Escalation of daunorubicin and addition of etoposide in the ADE regimen in acute myeloid leukemia patients aged 60 years and older: Cancer and Leukemia Group B Study 9720

Abstract

Untreated de novo (n=421) and secondary (n=189) acute myeloid leukemia (AML) patients≥60 years received intensified chemotherapy, including daunorubicin 60mg/m² and etoposide 100mg/m² during days 1, 2, 3 with cytarabine 100mg/m² during days 1–7, with a second induction if needed and one consolidation course with these drugs and doses for 2, 2 and 5 days, respectively. In all, 287 (47%) achieved complete remission (CR), 136 (22%) died and 187 (31%) were non-responders. CR rates were 27, 44 and 52% for complex karyo-types, rare aberrations and neither (P<0.001), 52 and 37% for de novo and secondary AML (P=0.003), and 53 and 42% for age 60–69 and ≥70 years (P=0.015). In multivariable analysis, CR predictors included non-complex/non-rare karyotypes (P<0.001), de novo AML (P<0.001), better performance status (PS) (P<0.001) and younger age (P=0.001). Disease-free (DFS) and overall (OS) survival medians were 6.8 (95% CI: 6.2, 7.8) and 7.2 (95% CI: 6.4, 8.6) months. In multivariable analysis, DFS was shorter for complex karyotypes (P<0.001) and increasing white blood count (WBC) (P<0.001) and age (P=0.038), and OS for complex karyotypes (P<0.001), increasing WBC (P=0.001) and age (P<0.001), poorer PS (P<0.001) and secondary AML (P=0.010). Outcomes and prognostic factors were similar to those in previous Cancer and Leukemia Group B studies.

Introduction

The incidence of acute myeloid leukemia (AML) increases with age, with a median age of 67 years at diagnosis.¹ Older AML patients benefit from chemotherapy relative to supportive care,^2,3 but their outcome from chemotherapy remains significantly worse than that of younger patients.^4,5 Inferior outcomes in older AML patients reflect both impaired ability to withstand intensive chemotherapy due to inferior performance status (PS), comorbidities^5–9 and adverse biological features of the disease,^5,8,9 including high frequencies of karyotypes that are prognostically unfavorable in younger patients, expression and function of the multidrug resistance protein P-glycoprotein (Pgp)⁵ and secondary AML, defined by antecedent bone marrow (BM) disorders including myelodysplastic syndromes (MDS) and/or therapy-related AML.⁸ Even older patients with AML features that are favorable in younger patients, such as core-binding-factor (CBF) cytogenetic abnormalities, have worse outcomes than younger patients, possibly reflecting properties of the stem cells in which leukemia arises in older patients.⁵

Dose intensification is being explored as an approach to improving outcome in older AML patients. In a recent European cooperative group study, increasing the daily daunorubicin dose from 45mg/m² to 90mg/m² in conjunction with cytarabine (AraC) in AML patients ≥60 years benefited patients 60–65 years, increasing their complete remission (CR) rate from 35 to 52% after one induction course and from 54 to 64% after a subsequent high-dose AraC course, and prolonging event-free and overall survival (OS).¹⁰ Another recent study, ALFA-9801, compared daunorubicin 80mg/m² for 3 days and idarubicin 12mg/m² for 3 days and for 4 days in de novo AML patients 50–70 years old, and found the highest CR rate (83%) with idarubicin for 3 days, albeit with no difference in event-free survival or OS.¹¹

Cancer and Leukemia Group B Study 9720 (CALGB 9720) was a phase III trial evaluating both the Pgp modulator PSC-833 (Valspodar) in conjunction with AraC, daunorubicin and etoposide (ADE) in induction and consolidation therapy¹² and subcutaneous recombinant interleukin-2 maintenance immunotherapy ¹³ in previously untreated AML patients ≥60 years. A total of 669 patients were enrolled. The PSC-833-containing arm was closed owing to excessive toxicity after randomization of 120 patients, including 61 to ADE and 59 to ADE with PSC-833 (ADEP).¹² ADE was subsequently administered to 549 additional patients without PSC-833. One-hundred-forty-seven of the 610 ADE patients and 10 of the 59 ADE with PSC-833 were randomized to recombinant interleukin-2 or no maintenance therapy, without a difference in outcome.¹³ The two previous publications from CALGB 9720 included only randomized patients (for induction or interleukin-2).

In this study, we report the outcome of all 610 previously untreated AML patients ≥60 years old treated with ADE on CALGB 9720, including the 61 patients randomized to ADE and the 549 patients assigned to ADE after the induction randomization was stopped. The ADE regimen was developed in an earlier Phase I trial, CALGB 9420,¹⁴ in untreated AML patients ≥60 years. The purpose was to develop a regimen incorporating two Pgp substrate drugs, daunorubicin and etoposide, in addition to AraC, in order to optimize subsequent testing of Pgp modulation, and to dose-escalate daunorubicin, in conjunction with AraC 100 mg/m² and etoposide 100 mg/m². The maximum tolerated dose of daunorubicin in this regimen in patients ≥60 years was 60 mg/m².

Materials and methods

Eligibility

Eligibility criteria for enrollment on CALGB 9720 were previously described^12,13 and included a diagnosis of AML with French-American-British (FAB) types M0-M2 or M4-M7, no prior treatment for AML and age ≥60 years, without an upper age limit. The only explicit eligibility criteria were diagnosis, age and prior treatment; there was no requirement for PS, but it was specified that physicians should only enroll patients for whom the agents to be administered were appropriate. Patients with de novo and secondary (antecedent MDS or therapy-related) AML were eligible. Antecedent MDS was defined by cytopenias and BM morphology diagnostic of MDS at least 3 months before the AML diagnosis. Prior treatment for AML or MDS, including demethylating agents, was not permitted, except for growth factor or cytokine support and leukapheresis, hydroxyurea or cranial irradiation to manage hyperleukocytosis. The protocol was approved by local institutional review boards, and written informed consent was obtained from all patients.

Induction and consolidation chemotherapy

This paper reports outcomes for all CALGB 9720 patients who received ADE induction therapy,^12,13 including Ara-C 100 mg/m² per day by continuous infusion for 7 days and daunorubicin 60 mg/m² by intravenous bolus injection and etoposide 100 mg/m² by 2-h infusion, both daily for 3 days, initiated concurrently with the AraC.

Patients with persistent marrow leukemia, defined by ≥5% blasts in the aspirate and ≥20% cellularity in the biopsy on day 14 following initiation of induction therapy, received a second induction course identical to the first, but with AraC for 5 rather than 7 days and two rather than three doses of daunorubicin and etoposide.

Patients who achieved CR received a single course of post-remission chemotherapy identical to the second induction regimen (that is, 5+2+2), initiated as soon as possible following attainment of CR, but no sooner than day 28 of induction. Full recovery from infection and stomatitis and attainment of CALGB PS 0 or 1 were required. Repeat BM testing was required if post-remission chemotherapy was delayed more than 4 weeks following initial documentation of CR. Administration of hematopoietic growth factors was not encouraged, but was allowed per the guidelines of the American Society of Clinical Oncology.

The design of CALGB 9720 included, for patients who completed consolidation therapy, a randomization between interleukin-2 and no maintenance therapy. As previously reported, outcomes did not differ between these two approaches,¹³ and herein these two groups are pooled.

Response and toxicity definitions

CR was defined according to 1990 National Cancer Institute workshop criteria,¹⁵ including absolute neutrophil count ≥1.5×10⁹ per l, platelet count >100×10⁹ per l, absence of leukemia cells in the blood, no extramedullary disease and BM cellularity >20%, with <5% blasts and normal myeloid maturation. Induction outcomes were also analyzed by 2003 National Cancer Institute workshop criteria,¹⁶ with CR with incomplete count recovery defined by meeting all criteria for CR except for incomplete platelet or neutrophil recovery. Induction deaths were defined as deaths while cytopenic without evidence of persistent leukemia. Relapse was defined as marrow infiltration by ≥5% blasts not attributable to another cause or development of extramedullary disease in a patient previously in CR. Disease-free survival (DFS) was measured as the interval from achievement of CR until relapse or death, regardless of cause, and OS as the interval from on-study date until death. Patients alive at last follow-up were censored for both DFS and OS. Toxicity was graded according to Common Toxicity Criteria version 2.0 (National Cancer Institute, Bethesda, MD, USA).

Auditing

As part of the CALGB quality assurance program, Data Audit Committee members visit all participating institutions at least every 3 years to review source documents and verify compliance with federal regulations and protocol requirements, including those pertaining to eligibility, treatment, adverse events, tumor response and outcome in a sample of protocols at each institution. Such on-site review of medical records was performed for 226 (34%) of the 669 patients enrolled on CALGB 9720.

Cytogenetic analysis

Cytogenetic analysis was performed as part of a prospective karyotyping study, CALGB 8461, ‘Cytogenetic Studies in Acute Leukemia,’ which was a mandatory companion study for CALGB 9720, as previously described.^12,13 The cytogenetic risk groups used in the analysis were those established by Farag et al.¹⁷ for AML patients ≥60 years. Specifically, complex karyotypes with ≥3 abnormalities and rare aberrations predicted lower CR rates, complex karyotypes with ≥5 abnormalities predicted shorter DFS, complex karyotypes with ≥5 abnormalities and rare aberrations predicted shorter OS, and CBF abnormalities predicted longer OS.

Statistical analyses

Patients with CRs lasting <4 weeks or achieving CR with incomplete count recovery were classified as non-responders. Comparisons of response status across baseline characteristics were performed using the Fisher’s exact test. Univariable survival analyses employed the Kaplan–Meier estimation method¹⁸ and the logrank test for comparing survival distributions. Multivariable analyses used the Cox proportional hazards model and theWald χ²-test. Variables considered for inclusion in the multivariable analyses were age, sex, disease onset, white blood count (WBC), race, FAB, PS and cytogenetics. The final multivariable models included only those variables with an individual P-value <0.05. Analyzes were performed by CALGB statisticians and were based upon data available as of October 2009. The reported P-values are two-sided.

Results

Six-hundred-ten AML patients ≥60 years were treated with ADE on CALGB 9720 between March 1998 and April 2002 (Supplementary Figure 1). Pretreatment characteristics are shown in Table 1. Median age was 71 years (range, 60–90). Four-hundred-twenty-one patients (69%) had de novo AML and 189 (31%) secondary AML. By the Farag et al.¹⁷ OS risk criteria, 88 (19%) had complex karyotypes with ≥5 abnormalities, 15 (3%) CBF abnormalities, 9 (2%) rare aberrations, and 362 (76%) none of the above. Cytogenetic results were not available for 136 patients (22%); response rates, DFS and OS for these patients were quite similar to those for patients with known karyotypes.

Table 1.

Pretreatment characteristics

Age (years), median (range)	71 (60–90)
Sex, n (%)
Male	366 (60)
Female	244 (40)
Race, n (%)
White	533 (87)
African–American	43 (7)
Other	34 (6)
Performance status, n (%)
0	169 (29)
1	265 (46)
2	117 (20)
3+	29 (5)
AML onset,a n (%)
De novo	421 (69)
Secondary	189 (31)
Antecedent MDS	138 (23)
t-AML	47 (8)
Antecedent MDS and t-AML	4 (<1)
White blood count × 109 per l, median (IQR)	7.0 (2.2–33.1)
Cytogenetics,b n (%)
Complex karyotype, ≥5 abnormalities	88 (19)
CBF	15 (3)
Rare aberrations	9 (2)
Other	362 (76)

Abbreviations: AML, acute myeloid leukemia; CBF, core binding factor; IQR, interquartile range; MDS, myelodysplastic syndromes.

^aData available for 564 patients.

^bData available for 474 patients. The cytogenetic categories are those of Farag et al.¹⁸ for overall survival.

CRs were achieved in 287 patients (47%), with 136 (22%) induction deaths and 187 (31%) non-responders. Non-responders included 37 (6%) with CR with incomplete count recovery and 5 (<1%) who met CR criteria for <4 weeks. Among the 136 induction deaths, 96 occurred in the first 4 weeks, 37 between 4 and 8 weeks and 3 beyond 8 weeks. Looking at induction deaths and deaths with persistent leukemia, 108 patients (18%) died in the first 4 weeks, and 57 (9%) between 4 and 8 weeks. 4-week induction death rates were 10, 14, 19, 14% and 33% for ages 60–64, 65–69, 70–74, 75–79 and ≥80 years, and 4–8-week death rates 3, 8, 6, 8 and 3%. For all deaths, 4-week rates were 11, 17, 21, 17 and 36%, and 4–8-week rates were 4, 12, 10, 13 and 3%. The major causes of induction deaths were sepsis and multi-organ failure

CR rates were 27% for patients with complex karyotypes (≥3 abnormalities), 44% for those with rare aberrations¹⁷ and 52% for those with neither finding (P<0.001); 52% for de novo AML and 37% for secondary AML (P=0.003) and 53% for ages 60–69 years (61% for 60–64 years and 47% for 65–69 years), 43% for ages 70–79 years (41% for 70–74 years and 47% for 75–79 years) and 27% for ages ≥80 years (P<0.001; Table 2). In a multivariable logistic regression model, predictors of CR included non-complex (<3 abnormalities) non-rare karyotypes (P<0.001); de novo AML (P<0.001); younger age (P=0.001); better PS (P<0.001) and lower WBC (P=0.026; Table 3).

Table 2.

Outcomes of induction therapy by baseline characteristics

Characteristic	CR (n=287), N (%)	NR (n=187), N (%)	Death (n=136), N (%)	Total (n=610) no.	Pa
Age, years
60–69	152 (53)	81 (28)	53 (19)	286	0.014
60–64	79 (61)	33 (25)	18 (14)	130
65–69	73 (47)	47 (30)	36 (23)	156
70–79	126 (43)	94 (32)	71 (24)	291
70–74	72 (41)	59 (33)	46 (26)	177
75–79	54 (47)	35 (31)	25 (22)	114
≥80	9 (27)	12 (36)	12 (36)	33
Sex
Male	177 (48)	112 (31)	77 (21)	366	0.6066
Female	110 (45)	75 (31)	59 (24)	244
Performance status
0	96 (57)	50 (30)	23 (14)	169	<0.0001
1	130 (49)	80 (30)	55 (21)	265
2	45 (38)	41 (35)	31 (26)	117
3+	8 (28)	7 (24)	14 (48)	29
Disease onset
De novo	217 (52)	119 (28)	85 (20)	421	0.004
Secondary	70 (37)	68 (36)	51 (27)	189
WBC, ×109 per l
<10k	156 (46)	102 (30)	83 (24)	341	0.404
≥10k	131 (49)	85 (32)	53 (20)	269
Race
White	253 (47)	164 (31)	116 (22)	533	0.702
Other	34 (44)	23 (30)	20 (26)	77
FAB
M1	65 (47)	43 (31)	31 (22)	139	0.458
M2	93 (47)	61 (31)	46 (23)	200
M4	53 (56)	21 (22)	21 (22)	95
Other	76 (43)	62 (35)	38 (22)	176
Cytogeneticsb					+
Complex ≥3	29 (27)	43 (41)	34 (32)	106	<0.0001
Rare aberrations	185 (52)	105 (29)	69 (19)	359
Otherc	4 (44)	5 (56)	0 (0)	9

Abbreviations: CR, complete remission; FAB, French-American-British classification; NR, patients alive without a CR following induction; WBC, white blood cell.

Deaths refer to patients who died during the induction phase of treatment. NR category includes CR <4 weeks, CRi, and patients whose status was indeterminate.

^aP-values refer to overall comparison of categories versus outcomes using Fisher’s Exact Test.

^bCytogenetic categories based on the risk groups defined by Farag et al.¹⁸ for complete response. Cytogenetic classifications were available for 474 of the 610 ADE patients.

^cAML patients with fewer than three non-rare aberrations.

Table 3.

Multivariable analysis of complete remissions

Variable in the final model	Adjusted OR	95% CI	Pa
Performance statusb	0.54	(0.34, 0.85)	0.008
Cytogenetics
Rare aberrationsc	3.63	(0.86, 15.27)	0.079
Non-complex, non-rare karyotypec	3.21	(1.94, 5.33)	<0.001
De novo versus secondary AML	2.26	(1.46, 3.51)	<0.001
Aged	0.78	(0.66, 0.92)	0.004
Log(WBC)	0.87	(0.77, 0.99)	0.036

Abbreviations: AML, acute myeloid leukemia; CI, confidence interval; OR, odds ratio; WBC, white blood cell.

^aP-values refer to Wald χ²-test.

^bPS=(2,3,4) versus PS=(0,1)

^cVersus complex karyotype (≥3 aberrations).

^dIncrease in risk for each increment of 5 years.

Of 402 patients with day-14 BM data available, 322 (80%) achieved marrow hypoplasia. Of the 80 patients whose day-14 marrows were not hypoplastic, only 49 (61%) received second induction courses, but 37 patients who achieved marrow hypoplasia also received second induction courses. CR rates for patients receiving one and two induction courses were 52 and 34%, respectively (P<0.001). In addition, CR was achieved in 6 of 12 patients (50%) with number of induction cycles not reported. The comparison of patients who received one versus two induction courses is confounded by the facts that those who received a second course did so because they did not respond to one course, and, additionally, some patients who did not respond to the first course were unable to receive a second course. The question of benefit of a second intensive induction course in older patients cannot be resolved from our data.

Median times to absolute neutrophil count ≥1.0 × 10⁹ per l were 34 days (95% CI: 32, 35) following one induction course, and 53 days (95% CI: 47, 55) following two. Median times to platelet count ≥100 × 10⁹ per l were 34 (95% CI: 33, 36) and 51 (95% CI: 46, 55) days.

Non-hematologic toxicities associated with ADE induction therapy are summarized in Supplementary Table 1. The regimen was generally well tolerated.

Consolidation chemotherapy was administered to 237 patients (39%), of whom 147 were randomized to recombinant interleukin-2 maintenance immunotherapy (n=73) or no maintenance therapy (n=74), with no difference in outcome.¹³ DFS for patients achieving CR was short, a median of 6.8 months (95% CI: 6.2, 7.8) (Figure 1a), as was OS for all patients, a median of 7.2 months (95% CI: 6.4, 8.6) (Figure 2a). The median OS for patients achieving a CR was 15.0 months; for patients with a CR with incomplete count recovery or CR lasting <4 weeks, the median survival was 11.1 months; for other non-responding patients the median OS was 4.7 months. OS was also age-dependent in older patients. For patients aged 60–64, 65–69, 70–74, 75–79 and ≥80 years, median OS was 11.3 (9.5–13.3), 6.6 (5.1–8.8), 6.1 (4.4–7.5), 6.2 (4.7–9.8) and 4.4 (0.8- 9.5) months, and 1-year survival was 0.462 (0.374–0.544), 0.325 (0.252–0.399), 0.297 (0.231–0.365), 0.377 (0.289–0.465) and 0.212 (0.094–0.363), respectively.

Figure 1.

(a) Disease-free survival (DFS) in the 287 patients who achieved complete remission (CR). There were 274 events (95%). Median DFS was 6.8 months, with a 95% confidence interval of 6.2–7.8 months. (b) DFS by cytogenetics in the 218 responding patients with known karyotypes. (c) DFS by age in all 287 patients. (d) DFS by disease type (de novo acute myeloid leukemia (AML) versus secondary AML) in the 287 patients, with a CR. (e) DFS by performance status ((0,1) versus (2,3,4)).

Figure 2.

(a) Overall survival (OS) in all 610 patients. There were 588 deaths (96%). Median OS was 7.3 months, with a 95% confidence interval of 6.4–8.6 months. (b) OS by cytogenetics in the 474 patients with known karyotypes. (c) OS by age in all 610 patients. (d) OS by disease type (de novo acute myeloid leukemia (AML) versus secondary AML) in all 610 patients. (e) OS by performance status ((0,1) versus (2,3,4)).

In univariable analyses, summarized in Figure 1, DFS was shorter for complex (≥5 abnormalities) versus non-complex karyotypes (P<0.001) (Figure 1b). Other factors did not reach the traditional nominal P-value cut-off of 0.05 although it was close for secondary versus de novo AML (P=0.063) (Figure 1d). The results of the multivariable analysis, summarized in the top part of Table 4, indicated that DFS was shorter for patients with complex karyotypes (P<0.001), for those with higher WBC (P<0.001) and for older patients (P=0.038).

In univariable analyses, summarized in Figure 2, OS differed by cytogenetic risk group (P<0.001) (Figure 2b), and was shorter for older patients (P<0.001) (Figure 2c), those with secondary AML (P<0.001) (Figure 2d) and those with poorer PS. The results of the multivariable analysis, summarized in the bottom part of Table 4, indicated that OS was shorter for patients with complex (≥5 abnormalities) karyotypes (P<0.001), higher WBC (P<0.001), older patients (P<0.001), those with de novo AML (P=0.011) and those with a poorer PS (P<0.001). Of note, CBF karyotypes, rather than normal karyotypes, were chosen as the cytogenetic reference group because of their greater biological and clinical homogeneity.

Analyses of treatment effect by race and sex, not reported here, revealed no suggestion of a treatment effect in any of the four subgroups defined by race (white, non-white) and sex (male, female).

Results of two major CALGB clinical trials in previously untreated AML patients, CALGB 8525⁴ and CALGB 8923,¹⁹ were used in historical comparisons with the results reported here. Both had ‘7+3’ induction regimens, with daunorubicin 30 mg/m² and 45 mg/m², respectively. Eligibility requirements were not identical to those of CALGB 9720, as they excluded patients with secondary AML, but included patients <60 years old (CALGB 8525) and patients with acute promyelocytic leukemia. Thus for valid historical comparisons, attention was restricted to de novo AML patients in CALGB 9720 and to CALGB 8525 and 8923 patients who would have met the eligibility requirements for CALGB 9720. There were 728 such patients in the historical control group and 421 in the current study. The increased dose of daunorubicin and the addition of etoposide in the ADE regimen did not improve outcomes. The pooled CR rate of 49% (95% CI: 46%, 53%) on the two historical studies was similar to the CR rate of 52% (47%, 56%) on the current study. Median DFS was 10.2 months (8.8 months, 11.8 months) on the historical studies, significantly longer than the median of 7.0 months (6.4 months, 8.6 months) on the current study. Median OS was 7.6 months (95% CI: 6.1 months, 9.0 months) on the historical studies, similar to the 8.4 months (6.9 months, 9.6 months) on the current study.

Discussion

In the study reported here, we showed that intensification of standard ‘7+3’ AraC and daunorubicin therapy by dose escalation of daunorubicin and addition of etoposide was feasible in older AML patients in a cooperative group setting, but did not improve outcomes compared with historical ‘7+3’ trials, nor alter the influence of previously established prognostic factors.

Daunorubicin dose escalation has been a focus of several recent clinical trials in younger AML patients, with evidence of beneficial effects. In a comparison of previously untreated patients aged 13 to 67 years treated on two sequential clinical trials with AraC 100 mg/m² daily for 7 days, etoposide 100 mg/m² daily and daunorubicin at either 45 or 75 mg/m² for 3 days,²⁰ CR was achieved in 59 and 77% of patients treated with the lower and higher daunorubicin doses (P=0.03), with 64 and 88% (P=0.02), respectively, of CRs occurring after a single induction course, and with a survival advantage for the higher-dose cohort (40 versus 23%, P=0.03). Older age was an adverse factor for survival. CALGB has also escalated the daunorubicin dose and added etoposide in prospective studies in AML patients <60 years old. CALGB study 9621 evaluated an ADE regimen with a daunorubicin dose of 90 mg/m² for 3 days, yielding a CR rate of 78%, with most CRs achieved after a single induction course.²¹ This response rate was confirmed in a subsequent study with the same regimen, CALGB 19808.²² Finally, in a recent Eastern Cooperative Oncology Group phase III trial (E1900) of 90mg/m² versus 45mg/m² daunorubicin with cytarabine as induction therapy for untreated AML patients <60 years old, the higher daunorubicin dose regimen resulted in a higher CR rate (70.6 versus 57.3%,) and longer OS (median, 23.7 versus 15.7 months), with similar rates of serious adverse events.²³

As noted above, in a recent European cooperative group study an increase in daunorubicin dose from 45 mg/m² to 90 mg/m² in conjunction with AraC in AML patients aged ≥60 years, increased the CR rate from 35 to 52% after a single induction course and from 54 to 64% after a subsequent course of high-dose AraC, and prolonged event-free survival and OS in patients 60–65 years old.¹⁰ In contrast, other studies of dose intensification in older patients have not shown benefit. As also noted above, another recent study, ALFA-9801, compared daunorubicin 80 mg/m2 for 3 days and idarubicin 12 mg/m² for 3 days and for 4 days de novo AML patients 50 to 70 years old, and found the highest CR rate (83%) with idarubicin for 3 days, albeit with no difference in event-free survival or OS.¹¹ Another recent study from the United Kingdom compared daunorubicin at 50 mg/m² and 35 mg/m², and found no differences.²⁴ Other investigators compared a single high dose of mitoxantrone (80 mg/m² on day 2) with standard doses (12 mg/m² on days 1–3) in older AML patients, and found no differences.²⁵ Comparisons among daunorubicin, idarubicin and mitoxantrone in older patients have also generally not shown significant differences,^26–29 despite the fact that idarubicin may be less susceptible to resistance mediated by Pgp,^30,31 whereas idarubicin may be superior in younger patients.³² Similarly, in our study, dose escalation of daunorubicin to 60 mg/m² did not improve induction outcome or survival in older AML patients in relation to historical studies with daunorubicin 30 mg/m² or 45 mg/m².4,19

In our study, the induction regimen was also intensified by adding etoposide. In a randomized trial by the Australian Leukemia Study Group, adding etoposide to AraC and daunorubicin induction and consolidation therapy for patients aged 15–70 years benefited patients <55 years, but not older patients.³³ This is consistent with the lack of benefit for older patients in our study.

Daunorubicin and etoposide are both substrates for Pgp. Pgp expression and function in AML blasts are more common in older than in younger AML patients,³⁴ and have strong adverse prognostic significance in older patients.^5,34,35 Nevertheless, co-administration of the potent Pgp inhibitor valspodar (PSC-833) with chemotherapy regimens that included Pgp substrate drugs failed to improve treatment outcome in older AML patients in the randomized portion of CALGB 9720¹² and in other studies.³⁵ Cyclosporine showed efficacy in a clinical trial in older AML patients,³⁶ but may act by other mechanisms in addition to Pgp modulation.³⁷ One possible explanation for enhanced efficacy of daunorubicin at 90mg/m² is that this dose might partially overcome Pgp-mediated drug efflux.

Finally, not only were results of CALGB 9720 not better than those of 8525 and 8923, but also DFS on 9720 was in fact shorter. One possible explanation for this observation is that 9720 included only one consolidation course, whereas 8525 included four and the two arms of 8923 included either four or two. Whether additional consolidation courses prolong DFS in older AML patients would need to be tested in a randomized clinical trial. Of note, a German AML Cooperative Group trial demonstrating longer relapse-free survival with one consolidation course followed by maintenance, compared with two consolidation courses,³⁸ raises the possibility that duration, rather than intensity, of post-CR therapy may be important in poor-risk, including older, AML patients.

New treatment approaches continue to be needed for AML in older patients. Possible approaches include substitution of non-Pgp-substrates, such as amonafide³⁹ for daunorubicin and etoposide, co-administration of new multidrug resistance modulators or alternative approaches to chemosensitization, such as epigenetic therapies and signal transduction inhibitors, as well as stratification of treatment based on cytogenetic and molecular subsets, as with younger patients, as additional prognostic information⁴⁰ becomes available. Additionally, reduced intensity-conditioning regimens followed by allogeneic hematopoietic stem-cell transplantation may hold promise for appropriate older AML patients.

Table 4.

Multivariable Cox proportional hazards modeling of DFS and OS

Variable in the final model	Adjusted HR	95% CI	Pa
DFS (n=211; 202 events)
Performance statusb	1.03	0.72, 1.47	0.893
Cytogenetics
Complex (≥5 aberrations)c	5.15	3.18, 8.35	<0.001
De novo AML	0.78	0.56, 1.10	0.161
Aged	1.11	0.99, 1.25	0.088
Log(WBC)	1.17	1.06, 1.30	0.001
OS (n=449; 433 events)
Performance statusb	1.41	1.13, 1.75	0.002
Cytogeneticse
Complex (≥5 aberrations)	3.93	2.08, 7.42	<0.001
Rare aberrations	0.89	0.36, 2.22	0.807
Non-CBF, <5 non-rare aberrations	1.21	0.62, 2.02	0.705
Log(WBC)	1.14	1.06, 1.21	<0.001
Aged	1.15	1.07, 1.25	<0.001
De novo	0.76	0.62, 0.94	0.010

Abbreviations: AML, acute myeloid leukemia; CBF, t(8;21)(q22;q22) and inv(16)(p13q22) or t(16;16)(p13;q22), cytogenetic abnormalities characteristic of core-binding-factor AML; DFS, disease-free survival; HR, hazard ratio; OS, overall survival; WBC, white blood cell.

^aP-values refer to Wald χ²-test.

^bPS=(2,3,4) versus PS=(0,1).

^cVersus all other karyotypes.

^dIncrease in risk for each increment of 5 years.

^eEach category versus CBF AML.