Early increase in apparent diffusion coefficient predicts favorable response to dordaviprone (ONC201) therapy in H3K27M-mutant diffuse midline glioma

Rezumat

Abstract Background Dordaviprone (ONC201) is the first FDA-approved therapy for H3K27M-mutant diffuse midline glioma (DMG). However, therapeutic response varies among patients, and no imaging biomarkers currently predict long-term benefit. Diffusion-weighted MRI and apparent diffusion coefficient (ADC) obtained during standard clinical imaging reflect tissue cellularity and necrosis, offering potential as an early response indicator. Methods Clinical and imaging data from patients treated with dordaviprone (n = 83) on completed studies underwent centralized neuroradiology review, ADC histogram analysis, and functional diffusion mapping (fDM), which were compared and correlated to CSF tumor DNA (tDNA) and metabolomics when available. Final analysis included 38 patients receiving standardized dordaviprone dosing and treatment schedule. Results In the pre-progression cohort, early changes in volumetric or cross-sectional RAPNO criteria were not predictive of overall (OS) or progression-free survival (PFS). Importantly, in patients treated pre-progression with dordaviprone post-radiation, tumors with increased diffusion ADC (reduced cellularity) at cycle 3 demonstrated median OS of 732 versus 360 days for those with decreased ADC (increased cellularity, P = .0149); with corresponding PFS of 334 versus 110 days (P = .0132). fDM analysis of spatio-temporal changes of ADC did not correlate with outcomes in a sub-cohort analysis but demonstrated predictive patterns in individual patient time course series. CSF metabolomics stratified by ADC change revealed differences in reductive stress metabolites among dordaviprone responders, linking ADC elevation to dordaviprone-driven metabolic effects. Conclusion Early increase in ADC correlates with improved response to dordaviprone in H3K27M-DMG and correlates with CSF metabolic changes, suggesting ADC warrants prospective validation as a biomarker.