Prediction Of Treatment Failure in Patients with Glioblastoma with Perfusion MRI and Molecular Biomarkers

Abstract Background Predicting early treatment failure in glioblastoma remains challenging. Given the median survival of 13.5 months, better prognostic biomarkers are needed. This study aimed to identify predictors of early treatment failure and their association with overall survival (OS). Methods We performed a retrospective analysis of consecutive patients with newly diagnosed glioblastoma who underwent DSC- and DCE-MRI before surgery and radiochemotherapy. Treatment failure was defined as tumor progression per RANO 2.0 criteria within six months of surgery. Factors independently associated with outcome among clinical, MRI perfusion at the diagnosis and molecular parameters were identified using multivariable logistic regression models. OS was evaluated using an 180-day landmark analysis to avoid time-dependent bias. Results Among 62 patients diagnosed between 01/2017 and 12/2021, 44 patients (71%) were non-responders and 18 patients (29%) responders. Elevated pretreatment rCBV and Ktrans were associated with treatment response (P < .05), with non-responders showing lower median rCBV (4.15 vs. 5.90) and Ktrans (0.15 vs. 0.17 min−1). A model, including age, perfusion metrics, and MGMT methylation status showed robust discrimination (AUC = 0.84, 95% CI: [0.73; 0.95]) to identify non-responders. Landmark analysis showed no evidence for early treatment failure itself to be associated with OS (P = .40), whereas MGMT promoter methylation status was associated with long-term survival and reduced hazard of death (HR 0.31, 95% CI 0.17–0.55; P < .001). Conclusion Pretreatment MRI perfusion metrics may predict short-term treatment failure and may guide early experimental therapy, while the MGMT status remains the primary determinant of long-term survival.