Multiparametric static hybrid ¹⁸F-FET PET/MR for preoperative prediction of TERT promoter mutation in glioblastoma.

PURPOSES: To determine whether multiparametric hybrid 18F-FET PET/MRI enables noninvasive prediction of telomerase reverse transcriptase (TERT) promoter mutation in glioblastoma and to assess the incremental value of combined clinical and imaging parameters. METHODS: We retrospectively included 220 patients with newly diagnosed glioblastoma at our institution between June 2021 and December 2025. Parametric maps, including ¹⁸F-FET PET, cerebral blood flow (CBF), and apparent diffusion coefficient (ADC), were co-registered with fluid-attenuated inversion recovery (FLAIR) images. Tumor volumes of interest (VOIs) were delineated on FLAIR and transferred to aligned maps for quantitative analysis. Contralateral normal white matter served as the reference region to calculate normalized ADC and CBF values (nADC and nCBF), defined as tumor-to-normal ratios, while PET uptake was expressed as a tumor-to-brain ratio (TBR). Independent predictors were identified using multivariable logistic regression. Diagnostic performance was evaluated using receiver operating characteristic (ROC) analysis with calculation of the area under the curve (AUC). RESULTS: In multivariable logistic regression analysis, age (OR 1.088, 95% CI 1.030-1.148; P = 0.002), TBRmax (OR 2.586, 95% CI 1.126-6.015; P = 0.033), and nADCmax (OR 0.130, 95% CI 0.024-0.700; P = 0.018) were independent predictors of TERT promoter mutations. The combined model incorporating clinical and imaging variables achieved the highest discriminative performance (AUC = 0.893), with a sensitivity of 84.3% and a specificity of 89.4%, outperforming the clinical-only model (AUC = 0.749) and the imaging-only model (AUC = 0.786). CONCLUSION: Combining age with multiparametric 18F-FET PET/MRI improves prediction of TERT promoter mutations in glioblastoma and may enable noninvasive preoperative stratification.