future. In the absence of integrated computational methods, large genomic datasets
could be analysed only by professional programmers possessing advanced com-
puter programming skills.
Also of great significance have been advances in the sophistication, range and
“user-friendliness” of pharmacokinetic modelling software programmes. There
have been further parallel developments of pharmacodynamic models. Together
these permit an increased throughput in the conduct of ADME and explorative
studies that previously were time, as well as cost, prohibitive. Of equal if not greater
significance, this increased ability to conduct such studies, using techniques which
can be grouped under the umbrella term ofpharmacometrics(Ette and Williams
2007 ), increases our understanding of the biological determinants of ADME
processes. This, in turn, facilitates development of structure–activity relationships
(SAR) for ADME as well as pharmacodynamic endpoints.
It is not possible to track all the innumerable developments occurring in these
areas across the fields of pharmacology and toxicology, where such developments
fall under the description of computational toxicology (Elkins 2007 ). A pivotal
report from the U.S. National Research Council ( 2007 ) outlined a mechanistic-
based and quantitative approach for toxicology in the twenty first century that melds
in vitro testing with pharmacokinetic and pharmacodynamic modelling to create a
more realistic, defensible and precise approach to chemical risk assessment. How-
ever, the integration of such diverse approaches is tedious, expensive, time con-
suming and fraught with difficulties when attempts to adopt such approaches are
made by regulatory authorities. This is especially difficult for constantly evolving
techniques. Despite their use for formal regulatory approval, they are outstanding
tools for probing mechanisms of disease and drug action, as well as screening for
and developing safe and effective drug candidates for submission through the
traditional regulatory channels. Some such techniques may not be used for formal
drug approval, but will nevertheless be adapted to compounds and devices that
perform better in regulatory testing as well as in the market place.
One example of enablement through the development of such powerful software
is the increasing application of mechanistic physiological based pharmacokinetic
models (PBPK) to more compounds, leading to integration of biological processes
with quantitative endpoints (Reddy et al. 2005 ). PBPK models, rather than being
founded on empirical curve-fitting approaches, are constructed using biological
data from organs linked by systemic blood flow. They lend themselves well to
extrapolations between species, as species-specific physiological and metabolic
variables (e.g. blood flow to specific organs, hepatic biotransformation enzymes)
can readily be incorporated. Recent examples of the PBPK approach in veterinary
medicine include the prediction of drug withdrawal times for oxytetracycline in
sheep and sulfamethazine in swine (Craigmill 2003 ; Buur et al. 2006 ). These
approaches allow simulation of the effect of inter- and intra-species variability on
the ultimate outcome. They can greatly improve the design of experimental studies
and even of clinical trials. As seen with the compound melamine, they also allow
cross species extrapolations to be made based on underlying physiology in the
face of minimal data (Buur et al. 2008 ). As discussed earlier, melamine was the
198 J.E. Riviere