Time to Grade 4 Transformation and Subsequent Survival in Diffuse Glioma

Background/Objectives: Diffuse gliomas can reach grade 4 disease through markedly different temporal trajectories, yet prognosis after transformation is generally assessed from the characteristics of the transformed tumor itself. Whether the duration of the preceding disease course retains prognostic information after grade 4 transformation remains incompletely understood. Methods: We conducted a retrospective cohort study of patients with an antecedent WHO grade 2 or 3 diffuse glioma who subsequently underwent histopathologically confirmed grade 4 transformation. Transformation interval was defined as the time between the antecedent diagnosis and grade 4 transformation and was modelled continuously per 12-month increments. The primary endpoint was overall survival (OS) from transformation; progression-free survival (PFS) was secondary. Multivariable Cox models were adjusted for age, Eastern Cooperative Oncology Group performance status, antecedent grade and IDH status. Functional form, molecular context, landmark robustness and incremental prognostic information were additionally examined. Results: Fifty-one patients were included; 33 (64.7%) had antecedent grade 2 and 18 (35.3%) grade 3 disease. Median transformation interval was 51.6 months. Median post-transformation OS and PFS were 20.9 and 14.1 months, respectively. Each additional 12 months before grade 4 transformation was associated with lower mortality in univariable analysis (hazard ratio [HR], 0.845; 95% CI, 0.755–0.946; p = 0.003) and remained independently associated with OS after multivariable adjustment (HR, 0.858; 95% CI, 0.748–0.983; p = 0.028). No evidence of nonlinearity was observed (p = 0.407). The corresponding association with PFS was weaker and not statistically significant (adjusted HR, 0.943; 95% CI, 0.840–1.060; p = 0.327). Exploratory interaction analysis indicated that the inverse transformation interval–OS association was concentrated in IDH-mutant tumours (adjusted HR per 12 months, 0.758; 95% CI, 0.630–0.912), whereas no significant inverse association was demonstrated in IDH-wildtype tumours (adjusted HR, 1.225; 95% CI, 0.923–1.626; interaction p = 0.007). Adding transformation interval improved overall model fit but did not materially improve optimism-corrected discrimination. Conclusions: In this retrospective, single-centre cohort, a longer transformation interval was associated with improved post-transformation OS after adjustment for selected clinical and molecular factors, whereas the association with PFS remained inconclusive. The exploratory IDH interaction and the absence of meaningful improvement in optimism-corrected discrimination warrant cautious interpretation. These findings are hypothesis-generating and require validation in larger, molecularly defined longitudinal cohorts before clinical prognostic utility can be established.