Dose optimization of stereotactic radiosurgery by 18F-FHBG PET/CT tracking of HSV-TK gene vectors in recurrent glioblastoma: A prospective controlled study

To test whether 18 F-FHBG PET tracking of HSV-TK gene-vector expression can guide stereotactic radiosurgery (SRS) dose escalation to improve outcomes versus MRI-only SRS. In a single-centre, open-label, randomized phase II trial, 90 patients with recurrent glioblastoma were assigned (1:1) to PET-guided biological target volume SRS (PET-BTV) or standard single-fraction SRS (S-SRS). Primary endpoint was 6-month progression-free survival (PFS 6 ); key secondary endpoints included local control, 12-month overall survival (OS 12 ), ≥grade-3 toxicity, neurocognition (Trail-Making Test-B), and PET reproducibility. PET test–retest reliability and inter-observer agreement were assessed. Time-to-event outcomes used Kaplan–Meier/Cox models; toxicity used Firth logistic regression; repeated measures used appropriate ANOVA. PET metrics (SUV max , SUV mean , MTV) showed excellent test–retest reliability (intraclass correlation coefficients >0.85) and robust contouring agreement. Versus S-SRS, PET-BTV improved PFS (hazard ratio [HR] ≈ 0.54) and local control (HR ≈ 0.48; both P < 0.05), with no significant differences in OS 12 (HR ≈ 0.71) or ≥ grade-3 toxicity (HR ≈ 0.43; both P > 0.05). Multivariable Cox confirmed treatment group (HR ≈ 0.55), smaller tumour volume (HR ≈ 1.06 per unit), and higher Karnofsky score (HR ≈ 0.97) as independent predictors of longer PFS. PET response (decrease in SUV max ) correlated inversely with risk of progression. Trail-B favored PET-BTV (significant group effect and interaction), suggesting preserved neurocognition. 18 F-FHBG PET-guided intratumoral dose painting is feasible, reproducible, and improves PFS and local control without increasing high-grade toxicity, supporting imaging-driven precision re-irradiation for recurrent glioblastoma.