Olutasidenib plus azacitidine in IDH1-mutant higher-risk MDS, CMML, and advanced MPN

What was tried

NCT06597734 was an investigator-initiated Phase 2 study at MD Anderson Cancer Center evaluating olutasidenib (FT-2102, REZLIDHIA), a selective oral inhibitor of mutant isocitrate dehydrogenase 1, in combination with azacitidine for patients with IDH1-mutated higher-risk myelodysplastic syndrome, chronic myelomonocytic leukemia, or advanced myeloproliferative neoplasms. The protocol was posted in 2024 with a single arm of olutasidenib plus azacitidine and an estimated primary completion date of August 2027. The trial status was changed to WITHDRAWN with an actual enrollment of zero. The posted reason was "PI request."

Olutasidenib received accelerated FDA approval in December 2022 for relapsed or refractory IDH1-mutated AML at 150 mg twice daily. The MDS, CMML, and advanced MPN expansion would have been the first prospective registrational-type evidence for this molecule outside AML and cholangiocarcinoma.

The biological hypothesis

IDH1 R132 mutations produce a neomorphic enzyme that converts alpha-ketoglutarate to D-2-hydroxyglutarate. D-2-HG inhibits TET2 and histone demethylases, blocks myeloid differentiation, and drives DNA hypermethylation. Adding a mutant-IDH1 inhibitor to a hypomethylating agent compounds the antileukemic signal: olutasidenib lifts the differentiation block while azacitidine continues to depress global methylation. The strategy was validated in IDH1-mutated AML, where the Lancet Haematology Phase 1/2 dataset reported composite complete remission rates of 33 percent for monotherapy and 46 percent for the olutasidenib plus azacitidine combination in untreated AML (Cortes 2023, doi:10.1016/S2352-3026(22)00292-7).

The translational case for extending into CMML and higher-risk MDS rested on three observations. IDH1 mutations occur in roughly 5 percent of higher-risk MDS and 1 to 2 percent of CMML. Ivosidenib has Phase 1 data in IDH1-mutated relapsed or refractory MDS showing a 75 percent overall response rate among 12 evaluable patients (DiNardo 2024). The recently reported olutasidenib MDS expansion suggested benefit in the relapsed setting (Watts 2025). The MD Anderson protocol intended to fill the upfront combination gap.

What actually happened

The trial was withdrawn before any patient was enrolled. ClinicalTrials.gov lists the reason as "PI request" with no further detail, which is the standard shorthand at academic sponsors when an investigator-initiated protocol is replaced, redirected, or shelved. Industry-sponsored olutasidenib programs in IDH1-mutated MDS were already underway during the same window (NCT05445479 olutasidenib plus azacitidine in newly diagnosed IDH1-mutant AML, NCT05224258 olutasidenib monotherapy in IDH1-mutated MDS), and MDACC has a long history of pausing investigator-led work when an industry sibling trial opens at the same site. There were no patients treated, no adverse events recorded, and no efficacy signal generated.

Failure mechanism, best guess

The most likely cause is portfolio rationalization: the protocol was eclipsed by industry-sponsored studies covering an overlapping eligibility envelope, and the academic team chose to channel patients into those better-resourced trials. A secondary factor is the eligibility math. For a single-site trial requiring both higher-risk disease and an IDH1 mutation, the screening burden is roughly 25 to 50 patients screened per enrolled patient.

This is a sponsor_decision archetype. It does not implicate IDH1 biology, the olutasidenib molecule, or the hypomethylating agent combination. It does reveal a structural problem: IDH1-mutated non-AML myeloid neoplasms are an orphan-within-an-orphan population, and single-site investigator-initiated trials are not the right instrument to study them.

How to prevent this next time

Two changes would have raised the probability of completion.

First, a Bayesian feasibility design with explicit screening number-needed-to-screen (NNS). The NNS for an IDH1-mutated higher-risk MDS patient at a referral center is roughly 20 to 25 (5 percent prevalence, 25 percent willingness-to-enroll). For CMML it is 65 to 100. Enrolling 30 patients across all three indications would have required screening 700 to 1,500 candidates over 24 months, feasible only at consortium scale. A pre-activation Bayesian feasibility audit with a Beta(2, 18) prior on monthly accrual would have flagged a posterior probability of meeting target enrollment near 8 percent at a single site:

PP(success)=∫θ​P(final success∣θ,interim data)⋅π(θ∣interim data)dθ

That calculation, executed before activation, would have prompted either expansion to a multi-site network or migration to an industry-led structure.

Second, biomarker enrichment with consortium prescreening. The MDS Clinical Research Consortium and the MPN Research Foundation both run prospective molecular profiling networks. Routing prescreened IDH1-mutated patients across 8 to 12 sites into a single open protocol changes NNS from per-site to per-network, moving expected accrual from 0.4 to roughly 1.8 patients per month and completing a 30-patient cohort in 18 months. Adaptive randomization between olutasidenib monotherapy and the azacitidine combination, with response-adaptive arm dropping, would have made the trial informative even at smaller samples.

The single highest leverage change would have been activating the trial inside a consortium with shared molecular prescreening before opening at any individual site.

What this means for similar programs

Ivosidenib MDS expansion data is now the principal source of prospective evidence for IDH1 inhibition outside AML. Olutasidenib relies on retrospective and single-arm data in MDS. Programs in CMML have effectively no prospective IDH1-targeted dataset. Vorasidenib, approved in IDH-mutated low-grade glioma in 2024, is unlikely to be redeployed into hematologic disease given its brain-penetrant pharmacology.

The Mechanism Risk Score for IDH1 sits at 36 in the yellow band, driven by programmatic saturation across diseases and a moderate genetic-evidence deficit in CMML rather than mechanistic failure. Translational teams designing the next IDH1 trials in MDS and CMML should price in 24 to 36 months of consortium-level molecular prescreening before activation and consider basket designs that pool CMML, MDS, and MPN under a shared protocol with disease-stratified arms.

Open questions

How deep is olutasidenib mutant-allele clearance in non-AML IDH1-mutated myeloid disease at combination doses, and does it differ from the AML setting? Is there a synergistic differentiation effect in CMML monocyte progenitors distinct from AML myeloblasts? Will the ongoing industry-sponsored studies replicate the withdrawn MDACC eligibility envelope, or will they leave a CMML-specific knowledge gap that warrants reactivation?

Sources

1. • ClinicalTrials.gov NCT06597734 record (status WITHDRAWN, reason "PI request", enrollment 0, primary completion 2027-08-31). - ClinicalTrials.gov NCT05445479 and NCT05224258 (overlapping industry-sponsored IDH1-mutant MDS/AML olutasidenib programs). - Cortes JE et al. Olutasidenib alone or with azacitidine in IDH1-mutated AML. Lancet Haematology, 2023.DiNardo CD et al. Final Phase 1 substudy results of ivosidenib for mutant-IDH1 relapsed MDS. Blood Advances, 2024.Watts JM et al. Olutasidenib alone or combined with azacitidine in IDH1-mutant myelodysplastic neoplasms. Blood Advances, 2025.Patnaik MM et al. Genomic landscape of CMML.- Lindsley RC et al. Acute myeloid leukemia ontogeny is defined by distinct somatic mutations. NEJM 2017. - FDA prescribing information, REZLIDHIA (olutasidenib), revised 2024. - Open Targets v23.12: IDH1 target page, IDH1-CMML association score 0.283, IDH1-AML 0.791, IDH1-MDS 0.370. - ChEMBL CHEMBL4297610 (olutasidenib), small molecule, max phase 4. - openFDA FAERS, REZLIDHIA reactions through April 2026: 130 serious reports, 39 deaths, top events include off-label use (n=69), nausea (n=27), fatigue (n=26). Available from: https://clinicaltrials.gov/study/NCT06597734.