Volterra Series in Pharmacokinetics and Pharmacodynamics

Nonparametric black-box modeling has a long successful history of applications in pharmacokinetics (PK) (notably in deconvolution), but is rarely used in pharmacodynamics (PD). The main reason is associated with the fact that PK systems are often linear in respect to drug inputs, while the reverse is true for many PK/PD systems. In the PK/PD field existing non-parametric methods can deal with linear systems, but they cannot describe non-linear systems. Our purpose is to describe a novel implementation of a general nonparametric model which can represent non-linear systems, and in particular non-linear PK/PD systems. The model is based on a Volterra series, which is an integral series expansion of the response of a system in terms of its kernels and the inputs to the system. In PK we are familiar with the first term of the Volterra series: the convolution of the first kernel of the system (the so-called PK disposition function) with drug input rates. The main advantages of higher order Volterra representations is that they are general representations and can be used to describe and predict the response of an arbitrary (PK/PD) system without any prior knowledge on the structure of the system. The main problem of the representation is that in a non-parametric representation of the kernels the number of parameters to be estimated grows geometrically with the order of the kernel. We developed a method to estimate the kernels in a Volterra-series which overcomes this problem. The method (i) is fully non-parametric (the kernels are represented using multivariate splines), (ii) is maximum-likelihood based, (iii) is adaptive (the order of the series and the dimensionality of each kernel is selected by the method), and (iv) allows for non-equispaced observations (thus allowing a reduction of the number of parameters in the representation, and the analysis of, e.g., PK/PD observations). The method is based on an adaptation of Friedmans’s Multivariate Adaptive Regression Spline method. Examples demonstrate the possible application of the approach to the analysis of different PK/PD systems.

[1]  Carl de Boor,et al.  A Practical Guide to Splines , 1978, Applied Mathematical Sciences.

[2]  L B Sheiner,et al.  Modeling a bivariate control system: LH and testosterone response to the GnRH antagonist antide. , 1996, The American journal of physiology.

[3]  Leo Breiman,et al.  [The ∏ Method for Estimating Multivariate Functions from Noisy Data]: Response , 1991 .

[4]  R. Tibshirani,et al.  The II P method for estimating multivariate functions from noisy data , 1991 .

[5]  S. Gupta,et al.  Statistical decision theory and related topics IV , 1988 .

[6]  G. Segre,et al.  Kinetics of interaction between drugs and biological systems. , 1968, Il Farmaco; edizione scientifica.

[7]  D. Verotta,et al.  LINEAR MIXED-EFFECT MULTIVARIATE ADAPTIVE REGRESSION SPLINES APPLIED TO NONLINEAR PHARMACOKINETICS DATA , 2000, Journal of biopharmaceutical statistics.

[8]  R. Stephenson.  A and V , 1962, The British journal of ophthalmology.

[9]  D. Brillinger.  The maximum likelihood approach to the identification of neuronal firing systems , 2006, Annals of Biomedical Engineering.

[10]  Lewis B. Sheiner,et al.  A general conceptual model for non-steady state pharmacokinetic/Pharmacodynamic data , 2006, Journal of Pharmacokinetics and Biopharmaceutics.

[11]  C. R. Deboor,et al.  A practical guide to splines , 1978 .

[12]  Davide Verotta,et al.  Two constrained deconvolution methods using spline functions , 1993, Journal of Pharmacokinetics and Biopharmaceutics.

[13]  H. Akaike.  A new look at the statistical model identification , 1974 .

[14]  J. Freidman,et al.  Multivariate adaptive regression splines , 1991 .

[15]  Vito Volterra,et al.  Theory of Functionals and of Integral and Integro-Differential Equations , 2005 .

[16]  S. Sardy,et al.  Triogram Models , 1996 .

[17]  J. Rice.  Choice of smoothing parameter in deconvolution problems , 1986 .

[18]  John Rice,et al.  Deconvolution of Microfluorometric Histograms with B Splines , 1982 .

[19]  Leo Breiman,et al.  Classification and Regression Trees , 1984 .

[20]  E. Hannan.  Rational Transfer Function Approximation , 1987 .

[21]  Grace Wahba,et al.  Statistical Decision Theory and Related Topics III , 1982 .

[22]  D Verotta,et al.  Concepts, properties, and applications of linear systems to describe distribution, identify input, and control endogenous substances and drugs in biological systems. , 1996, Critical reviews in biomedical engineering.

[23]  H. Chipman,et al.  Bayesian CART Model Search , 1998 .

[24]  G. Wahba,et al.  The computation of generalized cross-validation functions through householder tridiagonalization with applications to the fitting of interaction spline models , 1989 .

[25]  V. Marmarelis.  Identification of nonlinear biological systems using laguerre expansions of kernels , 1993, Annals of Biomedical Engineering.

[26]  D. Titterington.  Common structure of smoothing techniques in statistics , 1985 .

[27]  F. O’Sullivan.  A Statistical Perspective on Ill-posed Inverse Problems , 1986 .

[28]  Edmund Taylor Whittaker.  On a New Method of Graduation , 1922, Proceedings of the Edinburgh Mathematical Society.

[29]  Trevor Hastie,et al.  Polynomial splines and their tensor products in extended linear modeling. Discussion and rejoinder , 1997 .

[30]  W J Jusko,et al.  Physiologic indirect response models characterize diverse types of pharmacodynamic effects , 1994, Clinical pharmacology and therapeutics.