Correlation of treatment effects on progression-free survival at 6 months (PFS-6) and overall survival (OS) in patients with newly diagnosed glioblastoma (GBM).

Abstract

2008 Background: PFS-6 is commonly proposed as a primary endpoint in phase II trials to evaluate the efficacy of new therapies in patients with GBM. Two conditions must be fulfilled for validity: a strong correlation between PFS-6 and OS, and consistency in the effect of treatments on PFS-6 and OS. We examined the validity of this second condition using phase II data. Methods: OS and PFS at 6, 9, and 12 months were collected from 352 patients with newly diagnosed GBM treated in NABTT phase II trials with similar eligibility criteria and OS as the primary endpoint. 134 subjects were from historical controls (before temozolomide [TMZ]) and 218 were from 3 trials with TMZ. Treatment effects on OS compared to historical control were estimated using proportional hazards regression. Treatment effects on OS at 24 months (OS-24), PFS-6, 9, and 12 were estimated by the absolute difference in improvement of the corresponding proportions over control. Potential surrogacy of PFS-6, 9 12 for OS with respect to the effect of treatment was evaluated using rank correlation of the outcomes by bootstrap method. Results: The median age was 56 (range: 21-89), median KPS was 90 (60-100), and 80% underwent debulking surgery. Progression dates were captured in 99% of patients before death occurred. Treatment effects (percent reduction in hazard of death or OS-24) and PFS are tabulated below. The rank correlation between hazard of death (OS) and PFS-6 is -0.03 (95%CI: -0.16 to 0.11), for PFS-9 is -0.40 (95%CI:-0.51 to -0.28), and for PFS-12 is -0.21 (95%CI:-0.35 to -0.07). Similar correlations were observed between OS-24 and PFS-6, 9 or 12. Conclusions: We observe a near zero correlation between the magnitude of treatment effects on PFS-6, 9, 12 and OS. PFS-6, 9 or 12 do not appear to be valid surrogates for OS in clinical trials for newly diagnosed GBM. Improvement on treatments (95%CI) % reduction in hazard of death Absolute improvement %OS-24 %PFS-6 %PFS-9 %PFS-12 High (NABTT 0306, n=66) 49 (47-50) 28 (27-29) 23 (21-24) 27 (25-28) 21 (20-22) Median (NABTT 0304, n=85) 46 (45-48) 25 (24-26) 14 (13-15) 24 (23-26) 22 (21-23) Low (NABTT 0501, n=67) 39 (37-41) 15 (14-16) 21 (19-22) 36 (35-37) 25 (24-26)