PREDICTING POTENTIAL CYP450 ENZYME INHIBITION BASED DRUG-DRUG INTERACTION DURING DRUGS PRESCRIPTION USING A COMPUTER AID
Abstract
Introduction: The increase in the number of drugs on the market and concomitant
treatment of co-infections has increased the potential for drug interactions making it
difficult for healthcare professionals to minimize the potential adverse effects of every
drug. Fortunately, Medical Informatics has been evolving to match this increase in
complexity in medical delivery. Pharmacoinformatics has become particularly relevant in
addressing some of the undesirable effects associated with the increased practice of
polypharmacy. Therefore, the major aim of this study was to develop a computer based
pharmacoinformatic tool for use by clinicians and pharmacists in the prediction of in vivo
drug-drug interactions (DDIs) using in vitro data.
Materials and methods: The prototypic tool was developed using Standard Query
Language (SQL) database and Delphi 6.0 as the programming language. Literature
sources were assembled, both as databases and symposia abstracts, original publications
of drug-enzyme or drug-drug interactions for competitive and mechanism-based
inhibition. Sources with validated in vitro methods and having the following parameters:
inhibition constant (Ki); maximum enzyme velocity (Vmax); substrate concentration
needed to reach half maximal velocity (Km); fraction metabolized by cytochrome P450
(fm) and fraction cleared by cytochrome P450 (fh), were considered. Different plasma
concentrations of the inhibitor available to the enzyme site for interaction were tested
with and without taking into account protein binding. The concentrations included the
average maximum plasma concentration (Cmax) and the estimated of maximum
concentration of the inhibitor at entrance to the liver (Iin.max), both bound and unbound. A
pilot study was carried out among 10 doctors and 10 pharmacists to test the medical
relevance of the tool using a questionnaire with scores ranging from 1 (best) to 6 (worst).
Results and discussion: Various drug combinations were tested. The best predictions of
in vivo drug-drug interactions were achieved when the concentration of inhibitor was set
at the unbound maximum concentration at entrance to the liver enzymes with better
overall geometric mean fold error (GMFE) values of 0.68 and overall root mean square
error (RMSE) of 3.13 without considering mechanism-based inhibition (MBI). There was
improvement in overall GMFE (0.49) and RMSE (1.71) for steady-state unbound Cmax
when MBI was incorporated. A preliminary evaluation of the tool by medical
professionals has highly recommended application in private practice and in academia as
a teaching tool, and with mixed reactions in public sector. The survey recommended that
modifications be made on details captured under product composition.
Conclusion: The pharmacoinformatic tool developed during this work is likely to be well
received by the medical community starting as a teaching tool. More drugs used routinely
need to be added, and a high sample size evaluation of relevance and acceptability
conducted. The predictive capacity of the tool had low levels of bias when the
concentration of inhibitor was set at the unbound maximum concentration at entrance to
the liver enzymes. However more work needs to be done to include Drug-Drug
Interactions (DDIs) due to induction and irreversible enzyme inhibition or through
inhibition of other enzymes not considered in this study.