AAML0523: A report from the Children's Oncology Group on the efficacy of clofarabine in combination with cytarabine in pediatric patients with recurrent acute myeloid leukemia




The discovery of new, effective non–anthracycline-based reinduction regimens for children with recurrent acute myeloid leukemia (AML) is critical. In this phase 1/2 study, the tolerability and overall response rate of clofarabine in combination with cytarabine was investigated in children with recurrent/refractory AML.


AAML0523 enrolled 49 children with AML in first recurrence or who were refractory to induction therapy. The study consisted of a dose-finding phase (9 patients) and an efficacy phase (40 patients). Two children received clofarabine at a dose of 40 mg/m2/day and 47 children at a dose of 52 mg/m2/day.


Toxicities typical for intensive chemotherapy regimens were observed at all doses of clofarabine. The recommended pediatric phase 2 dose of clofarabine in combination with cytarabine was 52 mg/m2/day for 5 days. Of 48 evaluable patients, the overall response rate (complete remission plus complete remission with partial platelet recovery) was 48%. Four patients met conventional criteria for complete remission with incomplete count recovery. Twenty-one of 23 responders subsequently underwent hematopoietic stem cell transplantation. The overall survival rate at 3 years was 46% for responders compared with 16% for nonresponders (P < .001). Patients found to have no minimal residual disease at the end of the first cycle by flow cytometric analysis had superior overall survival after 1 year (100% vs 38%; P = .01).


The combination of clofarabine and cytarabine yielded an acceptable response rate without excess toxicity in children with recurrent AML. The nearly 50% survival rate reported in responders is highly encouraging in these high-risk patients and suggests that this combination is an effective bridge to hematopoietic stem cell transplantation. Cancer 2014;120:2482–2489. © 2014 American Cancer Society.


Cure rates for children with acute myeloid leukemia (AML) have dramatically improved with the advent of anthracycline-based chemotherapy regimens and improvements in supportive care.[1, 2] Nevertheless, approximately 50% of children still develop disease recurrence with resistant disease. It is critical to develop effective non–anthracycline-based reinduction regimens to avoid further risk of cardiotoxicity while allowing for curative hematopoietic stem cell transplantation (HSCT). Clofarabine is a second-generation purine nucleoside analog designed to integrate the mechanistic properties of fludarabine and cladribine.[3] Clofarabine requires intracellular phosphorylation by deoxycytidine kinase to the active triphosphate form (clo-CTP) before inhibition of DNA polymerase.[4, 5] Clofarabine has shown single-agent activity in phase 1 and 2 studies in pediatric patients with recurrent or refractory acute lymphoblastic leukemia (ALL) and AML.[6, 7]

Clofarabine is a potent inhibitor of ribonucleotide reductase, leading to increased accumulation of the cytotoxic triphosphate form of cytarabine (ara-CTP) in leukemia cells. This biochemical modulation is well established in vitro and has been studied in clinical trials in adults with AML.[8-10]

In the current study, we have described the findings of the AML stratum of the Children's Oncology Group (COG) phase 1/2 study AAML0523. The primary objective of this study was to define the overall response rate (ORR) to clofarabine in combination with cytarabine in children with recurrent or refractory AML or ALL. The current study reports only the results for children with recurrent AML; the findings for children with ALL have been published previously.[11]



AAML0523 was open to accrual between March 12, 2007 and January 5, 2012. Data analyses for patients with AML were current as of September 30, 2012. Patients with AML were required to be aged 1 to 21 years, and in first recurrence or with disease that was refractory to reinduction. Patients were required to have histologically proven AML according to the French-American-British classification system and ≥ 5% bone marrow blasts. Other requirements included adequate liver (serum bilirubin ≤ 1.5 times the upper limit of normal [ULN] for age, alanine aminotransferase ≤ 2.5 times the ULN for age), renal (derived from the Schwartz formula), cardiac (echocardiogram with a shortening fraction ≥ 27%), and pancreatic function (serum amylase and lipase ≤ 1.5 times the ULN), and adequate performance status (Karnofsky or Lansky ≥ 50%). Exclusion criteria included uncontrolled systemic infection and active central nervous system involvement. Due to the severe hepatotoxicity observed in a concurrent pediatric clinical trial in patients with recurrent ALL receiving treatment with concurrent clofarabine, cyclophosphamide, and etoposide,[12] AAML0523 was amended to exclude patients who had undergone HSCT within 12 months of study entry. Institutional Review Boards at participating centers approved the study, and participating patients or their parents provided written informed consent. The original clinical trial was registered at www.clinicaltrials.gov as NCT00372619.

Study Design

The study was conducted in 2 phases: a dose-finding phase and an efficacy phase. The dose-finding phase consisted of a single dose escalation/deescalation of clofarabine in combination with a fixed dose of cytarabine (1 g/m2/day for 5 days). For each dose level, 10 patients (with both ALL and AML) were enrolled. The first cohort received clofarabine at a dose of 40 mg/m2/day for 5 days. Based on safety data from the first cohort, the dose of clofarabine would either be escalated to 52 mg/m2/day or deescalated to a dose of 30 mg/m2/day. The recommended phase 2 dose (RP2D) was used in the efficacy portion of the study with separate ALL and AML cohorts. An optimal 2-stage Simon design was implemented to test the null hypothesis that the ORR (complete remission [CR] plus complete remission with partial platelet recovery [CRp]) is ≤ 40% versus the alternative hypothesis that the ORR is ≥ 60% based on the ORR from the Children's Cancer Group 2951 study.[13] Eighteen patients were to be enrolled in the first stage; if at least 8 overall responses were observed, an additional 28 patients would be enrolled in the second stage. The null hypothesis would be rejected if at least 23 overall responses were observed in patients treated at a dose of 52 mg/m2/dose. Patients who received therapy at the RP2D in the first phase were included in the efficacy phase.

Treatment Plan

Induction therapy consisted of up to 2 cycles. If a bone marrow aspirate (BMA) performed between day 14 and day 21 of the first cycle revealed ≥ 5% blasts, cycle 2 was administered without waiting for count recovery. BMA was repeated at least once every 14 days, until response assessment was possible. If the day 14 to day 21 BMA revealed < 5% blasts, patients received therapy once they attained adequate peripheral blood count recovery, defined as an absolute neutrophil count(ANC) > 1000/μL and a platelet count > 100,000/μL. Patients without ANC recovery by day 42 were to proceed to cycle 2 if there was no bone marrow aplasia noted. In the dose-finding phase, cycle 1 consisted of clofarabine on days 2 to 6 and cytarabine on days 1 to 5 for correlative study purposes. Correlative biology studies will be described in a separate article. In the efficacy phase of the study, both clofarabine and cytarabine were administered intravenously (iv) over 2 hours on days 1 to 5. Cytarabine was administered beginning 4 hours after the initiation of clofarabine. Because systemic inflammatory response is a known side effect of clofarabine,[6, 7] patients experiencing respiratory distress, unexplained hypotension, or tachycardia were to receive iv hydrocortisone pretreatment (50-100 mg/m2/day) before each clofarabine dose for remainder of that cycle. Prophylactic intrathecal cytarabine was administered to all patients at the time of diagnostic lumbar puncture or on day 0 of cycle 1 (at least 24 hours before administration of iv cytarabine). On the first day of cycle 2, patients were allowed to receive intrathecal cytarabine dosed according to age, at the discretion of the treating physician.

Response Criteria

The ORR consisted of patients who achieved a CR and a CRp. Criteria did not include those with CR with incomplete hematologic recovery (CRi). Response was considered as the best response after up to 2 induction cycles. CR was defined as attainment of an M1 bone marrow (< 5% blasts) with an ANC of > 1000/μL and a platelet count of > 100,000/μL. CRp was defined as attainment of an M1 bone marrow (< 5% blasts) with recovery of an ANC > 1000/μL and platelet transfusion independence, which was defined as no platelet transfusions for 1 week. Stable disease (SD) was defined as being present when a patient did not qualify for either a CR, CRp, or progressive disease (PD). PD was defined as an increase in the extent of bone marrow infiltration by leukemic cells of ≥ 20% blasts or extramedullary disease.

Detection of Minimal Residual Disease

Response to initial therapy was assessed by multidimensional flow cytometry (MDF) as previously described.[14] Samples were to be collected at the time of study entry, at the end of induction 1, and at the end of induction 2. The detection threshold of residual disease has been demonstrated to be 0.01%.[14]

Statistical Analysis

Data were analyzed through September 30, 2012. If therapy was associated with a 40% ORR after up to 2 cycles of therapy, the null hypothesis (40% ORR) would be rejected in the 2-stage design with a probability of .10. If therapy was associated with a 60% ORR, the null hypothesis would be rejected in the 2-stage design with a probability of .90. The Kaplan-Meier method was used to estimate overall survival (OS), which was defined as time from study entry until death, and the log-rank test was used to compare OS. Patients who were alive at the time of last follow-up were censored for OS analyses at the date of last contact. Patients defined as responders (best response of CR or CRp) were compared with nonresponders. The significance of observed difference in percentages was tested using the chi-square test and Fisher exact test when data were sparse comparing groups of patients. The Kruskal-Wallis test was used to determine the significance between differences in medians of groups.


Study Population

As outlined in Table 1, a total of 51 patients were enrolled on AAML0523, 49 of whom were eligible for response assessment. Of the 49 eligible patients, 44 were in first morphologic recurrence and 5 had disease that was refractory to induction chemotherapy. Patients in first recurrence had a median length of initial CR of 306 days (range, 55 days-2212 days), with 27 patients (61%) having developed disease recurrence within 1 year of achieving their initial CR. Of 40 patients with cytogenetic data, 3 (7.5%) had high-risk cytogenetics (monosomy 7 and del(5q0)), 11 patients had core-binding factor AML (27.5%), 4 had 11q23 abnormalities (10%), and 4 patients had extra copies of chromosome 8 (10%).

Table 1. Patient Characteristics
 AAML0523 (AML)AAML0523 (40 mg/m2)AAML0523 (52 mg/m2) 
  1. Abbreviations: AML, acute myeloid leukemia; CR1, first complete remission; HSCT, hematopoietic stem cell transplant; WBC, white blood cell.

No. of patients with AML enrolled51 24%4996% 
Ineligible2 0 2  
Black or African American613%00%614%1.000
American Indian or Alaska Native12%150%00%.044
Unknown4 0 4  
Patients in first recurrence4490%2100%4289%1.000
Refractory to reinduction5100% 5100% 
Prior HSCT48%00%49%1.000
Median age at time of study entry (range), y14.11.4-
Median WBC count (range), ×103/µL3.30.5-14001.651.5-
Median length of CR1 (range), d (n=44)30635-2212361.5213-510)30635-2212.955
Monosomy 713%00%13%1.000
Unknown9 0 9  


Nonhematologic toxicities of grade 3 and higher as defined in the National Cancer Institute's Common Terminology Criteria for Adverse Events (version 3) were collected for all patients. Those toxicities occurring in > 1 individual at clofarabine doses of 40 mg/m2 and 52 mg/m2 are included in Table 2. Toxicity data regarding 20 patients (including 2 patients with AML who were treated at a dose of 40 mg/m2 and 7 patients with AML who were treated at a dose of 52 mg/m2) treated in the dose-finding portion of the study have been published previously.[11] There were no dose-limiting toxicities (DLTs) for the 2 patients with AML who were treated at a dose of 40 mg/m2, whereas 1 patient with AML who was treated at a dose of 52 mg/m2 experienced a DLT (bone marrow aplasia, grade 4 hypokalemia, grade 3 nausea, and grade 3 dehydration). Therefore, the recommended dose of clofarabine for the efficacy portion (RP2D) of AAML0523 was 52 mg/m2. The most common toxicities were infection, nausea/diarrhea/anorexia, fever/neutropenia, transaminitis, hyperglycemia, and hypokalemia, which are consistent with prior studies on clofarabine. No deaths occurred among patients with AML within 30 days of protocol therapy.

Table 2. Toxicities of Grade 3 or Higher Observed in >10% of Patients During Induction Cycles 1 and 3a
 Induction, Cycle 1 (N=49)Induction, Cycle 2 (N=31)
Adverse Event ≥Grade 3No.%No.%
  1. a

    Toxicities were graded according to the National Cancer Institute's Common Terminology Criteria for Adverse Events (version 3).

Febrile neutropenia2041%929%
Infections and infestations2347%1342%
Alanine aminotransferase increased612%619%
Aspartate aminotransferase increased48%310%

Of 47 patients with AML treated at the RP2D, 12 had ANC recovery (> 1000/μL) and 16 had platelet recovery (≥ 100,000/μL) after cycle 1 at a median of 29.5 days and 26.5 days, respectively. Of those who received a second cycle, the median times to ANC recovery (15 patients) and platelet recovery (16 patients) were 35 days and 27.5 days, respectively. The remainder of the patients who did not have ANC recovery (35 patients) or platelet recovery (31 patients) after 1 cycle of therapy had median cycle lengths of 23 days and 29 days, respectively. Of 30 patients who received cycle 2 at the RP2D, 18 patients did not have ANC recovery (4 of whom achieved SD), platelet recovery (3 of whom achieved CRp), or both (6 with PD and 5 with SD). Of these 18 patients, 7 received more chemotherapy and HSCT, 4 underwent HSCT only, 6 received chemotherapy only, and 1 did not receive additional therapy.


Table 3 shows the response of eligible patients at the RP2D. A sufficient number of responses were observed at the RP2D in the first stage of the 2-stage design to proceed to the second stage. A total of 46 evaluable patients were treated at the RP2D, 16 of whom achieved CR and 5 of whom achieved a CRp for an ORR of 45.7% (95% confidence interval, 30.9%-61.0%). The 2 patients with AML who were treated on the dose-finding portion of the study at a dose of 40 mg/m2 responded (1 with CR and 1 with CRp), thus increasing the response rate for all patients with AML to 48%. Of 23 responders, 11 (48%) had SD after the first course and then achieved CR or CRp after cycle 2. However, 6 patients were removed from protocol therapy by their physician after the first cycle after achieving SD. Not included among responders were 4 patients with SD who achieved CRi, defined as having fulfilled criteria for bone marrow CR but without recovery of their ANC (< 1.0 × 109/L) and platelet count (< 100 × 109/L). All 4 patients with CRi proceeded to undergo HSCT and only 1 patient was alive at the time of last follow-up. There were 30 patients who received cycle 2 at the RPD2. Fifteen patients received a bone marrow evaluation between days 14 and 21 and 16 patients at day 23 or later. Response by length of first CR (CR1) is reported in Table 4 and is consistent with previous studies demonstrating that a CR1 of < 1 year predicts poor response.[15]

Table 3. Overall Response for Eligible Patients at Dose Level of 52 mg/m2a
  1. Abbreviations: CR, complete remission; CRp, complete remission with partial platelet recovery; PD, progressive disease; SD, stable disease.

  2. a

    One patient not evaluable for response because therapy was withdrawn before response assessment.

Course 1 response (N=47)  
Not evaluable12%
Course 2 response (N=30)  
Best response  
Not evaluable1a2%
Table 4. Overall Response for Eligible Patients at Dose Level of 52 mg/m2 (Patients in First Recurrence Only, N=42) by Duration of CR1
 Duration of CR1 ≤365 DaysDuration of CR1 >365 Days 
  1. Abbreviations: CR, complete remission; CR1, first complete remission; CRp, complete remission with partial platelet recovery; PD, progressive disease; SD, stable disease.

Course 1 response (N=42)     
Not evaluable00%00%
Course 2 response (N=29)     
Best response     
Not evaluable00%00%

Characteristics of Responders

The clinical features of all responders are listed in Table 5. Responders generally had lower-risk features including a longer median duration of CR1 (374.5 days vs 258 days; P = .085) and a higher percentage of favorable cytogenetics (40% vs 15%; P = .077). There were responses noted in higher-risk patients, because 4 of 12 patients with an initial CR of < 6 months achieved CR (1 patient) or CRp (3 patients). In addition, there were responses in a patient with monosomy 7, and 2 with 11q23 rearrangement. Of 23 responders, 21 proceeded to HSCT (10 after the first cycle and 11 after 2 cycles) compared with 12 of 25 nonresponders. For these patients, responses were found to be durable, with a 3-year OS rate of 46% ± 27% compared with 16% ± 16% in nonresponders. (P < .001) (Fig. 1).

Table 5. Characteristics of Responders and Nonresponders Among Eligible Patientsa
 Responders(CR + CRp)Nonresponders
  1. Abbreviations: CR, complete remission; CR1, first complete remission; CRp, complete remission with partial recovery of platelet count; HSCT, hematopoietic stem cell transplant; WBC, white blood cell.

  2. a

    One patient was not evaluable for response because therapy was withdrawn before response assessment.

Patients in first recurrence2296%2288%
Patient refractory to induction14%312%
Patients with prior HSCT29%28%
Cytogenetics (N=40)    
Monosomy 715%00%
Median age at study entry (range), y14.71.4-
Median WBC count (range), ×103/µL2.71.1-
Median length of CR1 (range), d (n=44)374.542-221225835-1066
Received HSCT in follow-up2191%1248%
No. of received courses to best response    
1 cycle1252%1144%
2 cycles1148%1456%
Figure 1.

Overall survival from the time of study entry for all responding patients with acute myeloid leukemia versus nonresponders on AAML0523. SE indicates standard error.

Minimal Residual Disease

Of 46 evaluable patients at the RP2D, specimens for MRD evaluation after the end of induction 1 (EOI1) were submitted for 13 patients, 3 of whom were in morphologic CR/CRp and 10 of whom had SD or PD. In these 13 patients, 8 had disease that was detectable by MDF after the first course (62%) (Table 6). Six of 8 patients who were positive for MRD at EOI1 had died at the time of last follow-up, whereas all the patients without MRD were alive. Of 10 patients with morphologic SD or PD at EOI1, 3 were negative for MRD, all of whom were long-term survivors compared with only 1 of 7 patients found to have MRD (14%). This suggests that MDF may provide clinically relevant data beyond morphology. MDF data at the EOI2 were available for 10 patients with an MRD rate of 40%. All patients without evidence of MRD at the EOI2 were long-term survivors, whereas 2 of the 4 patients with MRD at EOI2 had died at the time of last follow-up.

Table 6. MRD Status for the Evaluable Patients at the Dose Level of 52 mg/m2a
Patient No.EOI1 ResponseEOI1 Bone Marrow Blast, %MRD at EOI1EOI2 ResponseEOI2 Bone Marrow Blast, %MRD at EOI2Best ResponseSurvival Status
  1. Abbreviations: CR, complete remission; CRp, complete remission with partial platelet recovery; EOI1, end of induction 1; EOI2, end of induction 2; MRD, minimal residual disease; NA, not available; PD, progressive disease; SD, stable disease.

  2. a

    Patients 3 and 8 had MRD measured at EOI2 only.



Cure rates in children with AML in first recurrence are poor, especially among those who develop disease recurrence within 1 year of therapy.[13, 16] An important factor directing therapy after disease recurrence is prior anthracycline exposure, because the majority of children treated on current COG AML protocols have a cumulative anthracycline exposure of approximately 440 mg/m2. Emerging data have revealed a significant risk of cardiovascular death in pediatric survivors of AML.[17, 18] Therefore, effective non–anthracycline-based reinduction regimens are desperately needed to decrease treatment-related morbidities.

AAML0523 demonstrated that the combination of clofarabine and cytarabine is an active non–anthracycline-based chemotherapy regimen for children with AML in first recurrence. The combination proved to be an effective bridge to more definitive therapy, with 21 of 23 responders proceeding to HSCT. The significance of this “bridge to transplant” is demonstrated by the 3-year OS rate of 46% ± 27% for responders versus 16% ± 16% for nonresponders (P < .001) (Fig. 1). Although the ORR in the current study (45.7% at RP2D, 48% overall) was suboptimal, nonadherence to protocol recommendations may have affected the results. Because single-agent clofarabine trials have demonstrated that many patients required > 1 cycle to achieve best response,[6, 7] the protocol recommended that patients receive 2 cycles of protocol therapy. Of those patients in compliance with this recommendation, 11 (48% of responders) with SD after the first cycle achieved CR or CRp after cycle 2. However, 6 patients (29%) with SD after the first course were removed from protocol therapy by treating physicians without receiving a second cycle. In addition, 4 patients met conventional criteria for CRi, but came off the study before meeting parameters for CR or CRp. These 4 patients proceeded to undergo HSCT but only 25% were alive at the time of last follow-up. It is possible that improved protocol compliance may have resulted in an improved response rate. It is interesting to note that 2 patients with acute leukemia of ambiguous lineage were treated on study and both responded. The toxicity profile of this combination was consistent with other cytotoxic AML reinduction regimens, without unexpected grade 3 or 4 toxicities. After 2 DLTs identified as fungal infections in the phase 1 portion of the study (both in patients with ALL), the supportive care guidelines were amended to include antifungal prophylaxis. Venoocclusive disease, a concern with the combination of clofarabine, etoposide, and cyclophosphamide, was not observed on this study. Results from Children's Cancer Group 2951 study demonstrated that the combination of mitoxantrone and cytarabine in children with recurrent/refractory AML achieved an ORR of 58% (unpublished data).[13] Patients generally received less-intensive therapy and fewer cumulative anthracyclines compared with those treated on AAML0523. In addition, the majority of patients on AAML0523 had already received the combination of mitoxantrone and cytarabine, and therefore the response rate in the current study is not surprising. The combination of fludarabine and cytarabine has recently demonstrated efficacy in patients with recurrent pediatric AML.[15] Kaspers et al recently published their results of a randomized study of fludarabine, cytarabine, and granulocyte–colony-stimulating factor (FLAG) versus FLAG with liposomal daunorubicin.[15] Randomizing 394 patients aged > 10 years who developed disease recurrence after a variety of de novo therapeutic regimens, the CR rate (ANC > 1000/μL and a platelet count > 50,000/μL) after 2 cycles was 69% with the combination of FLAG and liposomal daunorubicin, and 59% with FLAG alone (P = .07). In this trial, every responder received 2 cycles whereas many patients on AAML0523 did not, a finding that reemphasizes an important potential confounding factor in the response rates of patients in the current study to the combination of clofarabine and cytarabine.

The results of the current study demonstrate that the quality of remission achieved by salvage chemotherapy, as defined by MDF, impacts long-term survival. The use of MDF for the assessment of response is well-established in predicting outcome in children with de novo AML.[14, 19, 20] In COG study AAML03P1, nearly one-third of patients in morphologic CR were found to have residual disease by MDF and had a significantly higher rate of disease recurrence. Conversely, this same study revealed that nearly 25% of the patients in morphologic failure (> 5% blasts) had no evidence of disease by MDF. This lack of immunophenotypic disease correlated with favorable outcome, highlighting the difficulty in distinguishing normal regenerating cells from leukemic blasts by morphology alone.[14] Given the paucity of available data, the usefulness of MRD by MDF in the setting of recurrent AML is unclear. All submitted samples were evaluated for MRD. However, the number of optional biology samples was limited because the majority of patients had received a bone marrow examination to confirm disease recurrence before providing consent for trial participation. There were no significant differences noted in response when comparing those who received MRD testing with those who did not. Despite the limited sample size, there was a statistically significant difference in 1-year OS from the end of the first cycle between patients who were negative for MRD (100%) versus patients who were positive for MRD (38%) (P = 0.01). This suggests that MDF may provide a valuable tool in response assessment in patients with recurrent disease, and further supports the need for accurate and sensitive response assessment beyond morphologic response in the setting of retrieval regimens. This will allow for improved patient care and facilitate the more rapid evaluation of new agents under study based on MDF data as a surrogate marker.

In conclusion, the results of the current study demonstrated that the non–anthracycline-based combination of clofarabine and cytarabine has clinically significant activity in pediatric patients with refractory/recurrent AML. The durability of these responses and the relatively high 3-year OS rate of 46% indicate that this combination is an effective therapy as a bridge to HSCT in this patient population.


Supported by National Institutes of Health (NIH) Children's Oncology Group Chair's grant NIH U10 CA98543 and SDC U10 CA98413. Clofarabine was supplied by Genzyme Oncology/Sanofi.


In Dr. Whitlock's role as a leader in drug development for pediatric acute leukemias, he has served on several advisory boards and engaged in other consulting relationships with Genzyme and several other pharmaceutical companies, including Amgen, Ariad, AstraZeneca, Bristol-Myers Squibb, EUSA Pharma, Genzyme, GlaxoSmithKline, Jazz Pharmaceuticals, MedImmune, Micromet, Novartis, Pfizer, Sanofi-Aventis, Seattle Genetics, and Talon. He reports no financial relationship or other conflict of interest with Genzyme or with any of the other aforementioned entities because he has not accepted any honoraria or consulting fees for this work since the beginning of 2011, which will encompass the 36 months before the publication of this article. Dr. Whitlock does report research funding from Amgen, Bristol-Myers Squibb, GlaxoSmithKline, and MedImmune.