Deferred use of bevacizumab for recurrent glioblastoma is not associated with diminished efficacy

We would like to thank D'Alessandris and colleagues for their interest in our study. We agree that following 2 negative phase III trials of upfront bevacizumab, both with significant crossover in the placebo arms, the optimal timing of bevacizumab administration is unclear.^1,2 Our study showed that the deferred use of bevacizumab to later recurrences was not associated with diminished efficacy. Additionally, we suggested that deferring bevacizumab to later recurrences may be beneficial for some, but not all patients.

D'Alessandris and colleagues observed that the time between diagnosis and initiation of bevacizumab was progressively longer in each recurrence cohort, and the overall survival (OS) from bevacizumab administration was statistically similar, which would imply that delayed bevacizumab resulted in an increase in OS. While it is true that the OS from time of diagnosis was progressively longer in each of the recurrence cohorts (first, second, and 3+ recurrences), we intentionally avoided making the conclusion that deferred use of bevacizumab led to an improved OS. Since it is likely that patients receiving bevacizumab at later recurrences were subject to selection bias enriching for those with greater OS, we strongly believe that the limitations of a retrospective study prohibit this conclusion, and suggest that a randomized prospective study is needed to answer this question. In a similar study by Hamza and colleagues, deferring bevacizumab to later recurrences did not change progression-free survival and showed a longer OS from date of surgery, but again this must be interpreted with caution.³ In both studies, the clinician's judgment to defer bevacizumab to later recurrences presumably selected for less aggressive disease, such as tumors with less edema, mass effect, enhancement, or steroid requirement. Moreover, this highly selected group of patients who deferred bevacizumab to later progressions may be confounded by markers of improved survival such as isocitrate dehydrogenase 1 (IDH1) mutation and methylation of O⁶-DNA methylguanine-methyltransferase (MGMT). Our study did not find a difference in MGMT methylation among the cohorts, but the percentage of IDH1 mutations increased with each recurrence cohort, suggesting a selection bias. Hamza et al also point out that data for prognostic markers were not included in their study and that the question of deferring bevacizumab to later recurrences must be examined prospectively.

From our study, we conclude that delaying bevacizumab in recurrent glioblastoma patients is not associated with diminishing effectiveness. Furthermore, we identified patients who were unable to continue therapy at early progressions, and suggest this population might benefit from early bevacizumab. More work is needed to identify both the optimal timing and patient subpopulations to maximize the benefits of bevacizumab therapy.