both 37% (Tables 3 and 4). OS for the patients treated in R1 with BU/CY at 3 and 5 years was 38% (95% CI: 18%-58%).

Eleven CR1 patients survive disease-free at a median follow-up time of 49 months (range 3-126 months). Twenty-seven patients transplanted in CR2/3 survive disease-free at a median time of 55 months (range 1-149 months). Ten patients transplanted in R1 survive disease-free at a median time of 47 months (range 5-96 months). For 48 of 115 CR2/3 and R1 patients, the duration of the post-ABMT remission exceeded the duration of CR1 or CR2.

Relapse

The actuarial relapse rates for different patient groups were calculated using the Kaplan-Meier method. For all patients (n = 138) the relapse rates (RR) were 58% and 61% at 3 and 5 years. For patients in CR1, the RR were 50% and 55% at 3 and 5 years. For patients in CR2/3 the rates were 62% and 65% at 3 and 5 years. For patients in R1 the rates were 56% and 56% at both 3 and 5 years.

Examined by regimen the results were as follows: CR1: CY/TBI, 29% and 43%, at 3 and 5 years; BU/CY: 55% and 55% at 3 and 5 years. CR2/3: CY/TBI: 56% and 56% at 3 and 5 years; BU/CY: 62% and 65% at 3 and 5 years. Patients in R1 treated with the BU/CY regimen: 54% and 54% at 3 and 5 years.

Median time to relapse for patients who relapsed (n = 65) was 5.7 months. Median time to death or relapse (n = 92) was 4.5 months.

Multivariate analysis

Overall survival

All Patients (n = 138): When the factors in Table 2 were evaluated in univariate fashion, age and age 30 were the only variables significantly related to OS (P = 0.047 and 0.025, respectively). As expected, higher age is associated with increased risk of death. When age 30 was 'forced into' the proportional hazards model (ie in a 'multivariate' analysis), no additional factors were found to be significant.

CR2/3 patients conditioned with BU/CY (n = 55): Among these patients, the strongest single predictor of OS was time in CR1 (P = 0.002). Longer CR1 duration was associated with decreased risk of death. In addition, male patients had a higher risk of death (P = 0.045).

When CR1 duration was included in a proportional hazards model, BMI (P = 0.047), gender (P = 0.035), FAB 1 or 2 (P = 0.022) and an extra chromosome 8 (P = 0.048) were all associated with risk of death. The directions of these associations depended on the CR1 duration (ie each factor showed an interaction with CR1). Females with CR1 duration longer than 6 months showed a decreased risk of death compared to males. However, the reverse association was true for females with shorter CR1 duration times. FAB 1 or 2 was a risk factor for death among patients with CR1 duration longer than 6 months. However, FAB 1 or 2 patients with shorter CR1 duration had a decreased risk of death. Patients with high BMI and CR1 duration between 6 and 9 months had a decreased risk of death. Such patients with shorter and longer CR1 duration times were at increased risk. Finally, there was apparently an effect of CR1 time on the risk of death due to an extra chromosome 8. However, because of model fitting problems, it was not possible to describe the interaction between those factors.

Disease-free survival

All patients: Patient age and age 30 were the only variables significantly related to DFS (P = 0.026 and 0.015, respectively). As was observed in proportional analysis of OS, higher age was associated with increased risk of death or relapse. When age 30 was 'forced into' the proportional hazards model, no additional factors were found to be significant.

CR2/3 patients conditioned with BU/CY: The only univariate predictor of DFS was time in CR1 (P = 0.003). Longer CR1 duration was associated with decreased risk of death or relapse. When CR1 duration was forced into the model predicting DFS, several additional factors were significantly associated with that outcome: FAB class 1 or 2 (P = 0.045), MDS or secondary AML (P = 0.017), and PM-81 sensitivity (P = 0.025). MDS or secondary AML was strongly associated with increased risk of death or relapse. There were interactions between the other two factors and CR1 duration, although the relationship between PM-81 sensitivity and DFS was obscured by model-fitting problems. Patients with FAB class 1 or 2 and CR1 durations less than 6 months had a lower risk of death or relapse than did patients in the other FAB classes. In contrast, FAB class 1 or 2 patients with longer CR1 durations had a lower risk of death or relapse.

Time to relapse

All patients: In the univariate analysis, conditioning regimen (P = 0.04), age at ABMT (P = 0.03), extramedullary disease (P = 0.02), -7,7q karyotype (P = 0.04), normal karyotype (P = 0.01), and karyotype-based prognosis (P = 0.01) were associated with time to relapse. Specifically, the Bu/CY and Bu/VP16 regimens had slightly stronger associations with relapse than did the Cy/TBI regimen. Increased age, extramedullary disease, -7,7q and normal karyotypes were all positively associated with relapse. In particular, a 'normal' karyotype was a worse prognostic factor than either 'good' or 'indeterminate' karyotype. Those karyotypic factors considered 'poor' for prognosis were apparently no worse than a normal karyotype.

In the proportional hazards model resulting from the forward stepwise selection process, several factors were jointly significant in predicting time to relapse: age at ABMT (P = 0.001), extramedullary disease (P = 0.013), karyotype (normal, -7,7q, or other abnormality; P = 0.001), and history of MDS (P = 0.019). Increasing age was associated with shorter relapse times, although the association was stronger among patients with extramedullary disease. Extramedullary disease was associated with shorter relapse times, with the effect more pronounced among younger patients than among older patients. Patients with a history of MDS tended to have shorter relapse times than did those without such a history. Finally, patients with a -7,7q karyotype were more likely to relapse than were patients with a normal karyotype. However, patients with other chromosomal abnormalities were less likely to relapse than were patients with a normal karyotype.

Bu/Cy patients transplanted in CR2/3: When potential risk factors were tested individually for association with time to relapse in this patient group, time in CR1 was the strongest individual predictor (P = 0.002). Patients in CR1 longer than 208 days had longer times to relapse than did patients with shorter CR1 durations. In addition, patients with normal karyotypes were at greater risk for relapse than were patients with karyotypic abnormalities (P = 0.037).

Duration in CR1 (P < 0.001), karyotype prognosis (P = 0.002), and number of cells transplanted (P = 0.004) were jointly predictive of relapse in the 'multivariate' analysis. As in the univariate analysis, patients in CR1 longer than 208 days had longer times to relapse than did patients with shorter CR1 durations. A higher number of cells transplanted was associated with a shorter relapse time. As with the total patient group, a 'normal' karyotype was a worse prognostic factor than either 'good' or 'indeterminate' karyotype. However, karyotypic factors considered 'poor' for prognosis carried a stronger association with relapse than did a normal karyotype.

Discussion

Bone marrow transplantation offers the potential for complete elimination of occult leukemia cells after initial remission induction, and BMT is probably the only curative treatment for patients with AML after R1.⁴ Allogeneic BMT is established as a potentially curative therapy for patients in first remission, but the majority of patients with AML cannot undergo this therapy due to lack of an HLA-matched donor and/or advanced age.⁴ This report and others^{5,6,7,8,9,10,11,12,13,14,15,16,17,18} show that ABMT is a viable alternative.

Since reinfused marrow may be contaminated with residual malignant cells after ABMT, we have used ex vivo purging to eliminate residual neoplastic cells from the graft. Long-term survival for AML patients after ABMT using various methods for removing occult leukemia cells has been reported.^{12,13,14,15,16,17,18} The benefit of mafosfamide purging for patients transplanted in first CR within 6 months of attaining CR has been reported.¹⁶ Chao et al⁹ published phase II trial results showing that patients who received purged bone marrow (4-hydroperoxycyclophosphamide (4-HC) and/or etoposide) had an actuarial DFS of 57% compared with a DFS of 32% in patients who received unpurged bone marrow. Yeager et al¹³ have reported favorable results similar to allogeneic BMT with 4-HC marrow purging in patients with AML who underwent ABMT. Miller et al²⁶ recently demonstrated that 4-HC purged ABMT, reported to the Autologous Bone Marrow Transplantation Registry (ABMTR), had significantly better outcomes than unpurged ABMT in both CR1 and CR2/3.

Brenner et al,²⁷ using the neomycin-resistant gene as a marker for AML relapse, demonstrated that autologous marrow harvested from leukemia patients in remission may harbor malignant cells capable of contributing to relapse. Thus, effective marrow purging may be essential for improving the outcome of ABMT for AML.

MoAb-based techniques using anti-myeloid moAb have been used to purge AML marrow. This report updates our greater than 10 year multi-institutional clinical data of ABMT in AML with moAb and C-mediated purging. As with allogeneic BMT, the results are dependent on remission status. Five-year DFS for patients transplanted in R1 and CR2/3 who were conditioned with BU/CY2 were 37% and 28%, respectively. These results compare favorably with those of allogeneic BMT.^4,28

Significant prognostic factors operating in our study were length of first CR (for patients transplanted in R1 or CR2/3), and gender. Interestingly, patients with secondary AML, or extramedullary disease did not fare worse. Nor did cytogenetics of AML cells at diagnosis affect outcome (although some of the patient subgroups were small). These results are surprising since prior CNS disease and cytogenetic abnormalities have been associated with worse outcomes in previous studies.^29,30,31

We have conducted a phase II clinical trial of monoclonal antibody purging of bone marrow in patients with AML in remission at the time of harvest and in remission or relapse at the time of transplant. The overriding question raised by the favorable outcomes reported for purged autologous bone marrow transplants is that of how much the purging contributed to the outcomes. This is stimulated by the occasional reports that unpurged marrow transplants have been associated with results often comparable to purged transplants. However, recent evidence has been reported supporting the notion that purging is probably effective in reducing late relapses in AML.²⁶ Thus, a randomized trial comparing purged to unpurged stem cell transplants in patients with AML in remission may help resolve this issue.

Acknowledgements

This work was supported by a grant CA31888 from the National Institutes of Health (EDB). We wish to acknowledge the contributions of the following physicians who supervised the treatment of patients at the participating institutions: Robert McMillan (Scripps Clinic and Research Foundation, La Jolla, CA), Letha Mills (Dartmouth-Hitchcock Medical Center, Lebanon, NH), David Hurd (Bowman Gray Medical Center, Winston-Salem, NC), Eric Martin (Medical Center of Delaware, Newark, DE), Roger Gingrich (University of Iowa Hospitals, Iowa City, IO), Matthew Carabasi (University of Alabama Medical Center, Birmingham, AL), Elizabeth Shpall (University of Colorado Health Science Center, Denver, CO), Charles August (Miami Children's Hospital, Miami, FL), Nayesh Kamani (University of Texas, San Antonio, TX), Aly Abdel-Mageed (University of Florida, Gainsville, FL), Robert Negrin (Stanford University, Palo Alto, CA), and David Friedman (Harris Methodist Hospital, Ft Worth, TX).

1 Bishop JF. The treatment of adult acute myeloid leukemia. Semin Oncol 1997; 24: 57-69, MEDLINE

2 Woods WG, Kobrinsky N, Buckley JD et al. Timed-sequential induction therapy improves postremission outcome in acute myeloid leukemia: a report from the Children's Cancer Group. Blood 1996; 87: 4979-4989, MEDLINE

3 Ravindranath Y, Yeager AM, Chang MN et al. Autologous bone marrow transplantation versus intensive consolidation chemotherapy for acute myeloid leukemia in childhood. New Engl J Med 1996; 334: 1428-1434, MEDLINE

4 Appelbaum FR. Allogeneic hematopoietic stem cell transplantation for acute leukemia. Semin Oncol 1997; 24: 114-123, MEDLINE

5 Körbling M, Hunstein W, Fliedner TM et al. Disease-free survival after autologous bone marrow transplantation in patients with acute myelogenous leukemia. Blood 1989; 74: 1898-1904, MEDLINE

6 Löwenberg B, Abels J, van Bekkum DW et al. Transplantation of non-purified autologous bone marrow in patients with AML in first remission. Cancer 1984; 54: 2840-2843, MEDLINE

7 McMillan AK, Goldstone AH, Linch DC et al. High-dose chemotherapy and autologous bone marrow transplantation in acute myeloid leukemia. Blood 1990; 76: 480-488, MEDLINE

8 Labopin M, Gorin NC. Autologous bone marrow transplantation in 2502 patients with acute leukemia in Europe: a retrospective study. Leukemia 1992; 6: (Suppl. 4) 95-99, MEDLINE

9 Chao NJ, Stein AS, Long GD et al. Busulfan/Etoposide - initial experience with a new preparatory regimen for autologous bone marrow transplantation in patients with acute nonlymphoblastic leukemia. Blood 1993; 81: 319-323, MEDLINE

10 Zittoun RA, Mandelli F, Willemze R et al. Autologous or allogeneic bone marrow transplantation compared with intensive chemotherapy in acute myelogenous leukemia. New Engl J Med 1995; 332: 217-223, MEDLINE

11 Harousseau JL, Cahn JY, Pignon B et al. Comparison of autologous bone marrow transplantation and intensive chemotherapy as postremission therapy in adult acute myeloid leukemia. The Groupe Ouest Est Leucemies Aigues Myeloblastiques (GOELAM). Blood 1997; 90: 2987-2986, MEDLINE

12 Burnett AK, Goldstone AH, Stevens RMF et al. Randomized comparison of addition of autologous bone-marrow transplantation to intensive chemotherapy for acute myeloid leukaemia in first remission: result of MRC AML 10 trial. Lancet 1998; 351: 700-708, MEDLINE

13 Yeager AM, Kaizer H, Santos GW et al. Autologous bone marrow transplantation in patients with acute nonlymphocytic leukemia, using ex vivo marrow treatment with 4-hydroperoxycyclophosphamide. New Engl J Med 1986; 315: 141-147, MEDLINE

14 Ferrero D, DeFabritis P, Armadori S et al. Autologous bone marrow transplantation in acute myeloid leukemia after in vitro purging with an anti-lacto-n-fucopentaose III antibody and rabbit complement. Leukemia Res 1987; 11: 265-272,

15 Rowley SD, Jones RJ, Piantadosi S et al. Efficacy of ex vivo purging for autologous bone marrow transplantation in the treatment of acute nonlymphoblastic leukemia. Blood 1989; 74: 501-507, MEDLINE

16 Gorin NC, Aegerter P, Auvert B et al. Autologous bone marrow transplantation for acute myelocytic leukemia in first remission: a European survey of the role of marrow purging. Blood 1990; 75: 1606-1614, MEDLINE

17 Selvaggi KJ, Wilson J, Mills LE et al. Improved outcome for high risk acute myeloid leukemia patients using autologous bone marrow transplantation and monoclonal antibody purged bone marrow. Blood 1994; 83: 1698-1705, MEDLINE

18 Ball ED, Phelps V, Wilson J. Autologous bone marrow transplantation for acute myeloid leukemia in remission or first relapse using monoclonal antibody-purged marrow. In: Dicke KA, Keating A (eds). Autologous Marrow and Blood Transplantation. Carden Jennings Publishing: Charlottesville, VA, 1997, pp45-53.

19 Ball ED, Graziano RF, Fanger MW. A unique antigen expressed on myeloid cells and acute leukemia blast cells defined by a monoclonal antibody. J Immunol 1983; 130: 2937-2941, MEDLINE

20 Ball ED, Fanger MW. The expression of myeloid-specific antigens on myeloid leukemia cells: correlations with leukemia subclasses and implications for normal myeloid differentiation. Blood 1983; 61: 456-463, MEDLINE

21 Sabbath KD, Ball ED, Larcom P et al. Heterogeneity of clonogenic cells in acute myeloblastic leukemia. J Clin Invest 1985; 75: 746-753, MEDLINE

22 Howell AL, Fogg-Leach M, Davis BH, Ball ED. Continuous infusion of complement by an automated cell processor enhances cytoxicity of monoclonal antibody sensitized leukemia cells. Bone Marrow Transplant 1989; 4: 317-322, MEDLINE

23 Howell AL, Ball ED. Monoclonal antibody-mediated cytotoxicity of human myeloid leukemia cells: an in vitro model for estimating efficiency and optimal conditions for cytolysis. Blood 1985; 66: 649-654, MEDLINE

24 Armitage P, Berry G. Statistical Methods in Medical Research. Blackwell Scientific Publishers: Boston, MA, 1987,

25 Collett D. Modelling Survival Data in Medical Research. Chapman and Hall: London, 1994,

26 Miller CB, Rowlings PA, Jones RJ et al. Autotransplants for acute myelogenous leukemia (AML): effect of purging with 4-Hydroperoxycyclophosphamide (4 HC). Proc ASCO 1996; 15: 338-000,

27 Brenner MK, Rill DR, Moen RC et al. Gene-marking to trace origin of relapse after autologous bone marrow transplantation. Lancet 1996; 341: 85-86,

28 Buckner CD, Clift RA. Clinical studies of allogeneic marrow transplantation in patients with acute nonlymphoblastic leukemia. Seattle Marrow Transplant team. Bone Marrow Transplant 1989; 3: (Suppl) 82,

29 van Besian K, Przepiorka D, Mehra R et al. Impact of preexisting CNS involvement on the outcome of bone marrow transplantation in adult hematologic malignancies. J Clin Oncol 1996; 14: 3036-3042, MEDLINE

30 Mrozek K, Heinonen K, de la Chapelle A et al. Clinical significance of cytogenetics in acute myeloid leukemia. Semin Oncol 1997; 24: 17-31, MEDLINE

31 Barnard DR, Kalousek DK, Wiersma SR et al. Morphologic, immunologic, and cytogenetic classification of acute myeloid leukemia and myelodysplastic syndrome in childhood: a report from the Childrens Cancer Group. Leukemia 1996; 10: 5-12, MEDLINE

Figure 1 Disease-free survival for patients transplanted in CR1 with three different preparative regimens.

Figure 2 Disease-free survival for patients transplanted in CR2/3 with three different preparative regimens.

Figure 3 Disease-free survival for patients transplanted in R1 with two different preparative regimens.

Table 1  Clinical characteristics of patients

Table 2  Factors investigated using proportional hazards regression

Table 3  1, 2 and 3 year survival by regimen and remission status

Table 4  Actuarial overall and disease-free survival (DFS) at 5 years in patients grouped by preparative regimen and remission status at time of ABMT