The outcome of acute myeloid leukemia (AML) is extremely variable, with survival ranging from a few days to cure. A number of clinical and biological features at presentation have been reported as useful for the prediction of final outcome.1 Age represents the most relevant prognostic factor in AML, in that the prognosis of the disease steadily worsens in elderly patients.2 Comorbidity related to advanced age, more frequent unfavorable cytogenetic findings at diagnosis, and presence of an antecedent hematologic disorder, result in lower probabilities of complete remission (CR) and long-term survival after conventional intensive chemotherapy (CHT). The actual benefit of CHT is still debated, mainly in patients aged over 70 years.3 Accordingly, the research with new agents is particularly active, and AML has been recently proposed as a model for shortcomings in the clinical evaluation of new drugs.4
In patients with high-risk myelodysplastic syndromes (MDS), the use of the hypomethylating agents (HMA) 5-azacytidine (AZA) and 2′-deoxyazacitidine (decitabine: DAC) has been clearly associated with an improved outcome. These agents are now approved for use in this indication.5-7 Results in AML are less convincing, even though different studies suggest a potential utility of these agents. In this article, we critically appraise some recent clinical data reported on the use of HMA for the treatment of older patients with AML.
In a subanalysis of the AZA-001 trial, an advantage in terms of overall survival (OS) and other patient morbidity measures had been reported for AZA in older patients with low marrow blast count (20% to 30%) World Health Organization (WHO) -defined AML as compared to other conventional care regimens (CCR).8 At a median follow-up of 20.1 months, the median OS for AZA-treated patients was 24.5 months compared with 16.0 months for CCR treated patients (P = .005); the 2-year OS rates were 50% and 16%, respectively (P = .001). Two-year OS rates were higher with AZA versus CCR in patients considered unfit for intensive CHT (P = .0003). In addition, AZA was associated with fewer total days in hospital (P <.0001) than CCR.
While these data suggest a potential utility of AZA in AML, they also raise different points of methodological criticisms. First, while the authors refer to elderly patients with AML, they consider patients aged 50 years or older, who certainly are not elderly. Most patients were in the age category of 55 to 64 years, not strictly an elderly population, accordingly to the current AML literature.9 Second, the trial was based on the comparison of AZA with CCR. which encompassed 3 completely different therapeutic strategies for AML: best supportive care (BSC), low-dose cytosine-arabinoside (LDARA-C), and intensive CHT. While either LDARA-C or CHT aim at CR achievement, BSC only aims at improving the quality of life and, whenever possible, at avoiding hospitalization. Accordingly, comparison of morphological CR rate between AZA and CCR is misleading. The response assessment should exclude patients managed with BSC and limited only to patients “actively” treated with LDARA-C and CHT. The CR rate would then be 29% for the control group versus 15% for AZA. Although, as previous studies demonstrated, the improvement in survival in patients with high risk MDS treated with AZA is seen even in those who do not achieve CR,6 translation of this concept to AML needs more robust demonstration. On the other hand, it should be also considered that many hematologists are today reluctant to administer CHT for this population, as exactly happened in AZA-001 study. The above considerations should be also applied to the impact on transfusion requirement: patients managed with BSC would not achieve transfusion independence; therefore, they should be excluded from the comparison. Third, the trial was designed for patients initially classified as MDS, not AML. The investigators could have been influenced by the initial diagnosis, and might have been more reluctant to administer CHT to patients diagnosed as having MDS rather than AML. This could partially explain the very low number of patients selected for CHT (n = 11). As a consequence, the comparison with intensive CHT is unreliable, unless a control group including 11 patients is considered acceptable for the statistical analysis. The low number of patients results in even less reliable comparisons as distinct cytogenetic groups were considered: in the comparison with CHT, 100% of patients treated with AZA had intermediate cytogenetics, a very uncommon finding in elderly AML where at least 30% of patients present with adverse karyotype.2 Fourth, the control group had a median survival of 16 months, which is longer than that reported in most series of AML of the elderly (generally 5-7 months). The exclusion in the study of the proliferative variant of AML, characterized by high white blood cell at diagnosis, which represents a main adverse prognostic factor,10 likely contributes to the unusual favorable outcome seen either in the AZA or in the control group. Finally, in the study the survival curve of patients treated with AZA showed a plateau after 2 years. More information should have been given as long-term surviving patients are concerned. Were they in morphological CR? Did they achieve a consistent reduction of bone marrow blast count? This information is important, to define whether in AML with low blast count, the achievement and maintenance of CR represent an essential perquisite for long-term survival as in AML of young-adults patients.
In a multicenter, phase 2 study, 55 untreated patients older than 60 years with AML were treated with DAC 20 mg/m2 intravenously for 5 consecutive days of a 4-week cycle.11 Patients continued to receive the drug until disease progression or an unacceptable adverse event occurred. The mean age was 74 years, and patients were treated with a median of three cycles (range, 1 to 25 cycles) of DAC. The overall response rate was 25% (CR rate, 24%), and the response rate was consistent across subgroups, including patients with poor-risk cytogenetics and those with a previous history of MDS. The median OS was 7.7 months, and the 30-day mortality rate was 7%. The most common toxicities were myelosuppression, febrile neutropenia, and fatigue. The conclusion of the study was that DAC, given in a low-dose 5-day regimen, has activity as upfront therapy in older patients with AML with acceptable toxicity and mortality.
Interesting results were also reported by adopting an alternative schedule of DAC (20 mg/m2 intravenously over 1 hour on days 1 to 10) in a series of 53 subjects with AML and median age of 74 years.12 A remarkable CR rate of 47% was achieved after a median of 3 cycles of therapy and responses were seen in all age groups, in subjects with both low and high presenting WBC count, in both de novo and secondary AML, and in all cytogenetic subsets. Median OS for all patients was 55 weeks, median disease-free survival (DFS) for patients achieving CR was 46 weeks. The main causes of death were refractory/relapsed disease (56%) and infections (19%). Of interest, in this study, higher levels of miR-29b were associated with clinical response.
Finally, a superior outcome with either AZA or DAC, in comparison to conventional CHT, was demonstrated in patients with AML or high risk MDS with chromosome 5 and 7 abnormalities by a retrospective comparative analysis versus conventional aggressive CHT, including cytarabine-based regimens in 72% and other regimens in 28%.13 In this study, 37 patients had AML and most had complex karyotypes at presentation. Of note, while HMA results in terms of OS were superior for patients with MDS (46 weeks vs 19 weeks), the median OS was similar when comparing patients with AML who received HMA with those who received CHT (21 weeks vs 24 weeks, respectively). In addition, the median OS was significantly longer for patients with lower presenting WBC count (<10 × 109/L) when treated with HMA as compared with CHT (192 weeks vs 40 weeks, respectively). In this study, early death rate was similar (17% for both groups); however, patients treated with HMA had a significantly higher median age and a larger proportion of them presented with secondary or therapy-related AML/MDS, both of which are expected to adversely affect the clinical outcome.
Thus, though a comparison between phase 2 vs phase 3 studies should be always considered with great caution, the response rate (namely CR rate) seems to be superior with DAC than with AZA in elderly AML, while data on survival are worse in all DAC studies as compared to the trial based on AZA by Fenaux et al.8 The selection of patients with hypoproliferating disease, age less than 70 years in a considerable number of patients, no high risk karyotype and, last but not least, a low number of patients, may account for the above differences. Accordingly, a head-to-head comparison of AZA versus DAC is warranted.
All the above studies suggest a potential benefit of HMA in AML in older patients; indeed, the risk/benefit profile of either AZA or DAC offers a treatment potentially able to alter the natural history of the disease (in particular improving OS) in a patient population that otherwise receives supportive treatment only or hydroxyurea. The possibility of overcoming the adverse prognostic relevance of adverse cytogenetic findings at diagnosis represents an additional reason for pursuing the biologic and clinical development of HMA in AML. In this regard, it should be emphasized that, despite the high frequency of factors predicting poor outcome, clinical results from some of AZA or DAC trials compare favorably with those achieved with conventional regimens in “fit” patient populations treated intensively. In addition, data demonstrating a survival advantage in absence of CR are exciting and suggest an alternative mechanism of disease control in AML. This needs to be confirmed in larger and well-conducted new trials. Finally, a true superiority with respect to intensive CHT or other any new agent for AML in older patients with unfavorable cytogenetics or other adverse prognostic factors remains to be definitively demonstrated.