The Parachute Principle of Drug Interactions
In an article published in theDecember 20, 2003, issue of theBritish Medical Journal, Gordon Smithand Jill Pell pointed out that no double-blindrandomized controlled trials havebeen completed on the efficacy of theparachute in preventing death from the"gravitational challenge" of jumping outof an airplane. They suggest that the benefitof parachutes in such situations ispurely anecdotal; it is based merely oncommon sense instead of scientificinvestigation, and thus it cannot be trustedto be true.
Although their article was tongue-in-cheek,it raises an important issue in theevaluation of drug interactions—namely,the extent to which one can useinformed common sense and reason tomake clinical decisions about drug interactionsin the absence of actual scientificstudies. Put another way, are inferencesabout the danger of specific druginteractions justified based on the knowninteractivepropertiesof the 2drugs in theabsence ofpublishedstudiesinvolvingthe 2 drugsused concurrently?
When drug interactionsfirst became the object of intense studyin the mid 1960s and the 1970s,we knewonly that some drugs were "enzymeinhibitors" and some were "enzymeinducers," with little information on thespecific enzymes involved. It was thereforeoften mysterious why a particularenzyme inhibitor reduced the metabolismof one drug and not another.Inferences and generalizations wereproblematic during this time, becausethe patterns were obscured by our ignoranceof CYP450 isozymes and membranetransport proteins such as P-glycoprotein.
As the mechanisms of drug interactionsbegan to be understood in the1980s and the 1990s, it became possibleto predict that certain drug pairs wouldinteract even before the interactionswere studied. Old habits die hard, however,and even to this day some peoplecontinue to insist that the only realityregarding drug interactions comes fromactual clinical studies of the 2 drugs. Thisis knowledge trumping wisdom.
This focus on the published drug interactionliterature to the exclusion of commonsense has also been reinforced bycalls for making all drug-therapy decisionsbased on "evidenced-based medicine." But "evidenced-based" does notapply to the drug-drug interaction literature,where perhaps 90% of the reportsare in the form of pharmacokinetic studiesin healthy subjects and isolated casereports. Controlled outcome studies ofdrug interactions are rare, and we areusually forced to make clinical decisionsabout drug interactions with less informationthan we would like.
Inferences, therefore, must be madebased on what we know about theinteractive properties of drugs. Forexample, the metabolism of carbamazepineis known to be highly sensitiveto inhibition by CYP3A4 inhibitors.So if a new drug comes on the marketthat is known to inhibit CYP3A4, weknow that it is highly likely that it willcause carbamazepine toxicity even inthe absence of actual studies. If someonedrives a Buick off a high cliff into theocean and dies instantly, we do nothave to have someone else drive aChevrolet off the same cliff to see if itwill also be lethal. Inferences are justifiedfrom known data.
Similarly, we know that P-glycoproteininhibitors are likely to increase serumdigoxin levels; enzyme inducers are likelyto reduce verapamil levels; CYP2C9inhibitors are likely to increase warfarinlevels; CYP1A2 inhibitors are likely toincrease tizanidine levels; and indirect-actingsympathomimetics are likely toresult in a hypertensive crisis in patientson nonselective monoamine oxidaseinhibitors. Many patients have beenharmed by predictable (but unstudied)drug interactions such as these. Thus, tominimize the risk to the patient, we mustassume that these interactions will occurand act accordingly. Even for those rareoccasions when subsequent studyproves that the 2 drugs do not interact,we still have acted appropriately, giventhe data we had at the time.
The "parachute principle" of druginteractions says that we have reacheda point where—for many drugs withwell-known interactive properties—it ispossible to predict other drugs withwhich they are likely to interact. Todemand actual clinical studies beforetaking action is like asking for controlledstudies of the efficacy of parachutesbefore recommending that parachutesbe used by people jumping out of airplanes.In the absence of data, informedcommon sense may be our only defenseagainst an adverse outcome.
Drs. Horn and Hansten are both professorsof pharmacy at the Universityof Washington School of Pharmacy.For an electronic version of this article,including references if any, visitwww.hanstenandhorn.com.