The Why, Which, When, and How of Azacytosine Nucleosides in Myeloid Malignancies

5AC and DAC comprise the most active class of drugs for the treatment of unselected patients with MDS. Randomized trials of each agent compared to observation have firmly established meaningful hematologic responses, ranging from well-documented single lineage hematologic improvement, which can provide important palliation of anemia, thrombocytopenia, and neutropenia, to less common complete hematologic, and at times, cytogenetic remissions ^1-3 . While the median duration of response in these studies (15 months for 5AC and 10 months for DAC) was less than optimal, both agents prolonged the time-to-progressive-disease or death (progressive disease in these cases defined as the development of 30 percent marrow blasts). While this endpoint did not achieve statistical significance in the DAC study, most likely due to under-treatment of patients (see ‘Which’), the observation of this trend in the latter study strengthens the conclusion from the former study that DNA methyltransferase inhibitors positively impact the natural history of MDS. Despite the toxicities of these drugs, treatment with 5AC was associated with improved quality of life compared to observation4.

Both 5AC and DAC inhibit DNA methyltransferase (DNMT), but it remains unclear how — indeed, whether — this mediates the clinical activity of these drugs (see ‘How’). DNMT inhibition requires incorporation of the azacytosine residue into DNA in lieu of native cytosines — these incorporated residues inactivate DNMT through the formation of irreversible inactive adducts. On a molar basis, DAC is approximately five times more potent in this activity. In contrast to DAC, 5AC is also incorporated into RNA — it is not clear whether this is associated with either beneficial or deleterious effects of the drug. In phase II studies, the clinical activities of these drugs appear comparable^5,6 .

Comparison of the phase III studies is difficult. The DAC study included fewer patients with low-risk disease and administered no greater than eight cycles of therapy. In the 5AC study, patients achieving complete response (CR) were treated with three cycles of therapy beyond CR; patients achieving responses less than CR received ongoing treatment until progression. The median number of cycles of treatment administered in the DAC study was three. Importantly, previous studies had shown that both agents require a minimum of four cycles to demonstrate hematologic response. The median time to exit from the 5AC arm in that trial was nine months. Given these important differences in study design and execution, response rate and duration were superior in the 5AC study, while toxicity was significantly less (< 1 percent mortality versus 10 percent). The reasons for the apparent under-treatment in the DAC study included disease progression (32 percent), death (23 percent), adverse events (7 percent), persistent cytopenias (21 percent), withdrawal of consent in four patients (9 percent), and physician choice (6 percent). Thus, it seems likely that the dose and schedule of DAC investigated and now FDA-approved may be more toxic and difficult to administer than the approved dose schedule of 5AC. A recent randomized phase II trial of three dose schedules of decitabine (100 mg/m2 total dose administered over five days) suggests a lower toxicity rate and higher response rate, with 95 percent confidence intervals around the response rate in the “best” schedule ranging from 27 – 51 percent (reported CR rate 40 percent). This important single-center phase II study will require confirmation in a multicenter setting⁷ .

Many clinicians appear to believe that azacytosine nucleosides should be reserved for high-risk or progressive MDS. However, the Silverman CALGB study of 5AC enrolled patients with all FAB subtypes, including approximately 60 percent with low-or intermediate-1-risk MDS classified by the International Prognostic Scoring System (IPSS)⁸ . Thirty percent of the patients on the randomized trial of DAC had lower-risk disease. In both studies, eligibility required hematologic impairment sufficient to potentially benefit from therapeutic intervention. Given that these agents are active in lower-risk disease and appear to retard progression of the disease, it does not seem appropriate to limit the use of these drugs to patients with higher-risk disease. Demonstrating change in natural history in response to these agents in lower-risk disease would require randomized trials specifically in that patient population. Both agents clearly have significant single-agent activity in AML; ongoing studies will be exploring the optimal application of these drugs for that indication. Regardless of the timing of the decision to commence therapy with an azacytosine analogue, a commitment to administering a minimum of four to six cycles of drug must be made, since it may require many cycles to see the first evidence of hematologic improvement. There is currently no evidence suggesting benefit of ongoing therapy if no objective response develops within that first six-month window. Several authors advocate the administration of azacytosine analogues on schedule despite persistent or worsening cytopenias during the first four cycles of treatment⁷ . At Hopkins, we have delayed therapy only if patients begin treatment with greater than 500 neutrophils and/or or greater than 20,000 platelets and have failed to recover to that baseline⁹ . Responses clearly develop in patients with critical neutropenia and thrombocytopenia. Indeed, these drugs may be critically important for such patients.

Based on their ability to inhibit DNMT and reverse aberrant methylation in vitro, both agents are frequently referred to as “hypomethylating agents.” It is important to note that while reversal of aberrant methylation in malignant cells has been reported associated with treatment with 5AC⁹  and DAC¹⁰ , it remains unclear whether this mediates or is required for response. Administration of 5AC to patients with MDS and AML has resulted in the unexpected induction of histone acetylation⁹ . DAC induces expression of phosphorylated histone H2AX, closely associated with double-stranded breaks in DNA (Fandy and Gore, unpublished data). Dissection of the operant molecular mechanisms responsible for the clinical responses to azacytosine analogues is critical both for the development of more active congeners and for the rational development of appropriate combination therapies.