Salvage induction chemotherapy after azacitdine treatment failure in patients who received azacitidine as a bridge to allogeneic stem cell transplantation


  • Presented in part at the 2013 Annual Meeting of the American Society of Clinical Oncology, Chicago, Illinois, May 31, 2013–June 4, 2013.

Hypomethylating agents (HMAs), such as azsacitidine (AZA), are considered the standard of care for patients with International Prognostic Scoring System (IPSS) Intermediate-2 (Int-2) or High-risk myelodysplastic syndrome (MDS) and chronic myelomonocytic leukaemia (CMML) who are not candidates for allogeneic stem cell transplantation (alloSCT) (Kantarjian et al, 2006; Silverman et al, 2006; Fenaux et al, 2009). Therapy with AZA prolongs overall survival (OS) in this group of patients (including acute myeloid leukaemia [AML] with 20–30% blasts).(Fenaux et al, 2009) Nevertheless, these agents are not curative, with a median OS of 5·6 months for patients that fail AZA therapy. (Prebet et al, 2011) Many centres have used HMAs as a bridge to alloSCT in MDS patients in an attempt to reduce tumour burden and prolong the time to AML progression. (Damaj et al, 2012) However, this strategy leaves patients at risk of AZA treatment failure, which may delay or prevent subsequent alloSCT. Recent data, in a very small number of patients, have demonstrated poor response rates to salvage therapy with intensive chemotherapy after AZA failure. (Prebet et al, 2011, 2012; Duong et al, 2013)

This retrospective analysis evaluated the outcome of 18 patients with MDS, CMML and AML, who failed AZA therapy, whilst awaiting a donor to be found for alloSCT, and subsequently received induction chemotherapy (between 1 July, 2009 and 31 December, 2012) at the Princess Margaret Cancer Centre.

Seventeen of these patients received AZA as a bridge to alloSCT, while awaiting a donor to be found. Inclusion criteria were: (i) diagnosis of AML with 20–30% blasts, MDS or CMML by the World Health Organization classification, (ii) IPSS Int-2 or High-risk disease (Greenberg et al, 1997), and (iii) at least 1 cycle of AZA received. For MDS and AML patients, responses were defined according to the International Working Group Criteria.(Cheson et al, 2003, 2006) Survival was calculated from the date of AZA failure, defined as the date of evaluation of response after the last cycle of AZA. Primary AZA failure was defined as the absence of any response to AZA and secondary AZA failure as disease progression after a first response. Kaplan-Meier curves and median survival times were used to summarize the OS and time to progression outcomes. Continuous patient demographics were summarized by their median and range, while categorical demographics were summarized by their frequency and proportion.

Patient characteristics are listed in Table 1. Only one patient was over the age of 70 years. Of the 18 patients, 44% had secondary or therapy-related (s/t-) MDS or AML at the time of AZA therapy (including two patients who received a prior autologous SCT [autoSCT] for non-Hodgkin lymphoma and AML [different karyotype when developed MDS] and 1 an autoSCT followed by syngeneic alloSCT for multiple myeloma); 67% had IPSS poor-risk cytogenetics. Median number of AZA cycles received was 5·5 (range 1–18) with seven patients (39%) achieving a response (including one complete remission [CR], one marrow CR and haematological improvement [HI], and five HI). Nine patients were primary AZA failures and seven secondary AZA failures. No patients were intolerant of AZA. Two patients were taken off AZA after 1 cycle of therapy to receive induction chemotherapy as a human leucocyte antigen-identical donor was found.

Table 1. Patient characteristics.
CharacteristicAll (n = 18)
  1. AML, acute myeloid leukaemia; AZA, azacitidine; CMML-2, chronic myelomonocytic leukaemia, type 2; EPO, erythropoietin; F, female; G-CSF, granulocyte-colony stimulating factor; IPSS, International Prognostic Scoring System; M, male; MDS, myelodysplastic syndrome; MRC, Medical Research Council; RAEB-2, refractory anaemia with excess blasts, type 2; s/t-AML, secondary or therapy-related-AML.

Median age, years (range)57·5 (19–75)
Gender8 F:10 M
Disease duration before AZA, months (range)1·3 (0·6–24·6)
Diagnosis at time of AZA therapy
De novo10 (56%)
Secondary/Therapy-related8 (44%)
MDS15 (83%)
CMML-21 (6%)
AML2 (11%)
Prior therapy before AZA3 (17%)
Growth factors (G-CSF or EPO)2 (11%)
Hydroxycarbamide1 (2%)
Frontline AZA therapy15 (83%)
Azacitidine-based combination therapy 2 (11%)
Hydroxycarbamide + AZA2 (11%)
Median number of cycles of AZA (range)5·5 (1–18)
≤613 (72%)
7–124 (22%)
>121 (6%)
Type of AZA failure
Primary AZA failure9 (50%)
Secondary AZA failure7 (39%)
AZA discontinued for induction chemotherapy2 (11%)
Diagnosis at time of induction therapy
RAEB-22 (11%)
CMML-21 (6%)
s/t-AML15 (83%)
Cytogenetic risk group at time of induction therapy (MRC AML stratification)
Intermediate risk4 (22%)
Normal1 (6%)
Adverse risk 11 (61%)
Unknown 3 (17%)

At the time of induction chemotherapy, 15 patients had s/t-AML (12 had progressed from MDS), two had refractory anaemia with excess blasts type 2 (RAEB-2), and one CMML-2. Two patients with AML received cytoreductive therapy prior to induction chemotherapy: 1 with single agent cytarabine followed by hydroxycarbamide and 1 with hydroxycarbamide alone. Cytogenetic risk, as defined by the Medical Research Council AML classification, was intermediate in 4 and poor in 11 patients, respectively. Induction chemotherapy regimens are listed in Table 2. The overall response rate was 50% and 43·8% in all patients and AZA failures only, respectively (Table 3). The CR rate was 22% and 25%, CR with incomplete platelet recovery (CRp) was 6% and 0%, CR with incomplete count recovery (CRi) was 17% and 12·5%, and morphological leukaemic-free state 17% and 6·3% in all patients and in AZA failures only, respectively. One patient in CRi relapsed after 2·5 months and achieved CR2 after salvage chemotherapy with mitoxantrone, etoposide and cytarabine (NOVE-HiDAc). All seven patients achieved a CR/CRp/CRi after one induction chemotherapy cycle; four of these patients had intermediate-risk and 1 poor-risk cytogenetics, with no karyotype results available for the remaining two patients. Three patients with AML who were primary refractory to initial daunorubicin and cytarabine induction failed to achieve a CR with NOVE-HiDAc salvage chemotherapy. Four patients died during induction. Response duration for patients who achieved a CR/CRi/CRp was 4 months (range, 1·7–9 months). Four of nine responders received an alloSCT, with the remaining five patients relapsing (three while awaiting an alloSCT). After a median follow-up of 5·2 months, 17 patients have died. The one surviving patient has been lost to follow-up 25 weeks after induction chemotherapy. There was no difference in OS between patients who received an alloSCT compared with those who did not (= 0·55). All four patients who received an alloSCT died: two from relapsed AML, one from graft-versus-host disease and one from progressive respiratory failure. Median OS for all patients was 6·2 months (Fig 1). Median OS for patients who achieved a CR/CRi/CRp was 11·1 months, after a median follow-up of 10·2 months. No pretreatment variables were associated with improved OS or likelihood of achieving a CR, although the patient numbers were small.

Table 2. Induction chemotherapy details.
CharacteristicFirst inductiona All (n = 18)Second inductiona All (n = 18)Third inductiona All (n = 18)
  1. FLAG-Ida, fludarabine, cytarabine, G-CSF, idarubicin; NOVE-HiDAc, mitoxantrone, etoposide, cytarabine; modified Capizzi, cytarabine, etoposide, L'asparaginase

  2. a

    Five patients received a second induction and 1 a third induction

Induction chemotherapy regimen
Daunorubicin + cytarabine11 (61%)
NOVE-HiDAc3 (17%)5 (28%)
NOVE-HiDAc + Imatinib2 (11%)
Idarubicin + cytarabine1 (6%)
Modified Capizzi1 (6%)
FLAG-Ida1 (6%)
Induction Deaths4 (22%)00
Table 3. Responses with induction chemotherapy.
 All patients (n = 18)AZA failures only (n = 16)
  1. alloSCT, allogeneic stem cell transplantation; AZA, azacitidine; CR, complete remission; CRi, CR with incomplete count recovery; CRp, CR with incomplete platelet recovery; MLFS, morphological leukaemia-free-state; ORR, overall response rate; OS, overall survival.

Overall response rate 9 (50%)7 (43·8%)
CR4 (22%)4 (25%)
CRi2 (17%)2 (12·5%)
CRp1 (6%)0
MLFS2 (17%)1 (6·3%)
No. induction regimens to achieve CR/CRi/CRp
17 (39%)6 (37·5%)
Response duration (CR/CRi/CRp), months (range)4 (1·7–9)5·4 (1·7–9)
Responders who received alloSCT4 (44%)3 (43%)
Median OS (months)6·3 
Figure 1.

Kaplan–Meier curve for overall survival on all patients.

Contrary to prior reports (Prebet et al, 2011, 2012), our results demonstrate that induction chemotherapy can yield responses in a significant number of patients who have failed AZA therapy. However, response duration and OS remain poor for AZA treatment failures. Therefore, despite retrospective studies suggesting that it is not detrimental to receive hypomethylating agents prior to alloSCT (Damaj et al, 2012), one should not delay alloSCT for eligible patients with IPSS Int-2 and High–risk MDS until a loss or lack of response to AZA is observed. For those transplant-eligible patients who have a donor and may require tumour burden reduction prior to alloSCT, induction chemotherapy may be preferable to AZA. Patients who fail AZA therapy should be enrolled onto clinical trials.

Author contributions

KWLY treated patients, conceived research, collected and analysed data and wrote the manuscript. JB, ADS, VG, MDM and ACS treated patients and participated in writing of the manuscript. RDB and WX analysed the data.

Disclosures of conflicts of interest

KWLY, JB, and ACS received honoraria from Celgene; KWLY and ACS received research support from Celgene. The remaining authors declare no conflict of interest.