NorthWest Biotherapeutics Inc (NWBO)

[flipper44]

How statisticians handle cross-over arms:

Using Provenge as an example -- Crossover analysis is extrapolated to show Provenge treatment was even more efficacious than previously determined.

The Phase 3 IMPACT trial included a crossover design that allowed patients who were randomized to the control arm and experienced disease progression the opportunity to participate in an open label Phase 2 protocol to receive APC8015F, an investigational autologous cellular immunotherapy made from cells that were cryopreserved at the time the control was manufactured. As a result, 109 out of the 171 control patients (64%) received APC8015F.

In this exploratory analysis, researchers used a rank-preserving structural failure time (RPSFT) model, to quantify how treatment with APC8015F might have impacted the overall survival of the Phase 3 IMPACT trial by adjusting for the positive treatment effect of APC8015F in the control arm. The previously published intent to treat analysis, which is described in the Food and Drug Administration approved prescribing information for PROVENGE, did not account for cross-over and demonstrated a 4.1 month median survival benefit (HR=0.775, 95% CI: 0.614, 0.979). Using the RPSFT model, and assuming that APC8015F was equally effective as PROVENGE, the median overall survival benefit of PROVENGE in the Phase 3 IMPACT trial was estimated to be 7.8 months, had there been no cross-over to APC8015F (HR=0.60, 95% CI: 0.41, 0.95).

"This exploratory analysis provides important insight into how the cross-over design of the IMPACT trial may have affected the overall survival findings," said Leonard Gomella, MD, Kimmel Cancer Center of Thomas Jefferson University. "These data support the use of PROVENGE as a foundation of care for the treatment of metastatic castrate resistant prostate cancer."

http://investor.dendreon.com/releasedetail.cfm?ReleaseID=675334