Modeling tumor volume trajectory and outcome identifies 3 distinct response classes in lower-grade IDH-mutant glioma receiving first-line chemotherapy.

BACKGROUND: Isocitrate dehydrogenase (IDH)-mutant gliomas are infiltrative brain tumors with heterogeneous clinical courses. Predicting individual responses to chemotherapy and long-term outcomes remains challenging. We aimed to characterize distinct patterns of tumor evolution during first-line chemotherapy using longitudinal tumor volume dynamics and clinical outcomes. METHODS: We retrospectively included 195 patients treated at a single institution and analyzed 1956 MRI examinations obtained during chemotherapy. Tumor volume and mean tumor diameter (MTD) were quantified on T2/FLAIR-weighted images using an automated nnU-Net-based segmentation algorithm. Tumor evolution was analyzed using latent class joint models integrating longitudinal tumor measurements and time-to-event outcomes. RESULTS: Modeling based on MTD trajectories and overall survival (OS) identified 3 distinct response classes. The non-response class (13.8%) was characterized by early tumor growth from treatment initiation and poor outcomes (median progression-free survival [PFS]: 8.1 months; median time to anaplastic transformation [TAT]: 19.9 months; median OS: 41.1 months). The durable response class (57.4%) showed stable or slightly decreasing MTD trajectories and favorable outcomes (median PFS: 73.2 months; median TAT: 93.2 months; median OS: 169.3 months). Conversely, the transient response class (28.7%) exhibited an initial decrease in tumor size followed by rapid regrowth within 2 years (median PFS: 31.4 months; median TAT: 27.9 months; median OS: 54.2 months). CONCLUSIONS: Longitudinal modeling of MTD trajectories together with OS identified clinically meaningful chemotherapy response patterns in IDH-mutant gliomas, strongly associated with progression and malignant transformation. While most patients exhibited durable responses, others showed unfavorable trajectories, highlighting the need for early biomarkers to tailor treatment strategies.