This article provides an overview of the clinical evidence of interactions between herbal and conventional medicines. Herbs involved in drug interactions – or that have been evaluated in pharmacokinetic trials – are discussed in this review. While many of the interactions reported are of limited clinical significance and many herbal products (e.g. black cohosh, saw palmetto, echinacea, hawthorn and valerian) seem to expose patients to minor risk under conventional pharmacotherapy, a few herbs, notably St. John’s wort, may provoke adverse events sufficiently serious to endanger the patients’ health. Healthcare professionals should remain vigilant for potential interactions between herbal medicines and prescribed drugs, especially when drugs with a narrow therapeutic index are used.
According to the World Health Organisation, herbal medicines are defined as ‘finished, labelled medicinal products that contain as active ingredients aerial or underground parts of plants, or other plant material, or combinations thereof, whether in the crude state or as plant preparations. Plant material includes juices, gums, fatty oils, essential oils, and any other substances of this nature. Herbal medicines may contain excipients in addition to the active ingredients. Medicines containing plant material combined with chemically defined active substances, including chemically defined, isolated constituents of plants, are not considered to be herbal medicines’ . Thus, herbal medicines contain a combination of pharmacologically active plant constituents that are claimed to work synergistically to produce an effect greater than the sum of the effects of the single constituents [2,3,4,5]. There is a general belief by the public that herbal medicines are safe because they are natural. However, this is a hazardous oversimplification. Many different side effects to herbs have been reported and recently reviewed [6,7], including adverse events caused by herb-to-drug interactions [6,7,8]. Since all herbal medicines are mixtures of more than one active ingredient, such combinations of many substances obviously increase the likelihood of interactions taking place. Hence, theoretically, the likelihood of herb-to-drug interactions is higher than drug-to-drug interactions, if only because synthetic drugs usually contain single chemical entities.
The aim of this article is to provide an overview of the clinical data regarding the interactions between herbal remedies and prescribed drugs. Detailed considerations on the mechanisms and molecular explanations of the clinical observations of herb-to-drug interactions can be found elsewhere [9,10,11]. The herbal remedies involved in clinical herb-to-drug interactions are given in table 1, which also reports the level of evidence for each interaction. The ultimate goal of this article is to raise the awareness of pharmacists and physicians regarding this topic and thus protect the health of consumers.
Mechanisms of Herb-to-Drug Interactions: General Considerations
Herb-to-drug interactions are based on the same pharmacokinetic (changes of plasma drug concentration) and pharmacodynamic (drugs interacting at receptors on target organs) principles as drug-to-drug interactions.
The pharmacokinetic interactions that have been identified so far all point towards the fact that a number of herbs, most notably St. John’s wort, can affect the blood concentration of different conventional medicines that are metabolized by cytochrome P450 (CYP, the most important phase I drug-metabolizing enzyme system) and/or are transported by P-glycoprotein (a glycoprotein which influences drug absorption and elimination by limiting the cellular transport from the intestinal lumen into epithelial cells and by enhancing the excretion of drugs from hepatocytes and renal tubules into the adjacent luminal space). Polymorphisms in the genes for CYP enzymes and P-glycoprotein may influence the interactions mediated through these pathways . Probe drugs used in pharmacokinetic trials include midazolam, alprazolam, nifedipine (CYP3A4), chlorzoxazone (CYP2E1), debrisoquine, dextromethorphan (CYP2D6), tolbutamide, diclofenac and flurbiprofen (CYP2C9), caffeine, tizanidine (CYP1A2) and omeprazole (CYP2C19). Fexofenadine, digoxin and talinolol have been extensively used in pharmacokinetic trials as P-glycoprotein substrates.
Pharmacodynamic interactions have been less studied but may be additive (or synergetic), i.e. the herbal medicines potentiate the pharmacological/toxicological action of synthetic drugs, or antagonistic, i.e. the herbal medicines reduce the efficacy of synthetic drugs. Warfarin interactions are a classical example of pharmacodynamic interactions. Theoretically, increased anticoagulant effects could be expected when warfarin is combined with coumarin-containing herbs (some plant coumarins exert anticoagulant effects) or with antiplatelet herbs. Conversely, vitamin K-containing herbs can antagonize the effect of warfarin (the action of warfarin is due to its ability to antagonize the cofactor function of vitamin K).
Comprehensive review articles specifically highlighting the mechanisms of herb-to-drug interactions, including evidence of herbs that can modulate CYP or P-glycoprotein have recently been published [9,10,11,12].
Level of Clinical Evidence
In this article, clinical evidence has been categorized into the following levels:
Level 1: incomplete case report, presence of other explanatory factors for the adverse reaction, adverse event unlikely from a pharmacological viewpoint.
Level 2: case report providing some evidence for an interaction, other causes not fully excluded (e.g. interactions indicated as ‘probable’ or ‘possible’ by the Naranjo probability scale).
Level 3: well-documented case report; multiple case reports, case series.
Level 4: pharmacokinetic trials in patients or healthy volunteers.
Level 5: interaction highlighted by case report(s) and confirmed by clinical pharmacokinetic trials.
Level of evidence ‘not applicable’: adverse event highlighted by case report(s) and not confirmed by clinical trials, contradictory data from different clinical trials.
Clinical Interactions between Herbs and Conventional Drugs
An overview of the clinical data regarding herb-to-drug interactions for a number of herbal remedies known to interact with conventional medicines is reported below.
Aloe vera (Fam. Liliaceae) is used in western countries as a laxative (A. vera latex, which contains anthraquinones) and for dermatologic conditions (A. vera gel, containing mainly mucilages) [2,4]. In traditional Chinese medicine, A. vera is mainly employed for inflammatory conditions, diabetes and hyperlipidaemia. Blood loss during surgery as a result of a possible interaction between A. vera and the anaesthetic sevoflurane has been reported . An additive effect on platelet function has been hypothesized but not proven since both sevoflurane and A. vera ingredients may inhibit platelet aggregation.
Black Cohosh (Cimicifuga racemosa)
Black cohosh (Cimicifuga racemosa rhizome and roots, Fam. Ranunculaceae), mostly used to treat symptoms of menopause [2,3], has been associated with serious safety concerns, such as hepatotoxicity, which urgently require further investigation [3,4].
The effect of black cohosh extract on the activity of human CYP enzymes as well as on P-glycoprotein has been evaluated in a number of clinical trials [14,15,16,17] using different probe drugs, including caffeine, midazolam, chlorzoxazone, debrisoquin and digoxin. The results suggest that black cohosh is unlikely to affect the pharmacokinetics of conventional drugs that are metabolized by CYP1A2, CYP3A4, CYP2E1 and CYP2D6 or are substrates of P-glycoprotein. In addition, seven different brands of commercial black cohosh products were found not to affect human CYP using an in vitro liver microsomal technique . On the whole, black cohosh seems to pose only minor risks in patients undergoing conventional pharmacotherapy.
Cat’s Claw (Uncaria tomentosa)
Cat’s claw (Uncaria tomentosa, Fam. Rubiaceae) is a medicinal plant from the Amazon rainforest. Due to its immunostimulant and antiviral effects, it has been used for conditions, such as rheumatoid arthritis and AIDS . Cat’s claw has been shown to increase the plasma concentration of the protease inhibitors atazanavir, ritonavir and saquinavir . In vitro, cat’s claw has been shown to inhibit CYP3A4, which is responsible of the metabolism of the protease inhibitors. However, no human data on the possible modulation of CYP enzymes by cats’ claw have been provided to date.
Chamomile (Matricaria recutita)
Chamomile, consisting of fresh or dried flower heads of Matricaria recutita (Fam. Asteraceae), is used both externally (for skin and mucous membrane inflammations) and internally (for the treatment of gastrointestinal spasms and inflammatory disease of the gastrointestinal tract) [4,5]. Chamomile contains coumarins, a large class of over 1,300 natural compounds. Some, but definitely not all, coumarin compounds may exert an anticoagulant effect . A case of rectus sheath and retroperitoneal haematomas was reported in a patient under warfarin therapy . It was believed, but not proven, that the coumarin constituents of chamomile may have worked synergistically or additively with warfarin, resulting in overanticoagulation.
Cranberry (Vaccinium macrocarpon)
Cranberry is the American name of the fruit of Vaccinium macrocarpon (Fam. Ericaceae); it has been used for decades to prevent urinary tract infections [3,4], generally in the form of an encapsuled standardized extract, a dilute juice or a dried-juice capsule .
On the basis of multiple published cases (including 2 cases of fatal interaction) reporting increased international normalized ratio (INR) and haemorrhage [21,22,23,24,25,26,27,28,29,30,31], serious concerns have been raised regarding a possible interaction with the anticoagulant warfarin. However, these warnings may possibly be attributed to misleading conclusions .
With the exception of one study, which showed that capsules containing concentrated cranberry juice increased the area under the INR-time curve of warfarin by 30% , a number of clinical trials have consistently shown that cranberry juice, even administered at high doses, did not cause any clinically relevant changes in warfarin pharmacokinetics and pharmacodynamics [34,35,36,37,38]. Clinical evidence indicates the lack of interaction between cranberry juice and some CYP isoenzymes, e.g. CYP2C9, CYP1A2 and CYP3A4 [36,37,38]necessary for warfarin metabolism . Finally, a clinical trial found that pomelo juice, but not cranberry juice, affected the pharmacokinetics of cyclosporine (CYP3A4 and P-glycoprotein substrate) in humans .
Danshen (Salvia miltiorrhiza)
Danshen, also known as Chinese salvia or red salvia, are preparations derived from the roots and rhizome of Salvia miltiorrhiza (Fam. Lamiaceae). Danshen is widely used in traditional Chinese medicine to prevent and treat cardiovascular conditions, such as acute ischemic stroke and myocardial infarction [2,3,4]. Danshen can affect haemostasis in several ways, including inhibition of platelet aggregation. Case reports have highlighted the possibility of interactions between warfarin and danshen, resulting in an increased anticoagulant effect [41,42,43,44]. A pharmacokinetic mechanism seems unlikely since danshen has been shown to induce intestinal CYP3A4 in 14 healthy volunteers .
Dong Quai (Angelica sinensis)
Angelica sinensis (Fam. Apiaceae), commonly known as ‘dong quai’, is one of the most popular traditional Chinese medicines . Preparations from its roots are used mainly for dysmenorrhoea, amonorrhoea or excessive menstrual flow. The actions of dong quai are said to be due to the presence of a number of chemical constituents, including coumarins , which may have anticoagulant actions . Two well-documented case reports suggest overanticoagulation following co-administration of warfarin and dong quai [46,47].
Echinacea (Echinacea spp.)
Echinacea preparations derive from underground as well as aerial parts of several species of Echinacea (Fam. Asteraceae), e.g. E. angustifolia, E. pallida and E. purpurea . Due to its immunostimulant properties, echinacea is widely used for the prevention and treatment of common infections, such as respiratory tract infections [2,3,4].
Echinacea seems to pose no serious risk for drug interactions in humans. No verifiable case reports of drug-to-herb interactions with any echinacea product have been published to date. Echinacea did not change the pharmacokinetics of digoxin, a P-glycoprotein substrate  nor did it alter the pharmacokinetics of chlorzoxazone (CYP2E1 probe) , debrisoquine (CYP2D6 probe) [17,49], dextromethorphan (CYP2D6 probe)  or tolbutamide (CYP2C9 probe) . Some studies have found that echinacea affects caffeine (CYP1A2 probe) and midazolam (CYP3A4 probe) pharmacokinetics; however, this has not been confirmed by other clinical trials [17,49].
Finally, a recent clinical trial showed that E. purpurea root extract did not affect the overall darunavir or ritonavir (a combination of protease inhibitors) pharmacokinetics in HIV patients . Protease inhibitors are mainly metabolized by CYP3A4 and are P-glycoprotein substrates.
Eleuthero (Eleutherococcus senticosus)
Eleuthero, also named ‘Siberian ginseng’, belongs to the same family (Araliaceae) as Asian ginseng (Panax ginseng). Like Asian ginseng, eleuthero is promoted as a ‘tonic for invigoration and fortification in times of fatigue and debility or declining capacity for work and concentration, also during convalescence’ .
Eleuthero, at generally recommended over-the-counter doses, is unlikely to alter the disposition of co-administered medications primarily metabolized by CYP2D6 or CYP3A4 .
Increased levels of digoxin have been associated with ingestion of eleuthero . In this case, the patient was asymptomatic for digoxin toxicity despite high plasma levels of the cardiotonic drug. Since eleuthero contains glycosides with structural similarities to digoxin that interfere with digoxin assays, this is not a real clinical herb-to-drug interaction, but rather represents an artefact of digoxin assays.
Garlic (Allium sativum)
Garlic (Allium sativum L., Fam. Alliaceae) is used in modern phytotherapy to treat hypercholesterolaemia and prevent arteriosclerosis although the clinical evidence is far from compelling [2,3]. Garlic preparations include garlic powder standardized to contain 1.3% alliin and 0.6% allicin, garlic aged extract, which does not contain allicin but is high in water soluble phytochemicals, such as diallyl sulphides and garlic oil (i.e. essential oil obtained from the distillation of the cloves) .
Two garlic preparations, namely garlic oil and garlic powder, have been evaluated for their potential to affect CYP enzymes in clinical trials. The results suggest that garlic oil may selectively inhibit CYP2E1, but not other CYP isoforms (such as CYP1A2, CYP3A4 or CYP2D6) and that garlic powder has no effect on CYP3A4 [54,55,56,57,58]. Recently, it has been shown that a 21-day garlic treatment (aged garlic extract) induces intestinal expression of P-glycoprotein without affecting intestinal or hepatic CYP34A in humans .
The most thoroughly studied garlic interactions with conventional drugs include interactions with the anticoagulant warfarin, which, in any case, have not been confirmed by controlled clinical trials or antiretroviral drugs (see details in table 1) [60,61,62,63,64,65,66,67]. Other irrelevant and/or poorly documented interactions include changes in paracetamol pharmacokinetics  and hypoglycaemia when combined with the antidiabetic drug chlorpropamide .
Ginger (Zingiber officinale)
Ginger (rhizome of Zingiber officinale, Fam. Zingiberaceae) preparations are effective in attenuating nausea and vomiting during pregnancy and during the post-operative period [2,3,4]. They showed considerable antiplatelet effects in preclinical studies  and this might explain the elevated INR in a patient taking it concomitantly with the anticoagulant phenprocoumon . However, such an interaction has not been confirmed by a clinical trial .
Ginkgo (Ginkgo biloba)
Extracts from the leaves of the ginkgo tree (Ginkgo biloba, Fam. Ginkgoaceae) are used for the treatment of cognitive impairments, dementia, intermittent claudication and tinnitus [2,3,4,5]. The effect of ginkgo on various CYP isoforms as well as on P-glycoprotein has been investigated in a number of clinical trials by using different probe drugs, such as alprazolam, midazolam, diazepam, nifedipine (CYP3A4), caffeine (CYP1A2), chlorzoxazone (CYP2E1), debrisoquine (CYP2D6), tolbutamide, diclofenac, flurbiprofen (CYP2C), omeprazole, voriconazole (CYP2C19), fexofenadine, digoxin and talinolol (P-glycoprotein substrates) [55,56,72,73,74,75,76,77,78,79,80,81,82]. Given the heterogeneity of the results, firm conclusions cannot be drawn. Nevertheless, the results seem to suggest minor or no effect of ginkgo on the various CYP isoforms or on P-glycoprotein.
It is often mentioned that ginkgo can interact with anticoagulant drugs [2,3,4]. However, clinical evidence refuted this notion since this herbal product has been shown not to affect blood coagulation or platelet function in humans . Clinical trials have also shown that ginkgo has no additive effect with aspirin on platelet aggregation , does not change the antiplatelet activity of clopidogrel and cilostazol  and has no effect on warfarin INR and platelet aggregation [71,86]. In light of these recent controlled clinical data, causality and mechanisms advanced in previous case reports, in which ginkgo was suspected to cause spontaneous hyphaema when associated with aspirin , intracerebral haemorrhage when associated with warfarin  and intracerebral mass bleeding when associated with ibuprofen , should be re-examined.
Finally, single cases suggest that ginkgo may cause priapism when combined with the antipsychotic drug risperidone , coma when combined with the atypical antidepressant trazodone , fatal seizure when combined with the anticonvulsant drugs valproic acid and phenytoin  and virological failure when combined with efavirenz, a non-nucleoside reverse transcriptase inhibitor .
It should be noted that, in clinical trials, EGb 761, a well-defined extract of Ginkgo biloba leaves, standardized to contain 24% flavone glycosides and 6% terpene lactones, has been used. EGb 761 has generally not been implicated in case reports .
Ginseng (Korean Ginseng, Panax ginseng)
Preparations of Asian ginseng, obtained from the roots of Panax ginseng (Fam. Araliaceae), are used to reduce susceptibility to illness, promote health and longevity, restore male sexual function and aid convalescence [4,5]. Pure ginsenosides can inhibit platelet aggregation in vitro . However, clinical studies have consistently demonstrated that ginseng extracts have had no significant effect on platelet function in humans  and did not change the pharmacokinetics or pharmacodynamics of warfarin [95,96,97]. Surprisingly, a decreased anticoagulant effect has been reported in a patient taking both ginseng and warfarin .
Case reports suggest potentially serious interactions when ginseng is used with the antidepressant phenelzine and the anticancer drug imatinib [99,100,101] (see table 1 for details). Finally, the clinical results consistently showed that ginseng does not affect CYP enzymes although a slight inhibition of CYP2D6 has been observed [55,56].
Ginseng (American Ginseng, Panax quinquefolius)
Panax quinquefolius (Fam. Araliaceae), commonly known as ‘American ginseng’, is a herbaceous perennial herb native to North America [4,5]. A clinical study showed that American ginseng reduced the anticoagulant effect of warfarin in healthy volunteers  (see table 1 for further details). On the other hand, two clinical trials have recently shown that American ginseng did not affect the pharmacokinetics of the antiretroviral drugs indinavir and zidovudine [103,104].
Goldenseal (Hydrastis canadensis)
Goldenseal (Hydrastis canadensis, Fam. Ranunculaceae) has a history of folk medicine use in the treatment of gastrointestinal disturbances, urinary disorders, skin ailments and various infections [2,4]. A clinical trial showed that goldenseal did not change the disposition of digoxin, suggesting that this herb has no effect on P-glycoprotein . Although one study did not yield the same conclusions , convincing clinical evidence suggests that adverse herb-to-drug interactions may result with concomitant ingestion of goldenseal and drugs that are metabolized by CYP3A4 or CYP2D6 [14,17,107,108]. Therefore, although no clinical case report of herb-to-drug interaction has been published to date, goldenseal should be not administered concomitantly with drugs that are metabolized by CYP3A4 or by CYP2D6.
Green Tea (Camellia sinensis)
Green tea (Camellia sinensis leaves, Fam. Theaceae) is used both as a beverage and as a herbal drug . Possibly due to its vitamin K content, green tea might reduce the anticoagulant effect of warfarin . Furthermore, green tea has been shown to reduce acid folic and the plasma level of statins through a mechanism that remains to be clarified [110,111]. Lastly, green tea has minor effects on human CYP3A4 [112,113].
Kava (Piper methysticum)
Preparation from the rhizome and roots of Piper methysticum (Fam. Piperaceae) are used for the treatment of anxiety, and the available evidence suggests that kava extracts are superior to placebo for treating patients with anxiety disorders [2,3,4]. Unfortunately, in the UK and various other European countries, the sale of kava is currently prohibited due to reports of potential hepatotoxicity .
In vitro, kavalactones, the active ingredients of kava, have been shown to be potent inhibitors of several enzymes of the CYP450 system . However, clinical trials have shown that, at therapeutic doses, kava inhibits CYP2E1 but not other CYP isoforms, such as CYP3A4, CYP2D6 or CYP1A2. Kava does not affect P-glycoprotein [14,17,105,107,108].
Some possible pharmacodynamic interactions, highlighted by single case reports have been postulated to occur when combining kava with benzodiazepines, anti-Parkinson or antidepressant drugs (see table 1 for further details) [115,116,117].
Licorice (Glycyrrhiza glabra)
The roots and rhizomes of Glycyrrhiza glabra(Fam. Fabaceae) are mainly used for the treatment of peptic ulcer and catarrhs of the upper respiratory tract [2,3,4,5]. A preliminary report, published in abstract form only, showed that the ingestion of aqueous licorice extract for 7 days did not significantly alter the pharmacokinetics of midazolam, a CYP3A4 substrate . However, both glycyrrhizin and glycyrrhetic acid (i.e. chemical components of licorice) have recently been shown to induce CYP3A4 in humans [119,120]. In the absence of definitive data for standardized licorice extracts, it is suggested that this herbal remedy should be used with caution when taken concomitantly with other drugs that interact with CYP3A4.
There is some indirect evidence that licorice may affect the pharmacokinetics of prednisolone. Glycyrrhizin is known to increase the plasma prednisolone concentration in humans and is one of the ingredients of three major traditional Chinese formulations, namely Sho-saiko-To, Saiboku-to, and Sairei-To, which all affected prednisolone pharmacokinetics in healthy volunteers [121,122].
Milk Thistle (Silybum marianum)
Phytotherapeutic milk thistle preparations are obtained from Silybum marianum (Fam. Asteraceae) and are used to treat liver diseases [2,3,4]. S. marianum extracts seem to have minor effects on the pharmacokinetics of drugs metabolized by CYP enzymes or transported by P-glycoprotein. With the exception of one study , several clinical trials have reliably shown that S. marianum extracts did not affect the pharmacokinetics of a number of drugs metabolized by various CYP isoforms (e.g. CYP1A2, CYP2D6, CYP2E1 and CYP3A4) and/or transported by P-glycoprotein [15,16,124,125,126,127,128,129,130]. Overall, milk thistle seems to pose no risk for drug interactions in humans.
Peppermint (Mentha piperita)
Peppermint leaf and oil from Mentha piperita (Fam. Labiateae) have a long history of use in digestive disorders [3,4]. Recent evidence suggests that enteric-coated peppermint oil may be effective in relieving some of the symptoms of irritable bowel syndrome . Some clinical data suggest that peppermint might increase the levels of drugs metabolized by CYP3A4, such as felodipine .
Red Yeast Rice
Red yeast rice is produced by fermentation of washed and cooked rice using the fungus Monascus purpureus and is used to lower blood cholesterol [3,4]. Red yeast rice has been suspected to cause rhabdomyolysis in a a stable renal-transplant patient under cyclosporine treatment  (see table 1 for further details). It should be noted that red yeast rice may cause myopathy even when administered alone .
Saw Palmetto (Serenoa repens)
Serenoa repens (Fam. Arecaceae) preparations are well tolerated by most users and are not associated with serious adverse events [2,3,4]. No evidence for drug interactions with saw palmetto has been published. Two clinical studies demonstrated that saw palmetto had no significant effect on CYP1A2, CYP2D6, CYP2E1 or CYP3A4 in healthy volunteers [50,134]. Extracts from S. repens berries are the most widely used herbal preparations for the treatment of benign prostatic hyperplasia [2,3,4,5,200]. Saw palmetto, pumpkin and vitamin E are ingredients of curbicin, a herbal formulation used to relieve symptoms associated with benign prostatic hyperplasia. Two cases of increased INR were reported after co-administration of curbicin and warfarin ; the INR normalized after discontinuation of curbicin. No anticoagulant effect has been found in the literature associated with both saw palmetto and pumpkin. However, vitamin E has been shown to antagonize the effect of vitamin K and may lead to an increased risk of bleeding, particularly in patients taking oral anticoagulants . The currently available evidence suggests that saw palmetto is unlikely to pose serious health threats to patients combining it with conventional drugs.
Schisandra chinensis (Wuweizi, Fam. Schizandraceae) is used in modern Chinese medicine as an adaptogenic drug . A clinical trial showed that the herb increased the area under the curve and Tmax of tanilolol, a P-glycoprotein substrate . Thus, patients receiving S. chinensis might require dose adjustments when treated with drugs primarily transported by P-glycoprotein.
Schisandra sphenanthera (Nan-Wuweizi) is widely used to treat viral and drug-induced hepatitis in China . Two clinical trials showed that extracts obtained from S. sphenanthera increased the oral bioavailability of the immunosuppressive drug tacrolimus, which is metabolized by CYP3A4 and P-glycoprotein [138,139]. A further study showed that S. sphenanthera increased the oral bioavailability of midazolam (CYP3A4 substrate) . Overall, S. sphenanthera preparations should not be co-administered with CYP3A4-metabolized drugs.
Soy (Glycine max)
Soy beans, obtained from Glycine max (Fam. Fabaceae), are very rich in phytoestrogens, i.e. non-steroidal plant-derived compounds possessing a weak oestrogenic activity. Soy phyto-oestrogens are claimed to exert beneficial effects in the treatment of menopausal symptoms and prevention of heart disease and cancer [2,4]. Decreased INR has been reported in a patient under warfarin therapy . On the other hand, a clinical study showed that a 14-day treatment with soy extract did not significantly influence the pharmacokinetics of losartan and its active metabolite E-3174 in 18 healthy Chinese female volunteers .
St. John’s Wort (Hypericum perforatum)
Hypericum perforatum L. (St. John’s wort) extracts are widely used as a safe alternative to conventional antidepressant drugs for mild to moderate forms of depressive disorders [2,5]. The herb contains numerous compounds with documented biological activity, including the naphthodianthrone hypericin, a broad range of flavonoids, and the phloroglucinol hyperforin, which inhibits the re-uptake of several brain neurotransmitters, including 5-hydroxytryptamine (5-HT, serotonin) .
The possible interactions with conventional medicines are the most important risk associated with the intake of H. perforatum extracts . St. John’s wort represents the herbal product that is most involved in herb-to-drug interactions. Clinical evidence suggests that St. John’s wort may cause both pharmacokinetic and pharmacodynamic interactions. Using well-established probe drugs, a great number of clinical trials have consistently shown that St. John’s wort induced P-glycoprotein as well as CYP3A4, CYP2E1 and CYP2C19, with no effect on CYP1A2, CYP2D6 or CYP2C9 [144,145,146,147,148,149,150,151,152,153,154,155,156,157]. Induction of CYP enzymes and P-glycoprotein is caused by hyperforin via activation of the pregnane X receptor [158,159,160,161].
Pharmacodynamic interactions may occur when St. John’s wort is given together with drugs that enhance 5-HT signaling in the brain (e.g. 5-HT re-uptake inhibitors, 5-HT ligands). St. John’s wort has been shown to clinically interact with a number of conventional drugs mostly via these pharmacokinetic and/or pharmacodynamic mechanisms; such interactions take place with immunosuppressants (cyclosporine, tacrolimus, prednisone), hormones (oral pill, tibolone), cardiovascular drugs (the anticoagulants warfarin and phenprocoumon, the cardiac inotropic drug digoxin, the antilipidaemic drugs simvastatin, rosuvastatin and atorvastatin, the calcium blockers nifedipine and verapamil, the β1-adrenoreceptor blocker talinolol, the anti-anginal drug ivabradine), antiretroviral drugs (indinavir, nevirapine), anticancer drugs (irinotecan, imatinib), drugs acting on the CNS (anaesthetics, the anxyolityc drugs alprazolam, midazolam, quazepam and buspirone, the antidepressants sertraline, nefazodone, paroxetine, venlafaxine and amitriptyline, the anti-epileptic drugs mephenytoin, drugs for addicted patients, such as methadone and bupropion, the centrally acting muscle relaxant chlorzoxazone, the antitussive drug dextromethorphan), anti-ulcer medications (omeprazole), antidiarrhoeal drugs (loperamide), drugs acting on the respiratory system (theophylline, fexofenadine), antifungal drugs (voriconazole) and antimigraine medicines (eletriptan) [55,56,143,146,147,148,149,150,151,154,162,163,164,165,166,167,168,169,170,171,172,173,174,175,176,177,178,179,180,181,182,183,184,185,186,187,188,189,190,191,192,193,194,195,196,197,198,199,200,201,202,203,204,205,206,207,208,209,210,211,212,213,214,215,216,217,218,219,220,221] (see table 1 for further details).
Well-documented and clinically relevant interactions include: (1) reduced blood cyclosporine concentration associated in some cases to rejection episodes; (2) reduced efficacy of the oral pill, resulting in unwanted pregnancy; (3) reduced plasma concentration of antiretroviral (e.g. indinavir, nevirapine) and anticancer drugs (e.g. imatinib, irinotecan).
Valerian (Valeriana officinalis)
Valerian (Valeriana officinalis, Fam. Valeraniaceae) root preparations are widely available in a variety of commercial preparations as a sleep aid. Clinical evidence supports the notion that valerian is a safe herb associated with only rare adverse events [2,3,4]. Valerian has no impact on a number of CYP isoenzymes, including CYP3A4, CYP2D6, CYP2E1, CYP1A2 .
Valerian might theoretically potentiate the effect of CNS depressants.Hand tremor, dizziness, throbbing and muscular fatigue have been reported in a patient self-medicated with valerian and passion flower (Passiflora incarnata) while on lorazepam treatment. Also, a brief episode of acute delirium has been reported in a patient taking the antidiarrhoeal drug loperamide in combination with St. John’s wort and valerian .
Other Herbs Involved in Drug Interactions
Other herbal products that have been implicated in drug interactions include betel nut (Areca catechu, used for the preparation of a relaxing/refreshing beverage) , chlorella (Chlorella pyrenoidosa), a unicellular fresh water green alga used mainly as a potential source of food and energy and also believed to have some therapeutic benefits , boldo (Peumus boldus) used as a choleretic/cholagogue drug , fenugreek (Trigonella foenum-graecum), mostly used for the treatment of hypercholesterolaemia and diabetes mellitus , evening primrose oil (Oenothera biennis), mostly used in dermatology as well as for the treatment of rheumatoid arthritis , maitake (Grifolia frondosa), an edible mushroom with potential anticancer benefits , mistletoe (Viscum album) used as a palliative therapy for malignant tumors , prickly pear cactus (Opuntia polyacantha), traditionally used in Mexico for the treatment of diabetes , goji (Lycium barbarum), used in traditional Chinese medicine in cases of loss of energy, diabetes and liver disorders) [231,232], and hibiscus (Hybiscus sabdariffa), used in folk medicine for the treatment of hypertension [233,234]. Details of such interactions are reported in table 1.
Gums, mucilages, pectins or fibers contained in several medicinal plants have the ability to bind, trap and form viscous matrices with concurrently administered drugs. Hence, they may reduce their absorption. For example, a decrease in the absorption of lovastatin (associated to increased LDL levels) was observed in patients who took the statin concomitantly with pectin or oat bran . Clinical data have shown that plant products, such as gum guar (from Cyamopsis tetragonolobus), acacia gum (from Acacia senegal), or guggulipid (a standardized neutral fraction extract of gum guggul, an oleoresin obtained from Commiphora mukul) may reduce the absorption of drugs, such as metformin , amoxicillin , propranolol , and digoxin . A case of a decreased INR, suggestive of decreased anticoagulant effect, has been reported in a 57-year old man who began treating himself with an aqueous extract of the boiled roots of Commiphora molmol with his usual warfarin . C. molmol, one of the primary trees used in the production of myrrh, is traditionally used for the treatment of diabetes mellitus.
Finally, clinical studies have shown that hawthorn (Crategus oxyacantha), used for the treatment of congestive heart failure), had no effect on the pharmacokinetics of digoxin (P-glycoprotein substrate)  and Citrus aurantium subspecies amara (bitter orange peel) used for dyspeptic ailments, had no effect on various CYP isoforms, namely CYP3A4, CYP1A2, CYP2E1, and CYP2D6 .
Characteristics of the patient, such as age, frailty, infrequent genotypes, ethnicity, gender, and comorbidity  should be taken into account when considering herb-to-drug interactions.
It is well known that polymorphisms in the genes for drug-metabolizing enzymes or transporters may influence herb-to-drug interactions . For example, ginkgo can induce omeprazole hydroxylation in a CYP2C19 genotype-dependent manner (i.e. the effect has been shown to be more pronounced in poor metabolizers than in extensive metabolizers) . Wang et al.  also found that St. John’s wort increased CYP2C19 activity, as revealed by the increased urinary 4′-hydroxymephenytoin excretion in CYP2C19 wild-genotype subjects, but not in CYP2C19 poor metabolizers. Conversely, another clinical trial found that St. John’s wort increased the clearance of fexofenadine (P-glycoprotein substrate) and midazolam (CYP3A4 substrate) in six ethnic groups (i.e. Caucasian, African American, Hispanic, Chinese, Indian and Malawy) and there was no significant difference in the extent of induction between the ethnic groups .
The concomitant use of prescription medications and herbal products by older adults is a common situation in western countries . In addition, because older adults have multiple health problems, they are at particular risk for herb-to-drug interactions. Despite this, clinical studies aimed at investigating the potential of drug interaction in elderly patients are rare. Gurley et al.  found that elderly subjects, like their younger counterparts, are susceptible to herb-mediated changes in CYP activity and that some age-related changes in CYP responsivity to herbal products may exist. Specifically, it was found that ginseng slightly inhibited CYP2D6 in elderly subjects , in contrast to young subjects where no such inhibition was observed .
It is well established that the pharmacokinetics of many drugs may vary between men and women. Gender differences in herb-to-drug interactions have been reported both experimentally and in clinical trials. For example, a differential inductive profile of hepatic cytochrome P450s by the extracts of Sophora flavescens in male and female mice have recently been observed . More importantly, Gurley et al.  reported a significant sex-related difference in the inductive ability of St. John’s wort on CYP3A4 activity (i.e. St. John’s wort induced CYP3A4 more marked in male than in female subjects).
The use of herbal medicines is widespread. A survey found that approximately 15% of patients receiving conventional pharmacotherapy also take herbal products, and among these, potential adverse herb-to-drug interactions were observed in 40% of patients . It is therefore incumbent upon health care professionals to ask their patients about their use of herbal remedies. Patients erroneously believe that herbal products are natural and therefore safe. Probably for this reason, they are reluctant to disclose fully herbal use to their physicians. A recent study found that only 51.8% of women using complementary medicine, including herbal medicine, disclosed this use to their physician . It is therefore imperative that patients, especially those under cardiovascular, immunosuppressant or antiretroviral therapy, be informed of the possible adverse effects caused by interactions between herbal products and conventional medicines.
This review article has several limitations: interactions were searched by consulting PubMed and Embase and by checking the reference list of relevant review articles dealing with herb-to-drug interactions. Only clinical reports were considered. Preclinical studies, including human in vitro experiments, were not considered. Even though the search strategy was meticulous, the author cannot affirm that all relevant clinical data have been retrieved.
A good deal of the evidence on herb-to-drug interactions discussed in this article is based on case reports, which are sometimes incomplete and do not allow one to infer a causal relationship. It is worth noting that even documented case reports can never establish a causal relationship between drug administration and an adverse event; in addition for many interactions listed in table 1, the evidence is far from conclusive, as sometimes only one case report has been used and in many cases, a poorly documented case report may have been published. In this article, the level of evidence has been categorized using a 5-point scoring system. The highest level of clinical evidence (i.e. level of evidence: 5) has been considered when an adverse event described in a case report has been confirmed by a clinical pharmacokinetic trial. On the other hand, many adverse events are supported by poorly documented case reports (level of evidence 1, see table 1 for further details). When pharmacokinetic trials have not confirmed the adverse event hypothesized on the basis of the published case report(s) (e.g. interactions between warfarin and cranberry or ginkgo) or when contradictory pharmacokinetic data were published, the level of evidence was defined as ‘not applicable’. Although this scale has not been validated, it may be helpful as a guide for assessing whether an interaction is supported by adequate reliable clinical information.
In many instances, the extract type, standardization of extract, part of the plant used and the scientific (Latin) name of the plant have not been specified in clinical papers. This is an important omission because preparations obtained from the same plant may have different chemical compositions and hence different biological actions. Herbal preparations are not subject to the same regulations as prescription drugs and thus the content of the active ingredients may vary among manufacturers, potentially causing a large variation in efficacy and safety [247,248].
The often underregulated quality of herbal medicines is another safety issue. Contamination or adulteration of herbal medicines, including adulteration with synthetic drugs, may be relatively frequent and can cause drug interactions [2,3]. In other words, the possibility that a contaminant/adulterant and not an herbal ingredient causes drug interactions cannot be ruled out.
As highlighted above, people who use herbal medicines tend to conceal this use to their physicians or pharmacists. This observation, together with the fact that in many countries there are no central mechanisms for mandatory reporting as there is for conventional medicine complicate the identification of most herb-to-drug interactions.
Clinical reports clearly indicate that herbal medicines can interact with conventional drugs. While the majority of such interactions may have a negligible clinical significance, some may pose a serious threat to public health. For example, combining St. John’s wort with antiretroviral, immunosuppressive or anticancer agents that are metabolized by CYP enzymes and/or are substrates of P-glycoprotein may lead to drug failure. Serious health problems may occur when patients take herbal products before surgery. Cases of delayed emergence, cardiovascular collapse and loss of blood have been documented. A recent retrospective review of surgery patients presenting to the Anesthesia Preoperative Evaluation Clinic at the University of Kansas Hospital reported that approximately one-fourth of patients indicated the use of natural products prior the surgery . It is therefore incumbent on clinicians to screen patients before surgery for use of these supplements.
In conclusion, herbal medicines may be used by patients concomitantly receiving conventional drugs, which can result in potentially serious adverse events. It is incumbent upon healthcare professionals to be well informed about the growing clinical evidence of herb-to-drug interactions.