Median overall survival reached 34.9 months with eflornithine plus lomustine versus 23.5 months with lomustine alone in patients with recurrent IDH‑mutant grade 3 astrocytoma (HR 0.64; p=0.01). Median progression‑free survival was 15.8 months versus 7.2 months (HR 0.57; p=0.01). In the overall 343‑patient intent‑to‑treat population, the primary endpoint of OS was negative (HR 0.95), and no benefit was seen in patients now classified as grade 4 disease. Safety was consistent with expectations, with Grade 3+ myelosuppression and hearing impairment the most relevant adverse events and no new signals reported.
The core development is publication of Phase 3 STELLAR in the Journal of Clinical Oncology, evaluating eflornithine combined with lomustine against lomustine alone after surgery, radiation, and temozolomide. The trial began under WHO CNS4 criteria and stratified by IDH status, but the clinically positive signal aligns with WHO CNS5, in which IDH‑mutant grade 3 astrocytoma constitutes a molecularly defined subgroup. A prospective analysis of this intended target population, planned prior to unblinding, demonstrated the OS and PFS gains that were absent in the broader ITT cohort.
Strategically, the path forward hinges on whether regulators accept a prospectively defined subgroup aligned to a reclassified disease taxonomy when the original ITT primary endpoint was negative. The sponsor’s decision to anchor its case in a CNS5‑concordant, IDH‑mutant grade 3 population reflects the field’s shift toward molecularly precise labels, but it also places scrutiny on the statistical analysis plan amendments, control of type I error, and the exclusion of CDKN2A/B homozygous deletions that now upstage disease. The magnitude and internal consistency of the effect help the argument; the misalignment between historical design and current biology complicates it.
For sites and CROs, this is another example of how classification updates can reshape trial execution midstream. Operationally, routine confirmation of IDH status and CDKN2A/B deletion will be table stakes for enrollment and for any post‑marketing use, pushing more central molecular testing and turnaround time management in the recurrent setting. Safety management will require hematology monitoring and audiology assessments, with clear dose‑modification algorithms to preserve time on therapy. For sponsors, the results underscore a broader trend: legacy datasets are being re-annotated to fit modern taxonomies, which demands robust data curation, preplanned subgroup analyses, and pragmatic biomarker workflows that sites can execute without slowing accrual.
If regulators entertain a registration path, expect a discussion around a restricted indication in IDH‑mutant grade 3 astrocytoma and an expectation for confirmatory evidence in a prospectively defined CNS5‑pure population. The competitive context remains relatively open in recurrent IDH‑mutant grade 3 disease, though any label will need to navigate evolving use of alkylators, re‑irradiation, and targeted options in earlier lines. Key risks include regulatory skepticism of a negative ITT readout, the requirement for robust companion or standardized testing infrastructure, and real‑world tolerability given ototoxicity and myelosuppression. Watch for alignment on endpoint selection and SAP particulars with FDA and EMA, the design of a confirmatory study, and whether the company secures a partner to finance and operationalize a tightly biomarker‑defined program.
Jon Napitupulu is Director of Media Relations at The Clinical Trial Vanguard. Jon, a computer data scientist, focuses on the latest clinical trial industry news and trends.
