Matts Kagedal (1) , Russ Wada (2), Shang-Chiung Chen (1), Jin Jin (1)
1. Genentech 2. Quantitative Solutions
Objectives: The aims of an exposure-response (E-R) analysis in oncology includes 1) to describe the relationship between drug exposure and survival and 2) to assess the risk of loosing efficacy in low exposure patients by comparing efficacy in patients with low exposure to control. The analysis is often performed as a time-to-event analysis where patients are binned into equally sized groups based on exposure. Few bins may result in a low probability in picking up an E-R trend since a wide range of exposures with different efficacy will be pooled into one group, while many bins will result in few patients per bin and large variability/uncertainty in the hazard ratio (HR) estimation. This simulation study was performed to assess the impact of binning by quartiles, tertiles or twotiles on the ability to 1) detecting an E-R trend; and 2) compare efficacy in the low exposure treatment bin over control.
Methods: A simulation study was performed assuming a 2-arm oncology trial with n=100 per arm. Two E-R scenarios were assumed: one with no E-R trend and a hazard ratio (HR) of 0.67 across all exposure levels, and one with an E-R trend where the HR ranged from 1 to 0.4 from low to high exposure. The simulated trials were evaluated after binning exposure by quartiles, tertiles and twotiles.
Results: In the first scenario of no E-R trend (True HR=0.67), it was concluded that efficacy was similar to or better than control (HR point estimate<1, CI<1.25) in 79% (quartiles), 87% (tertiles) and 94% (two-tiles) of the simulated studies, suggesting a quartile based analysis would fail to fulfill this criterion in 21% of the cases, while a two-tile based analysis would fail in only 4% of the cases. “No substantial E-R trend” was correctly concluded in 82% (quartiles), 85% (tertiles) and 89% (twotiles) of the simulated studies. In the second scenario with an E-R trend, there was no indication of worse efficacy (HR point estimate<1.25) than control in the low-exposure bin in 84% (quartiles), 91% (tertiles) and 98% (two-tiles) of the simulated studies. A “Substantial E-R trend” was correctly concluded in 92% (quartiles), 88% (tertiles) and 77% (twotiles) of the simulated studies.
Conclusions: The proposed clinical trials simulations approach enabled a quantitative evaluation of a key component of the E-R analysis plan. Binning by tertiles appears to provide the best ability to draw correct exposure-response conclusions in the tested oncology trial design.
Reference: PAGE 24 () Abstr 3591 [www.page-meeting.org/?abstract=3591]
Poster: Drug/Disease modeling - Oncology