Next‐generation sequencing of baseline genetic mutations and outcomes of eltrombopag and azacitidine therapy in patients with myelodysplastic syndromes and thrombocytopenia: Data from the SUPPORT clinical trial

Abstract Eltrombopag has been previously shown to be effective in reversing azacitidine‐mediated thrombocytopenia. This was further investigated in the SUPPORT trial, a phase III study assessing the efficacy/safety of eltrombopag plus azacitidine in patients with intermediate‐ to high‐risk myelodysplastic syndromes and thrombocytopenia. The results did not support a clinical benefit for the addition of eltrombopag to azacitidine. We investigated if the somatic mutational profiles in the patient cohort were associated with treatment outcomes. Based on the available data, we observed no imbalance in the mutational profiles between treatment arms or a clear association between identified somatic mutations and clinical outcomes.

Hypomethylating agents used to treat advanced MDS, such as azacitidine, are associated with the development or exacerbation of thrombocytopenia [2]. It was previously shown in phase I and II trials that eltrombopag can reverse thrombocytopenia caused by azacitidine [3,4].
The phase III SUPPORT trial (NCT02158936) investigated the efficacy and safety of eltrombopag plus azacitidine compared with placebo plus azacitidine in patients with intermediate-to high-risk MDS and thrombocytopenia. The primary endpoint of platelet transfusion independence in cycles 1-4 of azacitidine therapy was inferior in the eltrombopag versus placebo arm (16%, n = 28/179 vs. 31%, n = 55/177, respectively), and the proportion of patients with progression to acute myeloid leukaemia (AML) tended to be higher in the eltrombopag than in the placebo arm (15%, n = 27/179 vs. 9%, n = 16/177, respectively) [5]. This report focusses on whether genetic factors contributed to the findings.
Recently, somatic mutations in >40 genes have been identified in the molecular profile of MDS, and distinct molecular subgroups with different prognoses or risks of progression to AML have emerged [6].
Point mutations in RUNX1, TP53 and NRAS were associated with severe thrombocytopenia and an increased proportion of bone marrow blasts, whereas point mutations in TP53, EZH2, ETV6, RUNX1 and ASXL1 were associated with shorter overall survival [8]. To address the possibility of genetic factors contributing to the outcome of SUPPORT, we performed biomarker analyses by next-generation sequencing (NGS) on 53 genes previously reported to be commonly mutated in MDS, including a set of 18 genes associated with poor prognosis (Table   SI) [5]. We analysed the mutational profiles of individuals within SUP-PORT to investigate whether (1) specific genetic mutations could be associated with the rate of progression to AML in each of the treatment arms; (2) an imbalance in genetic mutations might explain the increased progression to AML in the eltrombopag arm; or (3) treatment may have influenced patterns of clonal evolution.
Out of a total of 356 patients randomised to either eltrombopag or placebo, both in combination with azacitidine (intent-to-treat population; Figure S1), peripheral blood samples were taken at baseline in 211 patients (NGS population; eltrombopag, n = 101; placebo, n = 110). Both arms in the NGS population showed similar demographic and disease characteristics (Table S2) A previous study reported an association between patient outcome and the number of oncogenic mutations [9]. Therefore, we investigated the relative distribution of patients in the two arms by the number of mutations and observed similar frequency in both arms for all groups (Table S3).
Subsequent analyses focussed on a set of 11 genes for which mutations were found in >5% at baseline in at least one arm. The analyses showed no evident differences between the arms in the proportion of patients with mutations, although baseline frequencies of mutations in TET2, TP53 and DNMT3A were numerically (5%-6%) higher in the placebo versus the eltrombopag arm ( Figure 1A     in other genes ( Figure S5). Similar results were obtained for times to progression to AML ( Figure S6).
The clinical results from SUPPORT were unexpected, considering that other studies in similar patient populations have demonstrated that eltrombopag monotherapy has a well-tolerated safety profile and favourable outcomes in patients with MDS with thrombocytopenia [11][12][13]. Moreover, there were no signals that were likely to predict the outcomes of SUPPORT in the earlier trial phases of eltrombopag in combination with azacitidine in patients with MDS [3,4].
NGS revealed a highly variable genetic landscape across both arms, which is consistent with the genetic landscapes reported in other studies [6,9]. The mutational landscape was broadly similar between the two treatment arms.