Identification of prognostically relevant residual tumor in glioblastoma after surgery: A comparative analysis of MR-based RANO resect classes vs. [18F]FET PET.
Abstract
Background
Assessment of postoperative tumor burden is essential in IDH-wildtype glioblastoma. The RANO resect classification stratifies patients based on residual tumor volume on postoperative MRI. [¹⁸F]FET PET can detect metabolically active tumor-suspicious tissue invisible on MRI. Its additional prognostic contribution remains ill-defined.
Methods
This retrospective bicentric study analyzed 140 patients with newly diagnosed IDH-wildtype glioblastoma and postoperative MRI and [¹⁸F]FET PET. MRI-derived residual tumor was classified by RANO resect; PET-derived volumes were defined using PET RANO 1.0 criteria. Spatial concordance was quantified using Dice coefficients. Prognostic associations were assessed by Cox regression. PET's incremental prognostic value was evaluated using a two-stage framework comparing MRI-only, PET-augmented, and fully adjusted models.
Results
Postoperative PET-defined volumes were substantially larger than MRI-defined volumes and showed minimal spatial overlap across RANO resect classes. PET-, T1CE-, and T2/FLAIR-based tumor volumes were independently associated with OS. Residual PET-volume was associated with increasing mortality risk and remained independently associated with OS (HR 1.021 per cm³, **p = 0.002). In an imaging-only model addition of PET improved time-dependent discrimination (ΔAUC(t) 0.03-0.07) and increased Harrell's C (ΔC ≈ 0.03), indicating improved prognostic discrimination. In the fully adjusted model, adding PET significantly improved model fit (*p = 0.0049), indicating prognostic information beyond MRI-derived tumor burden and clinical covariates.
Conclusions
Postoperative PET visualized tumor-suspicious volume not identified by MRI and provided additional prognostic information. These findings support the potential role of PET for refined postoperative risk stratification, while prospective studies are needed to determine whether PET-informed postoperative management improves clinical outcomes.