Patients with recurrent glioblastoma benefiting from bevacizumab are often treated indefinitely due to concerns regarding rebound tumor recurrence upon discontinuation. However, treatment is discontinued for reasons other than disease progression in a subset of these patients, the characteristics and outcomes of which are poorly defined.
Of 342 adults with recurrent glioblastoma in our database treated with bevacizumab, 82 received treatment for ≥6 months; of these, bevacizumab was discontinued for reasons other than tumor progression in 18 patients (Bev-D) and for disease progression in the remainder (Bev-S). The impact of discontinuation on outcome was assessed with discontinuation as a time-dependent covariate in a Cox hazards model for progression-free survival.
There was no difference in hazard rates for progression between Bev-D and Bev-S groups; the adjusted hazard ratio for progression using discontinuation as a time-dependent covariate was 0.91 (95% CI:0.47, 1.78). The median PFS after bevacizumab-discontinuation was 27 weeks (95% CI:15-NR). At progression, a higher proportion of Bev-D patients had local progression compared with the Bev-S patients. Salvage therapy in Bev-D patients yielded a PFS-26 weeks of 47% (95% CI:23%-94%) with a median PFS of 23 weeks (95% CI:12-NR), vs. 5% (95% CI: 1%-21%) and 9 weeks (95% CI: 6-11) in Bev-S patients (HR:0.3;CI, 0.1-0.6) (P = .0007).
Bevacizumab discontinuation unrelated to disease progression does not appear to cause rebound recurrence or worsen PFS in patients who benefit from bevacizumab. Additionally, Bev-D patients had an improved response to salvage therapy, findings which provide a strong basis for a prospective study.