Fax.: 011 (49) 8999017209
High-dose cytosine arabinoside in the treatment of acute myeloid leukemia
Review of three randomized trials
Article first published online: 23 MAY 2006
Copyright © 2006 American Cancer Society
Volume 107, Issue 1, pages 116–124, 1 July 2006
How to Cite
Kern, W. and Estey, E. H. (2006), High-dose cytosine arabinoside in the treatment of acute myeloid leukemia. Cancer, 107: 116–124. doi: 10.1002/cncr.21543
Fax.: 011 (49) 8999017209
- Issue published online: 16 JUN 2006
- Article first published online: 23 MAY 2006
- Manuscript Accepted: 16 JUL 2005
- Manuscript Revised: 26 JUN 2005
- Manuscript Received: 9 MAY 2005
- acute myeloid leukemia;
- high-dose cytosine arabinoside;
- standard-dose cytosine arabinoside
The use of high-dose cytosine arabinoside (HDAraC) during induction may improve outcomes in patients with acute myeloid leukemia (AML) compared with standard-dose AraC (SDAraC). The objective of this review was to assess the impact of HDAraC during induction therapy for patients with AML based on results from randomized trials.
All randomized trials in the field were identified by using a predefined search strategy. Trials that assessed the impact of HDAraC compared with SDAraC as induction therapy for adult patients with AML in a randomized fashion and that reported the relevant endpoints were included. Data were extracted from each trial by both reviewers according to prespecified criteria.
No differences between HDAraC and SDAraC were found with regard to complete remission rates (relative risk, 1.00; 95% confidence interval [95% CI], 0.92-1.10). The weighted mean difference (WMD) for median recurrence-free survival (RFS) was 4.19 in favor of HDAraC (95% CI, 0.59-7.78; P = .02). The WMD for 4-year RFS was 10.98 in favor of HDAraC (95% CI, 1.02-20.94; P = .03). The WMD for median overall survival (OS) was − 0.22 for HDAraC compared with SDAraC (95% CI, − 2.76-2.32; P = .9). Data regarding the median OS was heterogeneous between studies (chi-square P = .00), with 2 studies in favor of HDAraC and 2 studies in favor of SDAraC. The WMD for 4-year OS was 6.21 in favor of HDAraC (95% CI, 2.70-9.72; P = .0005).
Induction therapy with HDAraC improved long-term disease control and overall survival in adults age < 60 years with de novo AML. It remains unknown whether patients should receive HDAraC during induction or if it is to be given during postremission therapy. Further analyses should focus on this issue and on the effects of HDAraC in prognostically different subgroups of patients with AML. Cancer 2006. © 2006 American Cancer Society.
Acute myeloid leukemia (AML) is a clonal disorder resulting from acquired somatic mutations in hematopoietic progenitor cells that lead to the dysregulation of differentiation and proliferation.1 The incidence of AML is 2.7 per 100,000 population and increases with age. Based on Surveillance, Epidemiology, and End Results Registry data, the 5-year survival rates for patients with AML rose from 5.8% in the 1970s to 14.6% in the 1990s.2 The 5-year survival rates range from 35.9% in patients age <45 years to 1.3% in patients age >75 years. The etiology of AML is unknown in most patients. However, in 8% of patients, the disease is preceded by a myelodysplastic syndrome; and, in another 6% of patients, AML is associated with prior cytotoxic therapy.
For distinct subgroups of patients with AML who have specific balanced chromosome translocations, the associated fusion genes have been well characterized, and their role in leukemogenesis has been demonstrated. Thus, fusion of the AML1 and ETO genes, of the PML and the RARα genes, and of the CBFβ and the MYH11 genes are the basis for the type of AML with chromosome translocations t(8;21) and t(15;17) and for the type of AML with the inversion inv(16).3 The basis for malignant transformation has been characterized less well in AML with other cytogenetic aberrations and particularly in AML with normal karyotypes, which comprise approximately 50% of all patients with AML. In those patients, genetic alterations, like mutations of the FLT3 and the MLL genes, that are not detectable at the chromosomal level, may have leukemogenic potential.
Along with their role in pathogenesis, cytogenetic aberrations comprise the most important disease-related prognostic factors, with t(8;21), inv(16), and t(15;17) representing the group of patients who have relatively favorable outcomes irrespective of the treatment modality applied (i.e., chemotherapy or autologous or allogeneic transplantation).4 Conversely, patients who have AML associated with aberrations of chromosomes 5 and 7 and with complex karyotype aberrations in particular generally have an unfavorable outcome, and only approximately 50% of those patients achieve a remission, which is a long-lasting remission only in a minority of patients.5 Among the patient-related prognostic factors, it has been documented that age is very significant with respect to both early response and long-term outcome; however, the associations of older age with unfavorable cytogenetics and with a greater frequency of impaired performance status suggest that the prognostic impact of age, by itself, may be based on other factors.
The treatment of patients with AML includes of at least 1 course of intensive myelosuppressive induction chemotherapy, followed by additional courses of intensive consolidation and maintenance therapies, and by autologous and allogeneic transplantation procedures.6 In the 1970s, 3 + 7 induction therapy regimen, which consists of 3 days of daunorubicin and 7 days of standard-dose cytosine arabinoside (SDAraC), became standard. Although the dose intensity of daunorubicin may have an impact on patient outcomes, various studies that focused on alternatives to daunorubicin and on the addition of other drugs to the 3 + 7 regimen did not result in an advantage in the treatment of AML. However, the introduction of high-dose AraC (HDAraC) (i.e., ≥1 g/m2 per dose) has been a standard part of salvage therapies for a long time and has yielded favorable results during first-line therapy compared with SDAraC. However, a comprehensive analysis of the effects of HDAraC during induction therapy has not been performed to date and is the focus of the current review. The objectives of this review were to compare the effectiveness of HDAraC versus SDAraC as induction and consolidation therapy in the management of adult patients with AML.
MATERIALS AND METHODS
Types of studies
All relevant randomized controlled trials were considered. Nonrandomized trials and Phase II trials were excluded.
Types of participants
Patients had to be age >16 years and were required to have untreated, de novo AML or AML secondary to prior hematologic diseases or to prior chemotherapy or radiotherapy according to the French–American–British criteria.
Types of interventions
Trials were included if they compared HDAraC (≥1 g/m2 per dose) with SDAraC (≤0.2 g/m2 per dose) in induction therapy.
Types of outcome measures
Clinical efficacy was defined according to the following parameters: 1) overall survival (OS) (the time from randomization to death), 2) event-free survival (EFS) (the time from randomization to death, recurrence, or nonachievement of a complete response [CR], whichever was first), 3) disease-free survival (DFS) (the time from the achievement of a CR to death or recurrence, whichever was first), and 4) the CR rate.
Search Strategy for Identification of Studies
Trials were identified by computerized searches of the Cochrane Controlled Trials Register (CENTRAL/CCTR), MEDLINE (2000-2001), and EMBASE (1988-2001). In data bases other than CCTR, we used the highly sensitive search strategy for identifying reports of randomized controlled trials developed by Robinson et al.7
The following search strategy was used to search the Cochrane Library:
#1 (ACUT* AND ((((MYELO* and LEUKAEM*) OR LEUKEM*) OR LEUCEM*) OR LEUCAEM*))
#7 (((((#1 or #2) or #3) or #4) or #5) or #6)
#12 (((#8 or #9) or #10) or #11)
#13 (#7 and #12).
This search strategy was adapted for the data bases as outlined below. In addition, we searched the Internet data base of Grey Literature (SIGLE) and ongoing trials, as follows: www.controlled-trials.com; http://clinicaltrials.nci.nih.gov; http://clinicaltrials.gov/ct/gui; www.eortc.be/; www.ctc.usyd.edu.au/; and www.trialscentral.org/index.html.
We hand searched the conference proceedings of the American Society of Clinical Oncology (1983-2001) and the American Society of Hematology (1983-2001). Citations of all trials that were identified in the search were checked for additional references. We contacted experts in the field and pharmaceutical companies for further unpublished or ongoing trials. All searches were conducted between 1983 and 2001. No language restrictions were used.
The following search strategy was designed for SilverPlatter ASCII 3.0 WIN:
#1 randomized-controlled-trial in pt
#2 controlled-clinical-trial in pt
#3 explode “Randomized-Controlled-Trials”/all subheadings
#7 (#1 or #2 or #3 or #4 or #5 or #6)
#8 clinical-trial in pt
#9 explode “Clinical-Trials”/all subheadings
#10 (clin* near trial*) in ti
#11 (clin* near trial*) in ab
#12 ((singl* or doubl* or trebl* or tripl*) near (blind* or mask*)) in ti
#13 ((singl* or doubl* or trebl* or tripl*) near (blind* or mask*)) in ab
#14 explode “Placebos”/all subheadings
#15 placebo* in ti
#16 placebo* in ab
#17 random* in ti
#18 random* in ab
#19 explode “Research-Design”/all subheadings
#20 (#8 or #9 or #10 or #11 or #12 or #13 or #14 or #15 or #16 or #17 or #18 or #19)
#21 tg = comparative-study
#22 explode “Evaluation-Studies”/all subheadings
#25 (control* or prospectiv* or volunteer*) in ti
#26 (control* or prospectiv* or volunteer*) in ab
#27 (#21 or #22 or #23 or #24 or #25 or #26)
#34 (#28 or #29 or #30 or #31 or #32 or #33)
#35 (#7 or #20 or #27 or #34)
#36 (tg = animal) not (tg = human)
#37 (#35 not #36).
The following search strategy was designed for EMBASE:
#2 MAJOR-CLINICAL-STUDY (EMTAG 0150)
#3 (#1 or #2)
#4 (ACUT* AND MYELO* AND (LEUKAEM* or LEUKEM* OR LEUCEM* OR LEUCAEM*))
#6 (ACUT* AND NON-LYMPHO* AND (LEUKAEM* or LEUKEM* OR LEUCEM* OR LEUCAEM*))
#7 (#4 or #5 or #6)
#12 (#8 or #9 or #10 or #11)
#13 (#3 and #7 and #12).
Methods of the Review
Selection of trials
Each trial was assessed independently by the 2 reviewers, who decided whether the trial fulfilled the inclusion criteria. The reviewers were not blinded to the authors' names or institutions. Disagreements between the 2 reviewers were discussed, and agreement was reached. Trials were studied separately when patient subgroups were established according to cytogenetic criteria.
The data extraction was done independently by each reviewer. Excluded studies and the reasons for their exclusion were collected in a specific register. Characteristics of the included patients with AML were recorded. Direct contact with the authors may have been required if the information published in the study was not sufficient for its classification.
Assessment of methodological quality
The 2 reviewers assessed the methodological quality of each trial. Studies were rated by using the following criteria: method of randomization; allocation concealment; intention-to-treat analysis; description of withdrawals/drop-outs; preplanned α error; preplanned β error; and risk of selection bias because of inclusion criteria.
Data were synthesized by using the Cochrane Statistics package RevMan (version 4.0.4).
Heterogeneity of the selected trials was assessed by calculating the Q statistic for heterogeneity8 and by inspecting graphic presentations of the study results through the use of Forrest plots. Complementary graphic techniques to assess for heterogeneity, such as l'Abbe plot,9 were used when necessary.
Subgroups analyses were performed according to patient age and cytogenetics. A test for interaction was used.
Description of Studies
|Median age of patients (range), y||41 (15–60)||ND (15–64)||44 (16–60)|
|De novo AML/secondary AML||Yes/no||Yes/yes||Yes/no|
|ECOG performance status||0-3||NA||NA|
|No. of induction therapy courses||1||1||2|
|Ara-C dose (mg/m2 per day) during induction therapy|
|Standard arm||100 × 7 Days||200 × 7 Days||200 × 7 Days|
|Experimental arm||6000 × 4 Days||2000 × 6 Days||6000 × 3 Days|
|Combination partners during induction therapy|
|Standard arm||Daunorubicin and etoposide||Daunorubicin||Daunorubicin and thioguanine|
|Experimental arm||Daunorubicin and etoposide||Daunorubicin||Mitoxantrone|
|Uniform therapy in complete remission||Yes||No||Yes|
Methodological quality of included studies
All 3 trials met the inclusion criteria. The quality of the included trials was assessed by using previously specified criteria.
Method of randomization
Patients were randomized at the trial centers in the Australian Leukemia Study Group (ALSG) and the German AML Cooperative Group (AMLCG) studies. The method of randomization was not specified in the Southwest Oncology Group (SWOG) study.
There was no indication of impaired allocation concealment in any of the 3 studies.
In all 3 studies, analyses were performed on an intent-to-treat basis.
Description of withdrawals/drop-outs
In all 3 studies, withdrawn patients and drop-outs were described adequately.
Preplanned α error
A preplanned α error was indicated in the ALSG (0.05) and AMLCG (0.05) studies but was not indicated in the SWOG study.
Preplanned β error
A preplanned β error was indicated all 3 studies (0.80).
Risk of selection bias from inclusion criteria
The risk of selection bias because of inclusion criteria was present in all studies. The most important criterion was the upper age limit of 60 years (64 years in the SWOG study). Consequently, the results of this review apply only to adult patients younger that age 60 years, and no data are provided on the large group of older patients with AML. Another important aspect is the inclusion of patients with secondary AML in the SWOG study only. Thus, because the great majority of patients had de novo AML, the current review applies only to adult patients age <60 years with de novo AML.
The effect of HDAraC compared with SDAraC as induction therapy for patients with AML is evident in light of improved long-term outcomes, although induction results are not affected. Thus, there is virtually no effect on the rates of complete remission (relative risk [RR], 1.00; 95% confidence interval [95% CI], 0.91-1.10) (Fig. 1). There were no differences observed in the rates of persisting leukemia (RR, 0.88; 95% CI, 0.73-1.07) (Fig. 2) or early death (RR, 1.53; 95% CI, 0.84-2.78) (Fig. 3).
The data on induction results were highly homogeneous between studies, with the rates of complete remission and persistent leukemia close to an RR of 1.00 in each study. The rates of early death, however, were lower for HDAraC in the AMLCG study only, whereas they were higher for HDAraC in the other studies. Overall, however, there was no effect of HDAraC on the early death rate. Similarly, there was only a marginal effect of HDAraC on the median overall survival (weighted mean difference [WMD], − 0.218; 95% CI, − 2.761-2.325) (Fig. 4), although there was a more obvious effect on median recurrence-free survival (WMD, 4.225; 95% CI, 0.511-7.939) (Fig. 5).
Again, the results were heterogeneous between studies with a shorter median overall survival for HDAraC in the SWOG study and a longer median survival for HDAraC in the ALSG and AMLCG studies. The median recurrence-free survival was longer for patients who received HDAraC in all studies with the exception of the subgroup of elderly patients in the SWOG study.
The impact of HDAraC was strongest with regard to patient outcomes at 4 years. Thus, the effect of HDAraC was highly significant on 4-year overall survival (WMD, 6.211; 95% CI, 2.701-9.721) (Fig. 6) and on 4-year recurrence-free survival (WMD, 10.982; 95% CI, 1.025-20.939) (Fig. 7).
These effects were present in all studies, with WMDs ranging from 2.000 to 10.000 for 4-year overall survival and from 2.000 to 18.000 for 4-year recurrence-free survival. Because of the increased antileukemic efficacy of HDAraC versus SDAraC, it is most important to assess the effect of HDAraC on nonhematologic toxicities. Severe infections (World Health Organization [WHO] Grade 3, 4, or 5) occurred more frequently after HDAraC (RR, 1.21; 95%CI, 1.08-1.36) (Fig. 8), which must be considered a possible cause for the higher early death rate after HDAraC, although the latter difference was not significant. In addition, nausea and emesis (WHO Grade 3, 4, or 5) were recorded more frequently after HDAraC (RR, 1.49; 95% CI, 1.19-1.85) (Fig. 9). It is noteworthy that central nervous system toxicity (WHO Grade 3, 4, or 5) occurred more frequently after HDAraC (RR, 3.63; 95% CI, 1.72-7.66) (Fig. 10), which also possibly contributed to the higher early death rate.
The treatment of patients with AML remains unsatisfactory despite of improvements in chemotherapy and supportive care, because most patients do not achieve a cure and die from refractory leukemia. The application of HDAraC has been tested extensively beyond first-line therapy and is considered standard therapy in combination with an anthracycline in this setting as long as the therapeutic objective is the achievement of a complete remission. Although there have been many nonrandomized studies of HDAraC as first-line induction therapy for patients with AML, there has been no general agreement on this approach. For the current review, we analyzed the data available from randomized trials in this field.
The current analysis clearly demonstrated the superiority of HDAraC compared with SDAraC for the long-term control of AML and for long-term overall survival. However, it is noteworthy there was no evidence to show that HDAraC leads to an increase in complete remission rates. Accordingly, there were no significant differences in the rates of persistent leukemia or early death. Thus, it remains possible that HDAraC has no effect on overcoming the primary resistance of leukemic blasts, which are present even before the application of induction therapy. Rather, HDAraC eliminates leukemic cells that, otherwise, would persist during complete remission and give rise to a greater rate of hematologic recurrence compared with SDAraC. Clearly, these issues should be kept in mind when designing new trials, which should consider the use of HDAraC as induction therapy in patients who have a high recurrence rate; however, patients who have a low probability of achieving first complete remission may be candidates for alternative antileukemic regimens, but not for HDAraC.
The quality of the trials in general allowed us to consider the results of this review valid. Most important, all trials were reported on an intent-to-treat basis and included a complete description of withdrawals and drop-outs. With the exception of the SWOG trial, the methods of randomization and a preplanned α error were stated, and all trials reported a preplanned β error. Overall, there was no indication that, in view of these issues, the results of the review may be compromized. However, certain limitations should be kept in mind. One limitation of the trials and, thus, of the current review is the restricted inclusion criteria, which resulted in an analysis of adult patients younger than age 60 years (up to age 64 years in the SWOG trial). Furthermore, the majority of patients had de novo AML. Therefore, the results of this review must be considered valid only for adult patients younger than age 60 years with de novo AML. Based on the current data, we cannot speculate about whether HDAraC would have the same effects on older patients or on patients with AML as a secondary disease after a preceding hematologic disease or after prior chemotherapy or radiation therapy.
Overall, there was only limited heterogeneity between the 3 trials with regard to the endpoints analyzed. None of the trials demonstrated an effect of HDAraC on the results of induction therapy. Furthermore, whereas there was no improvement reported in the median overall survival from HDAraC in the SWOG trial, the 2 other trials demonstrated reproducible improvements in overall survival at 4 years. Similarly, a clear benefit from HDAraC with regard to the median recurrence-free survival was demonstrated in the AMLSG and AMLCG trials, whereas the other trials reported reproducible improvements in recurrence-free survival only at 4 years. With regard to differences in the side effects between HDAraC and SDAraC, there was no heterogeneity between trials at all. Therefore, with regard to the most important findings of the current review, i.e. the lack of impact on remission rates and the improvement of long-term disease control with the use of HDAraC as induction therapy for patients with AML, there is no heterogeneity between trials, and this may limit the validity of the findings.
Implications for Practice
The current review data indicate that the use of HDAraC, compared with SDAraC, as induction therapy for patients with AML improves the long-term control of the disease, although it has no effect on the rate of complete remissions. These results are limited to adult patients younger than age 60 years with de novo AML. Therefore, the administration of HDAraC should be considered in adult patients younger than age 60 years who have de novo AML, particularly in patients who have a high risk of disease recurrence. Patients who have a low probability of achieving complete remission cannot be expected to benefit from HDAraC induction therapy and, rather, should be considered for experimental treatment approaches. Given the large prognostic heterogeneity among patients with AML and the crucial importance of cytogenetics in this regard, HDAraC as induction therapy may be the best therapy for patients who have prognostically favorable and intermediate cytogenetics, whereas patients who have unfavorable cytogenetics have a low probability of benefiting from HDAraC given their poor remission rates.
Implications for Research
Further analyses should be performed to assess differences in the effects of HDAraC induction therapy between prognostically different subgroups of patients with AML. Thus, it is anticipated that patients with favorable and intermediate cytogenetics particularly will benefit from HDAraC, whereas patients with unfavorable cytogenetics will not. Clearly, these hypotheses should be analyzed based on the results from trials that already have been completed and from future randomized trials. An addition field of investigation should be the extension to older patients and patients with secondary AML. The modern management of patients with AML is led by the most important disease-specific and patient-specific factors (i.e., cytogenetics and performance status) rather than on the basis of age and AML as secondary disease, both of which are dependent on additional parameters. Thus, the limitation to younger patients with de novo AML, which was standard in the 1980s, should no longer be the practice in modern trials.
- 2National Cancer Institute. SEER cancer statistics review, 1973-1996. NIH Pub. No. 99-2789. Bethesda: National Cancer Institute; 1999.
- 12Double induction strategy for acute myeloid leukemia: the effect of high-dose cytarabine with mitoxantrone instead of standard-dose cytarabine with daunorubicin and 6-thioguanine: a randomized trial by the German AML Cooperative Group. Blood. 1999; 93: 4116–4124., , , et al.