Clinical, molecular and metabolic assessment in vorasidenib-treated IDH-mutant gliomas: the VORAFET study

Background: Vorasidenib, a brain-penetrant inhibitor of mutant IDH1/2, has become a treatment option for IDH-mutant, low-grade gliomas after surgery. Its effect is difficult to capture with conventional MRI, and its impact on seizures and the molecular determinants of response in routine practice are poorly known. We evaluated metabolic and structural response, safety, seizure outcomes and molecular features associated with response in a real-world cohort of IDH-mutant glioma patients treated with vorasidenib. Methods: VORAFET is a single-centre longitudinal study of consecutive adult patients with predominantly non-enhancing WHO grade 2-3 IDH-mutant gliomas, without prior radiotherapy or chemotherapy, treated with vorasidenib at the Veneto Institute of Oncology (IOV-IRCCS), Padua, Italy, within an expanded access program between June 2024 and October 2025. Integrated [¹⁸F]FET PET/MRI was performed at baseline and every 3 months up to 12 months, and response was assessed according to PET RANO 1.0 and volumetric RANO 2.0. Safety was assessed according to CTCAE v5.0. Seizure course was descriptively assessed in 3-month intervals in patients with (population A) and without (population B) active epilepsy at baseline, defined as at least one seizure in the 6 months before vorasidenib initiation. Tumours were characterised by next-generation sequencing, and the clonality of the IDH mutation was subsequently estimated as its cancer cell fraction (CCF). Results: Thirty-nine patients were included (16 astrocytomas, 23 oligodendrogliomas; 4 grade 3 gliomas); median follow-up was 11 months. Among 23 patients with PET-measurable disease, PET RANO 1.0 identified 2 complete and 12 partial responses (objective response rate 60.9%), with a median time to first response of 4.5 months. RANO 2.0 showed no complete or partial responses; 5 of 29 evaluable patients (17.2%) had a minor response. PET-PD occurred in 6 of 39 patients (15.4%), 5 of whom had stable disease on concurrent MRI. No baseline factor was clearly associated with metabolic response, whereas patients who developed PET-PD more often had ECOG performance status > 0 (50.0% vs 3.0%) and subtle contrast enhancement (33.3% vs 3.0%), and had numerically higher baseline FLAIR volume, MTV and TBR. Transaminase elevation occurred in 29 patients (74.4%; grade 3 in 1), requiring temporary interruption in 7 (17.9%) and dose reduction in 1 (2.6%), with no permanent discontinuation for toxicity. Among 13 patients with active epilepsy (population A), 7 (53.8%) remained seizure-free throughout the observation period; seizures during treatment were associated with larger baseline FLAIR volume (40.8 vs 8.1 cm³; AUC 0.93) rather than with early metabolic response. Among 26 patients without active epilepsy (population B), 6 (23.1%) had seizures during treatment. The IDH mutation was clonal in 20 of 21 evaluable patients (95.2%); higher CCF correlated with greater TBRmax reduction (ρ = −0.81; p < 0.001), and the only patient with subclonal-range CCF developed PET-PD. Conclusions: [¹⁸F]FET PET/MRI captured early metabolic changes during vorasidenib treatment not reflected by RANO 2.0, and liver toxicity was manageable and consistent with the INDIGO trial. Seizure outcomes were associated with baseline tumour burden rather than with early metabolic response, and molecular features such as the clonality of the IDH mutation emerged as potential determinants of metabolic response. Longer follow-up and larger cohorts are needed to confirm these hypothesis-generating findings.