Saddle-Reset for Robust Parameter Estimation and Identifiability Analysis of Nonlinear Mixed Effects Models

Parameter estimation of a nonlinear model based on maximizing the likelihood using gradient-based numerical optimization methods can often fail due to premature termination of the optimization algorithm. One reason for such failure is that these numerical optimization methods cannot distinguish between the minimum, maximum, and a saddle point; hence, the parameters found by these optimization algorithms can possibly be in any of these three stationary points on the likelihood surface. We have found that for maximization of the likelihood for nonlinear mixed effects models used in pharmaceutical development, the optimization algorithm Broyden–Fletcher–Goldfarb–Shanno (BFGS) often terminates in saddle points, and we propose an algorithm, saddle-reset, to avoid the termination at saddle points, based on the second partial derivative test. In this algorithm, we use the approximated Hessian matrix at the point where BFGS terminates, perturb the point in the direction of the eigenvector associated with the lowest eigenvalue, and restart the BFGS algorithm. We have implemented this algorithm in industry standard software for nonlinear mixed effects modeling (NONMEM, version 7.4 and up) and showed that it can be used to avoid termination of parameter estimation at saddle points, as well as unveil practical parameter non-identifiability. We demonstrate this using four published pharmacometric models and two models specifically designed to be practically non-identifiable.

[1]  R. Fletcher,et al.  A modified Newton method for minimization , 1977 .

[2]  L. Aarons,et al.  What do we mean by identifiability in mixed effects models? , 2015, Journal of Pharmacokinetics and Pharmacodynamics.

[3]  Danny C. Sorensen,et al.  On the use of directions of negative curvature in a modified newton method , 1979, Math. Program..

[4]  Lewis B. Sheiner,et al.  Evaluation of methods for estimating population pharmacokinetic parameters. I. Michaelis-menten model: Routine clinical pharmacokinetic data , 1980, Journal of Pharmacokinetics and Biopharmaceutics.

[5]  Leon Aarons The estimation of population pharmacokinetic parameters using an EM algorithm. , 1993, Computer methods and programs in biomedicine.

[6]  Preconditioning of Nonlinear Mixed Effects Models for Stabilisation of Variance-Covariance Matrix Computations , 2016, The AAPS Journal.

[7]  V. K. Siripuram,et al.  Deterministic identifiability of population pharmacokinetic and pharmacokinetic–pharmacodynamic models , 2017, Journal of Pharmacokinetics and Pharmacodynamics.

[8]  R. Gomeni,et al.  A two-step iterative algorithm for estimation in nonlinear mixed-effect models with an evaluation in population pharmacokinetics. , 1995, Journal of biopharmaceutical statistics.

[9]  W. Vach,et al.  Impact of CYP2C8*3 on paclitaxel clearance: a population pharmacokinetic and pharmacogenomic study in 93 patients with ovarian cancer , 2011, The Pharmacogenomics Journal.

[10]  R. Fletcher Practical Methods of Optimization , 1988 .

[11]  M. Karlsson,et al.  Population Pharmacokinetic Modelling and Estimation of Dosing Strategy for NXY-059, a Nitrone Being Developed for Stroke , 2005, Clinical pharmacokinetics.

[12]  Neil D. Evans,et al.  Parameter Identifiability of Fundamental Pharmacodynamic Models , 2016, Front. Physiol..

[13]  D. Bates,et al.  Newton-Raphson and EM Algorithms for Linear Mixed-Effects Models for Repeated-Measures Data , 1988 .

[14]  J F Boisvieux,et al.  Alternative approaches to estimation of population pharmacokinetic parameters: comparison with the nonlinear mixed-effect model. , 1984, Drug metabolism reviews.

[15]  Roger Fletcher,et al.  Practical methods of optimization; (2nd ed.) , 1987 .

[16]  Claudio Cobelli,et al.  A Priori Identifiability Analysis in Pharmacokinetic Experiment Design , 1981 .

[17]  S. S. Wilks The Large-Sample Distribution of the Likelihood Ratio for Testing Composite Hypotheses , 1938 .

[18]  A Racine-Poon,et al.  A Bayesian approach to nonlinear random effects models. , 1985, Biometrics.

[19]  Arild Thowsen,et al.  Structural identifiability , 1977, 1977 IEEE Conference on Decision and Control including the 16th Symposium on Adaptive Processes and A Special Symposium on Fuzzy Set Theory and Applications.

[20]  R J Keizer,et al.  Modeling and Simulation Workbench for NONMEM: Tutorial on Pirana, PsN, and Xpose , 2013, CPT: pharmacometrics & systems pharmacology.

[21]  Mats O Karlsson,et al.  Models for time-varying covariates in population pharmacokinetic-pharmacodynamic analysis. , 2004, British journal of clinical pharmacology.

[22]  Marie Davidian,et al.  Smooth nonparametric maximum likelihood estimation for population pharmacokinetics, with application to quinidine , 1992, Journal of Pharmacokinetics and Biopharmaceutics.

[23]  T H Grasela,et al.  Neonatal population pharmacokinetics of phenobarbital derived from routine clinical data. , 1985, Developmental pharmacology and therapeutics.

[24]  Robert J. Bauer,et al.  Monte Carlo Parametric Expectation Maximization (MC-PEM) Method for Analyzing Population Pharmacokinetic/Pharmacodynamic Data , 2004 .

[25]  V Shivva,et al.  An Approach for Identifiability of Population Pharmacokinetic–Pharmacodynamic Models , 2013, CPT: pharmacometrics & systems pharmacology.

[26]  Tim Hesterberg,et al.  Introduction to Stochastic Search and Optimization: Estimation, Simulation, and Control , 2004, Technometrics.

[27]  David J. Spiegelhalter,et al.  Estimation of population pharmacokinetics using the Gibbs sampler , 1995, Journal of Pharmacokinetics and Biopharmaceutics.

[28]  Robert J. Bauer,et al.  A survey of population analysis methods and software for complex pharmacokinetic and pharmacodynamic models with examples , 2007, The AAPS Journal.

[29]  M. J. D. Powell,et al.  Nonlinear Programming—Sequential Unconstrained Minimization Techniques , 1969 .

[30]  Alexander H. G. Rinnooy Kan,et al.  A stochastic method for global optimization , 1982, Math. Program..

[31]  V. Carey,et al.  Mixed-Effects Models in S and S-Plus , 2001 .

[32]  F Mentré,et al.  Hyperparameter estimation using stochastic approximation with application to population pharmacokinetics. , 1988, Biometrics.