Document Type



Background: Glioblastoma is an aggressive and almost universally fatal tumor. The prognosis at the time of recurrence has generally been poor, with overall survival typically in the range of 4-40 weeks. The merits of surgical resection (vs. open biopsy, to confirm recurrence via histology) in addition to conventional adjuvant chemotherapy have been the subject of longstanding debate. We wondered whether it would possible to conduct a trial at our institution to settle this question definitively with Class I evidence. Results: Initially, we had hoped to conduct a randomized, unblinded prospective clinical trial. However on closer inspection it appeared that such an undertaking would pose significant practical challenges. Thus we present our protocol in draft form. In keeping with recommended outcomes for these tumors, the primary endpoint would be median progression free survival. Secondary end points would be: median overall survival (mOS, from time of recurrence) and change in Karnofsky Performance Status over time. Patients would be eligible at the time of first recurrence if they had received conventional treatment until that point and at least 1 month had elapsed since the time of radiation. All patients would be considered potentially eligible for enrollment (unless the decision regarding resection was already clear-cut in view of other factors). Using Cox's proportional hazards model, we estimate that at least 456 patients would be necessary to demonstrate an increase in the hazard ratio to 1.3 for those undergoing biopsy alone. This magnitude of benefit is estimated based on a review of retrospective studies. Discussion: If restricted to our Institution alone, which sees approximately 100-150 new cases of glioblastoma each year, a trial of this nature would be likely to take around 10 years. Furthermore, there may be significant reluctance on the part of patients and physicians to participate. There is also the opportunity cost of excluding patients from other trials to consider. We recognize that the estimate of the magnitude of effect may be conservative. As things stand, we feel that multi-institutional collaboration would almost certainly be required for an undertaking of this kind.

Medical Subject Headings


Publication Date


Publication Title

BMC Research Notes







Digital Object Identifier (DOI)