1391P - Interpreting overall survival (OS) results when progression free survival (PFS) benefits exists in today's oncology landscape - metastatic renal cel...
|Date||30 September 2012|
|Event||ESMO Congress 2012|
|Session||Poster presentation II|
|Topics|| Renal Cell Cancer
S. Negrier1, Y. Tang2, C. Chen3, P. Bycott2
New cancer treatments (tx's) have clinical and health policy challenges to demonstrate a survival benefit over active tx after showing a PFS benefit. Subsequent tx and long post progression survival periods can lead to significant variability posing hurdles in showing OS benefit. The objective of this analysis was to examine the likelihood of OS benefit when a PFS benefit exists. We analyzed Phase III (P3) AXIS 2nd line trial data of axitinib versus sorafenib in mRCC employing methods described by Broglio et al 2009.Methods
We conducted simulations investigating the OS and PFS relationship utilizing the AXIS trial. OS was partitioned into 2 parts, expressed as the sum of PFS and survival post progression (SPP). We assumed a median SPP ranging from 6 to 18 months. The actual observed trial pooled SPP was 12.9 months. Median SPP was assumed equal in both arms. The impact on OS was directly reflective of PFS translated into a survival benefit. We further assumed a median improvement in PFS to 6.8 months in the axitinib arm from 4.7 months for sorafenib which translated into a HR = 0.69, based on P3 results. By bootstrapping the censoring times the simulations preserved and mimicked observed accrual, follow-up patterns, and censoring percentage of the P3 trial.Results
Based on 10,000 simulations per median SPP value and a PFS HR = 0.69, the likelihood of randomly observing an OS HR >0.9 ranged from 65% to 74% and increased with increasing median SPP up to 16 months and then appeared to plateau.Conclusion
Between 65% and 74% of the time, the HR for OS was >0.9 even when the HR was 0.69 for PFS. This gain was assumed to translate directly into a survival advantage (median SPP per tx arm was assumed identical). This may be due to variability during SPP follow-up (e.g. subsequent tx) which dilutes the OS comparison making it difficult to show statistical significance (Broglio et al 2009). The probability increased with increasing SPP up to 16 months. These results have important health policy implications. Powering a P3 trial and the long follow-up needed to show a statistically significant OS improvement may be costly, and prolong clinical trial read out in RCC.Disclosure
S. Negrier: Recieved honoraria from Pfizer and Novartis, and grants/research support from GlaxoSmithKline and Roche.
Y. Tang: Pfizer employee and stock shareholder.
C. Chen: Pfizer employee and stock shareholder.
P. Bycott: Pfizer employee and stock shareholder.