Anca‐Ligia Grosu, Wolfgang Weber, Erika Graf, Michael Mix, Ursula Nestle, Tanja Schimek‐Jasch, R. Wiehle, Irina Mader, Urs Würtemberger, Karl‐Josef Langen, Maximilian Niyazi, Frank Paulsen, Liane König, Frank A. Giordano, Irina Spehl, Denise Bernhardt, Markus Schymalla, Christoph Pöttgen, Sabine Semrau, Thomas Brunner, Beatrix Hültenschmidt, Bernd J. Krause, I. Frank Ciernik, Jürgen Beck, Brigitta G. Baumert, Philipp T. Meyer, Horst Urbach, Ilinca Popp, Claus Belka, Franziska Eckert, Michael J. Eble, Elena Sperk, Felix Momm, Stephanie E. Combs, Benedikt Wiestler, Rita Engenhart‐Cabillic, Martin Stuschke, Rainer Fietkau, Stephan Nadji, Guido Hildebrandt, Arnab Chakravarti, Susan Short, Carsten Nieder, Sabine Schneider-Fuchs, Marco Prinz
BACKGROUND: F]fluoroethyl)-L-tyrosine (FET)-PET has a higher specificity than contrast-enhanced T1-weighted MRI (CE-T1MRI) in diagnosing recurrent glioblastoma. We aimed to evaluate whether a FET-PET-based target volume delineation, compared with CE-T1MRI, improves outcomes in patients with recurrent glioblastoma scheduled for re-irradiation. METHODS: GLIAA was a multicentre, open-label, parallel randomised study done in 15 radiation oncology centres in Germany. Patients aged 18 years or older with a Karnofsky performance score greater than 60% and a macroscopic WHO grade IV recurrent glioblastoma (1-6 cm) were randomly assigned (1:1) to receive either FET-PET-based or CE-T1MRI-based target volume delineation followed by re-irradiation with 39 Gy in 13 fractions. Randomisation was performed centrally, using a minimisation technique with a random element and a computer-assisted randomisation tool, stratified by time since first radiotherapy, previous chemotherapy, tumour diameter, MGMT status, and planned chemotherapy. The primary endpoint was progression-free survival from randomisation, assessed in the per-protocol population (patients who initiated treatment per their assigned group). Adverse events were systematically assessed in all patients who commenced therapy. The trial was registered with ClinicalTrials.gov (NCT01252459), German Clinical Trials Registry (DRKS00000634), and European Clinical Trials Database (EudraCT 2012-001121-27), and is completed. FINDINGS: Between Nov 22, 2013, and Aug 18, 2021, 271 patients were recruited and screened for eligibility, 200 of whom were randomly assigned to re-irradiation based on FET-PET (n=100) or CE-T1MRI (n=100). 85 (43%) participants were female and 115 (58%) were male. 98 patients in the FET-PET group and 97 in the CE-T1MRI group were treated per protocol. Median follow-up for censored patients was 12·2 months (IQR 6·6-20·7). Median progression-free survival was 4·0 months (95% CI 3·7-5·2) in the FET-PET group and 4·9 months (3·7-6·0) in the CE-T1MRI group (one-sided stratified log-rank p=0·98; adjusted hazard ratio 1·14 [95% CI 0·85-1·52]; p=0·39; median follow-up for six censored patients 4·1 months [IQR 2·3-6·6]). The most common grade 3-4 adverse event was radionecrosis (eight [8%] of 99 in the FET-PET group vs seven [7%] of 99 in the CE-T1MRI group). Acute and subacute serious adverse events occurred in 15 (15%) of 99 patients in each group; possibly re-irradiation-related late serious adverse events occurred in ten (10%) of 97 patients in the FET-PET group and 18 (19%) of 96 in the CE-T1MRI group. There were no treatment-related deaths. INTERPRETATION: FET-PET-based target volume delineation for re-irradiation did not lead to a significant clinical benefit compared with CE-T1MRI-based treatment in patients with recurrent glioblastoma. Thus, CE-T1MRI remains the preferred delineation method in this setting. FUNDING: Deutsche Krebshilfe.