• Randomized comparisons of bevacizumab(B) and irinotecan(I),added to temozolomide(T), in children with relapsed/refractory high-risk neuroblastoma(RR-HRNB); survival results of the ITCC-SIOPEN beacon-neuroblastoma phase 2 trial

      Wheatley, K.; Holt, G.; Owens, C.; Valteau-Couanet, D.; Gambart, M.; Castel, V.; Van Eijkelenburg, N.; Castellano, A.; Nysom, K.; Gerber, N.; et al. (2020)
      Background and Aims: BEACON is a randomized phase 2 trial assessing whether bevacizumab adds to chemotherapy and evaluating chemotherapy regimens for RR-HRNB. Methods: Patients with RR-HRNB underwent randomizations, in a 3x2 factorial design, to: T, IT or topotecan (To)-T, +/- B. Toxicity and response were already reported. Survival outcomes – progression-free (PFS) and overall (OS) – for the I and B randomizations are reported here (To rand continued). The B randomization used a relaxed alpha (1- sided p=0.2) for PFS as its phase 2 success criterion; the I randomiza- tion was Bayesian. Cox model hazard ratios (HR) <1.0 indicate benefit for I or B. Heterogeneity tests (HT) assessed interactions between B and I. Analysis was intention-to-treat. Results: From 2013-19, 160 patients were randomized to B v. no B, including 121 to I v. no I, with median follow-up 15.4 months. B ran- domization: HR (95%CI) for PFS and OS were 0.85(0.59-1.23) and 0.84(0.56-1.24) respectively. I randomization: HR (95%CI) for PFS and OS were 0.66(0.43-1.00) and 0.63(0.41-0.99). I improved PFS and OS (98% probability that true HR<1.0 for both) and B just met its success criterion (PFS: 1p=0.20; OS: 1p=0.19).There was some, but not conclu- sive, evidence of a positive interaction between B and I for both PFS (HT:p=0.06) and OS (HT:p=0.12). If real, this would suggest that adding either I (IT) or B (BT) to T does not improve outcome, but adding both BIT) does. Conclusions: The BEACON results show that single agent T is subopti- mal. Statistical uncertainty about an interaction between I and B means two further interpretations are possible: 1) IT and possibly BT are bet- ter than T; 2) IT and BT are not better than T, but I and B together (BIT) are better. Hence, a definitive conclusion on the best combination(s) to take forward is not currently possible and further randomized evalua- tion is needed