`Induction-Consolidation Chemotherapy-The Value of
`Low-Dose Cytarabine for Maintenance of Remission, and an
`Assessment of Prognostic Factors in Acute Myeloid
`Leukemia in the Elderly: Final Report of the
`Leukemia Cooperative Group of the European Organization
`for the Research and Treatment of Cancer and the
`Dutch-Belgian Hemato-Oncology Cooperative Hovon Group
`Randomized Phase III Study AML-9
`
`By B. L[wenberg, S. Suciu, E. Archimbaud, H. Haak, P. Stryckmans, R. de Cataldo, A.W. Dekker, Z.N. Berneman,
`A. Thyss, J. van der Lelie, P. Sonneveld, G. Visani, G. Fillet, M. Hayat, A. Hagemeijer, G. Solbu, and R. Zittoun
`
`Purpose and Methods: Optimization of remission-in-
`duction and postremission therapy in elderly individuals
`with acute myeloid leukemia (AML) was the subject of a
`randomized study in patients older than 60 years. Remis-
`sion-induction chemaotherapywas compared between dau-
`nomycin (DNR) 30 mg/m 2 on days 1, 2, and 3 versus
`mitoxantrone (MTZ) 8 mg/m 2 on days 1, 2, and 3, both
`plus cytarabine (Am-C) 100 mg/m 2 on days 1 to 7. Follow-
`ing complete remission (CR), patients received one addi-
`tional cycle of DNR or MTZ chemotherapy and were then
`eligible for a second randomization between eight cycles of
`low-dose (LD)-Ara-C 10 mg/m 2 subcutaneously every 12
`hours for 12 days every 6 weeks or no further treatment.
`Results: A total of 242 patients was randomized to
`DNR and 247 to MTZ. Median age of both study groups
`was 68 years. Secondary AML was documented in 26%
`and 25% of patients in either ann. The probability of
`attaining CR was greater (P = .069) with MTZ (47%) than
`with DNR (38%). Median duration of neutropenia was 19
`(DNR) and 22 days (MTZ). The greater response rate to MTZ
`therapy correlated with reduced occurrence of chemother-
`apy resistance (32% v 47%, P = .001). With a median
`follow-up of 6 years, 5-year disease-free survival (DFS) is
`
`From the Daniel den Hoed Cancer Center; University Hospital; Depart-
`ment of Cell Biology and Genetics, Erasmus University, Rotterdam; Leyen-
`burg Hospital, The Hague; University Hospital, Utrecht; University Hospi-
`tal, Amsterdam, the Netherlands; European Organization for Research and
`Treatment of Cancer Data Center, Institut Bordet, Brussels; University
`Hospital, Antwerpen; Centre Hospitalier Universitaire, Liege, Belgium;
`H6pital Edouard Herriot, Lyon; Centre A. Lacassagne, Nice; Institut
`Gustave Roussy, Villejuifl Hpital Hbtel Dieu, Paris, France; Ospedale
`Maggiore, Milano; and Ospedale San Orsola, Bologna, Italy.
`Submitted April 2, 1997; accepted October 16, 1997.
`Supported in part by grants no. 2U10 CA 11488-16 through 5U10 CA
`11488-26 from the National Cancer Institute, Bethesda, MD.
`Address reprint requests to B. Ltwenberg, MD, Dr Daniel den Hood
`Cancer Center, PO Box 5201, 3008 AE Rotterdam, the Netherlands;
`Email lowenberg@haed.azr nl.
`© 1998 by American Society of Clinical Oncology.
`0732-183X/98/1603-0004$3.00/0
`
`8% in each arm. Overall survival estimates are not differ-
`ent between the groups (6% v 9% at 5 yrs). Poor perfor-
`mance status at diagnosis, high WBC count, older age,
`secondary AML, and presence of cytogenetic abnormalities
`all had an adverse impact on survival. SecondaryAML and
`abnormal cytogenetics predicted for shorter duration of
`CR. Among complete responders, 74 assessable patients
`were assigned to Ara-C and 73 to no further therapy.
`Actuarial DFS was significantly longer (P = .006) for Ara-C-
`treated (13% [SE = 4.0%] at 5 years) versus nontreated
`patients (7% [SE =3%]), but overall survival was similar (P =
`.29): 18% (SE = 4.6%) versus 15% (SE = 4.3%). Meta-
`analysis on the value of Ara-C postremission therapy con-
`firms these results.
`In previously untreated elderly patients
`Conclusion:
`with AML, MTZ induction therapy produces a slightly better
`CR rate than does a DNR-containing regimen, but it has no
`significant effect on remission duration and survival. Ara-C
`in maintenance may prolong DFS, but it did not improve
`survival.
`J Clin Oncol 16:872-881. o 1998 by American Society of
`Clinical Oncology.
`
`IN RECENT YEARS, there has been an intensified
`
`interest in the development of treatment in the elderly
`with acute myeloid leukemia (AML).1,2 The outcome of
`these patients following remission-induction chemotherapy
`has remained disappointing. While complete remission (CR)
`rates in middle-aged adults have improved to values of 70%
`to 80%. response rates in patients of aged 2 60 years
`generally range between 40% and 50%. Overall survival at 2
`years following start of treatment is often less than 20%. The
`results of a palliative wait-and-see approach in patients older
`than 65 years of age are worse than those of remission-
`induction chemotherapy.3 Scarce studies have especially
`dealt with the question of treatment development in the
`elderly.3-9 An important question relates to the choice of drug
`or the dose applied in remission-induction treatment. Usu-
`
`872
`
`Journal of Clinical Oncology, Vol 16, No 3 (March), 1998: pp 872-881
`
`Copyright © 1998 by the American Society of Clinical Oncology. All rights reserved.
`
`
`
`
`
`MITOXANTRONE AND LOW-DOSE CYTARABINE IN ELDERLY AML
`
`873
`
`ally, a combination of an anthracycline (eg, daunomycin
`[DNR] or doxorubicin) and cytarabine (Ara-C) has been
`applied. Variation of the dose of DNR has not brought any
`significant benefit. 5 In a noncontrolled study, mitoxantrone
`(MTZ) plus Ara-C yielded promising CR rates of 58%.6 A
`phase III study in adults of all ages that compared DNR plus
`Ara-C versus MTZ plus Ara-C included 99 patients older
`than 60 years of age. 10 In this cohort of patients, the MTZ
`chemotherapy regimen was suggested to produce higher CR
`rates (46% v 37%), but the numbers were small and the
`results were based on a subgroup analysis. Therefore, the
`question has remained as to whether MTZ would produce
`greater response rates and also prolonged survival in elderly
`patients with AML.
`Another issue of the therapeutic management of elderly
`AML concerns the employment of postremission treatment.
`Should patients of higher age receive additional chemother-
`apy once a CR has been obtained and, if so, what would be
`the postremission therapy of choice? The application of
`repeated cycles of high-dose Ara-C postremission showed
`efficacy in reducing the recurrence of leukemia and improv-
`ing survival in a large study of adults, but did not prove
`successful in elderly patients." Conventional-dose (100
`mg/m2/d), intermediate-dose (400 mg/m 2), and high-dose (3
`g/m 2) schedules of Ara-C resulted in approximately equiva-
`lent outcome in 60+-year-old patients." High-dose Ara-C
`was associated with excessive toxicity in the elderly in the
`latter study. The use of low-dose (LD)-Ara-C (10 mg/m 2/d)
`has been favored for some time in the treatment of patients
`with myelodysplastic syndromes and shown to offer active
`therapy.12-14 It has also been advocated in
`antileukemic
`elderly patients with AML, of whom a significant proportion
`may have a hidden history of prior myelodysplasia (MDS).
`LD-Ara-C as post-remission therapy has not been critically
`evaluated yet.
`We report here the results of a phase III study in which
`MTZ induction therapy was compared with a schedule of
`
`DNR. Complete remitters were then eligible for a second
`randomization and received LD-Ara-C for 12 days for eight
`cycles as maintenance chemotherapy or no maintenance
`chemotherapy.
`
`MATERIALS AND METHODS
`
`Study Design and Chemotherapy
`
`In a collaborative phase III study of the European Organization for
`the Research and Treatment on Cancer-Leukemia Cooperative Group
`(EORTC-LCG) and the Dutch-Belgian Hemato-Oncology Cooperative
`- 60 years of age were enrolled and
`Group (HOVON), patients
`randomized to receive as induction therapy either DNR 30 mg/m2
`by intravenous bolus on days 1, 2, and 3 plus Ara-C 100 mg/m 2 by
`continuous infusion on days 1 through 7, or MTZ 8 mg/m 2 by
`intravenous bolus on days 1, 2, and 3 plus Ara-C 100 mg/m 2 by
`continuous infusion on days 1 through 7 (Fig 1). The choice of these
`dosages was based on a previous pilot study.3 In case of a partial
`response (PR) to the latter induction cycle, patients were planned to
`receive a second identical course of treatment. Complete responders
`were to receive one cycle of consolidation therapy that consisted of the
`same agents, but with 1 day of DNR or MTZ depending on the treatment
`arm. The objective of the study was to compare DNR and MTZ
`induction chemotherapy as regards the response rate, and in addition,
`the duration of survival, disease-free survival (DFS), postchemotherapy
`cytopenia, frequency of infectious complications, and number of days
`spent in the hospital during and following induction chemotherapy.
`Patients after consolidation and continuing in CR were eligible for a
`second randomization between no further therapy (arm A) and LD-
`(arm B) (Fig 1). Arm B patients received
`Ara-C chemotherapy
`LD-Ara-C at 10 mg/m2 subcutaneously every 12 hours on days 1
`through 12 at 42-day intervals for a total of eight cycles or until relapse.
`The effects of LD-Ara-C maintenance chemotherapy on DFS and
`overall survival were evaluated.
`
`Eligibility
`Patients > 61 years of age with AML were eligible if they had
`MO-M7 AML according to the French-American-British (FAB) classifi-
`cation. 15,16 Patients with secondary
`leukemias following MDS or
`following chemotherapy for solid tumors or lymphomas more than 1
`year before entry onto the study were also eligible. They were not
`eligible if they had been treated with chemotherapy for AML or MDS.
`They were also not eligible if they were refractory to platelet transfu-
`
`N
`
`cc
`
`entry
`
`Fig 1. Study scheme.
`
`DNR 30 mg/m2 day 1-3
`Ara-C 100 mg/m 2 continuous IV day 1-7
`
`A no further treatment
`
`E,,
`
`B LD= Ara-C
`Ara-C 10 mg/m2
`SC 12 hrs for 12 days
`8x 6 wk cycles
`
`MTZ 8 mg/m 2 day 1-3
`Ara-C 100 mg/m 2 continuous IV day 1-7
`
`Copyright © 1998 by the American Society of Clinical Oncology. All rights reserved.
`
`
`
`
`
`874
`
`sion; had severe hepatic (bilirubin level > three times normal value),
`pulmonary, or renal disease (serum creatinine concentration >
`two
`times normal value or creatinine clearance < 50 mL/min); or had
`symptomatic heart disease that required therapy.
`
`Criteria of Response and Evaluation of Outcome
`
`CR was defined by a normocellular bone marrow that contained -5%
`blast cells, including monocytoid cells, less than 10% blast cells and
`promyelocytes, and less than 50% erythroid cells, no evidence of
`extramedullary leukemia, and recovery of peripheral-blood values to
`platelet counts of at least 100 X 109/L and neutrophils of at least 1.5 X
`109/L. PR was defined by bone marrow smears that contained between
`5.1% and 25% blasts and less than 5% circulating blast cells. Failure to
`respond was classified as treatment resistance when there was no
`reduction of the leukemic cell infiltration in the marrow or a reduction
`that would not meet the criteria for a PR or CR. Hypoplasia followed by
`leukemic regrowth was also classified as treatment failure. Regenera-
`tion failure was defined as a prolonged hypoplasia of -8 weeks without
`evidence of medullary leukemia. Early death was defined as death
`before the completion of the induction cycle of therapy, and hypoplastic
`death as death during the 4- to 5-week recovery interval after the
`completion of chemotherapy. Survival duration and DFS were impor-
`tant parameters of evaluation (defined later). Frequencies of excessive
`toxicities, numbers of nights spent in hospital, frequencies of hemor-
`rhages and infections, number of days to hematopoietic recovery, and
`duration of fever were evaluated separately. Standard cytogenetic
`techniques, including direct preparations, incubation of cultures for 24
`or 48 hours, and banding techniques, were used at diagnosis
`to
`karyotype the leukemia." Normal (NN) cytogenetics (this category
`included the deletion of the Y chromosome), abnormal cytogenetics
`(AA), and a mosaicism of abnormal and normal karyotypes (AN) were
`recorded. Deletions of the long arm of chromosomes 5 and 7 (5q-, 7q-)
`or the entire chromosomes (-5, -7), and abn 11q23, as well as +8
`abnormalities, were regarded as poor-risk abnormalities, whereas
`inv16(pl3q2 2), t(16;16)(p13q22), t(15;17)(q22;q21), and t(8;21) (q2 2;
`q22) were considered as good-risk features. Karyotypic abnormalities
`that involved three or more chromosomes but without any of the
`aforementioned specific poor-risk or good-risk aberrations were classi-
`8 -20
`fied as complex anomalies.
`
`Statistical Analysis
`
`The relationship between the initial categorized ordered variables
`(WBC count, age, and performance status) and the CR rate after
`induction was statistically tested using the X2 test for linear trend.2 1 For
`unordered variables, the usual X2 test, with correction for continuity,
`was used. The relationship between treatment randomized and response
`(CR, resistance, or death during induction or during the hypoplastic
`phase) was tested using Fisher's exact test. The 95% confidence interval
`(CI) of the treatment difference was computed using the Confidence
`Interval Analysis (CIA) program. 22
`Overall survival was calculated from the date of randomization until
`the date of death, whatever the reason. DFS was calculated from the
`date of first CR achieved after the induction course(s), until the date of
`first relapse or date of death without confirmed relapse. For patients
`randomized for the second question (Ara-C v no Ara-C), the starting
`point for these analyses was the date of second randomization.
`Actuarial curves were computed according to the Kaplan-Meier
`technique. 2" The standard error was calculated according
`to the
`Greenwood formula.2 1 The log-rank test was used to perform the
`treatment comparison."2 The prognostic importance of different vari-
`ables was assessed using the log-rank test (for binary variables) or the
`
`LOWENBERG ET AL
`
`log-rank test for linear trend21 (for ordered variables). The relative risk
`(RR) of having an event per time unit in the MTZ treatment group
`versus DNR group, along with its 95% CI, was computed using the odds
`ratio technique. 22 The intention-to-treat principle was applied in the
`statistical analyses.
`The aim of the trial was to detect a difference in the CR rate from 40%
`to 55% (using the usual X2 test, at = 0.05, / = 0.10) between the two
`induction arms. The assumption was that such a difference in the CR
`rate, if it truly existed, would lead to a difference in the survival at 3
`years from 10% to 20%. Therefore, it was planned to enter 488 patients,
`to evaluate their remission status after the induction course, and to
`monitor them until relapse and death. The final analysis was planned to
`be performed once 425 deaths had been reported within 3 years from
`randomization (log-rank test, a = 0.05, P = 0.10).
`To detect a 15% difference (10% v 25%) in DFS rates at 3 years
`between the two maintenance groups (LD-Ara-C v no Ara-C), a total of
`208 patients was required to be randomized. The final analysis was
`planned once 171 events (relapses/deaths) had been reported (log-rank
`test, at = 0.05, / = 0.10). As an insufficient number of patients were
`randomized to address this question in the AML-9 trial (147 in total), 86
`additional patients were randomized
`in the subsequent EORTC-
`HOVON AML-I I trial in the elderly.2 3 Randomization was performed
`centrally at the EORTC Data Center, based on the minimization
`technique, with the stratification factors being patient age (60 to 70, 71
`to 80, or >80 years) and treating center. For the second randomization,
`LD-Ara-C versus no maintenance, first treatment allocated by random-
`ization and treating center were used as stratification factors.
`
`RESULTS
`
`A total of 539 patients were registered between April 1986
`and November 1993, of whom 270 individuals were random-
`ized to induction therapy with DNR and 269 to treatment
`with MTZ. Of these, four patients were considered to be
`nonassessable because of incomplete data (n = 3), and in
`one case, the dose of Ara-C administered was 10 times
`greater than the protocol dose. Forty-six subjects were
`ineligible, of whom 26 had been assigned to DNR and 20 to
`MTZ treatment. Reasons for ineligibility were incomplete
`data (n = 18), incorrect or inadequate diagnosis (n = 18),
`insufficient organ function (n = 8), and exclusions (n = 2)
`because of chemotherapy for AML or chemotherapy for
`breast cancer during the year before registration. The clinical
`and hematologic characteristics are listed in Table 1 for the
`242 patients randomized to DNR treatment and the 247 to
`MTZ treatment who could be evaluated, The median age of
`the study population was 68 years (range, 60 to 88), of
`whom only 5% were - 80 years. Less than 10% of patients
`enrolled had a performance status that kept them in bed for
`more than 50% of the time.
`
`Response to Remission-Induction Chemotherapy
`
`CR probabilities were 46.6% for patients on MTZ treat-
`ment and 38.0% for patients on DNR (P = .067) (Table 2).
`Interestingly, MTZ-treated patients showed a reduced prob-
`ability of primary resistance to chemotherapy (47% v 32%,
`
`Copyright © 1998 by the American Society of Clinical Oncology. All rights reserved.
`
`
`
`
`
`MITOXANTRONE AND LOW-DOSE CYTARABINE IN ELDERLY AML
`
`875
`
`Table 2. Response to Induction Chemotherapy
`
`DNR
`(n = 242)
`
`MTZ
`(n = 247)
`
`Response
`
`CR
`Resistance*
`Deatht
`
`No.
`
`92
`114
`36
`
`%
`
`No.
`
`%
`
`38.0
`47.1
`14.9
`
`115 46.6
`80 32.4
`52 21.1
`
`Between
`Groups (%)
`
`8.6
`-14.7
`6.2
`
`95% CI
`
`-0.2-17.3
`-23.3--6.1
`-0.6-13.0
`
`Pt
`
`.067
`.001
`.079
`
`following
`NOTE. 15 patients who did not receive any chemotherapy
`randomization because of early deteriorating condition (3 on DNR arm and 6
`on MTZ arm) or subsequent refusal (one on DNR and 5 on MTZ) are included in
`the analysis based on the intention-to-treat principle.
`*Includes absolute resistance, PR (8% and 7%), and transient hypoplasia
`followed by leukemic regrowth.
`tincludes early death during chemotherapy (6% on both arms) and
`postinduction death (9% and 15%).
`+Fisher's exact test.
`
`effect of age was associated with a reduced responsiveness
`of the leukemia to chemotherapy (ie, resistance). In contrast,
`poor performance status and (hyper)leukocytosis correlated
`with a greater death rate (Table 4). Secondary leukemia
`predicted for a greater probability of resistance to chemother-
`apy (Table 4). Sex of the patient or FAB subtype of AML
`
`Table 3. Analysis of Prognostic Factors for Response to
`Induction Chemotherapy
`
`Factor
`
`No. of Patients
`
`%CRs
`
`P
`
`Sex
`Male
`Female
`Age, years
`60-69
`70-79
`80-88
`WHO performance status
`Normal
`Ambulatory
`In bed < 50% of time
`In bed > 50% of time
`Entirely disabled
`WBC count (x 109/L)
`< 25
`25-99
`- 100
`Antecedent history
`No
`Secondary leukemia or prior MDS
`Cytology-FAB type
`MO/MI
`M2
`M3
`M4
`M5
`M6/M7
`Cytogenetics
`NN
`AN-NN
`P values according to the *x2 test o
`
`274
`215
`
`300
`167
`22
`
`94
`247
`107
`36
`5
`
`299
`120
`70
`
`365
`124
`
`101
`181
`19
`74
`100
`11
`
`73
`137
`
`.31*
`
`.074t
`
`.014t
`
`.006t
`
`.21"
`
`.4*
`
`.057'
`
`45
`40
`
`44
`43
`14
`
`46
`46
`36
`28
`20
`
`48
`32
`36
`
`44
`37
`
`45
`45
`32
`38
`40
`55
`
`53
`39
`
`r t2
`
`test for linear trend.
`
`Table 1. Characteristics of Patients by Treatment Arm
`% of Patients
`
`DNR
`(n = 242)
`53
`
`MTZ
`(n = 247)
`
`61
`34
`5
`
`20
`52
`22
`
`5 1
`
`59
`27
`14
`
`0.4
`18
`38
`5
`14
`21
`
`2 1
`
`19
`7
`
`31
`
`4 1 3
`
`12
`
`9 2 84
`
`26
`
`Sex, male
`Age, years
`61-69
`70-79
`80-88
`WHO performance status
`Normal
`Ambulatory
`In bed < 50% of time
`In bed > 50% of time
`Entirely disabled
`WBC count (x I0 9/L)
`< 25
`25-99
`- 100
`FAB cytology
`MO
`M1
`M2
`M3
`M4
`M5
`M6
`M7
`Antecedent history (secondary AML)
`Prior MDS
`Prior hematologic disease or chemotherapy
`Cytogenetics*
`Normal
`t(8;21)
`t(15;17)
`inv/dic(16)
`-5, 5q-
`-7, 7 q-
`abn 11 q2 3
`+8
`Complex
`Other abnormalities
`
`Abbreviation: WHO, World Health Organization.
`*Adequate cytogenetic examination was performed in 47.5% of patients
`(n = 115) assigned to the DNR arm and 38.5% (n = 95) of the MTZ arm.
`Percentage distributions of specific cytogenetic karyotypes are expressed
`relative to the subgroup of patients (set at 100%) in whom adequate and
`assessable chromosome analysis was performed.
`
`P = .001) in the context of a slightly greater death rate (21%
`v 15%) (Table 2).
`
`Prognostic Factors for Response to
`Remission-Induction Chemotherapy
`
`Table 3 lists the factors at diagnosis that showed prognos-
`tic value for CR. Age
`- 80 years, a progressively worse
`performance status, and high WBC count (>25 X 10 9/L) all
`predicted for a reduced CR probability. The unfavorable
`
`Copyright © 1998 by the American Society of Clinical Oncology. All rights reserved.
`
`
`
`
`
`876
`
`Table 4. Prognostic Factors for Drug Resistance or Death Following
`Induction Chemotherapy
`
`Treatment Resistance
`
`Death
`
`Factor
`
`%
`
`P
`
`%
`
`P
`
`39
`38
`68
`
`49
`40
`36
`25
`40
`
`37
`48
`34
`
`.02*
`.14t
`
`.12*
`.01t
`
`.07*
`.72t
`
`17
`19
`18
`
`5
`14
`28
`47
`40
`
`14
`20
`30
`
`.89*
`.68t
`
`< .00001*
`< .000011t
`
`.007*
`.002t
`
`LCWENBERG ET AL
`
`randomized to the MTZ arm (median, 22 days) than for
`those on the DNR arm (median, 19 days). Patients treated on
`either arm had fever for a median of 6 days (P = .10). The
`median number of days spent in the hospital was 31 days in
`both the MTZ and DNR treatment groups (P = .71).
`
`Survival and DFS
`
`The median follow-up duration at the time of statistical
`analysis was 6 years. The duration of survival was similar (P =
`.23) in the two treatment groups (Fig 2), the RR of the death rate
`per time unit of MTZ group versus DNR group was 0.893 (95%
`CI, 0.742 to 1.076). Median survival estimates were 36 weeks
`(DNR) and 39 weeks (MTZ). The percentages of patients still
`alive at 5 years were 6% and 9%, respectively. For patients who
`achieved CR after induction, the DFS probabilities between the
`treatment arms were not different (Fig 3). The median DFS
`estimates were 39 weeks in both groups. The DFS rate at 5 years
`was 8%. Atotal of 174 patients (77 v 97) relapsed and 15 patients
`died without relapse (seven v eight). The causes of death of the
`latter patients in continuous CR were infection (n = 4), cardiac
`arrest and myocardial infarction (n = 2), hemorrhage (n = 3),
`and other or unknown cause (n = 6). The duration of survival
`from CR was slightly longer (P = .29) in the MTZ arm. Median
`survival estimates for complete responders were 74 weeks
`(MTZ) versus 55 weeks (DNR), survival estimates at 5 years
`were 12% versus 16%, and the RR was 0.85 (95% CI, 0.633 to
`1.149).
`The following initial factors appeared to relate to a shorter
`duration of survival from start of treatment (Fig 4): older age
`(P = .01), poor performance status (P < .001), high WBC
`.001), secondary AML (P = .02) (data not
`count (P <
`shown), and the presence of cytogenetic abnormalities (P =
`.002). Interestingly, not only patients with a complete or
`partial deletion of chromosomes 5 and 7 and complex
`chromosome abnormalities had a poor prognosis, but also 12
`
`s
`
`(yers)
`
`2
`
`3
`
`Number of patients at risk:
`
`242
`247
`
`77
`101
`
`41
`48
`
`21
`32
`
`16
`22
`
`10
`16
`
`5
`8
`
`3
`5
`
`2
`2
`
`DNR
`MTZ
`
`Fig 2. Duration of survival according to induction Ireatment: DNR versus MTZ.
`
`Age, years
`< 70
`70-79
`> 80
`WHO performance status
`Normal
`Ambulatory
`In bed < 50% of time
`In bed > 50% of time
`Entirely disabled
`WBC count (x 109/L)
`< 25
`25-99
`> 100
`Secondary AML
`19
`37
`No
`.45*
`15
`.03*
`48
`Yes
`NOTE. P values according to the *x2 test or tx2 test for linear trend.
`
`showed no predictive value for induction response, nor did
`cytogenetics. However, of 489 patients, adequate cytoge-
`netic examination was performed in only 210 (43%). Among
`these, only 12 patients showed the favorable karyotypes
`t(8;21) or abnl6(q22) or t(15;17), 42 had deletions of
`chromosomes 5 or 7 (-5,5q-,-7,7q-), 25 presented with
`1 1q23 or + 8 abnormalities, and seven other patients showed
`complex chromosomal abnormalities. Because of the limited
`numbers of the specific cytogenetic subgroups, most of the
`subsequent analyses are based on the comparison of normal
`(NN) versus abnormal (AN and AA) cytogenetics.
`
`Toxicities Associated With MTZ Versus DNR
`Remission-Induction Therapy
`
`Since the great majority of patients (n = 489) received
`induction cycle no. 1 and only 73 patients received cycle no.
`2, the toxicities of DNR and MTZ chemotherapy were
`directly compared after the first chemotherapy cycle. There
`were no significant differences between the two treatment
`groups as regards the frequencies of mild, gross, or debilitat-
`ing hemorrhages (mean, 6% of cases); serious infections
`(mean, 22%); liver function abnormalities (bilirubin level >
`2.5 times normal; mean, 13%); renal toxicity (serum creati-
`nine concentration > 2.5 times normal; mean, 4%); vomit-
`ing and nausea, and severe intractable diarrhea (mean, 2%);
`or severe oral toxicity that required liquid food intake or
`parenteral nutrition (mean, 2%). The incidence of severe
`infections among patients randomized to MTZ treatment
`exceeded that in patients on DNR therapy: 25.1% versus
`18.6% (P = .036). The duration of aplasia for patients who
`achieved a CR was slightly longer (P = .06) for patients
`
`Copyright © 1998 by the American Society of Clinical Oncology. All rights reserved.
`
`
`
`
`
`MITOXANTRONE AND LOW-DOSE CYTARABINE IN ELDERLY AML
`
`877
`
`% l00
`
`,90
`
`70
`
`Y 60
`
`50
`
`40
`
`30
`
`20
`
`10
`
`3
`
`1
`
`2
`
`3
`
`Number of patients at rk:
`
`92
`115
`
`30
`47
`
`13
`22
`
`10
`13
`
`7
`11
`
`5
`6
`
`3
`3
`
`3
`L
`
`2
`0
`
`DNR
`MTZ
`
`Fig 3. DFS from CR according to treatment: DNR versus MTZ.
`
`patients who presented with good-risk cytogenetic features.
`The Cox proportional hazards model showed that initial
`performance status (P < .0001) and WBC count (P = .003)
`were the two most important independent prognostic factors.
`Secondary AML and abnormal cytogenetics predicted for
`shorter duration of remission. The ability to attain CR after
`
`one versus two cycles of induction, or initial leukocytosis,
`had no predictive value for DFS.
`
`LD-Ara-C Versus No Maintenance Chemotherapy
`
`Of the complete responders, 76 patients were randomized to
`no further therapy (no Ara-C), and 75 patients to eight cycles of
`LD-Ara-C. Four patients were ineligible because of no CR at the
`time of second randomization, so that 73 (no Ara-C) and 74
`(LD-Ara-C) patients could be evaluated. Comparative DFS and
`overall survival are plotted in Fig 5, which shows an advantage
`of DFS for patients assigned to LD-Ara-C maintenance therapy.
`The RR estimate was 0.61 (95% CI, 0.43 to 0.87). The median
`DFS duration following the second randomization was 29 weeks
`for no-Ara-C and 51 weeks for LD-Ara-C treatment. At 3 years,
`DFS was estimated at 7% (SE = 3.0%) for no maintenance and
`20% (SE = 4.7%) for LD-Ara-C treatment; at 5 years, DFS was
`estimated at 7% (SE = 3.0%) and 13% (SE = 4.0%), respec-
`tively (P = .006). Four patients died without relapse: zero (no
`Ara-C) versus four (LD-Ara-C). The median overall survival
`estimates were 62 weeks (LD-Ara-C) versus 79 weeks (no
`Ara-C). The RR was estimated at 0.83 (95% CI, 0.58 to 1.18) for
`
`A
`
`o
`
`70
`
`t o
`
`30
`
`20
`
`10
`
`m
`
`. M
`
`so
`
`Y 6
`
`50
`
`30
`
`20
`
`10
`
`"(tread)
`
`(y-)
`
`2
`
`3
`
`2
`
`3
`
`4
`
`LATORY
`
`----
`
`I (trend)
`
`(ye)t
`
`Number of parents a ril :
`
`300 120
`54
`167
`4
`22
`
`63
`24
`2
`
`36 6
`16
`I
`
`12
`0
`
`18
`8
`0
`
`9
`4
`0
`
`6
`2
`0
`
`3
`t
`0
`
`<70
`<80
`>0=0
`
`Namebr of p me a*.k :
`
`41
`94
`247 100
`107
`28
`9
`41
`
`22
`48
`15
`4
`
`14
`25
`10
`4
`
`12
`15
`8
`3
`
`9
`9
`6
`2
`
`5
`5
`2
`1
`
`4
`2
`2
`0
`
`2
`1
`1
`0
`
`NORMAL
`AMBULATORY
`BED 505
`BED >50%
`
`7 7
`
`D
`
`I (trend)
`
`.
`
`.
`
`Numlxr ol paelms at risk
`
`299
`120
`70
`
`131
`31
`16
`
`68
`12
`9
`
`38
`8
`7
`
`28
`7
`3
`
`20
`4
`2
`
`10
`3
`0
`
`6
`2
`0
`
`2
`2
`0
`
`<2
`<00
`>-00
`
`i
`
`2
`
`Number of apa at riL
`
`:
`
`02
`
`9
`
`(ye )
`
`73
`137
`
`35
`33
`
`21 1
`17
`
`3
`10
`
`9
`8
`
`8
`5
`
`6
`2
`
`4
`
`2
`t
`
`NN
`AN-AA
`
`Fig 4. Duration of survival according to age (A), performance status (B), WBC count (C), and cytogenetic abnormalities (D).
`
`Copyright © 1998 by the American Society of Clinical Oncology. All rights reserved.
`
`
`
`
`
`878
`
`A
`
`b
`
`B
`
`%
`
`,
`S,
`
`S70
`
`50
`
`40
`
`30
`
`4
`
`----
`
`Numeroofpatiens t rik
`
`73
`74
`
`20
`37
`
`8
`20
`
`5
`14
`
`4
`14
`
`2
`6
`
`2
`2
`
`2
`2
`
`1
`I
`
`NoAraC
`LD-Ar-C
`
`C
`i-C
`
`-
`
`.
`
`'
`
`Numbr of palienls at risk
`
`73
`74
`
`38
`53
`
`24
`28
`
`t
`
`16
`23
`
`10
`16
`
`0
`
`4
`2
`
`2
`
`(cid:127)1
`I
`
`N A
`LD-Ara-C
`
`Fig 5. Outcome according to treatment: no maintenance treatment (no
`Ara-C) versus LD-Ara-C. (A) DFS and (B) overall survival.
`
`LD-Ara-C therapy. Overall survival probabilities at 5 years were
`18% (SE = 4.6%) in with LD-Ara-C and 15% (SE = 4.3%) for
`no-Ara-C maintenance treatment (P = .29).
`
`DISCUSSION
`
`We present the results in 489 assessable patients of a
`prospective randomized study that specifically deals with
`treatment development in the elderly with AML. The results
`of this study indicate that specific changes in the choice of
`induction chemotherapy may affect the effect of treatment in
`aged patients. They also show that LD-Ara-C maintenance
`may modify DFS. The dependence of elderly patients on the
`dose and the choice of the type of chemotherapeutic agents
`in remission-induction therapy may differ from those in
`- 60
`younger adults with AML. In one study of patients
`years of age, DNR at a 30-mg/m2 dose (combined with
`cytarabine) produced higher response rates than a regimen
`with DNR 45 mg/m 2 or doxorubicin 30 mg/m 2. 24 In contrast,
`in patients younger than 60 years, the 45-mg/m 2 DNR dose
`gave the best response results. In another study, DNR 50
`
`LCWENBERG ET AL
`
`mg/m2 on day 1 only and DNR 60 mg/m 2 on days 1 to 3
`produced differences, but the numbers of patients enrolled in
`the latter study were small.5 MTZ has been evaluated in a
`single-institution study, but induction with MTZ did not
`result in a greater likelihood of response. 6 A second study in
`adults with AML that included a considerable number of
`elderly patients had suggested better response probabilities
`in case of MTZ induction therapy. 10 The results of the study
`reported here indicate a slightly improved CR rate following
`MTZ-based chemotherapy in comparison to DNR treatment,
`probably due to a better antileukemic effect. The trial was
`based on a comparatively large number of patients, the
`patient characteristics in the treatment arms were extremely
`well balanced, and the analysis showed an improvement of
`response of approximately 10% (from 38% to 47%). Optimi-
`zation of induction therapy specific for elderly patients
`appears of considerable clinical relevance, since more than
`50% of all newly diagnosed patients with AML are over 60
`years of age. From the results of this study, it is also clear
`that certain subgroups of patients are notoriously difficult to
`induce into remission. Poor performance status at diagnosis,
`high WBC count, and high age were all predictors of poor
`response. A poor performance condition and high cell counts
`mainly predicted for toxic deaths and high age and second-
`ary leukemias mainly for resistance to therapy. In a prospec-
`tive study of 104 patients, 6 secondary leukemia and high
`blast counts were shown to be unfavorable predictors for
`response, and in a retrospective study,25 high age and poor
`performance status predicted remission failure. In our study,
`patients > 80 years of age had a probability of attaining a
`CR of 520%, and patients with a World Health Organization
`(WHO) performance score - 3 had a CR rate of less than
`30%. On the other hand, patients with hyperleukocytosis and
`poor performance status showed greater frequencies of toxic
`deaths and appeared more prone to early disease-related or
`treatment-induced complications. Patients with secondary
`leukemia presented more frequently with treatment refrac-
`tory disease (Table 4). This was evident both from the failure
`analysis of induction treatment and also the inferior DFS
`results. Although the MTZ chemotherapy schedule provided
`for better response rates, overall survival and DFS probabili-
`ties did not improve. Why this effect did not translate into a
`survival advantage remains unclear. One possible explana-
`tion is that the quality of postremission treatment, which is
`critical for maintaining the remission attained after induction
`therapy, was inadequate.
`A second question addressed in the study was concerned
`with evaluating eight cycles of LD-Ara-C chemotherapy as
`postremission therapy. At the time of initiation of the study,
`
`Copyright © 1998 by the American Society of Clinical Oncology. All rights reserved.
`
`
`
`
`
`MITOXANTRONE AND LOW-DOSE CYTARABINE IN ELDERLY AML
`
`879
`
`LD-Ara-C therapy was considered an attractive treatment
`strategy in the elderly, because it had shown activity in
`patients with MDS.12- 14 In fact, there may be a considerable
`proportion of cases with an occult history of MDS among
`elderly patients with AML. Further, it is of note that
`high-dose Ara-C schedules, while showing improved activ-
`ity in middle