Yevgen Ryeznik (1,2), Andrew C. Hooker (2), Alex Sverdlov (3)
(1) Department of Mathematics, Uppsala University, (2) Department of Pharmaceutical Biosciences, Uppsala University, (3) EMD Serono
Objectives: Many clinical trials use time-to-event (TTE) outcomes as primary measures of efficacy or safety. For instance, in dose finding cancer trials the goals may be to estimate a dose-response relationship and to identify a dose level that yields the longest progression-free survival for testing in subsequent studies. Efficient designs for such trials are needed but finding such designs in practice may be complicated due to uncertainty about the model for event times, delayed responses and censored observations. In this work we develop optimal and adaptive designs for dose finding clinical trials with TTE outcomes.
Methods: We consider an accelerated failure time (AFT) model [1] assuming a quadratic dose-response model for log-transformed TTE outcomes with a Weibull distribution that are subject to right censoring with a fixed or random censoring time. We obtain the D-optimal design for the most precise estimation of the dose-response curve applying the general equivalence theorem [2]. Both the case when the dose is assumed to be a continuous factor and the case when the dose levels are chosen from a discrete set are studied. The censoring mechanism [3] is explored and the robustness of the designs to model misspecification is assessed. For implementing optimal designs in practice we propose a multi-stage adaptive design. The effect of delayed response and different recruitment patterns on statistical properties of adaptive designs is examined.
Results: The proposed optimal designs generate allocation of patients to the most informative dose levels and achieve higher efficiency in estimating the parameters of interest compared to the popular equal allocation designs in the presence of censoring. Adaptive designs can be used to approximate the optimal designs; however, a sufficient amount of outcome data must be observed during the recruitment phase of the trial to facilitate adaptations. Further research on the robustness of the proposed designs to model misspecifications is ongoing.
Conclusions: The proposed designs can improve efficiency of clinical trials with time-to-event outcomes by reasonable allocation of study patients to dose levels that are most informative for the given study objectives.
References:
[1] Wei LJ. The accelerated failure time model: A useful alternative to the Cox regression model in survival analysis. Statistics in Medicine 11(14-15):1871-1879 (1992).
[2] Keifer J, Wolfowitz J. The equivalence of two extremum problems. The Canadian Journal of Mathematics 12:363-366 (1960).
[3] Lawless JF. Statistical Models and Methods for Lifetime Data, 2nd Edition, Wiley, New York (2003).
Reference: PAGE 24 (2015) Abstr 3608 [www.page-meeting.org/?abstract=3608]
Poster: Methodology - Study Design