Supportive care for thrombocytopenia in patients receiving treatment for myelodysplasia: a challenge for the future

In this edition of Leukemia and Lymphoma, Greenberg and colleagues report the results of their effort to find a signal of activity of the thrombopoeitin (TPO)-mimetic agent romiplostim when combined with decitabine for the treatment of myelodysplasia (MDS) [1]. The clinical and methodological questions raised by this study are important, and speak to both the potential role of the TPO-receptor agonists as supportive care agents in myelodysplasia, and the scientific and methodological pitfalls of clinical development in this particularly heterogeneous target population. As the authors acknowledge, the sample size of this trial is insufficient to allow any statistically sound conclusions about the efficacy of the TPO-mimetic. Nevertheless, the observations regarding tolerability and a signal of drug activity are of interest.

The context of these trials is the historical concern that activation of c-mpl, the target of TPO, may promote acute myeloid leukemia (AML) blast survival [2–4] and increase the rate of progression of MDS to AML [5]. Transient elevations of the marrow blast count to greater than 20% were noted in four of 41 patients in the phase II study of romiplostim monotherapy in patients with lower risk MDS, all falling to ≤ 10% after withdrawal of the drug [6]. Another two of 41 patients progressed to AML, probably consistent with the natural history of the disease. One hypothesis for this effect is that activation of the mpl receptor has a stem-cell mobilizing-effect, a hypothesis supported by clinical observations from trials of recombinant TPO [7]. Amgen have recently issued a formal notice of a risk of progression of International Prognostic System Score (IPSS) low- and intermediate-1 MDS to AML [8] in patients receiving romiplostim, a consequence of early data from the placebo-controlled randomized trial of the drug as monotherapy in patients with lower risk MDS [9]. Meanwhile Glaxo-Smith-Kline (GSK) continues its development of the small molecule mpl agonist eltrombopag in this field, including studies in patients with higher risk MDS than those selected for the romiplostim studies, with no suggestion, so far, of a “blast signal” [10].

Severe thrombocytopenia (platelet count < 25 × 10⁹/L) affects only 2% of patients with IPSS low-risk MDS, rising to 25% of those with high-risk disease [11]. Thrombocytopenia in MDS is thought to be due to megakaryocytic cell death [12], defective thrombopoiesis (partly due to defective c-mpl signaling) [13], increased peripheral destruction [14,15] and possibly immune causes [16]. The risk of bleeding is difficult to quantify, but serious bleeding has been estimated to contribute to death in 20% of patients with MDS [11]. Perceived bleeding risk leads to considerable psychological anguish and inconvenience to the patient, especially where the traditional supportive care measures such as platelet transfusions are impractical or ineffective due to acquired transfusion refractoriness [11,17]. Until the advent of the TPO-mimetics and the DNA-demethylating agents, pharmacotherapy for reduced platelet counts has been limited to agents such as danazol, which is not particularly effective and has an undesirable side effect profile for many patients [18].

Azacitidine is, throughout much of the world, the standard of care for patients with intermediate-2 and high risk MDS [19], and is approved in some regions for the treatment of cytopenias in lower risk disease. Decitabine has approval for these indications in the USA, but is yet to achieve geographically widespread approval. Myelosuppression from either agent may contribute to the severity of thrombocytopenia in the early cycles of treatment, prior to hematological improvement if the patient has responsive disease [20,21]. Between 20 and 50% of patients with MDS will have hematological improvement following treatment with a demethylating agent, but the platelet response rate is lowest in those with the most severe thrombocytopenia at baseline [21]. These patients are most at need of platelet-directed supportive care during the first cycles of demethylating agents, where a safe platelet count facilitates on-time delivery of the demethylating agent and may reduce the risk of bleeding.

This study of romiplostim by Greenberg et al. [1] describes the results in patients receiving decitabine, and is best read together with the previously published report of the same clinical trial of romiplostim in a separate cohort of patients receiving azacitidine by Kantarjian et al. [22]. Patients with low- or intermediate-1 and -2 risk disease who had received fewer than four cycles of a hypomethylating agent were eligible. Subjects were randomized to treatment with placebo or weekly subcutaneous romiplostim. In the decitabine study presented here, romiplostim was dosed at 700 μg weekly; there were 500 and 700 μg weekly arms in the azacitidine study. In each, randomization was stratified by baseline platelet count (≥ 50 or < 50 × 10⁹/L). The primary end point of “clinically significant thrombocytopenic event” (CSTE) seems intuitively reasonable, and is defined as a platelet count of < 50 × 10⁹/L after day 15 or the requirement of platelet transfusions. However, neither study was adequately powered to prove a difference in the primary end point, with sample size calculations based on a clinically improbable absolute CSTE difference of ˜50% between the treatment and placebo groups, with only 12–15 patients randomized to each arm. These were exploratory studies.

In the Greenberg study, no difference in either platelet count or CSTE rate was found. However, platelet counts at the start of each decitabine cycle appeared numerically higher in the romiplostim treatment arm (Figure 2 in the Greenberg article). This suggests that romiplostim may be active in this disease but that it does not, at these doses, overcome the myelosuppressive swings following decitabine treatment. Unfortunately, the treatment arms were imbalanced, the higher number of patients with refractory anemia and excess blasts-II (RAEB-II) in the placebo arm (71 vs. 31%) hindering interpretation. However, the temporal pattern of platelet count is similar to that observed in the Kantarjian azacitidine article, which also demonstrated a dose-effect [23], and taken together one can see that romiplostim has some activity as a supportive care agent in this setting.

In the perfect world, both studies would have tested the hypothesis on a larger number of patients, or tested those with the lowest platelet counts at baseline. As it was, 60 centers were required, so higher enrolment in the study may not have been feasible. Furthermore, choosing a definitive end point to assess the efficacy of the TPO-mimetic is a problem: here the demethylating agent contributes both to myelosuppression and to the gradual improvement in blood counts that accompanies disease response. Does a decline in platelet count reflect loss of response to the romiplostim, myelosuppression, or disease progression? Perhaps romiplostim monotherapy for a period prior to exposure to the demethylating agent would have improved platelet counts in the early cycles. Finally, for patients with platelet counts not responding to romiplostim, there may be a case for intra-patient dose escalation of the agent based on the hypothesis that they have more defective marrow function requiring higher exposure to TPO.

The story of romiplostim in MDS appears to be indefinitely on hold, but there may still be a future for TPO-receptor agonists in MDS. Updates to the results of the randomized romiplostim monotherapy study [9] are expected at the American Society of Hematology (ASH) 2012 meeting and may provide further insight into the perceived risk of progression to AML.

Eltrombopag, an orally available small molecule activator of mpl, binds its target at the transmembrane domain of the receptor, and in this respect differs from romiplostim, a peptibody that binds TPO at the main receptor site [24]. Eltrombopag may have an alternative post-receptor effect and off-target mechanisms. Pre-clinical cell-line [25] and primary patient sample [26,27] experiments suggest an anti-proliferative effect on AML blasts, possibly independent of the effect on the TPO receptor [27,28]. A number of investigator- and industry-sponsored studies are investigating this agent in both MDS and AML. As was the case with romiplostim, the development pathway for this agent will likely prove difficult due to the intrinsic natural history and heterogeneity of the diseases under investigation. In studies of eltrombopag, patients with MDS at higher risk of transformation to AML could be included with more confidence. There may yet be a future for the TPO-receptor agonists for patients with MDS and AML.

Potential conflict of interest

A Disclosure form provided by the author is available with the full text of this article at www.informahealthcare.com/lal.