of St John’s wort on human CYP enzymes, in which caffeine, tolbutamide,
dextromethorphan, oral and intravenous midazolam were used for simulta-
neous phenotyping of CYP1A2, CYP2C9, CYP2D6, and intestinal and hepatic
CYP3A4. The study focused on the induction effect of St. John’s wort on a
cocktail of substrates, but it did not validate whether there were any drug
interactions among these substrates after adding midazolam to the previously
validated three substrates cocktail (Bruce et al., 2001; Wang et al., 2001).
The cocktail approach has been proposed as a screening tool for potentialin
vivodrug–drug interactions since in many cases a cocktail approach uses only
metabolic ratios to determine the interactions. The same equivalence criterion
of 80–125% confidence intervals should be applied to the changes in ratios for
plasma or urine to conclude positive or negative outcomes. Negative results
from a cocktail study can eliminate the need for further evaluation of
particular CYP enzymes. However, positive results can indicate the need for
furtherin vivoevaluation to provide quantitative exposure changes (such as
AUC,Cmax), if the initial evaluation only assessed the changes in metabolic
ratios. One must be aware that any modifications from a validated cocktail
protocol, such as dose increase, replacement of different substrates, and/or
adding additional substrates will require additional validation to rule out any
potential drug interactions. For example, Palmer et al. reported that
chlorzoxazone significantly increased the exposure of oral midazolam
(Palmer et al., 2001). In general, the data collected from a properly validated
cocktail study can supplement data from otherin vitroandin vivostudies to
help judge a drug’s potential to inhibit or induce CYP enzymes.
7.4.2.4 Population Pharmacokinetics Population pharmacokinetic analysis
from clinical trials with sparse or intensive blood sampling can be valuable to
evaluate the clinical impact of drug interactions, and to suggest dose
adjustments. The results from population pharmacokinetic studies are
informative, but not conclusive in most cases. They can be conclusive if the
population pharmacokinetic components in clinical trials are adequately
designed to detect significant changes in drug exposure due to drug–drug
interactions. Population pharmacokinetic evaluations may detect signals of
unsuspected drug–drug interactions. When the results from a population
pharmacokinetic analysis are different from that which a conventional
pharmacokinetic drug interaction study have strongly suggested, the results
from the conventional pharmacokinetic study overwrite the population
pharmacokinetic analysis. Because there are many confounding factors in
large-scale clinical trials, population pharmacokinetic studies must have
carefully designed study protocols and sample collections. It is unlikely for
the drug label to claim there are no drug interactions for comedications based
on population pharmacokinetic analysis, especially when a conventional
pharmacokinetic drug interaction study indicates the presence of an interac-
tion. To be acceptable for drug labeling, population pharmacokinetic studies
should be properly designed and be prospective in nature.
DRUG–DRUG INTERACTION STUDIES 229