Opinions and research priorities

intl.elsevierhealth.com/journals/thre

Opinions and research priorities

Henry I. Bussey a,*,B, Victor Tapson b, Richard O. Cannon III c,Paul E. Nolan Jr. d, Brian Gage e, Scott R. Penzak f, Sarah A. Spinler g,Peter De Smet h, Ann Wittkowsky i, Alison Pataky Lee j, Edzard Ernst k,Victor J. Marder l

aCollege of Pharmacy, The University of Texas at Austin, Austin,Texas and The University of Texas Health Sciences Center at San Antonio, San Antonio,Clinical Pharmacy Programs MC 6220, 7703 Floyd Curl Drive, San Antonio, Texas 78229-3900, United StatesbDivision of Pulmonary and Critical Care Medicine, Duke University Medical Center,Durham, North Carolina, United StatescNational Heart, Lung, and Blood Institute, National Institutes of Health, Bethesda, Maryland,United StatesdDepartment of Pharmacy and Science, College of Pharmacy, University of Arizona, Tucson,Arizona, United StateseInternal Medicine, Washington University School of Medicine, St. Louis, Missouri, United StatesfClinical Pharmacokinetics Research Laboratory, NIH Clinical Center Pharmacy Department,Bethesda, Maryland, United StatesgClinical Pharmacy, Philadelphia College of Pharmacy, University of the Sciences in Philadelphia,Philadelphia, Pennsylvania, United StateshThe Scientific Institute of Dutch Phamacists, The Hague, The NetherlandsiAnticoagulation Services, University of Washington Medical Center, Seattle, Washington, United StatesjWalled Lake, Michigan, United StateskComplementary Medicine, Peninsula Medical School, Exeter, Devon, EnglandlMedicine & Neurology, David Geffen School of Medicine at UCLA, Los Angeles, California, United States

Received 17 May 2005; received in revised form 17 May 2005; accepted 24 May 2005Available online 15 August 2005

Thrombosis Research (2005) 117, 155—169

0049-3848/$ - see front matter D 2005 Elsevier Ltd. All rights reserved.doi:10.1016/j.thromres.2005.05.030

* Corresponding author.E-mail address: [email protected] (H.I. Bussey).

B Co-Founder and President of ClotCare.

H.I. Bussey et al.156

Introduction

The following is a summary of presenters’ commentsfrom this session either in the form of a manuscriptprepared by the presenter or as edited from thetranscript of the session. Presenters were asked toidentify their top research priorities in this area. Themoderator for this session (HIB) later compiled theseresearch suggestions into a prioritized list with theranking of each proposed initiative being based onhow many presenters recommended a given initia-tive and how much discussion the initiative gener-ated (Table 1). It is interesting to note that whatappeared to be the top priorities from this sessionare the same priorities that were identified by afive-year toxicological study in the United Kingdomthat was published more than seven years earlier[1]. That study concluded that bSurveillance of alarge population is needed. . .to identify the uncom-mon and unpredictable adverse effects which arepotentially serious. . .(and that)a single regulatorysystem is required which would ensure the safetyand quality of all herbal remedies and food sup-

Table 1 Prioritized research recommendations(from the perspective of the session moderator-HIB)

1. Create a national registry to identify and define theproblem presented by dietary supplements and/orherbal agents.

2. Evaluate and improve web-based information resourcesto assure that patients and clinicians can obtain accurateinformation regarding these supplements.

3. Pursue standards that will assure consistency of contentand accurate labeling of available agents.

4. Determine the effects of various supplements onmetabolizing hepatic isoenzymes.

5. Study whether the more commonly used supplementsalter the pharmacodynamics and/or pharmacokinetics ofwarfarin.

6. Quantify the extent of use of supplements amonganticoagulated patients.

7. Evaluate the impact of programs designed to educatepatients about the potential risks of using these agents inconjunction with antithrombotic agents.

8. Initiate and evaluate methods to enhance the extent towhich health care providers ask patients about use ofsupplements.

9. Conduct prospective trials of agents that have beenreported to interact with warfarin.

10. Focus on development of alternatives to warfarin thatmay be affected less by supplements.

11. Evaluate the effect of vitamin C as an antiplateletagent and its role in preventing stroke.

12. Explore the effect of optunia (prickly pear cactus) onglucose tolerance, antiplatelet effects, hypoglycemicand hypolipidemic effects.

13. Explore the effect of fucoidan (sea weed) as anantiplatelet, anticoagulant, and inhibitor of neointimalproliferation.

plements. . .Q [1]. Finally, following the individualpresentations is a record of the panel discussion.

Presentations

Dr. Bussey

As we approach the end of this two-day confer-ence, this session gives us the opportunity to have anumber of leading authorities share their viewswith us as to what they see as the top researchpriorities in this area. The organizers of this sessionhave selected today’s presenters so that theyrepresent a wide array of clinical disciplines.

As I was thinking about all of the issuespresented in the sessions yesterday, I realized thatalmost any question you can ask in this area can beanswered with a fairly short answer of maybe,maybe not. With few exceptions, we really don’tknow a lot about dietary supplements and how theymay interact with anticoagulant theray. Therefore,I thought I would start off by telling you about apatient that we had in our clinic. The patient washospitalized with severe bleeding while takingwarfarin and ginkgo biloba. Our hematologist, thehead of our clinic, who is an expert in bleeding andclotting problems, was unable to stop the bleeding.Everything he tried, did not work. The patienteventually bled to death. Several case reports havesuggested that gingko biloba may increase thebleeding risk in warfarin-treated patients; but themechanism of this interaction is not clearly estab-lished. After the fatal hemorrhage that occurred inour patient, one of the other clinical pharmacists inour clinic found a report of ginkgo biloba beinglaced with phenylbutazone; a non steroidal anti-inflammatory drug with a bleeding risk greatenough that it is not used any longer. Could thephenylbutazone have played a role in the fatalhemorrhage in our patient? Maybe, maybe not.

Our policy now, particularly with the head of ourhematology group is, if a warfarin patient is takingginkgo biloba, they either stop it or they findsomebody else to manage their therapy. Now, Irealize that, in this atmosphere of being non-judgmental toward the patient and being accom-modating of their values and so forth, that may be apolitically incorrect thing to do. But if we allow ourpatients to take ginkgo biloba and they bleed, thenare we not liable for that bleed? To me, it is nodifferent than patients who continue to missappointments. We will not agree to continue tobe responsible for their care if they continuallyinsist on doing things that we believe may be

Opinions and research priorities 157

harmful. We may not be able to prove exactly whyit is potentially harmful but, on balance, we are notwilling to accept that responsibility.

Dr. Tapson

(note: Dr. Tapson was unable to give his presenta-tion and the following is the discussion of Dr.Tapson’s slides as presented by Dr. Bussey) Thereare concerns about bleeding tendencies with inhi-bition of platelet aggregation and also interactionswith warfarin.

Considering complementary and alternativemedicine, interactions with warfarin could poten-tially be pharmacokinetic or pharmacodynamic innature, and I think we have seen both. Regardlessof concomitant medications, and regardless ofinformation available about them, careful moni-toring is crucial for the patients on warfarin. Weneed to be following these patients closely anyway,and if the interaction is one that affects the INR,and that is the main mechanism that we areworried about, then we can simply follow themmore carefully and adjust the dose of warfarin ifnecessary.

Inadequate medication reporting by the patientmust be addressed. Clinicians must actively inquireabout the use of alternative agents and comple-mentary medicines, something that I know Dr.Wittkowsky addressed in her discussion yesterday.In Dr. Tapson’s approach, anticoagulation clinicsremain crucial.

As we have already heard, these agents (herbaland dietary supplements) are not standardized interms of purity or potency. They are also of varyingstrengths with batch to batch variability, and, insome cases, labeled as containing ingredients thatthey do not contain. Authentication of plantmaterials should be required and we have heardsome already about today. Principles of goodmanufacturing practice should be imposed, but Iam not sure they are. Then, the obligations of theFDA need to be clearly defined and measuresshould be taken to see that these obligations aremet.

In a lot of ways, I think we could argue that mostof these agents could be viewed as drugs, drugswithout known safety and efficacy data, whichperhaps would require written informed consent, ifour patients are going to be using them. Of theagents that I see used in my clinical practice,ginkgo biloba and ginseng are two that I’veidentified as agents that I find most worrisome inwarfarin patients. In addition to evaluating warfa-rin interactions, we should realize that warfarin is a

significant part of the problem. If you are notdealing with the patients on warfarin on a day today basis, perhaps you may not be aware of the riskhere. Warfarin is perhaps the most dangerous drugthat we use on an outpatient basis with chronictherapy. It is also the most highly litigated drug,and I would argue that, if it were just beingapproved today, only specially designated clinicianswould be allowed to use it, because it is sodangerous. Because warfarin use represents sucha problem, perhaps we should be focusing on thenew agents, direct thrombin inhibitors, direct anti-Xa inhibitors, and other new agents that may nothave the extent of interactions that we see withwarfarin. Recently, however, we heard that xime-lagatran was not approved by the FDA, partlybecause of liver toxicity deaths, and also partlybecause of a suspicion that it might actually inducemyocardial infarction (MI). So, the new agentsmight not have all the answers we hoped for.

Also, the newer agents are frequently referredto as not having drug interactions, but there havebeen one or two quite significant drug interactionsidentified with ximelagatran. So, I think the newagents may be improvements, but they are notgoing to eliminate the problems.

The most important research priority in this areais to identify and define what interactions existbetween dietary supplements and warfarin. Inorder to do this a large scale registry of the useof these agents in anticoagulated patients needs tobe developed. One estimate would be that we needperhaps 5000 patients in such a registry in order toevaluate the potential interactions. In order tohave such a registry work, we would need todevelop medication lists and case report formsthat would include medication lists, INR values,bleeding time, and bleeding and/or thromboem-bolic events. Such data could then be used to try tosort out which agents may tie in with the occur-rence of complications. Additionally, it would behelpful if we also could characterize patientsaccording to hepatic enzyme polymorphisms sincedifferent polymorphisms may represent a greaterpredisposition to certain interactions. Althoughroutinely gathering such data might be relativelyexpensive and/or cumbersome, it would helpcharacterize the scope of the problem.

I think establishing a registry to collect data onsupplement use and potential interactions is clearlysomething we could do, and we already have somenetworks in place that might facilitate this. Interms of proposed research efforts, we also needprospective studies of primary potential offenders.Table 2 list some examples of potential offendersthat I believe should be studied. Although it has

Table 2 Potentially warfarin-interacting supple-ments that warrant prospective study as identifiedby Dr. Tapson

Ginseng (American, Oriental, Siberian)Green tea (Camellia sinensis)Danshen (Saliva miltiorrhiza-root)Dong Quai (Angelica sinenisis)Coenzyme-QKangen-karyn (KGK)Devil’s claw (Harpagophytum procumbens)PapainHigh-dose vitamin E


been suggested that these agents interact withwarfarin, the nature of the interactions is not welldocumented. Do they interact with warfarin?Maybe, maybe not.

Dr. Cannon

Atrial fibrillation is the most common arrhythmiarequiring treatment, experienced by over twomillion patients in the U.S. each year [2,3]. Theprevalence of atrial fibrillation increases with age:approximately 4% of persons over 60 years of ageand 9% of persons over 80 years of age experienceeither paroxysmal or chronic atrial fibrillation,-sometimes unrecognized by the patient until anembolic event occurs due to clots dislodged fromthe fibrillating left atrium. In addition to age, riskfactors for atrial fibrillation include hypertension,diabetes, obesity and most importantly, underlyingheart disease (valvular heart disease; congestive,hypertrophic and constrictive cardiomyopathies;coronary artery disease) [4]. In many patients,however, atrial fibrillation may present withoutunderlying heart disease in otherwise healthypeople (bloneQ atrial fibrillation). The morbidityand mortality risks of atrial fibrillation are largelydetermined by the nature and severity of underly-ing heart disease, but the most devastating conse-quence of the arrhythmia is stroke [5,6]. Theannual risk of stroke in Framingham study partici-pants with atrial fibrillation ranged from approxi-mately 2% in persons aged 50 to 59 years to almost24% in those aged 80 to 89 years, with greater riskin patients with associated structural heart disease[6]. Chronic anticoagulation by warfarin and othercoumarin derivatives that antagonize vitamin Kdecreases the risk of stroke and other thromboem-bolic events [7—9]. This therapy, however, requiresfrequent monitoring because of: 1) the narrowtherapeutic window (as measured by the interna-tionalized normalized ratio-INR), above and belowwhich are increased risks of bleeding and throm-boemboli, respectively, and 2) fluctuations in INRs

over time, in part because of changes in consump-tion of vitamin K-containing foods and effects ofmedications—and patients with heart disease oftentake many— on warfarin pharmacokinetics. As aresult, patients taking warfarin must have INRschecked from a blood sample every few weeks,with adjustments in warfarin dosage often requiredto maintain INR values between 2—3 in patientswithout serious underlying heart disease, or higherin certain high-risk subgroups, such as patients withmechanical heart valves [9].

Even with frequent monitoring, major bleedingcontinues to impart significant morbidity withfrequent hospitalizations and occasional mortality[10]. Because of the bleeding risks, physicians maybe reluctant to initiate warfarin therapy for chronicuse in some patients with atrial fibrillation whomay be at risk for thromboemboli, especially ifelderly. In addition, some patients may declinewarfarin therapy because of the risks or theinconvenience of frequent monitoring [11].

Cardiologists have only recently become awarethat dietary supplements may affect warfarinanticoagulation, either decreasing (e.g., St. John’swort, ginseng, garlic, soy products) or increasing(e.g., gingko, ginger products, danshen) INRs [12].Further, case reports of catastrophic bleeding orthrombotic events in patients taking warfarin andherbal dietary supplements suggest that on occa-sion, coadministration of warfarin and some dietarysupplements may be clinically relevant [12]. Non-prescription supplement use by patients with heartdisease (with or without atrial fibrillation) iscommon, especially anti-oxidant vitamins, folate,fish oil and coenzyme Q10 [13—15]. Herbal supple-ment use is less clear, but may be as high as a thirdof patients by some estimates (possibly an under-estimation, as some patients may not wish to tellhealth care professionals about nonprescriptionsupplement use), and may also be common inanticoagulation clinics [13—16]. For patients onwarfarin therapy, the widespread belief is thatfrequent INR checks will obviate concerns aboutsupplement use, known or unknown, just as forchanges in diet, as any change in the INR can beaddressed by increasing or decreasing the warfarindose before complications occur.

Under investigation are new drugs for chronicanticoagulation that target specific coagulationfactors and are not protein cofactor dependent.Because of more predictable effects on hemosta-sis, agents such as the thrombin (factor II)inhibitor melagatran do not require laboratorymonitoring, potentially representing a majoradvantage over warfarin with regards to patientacceptance and compliance, and physician enthu-


siasm for its use, if bleeding risk is low. The mostwidely investigated agent to date is ximelagatran,the orally active pro-drug for melagatran [17].Based on favorable clinical trial data, ximelaga-tran is approved in several European countries forprevention of venous thromboemboli in patientsundergoing knee or hip surgery. Other trials haveshown efficacy of ximelagatran in treatment andsecondary prevention of acute lower-extremitydeep venous thrombosis, and prevention ofthromboemboli after acute myocardial infarction.A series of clinical trials in the internationalStroke Prevention with an Oral Thrombin Inhibitorin Atrial Fibrillation (SPORTIF) program haveshown efficacy comparable to warfarin therapy,with somewhat less bleeding risk [17]. Of concernis a 5—10% frequency of liver enzyme elevation,although the clinical implications of this findingare unclear as enzyme levels generally normalizewith continued treatment. One clinical trialcomparing ximelagatran with low-molecular-weight heparin and warfarin for treatment ofdeep venous thrombosis reported an increasedincidence of coronary events (hospitalization forangina or myocardial infarction) in ximelagatran-treated patients compared with enoxaparin/war-farin-treated group [18]. Clinical trials of ximela-gatran treatment in atrial fibrillation, with longerduration of treatment, have not reported anincreased risk of coronary events [19]. Otheragents that may achieve protection from strokein patients with atrial fibrillation without needfor chronic monitoring, such as the factor Xainhibitor idraparinux among others), are alsounder investigation. Lessons learned form inter-actions between warfarin and dietary supple-ments, with adverse outcomes in some patients,may extend to coagulation factor-specific inhibi-tors and other novel approaches to anticoagula-tion. FDA approval of agents to prevent strokeand other thromboembolic risk that do notrequire monitoring would likely increase theacceptance of chronic anticoagulation therapyby patients in atrial fibrillation and, accordingly,the numbers of patients treated. If this results inless frequent interactions with health care pro-fessionals (as intended) and if important interac-tions with dietary supplements exist for theseagents, adverse effects on hemostasis may not berecognized until serious bleeding or thrombosisoccurs.

Specific areas for research include: 1) betterdata on frequency of nonprescription supplementuse by patients with cardiovascular disease, espe-cially those on anticoagulant therapy, 2) impact ofcounseling by health care providers on supplement

use, and 3) potential interactions between coagu-lation factor-specific inhibitors and the most com-monly used supplements to determine implicationsfor pharmacokinetics and hemostasis.

Dr. Nolan

As Dr. Bussey indicated, there are three supple-ments or products, if you will, upon which I amgoing to focus, and with which I have had varyingdegrees of clinical and research experience. Thefirst item which I will focus upon is ascorbic acid.We heard a little bit about this yesterday. It is awater soluble vitamin which we need to take inon a daily basis. It has known antioxidantproperties, and perhaps some under-appreciatedantiplatelet effects. These antiplatelet effectsmay, in part, be mediated by increases in nitricoxide, either in polymorphonucleocytes or inendothelial cells, which can affect platelet aggre-gation and adhesion. From an epidemiologicalperspective, there are at least two studies thathave shown inverse relationship with either dosesor concentrations of vitamin C and the incidenceof stroke.

In a related issue of vitamin C having an anti-clotting effect, we, at the University of Arizona,have been involved in some pioneering work withartificial heart technology with a device that wasrecently approved by the FDA as a bridge totransplantation. The device is a replacement forthe normal heart with two artificial ventricles.More importantly, within these ventricles, residefour stainless steel valves, as well as polyurethanethat is in contact with the blood. In the manage-ment of these patients, I utilize a multi-drugtherapy approach with a variety of somewhatnovel monitoring techniques. Since 1999, I addedvitamin C to this regimen in order to enhance theabsorption of iron, because these patients all havea low grade hemolytic anemia as a function ofhigh sheer stress on blood vessels. I have used thisapproach in 67 consecutive patients we have onlywitnessed two strokes. More interestingly, whenwe break this down into vitamin C users and novitamin C users, in a group of 31 consecutivepatients to whom we administered vitamin C, themean aspirin dose was 272 mg with a median of127. I report the median, because the doses arenon-normally distributed. In the non-vitamin Cgroup, a group of 24, the first 24 since Septemberof 1994, the dose of aspirin was much higher, bothat the mean and median level, and the results arehighly significant, suggesting an additive, syner-gistic, or what have you, affect with our anti-


platelet regimen. From a research direction, andwhere there is smoke there is fire, there appearsto be enough data, both from an antiplateletperspective, and from an epidemiological perspec-tive to consider, perhaps, prospective randomizedtrials looking at vitamin C in addition to perhapsother platelet agents for the primary or secondaryprevention of stroke.

Moving on, another research interest of oursinvolves a substance obtained from the prickly pearcactus, a particular species which is native tosouthern Arizona, in particular. It is made up ofpads that contain aureoles and spines. It has amature fruit and also flowers. This cactus, callopuntia engelmannii, has been used by the NativeAmerican population and the Hispanic population insouthern Arizona and northern Mexico for many,many years. And there are varying reports suggest-ing that it has hypoglycemic effects, hypolipidemiceffects, antioxidant effects. Very recently it hasbeen shown to be good for alcohol hangover,perhaps due to anti-inflammatory effects, andthere are some recent data to suggest that it doeshave some mild antiplatelet effects. There areproblems, however, with respect to what areactually the active moieties in opuntia, and itwould be useful to help identify these activeconstituents that may posses either antiplateletor anti-inflammatory effects. A useful dosage formalso would be needed, following the determinationof active constituents, before follow up clinicaltrials could be conducted.

Last, our laboratory research group has beenlooking at fucoidins. Fucoidins are extracted frombrown seaweed which is a staple of the Japanesediet. But it is also a supplement for a variety ofindications or uses, if you will. Indications is toostrong a word. There are really important pharma-cological effects in the fucoidin class. These aresulfated polysaccharides, not too different fromheparins, and they possess anticoagulant effects bybinding to antithrombin and they also affectheparin co-factor 2. They have antiplatelet effects,in part because they block the effect of selectins,and there are some data to suggest that they canprevent neointimal hyperplasia in a rabbit model ofarterial disease. From a research direction, wewould like to examine the true activity of theseagents, although it would be useful to look atpharmacodynamic studies in combination withexisting antithrombotic and antiplatelet agents.We are not certain, however, if these agentns areorally available; they have been given mostlyparenterally. Alternatively, given their potentialpotency, they may be useful as stand alone alter-natives to existing therapies.

Dr. Gage

Here’s where I recommend we go from here:

First we should continue to do well done andwell-reported case studies. These carefulobservations are particularly important forhypothesis generation.

Second we need to get more information on the pre-valence of use. Registries or other mecha-nisms would be useful for that, and thosecould be accomplished through the antic-oagulation forum (ACF), through the Soci-ety of General Internal Medicine (SGIM),and other multi-centered collaborations.

Third I think we need pharmacokinetic and phar-macodynamic studies to elucidate themechanism, the action and interaction ofalternative and complementary agents.

Fourth we need in vivo studies to investigateputativae synergy or antagonism. Many ofthese studies would be cross over studies.Larger studies should archive DNA fromparticipants for further pharmacogeneticanalysis.

Fifth we need good public and physician educa-tion. Presently, it is hard for clinicians andpatients to find quality information aboutsupplements — web searches mostly returnsites that sell alternative and complemen-tary agents with variable quality of infor-mation. One can envision a web sitesupported by the NIH that has evidence-based information about effectiveness andsafety.

Six we may need greater regulation or removalof compounds with high abuse or high riskpotential — for example, drugs that havecoumarin activity.

Finally we need to periodically reassess. We needto reassess because new anticoagulantswill become available, so that supplement— warfarin interactions should become lesscommon, but others will become morerelevant.

Dr. Penzak

As a pharmacokineticist, I think it is only natural,then, that my interests would gravitate towardcytochrome P450-mediated drug interactions, aswell as drug transport and pharmacogenomics. As Iam sure everybody is now keenly aware, warfarin ispresent as a racemic mixture, in which the Scomponent is primarily metabolized by 2C9 with


some of the other 2C isoforms playing a lessimportant role.

The R isomer, which is about two to five timesless potent as an anticoagulant, is primarily me-tabolized through CYP1A2 and 3A4.

At this point, just doing a very quick Medlinesearch and looking at the natural products databases, there are approximately 36 different herbsat this point—and this is just a cursory search—that I found that had the potential to modulate—inother words, either inhibit or induce the enzymesthat are involved in the metabolism of warfarin. Alot of this information—the potential for herbalconstituents to modulate CYP450 enzymes—isbased on a series of in vitro investigations, andshortly I will talk a little bit about some of theshortcomings with regard to these types ofstudies, and why studies in humans are eventuallythe ultimate goal for the types of investigationsthat we would hope to do.

So, clearly, then, when we see inhibition ofCYP2C9 or CYP3A4 and CYP1A2, we subsequentlyexpect to see an increase in either S or Rwarfarin, respectively. Since warfarin is a drugwith a very narrow therapeutic range, smallincreases in the concentrations of S warfarin, nodoubt, can cause clinically significant pharmaco-dynamic effects. Just as an example, we heard alot about St. John’s wort today. There was astudy done at the NIH just a few years ago withthe HIV protease inhibitor, indinavir, which wasadministered to a group of eight healthy volun-teers alone and then, after two weeks of St.John’s wort administration. What we saw wasabout a 50% to 60% reduction in total area underthe curve, and a nearly 90% reduction in theextrapolated eight hour (trough) concentration.One of the important points here is that thelength of administration of St. John’s wort wastwo weeks. So, when we talk about the types ofstudies that we would like to do in humans, oneof the things to consider is the length ofadministration of the different herbal compo-nents. It usually takes at least a couple of weeksbefore we can see induction to a clinicallysignificant degree, because it takes time to makeenzymes, to make these proteins, as it were.That is why I think the PCR activation studiesthat we have heard about earlier today hold somuch promise and are very interesting, because Ithink they give us an idea of what types ofclinical studies, then, that we can pursue; theyguide us as to what are the most commonalternative medicines, then, that are likely tointerfere with warfarin metabolism and affect itspharmacodynamics.

To date, we have had relatively few studiesbetween warfarin and herbal preparations. So, nowthat we have some candidate compounds that weknow may affect the metabolism of warfarin, Ithink it is important to examine these types ofinteractions in humans.

It may sound a little bit scary, but thereactually are a number of studies now that havebeen done, giving warfarin to healthy volunteerswith co-administration of vitamin K. In fact, thereare several groups that use warfarin as a CYP2C9probe, given along with vitamin K, and they havefound that it has been safe to do those types ofstudies.

When we only rely on in vitro studies to assesspotential drug interactions, one of the shortcom-ings we encounter is that not all in vitro types ofmodels can identify enzymatic induction. If you goback to the literature, I think it was the year 2000,you can find the title of a paper that says St. John’swort is a potent inhibitor of cytochrome p450enzymes. Well, we know that the effect of St.John’s wort on CYP450 is that of an inducer, butstudies in hepatic or intestinal microsomes are notable to depict induction. This is where PXRactivation comes in as being an important preclin-ical test, to identify those compounds that may, infact, be inducers of CYP450.

Another problem with in vitro studies is the factthat occasionally you will see supraphysiologicalconcentrations of constituents used in these in vitromodels; that is another thing to keep in mind whentrying to make sense of this in vitro data and decidewhich herbs are the best candidates to assess fortheir interaction potential with warfarin, inhumans.

Standardization of herbal preps is certainlysomething that, when studies are conducted inhumans, we will want to take a look at. In otherwords, looking at the constituents of the herbalproducts, and making sure that they are present inthe quantity that is stated on the label. Theseproducts should also be examined to make sure thatthe most common adulterants are absent from thepreparations. Another thing, too, is with some ofthese in vitro tests, we see individual constituentsassessed for their drug interaction potential; anexample includes the St. John’s wort componentshyperforin and hypericin. We know that the multi-ple components of St. John’s wort may act syner-gistically or in an additive fashion. So, at least inhumans, we want to look at the sum of ingredientsof the herbal products that patients are going to betaking, and not simply look at the individualconstituents for their drug interaction potential —not that studying individual herbal constituents in


vitro doesn’t have its importance from a mecha-nistic standpoint.

One of the advantages of healthy volunteerstudies is that you can have individuals acting astheir own control subjects. Again, I stress theimportance of giving the herbal supplement for atleast 14 days administration in these studies. Also,as we have heard about today, genotyping all thestudy subjects is also important; number one, forsafety concerns so that we make sure no one has aCYP2C9 mutant allele that we are going to givewarfarin to and secondly, to try to amass ahomogeneous data base.

Another step might be phenotyping CYP2C9,CYP3A4, and CYP1A2, so we can help get a handleon, or at least confirm, the mechanism by which wefeel these interactions may be occurring.

Dr. Spinler

It is estimated that more than 38 m US adults(18.6%) are using herbal therapy [20]. Less than 50%of patients report discussing dietary supplementswith their physician [21—23]. More than one-thirdof patients surveyed want more information re-garding dietary supplement adverse effects. There-fore, two research priorities for the NationalInstitutes of Health (NIH) should be the improve-ment of health professional medication historytaking training programs and exploration of whatinformation sources, other than healthcare provi-ders, consumers and patients use to obtain infor-mation on dietary supplement drug interactions.

Areas for research into health professionalmedication history taking practices concerningdietary supplements should include

! Identification of the extent of education ofnurses, physicians and pharmacists regardingdietary supplement medication history takingtraining by colleges of nursing, medicine andpharmacy,

! Identification of the extent of education ofnurses, physicians and pharmacists receive re-garding dietary supplement — vitamin K antag-onist anticoagulation drug interactions throughcolleges of nursing, medicine and pharmacy,

! Identification of the knowledge of practicingnurses, physicians and pharmacists of sourcesof print, PDA-based and Internet-based sourcesof information on dietary supplement-vitamin Kantagonist anticoagulation drug interactions.

The American Society of Health System Pharma-cists has recommended some resources for phar-macists [24], but we need to have more information

on this topic available for the practicing clinician.Where can this information be found and where canwe look to get the best answer for the patient?Following identification of college programs andsources of information, best practices in theseareas should be identified and promoted to bothcolleges and practitioners.

A second area for research that I recommend isa review of the accuracy and completeness ofInternet-based programs accessible to consumersand patients for accuracy of information andrecommendations for vitamin, food and herbinteractions with vitamin K antagonists. For ex-ample, Healthnotesk is available to consumersand patients through kiosks at pharmacies, super-markets and health food stores, as well as via theInternet [25]. This web site helps the patientidentify prescription drug-food, vitamin and herbinteractions. Recommendations are provided re-garding the significance of the drug interactionand steps the patient should take. The sitecontains primary literature references supportingthe interaction. Other Internet sites are accessibleto patients as well. The National Institutes ofHealth offers information on dietary and herbalsupplements through its National Center for Com-plimentary and Alternative Medicines [26]. Irecommend that the accuracy and completenessof information provided be reviewed.

If current access by patients and clinicians toinformation regarding drug interactions betweenvitamin K antagonists and vitamins, foods and herbsis limited or is inaccurate, the NIH should considersponsorship and promotion of such a site.

Dr. De Smet

Yesterday, in my talk, I emphasized that we shouldimprove the quality of our case reporting and,above all, that we should do epidemiologicalstudies. The more I hear during the conference,the more I think that that is important, becausethey are needed for hypothesis verification, forquantification of the problem, which is veryimportant.

There is another thing that I would like toemphasize. Besides finding out what we do notyet know, I think we should act on what weknow.

That takes me to the point of implementation.You can come up with all the research that has beendone and is being proposed, with a guideline onhow to handle this topic. When you have aguideline, although you put this information on aweb site, that does not necessarily mean that every

Table 4 Modified drug interaction probability scale*

Question Yes No N/A

Previous conclusive reports of thisinteraction?

+1 0 0

Observed interaction consistent withknown interactive properties ofthe supplement

+1 �1 0

Observed interaction consistent withknown interactive properties ofthe anticoagulant?

+1 �1 0

Event consistent with known orreasonable time course ofinteraction?

+1 �1 0

Did interaction remit whensupplement was stopped?

+1 �2 0

Did interaction reappear whensupplement was readministeredconcurrently with theanticoagulant?

+2 �1 0

Are there alternative causes for theevent other than an interaction?

�1 +1 0

Was there a change in the INRassociated with the interaction?

+1 0 0

Did the interaction result inhemorrhage or thrombosis?

+1 0 0

Was the interaction greater whenthe supplement dose wasincreased, or less when thesupplement dose was decreased?

+1 �1 0

*Adapted from Hansten and Horn [27] and Naranjo, et al. [28](from Dr. Wittkowsky’s presentation).


user of an oral anticoagulant will benefit from thatinformation.

So, my recommendation would be to developand test strategies to make sure that every user ofantithrombotic agents benefits from our accumu-lated knowledge in daily health care practice.

Dr. Wittkowsky

I would like to make a specific proposal for anational internet-based registry of anticoagulantinteractions with dietary supplements. The goal ofthe registry would be to obtain case-based evi-dence of drug interactions between anticoagulantsand dietary supplements that occur in stableanticoagulated patients, using pre-defined criteriafor stable anticoagulant therapy. A secondary goalof the registry would be to define these interac-tions based on clinical and monitoring outcomes, aswell as by likelihood, using a drug interactionprobability scale. Potential participants in theregistry include members of the AnticoagulationForum, a national organization of anticoagulationclinics and practitioners, members of the www.clotcare.com list-serve which attracts multidisci-plinary providers interested in thrombosis, as wellas other interested parties. The registry might alsoprovide an opportunity for anticoagulation patientsto participate in case reporting.

Data elements that would be collected for eachcase would include information about the patient inwhom an interaction is observed, the stability of

Table 3 Data elements to be collected (from Dr.Wittkowsky’s presentation)

Observer !Name !Email address!Clinic location !Phone!Phone

Patient !Age !Indication!Anticoagulant !Goal INR!Duration of therapyprior to interaction

Stabilityriteria

!Same dose of warfarinfor 2 weeks!Last 2 INRs withinrange +/� 0.2 INR units

Interactingproduct

!Brand name !Manufacturer

!Ingredients !Dose!Duration of useprior to event

Observations !Change in INR!Clinical event(thrombosis,hemorrhage)!Healthcare resourcesused (ER visit, hospitaladmission).

oral anticoagulation according to pre-defined sta-bility criteria, and the interacting product. Inaddition, specific observations about changes inlaboratory monitoring parameters, clinical out-comes including hemorrhagic or thromboemboliccomplications, and healthcare resource use, in-cluding hospitalization or emergency room visits,would also be collected for each case (Table 3).

A modified drug interaction probability scalewould be used to objectively determine thelikelihood that each observation is in fact a druginteraction. Hansten and Horn [27] have devel-oped a drug interaction probability scale based ona modification of the Naranjo adverse drugreaction probability scale [28], and this systemcan be modified further to more specificallyaddress anticoagulant drug interactions (Table 4).Using this scoring system, each interaction wouldbe objectively scored with respect to likelihood(Table 5).

Limitations of the registry include the inabilityto document concurrent use of antithromboticagents and dietary supplements that do not leadto interactions. It would be difficult to indepen-dently verify collected data, although it would bepossible to obtain additional information from theobserver of the interaction if information about the

http://www.clotcare.com

Table 5 Drug interaction probability

Total score Probability

N8 Highly probable5—8 Probable2—4 Possibleb2 Doubtful

*Adapted from Hansten and Horn [27] and Naranjo, et al. [28](from Dr. Wittkowsky’s presentation).


reporter is collected. It also would not be possibleto control for differences in product ingredients ortablet contents.

The advances of the registry, in comparison tocurrent case reporting, is that all interactionswould be described only in stable patients, andthat objective criteria would be used to character-ize both the observed outcomes, and the likelihoodthat the observations represent drug interactions.

Dr. Lee

Please note: This discussion opens with a presen-tation of a database of Chinese herbs with potentialeffects on coagulation. The database in full detailwill be available on the Thrombosis Researchwebsite [57].

In composing my remarks, I was asked to compilea list of herbs with interest to those involved inclinical work and research on anticoagulation, andinteractions between dietary supplements andanticoagulants. I will present the database Icomplied first. I will then go the second area Iwas asked to address, which was to present ideasfor research.

Using Excel spreadsheet format, I composed alist of Chinese herbs of interest with regard tocoagulation, in Excel spread sheet format. By bofinterest with regard to coagulation,Q I am discuss-ing a spectrum of agents, ranging from onesclearly shown to affect coagulation, to thosestrongly suspected of affecting coagulation withlimited data, to those whose other propertiesindicate that there is potential to affect thecoagulation cascade.

Given the vast Chinese herbal pharmacopia, it isnot possible to account for every possible herbworthy of consideration, however I do indicate theones with the strongest research. Another reasonwhy it is not possible to list every product isbecause development of the herbal formulary isan area of ongoing research in China, and it iscontinually being expanded.

It also needs to be stated that Chinese herbalmedicine is a misnomer, as many agents usedtherapeutically derive from natural sources other

than herbs. These sources include food products,insects, animal products, and minerals [29—32].

I present this spreadsheet to stimulate ideas forresearch rather than as a guide for clinical use. Theproducts, which I will, for the sake of simplicity,refer to as herbs for most of my discussion, arelisted in the spreadsheet according to their differ-ent names including English and Pin Yin Englishtransliterations of their Chinese name. In thecolumns that follow, I indicate effect on coagula-tion. There is a key at the end, explaining what theinitials mean in terms of effects on coagulation,specifically categorizing according to platelet ac-tivity and prothrombotic activity in addition toanticoagulant activity, when these details of actionare known. Some agents have biphasic activity, andI have indicated those as well. At this time itunderstood that biphasic activity results can resultfrom an array of compounds within a singleproduct, but there is also the question of dosedependant biphasic activity of single constituents.

The list also includes use history. The spread-sheet indicates where there is research support forthe described use. Another column denotes classi-cal or a long history of empiric use. Data is scant, asmost of the primary source literature has not beentranslated into English.

Additional columns contain information on othereffects indirectly related to coagulation, with asmall r or a small c to tell you whether or not thoseeffects are based on studies or on classical expe-rience. These include features overlapping ininterest between study of coagulation, inflamma-tion, and related disease states. I specificallyindicate when herbs also have been studied or usedin myocardial ischemia, hyperlipidemia, hyperten-sion, migraine, or inflammatory states. As I dis-cussed in my other presentation, and will againmention here, the many activities of herbal con-stituents relate to the widespread identification ofinflammatory features underlying a variety ofdisease states [33]. It is also interesting to notethat many natural products contain constituentswith opposing activities. Consider one of the mostprominent in this regard, the leech, who must gainaccess to easy blood flow for feeding, and then sealit’s wound up again, as if to save the host’s bloodsupply for another day [34].

When constituents with potential effects oncoagulation have been identified, they are namedin the Compounds of Interest column. Here, there isagain scant data, because many herbs had notundergone extensive analysis, and even whencoumarins have been identified, not all coumarinshave anticoagulant activity [35]. When they havebeen identified, I see that as cause not to neces-


sarily attribute anticoagulant properties to theherb, but to look at that herb more closely.Certainly, from a clinical standpoint, patients takingsuch herbs need to be monitored with particularvigilance, until we better understand the specificpharmacology of the herbs.

Before discussing specific research ideas, I needto ask if it’s time to reevaluate our current viewthat the ideal pharmaceutical is an isolatedmolecule affecting a pinpoint physiologic target.Intact natural products have comprised the bulk oftherapeutic agents around the world, for most ofhistory. Paracelus (1493—1541) advocated extrac-tion of active principles from therapeutic plants,animals, or minerals. His ideas were not givencredence at the time, but in the 19th century,active principles were successfully isolation, and itbecame recognized that the pharmaceuticaleffects of plants and other products could beattributed to single molecules. Morphine led theway in 1804, by Serturner, followed by Salicin byBuchner in 1838. It was this molecule thatbenefited from some engineering in the laborato-ry, with conversion to acetylsalicylic acid, as Idiscuss in further detail in my other presentation.We know today that many drugs are either deriveddirectly from plant structures, or based on mod-ification of plant molecules [36]. This is also thecase with Warfarin, a molecule based on thenaturally occurring dicumerol. Synthetic gikngo-lide analogues have been studied as PAF antago-nists [37].

There is no question that much medical progresshas been made using this biochemical approach. Wehave paid a price with toxicity, however. Theextent is sometimes annoying, but at times,unacceptable, most recently in the case of theCOX-2 inhibitors [38]. I wonder if it is time toinclude both views of what an effective therapeuticagent might be; to continue working with themodern model, and in addition, to systematicallystudy intact herbs. The latter can be viewed eitheras complete therapeutic agents themselves, or asspringboards for new types of multi-constituentpharmaceuticals. The balance of componentswould need to be closely controlled; as I stated inmy other discussion, coumarins have been shown tohave potential as anti-inflammatory and anti-cancer agents, yet they also have provided us withaflatoxin and dicumerol.

An herb that would be interesting to studyfurther, bearing these ideas in mind, is feverfew.An area I believe is worthy of further clinical studyof feverfew for migraine. Despite a long history ofempiric use, it is unclear whether feverfew iseffective [39]. In traditional use, feverfew was felt

to have not only has specific anti-migraine proper-ties, but antispasmodic, antihypertensive, andanticoagulant activity as well, with fewer GI sideeffects than aspirin [40]. There are many areas topursue in such a study, ranging from a study of thequality of feverfew preparations themselves, toclinical effect on migraine, to both laboratory andclinical analysis of effects on inflammation andstroke risk. Might studies reveal feverfew to be analternative to aspirin in stroke prevention, espe-cially in particularly susceptible migraine patients?An additional area for study would be to evaluatewhether parthenolide, believed to be the primaryactive constituent of feverfew, is more effective asan isolate, or when given in the form of the wholeherb.

If research eventually shows that by usingseveral related molecules of the same planttogether, we can produce effective treatmentswith better safety, perhaps this opens the door todiscovery of a novel anticoagulant with a bettersafety profile than Warfarin?

Whether herbs are used as fragrant tablets inpretty bottles from the health food store, or as thebasis for drug development, there are several otherareas ripe for study. We need to characterizediffering effects depending on dose. It is reputedin Chinese, and other classical herbal literature,that different patient characteristics cause clinicaleffects to vary [41]. This is another area forresearch with potential for broad application. Wecan draw on our ability to characterize geneticqualities. Outcomes of such work might apply touse of pharmaceuticals as well, leading to furtherdelineation of how to maximize drug benefits,while minimizing risk.

In terms of enhancing safety of use, before aplan can be formulated, we need to look into theepidemiology of serious adverse effects. What arethe problems, who are the high risk patients? Aspart of this, we need to find methods to improvereporting.

The pharmaceutical industry is also strugglingwith how to improve reporting, information dis-semination, and action in case of problems. Cer-tainly we can follow along to see what they learn,especially as this country is looking to learn frompharmaceutical adverse event reporting practicesof other nations [42]. We may also wish to look atadverse event study practices in China, wheremultidisciplinary herb problem monitoring teameffectiveness is being studied. In Hong Kong, apharmacist, chemical pathologist, and a physicianhave jointly provided advisory services to all publichospitals, in the event of problems in patientstaking herbs. In this way, the causes of complica-

Table 6 Herbal medicines versus synthetic drugs

Herbal Synthetic

Invariably more than onepharmacologicallyactive compound

Usually one pharmacologicallyactive compound

Active ingredients oftennot known

Active ingredient known

Pure compound notavailable

Pure compound available

Raw material limited Raw material unlimitedQuality variable Quality consistentMechanism of actionoften unknown

Mechanism of action known

Toxicology often unknown Toxicology knownLong tradition of use Short tradition of useTherapeutic window wide Therapeutic window narrowAdverse effects rare Adverse effects frequent

(from Dr. Ernst’s presentation).


tions can be clearly understood and any problemswith herbal preparations, addressed [43].

As I discussed at the start, when presenting thespreadsheet, the list of medicinal plants is exten-sive. It will most likely be useful to narrow the listand target what plants might have maximal clinicalimportance. Earlier I mentioned feverfew, andmuch of what I said there, applies to any otherherb targeted for study. We require chemicalcharacterization and evaluation of normal variabil-ity among plants of the same species. Location ofgrowth, conditions of weather and soil, harvestconditions may all determine composition of plantconstituents [44]. Individual plant parts need to bestudied. Once constituent structures are charac-terized, we need to know pharmacokinetic proper-ties. What attributes of a particular plant are mostimportant for clinical use?

The extent of herbal anticoagulant effects,when taken P.O. at clinically relevant doses remainslargely undefined. It is unclear whether availablelaboratory assays detect changes in coagulationparameters caused by herbs [45]. We requirefurther study of the effects of potential anticoag-ulant herbs in vivo. Developing methods to monitormore subtle, subclinical indicators of coagulationstatus would facilitate research into effects ofanticoagulant herbs in human subjects. Certaintechniques now used in laboratory research mightbe suitable for modification for use in a clinicalsetting, at least perhaps for application in clinicalresearch [46].

To conclude my recommendations, we need away of effectively informing the public of riskpotential, and I emphasize effective, becausewarnings in small writing on bottles does not carrymuch weight among a population who thinksnatural substances are harmless. Devising methodsof effectively informing patients requires inputfrom the variety of professionals recommendingherbs, and when I say professionals, I meaneverybody from the physician, to the pharmacist,to the friendly salesperson in the health foodstore.

Dr. Ernst

Clinical trials are prospective experiments forassessing the nature and results of medical inter-ventions. Generally speaking, clinical trials ofherbal medicine should follow the same lines asany other such experiment [47]. In particular theyshould provide clear definitions or descriptions of:

! Research questions/hypothesis! Participants and sample size

! Intervention! Outcome measurements! Randomization procedures! Blinding! Statistical analysis.

General considerationsBut there are many differences between herbalmedicines and synthetic drugs (Table 6). Onesignificant logistical problem in relation to clinicaltrials of herbal medicine is the lack of researchfunds for such research. There is little patentprotection for medicinal plants and herbal manu-facturers are usually relatively small companies.Lack of funds leads to a paucity of research and theundeniable fact that studies are often not asrigorous as they could be. The effect is aggravatedby a lack of research culture and expertise in herbalmedicine. If one accepts that the effects (if any) ofherbal medicines are usually mild and take a longtime to become clinically manifest [48], it becomesobvious that large, long-term studies are oftenrequired—a fact which can significantly complicatethe existing logistical problems with trials of herbalmedicine. In the US, the NIH is now supporting largeclinical trials of herbal medicines. Other countriesare far less fortunate, and generally speaking suchinvestigations remain globally under-funded.

Specific considerationsIn addition to the general difficulties, clinical trialsof herbal medicines are confronted with a range ofproblems that are not specific to this area inprinciple but, in terms of their size, representunique obstacles to research.

Bias can, of course, be an issue in any research;in herbal medicine it is probably larger thanelsewhere. We have shown that countries such as

Table 7 Advice on designing herbal trials.

!Don’t re-invent the wheel!Secure (independent) funding!Make sure you have all the essential expertise don boardT,

e.g:-herbal expert-clinical expert-statistician-methodologist

!Formulate a clear research question/hypothesis!Take all steps to minimise bias and maximise reproducibil-

ity,e.g:-full characterisation of herbal remedy-optimal quality of herbal preparation

!Validated outcome measures

(from Dr. Ernst’s presentation)


Germany [49] or China [50], which are of obviousimportance to herbal medicine, publish predomi-nantly positive results (Fig. 1). This does not, ofcourse, prove the existence of bias but seems tosuggest it. Similarly, complementary medicinejournals are associated with a strong positivepublication bias [50]. The concern therefore isthat, perhaps due to the lack of research culturein this area, bias is more of an issue in herbal thanin conventional medicine [51].

Outcome measures used in clinical trials should,of course, be validated. In trials of herbal medi-cine, soft and non-validated outcome measures areoften employed, e.g. percentage of patients per-ceiving benefit or patients’ preference. Similarly,multiple outcomes are frequently used withoutaccounting for multiple statistical tests. Finally,surrogate endpoints are frequent and oftenresearchers seem to measure what is measurablerather than what is relevant.

Blinding can be a real problem in ddouble-blindTtrials of herbal medicines. Due to taste, odor orappearance, verum herbal medicines may bedistinguishable from their respective placebos.Thus unblinding can occur and exert an undueinfluence on the results. For instance, there arenumerous dplacebo-controlled, double-blindT trialsof garlic preparations for cholesterol lowering [52].Yet anyone who has ever been involved in such astudy knows that, due to the body odor caused bygarlic, blinding is not a realistic option.

The intervention has to be fully described in allclinical trials. The aim must be to disclose alldetails such that the study can be reproduced

CAM research: direction of conclusions

10-9-8-7-6-5-4-3-2-1-

Ger

man

y

10.0

2.5

4.0

1.32.0

0.8 0.4

4.5

UK

Italy

Fran

ceSpa

in NLBel

gium USA

n=137 n=183 n=39 n=47 n=24 n=17n=17

n=183

( n = 652, year 2002)Ratio

CAM research: direction of conclusions

Figure 1 Evaluation of direction of outcome of allMedline-listed articles on complementary (includingherbal) medicine by country (major European countrieswith USA as a comparison). Ratio=number of articleswith clearly positive conclusion divided by number ofarticle with clearly negative conclusions (a ratio of 10thus means that for every dnegativeT there were 10dpositivesT). (from Dr. Ernst’s presentation).

elsewhere. In herbal medicine, this can be morecomplex than in conventional studies. Herbalmedicines are natural products; their compositionand therefore effects could depend on a range offactors, e.g.:

! Source(s), e.g. soil, climate! Processing/extraction! Storage.

A degree of variability from batch to batch maybe unavoidable in herbal medicines. Additionalissues can be adulteration [53], and contamination,e.g. with heavy metals [54].

An often-voiced criticism of clinical trials is thatsuch experiments tell us little about individualpatients. In herbal medicine, treatments are oftenhighly individualized. Therefore, some proponentsdismiss the value of clinical trials of herbal medi-cine. This obviously ignores the existence of n =1studies. Such studies are possible in herbal medicineand they inform us about responses of individualpatients [55]. Similarly, modifications of the stan-dard design of clinical trials have been developed[56], some of which may be suitable for specificresearch questions encountered in herbal medicine.

ConclusionsClinical trials of herbal medicines are undoubtedlyfeasible. They should adhere to the standardsapplied to any other clinical trial (Table 7). Theproblems they face can be significant but, inprinciple, they can be solved.

Dr. Marder

I have four suggestions that are rather different,some of them, than what have been presented, andthat are more general and global.


One is that, for preclinical testing of potentialagents that have effects on bleeding or clotting, wecould use some better animal models to see whichagents might make the transition from basicscience into clinical study, by doing appropriateanimal models. It might be a nice way to interface,and perhaps decide which agents are worthy offurther study.

I had a mini-debate with Dr. Ernst before aboutwhether we should go after safety or efficacy and,as I am a neophyte in the field, I really put morecredence on his opinion than mine for this, but mypoint would be that, if a study did not show strongefficacy, I am not sure how much that would turnoff people who were used to using a given drug.

I would be especially interested in safety insubgroups, which is an item that has been raisedseveral times, namely, whether women, elderly,pediatric patients are especially vulnerable to theeffects of a pill, which might have a higherconcentration relative to their physiology.

Third would be to set up formal programs ofteaching. There, you might be the AmericanSociety of Hematology, where I think this subjecthas not really been broached yet, and we do needto get the word out to the community of physicians.

Last, I am quite impressed with the knowledgeand contributions of the Europeans in this field.They really do seem to have a better handle on itthan the Americans.

I don’t know how to go about this other than onthe personal level, but perhaps on organizationallevels we could establish some formal collabora-tions for analysis of drugs and especially, ulti-mately, for regulation of the herbal supplementswith persons as Dr. DeSmet and Dr. Ernst.

References

[1] Shaw D, Leon C, Kolev S, Murray V. Traditional remedies andfood supplements; a 5-year toxicological study (1991—1995). Drug Saf 1997;17:342–56.

[2] Ai AL, Bolling SF. The use of complementary and alternativetherapies among middle-aged and older patients. Am J MedQual 2002;17:21–7.

[3] Go AS, Hylek EM, Phillips KA, Chang Y, Henault LE, Selby JV,et al. Prevalence of diagnosed atrial fibrillation in adults:national implications for rhythm management and strokeprevention: the Anticoagulation and Risk Factors in AtrialFibrillation (ATRIA) study. JAMA 2001;285:2370–5.

[4] Page RL. Newly diagnosed atrial fibrillation. N Engl J Med2004;351:2408–16.

[5] Benjamin EJ, Levy D, Vaziri SM, D’Agostino RB, Belanger AJ,Wolf PA. Independent risk factors for atrial fibrillation in apopulation-based cohort. The Framingham Heart Study.JAMA 1994;271:840–4.

[6] Halperin JL, Hart RG. Atrial fibrillation and stroke: newideas, persisting dilemmas. Stroke 1988;19:937–41.

[7] Wolf PA, Abbott RD, Kannel WB. Atrial fibrillation as anindependent risk factor for stroke: the Framingham Study.Stroke 1991;22:983–8.

[8] Stroke prevention in atrial fibrillation study: final results.Circulation 1991;84:527–39.

[9] ACC/AHA/ESC guidelines for the management of patientswith atrial fibrillation: executive summary. Circulation2001;104:2118–50.

[10] Levine MN, Raskob G, Landefeld S, Kearon C. Hemorrhagiccomplications of anticoagulant treatment. Chest 2001;119(suppl):108S–21S.

[11] Protheroe J, Fahey T, Montgomery AA, Peters TJ. Theimpact of patients’ preferences on the treatment of atrialfibrillation: observational study of patient based decisionalanalysis. BMJ 2005;320:1390–4.

[12] Williamson EM. Drug interactions between herbal andprescription medications. Drug Saf 2003;26:1075–92.

[13] Ackman ML, Campbell JB, Buzak KA, Tsuyuki RT, MontagueTJ, Teo KK. Use of nonprescription medications by patientswith congestive heart failure. Ann Pharmacother 1999;33:674–9.

[14] Ai AL, Bolling SF. The use of complementary and alternativetherapies among middle-aged and older patients. Am J MedQual 2002;17:21–7.

[15] Woods MJ, Stewart RL, Merry H, Johnstone DE, Cox JL. Useof complementary and alternative medical therapies inpatients with cardiovascular disease. Am Heart J 2003;145:806–12.

[16] Zuckerman IH, Steinberger EK, Ryder PT, Haines S. Herbalproduct use among anticoagulation clinic patients. Am JHealth Syst Pharm 2002;59:379–80.

[17] Halperin JL. Ximelagatran: oral direct thrombin inhibitionas anticoagulant therapy in atrial fibrillation. J Am CollCardiol 2005;45:1–9.

[18] Fiessinger J-N, Huisman MV, Davidson BL, Bounameaux H,Francis CW, Eriksson H, et al, for the THRIVE TreatmentStudy Investigators. Ximelagatran versus low-molecular-weight heparin and warfarin for the treatment of deepvein thrombosis: a randomized trial. JAMA 2005;293:681–9.

[19] Gurewich V. Ximelagatran—promises and concerns. JAMA2005;293:736–9.

[20] Tindle HA, Davis RB, Phillips RS, Eisenberg DM. Trends in useof complementary and alternative medicine by US adults:1997—2002. Altern Ther Health Med 2005;11:42–9.

[21] Eisenberg DM, Kessler RC, Van Rompay MI, Kaptchuk TJ,Wilkey SA, Appel S, et al. Perceptions about complemen-tary therapies relative to conventional therapies amongadults who use both: results from a national survey. AnnIntern Med 2001;135:344–51.

[22] Robinson A, McGrail AR. Disclosure of CAM use tomedical practitioners: a review of qualitative andquantitative studies. Complement Ther Med 2004;12:90–8.

[23] Giveon SM, Liberman N, Klang S, Kahan E. Are people whouse bnatural drugsQ aware of their potentially harmful sideeffects and reporting to family physician? Patient EducCouns 2004;53:5–11.

[24] Anonymous. ASHP Statement on the use of dietary supple-ments. Am J Health Syst Pharm 2004;61:1701–11.

[25] Healthnotes Safety Checker Using Medicines with Vitaminsand Herbs. Available from http://www.healthnotes.info/qs_db/healthnotes.cfm?contentid=2411003. Accessed 04/12/05.

[26] Dietary and Herbal Supplements. Available from: http://nccam.nih.gov/health/supplements.htm. Accessed 04/13/05.

http://www.healthnotes.info/qs_db/healthnotes.cfm?contentid=2411003

http://nccam.nih.gov/health/supplements.htm


[27] Hansten PD, Horn JR. Hansten’s and Horn’s drug interac-tions analysis and management. Vancouver, WA7 AppliedTherapeutics; 1998.

[28] Naranjo CA, Busto U, Sellers EM, Sandor P, Ruiz I, Roberts EA,et al. A method for estimating the probability of adversedrug reactions. Clin Pharmacol Ther 1981;30:239–45.

[29] Chen JK, Chen TT. Chinese medical herbology and pharma-cology. Art of Medicine Press; 2004.

[30] Bensky D, Gamble A. Chinese herbal medicine materiamedica. Chichester-New York7 Eastland Press; 1993.

[31] Huang KC. Pharmacology of chinese herbs. Boca Raton-London7 CRC Press; 1998.

[32] Duke JA. Biologically active phytochemicals and theiractivities. Boca Raton-Ann Arbor7 CRC Press; 1992.

[33] Black PH. The inflammatory response is an integral part ofthe stress response: implications for atherosclerosis, insulinresistance, type II diabetes and metabolic syndrome X.Brain Behav Immun 2003;17(5):350–64.

[34] Faria F, Junqueira-de-Azevedo Ide L, Ho PL, Sampaio MU,Chudzinski-Tavassi AM. Gene expression in the salivarycomplexes from Haementeria depressa leech through thegenerations of expressed sequence tags. Gene 2005;349:173–185.

[35] Cooke D, Fitzpatrick B, O’Kennedy R, McCormack T, EganD. In: O’Kennedy R, Thornes RD, editors. Coumarins—multifaceted molecules and many analytical and otherapplications. Chichester-New York7 John Wiley & Sons;1997. 304 pp.

[36] Heinrich M, Barnes J, Gibbons S, Williamson EM. Funda-mentals of pharmacognosy and phytotherapy. Edinburgh-London7 Churchill Livingstone; 2004. p. 16–8.

[37] Zhu W, Chen G, Hu L, Luo X, Gui C, Luo C, et.al. QSARanalyses on ginkgolides and their analogues using CoMFA,CoMSIA, and HQSAR. Bioorg Med Chem. 2005;17;13(2):313—22.

[38] Block KI. The demise of the super-aspirins: an opportu-nity for integrative medicine. Integr Cancer Ther 2005;4(1):5–7.

[39] Pittler MH, Ernst E. Feverfew for preventing migraine.Cochrane Database Syst Rev 2004;1 [CD002286].

[40] Holmes P. The energetics of western herbs. Boulder7 SnowLotus Press; 1994. p. 556–7.

[41] Bensky D, Barolet R. Chinese herbal medicine formulas andstrategies. Eastland Press; 1990.

[42] Matthews AW. As drug-safety worries grow, looking overseasfor solutions. Wall Street J 2004;12/31.

[43] Chan TY, Tam HP, Lai CK, Chan AY. A multidisciplinaryapproach to the toxicologic problems associated with

the use of herbal medicines. Ther Drug Monit 2005;27(1):53–7.

[44] Dong TT, Cui XM, Song ZH, Zhao KJ, Ji ZN, Lo CK, et al.Chemical assessment of roots of Panax notoginseng inChina: regional and seasonal variations in its activeconstituents. J Agric Food Chem 2003;51(16):4617–23.

[45] Kohler S, Funk P, Kieser M. Influence of a 7-day treatmentwith Ginkgo biloba special extract EGb 761 on bleedingtime and coagulation: a randomized, placebo-controlled,double-blind study in healthy volunteers. Blood CoagulFibrinolysis 2004;15(4):303–9.

[46] Marton Z, Halmosi R, Alexy T, Horvath B, Toth A, Feher G,et al. Hemorheological methods in drug research. ClinHemorheol Microcirc 2004;30(3—4):243–52.

[47] Ioannidis JPA, Evans SJW, Gøtzsche PC, O’Neill RT, AltmanDG, Schulz K, et al. Better reporting of harms inrandomized trials: an extension of the CONSORT state-ment. Ann Intern Med 2004;141:781–8.

[48] Ernst E, Pittler MH, Stevinson C, White AR. The desktopguide to complementary and alternative medicine.Edinburgh7 Mosby; 2001.

[49] Giovannini P, Schmidt K, Canter PH, Ernst E. Research intocomplementary and alternative medicine across Europeand the United States. Forsch Komplement Med KlassNaturheilkd 2004;11:224–30.

[50] Tang JL, Zhan SY, Ernst E. Review of randomised con-trolled trials of traditional Chinese medicine. BMJ 1999;319:160–1.

[51] Ernst E, Pittler MH. Alternative therapy bias. Nature1997;385:480.

[52] Stevinson C. Letter to the editor garlic for total cholesterolreduction. Ann Intern Med 2001;135:65–6.

[53] Ernst E. Adulteration of Chinese herbal medicines withsynthetic drugs: a systematic review. J Int Med 2002;252:107–13.

[54] Ernst E, Thompson Coon J. Heavy metals in traditionalChinese medicines: a systematic review. Clin PharmacolTher 2001;70:497–504.

[55] Canter PH, Ernst E. Multiple n =1 trials in the identifica-tion of responders and non-responders to the cognitiveeffects of Ginkgo biloba. Int J Clin Pharmacol Ther 2003;41:354–7.

[56] Ernst E. Randomised clinical trials: unusual designs.Perfusion 2004;17:416–21.

[57] Lee A. Agenda item: dillemas in herbal medicine: theclinician’s viewpoint. Thromb Res 2005 [this issue].doi:10.1016/j.thromres.2005.05.003.

intl.elsevierhealth.com/journals/thre

Thrombosis Research (2005) 117, 170—174

Session V: Comments and discussion

DR. BUSSEY: I would particularly like to thank allof the speakers for being succinct and keeping uson our tight schedule. At this time, we invite anyand all questions and comments from the floor. Inmy opinion, I think we are getting down to theheart of the matter here. Where do we go, whatdo we do next, to deal with an area that is sopotentially devastating to patients as these sortsof interactions in patients on antithrombotictherapy?

DR. PALMER: I am Dr. Mary Palmer, GeorgeWashington University. My first comment is for Dr.Ernst. In the United States, if there was a casereport, because we have so little backgroundinformation to begin with, I think that is actuallyvery important. Otherwise, you look for informa-tion, you don’t find it, and then you chuck itbecause there is nothing to reinforce it. I thinkthat is what is happening with our safetyinformation, from a poison center perspective.There is just not enough consistency in what wehave, and in the sources. We are in a cycle ofnon-information. We get a case, we are confusedby it, we don’t find confirmation, because weare used to things that have pre-marketing andpost-marketing surveillance — pre-marketingsafety data and post-marketing surveillance—sothat I think we are not used to working with acollection of materials, and especially in thepermutations and combinations in which theycome. Rather, we are used to working with drugsthat have a specific pharmacologic activity. Ithink we are really much more primitive in ourdata on supplements, and case reports with theseagents shouldn’t be held to the same standardthat we are used to holding case reports. I thinkthat is very much a problem, but I don’t knowwhere else one is going to start capturing safetyinformation.

DR. ERNST: I am not even sure I hear a questionthere, but the relative importance of case reports,maybe that is the issue.

DR. PALMER: You had made a statement that youthought that, if there was a case report and youcannot find verification of it, then you don’t take itseriously. That is what I am responding to. I amsaying that, especially as compared to Europe,where you have more of a pharmacovigilence post-marketing system, that the American case reportsare not going to have the same kind of place tocheck and verify and validate.

DR. ERNST: Case reports are obviously poten-tially important. So, one has to take themseriously. They are often not very well writtenup, and that has been addressed by PeterDeSmet. But case reports are only the startingpoint for some deeper investigations. They cannotbe anything else. They are one single anecdote,and even the plural of anecdote is not data, butanecdotes. So, they are the starting point forlooking deeper.

Now, if you take ginkgo and you have a casereport of bleeding with ginkgo, then you have toview this in relation to millions of people takingginkgo. You could theoretically write millions ofcase reports of people bleeding without ginkgo.So, that tells you the importance of thatparticular case report, and it also tells you thatyou ought to look then into the best evidencethat is available, and there are some rather goodclinical studies showing that the effect may bemeasurable in the laboratory but is, most likely,clinically irrelevant.

DR. PALMER: The second, I would like to make acomment to add to what Dr. Lee was saying. Dr.Richard Crowe in the FDA in California keeps adatabase of adverse events with Asian and othercultural remedies, and will assay materials for

Session V: Comments and discussion 171

contaminants that are heavy metal. It is a goodresource. It is not easy to get because it is in hardcopy, but I just throw that out to add to hercompendium.

DR. BUSSEY: If I might just share a couple ofexamples of things that we worked with thataddressed the issue of case reports versus goodclinical trials, years ago we had a case reported fromour group in which sucralfate was added to a patienton warfarin. When the sucralfate was added, the INRfell, when they stopped it the INR came back up.They re-challenged. The INR fell again, they stoppedit and INR came up, a pretty convincing case. Wethen did a randomized, double blind placebo controltrial that did not show any interaction in any of thepatients getting the drug. Now, does that mean thatit doesn’t interact? No. It means it doesn’t interactpredictably in every patient.

Similarly, we did a randomized, double-blindplacebo-controlled trial of cyprofloxacin with war-farin. At the time we were trying to get the studyapproved, the FDA told us that they were consider-ing not approving the study because they thought itwas unnecessary and unethical. Their explanationfor their view was that they had enough case reportsthat they were sure that there was an interactionand they wanted to put a black box warning in theCipro package insert. Eventually, the FDA approvedthe trial which was completed and failed to showany interaction whatsoever. Having said that, I amconvinced that we have seen occasional patientswho have an interaction between warfarin andCipro. On the one hand, I think we all appreciatethe problems with poorly substantiated casereports, like the ones mentioned yesterday, thatthe gentleman stopped eating and started drinkinga liter and a half of cranberry juice a day. I am notsurprised the INR changed. I am not sure thecranberry juice had anything to do with it. On theother hand, we tend to look at randomized, double-blind, placebo-controlled trials as the gold stan-dard, but such trials really cannot, absolutely ruleout the possibility that an interaction may occur inan individual patient.

DR. PALMER: I just want to reinforce that I thinkthe state of the art is so very primitive, and thereare so many problems already with pharmaceuti-cals that I am not quite sure where the best placeto start is. Sometimes case reports are all you haveand, as poor as they may be, may be all that youhave as a sentinel marker.

DR. FUGH-BERMAN: I am Adrian Fugh-Berman,Georgetown University graduate program in town.A friendly suggestion. Certainly the issues withinadequately documented case reports have beenwell covered. Doctors have limited time and

perhaps educability. So, my friendly suggestion isthat perhaps we focus some of these efforts inimproving case reports to journal editors, to havethem throw back case reports that are inadequate-ly documented to the people who are writing them,with instructions on how to better them, thatwould improve case reports.

For Dr. Wittkowsky, or anybody else, there havebeen several people who have brought up the ideaof doing a registry of patients on anticoagulants andsupplements. I would be interested in what thepanel thinks and what sort of — how are changes indiet to be incorporated into a database such as this.

If somebody is using not only drugs, not onlyprescription drugs, of course, there are more than30 drugs that interact with warfarin, includingover-the-counter drugs like miconozole and likecimetidine, and if somebody is using cimetadineand ginger for an upset stomach and has a changein their INR, it would be pretty important to knowthat they are using cimetidine. We all know aboutvitamin K containing foods, but there are otherthings that people might not realize interact withwarfarin. There is the case report of someone whoreversed their warfarin by drinking a gallon ofgreen tea day. It apparently has a very smallamount of vitamin K in it, but if you drink agallon, you can get enough to reverse warfarin.There are case reports of interactions withavocado and other kinds of foods that we don’tnormally think of as interacting. Of course,fasting, changes in protein intake, all of theseother things can affect warfarin as well. So, if weare talking about doing a database on the effectof dietary supplements, how is the effect of dietalso being factored in.

DR. WITTKOWSKY: Part of the assessment ofchange in anticoagulation—over-anticoagulation orunder-anticoagulation—is a thorough assessment ofall potential confounders, from dietary issues tochanges in alcohol intake, changes in underlyingdisease states, et cetera. I would imagine that thepurpose of a registry for dietary supplementinteractions, or one of the goals of such a registry,would be to weed out cases in which there areconfounding factors. That would somehow have tobe incorporated into the data collection, so thatwhat you are looking at is a clean case, or as cleanas it can be, in which other potential confoundersare found not to be the issue, and that the mostlikely or more probable issue is the dietarysupplement. Part of that probability assessmentcomes from using a drug interaction probabilityscale, that would incorporate a scoring system forwhether or not other confounders might be part ofthe explanation.

Session V: Comments and discussion172

DR. BUSSEY: I might add to that — and Dr.Wittkowsky mentioned the scoring system mightneed to be somewhat modified by anticoag clinicsor anticoag patients. We found several years agothat roughly 10% to 15% of the INRs from ouruniversity based lab were seriously erroneouslyhigh. So, whenever I see a graph presented todescribe a drug interaction, the first thing I wonderabout is, did they repeat the INR on another sampleor was this a single elevated INR that could be dueto lab error? If the sample is not collected correctlyand not handled correctly, it is very easy for aperson who has an INR of three to have an INR of sixor seven reported back. So, often that is not reallyquestioned or validated. So, besides all the otherthings that go into that issue, I think that laboratoryvariability is a real issue, if we are focusing onlaboratory results, and we may not have clinicaloutcomes, as has been mentioned by several of thespeakers.

DR. DOFFREY: I am Nina Doffrey and I work for acompany that is a custom manufacturer of dietsupplements. So, I have a little bit differentperspective on things, but I think I appreciate,from being here a little while, the enormity of theproblem, and also a reflection on the problem withthe synergy or antagonism that can occur withmultiple ingredients in a product.

Where I work, I have oversight of formulation ofsome of these products, actually all of them, and Isee, because we manufacture for companies thatthen sell to the end consumer that, no matter whatwe say about our concerns about putting multipleingredients in a product, or about some of theconcerns about synergy or antagonism of thoseingredients, or side effects, or interactions withdrugs or even other herbals or foods themselves,the customer still believes that more is better andthey want to do that. So, it is a substantial problemfor those of us within the industry that are trying tofollow safe guidelines as well.

I also have oversight of adverse events, and Iparticularly appreciate the comments of Dr. King-ston. Because I do a lot of research in regard toformulation in regard to these adverse events, Iread a lot of clinical studies. Even months, years,before the JAMA study came out on HRT (hormonereplacement therapy), I knew that I did not want tofollow that accepted risk in taking HRT. Along thosesame lines, in regard to ephedra—and we don’tadvocate and never did advocate at our companythe use of ephedra—I think it points out theimportance of the media. Here, in the mostpublicized case, when ephedra involved a footballplayer, I think he was, that had multiple othercompounding events in his life that could have

caused an adverse event, could have caused theheat stroke that he had, and yet ephedra got thepredominance of blame for that, we see that thereis a great deal of problem involved in that.

I think part of the problem—I am hoping the FDApeople will discuss this, this afternoon—could be inthe way that dietary supplements have to belabeled. The way that they are labeled nowprevents us from making any type of statementconcerning any type of therapeutic use of thatproduct. So, in a way, anything that would indicatea therapeutic use says that that is a drug, and wedon’t want to get into the drug business and yet, atthe same time, it leads the public to believe thatthese products are totally safe, they are foods,more is better, I can take three, I can take four, andit is going to be safe, and they really don’t knowwhat the use may be for that dietary supplement,which may have a therapeutic use.

If we had some sort of intermediary way oflisting certain dietary supplements that haveresearch behind them, that goes not into the OTCor drug status, stays within the dietary supplementbut it is more than just totally safe, then that maybe a more effective way to tell the public what thatdietary supplement actually does, and what theprecautions should be in regard to other things thatthey are taking alongside it.

Now, down to my question. One of the thingsthat has been pointed out a little bit, most of whathas been discussed has been interactions involvingherbals and vitamin B, vitamin C, Co-Q10, some ofthose, but one of the focuses that we have, andsome of the research that we have done at thecompany that I work for regards high dose proteo-lytic enzymes. I know there was mention of a cowstudy, where the cows ate the sweet clover thatwas soured. Even though I understand that therewas warfarin found in that, it looks to me, sinceaspergillus was involved, that perhaps there wasalso a proteolytic enzyme there as well.

Also, there was reference made to the use ofcertain bloodsuckers for their saliva. There, again,you have a proteolytic enzyme. Then, withpapaya, pineapple, the bromelain from pineapple,you have another proteolytic enzyme. I know, foryears, it was considered that these would notsurvive, orally ingested enzymes would not survivethe gastric juices of digestion of endogenousenzymes. I think there is a preponderance ofresearch now that shows that they do, and wehave seen some effects on clotting from that. Iwas wondering if anyone has any experience onthat. I know there are at least two enzymes thatare being promoted. Incidentally, we do not carrythem because we are concerned about their

Session V: Comments and discussion 173

regulation, one derived from natto, a soy deriva-tive, and another derived from a silk worm, thatare promoted for their fibrinolytic and anticoag-ulant properties. I was wondering if there hasbeen any knowledge of research in this area. Thisis an area that we are very much interested in.

DR. BUSSEY: In the interest of time, we will keepthe answer short. Do any of the panelists have anyexperience with proteolytic enzymes being orallyadministered, and either benefits or harm? I knowwe were trying to find some information on papaya,and I am still looking for that one case report from1997, which is not available on line.

DR. HOFFMAN: We will be talking about thera-peutic aspects of these products in the 1:00 o’clocksession. So, please stay tuned, but let me addressthe group, the entire group. I am working on theheterogeneous products, and I know one of thebiggest areas of general concern with this confer-ence have been the botanicals and other naturalproducts. So, let me speak as the chair of thenatural health products special interest area for theDrug Information Association, which is a non-profitgroup that deals with regulatory issues and policies.It includes regulators and industry. One of the thingsthat I think has been sort of an ongoing problem inbringing science and looking at these problems hasbeen the lack of the basic scientist. If we weretalking about infectious organisms, we would havemicrobiologists at this meeting, and I do not seemany botanists or plant physiologists on the panel. Iknow there are some in the audience here.

Basically, one of the things that I have found indoing conference on botanicals and the regulationof botanicals, has been that the botanists and theplant physiologists have not really been vestedwith this area. There is a big gap between theactual basic research and the science and whatthey are off doing. They are doing some veryelegant work in terms of what plants do and don’tdo, and it is not being brought into the clinical,into the pharmaceutical, into the dietary supple-ment industry. It is being brought into the foodindustries and into the agriculture and biotechindustries, but you are not seeing a cross over. Ithink it is very evident here that a lot of theliterature that is missing is not literature youwould normally be looking at. Their questions areprobably not the questions that they should befocused on to answer your questions.

I think one of the things I would like to see out ofthe conference would be to get the basic scientistsin this area more involved in the research questionsthat you are interested in.

DR. BUSSEY: I think that is a good point. In fact,several speakers have made the point that products

frequently aren’t what they say they are, and itwould seem that basic scientists could be able toassure that we actually get what the product says.

We need to cut this off in just about six or sevenminutes. So, I want to try to get to the questions ofthose who are waiting, but try to be brief and tothe point, please.

DR. DWYER: Just 30 s. First of all, the idea ofimproving the quality of case reports by havingclinical pharmacologists and hematologists andother disciplines working together, I think, is veryimportant and I hope this happens.

Secondly, on the notion of the registry, I stronglyendorse the notion of not just including the dietarysupplements but also the confounders, the dietaryconfounders, the food as well as regimens and thecommon OTC drugs that I think you already do.

The nutritionists really aren’t very much in theanticoagulation clinics, and I think they do havesomething to contribute.

DR. GAGE: In terms of dietary factors, we saw norelationship between 48-h patient recall of foodsrich in dietary vitamin K and the INR or the warfarindose (Gage BF, et al. Use of pharmacogenetics andclinical factors to predict the maintenance dose ofwarfarin Thromb Haemost. 2004 Jan; 91:87—94).We could explain 39% of the variability in warfarindose. What that tells me is that, at least forpatients who are attending an anticoagulationservice (who have been educated to avoid bingeingon vitamin K rich foods), that it is difficult toquantify the vitamin K effect. We know that atmedicinal doses vitamin K affects warfarin, but toquantify the effect of the usual dietary intake isdifficult.

DR. BUSSEY: We participated in a prospectiverandomized trial where patients had to do dietarycalendars that required them to check off servingsof example foods. One patient, and I recallparticularly that she was a nurse, decided just toleave those foods off during the study because shedidn’t want to have to bother checking them off. Ofcourse her INR changed during the study. So,if youstart trying to be too aggressive in documentingdietary intake, you may inadvertently change theintake. In other words, you get into damned if youdo and damned if you don’t sometimes.

DR. KINGSTON: I would just like to, first of all,agree with Dr. Dwyer’s comments on the casereports and the registry, I concur with those. ThenI would like just expand that point a little bitfurther. I think, certainly, improving the quality ofcase reports would benefit us all and help ourassessments. On the other hand, I think that I am alittle concerned about making an exclusion criteriabased on case reports that don’t meet certain

Session V: Comments and discussion174

parameters, because I am not sure we want tothrow out information at an early stage. Certainlythat needs to be taken into consideration when weevaluate the relevance and the weight of variousreports because, after all, there is some judgmentthat needs to be applied in the end, and I think it isbetter applied on the basis of the full data.

Secondly, I think it is important to distinguishbetween our ability to control for factors in a casereport because, no matter how expert the judg-ment of the person evaluating those factors is, thatis nonetheless a subjective judgment, and that isdifferent than a controlled study in which we canascribe some statistical parameters to our controls,et cetera. I think in this entire area this isimportant. Once again—I think I have said this in anumber of ways today—but I think it is important tomake the distinctions between data and judg-ments, no matter how expert the judgements.

DR. YOUNG: I am Joan Young, from the PlateletDisorder Support Association. There aren’t manypatients here. I may be the only one. So, I feel Ihave to speak for them.

We hear a lot about patients being put onCoumadin, and then they do other things thatinterfere. I would like to say, in the suggestions,that patients—at least the ones I know and myself—many times aren’t trying to undermine what ishappening. They are trying to help. They are tryingto be well without some of the toxicities.

If I look at Coumadin, which I am learning a lotabout, if I look at myself as a patient I say, well,would I like to take rat poison or ginkgo, and maybeginkgo plus grape juice. I think if I knew enoughabout either one of them, knew enough to makethat judgment, I would certainly opt for somethingthat is not rat poison. I think, in the ultimatesolution of how to—let’s say how to maintainhemostasis in these patients, that perhaps thereis a solution that combines these two things,because obviously there is efficacy in a lot of thecompounds that we heard. So, maybe there is a wayof combining to, in fact, reduce the overall toxicityto the patient, and maybe even reduce the costs,but have it easier to be monitored.

So, that is my statement from a patient view-point. It is probably a little heretical, but I thoughtI had to say it.

DR. BUSSEY: Let me rebut just a little bit, andplease share this with any of your patients that maybe in need of warfarin. It is not a rat poison. Ratsare now immune to it. It rather is a natural productderived from herbs and, as Dr. Gannon was talkingabout with atrial fibrillation, for patients with

atrial fibrillation, if they take warfarin and arefairly well controlled, it reduces their risk of strokeby more than 80%, and we have very, very fewtherapies that are that effective. It also is verysafe, if it is managed correctly. So, we have proven,absolutely proven benefit, of a huge magnitude,scary side effects if it is not managed appropriately,but quite good if it is.

Unfortunately, I think we need to wrap things up.I have been jotting down questions the whole timeeveryone else was speaking, and I am sorry wecan’t get to those.

DR. WU: A very short question, but I think it is animportant question. When I visited China last year,they invited me to look at the intake rates,traditional and western. There is a huge hospital,a 1200 bed. My question, actually, is very relevantto this one. So, they have oncologists working, andthey select some of the medications to treatnausea, vomiting, pain, depression. They also havemedications to treat the very intangible symptoms.I like your idea about hypothesis driven testing, buteverything we do here will be extremely expensive.Would you comment on, how do you set thehypothesis, given that many of those herbs do noteven have anything that one can rely on to pose thehypothesis, yet, if you put up a very large clinicaltrial, that is a very expensive trial.

I think that is probably a very critical issue forthe western world. Even for the Chinese, they arevery difficult and they don’t even want to do it.Can you comment on that one?

DR. ERNST: Yes, I will try to comment. In a way,traditional views, if, in China, something has beenused for 1000 years, that, in itself, amounts to ahypothesis of efficacy and of safety. It is a strangetype of hypothesis. It is not what scientists call ahypothesis, but that is all that we have. So, I thinkwe should regard this as a hypothesis and take itfrom there.

Obviously, submitting everything to RCTs (ran-domized controlled trials) is hugely expensive, but Iam afraid that I don’t know any short cut for goodscience.

DR. BUSSEY: Thank you very much, and again, Iwould like to thank all the panelists. You have beenheroic and flexible in dealing with the rapid firesequence here.

Also, as Dr. Marder pointed out to me, thereshould be time later in the afternoon for morequestions and discussions on this.

I am not sure, we may have people leaving, buthopefully we can revisit some of these issues.Thank you very much.

Documents

Opinions and research priorities