Post-remission therapy of adult acute myeloid leukaemia: One cycle of high-dose versus standard-dose cytarabine

Annals of Oncology 8: 251-257, 1997. © 1997 Kluwer Academic Publishers. Printed in the Netherlands.

Original article Post-remission therapy of adult acute myeloid leukaemia: One cycle of highdose versus standard-dose cytarabine M. Fopp, M. F. Fey, M. Bacchi, F. Cavalli, J. Gmuer, E. Jacky, L. Schmid, A. Tichelli, A. Tobler, L. Tschopp, V. Von Fliedner & A. Gratwohl for the Leukaemia Project Group of the Swiss Group for Clinical Cancer Research (SAKK) Swiss Group for Clinical Cancer Research, SIAK/SAKK Coordinating Centre, Berne, Switzerland * See page 256 for list of collaborators and participating centres

Summary Background: Intensification of post-remission therapy improves the cure rate of acute myeloid leukemia (AML) but is often accompanied by unacceptable toxicity. From 1985 to 1992 the Swiss Group for Clinical Cancer Research (SAKK) performed a randomized phase III trial to evaluate the effectiveness of one single postremission course of high-dose cytarabine (HDAC) in terms of leukaemia-free and overall survival in adults with de novo AML. Patients and methods: Adult (15-65 years) AML patients in remission after two induction courses were randomly assigned to one consolidation course either with standard (SDAC: 100 mg/sqm 24 hours infusion over seven days) or with high-dose cytarabine (HDAC: 3000 mg/sqm every 12 hours as one-hourinfusion for six days). In addition, both arms included daunorubicin (45 mg/sqm daily on days 1 to 3). Thereafter, patients were observed without maintenance until relapse. Results: After two induction courses 208/276 eligible patients achieved remission (CR: 169, 61%, PR: 39,14%), 41 were resistant (15%) and 20 died early (7%). Seventy-one patients in remission were not randomized. One hundred thirty-seven were randomized in CR/PR (67 SDAC, 70 HDAC). 4/70 patients randomized to HDAC did not receive it. Treatmentrelated mortality in HDAC was 1.4% (1/66). WHO grade 3-4

Introduction

Although complete remission rates in adults with acute myeloid leukaemia (AML) are high [1] disease free survival (DFS) without postremission therapy rarely exceeds one year [2, 3] and most patients succumb to their disease within a few months from initial diagnosis. It has consistently been shown that consolidation therapy does improve survival [4-7] whereas maintenance chemotherapy after early consolidation does not [8]. Since AML is chemo-sensitive [9] it should be possible to improve the cure rate by intensification of early postremission treatment. A particularly attractive candidate drug for such endeavours would be cytosine arabinoside [10]. High-dose cytosine arabinoside (HDAC) has been shown to induce remissions even in refractory AML [11]

toxicities occurred in 14/67 SDAC and in 38/66 HDAC patients (P < 0.0001). The median event free survival was 10.8 (SDAC) vs. 12.2 months (HDAC; P = 0.18). The median overall survival was 24.6 (SDAC) vs. 32.6 months (HDAC; P = 0.07). Although statistically uncertain, HDAC reduced the hazard of progression (hazard ratio: 0.69, P = 0.08) and of death (hazard ratio: 0.70, P = 0.13). For 112 patients stratified as CR the estimated four-year disease-free survival was 25% ( ± 6 % ) with SDAC and 37% ( ± 6 % ) with HDAC (P = 0.09). The overall survival rates at four years were 38% (+7%) and 48% (+7%), respectively (P = 0.10). In multivariate analysis HDAC significantly reduced the hazard of relapse by 39% compared to SDAC (hazard ratio = 0.61, 95% CI: 0.37-0.99; P = 0.049). Conclusions: We conclude that early consolidation of adult AML in CR with a single course of HDAC is superior in terms of outcome to one cycle of SDAC. The results of our intensive, single course HDAC group compare favourably with less intensive, repetitive HDAC cycles, suggesting that Ara-C dose intensity may be more important than total dosage. In addition, our treatment strategy is much less toxic and less expensive. Key words: cytarabine Ara-C high dose, economic costs, leukaemia myelocytic acute adult, postremission consolidation therapy, randomized trial, toxicity

and was effective as consolidation therapy in nonrandomized studies [12,13]. Recent data from a CALGB study comparing three dose levels of cytarabine consolidation support the concept of a dose-response effect for this drug in patients of 60 years or younger [14]. These results were obtained with an extensive and expensive treatment programme including four consolidation cycles followed by four cycles of maintenance. In May 1985, we started a phase III study in adult AML comparing a single consolidation cycle of SDAC with HDAC (with daunorubicin in both treatment arms) after two courses of non-cross resistant induction chemotherapy [15, 16]. We postulated that intensive short-term consolidation with HDAC would yield considerably better results than SDAC without imposing the toxicity and the costs of prolonged treatment strategies.

252 Patients and methods Induction

Patients Course 1 DNM x 3d Ara-C x 7d VCRdlO

Previously untreated patients aged 15-65 years with de novo acute myeloid leukaemia (AML FAB Ml-6) classified according to the French-American-British (FAB) criteria [17] were registered from May 1985 to June 1992 from the eight main centres responsible for the care of virtually all cases of acute leukaemia of this age group in Switzerland. Patients with severe non-neoplastic conditions, and patients whose follow-up was not guaranteed were excluded. All patients gave their verbal informed consent to participate in this trial.

Course 2 m-AMSA x5d VP16-213x5d

r

Response evaluation

Treatment schedule Patients received two courses of induction treatment (study design: Figure 1). The first course consisted of daunomycin (DNM) at 45 mg/ sqm given daily on days 1 to 3, Ara-C at 100 mg/sqm as a 24-hour continuous infusion for seven days, beginning on day 1 and vincristin 0.8 mg/sqm i.v. on day 10. The second course was started after haematological recovery from the first course or at day 17 of the first course if a marrow aspirate taken at that time showed persistent leukaemia (marrow rating M2 or worse according to CALGB criteria [18]). The second course consisted of m-Amsacrine at 120 mg/sqm given i.v. daily for five days and VP16-213 at 80 mg/sqm day administered as a 24-hour infusion for five days. Patients not in CR/PR after two cycles went off study. Patients in remission (CR/PR) after two courses of induction who had an HLA-identical donor were scheduled to receive an allogenic bone marrow transplant (BMT). The other patients in CR/PR were stratified according to response to induction therapy (CR vs. PR), FAB classification (Ml-3 and M6 vs. M4-5) and age ( < 4 0 vs. > 4 0 years), and were centrally randomized to receive one course of SDAC (arm A; 100 mg/sqm given as a 24-hour continuous infusion for seven days) or HDAC (arm B; at 3 g/sqm given i.v. as a one-hour infusion every 12 hours for six days). In both arms DNM was given at 45 mg/sqm i.v. bolus on days 1, 2, and 3. No maintenance treatment was given. Patients who relapsed after study treatment were treated at the discretion of the responsible physician.

Stratification CR/PR FAB-TYPE Age

Randomization for Consolidation (Course 3)

DNM x 3d Ara-C x 7d

DNM x 3d HD-Ara-C x 6d

Figure 1. Design of the study. Patient numbers are given for each treatment step. Patients were stratified according to remission status after induction (CR vs. PR), FAB-type (FAB M1-3 and M6 vs. M4,5) and age ( < 4 0 vs. ^40 years). BMT denotes bone marrow transplantation.

Assessment of response and toxicity The response to treatment was assessed according to CALGB criteria [18]. Patients suffering early death (within 40 days) and cases inevaluable for response (lost to follow up; no marrow data) were classified as non-responders. Patients receiving bone marrow transplantation were evaluated for response and overall survival of the entire study population. Toxicity was classified using WHO criteria [19]. All toxicities and complications were centrally reviewed by the study chairman (MF). Event free survival (EFS) was calculated from the day of diagnosis and DSF from the day of clinical complete remission (CR) until the date of progression, relapse or death whichever occured first. OS was calculated from the day of diagnosis until the date of death due to any cause.

Study endpoints and statistical methods Study endpoints were CR rate, EFS, DFS, OS and toxicity. All randomized patients were analyzed on a 'intended to treat' basis. The study was planned to assess an absolute 30% increment of event-free survivors at three years (15% versus 45% EFS). With an alpha error of 5% and a power of 80% a minimum of 50 eligible patients per arm was required. Confidence intervals were calculated as described [20]. Ratios were compared by chi-square tests without continuity correction and Fisher's exact test where appropriate. Predictive factors for CR were analyzed with contingency tables and multivariate logistic regression [21]. EFS, DFS and OS probabilities were calculated by the Kaplan-Meier method [22]. Standard errors were estimated using Greenwood's formula [23]. The equality of EFS, DFS and OS distribu-

tions was tested as suggested by Gray and Tsiatis [24]. This linear rank test is specially powerful when the main interest is in differences of cure rates. Multivariate analyses for control of imbalances in prognostic factors were performed with Cox model for duration of EFS, DFS and OS [25] Hazard ratios were based on estimates from the Cox model. Overall results were analyzed for eligible registered patients, for all randomized patients and for patients stratified as CR at randomization and with confirmed CR. All P-values refer to two-sided tests.

Results Two hundred eighty-eight patients with newly diagnosed AML were registered from May 1985 to June 1992 from eight centres regularly treating adult AML patients. Twelve of the 288 patients (4%) were not eligible and excluded from analysis (FAB-classification impossible or other than M l - 6 : 5; age > 65 years: 6; and concomitant severe chronic lung disease: 1). One hundred thirtyseven patients in remission were randomized (67 in arm A and 70 in arm B); 119 stratified as CR (59 in arm A and 60 in arm B) and 18 as PR (8 in arm A and 10 in arm B), respectively. The baseline characteristics of eligible patients are

253 listed in Table 1. By the time of this evaluation the median follow-up from randomization was six years.

Table 2. Multivariate logistic regression model predicting CR (n = 276). Prognostic factor

Induction therapy Of 276 eligible patients, 208 responded to induction (CR + PR: 75%; 95% confidence limits: 70% to 80%) and 169 achieved CR (61%; 95% confidence limits: 55% to 67%). Forty-one patients had stable or progressive disease (15%). Twenty (7%) treatment-related early deaths occurred. Seven eligible patients (3%) were not evaluable for response because of lack of documents or early stopping of treatment and were considered as nonresponders. In a univariate analysis factors predictive for reaching CR after induction were age < 40 years (P - 0.006), female sex (P = 0.06), good initial performance status (P - 0.06) and presence of Auer rods (P = 0.01). When adjusting for the clinically relevant covariates in a logistic regression model, the same factors had a significant impact on the probability to achieve CR (Table 2). The odds ratios in this model indicate that, for a given set of covariate values, age <40 was associated with an increased (more than twice) chance of CR with respect to age ^40. Table 1 Baseline characteristics of 276 eligible patients with AML. Registered

Randomized

and not randomized n - 139

Arm A SDAC n = 67

Arm B HDAC n = 70

41 26 44(16-61)

33 37 47(17-64)

44 22 1

38 26 6

31 35 1

29 39 2

8 22 6 20 10 1

5 26 7 15 12 5

35 32 0

38 31 1

10

12

43 8.3

55 9

Sex Male 79 Female 60 Median age (years) 45(16-64) Performance status 0-1 69 2-3 61 9 4 or unknown Extramedullary disease Present 59 76 Absent 4 Unknown FAB classification 22 Ml M2 48 M3 11 M4 31 M5 21 M6 6 Auerrods Present 72 Absent 66 Unknown 1 Baseline haematological characteristics Median WBC(x 109/l) 21 Median platelet counts (x 1O9/1) 44 Median hemoglobin (g/dl) 8.3 Median peripheral blood leukemic cells (x 109/l) 12.8

3.6

4.3

Sex Male Female Age < 40 years > 40 years Performance status* 0-1 2*2° Auerrods Yes No

Odds ratio

95% confidence

/•-value

0 56 reference

0.33-0.94

0.03

2.27 reference

1.28-4 04

0.005

1.78 reference

1.06-2 98

0.03

2.10 reference

1.22-3.60

0.007

° Includes patients with unknown performance.

Toxicity and side effects are displayed in Table 3. Infections, haemorrhage, organ failure (mainly cardiac and renal) and transfusion-related complications (refractoriness to platelet transfusions and allergic/anaphylactic transfusion reactions) were the most frequent events during induction. Lethal infections with sepsis (mainly from Staphylococcus aureus and Escherichia coli) occurred during both induction cycles (2.9% in cycle one and 2.4% in cycle two) while lethal central nervous system hemorrhages occurred only during first cycle (2.5%). Three patients died of cardiac complications and one of renal failure. Consolidation therapy Of the 208 patients in remission eligible for BMT or consolidation, 33 (16%) received allogenic BMT, whilst 137 (67%) were randomized (67 in arm A and 70 in arm B) all of them with well balanced baseline prognostic factors (Table 1). Thirty-eight patients (18%) refused randomization or their physicians opted for elective consolidation treatment mainly because of toxicity during induction. Six randomized patients (two in A and four in B) were found not to satisfy randomization requirements (age and response), and were nevertheless analysed according to the assigned treatment arm. The median interval between beginning of induction (day one of treatment) and of consolidation (after reaching the protocol specified criteria for toxicity recovery in order to sustain HDAC) was 79 days (range: 54-161) in A and 76 days (range: 48-177) in B. Eleven patients in PR at randomization reached CR after consolidation (six in A and five in B). One patient randomized in arm B as PR achieved a CR in the interval before starting the consolidation. Four patients randomized to receive arm B (6%) did not receive the consolidation treatment due to toxicity during induction. Seven patients (one in arm A and six in arm B) received the assigned treatment with modifications such as dose reduction or suspension of some drugs. One patient in each arm received the other regimen.

254 Table 3. Summary of WHO grade > 3 toxicities and complications." Cycle 1 n = 216

Cycle 3

Cycle 2 n = 245

SDAC n = 67 Infection Grade 3 Grade 4 Grade 5 b Hemorrage Grade 3 Grade 4 Grade 5 Gastrointestinal Grade 3 Grade 4 Neurologic Grade 3 Grade 4 Cardiac Grade 3 Grade 4

HDAC n = 66

I s " V. 48 15 8

60 )4 6

11 0 0

31 6 1

13 6 7

1 2 &

Q 0: 0

1 0 0

10 2

17 7

1 0

11 0

4 5

3 1

0

4 4

3 1

2 2

ff 0

2 0"

I

ft

ft

2

0 0 0

8 0
Grade 5 1 Renal Grade 3 I Grade 4 1 0 Grade 5 1 Skin Grade 4 3 Allergy Grade 4 1 2 Transfusion related complications Grade 3-4 7

0

0 1

13

1

30

0

1

I

* Some patients developed more than one type of grade 3-4 and 5 toxicity or complication. b Lethal toxicity/comphcation.

The toxicity of the consolidation therapy is summarized in Table 3. WHO grade 3 or 4 side effects occurred in 14 of 67 patients treated with SDAC and in 38 of 66 patients with HDAC (P < 0.0001). One patient died due to septicemia in the HDAC arm. Toxicities under highdose Ara-C were significantly more severe including notably: infections (P < 0.0001), gastrointestinal (P < 0.002) and neurological side effects (P < 0.0001) including ataxia, syncope and desorientation. Vomiting, diarrhoea and mucositis were observed mainly in arm B. No major haemorrhages were observed in either arm with the exception of one patient with temporary blindness due to retinal bleeding. Event free and overall survival (registered and randomized patients) The median EFS of 276 registered patients (213 events) was 9.7 months and median OS (191 deaths) was 17.2 months with 42% + 3% of the patients alive at two years and 36% + 3% at three years. One hundred two of 137 randomized patients had progressed or died by the time of the current evaluation.The median EFS of randomized patients was 10.8 with SDAC versus 12.2 months with HDAC (Figure 2,

*

* //

/

/

/

IH

I

H

II

II

I

Conventions^ Ara-C High Dose Ara-C

Event Pree Time (yea"

Figure 2. Kaplan-Meier plots of Event-Free Survival (EFS) by treatment arm (all randomized patients). Conventional Ara-C, n = 67; high dose Ara-C, n - 70.

P = 0.18). Eighty-three randomized patients died by the time of the current evaluation. The median overall survival was 24.6 months in the SDAC arm and 32.6 months in the HDAC arm (P = 0.07) with 56% + 6% of the patients alive at two years and 49% + 6% at three years. Table 4a summarizes actuarial data of all randomized patients at three tofiveyears of follow up. Multivariate analysis of factors predictive of EFS and OS of all randomized patients showed that, after allowance for covariates (age, sex, performance status, extramedullary disease, auerrods, leukocyte count and FAB), the achievement of CR had the greatest impact on both outcomes. Achievement of CR corresponded to a hazard ratio of 0.43 for EFS (P = 0.001) and 0.37 for OS {P = 0.0005) compared with no achievement of CR. Although statistically uncertain, HDAC reduced the hazard of progression (hazard ratio = 0.69, P = 0.08) and of death (hazard ratio = 0.70, P - 0.13) in this group of patients. Disease free and overall survival (patients stratified as CR) Seven out of the 119 patients stratified as CR (four A and three B) never achieved a true CR (as assessed at review) and were excluded from these subgroup evaluations. The median DFS of CR patients was 8.5 with SDAC versus 15.2 months with HDAC (Figure 3, P = 0.09). The DFS curve of the HDAC arm shows a high risk of relapse for the first two years after diagnosis and subsequently a plateau after four to five years whereas the SDAC arm does not exhibit a plateau with late events due to relapsing leukaemia still occurring after four years and later. The median overall survival was 27.5 months in the SDAC arm and 43.3 months in the HDAC arm (P = 0.10). Table 4b summarizes actuarial data of all CR patients at three tofiveyears of follow up. Multivariate analysis of factors predictive of DFS and OS of CR patients showed that, after allowance for covariates (age, sex, performance status, extramedullary

255 Table 4a Summary of actuarial data for EFS and OS. All randomized patients: CR + PR. Time

Three-year % (SE) Four-year Five-year

EFS

OS

X

SDAC (67)'

HDAC (70)

SDAC (67)

HDAC (70)

22% (5) 22% (5) 22% (5)

33% (6) 31% (6) 31% (6)

40% (6) 34% (6) 29% (6)

49% (6) 45% (6) 41% (6)

Linear rank test

0.12

0 .07

K V C

o

Q

Conventional Ara-C High Dose Ara-C

* Patient numbers.

Table 4b. Summary of actuarial data for DFS and OS. Patients randomized in CR. Time

Three-year % (SE) Four-year Five-year Linear rank test

DFS

OS

SDAC (55)"

HDAC (57)

SDAC (55)

HDAC (57)

25% (6) 25% (5) 23% (5)

37% (6) 37% (6) 37% (6)

45% (7) 38% (7) 33% (7)

54% (7) 48% (7) 45% (7)

0.09

Table 5. Patient and study characteristics, DFS, OS and treatmentrelated mortality in two randomized trials comparing SDAC with HDAC as consolidation treatment. SAKK. (this study)

CALGB (Ref. [14])

Consolidation Eligibility for randomization SDAC

1 cycle CR + PR lOOmg/sqm x 7d

4 cycles

HDAC

3 g/sqm x 12

Schedule HDAC per cycle Addition DNM Maintenance Median follow up Number of patients Median age Treatment-related mortality in HDAC arm Actuarial data of randomized patients with confirmed CR DFS (four-year %)° SDAC HDAC OS (four-year % ) " SDAC HDAC

36 g in 6 days Yes No 72 months 137 (in 2 arms) 45 years 1/66 pts (course 1)

CR lOOmg/sqm x 5d (4 cycles) 3 g/sqm x 6 (4 cycles) 18 g in 5 days No 4 cycles 52 months 596 (in 3 arms) 50 years 9/187 pts (courses 1 to 4)

25% (13%-37%)° 37% (25%-49%)

21% (15%-26%) 39% (32%-46%)

38% (24°/^52%) 48% (34%-62%)

31%(24%-37%) 46% (38%-53%)

0.10

" Patient numbers, confirmed CR

disease, auerrods, leukocyte count and FAB), HDAC reduced the hazard of progression (hazard ratio = 0.61, P - 0.049) and of death (hazard ratio = 0.66, P = 0.13). In addition, we analyzed DFS of CR patients according to age group (<40 vs. ^40 years). In the younger age group, there was no significant difference between SDAC (n - 18, median 8.1 months) and HDAC (n - 18, median 10 months; P = 0.98). On the contrary, DFS was significantly longer in patients older than 40 years treated with HDAC (n - 39, median 16.6 months) compared to SDAC (n - 37, median 9.4 months; P = 0.04).

Discussion

Figure 3. Kaplan-Meier plots of Disease Free Survival (DFS) by treatment arm (CR patients only). Conventional Ara-C, n = 55; high dose Ara-C, n = 57.

The objective of this study was to investigate whether * 95% confidence interval. one cycle of HDAC consolidation would increase the proportion of AML patients in long term remission by transplantation; however, the two approaches result in postremission intensification. The multivariate analysis similar overall survival rates [28]. of disease-free survival suggests that, in CR patients, The optimization of dose and intensity of HDAC is one cycle HDAC may indeed reduce the hazard of far from being achieved. In this context, it is interesting relapse (hazard ratio = 0.61, P - 0.049) compared to to notice that our results of one single consolidation one cycle SDAC. There was also a trend for improved course of HDAC are comparable with those of repeated survival with HDAC (hazard ratio = 0.66, P = 0.13). The cycles of high dose Ara-C treatment in the CALGB higher efficacy (in terms of disease free survival) of AML study [14]. To our knowledge, this is the only other HDAC compared with SDAC or other postremission published randomized study comparing dose intensities intensifications is in accordance with findings previously of Ara-C as consolidation treatment of first complete reported for other randomized [14] and not randomized remission in adult AML. The patient characteristics and studies [26, 27]. HDAC seems associated with shorter median follow up of the two studies are comparable disease free survival only if compared with bone marrow except for a slight age difference and addition of dauno-

256 rubicin to consolidation in our study (Table 5). Our patients received a total dose of 36 g Ara-C in a single cycle whilst patients treated in the most intensive arm of the CALGB study received a total of 18 g per course repeated four times at monthly intervals. Thus, although the planned total dose of HDAC in the CALGB study was twice that of our study, our dose intensity of HDAC consolidation per course was twice that of the CALGB HDAC arm. The observation of comparable DFS and OS in our and the CALGB study suggests that dose intensity of Ara-C may be more important than total dosage [29]. Further studies are needed in order to verify whether additional courses of HDAC can improve the outcome compared to one single HDAC consolidation course. Whilst DFS and OS in the HDAC arms of these two studies appear to compare very well, toxicity and costs of treatment differ considerably. A single HDAC consolidation course (with or without daunorubicin) would be much less expensive than the CALGB multiple course strategy. As expected, significantly higher toxicity rates were observed in our HDAC arm compared with SDAC. High toxicity including sometimes unacceptably high treatment-related mortality under HDAC was observed in many other studies during intensive consolidation therapy [4, 12, 26, 30-32]. Few patients were able to sustain repeated and prolonged periods of HDAC-induced severe agranulocytosis without life-threatening complications with persistently high mortality rates during further HDAC-consolidation courses [12, 14]. In the CALGB study, only 56% of the patients allocated to the HDAC consolidation program received full treatment whereas in our study 84% of patients randomized to HDAC did receive the treatment as planned. This problem is particularly important when treating patients older than 60. Only 29% of 129 patients older than 60 years tolerated four courses of HDAC in the CALGB trial and only about one half of them received more than one course. The same problem arised in an other non-randomized study of HDAC consolidation, where only a minority of patients received the intended number of HDAC courses [12]. In summary, the outcome (DFS and OS) of our CR patients consolidated with a single course HDAC is better than with SDAC and compares favourably with the results of patients treated with much more aggressive and toxic consolidation such as repeated high-dose chemotherapy, autologous and allogeneic BMT [28, 31]. If confirmed, our results of short-term HDAC consolidation add an interesting option to the ongoing discussion on the most effective and least toxic consolidation treatment for adult AML patients in first complete remission.

Acknowledgements The authors thank the many physicians who cared for the patients and thus contributed data to this study. We

are also indebted to S. Hanselmann and E. Vogel for excellent data management and secretarial work. *The following persons and institutions participated in this trial and in the SAKK Leukaemia Project Group F. Cavalli, H. J. Senn, U Metzger and A. Goldhirsch (consecutive SAKK Chairmen), M. Castiglione (SAKK Coordinating Centre), M. Fopp (study coordinator), E. Schatzmann, S. Pampallona, M. Bacchi (SAKK statisticians), J. Fehr (morphology review), H. J. Muller, M. Jotterand Bellomo (cytogenetics); Haematology, University Hospital Basel: A Gratwohl, A. Tichelli; Haematology and Medical Oncology, Inselspital Berne' A. Tobler, G. Brun del Re, L. Tschopp, M. F Fey; Haematology-Oncology, University Hospital, Geneva: B. Chapuis; Haematology, University Hospital, Lausanne: V. von Fliedner, J. P. Grob; Haematology-Oncology, Cantonal Hospital, Neuchatel: P. Siegenthaler; Haematology-Oncology, Cantonal Hospital, St. Gallen. M. Fopp, L. Schmid; Oncology, Cantonal Hospital, Bellinzona: F. Cavalli, A. Pedrazzini; Haematology and Oncology, University Hospital, Zurich: J. Gmuer, J. Fehr, E. Jacky, U. Strebel.

References 1. Gale RP, Chne MJ. High remission induction rate in acute myeloid leukemia. Lancet 1977; 1- 497-9. 2. Bodey GP, Coltman CA, Hewlett JS et al. Progress in the treatment of adults with acute leukemia Arch Intern Med 1976; 136: 1383-8. 3. Rai KR, Holland JF, Glidewell OJ et al. Treatment of acute myelocytic leukemia: A study by cancer and leukemia group B. Blood 1981, 58: 1203-12 4. Bell R, Rohatiner AZS, Slevin ML et al. Short-term treatment for acute myelogenous leukaemia. BMJ 1982, 284. 1221—4. 5 Mayer RJ, Weinstein HJ, Coral FS et al. The role of intensive postinduction chemotherapy in the management of patients with acute myelogenous leukaemia. Cancer Treat Rep 1982; 66: 1455-62. 6. Preisler HD, Brecher M, Browman G et al. The treatment of acute myelogenous leukemia in patients 30 years of age and younger. Am J Hematol 1982; 13: 189-98. 7. Cassileth PA, Begg CB, Bennett JM et al. A randomized study of the efficacy of consolidation therapy in adult acute nonlymphocytic leukemia. Blood 1984; 63: 834-7. 8. Sauter C, Berchtold W, Fopp M et al. Acute myelogenous leukaemia: Maintenance chemotherapy after early consolidation treatment does not prolong survival. Lancet 1984; 1: 379-82. 9. Skipper HE, Schabel FM Jr, Wilcox WS. Experimental evaluation of potential anticancer agents. XIII. On the criteria and kinetics associated with 'curability' of experimental leukemia. Cancer Chemother Rep 1964; 35: 1-11. 10. Ellison RR, Holland JF, Weil M et al. Arabinosyl cytosine: A useful agent in the treatment of acute leukemia in adults. Blood 1968; 32: 507-23. 11. Herzig RH, Wolff SN, Lazarus HM et al. High-dose cytosine arabinoside therapy for refractory leukemia. Blood 1983; 62: 361-9. 12. Wolff SN, Herzig RH, Fay JW et al. High-dose cytarabine and daunorubicin as consolidation therapy for acute myeloid leukemia in first remission: Long-term follow-up and results. J Clin Oncol 1989; 7: 1260-7. 13. Mitus AJ, Miller KB, Shenkein DP. Improved survival for patients with acute myelogenous leukemia. J Clin Oncol 1995; 13: 560-9. 14. Mayer RJ, Davis RB, Schiffer CA et al. Intensive postremission chemotherapy in adults with acute myeloid leukemia. N Engl J Med 1994; 331: 896-903.

257 15. Tschopp L, von Fliedner V, Sauter C et al. Efficacy and clinical cross-resistance of a new combination therapy (AMSA/VP-16) in previously treated patients with acute nonlymphocytic leukaemia. J Clin Oncol 1986; 4: 318-24. 16. Fopp M, Gratwohl A, Tschopp L et al. Non-cross-resistant chemotherapy of adults with untreated acute myeloid leukaemia(AML) using three drug combinations including amsacrine (AMSA) etoposide (VP-16) and high-dose cytosine-arabinoside (Ara-Q. Proc Am Soc Clin Oncol 1987; 6: 152. 17. Bennet JM, Catovsky D, Daniel M et al. Proposals for the classification of acute leukaemias. Br J Haematol 1976; 33: 451-8. 18. Ellison RR, Holland JF, Weil M et al. Arabinosyl cytosine: A useful agent in the treatment of acute leukemia in adults. Blood 1968; 32: 507-23. 19. World Health Organization. WHO Handbook for Reporting Results of Cancer Treatment. Geneva. WHO 1979. 20. Simon R. Confidence intervals for reporting results of clinical trials. Ann Intern Med 1986, 105: 429-35 21 Walker SH, Duncan DB. Estimation of the probability of an event as a function of several independent variables. Biometrika 1967; 54: 167-79. 22. Kaplan EL, Meier P. Nonparametric estimation from incomplete observations. J Am Statist Assoc 1958; 53: 457-81. 23. Greenwood M. Reports on Public Health and Medical Subjects. The Natural Duration of Cancer. London: His Majesty's Stationary Office 1926; 33: 1-26. 24. Gray RJ, Tsiatis AA. A linear rank test for use when the main interest is in differences in cure rates. Biometrics 1989; 45. 899-904. 25. Cox DR. Regression models and life tables (with discussion). J R Stat Soc 1972; 34 (Ser B)- 187-220. 26. Cassileth PA, Lynch E, Hines JD et al. Varying intensity of

27

28.

29.

30.

31.

32.

postremission therapy in acute myeloid leukemia. Blood 1992; 79: 1924-30. Champlin R, Gajevski J, Nimer S et al. Postremission chemotherapy for adults with acute myelogenous leukemia: Improved survival with high dose cytarabine and daunorubicin consolidation treatment. J Clin Oncol 1990, 8. 1199-206. Zittoun RA, Mandelli F, Willemze R et al. Autologous or allogeneic bone marrow transplantation compared with intensive chemotherapy in acute myelogenous leukemia. N Engl J Med 1995; 332: 217-23. Coldman AJ, Goldie JH. Impact of dose-intense chemotherapy on the development of permanent drug resistance. Semin Oncol 1987; 14 (Suppl 4): 29-33. Harousseau JL, Milpied N, Briere J et al. Double intensive consolidation chemotherapy in adult acute myeloid leukemia. J Clin Oncol 1991; 9: 1432-7 Schiller GJ, Nimer SD, Territo MC et al. Bone marrow transplantation versus high-dose cytarabine-based consolidation chemotherapy for acute myelogenous leukaemia in first remission. J Clin Oncol 1992; 10. 41-6. Rubin EH, Andersen JW, Berg DT et al. Risk factors for highdose cytarabine neurotoxicity: An analysis of a cancer and leukemia group B trial in patients with acute myeloid leukemia. J Clin Oncol 1992; 10: 948-53.

Received 2 October 1996; accepted 28 January 1997. Correspondence to. M. Fopp, MD Regionales Blutspendezentrum SRK Kantonsspital CH-9007 St. Gallen, Switzerland