Jeff Barrett Improving Study Design and Conduct Efficiency of Event-Driven
Clinical Trials via Discrete Event Simulation: Application to Pediatric Oncology
Jeffrey S. Barrett, Jeffrey Skolnik, Bhuvana Jayaraman, Dimple Patel, Peter Adamson Laboratory for Applied PK/PD, Division of Clinical Pharmacology, The Children’s Hospital of
Philadelphia
Objectives: Discrete Event Simulation (DES) is a method used to model real world systems able to be decomposed into a set of logically separate processes that autonomously progress through time.
Each event occurs on a specific process, and is assigned a logical time (a timestamp). The result of this event can be an outcome passed to one or more other processes. The content of the outcome may result in the generation of new events to be processed at some specified future logical time.
Our objectives are to employ DES to the study of clinical trial design efficiency, specifically to the investigation of dose escalation / de-escalation and stopping rules based on the frequency of DLT occurrence in pediatric oncology trials.
Methods: The classic Phase I oncology trial design was decomposed into a series of discrete time events (accrual/enrollment, evaluation and/or time to DLT or inevaluability) with outcome
probabilities (DLT or inevaluability) assigned to each subject based on historical data from phase I pediatric oncology trials. A study population of available patients (to be enrolled population) was simulated based on their likelihood of being inevaluable, evaluable (having a DLT or completing w/o DLT) and distributions which describe various time indices (arrival time, enrollment time, evaluation time, time to DLT, time to inevaluability). All event outcomes are generated for all patients and logic to describe which event occurred based on the time to event comparison is run.
Population simulations were typically evaluated for 100 simulated trials with 15 patients per cohort (prob of DLT increases w/ cohort) although the effect of sample size (up to n = 1000 trials) was also explored. From the population dataset, logic for the conventional (3+3 design) and newly proposed (Rolling 6 design) was applied to evaluate the operating characteristics of each approach. The code (1) decides if the study is open to enroll subjects, (2) checks to see if there are subjects waiting to be enrolled, (3) evaluates if the subject can be enrolled by checking against events which determine cohort progression (open, escalation, de-escalation or termination) assessing if MTD criteria are met, (4) progressing on assessment of action (3), adds "waiting period" - random variable based on historical experience, and (5) summarizes study outcomes. Subjects are chosen for "on study" status from the population dataset; subjects can be skipped (from the population dataset) if they are available when a waiting period (study closed) is in effect. Simulated patient trials and decision rule logic were coded PC/Windows SAS v9.1. Metrics for study efficiency (time to reach MTD, time to complete various trial designs and number of patients necessary to complete the trial) were defined and compared.
Results: Study efficiency is highly dependent on enrollment times. The rolling 6 design is superior to the conventional 3+3 design for most typically encountered enrollment times and study
conditions. The magnitude of the benefit depends on distributional assumptions and the amount of time it takes the trial to achieve the MTD and the designs, while not hierarchical, do perform similarly when arrival times exceed 100 days.
Conclusions: DES can be employed to examine the dependency of design and escalation rules on study efficiency metrics. This technique may offer real design alternatives, especially when coupled with Bayesian approaches, to minimize drug exposures of agents with narrow therapeutic windows and when patients do not stand to benefit from the study and should be generalizable to any event-driven clinical trial design.
References:
[1] Barrett JS, Patel D, Skolnik J, Adamson P. Discrete Event Simulation (DES) as a Technique to Study Decision Rule Efficiency in Event-Driven Clinical Deigns. J. Clin. Pharmacol 46: 1092 (Abstr. 131), 2006
[2] Skolnik J, Patel D, Adamson PC, and Barrett JS. Increased efficiency in phase I trials:
improving trial design to expedite dosing guidance in pediatric oncology. J. Clin. Pharmacol 46:
1092 (Abstr. 130), 2006
Poster: Methodology- Design
caroline BAZZOLI Population design in nonlinear mixed effects multiple responses
models: extension of PFIM and evaluation by simulation with NONMEM and
MONOLIX
C. Bazzoli (1), S. Retout (1, 2), F. Mentré (1, 2)
(1) INSERM, U738, Paris, France; Université Paris 7, Paris, France. (2) AP-HP, Hôpital Bichat, Paris, France.
Objectives: Multiple responses are increasingly used in population analyses. In this context, efficient tools for population designs evaluation and optimisation are necessary. The objectives are 1) to extend the population Fisher information matrix (MF) for nonlinear mixed effects multiple responses models and 2) to evaluate it by a simulation study using the FO and FOCE methods in NONMEM and the SAEM algorithm of MONOLIX.
Methods: We first extend the expression of MF for multiple responses model using a linearisation of the model as proposed for a single response by Mentré et al. [1]. We implement this method in an extension of PFIM [2], a R function for population designs evaluation and optimisation. Using a PKPD model example [3], we evaluate the relevance of the predicted standard errors (SE) computed by PFIM. To do that, first, we compare the SE of PFIM to those computed under asymptotic convergence assumption by MONOLIX through a simulation of 10000 subjects.
MONOLIX is based on the SAEM algorithm [4], and uses two methods to compute SE: a
linearization method and the Louis method [5]. Then, we also compare the predicted SE of PFIM to the distribution of the SE obtained by estimation on 1000 data sets with FO, FOCE and with the two methods of SAEM. We also compare those predicted SE to the empirical SE obtained for each method defined as the standard deviation on the 1000 estimates. Last, we compute bias and root mean square errors (RMSE) for the estimates obtained by FO, FOCE and SAEM.
Results: The SE of PFIM are equivalent to those predicted by SAEM and to the empirical ones obtained with FOCE and SAEM; they are in agreement with the distribution of the SE for FOCE and for the linearization method of SAEM. Regarding FO, the range of the SE and the empirical ones are much larger than the SE of PFIM and those obtained with FOCE or SAEM. The
distribution of the SE for FOCE and the linearization method of SAEM are closed and are in accordance with the empirical SE. The Louis method of SAEM provides larger distribution of SE.
Last, we show large bias and large RMSE for the estimates obtained with FO, whereas with SAEM or FOCE they are good, especially for SAEM on the fixed effects.
Conclusions: We show the relevance of the extension of PFIM for multiple responses models.
Despite the linearization used to compute MF in PFIM, it adequately predicts the SE obtained by FOCE and SAEM but not those of FO which are much larger.
References:
[1] Mentré F, Mallet A, Baccar D. Optimal design in random effect regression models. Biometrika 1997; 84(2):429-442.
[2] Retout S, Mentré F. Optimization of individual and population designs using Splus. Journal of Pharmacokinetics and Pharmacodynamics 2003; 30: 417-443.
[3] Hooker A, Vicini P. Simultaneous optimal designs for pharmacokinetic-pharmacodynamic
[4] Kuhn E, Lavielle M. Maximum likelihood estimation in nonlinear mixed effects model, Computational Statistics and Data Analysis 2005; 49:1020-1038.
[5] Louis TA. Finding the observed information matrix when using the EM algorithm, Journal of the Royal Statistical Society: Series B 1982; 44:226-233. ISSN 0035-9246
Poster: Methodology- Design
Marylore Chenel Comparison of uniresponse and multiresponse approaches of