In HR-MDS, treatment goals include disease modification leading to a survival benefit. In the frontline setting, HMAs such as azacitidine (AZA), decitabine, and oral decitabine/cedazuridine remain the standard of care [1]. Of the HMAs, only AZA has Level I survival data supporting its use.

Relapsed/refractory (R/R) MDS have dismal outcomes with median survival between 4-6 months. In 2023, ivosidenib, an isocitrate dehydrogenase -1 (IDH-1) inhibitor, was approved to treat the small subset of patients (<5%) with R/R IDH1 mutated MDS based on a small, pivotal Phase 1, open-label study (n = 18). In this study, ivosidenib demonstrated a complete remission (CR) rate of 38.9% and the median duration of response was not reached [6].

The HR-MDS combination treatment space remains a boulevard of broken dreams on which HMA continues to walk alone. Between 2012-2014, SWOG 1117 randomized HR-MDS patients to AZA + lenalidomide, AZA + vorinostat, or AZA monotherapy with the primary endpoint of overall response [ORR – (CR+partial remission (PR) +hematological improvement (HI) per international working group (IWG) 2006 criteria] [7]. While previous phase 2 studies had an ORR of 72% (CR + HI), the S1117 ORR was 38% in the AZA monotherapy arm; 49% for AZA+lenalidomide (P = 0.14 vs. AZA monotherapy); and 27% for azacitidine+ vorinostat. The lack of benefit for combination arms was attributed to non–protocol–defined dose modifications, which compromised the efficacy of HMA-based therapy and response assessments, as many patients discontinued prior to the cycle 6 bone marrow assessment. This informed the conduct of future combination trials.

One such beneficiary was the PANTHER trial, conducted between 2017-2019, which randomized 454 patients with newly diagnosed HR-MDS, CMML, or oligoblastic AML to pevonedistat (a NEDD-8 activating enzyme inhibitor) +AZA vs. AZA monotherapy, with the primary endpoint of event-free survival (EFS). Strict language was included in the protocol to minimize dose modifications or premature drug discontinuation [8].

It worked - - CR rates in the AZA monotherapy arm were more than 50% higher than typically seen, which also may have compromised potential efficacy differences between arms. Consequently, at the time of prespecified interim analysis, the study failed to meet its primary endpoint. At a median follow up of 26.2 months for the pevonedistat+AZA arm (25.7 months for the AZA monotherapy arm), the ORR (CR + PR) was 24% (32% with AZA monotherapy) and the median EFS was 17.7 months (vs. 15.7 months for AZA monotherapy, P = 0.557). In patients with HR-MDS treated with pevonedistat+AZA, the median OS trended longer (21.6 months vs. 17.5 months for azacitidine; P = 0.092) and when analyzed by the number of treatment cycles received, the median OS in the pevonedistat+AZA arm was significantly longer in those who had received >3 cycles (median 23.8 vs 20.6 months for AZA monotherapy, P = 0.021), and for patients who received >6 cycles (27.1 vs 22.5 months for AZA monotherapy; P = 0.008) [8].

The number of drug combinations failing the ultimate randomized trial by fire, when compared to HMA monotherapy, is disheartening. Some of these include inhibitors of PI3 kinase, mutant TP53, and immunomodulatory therapies such as the Tim3 inhibitor (T cell immunoglobulin and mucin domain-containing protein 3) sabatolimab, and magrolimab (an anti-CD47 monoclonal antibody). All had previously shown efficacy (per IWG 2006 criteria) that appeared superior to HMA monotherapy in phase 2 studies. (Table 1).

What are the lessons learned from these combination trials? Keeping patients on HMA-based therapies for a minimum of six months leads to improved ORR and OS. Inflated early phase ORR [e.g., by incorporating clinically specious endpoints such as marrow CR(mCR)] may inflate perceptions of clinically meaningful efficacy of drugs that can translate to broader patient populations, and premature commitments to randomized trials. Focusing instead on meaningful treatment responses from phase 2 trials (such as CR) and improving response duration are essential prerequisites for registrational trials. Many of the trials above demonstrated a trend towards a significant improvement in EFS or OS for an HMA-based combination, and larger trials should be adequately powered to show smaller OS difference of 3-4 months, to allow these incremental benefits eventually to beget major survival benefits for patients with HR-MDS [9]. Most importantly, investigational agents must demonstrate single-agent activity before embarking on combination trials in HR-MDS.