"......So far, the evidence supporting the use of herbals to treat chronic liver diseases is insufficient and herbs should not be recommended outside clinical trials....."
Felix Stickela, Eleonora Patsenkerb, Detlef Schuppanb
a Department of Medicine, Salem Medical Center, Heidelberg, Germany
b Division of Gastroenterology and Hepatology, Beth Israel Deaconess Medical Center, Harvard Medical School, 330 Brookline Avenue, Boston, MA 02215, USA
Journal of Hepatology
Substantial progress has been made in understanding the interaction between herbal drugs and the liver. However, our knowledge of the potentials and risks of botanical drugs is still limited and efforts to elucidate them should be intensified.
Therapies developed along the principles of conventional (Western) medicine are often limited in efficacy and unaffordable for many individuals throughout the developing world. Therefore, treating diseases with plant-derived compounds which do not require extensive preclinical testing and laborious pharmaceutical synthesis is attractive. Furthermore, the interest in phytomedicine by the professional and lay public is increasing steadily. Several recent surveys from Europe and the US have demonstrated a sharp rise in the use of botanical drugs within a few years [1-3]. Thus, 21% of patients attending an out-patient liver clinic had taken herbal preparations, and 13% used herbs to treat liver diseases. These figures may reach 30-65% in distinct cohorts, since there exist large regional differences [3,4]. Another survey assessed the use of complementary medicine in the US and revealed a 380% rise in the use of herbals between 1990 and 1997, and estimated out-of-pocket expenditures for these medicines increased from $1.8 billion in 1990 to $5.1 billion in 1997 . Similar observations can be made for Europe where the expenses for silymarin, a herbal preparation used to treat chronic liver diseases, amounts to $180 million alone in Germany .
The increasing popularity of herbal medicines is based on their perceived effectiveness in the treatment and prevention of disease, the belief that these treatments are 'natural' and therefore safe, a feeling of better control of the disease and its management, and a holistic philosophy behind complementary medicine that pays tribute to the patients' desire for wellness and quality of life in nature's womb, which apparently is not provided by conventional health care. Dissatisfaction with conventional medicine due to lack of treatment success in many instances, unfavorable side effects, lack of time with the doctor and insufficient empathy from health care providers also contribute to herbal medicine's appeal. In addition, most countries do not impose prescription regulations upon herbal preparations and, therefore, access to this kind of therapy is unrestricted and cheap.
In the US, herbal products are labeled dietary supplements which are not expected to meet the standards for drugs specified in the Federal Food, Drug, and Cosmetic Act. The only requirement is that these preparations follow the standards set forth in the Dietary Supplement and Health Education Act issued in 1994 which allows marketing without prior approval of their efficacy and safety by the Food and Drug Administration (FDA). DSHEA allows manufacturers to claim that the sold product affects the body or its function, as long as there is no allegation that it prevents or treats certain diseases. Therefore, this simplified licensing practice does not guarantee efficacy and safety in the same strict way as with the approval of conventional medications and treatments. Herbal medicines are exempt from rigorous regulations in the United States leading to considerable variation in the composition of herbal remedies that leaves the door open for inferior products.
A somewhat different situation is found in many European countries and Canada which have implemented strategies for licensing herbal drugs. In Germany, for example, the Commission E as part of the Federal Institute for Drugs and Medicinal Devices (Bundestinstitut für Arzneimittel und Medizinprodukte) issues approval for herbals on the basis of information contained in more than 300 monographs on herbs with detailed information on terminology, composition, side effects, contraindications, interactions and mode of administration [7,8]. The Commission E also records voluntary reports on side effects which occurred along with the intake of herbal preparations. Similarly, the European Commission has recently drafted a directive on the licensing of traditional herbal preparations which allows premarketing assessment of the quality and safety of a drug allowing pharmacovigilance and product recalls [9,10]. Appropriate regulation guidelines and pharmacovigilance pay tribute to the undisputed fact that herbs may contain active ingredients that either treat diseases or cause unwanted adverse effects
The present review summarizes the current evidence on the potential hepatotoxicity from herbal drugs, and outlines the problems that need to be addressed to correctly assess their risk-benefit ratio.
2. Herbal hepatotoxicitygeneral considerations
Adverse hepatic reactions from xenobiotics are well documented and drug-related adverse hepatic events are the most frequent cause of post-marketing warnings and withdrawal. The problems in predicting and preventing drug-related idiosyncratic liver damage have been adressed extensively in recent review articles [11,12]. The reason for the release of hepatotoxic drugs relates to the fact that the frequency of liver abnormalities precipitated by drugs is usually low even for well-known hepatotoxins such as halothane and isoniazid (approximately 1/1000). Pre-marketing testing in healthy volunteers and post-marketing cohort studies usually fail to detect infrequent (1:10,000 or less) but potentially serious drug-induced hepatotoxicity since the cohort sample sizes are too low to let these adverse liver reactions occur . So, hepatotoxicity of certain drugs becomes evident only after a large number of patients have been exposed to them. Even then, hepatotoxicity often remains unnoticed since the causal relationship is not immediately recognized by users and prescribers. A recent example for such drug-related hepatotoxicity is that of troglitazone which was released to treat diabetes type II and even proved useful in lowering liver enzyme levels in patients with non-alcoholic steatohepatitis . Eventually, troglitazone was withdrawn from the market after numerous reports unequivocally proved its potential to precipitate fulminant hepatitis . Similar problems are likely to occur with herbal drugs even in case prescription rules are implemented.
The liver is the central drug-metabolizing organ and is, therefore, a prime target of drug-related pathologies. Foreign compounds are predominantly biotransformed in the liver by the action of drug-metabolizing enzymes including microsomal cytochrome P450 enzymes, mixed-function mono-oxygenases, glutathione-S-transferases, sulfotransferases and UDP-glucuronosyltrans-ferases. Some of these can be induced through variable mechanisms which may lead to large interindividual variability in susceptibility for drug-related liver damage. Hepatic damage from conventional drugs is widely acknowledged and most physicians are well aware of them. Herbals as a cause of adverse hepatic reactions, however, have only recently been recognized as their use has become more widespread. Certain herbals have been identified as a cause of acute and chronic hepatitis, cholestasis, drug-induced autoimmunity, vascular leasions and even hepatic failure and cirrhosis (Table 1). Risk factors for herbal toxicity have not been well identified, largely since hepatotoxic incidents have mostly been published as isolated case reports or small series. However, a certain risk pattern has become evident, such as the observation that most affected individuals were females. This gender difference does not reflect a higher likelihood of women to use these preparations, but their higher susceptibility towards herb-induced liver damage , as is observed for the majority of adverse hepatic reactions induced by conventional drugs. As with chemically defined drugs, adverse hepatic reactions towards herbals cannot be predicted through diagnostic means, which makes the early recognition of liver damage important. Most individuals who take herbals do not admit their intake, even on repeated questioning, either because they do not consider herbals as 'drugs', or because they fear not to be taken seriously by their doctors for using herbals. Furthermore, doctors who recommend herbals or patients who take them advocate the long-standing use of herbals in traditional medicine as proof of safety, in particular, since most herbals are available without prescriptions and at low costs. Therefore, self-medication is frequent and, sometimes, patients even increase the dosage as liver disease worsens.
Another problem is that herbals are usually mixtures of several ingredients or plants harvested during different seasons and extracted through variable procedures, which makes the identification of both the pharmacologically active and toxic compounds difficult. Also, contamination of herbals with microorganisms, fungal toxins such as aflatoxin, with pesticides, heavy metals, and synthetic drugs has been described .
Interactions between herbs and chemical drugs are another source of problems associated with the intake of herbal compounds. For example, St John's wort (Hypericum perforatum) which is promoted for the treatment of depression interacts with numerous conventional drugs including cyclosporin, simvastatin, warfarin, and digoxin due to its inducing properties on cytochrome P450 3A4, 2C9, and the drug transporter MDR1 . A number of other herbals such as Salvia miltiorrhiza, Angelica sinensis, and Papaya (papain) can impair coagulation with an increase of the INR  which poses a danger in cirrhotics with impaired liver function. Only rigorous pharmacological testing of herbals can prove their safety and efficacy, since they may contain highly active molecules that can incur significant side-effects.
3. Chinese herbs
Due to the rising popularity of oriental therapies, the use of Chinese herbal medicine has markedly increased. In China, the use of herbals dates back as far as 2100 BC with comprehensive records of Chinese medical theories dating back to 221 BC . Most Chinese medicines are blends of 4-5 different herbs of which 1 or 2 are the pharmacologically active 'King herbs'. The remaining constituents function as modifiers of toxicity, act synergistically with the King herb, improve the immune function or strengthen certain aspects of physical well being, which makes the identification of causative compounds extremely difficult.
Jin Bu Huan (Lypocodium serratum) which had been widely used in the US for mild sedation caused both acute and chronic hepatitis [18,19]. Liver histology in one patient revealed focal necrosis and moderate portal fibrosis. Evidence of hepatotoxicity is convincing, since other common etiologies were excluded and elevated liver enzymes normalized after withdrawal of the herbs. Nevertheless, final prove of causality is lacking, since no reexposure could be performed for ethical reasons.
Ma-Huang (Ephedra sp.) was marketed in the US to support weight loss therapies. A woman developed acute hepatitis together with elevated antinuclear antibodies (ANA) and smooth muscle antibodies (SMA) after only 3 weeks of intake of Ma-Huang, but liver abnormalities resolved after its discontinuation . Another report suggested that Ma-Huang hepatotoxicity is associated with the presence of compound heterozygosity for the hemochromatosis gene mutation, possibly through enhancing oxidative stress .
When studies that investigated the efficacy of Chinese herbal preparations for treatment of atopic dermatitis were performed, several cases of acute hepatitis and even lethal fulminant hepatic failure in one case were reported [22-26]. The investigators failed to identify the causative agent leading to liver failure but, as shown in Table 2, all herbal combinations that led to liver abnormalities contained Paeonia. If and how far Paeonia is the culprit needs urgent clearification, since it is considered the 'King herb' for treatment of atopic dermatitis .
4. Pyrrolizidine alkaloids
A dose-dependent hepatotoxicity of plants containing pyrrolizidine alkaloids is well established. 'Senecio disease' was first described in South Africa , followed by reports from Jamaica about children developing ascites, hepatomegaly and eventually cirrhosis after the ingestion of 'bush tea'. The Crotalaria species in tea leaves were identified as the causative herbs . Later, a series of pyrrolizidine alkaloid poisonings occurred in India  and Afghanistan , where 'Gondli' cereal contaminated with Heliotropium alkaloids was responsible for numerous cases of veno-occlusive disease (VOD), the typical liver injury evoked by pyrrolizidine alkaloids. Incidents of liver damage from pyrrolizidine alkaloid intoxication were noticed in Arizona, such as in two infants exposed to herbal tea containing Senecio longilobus[32,33]. Similar cases were also reported from Europe and elsewhere [34,35]. VOD is a non-thrombotic obliteration of the lumen of terminal centrilobular hepatic veins resembling Budd-Chiari syndrome (Fig. 1). The resulting venous outflow obstruction causes hepatic congestion and centrilobular necrosis which either results in acute liver failure or liver fibrosis and cirrhosis .
Pyrrolizidine alkaloids extracted from Heliotropium, Senecio, Symphytum (Comfrey) and Crotalaria species are particularly hepatotoxic. Their mechanism of action involves a direct toxic rather than an immunological mechanism. Acute toxicity is reproducible in animals and seems to be related to biotransformation of pyrrolizidine alkaloids by cytochrome P450s into pyrrole derivatives which then act as alkylating agents . In animal studies unbound pyrroles are highly reactive hepatocarcinogens [38,39]. Interestingly, toxicity of pyrrolizidines can be increased by phenobarbital, a potent inducer of cytochrome P450 3A4, 2B6, and several isoenzymes of the 2C family (Fig. 2) . A standard treatment does not exist except for avoiding the intake. As soon as chronic or acute liver failure is imminent, orthotopic liver transplantation may be the only effective therapy. Since hepatotoxicity of pyrrolizidines is well documented, it is difficult to understand why their use is not regulated. The sale of comfrey is banned in Germany and Canada, but it is still freely available in the USA .
Germander (Teucrium chamaedrys), a member of the Labiatae family is known as a herbal remedy for its assumed choleretic and antiseptic properties for more than 2000 years. It was considered as absolutely safe, although its active ingredients remained unknown. In 1986, Germander was approved as a drug for the supportive treatment of obesity and mild diarrhea in France . Germander preparations were available as capsules and tea bags, but also as an addition to liquors. Its large-scale use precipitated several reports to the French Pharmacovigilance Authorities in 1992 about Germander-associated acute, chronic and even fulminant hepatitis [43,44]. Most affected individuals were women attempting to loose weight. The daily dosage ranged between 600 and 1600mg/day and hepatitis usually developed after 2 months of intake . Histologically, acute cytolytic hepatitis was found and some patients with a more benign course of liver disease revealed patterns of chronic hepatitis with fibrosis and even cirrhosis [43,46]. However, all patients recovered after the discontinuation of treatment except for those with cirrhosis, but relapsed under accidental reexposure [46,47]. Consequently, Germander has been analyzed thoroughly and found to contain saponins, glycosides, flavonoids and several furane-containing neo-clerodane diterpenoids . Animal experiments in mice demonstrated the formation of toxic metabolites from these diterpenoids by cytochrome P450 3A. Their formation is enhanced by induction of cytochrome P450 3A and by glutathione depletion (Fig. 3) [49,50], and the toxified diterpenoids are potent inducers of hepatocyte apoptosis [51,52].
Lately, a case of fulminant hepatic failure after the ingestion of Teucrium polium has been described . This plant is of the same genus as Germander and has been used as an antiinflammatory and antimicrobial drug as well as for the treatment of scars.
6. Greater Celandine
Drugs containing Greater Celandine (Chelidonium majus) are frequently used in Europe for alleged choleretic and antispasmodic properties, particularly in biliary disorders and irritable bowel syndrome. Greater Celandine contains at least 20 different alkaloids including berberine, coptisine, chelerythrine, and chelidonine, of which the latter serves to standardize the extract. Most commercially available drugs are produced from the dried aerial parts harvested during blossoming since cell culture studies suggested increased antiviral, antitumour, antibacterial, and anti-inflammatory properties of these preparations. The therapeutic efficacy of Greater Celandine has never been tested in controlled trials . Several reports, mostly from Germany, where commercial drug preparations containing Greater Celandine alkaloids are widely available, demonstrated the potential hepatotoxicity of this herbal [55-58]. Nine out of 12 patients from a single center [57,58] revealed cholestatic hepatitis together with low titres of autoantibodies suggesting drug-induced autoimmunity after variable periods of ingestions. However, the exact mechanism remains unclear. Importantly, patients had taken Greater Celandine preparations from different manufacturers which strongly suggests that indeed the plant extract has been the cause of hepatotoxicity. Greater Celandine preparations are still on the market despite the clear evidence of potential hepatotoxicity.
Kava root (Piper methysticum rhizoma) has been used as a traditional psychotropic remedy in Hawaii, Polynesia and the Fidji Islands. In industrialized countries, Kava-containing preparations are marketed for the treatment of anxiety disorders and depression. Its sedative activity is caused by kavapyrones, including kavain, dihydrokavain, methysticin, dihydromethysticin, which are gamma-amino-butyric acid receptor agonists that inhibit activating neurons in the reticular formation and the limbic system [59,60]. The therapeutic potential was carefully addressed in a recent systematic review and in a meta-analysis which included the randomized, controlled trials with Kava for the treatment of anxiety. The results showed a significant anxiolytic effect as assessed by the Hamilton Rating Scale for Anxiety together with good tolerability . However, seven case reports describing hepatitis related to Kava ingestion raised doubts over the safety of Kava products, which until recently were still available without prescription in Germany and elsewhere [62-68].
The published seven cases and further 29 reported but unpublished cases of adverse hepatic reactions due to Kava were analyzed systematically using a clinical diagnostic scale developed and evaluated for the confirmation of drug-related liver damage in routine settings [69,70]. Hepatic necrosis or cholestatic hepatitis were noticed with both alcoholic and acetonic Kava extracts. Twenty-seven of the 36 patients were women. Both the cumulative dose and the latency to when the hepatotoxic reaction emerged were highly variable. Nine patients developed fulminant liver failure, of which eight patients underwent liver transplantation. Three patients died, two following unsuccessful liver transplantation and one without. In all other patients, complete recovery was noticed after the withdrawal of Kava. Pathophysiologically, both immunoallergic and idiosyncratic factors may be responsible. The long latency period and the absence of features of immune mechanisms in many patients combined with the fact that the doses ingested exceeded several times the recommended dose of 120mg/day suggested individual metabolic idiosyncrasy. Along this line, Russmann and collegues have recently found a poor-metabolizer phenotype of cytochrome P450 2D6 in two individuals who experienced Kava-induced hepatitis indicating that this genetic predisposition may be a risk factor for developing Kava-related liver damage .
Chaparral (Larrea tridentata)commonly referred to as 'creosote bush' or 'greasewood'grows in deserts and serves as a herbal remedy among Native Americans in the Southwestern United States and Mexico. After grinding the leaves, Chaparral is ingested as tea. A variety of ailments such as common cold, bone and muscle pain, and snake bites are treated mainly on the basis of anecdotal reports . Commercial plant extracts are available as tablets, capsules and ointments, and sold for their alleged antiinflammatory and 'blood purifying' potentials, as well as a liver tonic and as a treatment for skin disorders. In addition, presumed weight-reducing and antioxidant properties have led to the perception that Chaparral may retard the process of aging. Moreover, Chaparral is applied in alternative medicine regimens in the treatment of AIDS . Since 1990, incidents of Chaparral-related hepatotoxicity were reported to the FDA in the USA, and all cases were carefully reviewed generating a clearer picture of its hepatotoxic potential. Thus, 18 patients with Chaparral-associated toxicity were reported of which 13 revealed liver damage ranging from mild hepatitis to cirrhosis and fulminant liver failure . The predominant pattern of liver damage was cholestatic hepatitis with high serum transaminases and elevation of bilirubin and alkaline phosphatase. A minority of patients presented with cirrhosis and two patients required transplantation for fulminant hepatic failure. Some individuals were administered single-component Chaparral, while others took multi-ingredient combinations with various other herbs. In all preparations, Larrea tridentata was found, while biochemical and microbial contamination could be excluded. A causal relationship was postulated because of a temporal correlation between intake of Chaparral and the onset of liver disease, a consistent pattern of hepatic damage, and through the observation that reexposure to Chaparral or an increased dose led to relapse or aggravation of clinical signs of liver disease. Chaparral toxicity is supposed to be due to nordihydroguaiaretic acid which inhibits cycloogygenase and cytochrome P-450 [74,75].
9. Atractylis gummifera and Callilepsis laureola
Atractylis gummifera is a known cause of toxic hepatitis in the Mediterranean. It is used as an antipyretic, emetic and diuretic, and a bright fluid secreted from the plant is enjoyed by children as chewing-gum . The onset of hepatitis is usually acute and commences few hours after ingestion and following unspecific symptoms such as nausea, abdominal pain and headache. It is associated with a syndrome of neurovegetative symptoms, hepatorenal failure and pronounced hypoglycemia; the latter being caused by the inhibition of gluconeogenesis. Death due to fulminant hepatic failure is frequent. Consumption of A. gummifera is particularly dangerous during spring time when toxins are concentrated in roots or when the plant is confused with wild artichoc . Toxicity has been ascribed to atractylosides and gummiferin which are inhibitors of the Krebs cycle and other mitochondrial functions and exert selective toxicity to hepatocytes and kidney epithelia in vitro via induction of oxidative stress, as mirrored by glutathione depletion and increased lipid peroxidation .
Another plant species containing atractylosides is Callilepsis laureola which has led to several cases of fulminant hepatitis and renal tubular necrosis in South Africa .
Various other botanicals have been associated with toxic liver damage such as Senna (Cassia angustifolia), which was identified as a cause of a relatively benign hepatitis in a young woman taking approximately 10 times the recommended dose. The causal relationship between the preparation and hepatitis was confirmed through a positive rechallenge . Senna is biotransformed via intestinal bacteria to rhein anthrone which shows structural similarities to the laxative anthrone. Likewise, anthraquinones, including rhein anthrone, contained in rhubarb have been suspected to cause liver injury .
Pennyroyal, also referred to as 'squawmint oil', is a herb containing leaves from either Mentha pulegium or Hedeoma pulegoides and continues to be a source of intoxication since its use is widespread. It has long been applied as abortifacient or pesticide against fleas. There are several reports on fulminant hepatic necrosis, also with lethal outcome, due to Pennyroyal. Its primary constituents are pulegone and various other monoterpenes characteristically encountered in mint species . However, pulegone is found in greater concentrations in Pennyroyal than in other mints. Hepatotoxicity appears to be caused by both pulegone producing oxidative stress and via pulegone's primary metabolite menthofuran, the latter being transformed to a hepatotoxin through cytochrome P450 [82-84]. Direct depletion of glutathione by the formation of electrophilic metabolites seems to be the crucial step in pulegone toxicity.
The hepatotoxic effects of Margosa oil (Azadirachza indica) resemble Reye's syndrome which is histologically characterized by microvesicular steatosis, mitochondrial toxicity, and depletion of hepatic ATP and glycogen stores [76,85].
Tea made of Coltsfoot and Sassafras albidum (Sassafras), the latter containing safrole, which is known to be hepatocarcinogenic in rodents, may be another source of chronic liver injury rather than a cause of acute intoxication .
Ayurvedic herbal preparations from India have been studied for the treatment of chronic liver diseases either experimentally or clinically, including a herbal combination termed Liv.52. Liv.52 contains Capparis spinosa (capers), Cichorium intibus (wild chicory), Terminalia arjuna (arjuna), Solanum nigrum (black nightshade), Achillea millefolium (yarrow), and others. Liv.52 was reported to protect against experimental toxic liver damage , and suggested to be useful for human alcohol-related liver cirrhosis by lowering acetaldehyde, the highly toxic first metabolite of alcohol degradation . This initiated a 2-year European randomized controlled clinical trial in 188 patients with alcoholic liver cirrhosis . While no effect of Liv.52 on survival was noticed in cirrhotics with Child class A and B, increased liver-related mortality among those with child class C was found leading to premature termination of the study. This emphasizes the potential hazards of poorly defined and incompletely tested herbal medications.
The hepatotoxicity of mistletoe (Viscum album) is debated. A report from 1981 suggested potential adverse hepatic reactions but the composition of the suspected preparation was unclear. In fact, retrospective analysis showed that the suspected medication may not have even contained Mistletoe .
Another combination of herbal ingredients, known as 'Prostata', was suspected to have caused cholestatic hepatitis in a man using this medication for the treatment of benign prostatic hyperplasia . The assumed active ingredient is Serrenoa serrulata which exerts estrogenic and antiandrogenic effects. Either hormones may be liver toxic under certain circumstances.
Recently, the juice of Noni (Morinda citrifolia), a Polynesian tropical fruit which is highly publicized in the European tabloid press for its health promoting effects, was associated with severe hepatitis in a 45-year-old man .
The incidence of hepatic damage from botanicals is difficult to estimate, and the uncertainties to confirm causality in drug-induced liver injury are well known. Often, the diagnosis can only be made based on clinical evidence. Therefore, numerical scores that take certain clinical circumstances and findings into consideration were suggested for clinical routine use. The Council for International Organizations of Medical Sciences (CIOMS) developed a scale that aims to generate valid and reproducible cause-effect assessments of drug-induced adverse hepatic reactions . Maria and Victorino (MV) have proposed a simpler method to serve the needs in clinical practice . Both scoring systems (Tables 3 and 4) were recently compared and validated, and it was shown that the CIOMS score reached a better discriminative power and resulted in more exact causality assessments when compared with the MV scale .
12. Herbs to treat liver diseases
The popularity of herbal remedies is increasing and many patients with liver disease use them. Controlled treatment trials in diseases for which plant components claim to be effective are lacking, and most of the few studies so far published do not report the incidence of adverse effects. Nevertheless, a number of herbals show promising activity in the treatment of acute and chronic liver disease including silymarin as a potential antifibrotic, Phyllantus amarus as an antiviral in chronic hepatitis B, Glycyrrhizin as a hepatoprotectant in chronic viral hepatitis, and a number of herbal combinations from China and Japan that deserve testing in appropriate studies [96,97]. Therefore, there is the necessity of premarketing drug-testing and pharmacovigilance for all herbals just as with any other drug. So far, the evidence supporting the use of herbals to treat chronic liver diseases is insufficient and herbs should not be recommended outside clinical trials.