SURG-01IS UNRESECTED GLIOBLASTOMA TRULY UNRESECTABLE? AN ANALYSIS OF BASAL MRI'S OF PATIENTS WITH UNRESECTED GLIOBLASTOMA INCLUDED IN THE GENOM009 TRIAL AND THE VARIABILITY OF SURGICAL APPROACHES

Abstract

BACKGROUND: The percentage of patients with unresected glioblastoma ranges from 20 to 40%. There exists variability in the selection of the surgical approach. METHODS: The main inclusion criteria for the randomized neoadjuvant trial GENOM 009 was unresected disease. Both diagnosis and progression of all patients MRIs were reviewed. In this sub analyses we performed a blinded revision of basal MRIs from 77 patients included in the trial. MRIs were uploaded to the web where five neurosurgeons (NS) from 3 different institutions, blinded to clinical status and other NS evaluations, decided which surgical approach could be performed based solely on the MRI of each patient: B = biopsy, PR = partial resection or RR = radical resection. Finally, the neurosurgeons chose which pre-operative assessments they would ask for before surgery and the intra-surgical technology they would need to perform it. RESULTS: Each of the 77 cases had T1Gd and T2 or FLAIR sequences. Basal surgery performed to the patient before being included in the trial was: B = 68, PR = 9. For each NS the evaluation of the pre-planned surgery was: NS-1: B = 44, PR = 24, RR = 9; NS-2: B = 47, PR = 24, RR = 6; NS-3: B = 56, PR = 17, RR = 4; NS-4: B = 26, PR = 23, RR = 23;NS-5: B = 45, PR = 5, RR = 26. Evaluated pre-surgical assessments were: neuro-navigation sequences, functional MRI, tractography, neuropsicologic basal assessment. Evaluated intra-surgical tools were: neuro-navigation, mapping, physiologic monitoring, intra-surgical echography and fluorescence. The Fleiss kappa statistics about the surgical approach was 0.234 (95%CI = 0.156-0.312), P = 0.0001. (Fair, less than moderate). CONCLUSIONS: The approach to radical surgery in glioblastoma has a great variability among different neurosurgeons and even centers. An analysis of the causes of this variability is needed as well as an effort to develop guidelines based on tumor location, size and eloquence of adjacent brain based on MRI.