Simulation efficacy trials

Wientjes MG, Badalament RA, An JL. Use of pharmacologic data and computer simulations to design an efficacy trial of intravesical mitomycin C therapy for superficial bladder cancer. Cancer Chemother Pharmacol 1993,32(4) 255-62. 

The sections that follow provide examples of both model-based and trial-based input factors and output responses that should be considered for efficacy trial simulations. 

An additional component of the PK model that may warrant consideration in the simulation is the relative bioavaUabUity of the drug formulation to be used in the efficacy trial. Formulation changes may occur at this point in development, where a suboptimal formulation for commercial-scale manufacture may have been used in previous studies. If such changes have occurred between the dose-ranging study and current design, the relative bioavailabiUty between formulations (and associated 90% Cls) may also be considered for simulation evaluation. A sensitivity analysis (see Section 35.3.1) may be conducted to evaluate whether this effect will be influential on the simulation and/or if additional data may be required to provide acceptable precision. 

Taken together, the ER models for efficacy and safety will define the therapeutic window, where at the point of simulating an efficacy trial, an acceptable separation should exist. However, if multiple markers are being used to determine this separation, there may be a desire to weigh some markers more than others. For... 

Subjects will drop out of trials for either random (ignorable) reasons or perhaps for a reason attributable to their disease, trial conditions, or other nonignorable factor. Both conditions are important to consider for efficacy trial simulation. In the former case, subjects who drop out (are missing) at random will result in a decrease in total sample size and may affect the study power. In the latter case, nonrandom dropout is considered to be nonignorable in that the reason for dropout is informative to the trial outcome and may bias the results. In the seminal paper by Sheiner (25), an example of nonrandom dropout is presented for an analgesic trial, where those subjects not achieving adequate pain relief were more likely to drop out (i.e., to take rescue medication). 

Although some of the trial-based input factors will be fixed (e.g., if a design must be set a certain way due to unwavering logistics), many of the trial-based factors will be variables for which the simulation will attempt to find an appropriate combination to achieve the trials objectives. These variables are the what ifs of the efficacy trial simulation. An attempt to provide a thorough, albeit not all inclusive, list and brief description of trial-based factors to consider for efficacy trial simulation is provided below. 

Following discussion and acceptance of the SA results, including both model-based and trial-based input factor adjustments, the efficacy trial simulations may proceed as planned. For each possible trial design, the appropriate input factors and output responses are simulated and results are compared to determine the most appropriate design. As discussed previously, this final decision likely will not only be based on a specific p-value or trial power, but will also include valuations based on trial duration, monetary cost, or information gained or lost toward continuing development goals (e.g., an overall measure of clinical utility). 

A simulation of a hypothetical efficacy trial for a zidovudine analog (ZDVA) in HIV patients was completed to evaluate the probability of a successful Phase 3 trial if Phase 2b was skipped, given the Phase 2a results and prior knowledge from... 

TABLE 35.1 Description of Simulation Model Parameters Used for the Zidovudine Analog Efficacy Trial Simulation... 

FIGURE 35.1 Schematic of zidovudine analog efficacy trial simulation components. 

FIGURE 35.3 Results of the global SA from the zidovudine analog efficacy trial simulation displaying the mnltidimensional effect of parameter nncertainty on trial power (probabihty of snccess). Parameters displayed are placebo risk on the dropont hazard (HAZP) and maximnm dmg effect (ZDVSL) on the dropont hazard. 

The authors wish to acknowledge William R. Gillespie, Stuart L. Beal, and GloboMax LLC for their contributions to the efficacy trial simulation example. 

APPENDIX 35.1 NONMEM CODE FOR EFFICACY TRIAL SIMULATION... 

---