787O - External validation of the association of progression-free survival at 6 months (PFS6) with overall survival at 12 months (OS12) in second-line ther...

Date 30 September 2012
Event ESMO Congress 2012
Session Genitourinary tumors, bladder and testicular cancer
Topics Cytotoxic agents
Urothelial Cancers
Biological therapy
Presenter Guru Sonpavde
Authors G. Sonpavde1, B. Maughan2, K.M. Boucher3, R. Fougeray4, T. Choueiri5, G. Niegisch6, Y. Wong7, S.S. Sridhar8, C.N. Sternberg9, J. Bellmunt10
  • 1Medical Oncology, Texas Oncology, Baylor College of Medicine, 77598 - Webster, TX/US
  • 2Medicine, University of Utah, Salt Lake City/US
  • 3Biostatistics, University of Utah, Salt Lake City/US
  • 4Statistics, 5Institut de Recherche Pierre Fabre, Boulogne/FR
  • 5Medical Oncology, Dana Farber Cancer Institute, Boston/US
  • 6Urologic Oncology, 6Heinrich Heine University, Dusseldorf/DE
  • 7Medical Oncology, Fox Chase Cancer Center, Philadelphia/US
  • 8Medical Oncology, Princess Margaret Hospital, Toronto/CA
  • 9Medical Oncology, San Camillo Forlanini Hospital, Rome/IT
  • 10Medical Oncology, University Hospital del Mar-IMIM, Barcelona/ES



We hypothesized that PFS at 6 months (PFS6) correlates with OS12 and may be a robust intermediate endpoint for phase II trials of second-line therapy for advanced UC.


Ten phase II trials of second-line chemotherapy and/or biologics were pooled. Progression was defined as tumor progression or death from any cause. Adjustment of PFS6 and OS12 were performed for variability between trials using random effects models. The relationship between PFS6 and OS12 was assessed at the trial level using Pearson correlation and weighted linear regression. The relationship between PFS6 and OS12 at the individual level was assessed using Pearson chi-square test with Yates continuity correction. Statistical analyses employed “R” statistical computing software, version 2.8.0. External validation was conducted in a phase III trial comparing best supportive care (BSC) with vinflunine plus BSC.


Results from the pooled phase II dataset (n = 646) have been reported (ASCO 2012): the Pearson correlation between trial level PFS6 and OS12 was 0.66 (p = 0.037), and individual level agreement was seen in 82% of patients (kappa = 0.45), Pearson correlation between trial level response and OS12 was 0.37 (p = 0.30) and individual level agreement was seen in 78% (kappa = 0.36). Of the 357 patients in the external validation dataset, 17 had PFS censored before 6 months or OS censored before 12 months, leaving 340 evaluable patients. Of the progression events, 231 were objective progression, 104 were deaths and 5 were alive with PFS > 6 months and OS > 12 months. The PFS6 was 13.25% with BSC and 26.48% with vinflunine plus BSC. The individual level agreement of PFS6 and OS12 was 81% (κ= 0.44, p < 0.001) and of response and OS12 was 76% (κ= 0.17, p = 0.0002). Excluding the BSC arm showed an individual level agreement between response and OS12 of 82% (κ= 0.53, p < 0.0001).


PFS6 is robustly associated with OS12 at the trial and individual levels in the setting of second-line therapy for advanced UC. Given the suboptimal association of response and OS12 at the trial level, PFS6 may be a more optimal endpoint to capture the durable benefits of agents.


R. Fougeray: Employee of Pierre Fabre.

T.K. Choueiri: Research support from Astrazeneca.

Y. Wong: Research support from Imclone.

S.S. Sridhar: Research support from Celgene.

J. Bellmunt: Research support to institution from Pierre Fabre.

All other authors have declared no conflicts of interest.