which aml subsets benefit from leukemic cell priming during chemotherapy? long-term analysis of the...

Which AML Subsets Benefit From LeukemicCell Priming During Chemotherapy?Long-Term Analysis of the ALFA-9802GM-CSF StudyXavier Thomas, MD, PhD1,8; Emmanuel Raffoux, MD2,8; Aline Renneville, PhD3,8; Cecile Pautas, MD4,8;

Stephane de Botton, MD5,8; Christine Terre, MD6,8; Claude Gardin, MD7,8; Sandrine Hayette, PhD1,8;

Claude Preudhomme, PhD3,8; and Herve Dombret, MD2,8

BACKGROUND: Priming with granulocytic hematopoietic growth factors may modulate cell cycle kinetics of leukemic

cells and render them more susceptible to phase-specific chemotherapeutic agents. In a first report, we have shown

that priming with granulocyte-macrophage colony-stimulating factor (GM-CSF) may enhance complete remission

(CR) rate and event-free survival (EFS) in younger adults with acute myeloid leukemia (AML). METHODS: In this

randomized trial, 259 patients with AML were randomized at baseline to receive or not receive GM-CSF concurrently

with all cycles of chemotherapy. The effects of GM-CSF on survival were reported herein with a long-term follow-up

and studied according to distinct biological subgroups defined on cytogenetics and molecular markers. RESULTS:

The EFS rate was better in the GM-CSF group (43% vs 34%; P¼.04). GM-CSF did not improve the outcome in

patients from good risk subgroups, while patients from poor risk subgroups benefited from GM-CSF therapy. In this

population, the difference in terms of EFS probability was mainly observed in patients with high initial white blood

cell count and in those with FLT3-ITD or MLL rearrangement. When combining these 2 molecular abnormalities for

comparison of the effect of GM-CSF priming, the difference in terms of EFS was highly significant (5-year EFS, 39%

with GM-CSF vs 8% without GM-CSF; P¼.007). CONCLUSIONS: Sensitization of leukemic cells and their progenitors

by GM-CSF appears as a plausible strategy for improving the outcome of patients with newly diagnosed AML.

Patients with poor-prognosis FLT3-ITD or MLL rearrangement might be a good target population to further investi-

gate priming strategies. Cancer 2010;116:1725–32. VC 2010 American Cancer Society.

KEYWORDS: acute myeloid leukemia, priming, timed-sequential chemotherapy, GM-CSF, prognosis.

Priming with granulocytic hematopoietic growth factors may modulate cell cycle kinetics of acute myeloid leukemia(AML) blasts and render them more susceptible to phase-specific agents such as cytarabine. In vitro, the simultaneous exposureof leukemic cells to chemotherapy and hematopoietic growth factors increases the susceptibility of the cells to killing by chemo-therapy, especially by the cell-cycle-specific agent cytarabine.1-3 We first conducted a randomized double-blind study of theadministration of granulocyte-macrophage colony-stimulating factor (GM-CSF) in association with induction timed-sequentialchemotherapy (TSC) in relapsing AML patients that showed a time to progression that tended to be longer in the GM-CSFgroup.4 In the present Acute Leukemia French Association (ALFA)-9802 trial, newly diagnosed AML patients were

DOI: 10.1002/cncr.24943, Received: May 8, 2009; Revised: July 10, 2009; Accepted: July 31, 2009, Published online February 8, 2010 in Wiley InterScience

(www.interscience.wiley.com)

Corresponding author: Xavier Thomas, MD, PhD, Department of Hematology, Hopital Edouard Herriot, 5 place d’’Arsonval, 69437 Lyon, France; Fax: (011)

33 1 72117404; [email protected]

1Department of Hematology, Edouard Herriot Hospital, Lyon, France; 2Department of Hematology, Saint-Louis Hospital, Paris, France; 3Department of Hematol-

ogy, Claude Huriez Hospital, Lille, France; 4Department of Hematology, Henri Mondor Hospital, Creteil, France; 5Department of Hematology, Gustave Roussy Insti-

tute, Villejuif, France; 6Department of Hematology, Andre Mignot Hospital, Versailles, France; 7Department of Hematology, Avicenne Hospital, Bobigny, France;8Acute Leukemia French Association (ALFA), Saint-Louis Hospital, Paris, France

All authors participated actively in the study conception and design and acquisition of data. XT included patients, conducted the statistical analysis, interpreted

the data, and was the main author of the manuscript; ER, SdB, CP, and CG included patients; CT was responsible for coordinating cytogenetics; CP was responsi-

ble for coordinating molecular biology; SH and AR performed molecular biology; HD, president of the ALFA group, included patients, reviewed the manuscript,

and gave final approval.

The authors thank all ALFA investigators and most specially Mohamed Elhamri, PhD for collecting the data and providing technical support, and Quoc-Hung Le,

MD, PhD, for providing statistical advices.

Cancer April 1, 2010 1725

Original Article

randomized at baseline to receive or not GM-CSF duringTSC induction and each consolidation chemotherapy cycles.First published results were encouraging showing that pri-ming of leukemic cells with GM-CSF is a means of enhanc-ing the efficacy of chemotherapy in younger adults withAML.5 The effects of GM-CSF on event-free survival (EFS)are reported herein with a long-term follow-up and accord-ing to distinct biological subgroups.

MATERIALS AND METHODS

Patients and Treatments

A total of 259 patients aged 15-50 years with untreated denovo AML are included in the present GM-CSF study.This study was approved by the ethics committee of LyonB-Hotel Dieu Hospital (n�99017B). Informed consentwas obtained from each patient in accordance with theDeclaration of Helsinki. Treatment modalities have al-ready been detailed.5 After receiving a timed-sequentialinduction therapy, complete remitters without allogeneicstem cell transplantation indication and related donorwere randomized between the TSC ALFA-9000 consoli-dation6 or 4 courses of high-dose cytarabine, according tothe Cancer and Leukemia Group B (CALGB) schedule.7

All patients registered on the study were randomized atregistration to receive recombinant human GM-CSF(Leucomax, Schering Plough) given at a dose of 5 lg/kg/day intravenously over 6 hours, from Day 1 through thelast day of chemotherapy of each course (except salvage)of chemotherapy or no GM-CSF.

Cytogenetics, Mutations Detection,and Risk Classification

Karyotypes were classified into 3 categories (favorable, in-termediate, and unfavorable) according to the MedicalResearch Council (MRC) classification.8 The nucleo-phosmin mutations (NPM1), CCAAT/enhancer-bindingprotein-amutations (CEBPA), fms-like tyrosine kinase-3internal duplications (FLT3-ITD) and mutations(FLT3m), mixed-lineage leukemia gene partial tandemduplications (MLL/PTD), and Wilms tumor gene (WT1)mutations were detected as previously described.9-13 Inter-mediate-risk cytogenetics was further subdivided into agood risk-2 group (normal karyotypes with NPM1 orCEBPA mutation and no FLT3-ITD) and a poor risk-2group (other patients).

Definition of Clinical Endpoints

CR was defined as a normocellular bone marrow contain-ing less than 5% blasts and showing evidence of normal

maturation of other marrow elements, and the disappear-ance of all clinical signs of leukemia;14 95% of casesdefined as CR achieved a neutrophil count of 1� 109/land a platelet count of 100� 109/l. Remission failurecomprised resistant disease and induction death. EFS wascalculated from the date of first randomization, with CRachievement failures, deaths during induction or in firstCR, and relapses included as events. Overall survival (OS)was the time from first random assignment to death. Dis-ease-free survival (DFS) was defined from date of CR todate of first relapse or death, or last contact with patient incontinuous CR.

Statistical analysis

Patient characteristics and CR rates comparisons wereperformed using the Pearson v2 test for binary variablesand the Mann-Whitney U test for continuous variables.DFS, EFS, and OS were estimated by the Kaplan-Meiermethod and compared using the log-rank test. Statisticalanalyses were performed according to an intent-to-treatbasis. Survival comparisons were adjusted for covariatesusing the Cox model and tested by the likelihood ratiotest. Simultaneous effects of multiple covariates were esti-mated with the maximum-likelihood logistic regressionmodel for response to therapy and with the Cox model forEFS, DFS, and OS and tested by the likelihood-ratio test,also used in univariate analyses for continuous variables.Estimated hazard ratios (HRs) are reported as relative risks(RRs) with 95% confidence intervals (CIs). All computa-tions were performed using BMDP software (BMDP Sta-tistical Software, Los Angeles, Calif, USA).

RESULTS

General outcome

In this multicenter randomized trial, 124 patients were ini-tially randomized in the GM-CSF arm and 135 in the noGM-CSF arm. The main biological and clinical features ofthe 259 patients are presented in Table 1. First resultsshow that priming of leukemia cells with GM-CSF mayenhance CR rate and EFS have been previously published.5

With a median follow-up of 5.4 years, the 5-yearEFS rate was confirmed to be better in patients primedwith GM-CSF (43% vs 34%; P¼ .04) (Fig. 1), while thedifference remained not significant in terms of OS (49%vs 42%; P¼ .21).

Biological Subgroups Analysis

The effects of GM-CSF are reported herein according todistinct biological entities. Two risk groups were defined

Original Article

1726 Cancer April 1, 2010

according to cytogenetics and molecular biologicalmarkers: 1) good risk groups including favorable cytoge-netics, constituted by core binding factor (CBF) leukemiasand the good risk-2 subset (normal karyotypes with favor-able genotypes); and 2) poor risk groups including thepoor risk-2 subset and unfavorable cytogenetics (Fig. 2).

Results of the comparison between the 2 study arms ofinitial randomization (GM-CSF vs no GM-CSF) accord-ing to biological subgroups are summarized in Table 2.

Priming did not improve the outcome in patientswith CBF leukemias (5-year EFS at 61% with priming vs62% without priming; P¼ .9). In the present studycohort, good risk-2 genotypes were associated with a riskprofile comparable to that of CBF leukemias (5-year EFSat 67%). Comportment after priming with GM-CSF wasalso comparable, showing no difference in terms of long-term EFS compared with absence of priming (Fig. 3A).

Figure 1. Event-free survival (EFS) in the whole cohort ofrandomized patients is depicted according to the assignedtreatment (GM-CSF vs no GM-CSF). In the Cox model, a value>1 indicates that the outcome is worse in that category com-pared with the baseline. P-value was given by the Wald’stest. GM-CSF indicates granulocyte-macrophage colony-stim-ulating factor.

Table 1. Characteristics of Patients According to the Arm ofInitial Randomization

Characteristics GM-CSFStudy Arm124 Patients

No GM-CSFStudy Arm135 Patients

Age, y 37 (15-50) 39 (15-50)

BiologyWBC count (x109/L) 16.0 (0.8-400) 11.4 (0.9-270)

Hb level (g/L) 87 (41-149) 90 (33-143)

Platelets (x109/L) 58 (2-1500) 55 (4-277)

PB blasts (%) 35 (0-98) 48 (0-99)

BM blasts (%) 67 (20-98) 74 (20-99)

Clinical presentation (%)Hepatomegaly 5 (4) 9 (7)

Splenomegaly 11 (9) 13 (10)

CNS1 2 (2) 2 (1)

Bleeding 25 (20) 35 (26)

Fever 48 (39) 52 (39)

FAB classification (%)M0 2 (2) 8 (6)

M1 23 (19) 28 (21)

M2 37 (30) 38 (28)

M4 22 (18) 22 (16)

M5) 24 (19) 25 (18)

M6 5 (4) 4 (3)

M7 3 (2) 1 (1)

ND 8 (6) 9 (7)

WHO PS (%)0 45 (36) 54 (40)

1 61 (49) 63 (47)

2 18 (15) 18 (13)

Cytogenetics (%)Favorable risk 26 (22) 21 (16)

Intermediate risk 65 (53) 76 (57)

Unfavorable risk 27 (22) 30 (23)

Failure 4 (3) 6 (4)

Not performed 2 2

Gene mutations (%)NPM1 17/77 (22) 18/82 (22)

CEBPA 3/78 (4) 5/82 (6)

FLT3-ITD 10/78 (13) 13/87 (15)

FLT3m 3/76 (4) 5/81 (6)

MLL 13/84 (15) 13/87 (15)

WT1 3/73 (4) 6/79 (7)

GM-CSF indicates granulocyte-macrophage colony-stimulating factor;

WBC, white blood cell; Hb, hemoglobin; PB, peripheral blood; BM, bone

marrow; CNS, central nervous system; þ, positive; FAB, French-American-

British; M, myeloid; ND, not determined; WHO, World Health Organization;

PS, performance status.

Figure 2. Risk classification is defined according to cytoge-netic-risk groups and results of molecular biology. In the Coxmodel, a value >1 indicates that the outcome is worse in thatcategory compared with the baseline. P-value was given bythe Wald test.

GM-CSF Priming in Adult AML/Thomas et al


Conversely, there was a trend for better results afterpriming among poor risk groups (Fig. 3B). Among thesepatients, the effect of GM-CSF became significant whenconsidering patients with initial white blood cell (WBC)count above the median (>10� 109/L; 5-year EFS, 38% vs12%; P¼ .02), while there was no difference in patientswith lower initial WBC count. This was confirmed whenconsidering patients with initial percentage of circulatingblasts above the median (>45%; 5-year EFS, 33% vs 10%;P¼ .02) (Fig. 4).

In addition, the difference in terms of EFS washighly significant in patients presenting either FLT3-ITDor MLL abnormalities. In FLT3-ITDþ patients, 5-yearEFS was 50% after GM-CSF priming versus only 8% forthose randomized in the arm without GM-CSF (P¼ .04).Patients presenting MLL abnormalities also benefitedfrom GM-CSF priming with 5-year EFS at 31% versus

only 8% for those who did not receive GM-CSF(P¼ .03). When combining patients expressing 1 of these2 molecular markers, the difference in terms of EFS washighly significant (5-year EFS, 39% with GM-CSF vs 8%without GM-CSF; P¼ .007) (Fig. 5).

Risk Factors for Response to Therapyand Survival

Table 3 provides the results of multivariate analyses forresponse to therapy and survival with relative risks and Pvalues. When considering age, WBC count, percentage ofcirculating blasts, arm of first randomization, risk group,and postremission therapy (for survival), priming withGM-CSF appeared as an independent favorable prognos-tic factor for both response to therapy and survival interms of EFS and DFS.

Table 2. Comparison Between the Two Arms of Initial Randomization (GM-CSF Versus No GM-CSF)According to Biological Subgroups

Subgroups % CR After TSC P % 5-Year EFS P % 5-Year OS P

Cytogenetic groupsa

Favorable

GM-CSF 96 .20 61 .90 72 .99

No GM-CSF 86 62 75

Intermediate

GM-CSF 86 .19 46 .19 52 .24

No GM-CSF 78 35 41

Unfavorable

GM-CSF 89 .04 16 .21 16 .62

No GM-CSF 67 10 20

Molecular biologyMLL abnormality

GM-CSF 92 .27 31 .03 31 .25

No GM-CSF 77 8 9

FLT3-ITD1

GM-CSF 70 .96 50 .04 60 .05

No GM-CSF 69 8 15

MLL1 or FLT3-ITD1

GM-CSF 83 .42 39 .007 43 .02

No GM-CSF 73 8 13

Risk groupsb

Good risk

GM-CSF 98 .18 63 .73 70 .27

No GM-CSF 90 64 80

Poor risk

GM-CSF 84 .04 32 .06 37 .27

No GM-CSF 72 24 30

All patients

GM-CSF 88 .04 43 .04 51 .21

No GM-CSF 78 34 42

GM-CSF indicates granulocyte macrophage colony-stimulating factor; CR, complete remission; TSC, timed-sequential chemotherapy; EFS, event-free survival;

OS, overall survival.a According to the Medical Research Council classification criteria; CBF, core binding factor.a Risk groups as defined above are good risk groups (CBF leukemias þ good risk-2), poor risk groups (poor risk-2 þ unfavorable cytogenetics).

Original Article


When considering only patients after the secondrandomization (ALFA-9000 postremission vs CALGB-like consolidation), priming with GM-CSF still showed atrend for better results in poor risk groups (5-year EFS,45% with GM-CSF vs 32% without GM-CSF; P¼ .05).No difference was noted between the 2 study arms of ini-tial randomization among patients receiving the CALGB-like consolidation (5-year EFS, 39% with GM-CSF vs39% without GM-CSF; P¼ .7), while a significant differ-ence was noted for those receiving the TSC consolidation(5-year EFS, 35% with GM-CSF vs 18% without GM-CSF; P¼ .02).

DISCUSSIONRecent studies have shown that sensitization of leukemiccells with G-CSF or GM-CSFmay enhance the cytotoxic-ity of chemotherapy in younger adults with newly diag-nosed AML,5,15 while other published studies, especiallyin elderly AML patients, have failed to show enhancedtherapeutic efficacy of priming.16 The reasons for these

Figure 3. Event-free survival (EFS) according to the assignedtreatment (GM-CSF vs no GM-CSF) and the risk groups aredepicted. (A) Good risk groups comprise patients with CBFleukemias or with good risk-2 genetics; (B) poor risk groupscomprise patients with poor risk-2 genetics or unfavorablecytogenetics. In the Cox model, a value >1 indicates that theoutcome is worse in that category compared with the base-line. P-value was given by the Wald test. GM-CSF indicatesgranulocyte-macrophage colony-stimulating factor.

Figure 4. Shown is event-free survival (EFS) according to theassigned treatment (GM-CSF vs no GM-CSF) in patients frompoor risk groups (unfavorable-risk cytogenetics or poor risk-2 genetics) with initial circulating blasts �45%. In the Coxmodel, a value >1 indicates that the outcome is worse in thatcategory compared with the baseline. P-value was given bythe Wald test. GM-CSF indicates granulocyte-macrophagecolony-stimulating factor.

Figure 5. Shown is event-free survival (EFS) according to theassigned treatment (GM-CSF vs no GM-CSF) in patients withFLT3/ITD or MLL abnormalities at diagnosis. In the Coxmodel, a value >1 indicates that the outcome is worse in thatcategory as compared with the baseline. P-value was givenby the Wald test. GM-CSF indicates granulocyte-macrophagecolony-stimulating factor; FLT3/ITD, fms-like tyrosine kinase-3 internal duplications; MLL, mixed-lineage leukemia.



discrepant results have not been resolved but informationregarding the distinct biological entities of leukemia hasnot been available in those studies. Cytogenetics is cur-rently regarded as the most important prognostic factor inAML and was, therefore, herein taken into account forstudying the effects of priming with GM-CSF. However,research in defining prognostic factors has recently movedto an examination of molecular markers.17,18 Molecularprofiling develop models offering improvements in riskstratification and design novel approaches to treat patientsmore effectively. Molecular markers are particularly im-portant in patients displaying a normal karyotype.17 Thedata presented herein confirm the earlier studies showingthat 2 genotypes, NPM1þ FLT3-ITD� and CEBPAþFLT3-ITD� (which were classified here as good risk-2),displayed a relatively favorable outcome. This allowed usto split the intermediate cytogenetic groups into 2 differ-ent risk populations and to restructure our global risk clas-sification by grouping CBF leukemias and good risk-2leukemias in a good risk group, and poor risk-2 leukemias

and unfavorable cytogenetics in a poor risk group. Thus,when looking at the effect of GM-CSF, not all but onlycertain subsets of patients seem to benefit from priming.The subpopulation of younger adult patients from thepoor risk group seemed to benefit most from CSF pri-ming with respect of EFS, while limited or no effect wasobserved among patients from the good risk group. How-ever, other factors seem to be of importance for hemato-poietic growth factor priming and should be taken intoaccount. The schedule of administration seems importantwith a better benefit when repeating priming during allconsolidation chemotherapy courses, instead of a uniquesequence of administration during induction chemother-apy.16 It appeared in our study that there was a potentia-tion of GM-CSF priming with the theoretical effect ofTSC on cell cycle more than only on the cytotoxic efficacyof cytarabine. Indeed, a significant difference in terms ofCR was initially observed after the first course of induc-tion chemotherapy,5 and a significant difference betweenGM-CSF and no GM-CSF was also noted in terms ofEFS among patients receiving the ALFA-9000 TSC con-solidation course, while there were no differences for thoserandomized in the CALGB-like, high-dose cytarabineconsolidation study arm. These results also suggest a bet-ter efficacy in absence of priming of high-dose cytarabineas consolidation therapy in younger adults with AML.The type of growth factor may also influence the results ofpriming. Although priming with other hematopoieticgrowth factors such as G-CSF have shown efficacy,15 thismay not be identical in terms of efficacy to priming withGM-CSF that is, at least theoretically, more efficient byrecruiting earlier leukemia progenitors and potentiallystem cells instead of only bulk leukemic cells.19 AlthoughG-CSF stimulates the development of granulocyte colo-nies, GM-CSF stimulates the proliferation and differen-tiation of neutrophils and eosinophils and, at higherdoses, some megakaryocyte colonies. However, G-CSFstimulates very immature progenitors, and it is knownthat G-CSF cleaves SDF1.20 The same has not beendescribed with GM-CSF, although the capacity ofCXCR4 antagonists to promote mobilization of bonemarrow-derived leukocytes was demonstrated especiallywhen given in concert with GM-CSF.21

Although a better recruitment of leukemic cells hasbeen hypothesized, real mechanisms involved in the bene-ficial role of priming remain questionable. Recent studieshave started to elucidate the role of the bone marrowmicroenvironment in the pathogenesis of leukemias andhave demonstrated that leukemic cells settling in the

Table 3. Results of Multivariate Analysis for CR, EFS, DFS,and OSa

Outcome RR (95% CI) P

CR after TSCAll patients

Risk groupb 4.55 (1.54-13.50) .001

Arm of initial randomization 2.54 (1.21-5.33) .01

Poor risk groups


EFSAll patients

Risk group 2.43 (1.37-4.29) .001


Poor risk groups


DFSAll patients

Risk group 2.40 (1.20-4.81) .004


Poor risk groups


OSAll patients

Risk group 3.31 (1.68-6.54) <.001

CR, complete remission; EFS, event-free survival; DFS, disease-free sur-

vival; OS, overall survival; CI, confidence interval; RR, relative risk; WBC,

white blood cell; CBF, core binding factor.a Factors included in the model were age, WBC count at diagnosis, per-

centage of circulating blasts, the arm of first randomization, risk group, and

postremission therapy when appropriate. Only factors showing significant

results were reported.a Risk groups as defined above are good risk groups (CBF leukemias þgood risk-2), poor risk groups (poor risk-2 þ unfavorable cytogenetics).

Original Article


niches of the microenvironment are relatively protected

against the cytotoxic effects of chemotherapy. Chemo-

therapy resistance may be provoked by adherence of the

leukemic cells to the stromal environment via specific re-

ceptor and adhesion molecules, such as the CXCR4-

SDF1a axis. AMLs with high CXCR4 cell surface expres-

sion, and with, therefore, a high tendency of stromal pro-

tection, have been shown to have a poor prognosis.22,23

Inhibition of CXCR4 has been able to overcome resist-

ance to numerous drugs.24,25 CXCR4 inhibitors that tar-

get the stromal interaction and release the leukemic cells

from the microenvironment have both a mobilizing and

cell-cycle activating effect upon leukemic cells and may

sensitize AML for chemotherapeutic cell killing.24,26

CXCR4 inhibitors induced mobilization of AML cells

into circulation and enhanced antileukemic effects of

chemotherapy, resulting in markedly reduced leukemia

burden and prolonged survival.25 These effects resemble

those of AML cell stimulation by G-CSF or GM-CSF,

and synergy between the activities of hematopoietic

growth factors (at least G-CSF) and CXCR4 inhibitors

has been reported.27-29 A mechanism of action for the

beneficial effect of priming with GM-CSF could, there-

fore, be mediated through CXCR4-induced mobilization.CXCR4 expression in AML is correlated with FLT3

gene mutation,23 and CXCR4 inhibition increased thesensitivity of FLT3-mutated leukemic cells to the apopto-genic effects of FLT3 inhibitors. Although the limitednumber of patients in each subgroups defined by molecu-lar markers might lead to a cautious interpretation of data,our results regarding the advantage of priming in patientsdisplaying FLT3-ITD mutation or MLL abnormalitiesare supporting these observations. FLT3 gene mutationsactivate CXCR4 signaling. Furthermore, FLT3-ITD isknown to provide a proliferative advantage to a cell popu-lation, in which normal maturation is blocked,30 and tocause synergistic expansion of hematopoietic stem cellswhen combined with growth factors.31,32 Similarly, MLLfusion proteins confer a remarkable growth/survivaladvantage.31,33 Although none of our patients presentedsimultaneously both anomalies, a positive correlation hasbeen previously reported between MLL intragenic abnor-malities and FLT3 mutations.33 FLT3 might be a thera-peutic target in MLL-rearranged leukemias, in whichmutations have been described in the activation loop ofFLT3 resulting in constitutive activation. Furthermore,FLT3 inhibitors has been shown to be differentially cyto-toxic to leukemia cells withMLL translocations.34

Relations between FLT3-ITD mutation or MLLabnormalities and leukemic cell mobilization are also sus-tained by correlations found in our study between a bene-ficial effect of priming and high initial level of WBCcount and circulation blasts. Almost all patients withFLT3-ITD and more than half of those with MLL abnor-malities initially presented with such features. The imme-diate value of these observations will be to test theirpotential application in further trials with the aim to treatpatients more effectively.

The results of priming in the ALFA-9802 trial con-firm that chemotherapy and sensitization of leukemic cellsand their progenitors by GM-CSF is a plausible strategyfor improving the outcome of younger AML patients.Even if based on retrospective subgroup analysis, theseresults seem to indicate that patients with nonfavorablecytogenetics or genotypes displaying genetic alterationsimplicated in cell mobilization might be those who couldbenefit from a priming approach. Of course, additionallong-term randomized studies are necessary before thestrategy of priming with CSFs could be recommended asstandard of care, eventually only in certain patient subsets.On the basis of in vitro studies and on the clinical experi-ence of the greater sensitivity to chemotherapy of circulat-ing as compared with bone marrow resident leukemiccells, it is expected that mobilization of leukemic stemcells with CSF and anti-CXCR4, accompanied by chemo-therapy, will result in increased antileukemic effect. Thus,the mobilization of leukemic stem cells is a concept that ispresently being revisited. In this setting, intergroup col-laboration will be necessary to answer relevant questions.

CONFLICT OF INTEREST DISCLOSURESSchering Plough (Kenilworth, NJ, USA) and Amgen (Neuillysur Seine, France) provided grants for central data managementto the Edouard Herriot Hospital (Department of Hematology),and Schering Plough provided molgramostim (Leucomax, GM-CSF) free of charge. This trial is registered at www.clinicaltrials.gov as no. NCT00880243.

REFERENCES1. Tanaka M. Recombinant GM-CSF modulates the metabo-

lism of cytosine arabinoside in leukemic cells in bone mar-row. Leuk Res. 1993;17:585-592.

2. Te Boekhorst PA, Lowenberg B, Vlastuin M, Sonneveld P.Enhanced chemosensitivity of clonogenic blasts frompatients with acute myeloid leukemia by G-CSF, IL-3 orGM-CSF stimulation. Leukemia. 1993;7:1191-1198.

3. Reuter C, Auf der Landwehr U, Schleyer U, et al. Modula-tion of intracellular metabolism in cytosine arabinoside inacute myeloid leukemia by granulocyte-macrophage colony-stimulating factor. Leukemia. 1994;8:217-225.



4. Thomas X, Fenaux P, Dombret H, et al. Granulocyte-mac-rophage colony-stimulating factor (GM-CSF) to increase ef-ficacy of intensive sequential chemotherapy with etoposide,mitoxantrone and cytarabine (EMA) in previously treatedacute myeloid leukemia: a multicenter randomized placebo-controlled trial (EMA 91 trial). Leukemia. 1999;13:1214-1220.

5. Thomas X, Raffoux E, Botton S, et al. Effect of primingwith granulocyte-macrophage colony-stimulating factor inyounger adults with newly diagnosed acute myeloid leuke-mia: a trial by the Acute Leukemia French Association(ALFA) Group. Leukemia. 2007;21:453-461.

6. Castaigne S, Chevret S, Archimbaud E, et al. Randomizedcomparison of double induction and timed-sequentialinduc-tion to a ‘‘3 þ 7’’ induction in adults with AML: long-termanalysis of the Acute Leukemia French Association (ALFA)9000 study. Blood. 2004;104:2467-2474.

7. Mayer RJ, Davis RB, Schiffer CA, et al. Intensive postremis-sion chemotherapy in adults with acute myeloid leukemia:Cancer and Leukemia Group B. N Engl J Med. 1994;331:896-903.

8. Grimwade D, Walker H, Oliver F, et al. The importance ofdiagnostic cytogenetics on outcome in AML: analysis of1,612 patients entered into the MRC AML 10 trial. TheMedical Research Council Adult and Children’s LeukaemiaWorking Parties. Blood. 1998;92:2322-2333.

9. Preudhomme C, Sagot C, Boissel N, et al. Favorable prog-nostic significance of CEBPA mutations in patients with denovo acute myeloid leukemia: a study from the Acute Leu-kemia French Association (ALFA). Blood. 2002;100:2717-2723.

10. Boissel N, Cayuela JM, Preudhomme C, et al. Prognosticsignificance of FLT3 internal tandem repeat in patients withde novo acute myeloid leukemia treated with reinforcedcourses of chemotherapy. Leukemia. 2002;16:1699-1704.

11. Boissel N, Renneville A, Biggio V, et al. Prevalence, clinicalprofile and prognosis of NPM mutations in AML with nor-mal karyotype. Blood. 2005;106:3618-3620.

12. Poirel H, Rack K, Delabesse E, et al. Incidence and charac-terization of MLL gene (11q23) rearrangements in acutemyeloid leukemia M1 and M5. Blood. 1996;87:2496-2505.

13. Summers K, Stevens J, Kakkas I, et al. Wilms’ tumour 1mutations are associated with FLT3-ITD and failure of stand-ard induction chemotherapy in patients with normal karyo-type AML. Leukemia. 2007;21:550-551; author reply, 552.

14. Cheson BD, Cassileth PA, Head DR, et al. Internationalworking group for diagnosis, standardization of response cri-teria, treatment outcomes, and reporting standards for thera-peutic trials in acute myeloid leukemia. J Clin Oncol. 2003;21:4642-4649.

15. Lowenberg B, van Putten W, Theobald M, et al. Effect ofpriming with granulocyte colony-stimulating factor on theoutcome of chemotherapy for acute myeloid leukemia. NEngl J Med. 2003;349:743-752.

16. Amadori S, Suciu S, Jehn U, et al. Use of glycosylatedhuman recombinant G-CSF (lenograstim) during and/or af-ter induction chemotherapy in patients 61 years of age andolder with acute myeloid leukemia: final results of AML-13,a randomized phase-3 study. Blood. 2005;106:27-34.

17. Schlenk RF, Dohner K, Krauter J, et al. Mutations andtreatment outcome in cytogenetically normal acute myeloidleukemia. N Engl J Med. 2008;358:1909-1918.

18. Renneville A, Roumier C, Biggio V, et al. Cooperating genemutations in acute myeloid leukemia: a review of the litera-ture. Leukemia. 2008;22:915-931.

19. Grant SM, Neel RC. Recombinant granulocyte-macrophagecolony-stimulating factor (rGM-CSF). Drugs. 1992;43:516-561.

20. Carter BZ, Miella M, Altieri DC, Andreeff M. Cytokine-regulated expression of survivin in myeloid leukemia. Blood.2001;97:2784-2790.

21. Hu JS, Freeman CM, Stolberg VR, et al. AMD3465, anovel CXCR4 receptor antagonist, abrogates Schistosomalantigen-elicited (type-2) pulmonary granuloma formation.Am J Pathol. 2006;169:424-432.

22. Rombouts EJC, Pavic B, Lowenberg B, Ploemacher R. Rela-tion between CXCR-4 expression, Flt3 mutations andunfavorable prognosis of adult acute myeloid leukemia.Blood. 2004;104:550-557.

23. Konoplev S, Rassidakis GZ, Estey E, et al. Overexpressionof CXCR4 predicts adverse overall and event-free survival inpatients with unmutated FLT3 acute myeloid leukemia withnormal karyotype. Cancer. 2007;109:1152-1156.

24. Zeng Z, Samudio IJ, Munsell M, et al. Inhibition ofCXCR4 with the novel RCP168 peptide overcomes stroma-mediated chemoresistance in chronic and acute leukemias.Mol Cancer Ther. 2006;5:3113-3121.

25. Zeng Z, Shi YX, Samudio IJ, et al. Targeting the leukemiamicroenvironment by CXCR4 inhibition overcomes resist-ance to kinase inhibitors and chemotherapy in AML. Blood.2009;113:6215-6224.

26. Liesveld JL, Bechelli J, Rosell K, et al. Effects of AMD3100on transmigration and survival of acute myelogenous leuke-mia cells. Leuk Res. 2007;31:1553-1563.

27. Flomenberg N, Devine SM, Dipersio JF, et al. The use ofAMD3100 plus G-CSF for autologous hematopoietic pro-genitor cell mobilization is superior to G-CSF alone. Blood.2005;106:1867-1874.

28. Andreeff M, Konoplev S, Wang RY, et al. Massive mobili-zation of AML cells into circulation by disruption of leuke-mia/stroma cell interactions using CXCR4 antagonistAMD3100: First evidence in patients and potential for abol-ishing bone marrow microenvironment-mediated resistance[abstract]. Blood. 2006;108(suppl 1):171a. Abstract 568.

29. Fiegl M, Samudio I, Clise-Dwyer K, et al. CXCR4 expres-sion and biological activity in acute myeloid leukemia aredependent on oxygen partial pressure. Blood. 2009;113:1504-1512.

30. Yanada M, Matsuo K, Suzuki T, et al. Prognostic signifi-cance of FLT3 internal tandem duplication and tyrosine ki-nase domain mutations for acute myeloid leukemia: a meta-analysis. Leukemia. 2005;19:1345-1349.

31. Gilliland DG. Molecular genetics of human leukemias: Newinsights into therapy. Semin Hematol. 2002;39:6-11.

32. Dash A, Gilliland DG. Molecular genetics of acute myeloidleukaemia. Best Pract Clin Haematol. 2001;14:49-64.

33. Libura M, Asnafi V, Tu A, et al. FLT3 and MLL intragenicabnormalities in AML reflect a common category of geno-toxic stress. Blood. 2003;102:2198-2204.

34. Armstrong SA, Kung AL, Mabon ME, et al. Inhibition ofFLT3 in MLL. Validation of a therapeutic target identifiedby gene expression based classification. Cancer Cell. 2003;3:173-183.

Original Article


which aml subsets benefit from leukemic cell priming during chemotherapy? long-term analysis of the...

Documents