chemical adulterant of pet food that resulted in widespread kidney failure in pet
animals in Asia and North America. Contaminated pet food was subsequently used
as feed for food animals necessitating determination of melamine withdrawal times
to ensure that the entire chemical was eliminated from animals before human
consumption. A PBPK model allowed data collected in laboratory animals to be
combined with sparse swine data to make realistic estimates of withdrawal times.
A further adaptation of this approach comprises integrating genomic and pro-
teomic data with PBPK models to create systems biology approaches that attempt to
describe chemical and pharmacological actions by building models from the recep-
tor up to the whole animal (Kitano 2002 ). These models are in the early develop-
mental stages but, as they become more refined, they will have the potential to
dramatically advance the field of comparative pharmacology and toxicology. There
will be a resultant increase in drugs better targeted on receptors and disease
mechanisms and with improved safety profiles. Integration of sophisticated quanti-
tative SAR (QSAR) models as input into such approaches will greatly increase our
understanding of drug actions at the molecular level. The challenge and limitation
to developing such models in the present times is the lack of data in biological
systems reflecting the chemical diversity seen in QSAR models. However, systems
biology approaches will allow data to be evaluated at multiple levels and may
provide a method to expand their chemical inference space, that is the multiple
properties of a chemical (molecular weight, solubility, etc.) for which available data
might be applicable. Once validated, they would allow the so-called “in silico”
trials, that is simulated trials undertaken entirely on a computer, which could
potentially develop lead drug compounds using much fewer preclinical laboratory
animal studies. Although a smaller number of live animal studies would still have to
be conducted before approval to validate these predictions, these would be reduced
in number, as a consequence of the efficiency and accuracy inherent to the robust-
ness of the in silico analysis together with to the automated battery of pre-clinical
safety and efficacy tests.
Another field in which computational power has already exerted a major impact
is in the application of advanced statistical tools to analyse the much increased sets
of integrated data. In pharmacology, population pharmacokinetic models are now
being used to define the population factors that determine drug disposition and
activity (Ette and Williams 2007 ). It is increasingly recognised that effective
products must be based on the determinants of individual and sub-group suscepti-
bility. Population pharmacokinetic models allow the integration of kinetic models
describing ADME parameters to be linked to statistical models defining co-variates
identifying the source of variability in a population response. These models were
introduced to veterinary pharmacology a decade ago (Martin-Jimenez and Riviere
1998 ) but only recently have they begun to be considered in drug development and
therefore in regulatory submissions. These models offer the prospect of identifying
factors such as age, weight, gender, disease and breed, which significantly modify
drug disposition or activity. The role that inter- and intra-species differences in
drug transporters and metabolising enzymes exert on ADME parameters is
being actively researched (see chapters, “Pharmacogenomics in Domestic Animal
New Technologies for Application to Veterinary Therapeutics 199