Fax: (212) 604-6029
Combination of 5-azacytidine and thalidomide for the treatment of myelodysplastic syndromes and acute myeloid leukemia†
Article first published online: 20 AUG 2008
Copyright © 2008 American Cancer Society
Volume 113, Issue 7, pages 1596–1604, 1 October 2008
How to Cite
Raza, A., Mehdi, M., Mumtaz, M., Ali, F., Lascher, S. and Galili, N. (2008), Combination of 5-azacytidine and thalidomide for the treatment of myelodysplastic syndromes and acute myeloid leukemia. Cancer, 113: 1596–1604. doi: 10.1002/cncr.23789
The clinical trial and correlative laboratory studies were funded by a grant from Pharmion Corporation. Thalidomide was supplied for the study patients by Celgene Corporation.
- Issue published online: 17 SEP 2008
- Article first published online: 20 AUG 2008
- Manuscript Accepted: 20 MAY 2008
- Manuscript Revised: 19 MAY 2008
- Manuscript Received: 8 APR 2008
- acute myeloid leukemia;
- combined therapy;
- hematologic improvement;
- myelodysplastic syndromes
The treatment of myelodysplastic syndromes (MDS) remains a challenge to the clinician despite recent advances. Many patients will either not respond or will have only limited and/or brief responses to single-agent therapy. Eventually, 30% of patients with MDS will progress and develop acute myeloid leukemia (AML). New strategies are needed for these patients.
5-Azacytidine (AZA) and thalidomide were administered to 40 patients with MDS or AML. AZA was given at a dose of 75 mg/m2 subcutaneously for 5 of 28 days together with thalidomide starting at a dose of 50 mg per day and increasing to 100 mg per. Six patients had refractory anemia (RA), 2 patients had RA with ringed sideroblasts, 10 patients had RA with excess blasts (RAEB), 1 patient had RAEB in transformation, 4 patients had chronic myelomonocytic leukemia, 1 patient had chronic idiopathic myelofibrosis, and 16 patients had AML. Thirty-six patients were evaluable for outcome.
A hematologic improvement (HI) was observed in 15 of 36 patients (42%), stable disease was observed in 5 of 36 patients (14%), and 10 of 36 patients (28%) had disease progression. Six patients experienced complete remission (CR), 2 patients experienced an erythroid HI (HI-E), 1 patient experience an absolute neutrophil count HI (HI-ANC), 5 patients experienced a platelet HI (HI-P), and 7 patients had bilineage HI (HI-P and HI-ANC or an HI-E and HI-ANC). It was noteworthy that 9 of 14 patients with AML had a history of prior MDS, 2 of 9 patients achieved a CR, 4 of 9 patients had HI (HI-E and bilineage HI), and 1 patient had stable disease and was continuing treatment. DNA microarray analysis of 8 responders and 4 nonresponders revealed that the genes associated with cellular proliferation had higher expression levels in nonresponders.
The current findings indicated that a combination of low-dose AZA and thalidomide was well tolerated and was effective therapy for the treatment of patients with MDS and AML arising from prior MDS. Cancer 2008. © 2008 American Cancer Society.
The treatment of myelodysplastic syndromes (MDS) of all subtypes was changed significantly when, in 2004, the US Food and Drug Administration (FDA) approved the use of the hypomethylating drug, 5-azacytidine (AZA). Until that time, palliative care or bone marrow transplantation were the only options available to the patient. The Cancer and Leukemia Group B (CALGB) AZA phase 3 randomized trial indicated that the overall response to the drug was 60%, and 66% of responders achieved transfusion independence.1 To our knowledge, to date, AZA has been the only drug shown to alter the natural history of the disease.2 The drug also has been investigated in the acute myeloid leukemias (AML), mostly for recurrent or refractory disease.3, 4 The dose used in those early studies was higher than that used in patients with MDS, and the results were inconclusive. The studies suggested, however, that a lower dose treatment may be beneficial for a subset of patients with AML. The precise mechanism by which AZA affects a response in MDS remains an open question. The drug is a potent inhibitor of DNA methylation and is presumed to reactivate silenced genes that are critical regulators of growth, differentiation, angiogenesis, signaling, and DNA repair. Although this may be the mechanism underlying hematologic improvement (HI) in patients with MDS, the reactivated genes specific to MDS response remain unidentified.
It has been established that thalidomide, a drug that once was prohibited for use because of its teratogenic effects, is an effective modulator of immune functions, including the induction of T-cell and natural killer (NK) cell-mediated tumor cell killing; it has direct cytotoxicity effects (in multiple myeloma); it interferes with autocrine and paracrine cytokine functions involved in growth and inhibition of drug-induced apoptosis (in myeloma); and it interferes with the malignant cell microenvironment axis and angiogenesis.5 Some of these immune phenomena contribute to the evolution of MDS and prompted a pilot study in 83 patients with MDS. A response to thalidomide was observed in 16 patients, and 10 patients achieved transfusion independence. However, many patients could not tolerate the drug at the dosages used in the study.6
Microarray expression analysis has been used extensively to gain further biologic understanding of malignant disease and as an aid to diagnosis and prognosis. This type of analysis is especially difficult in MDS, because multiple lineages codominate the bone marrow. It is believed that the disease originates in a stem cell that then differentiates into all of the myeloid cell lineages. Prior attempts to purify the stem cells for analysis have not been totally successful, although useful biologic insight has been gained. However, a 30-gene erythroid gene expression signature has been correlated with responsiveness to lenalidomide.7 The profile, indicative of a block in erythroid differentiation, compliments the clinical response generated by the drug; improvement of anemia. The ability to identify which patients are less likely to respond to a specific treatment is beneficial for patients with MDS, in which drug-induced adverse side effects, such as myelosuppression, can be deleterious in a generally elderly population with comorbidities.
The treatment of MDS remains a challenge to the clinician despite recent advances. Many patients either will not respond or will have only limited and/or brief responses to single-agent therapy. Eventually, 30% of patients with MDS will progress and develop AML. New strategies are needed for these patients. The objective of this pilot study was to determine whether AZA combined with low-dose thalidomide was safe and whether the combination was superior to single-agent therapy in an outpatient setting for patients with MDS or AML. In addition, a microarray expression analysis was performed with pretreatment mononuclear bone marrow cells in an attempt to identify a profile that characterized which patients will and will not respond to treatment.
MATERIALS AND METHODS
Forty patients with any French-American-British (FAB) subtype of MDS as well as de novo or secondary AML were included in this study. Patients had not received hematopoietic growth factors for 14 days before the pretreatment bone marrow aspirate. AZA was given at a dose of 75 mg/m2 subcutaneously for 5 days every 28 days together with thalidomide starting at a dose of 50 mg per day and increasing to 100 mg per day. Responses were evaluated by using the International Working Group (IWG) 2000 criteria.
Mononuclear cells in pretreatment bone marrow aspirates from patients who were receiving AZA and thalidomide were obtained after informed consent according with Institutional Review Board-approved protocols from the University of Massachusetts Medical School. Cell pellets were dissolved in Trizol (Invitrogen) and stored at −80°C. Samples from the initial consecutive 28 patients who were enrolled in the study were used for RNA isolation and microarray analysis. Complementary DNA (cDNA) samples from these patients were hybridized to Affymetrix HG-U133AA microarrays, and data were analyzed using the GenePattern software package. Marker genes (those genes that best differentiate between different response types) were selected using the ClassNeighbors algorithm with the signal-to-noise metric.
Total RNA was purified from mononuclear cells using Trizol (Invitrogen). The Ovation Biotin RNA Amplification and Labeling System (Nugen, Inc.) was used for the linear amplification of 20 ng total RNA. Fragmented, labeled cDNA was hybridized to Affymetrix oligonucleotide microarrays as described previously.8, 9 Patient samples were profiled on HG_U133 Plus 2.0 microarrays; Raw expression values were normalized using Robust multiarray averaging (RMA).10
Marker Gene Selection
Raw gene expression values were preprocessed and normalized using RMA.10 Genes with minimal variation across the dataset were excluded by discarding genes for which the maximum gene expression value divided by the minimum value across all samples was <2 or if the difference between maximum and minimum values was <100. Marker genes were ranked using the signal-to-noise metric.9 For gene x, the signal-to-noise metric, Sx, was calculated as follows:
in which μ0 and σ0 are the mean and standard deviation for gene x in Class 0, and μ1 and σ1 are the respective values for Class 1. Statistical significance was determined by random permutation of the class labels.9 Significant markers were selected by using a false discovery rate threshold of 0.1, which was computed using the procedure described by Benjamini and Hochberg.11 Analyses were implemented with the GenePattern software package12 using the ComparativeMarkerSelection module.13
In total, 40 patients with either MDS or AML were recruited for this study (Table 1). The median patient age was 72 years (range, 49–84 years). There were 24 men and 16 women. Six patients had a FAB classification of refractory anemia (RA), 2 patients had RA with ringed sideroblasts, 10 patients had RA with excess blasts (RAEB), 1 patient had RAEB in transformation, 4 patients had chronic myelomonocytic leukemia, 1 patient had chronic idiopathic myelofibrosis, and 16 patients had AML. Among the 24 patients with MDS, 2 patients had low International Prognostic Scoring System (IPSS) scores, 9 patients had intermediate-1 (Int-1) scores, 9 patients had intermediate-2 (Int-2) scores, 3 patients had high scores, and 1 patient had an unclassified score (Table 1). Eighteen patients had a normal cytogenetic analysis.
|2||Died after 1 cycle||RAEB||Intermediate-2||46,XY|
|7||No TX, died early||AML||46xy,t(8;18)(q13;q21)/46,xy,add(18)(q21),add(21)(q22)/46,xy|
|9||HI-N minor, HI-E minor||AML||46,XX|
|14||HI-N minor||RAEB||Intermediate-2||46,XY.nuc ish 8p11.1-q11.2(D8Z2x2)|
|17||CR||AML||45, XX-16,der(18)t(1;18) (q31q22)der(21)t(16;21)(p11.2q21)+mar|
|18||HI-P major, HI-N minor||CMML||Low||46,XX|
|20||HI-P major, HI-N minor||AML||46,XY|
|22||Died after 1 cycle||AML||46,XY|
|23||HI-P minor||RAEB||High||44,xx,add(4)(p16),del(5)(q11.2),−7,−8,add(12)(p13),del(17)(p11.2), -18,add(19)(q13.4)|
|24||HI-N minor, HI-E minor||RAEB||Interemediate-2||46,XY|
|25||SD||RARS||Intermediate-2||44-47,xx,-5,dic(5;18)(q11.2;p11.3),+8,add(8)(p23),add(19)(p13.3), r(19)(p13.3q13.4), 120,add(20)(q11.2)|
|26||DP||AML||44 45,XX,add(2)(p13),der(2;3)(q10;q10),der(2;12)(p10;q10),der(3)t(3;12)(p12;q12), del(5)(q13q33),t(9;11)(q12;q13),−8,−12,−12, add(19)(q13.2),+add(19)(q13.2),+mar1, +mar2[cp16]/46,XX|
|27||CR||RA||Intermediate-2||44,X,-Y,del(3)(q13.2),add(3)(q21),add(5)(p13),del(5)(q13),-7,dic(12;13) (p11.1;p11.1), ring chromosome|
|28||HI-E major, HI-P major||RAEB||High||47,xy,+8/46,xy|
|31||On study 1 day only||RAEB||Intermediate-2||46,XY,del(6)(q21),der(17)t(11;17)(q13;p11.2)/46,sl,der(1)t(1;11)(p36.3;q13) /46, sl,der(8)t(8;11)(q24.3;q13)/46,XY|
|35||HI-E minor, HI-N minor||RA||Intermediate-1||47,XY,+8/46,XY|
|37||SD||RA||Intermediate-2||4-9,−12,add(13)(p11.2),−18,+mar4,XY,del(3)(p12p21),−5,−9, add(13)(p11.2)/45,sl1, +add(13)(p11.2)/43,X,−Y,del(3)(p12p21), del(5)(q13q33),add(6)(q21,add(7)(q22),|
|38||HI-N major, HI-P major||AML||46,XX|
|40||DP||AML||41 44,XY,del(5)(q13q33),der(7)t(7;15)(q21;q21),−12,−15,−17,add(19)(q13.4), −22[cp6]/46,XY|
Clinical Responses by IWG Criteria
Fourteen patients (35%) were removed from study because of disease progression (10 patients), declined therapy (1 patient), or death within ≤1 cycle of therapy (3 patients). There were 36 patients who could be evaluated for response (Table 2). Of these 36 patients, 21 (58%) had a response to treatment, and 15 (42%) did not (45 patients had stable disease, and 10 patients had disease progression) according to IWG criteria. In total, 6 patients (17%) experienced a complete remission (CR), including 2 patients with MDS and 4 patients with AML. Seven patients had a bilineage response, and 8 patients had a monolineage response. Two patients (6%) experienced an erythroid hematologic improvement (HI-E), 1 patient (3%) experienced an absolute neutrophil count HI (HI-ANC), 5 patients (14%) experienced a platelet HI (HI-P), and 7 patients (19%) had bilineage HI (HI-P and HI-ANC or HI-E and HI-ANC) (Table 2).
|IPSS/FAB||Total No. of Patients||CR||SD||HI-E||HI-N||HI-P||BI-LIN||DP|
Response in AML Patients
Of the 14 patients with AML, 5 patients (36%) left the study with disease progression, 1 patient (7%) had stable disease, and 8 patients (58%) responded with 4 CRs (29%), 3 bilineage HIs (22%), and 1 HI-E (7%) (Table 3). It is noteworthy that, of the 9 patients with AML who had a history of prior MDS (Table 3), 2 patients (22%) achieved a CR, 3 patients (33%) had bilineage HI, 1 patient (11%) had HI-E, and 1 patient (11%) had stable disease. Among the 5 patients with de novo AML, 2 patients (40%) experienced a CR, and 3 patients (60%) demonstrated evidence of disease progression.
|Response||No. of Patients|
|Total no. of patients|
|De novo AML (N = 5)|
|Secondary AML (N = 9)|
The median survival for all patients was 17.8 months. For patients with MDS, the median survival was 18.5 months; and, for patients with AML, the median survival was 13.2 months (Fig. 1). Eleven patients, including 3 patients with AML, were alive at the close of the study. Adverse events generally were mild (grade 1 and 2) and are listed in Table 4.
|Adverse Event||Relation to Drug Treatment, No. of Events|
|Clostridium difficile infection||2|
DNA Microarray Analysis
Total RNA was available from 8 patients who responded and from 4 patients with resistant disease (RD). Although the pattern in the responding group was somewhat heterogeneous, it was strikingly homogeneous for those with RD (Fig. 2). Most noteworthy, 21 of the 40 marker genes that were overexpressed (genes in red in Fig. 2 under the RD columns) in the resistant group were associated directly with cellular proliferation. These genes can be divided functionally into 3 groups. Group I (cell cycle) consists of 9 genes (BUB1B, CCNA2, TMPO, CDC2, CCNB1, UBE2C, TTK, CDC20 and MCM5); Group II (DNA replication/repair) is composed of 5 genes (TOP2A, RRM2, RFC3, FEN1, RNAH2A); and Group III (chromatin/spindle structure) contains 7 genes (KIF2C, NUSAP1, CENPF, MK167, SPAG5, CENPA and STK6) (gene names and symbols are defined in Table 5).
|Gene Name||Gene Symbol||Chromosome||Cellular Location||Function||Disease Associations|
|Group I: Cell cycle|
|Mitotic checkpoint kinase MAD3L||BUB1B/MAD3L||15q15||Cytoplasmic interphase cells; nuclear kinetochores||Delays anaphase until chromosomes attached to spindles||Neoplasms|
|Cyclin A2||CCNA2||4q25-q31||Nuclear||Essential for cell cycle control G1/S-phase and G2/M-phase transitions||Multiple neoplasms including myeloid leukemia|
|Thymopoietin||TMPO||12q22||Nuclear||Control of initiation of DNA replication; structural organization of nuclear envelope||Immunodeficiency diseases; neoplasms, viral diseases|
|Cell division cycle 2||CDC2||10q21.1||Nuclear||Essential for cell cycle control G1/S-phase and G2/M-phase transitions||Neoplasms|
|Cyclin B1||CCNB1||5q12||Nuclear||Control of cell cycle at G2/M-phase transition||Neoplasms|
|Ubiquitin-conjugating enzyme||UBE2C||20q13.12||Nuclear/cytoplasm||Required for destruction of mitotic cyclins||Unknown|
|Protein kinase TTK||TTK||6q13-q21||Spindle||Kinase functioning at mitotic spindle checkpoint||Unknown|
|Cell division cycle 20||CDC20||1p34.1||Nuclear||Association with cyclosome/anaphase-promoting complex||Neoplasms|
|DNA replication factor MCM5||MCM5||22q13.1||Nucleus||Cell cycle control of DNA replication||Neoplasms|
|Group II: Replication/repair|
|Topoisomerase (DNA) II-α||TOP2A||17q21-q22||Nuclear/cytoplasmic||Breakage rejoining double-stranded DNA; highly up-regulated in proliferating cells||Neoplasms|
|Ribonucleotide reductase MS||RRM2||2p25-p24||Cytoplasm||Biosynthesis of precursors for DNA replication||Immunodeficiency diseases by prevention of DNA replication and cell proliferation|
|Replication factor C3||RFC3||13q12.3-q13||Nuclear||Accessory protein for DNA polymerase activity||Tuberculosis|
|Flap endonuclease||FEN1||11q12||Nuclear||DNA replication, recombination and repair||Werner syndrome, leukemia, Bloom syndrome|
|RNase H||RNAH2A||19p13.13||Nuclear||Degrades RNA of RNA/DNA hybrids; participates in replication||Unknown|
|Group III: Structure|
|Kinesin family 2C||KIF2C||1p34.1||Cytoplasmic and nuclear||Associates with centromeres early prophase through telophase||Neoplasms|
|Nucleolar/spindle-associated protein 1||NUSAP1||15q15.1||Nuclear||Nuclear structural component||Unknown|
|Centromere protein F||CENPF||1q32-q41||Nuclear matrix||Involved in chromosome segregation||Unknown|
|Ki-67 antigen||MKI67||10q26.2||Nuclear||During mitosis associated with all chromosomes||Neoplasms|
|Sperm-associated antigen 5||SPAG5||17q11.2||Cytoplasmic||Necessary for mitotic spindle formation||Unknown|
|Centromere protein A||CENPA||2p24-p21||Nucleus||Histone H3-related, could act as core histone for centromere formation||Autoimmune disorders, scleroderma, telangiectasis|
|Serine/threonine kinase 6||STK6||20q13.2-q13.3||Centrosomes/spindles||Centrosome/spindle function during mitosis||Neoplasms|
If induction therapy for AML is to be taken as a model, then it is clear that the approximately 30% CR rate achieved with both cytosine arabinoside and doxorubicin as single agents was improved only to approximately 60% or 70% when the 2 drugs were combined in the now classic 7 + 3 regimen. Therefore, to improve the response rate in patients with MDS, it was considered reasonable to administer a combination of 2 agents with demonstrated efficacy in this disease and that act preferably through different mechanisms on the clonal population of cells.
MDS is a complex and heterogeneous disease in which the bone marrow microenvironment plays an important role in perpetuating the malignant clone. We combined 2 therapeutic agents, both of which target the bone marrow microenvironment as well as the cell. AZA is a prodrug with activity exerted through RNA and DNA. By binding to RNA, it causes inhibition of protein synthesis and, because these include several proinflammatory cytokines, the drug has a considerable effect on the bone marrow microenvironment. After conversion into its active form of decitabine, the drug acts by binding to the DNA with additional plieotropic effects on the MDS cells. These include cytotoxicity (especially in high doses), hypomethylation and gene reactivation (lower doses), inhibition of the enzyme methyl transferase, and differentiation (demonstrated for erythroleukemia cells).
Thalidomide, conversely, has anticytokine (especially antitumor necrosis factor), antiangiogenic, and immune-modulatory activity (especially differentiation of monocytes into NK and CD8-positive T cells, thereby enhancing autologous tumor response). Thus, thalidomide mainly targets the components of the bone marrow microenvironment. Combining thalidomide with AZA provides the potential to impact both ‘seed and soil.’ We previously published data on the combination of thalidomide with several agents, including amifostine, topotecan, etanercept, and arsenic trioxide. In each case, we demonstrated that combination therapy was more effective than single-agent therapy both for response and for the quality/duration of that response. In fact, the best survival we identified in an analysis of approximately 250 patients with MDS who were treated with thalidomide either alone or in combination, was observed in a trial of thalidomide and etanercept (Enbrel) in which no patient who responded had developed a recurrence almost 3 years later.14 Our experience, therefore, supported the combination of AZA with thalidomide as potentially useful in the treatment of all types of MDS and in patients with AML who have a history of prior MDS or who cannot receive conventional induction therapy.
In the current study, we commenced treatment with AZA at a dose of 75 mg/m2 every day for 5 days given every 4 weeks and thalidomide at a dose of 50 mg per day orally. On Day 8, the thalidomide dose was increased to 100 mg daily as tolerated. Patients with both MDS and AML were treated, and the results were encouraging, because 61% of patients had a response, including a 19% CR rate. In total, 40 patients were registered on the trial, but only 36 patients were evaluable for response. Of 14 patients with AML, 8 patients responded, 4 with a CR and 4 with a partial response. Of 22 patients with MDS (10 had higher risk IPSS scores and 11 had lower risk IPSS scores), 2 patients achieved CR, and 11 patients showed HI.
The results of the current study, albeit in a small cohort, established some important points: The drugs were tolerated well without additive toxicity. The treatment, by and large, was administered as outpatient therapy. Efficacy in high-risk MDS and AML patients was at least as good as 7 days of AZA alone. And, finally, survival after 1 year of treatment with AZA followed by maintenance with thalidomide appeared to be comparable to the survival of patients who were treated continuously with AZA in the phase 3 randomized survival study1 and appeared to be better than the survival in the conventional care arm of that study. The current study results cannot be compared with the AZA survival trial, because that trial only included higher risk patients.2
Although the sample size was small, gene expression profiling using DNA microarrays identified a molecular signature that appears to distinguish between patients who had some response (n = 8) and patients who manifested RD (n = 4). Of the 40 top marker genes that were overexpressed in the RD group, 21 are associated directly with cellular proliferation and can be divided functionally into 3 groups. Group I consists of 9 genes that have proteins located in the nucleus and that play a role in either the mitotic process or in the control of transition from the G1/S-phase to the G2/M-phase. Group II is comprised of 5 genes with proteins that are part of the DNA replication/repair pathway. Group III contains 7 genes with proteins that participate in chromatin and spindle structure. Aberrant expression and/or mutations in the majority of these genes have been identified in multiple malignant diseases. In general, the overexpression of these genes indicates that the cells are in a state of rapid proliferation, suggesting a possible underlying cause for the drug resistance observed in these patients. These results are especially striking because the treatment regimen consisted of a combination of 2 drugs, AZA and thalidomide, which purportedly act through different mechanisms. The finding that the RD patients appear to share a molecular signature, which also is consistent with a demonstrated mechanism of cell cycle resistance, is all the more promising for the future clinical usefulness of this methodology.
We thank Dr. Benjamin Ebert at the Broad Institute of Massachusetts Institute of Technology for computational analysis of the microarray data
- 2Azacitidine (AZA) treatment prolongs overall survival (OS) in higher-risk MDS patients compared with conventional care regimens (CCR): results of the AZA-001 phase III study. Program and abstracts of the 49th Annual Meeting of American Society of Hematology, Atlanta, Georgia, December 8-11, 2007. Abstract 817., , , et al.
- 11Controlling the false discovery rate: a practical and powerful approach to multiple testing. J R Stat Soc (B). 1995; 57: 289–300., .