Bevacizumab (Bev) to reduce the negative impact of glioblastoma (GBM) tumor size on survival from first recurrence.

Abstract

2052 Background: Bev was FDA approved for recurrent GBM in 2009. However, the survival benefit from Bev in GBM remains to be demonstrated. Methods: We retrospectively identified 168 primary GBM patients diagnosed between 2001-2015 at UCLA and Kaiser Permanente LA, who received upfront radio-chemotherapy, followed with Bev and/or Lomustine (CCNU) at 1st recurrence. We measured tumor size at 1st recurrent treatment initiation, using bi-dimensional (2D) and volumetric (3D) techniques. We analyzed overall survival (OS) from 1st recurrence by Kaplan-Meier analysis. Results: Three groups of patients diagnosed from 2009-2015 were identified: patients treated with Bev alone (n = 49), CCNU alone (n = 36), and concurrent Bev/CCNU (n = 53). Patients were statistically different in performance status at 1st recurrence and tumor size; the CCNU alone group had smaller tumor sizes at diagnosis compared to the Bev groups. The CCNU group showed substantially greater survival (median OS (mOS) = 14.1 mo) compared to the Bev and Bev/CCNU groups (mOS = 6.9 and 7.1 mo, respectively), which may be explained by the imbalance in tumor sizes among the groups, and high rate of crossover (69%) to Bev in subsequent recurrences. To minimize selection bias, we identified another control group (n = 30) diagnosed from 2001-2004 who received CCNU only (CCNU 01-04). These patients had tumor size and KPS more comparable to both Bev groups, and a low rate of crossover (7%). OS for CCNU 01-04 (mOS = 5.7 mo) was similar to the Bev groups. Across all patients, we observed poor OS associated with larger 2D size vs. those with small tumors (mOS = 6.7 vs. 8.8 mo; p = 0.003). In separate stratification of each treatment group by tumor size, this association was retained in the CCNU 01-04 group (mOS = 4.0 vs. 8.4 mo, p < 0.001), but not in either Bev (mOS = 6.7 vs. 7.3 mo) or BEV/CCNU (mOS = 7.0 vs. 8.8 mo). Analysis of effect of tumor size by 3D measurement yielded similar results. Conclusions: Bev appears to reduce the negative impact of large tumor size on GBM patient survival from 1st recurrence. 2D and 3D measurements were correlated, suggesting the adequacy of use of conventional tumor bi-dimensional measurement to predict benefit of Bev in patients based on tumor size.