Suggestions and discussions have emerged during the past years regarding possible causative mechanisms leading to the development of kava hepatotoxicity (2–4, 6, 7, 10, 11, 13, 14, 16, 17, 23, 25, 32, 37–66), and other relevant data (67–82) have to be considered (Table 1). Proposals include areas such as ethnic origin predisposing to liver injury (3, 6, 17, 42, 66); solvents and solubilizers used for kava extract preparation (2, 4, 6, 7, 10, 11, 14, 16, 32, 37, 43, 45–47); quality of the kava raw material regarding cultivar and used plant part, adulteration and impurities (2, 4, 6, 14, 16, 23, 25, 44); hepatic glutathione depletion (4, 11, 16, 43, 45, 48, 49, 63); cyclooxygenase inhibition (13, 50); P-glycoprotein alterations (51, 58); genetic enzyme deficiencies (3, 4, 6, 11, 40, 42, 66); interactions at the level of hepatic microsomal cytochrome P450 (CYP, P450) between drugs, kavalactones and alcohol (2–4, 6, 7, 11, 14, 16, 17, 19, 42, 52–60); comedication (1–7, 13, 16–20, 32, 33); daily overdose of kavalactones and prolonged treatment (3, 6, 7, 17, 19); and the existence of one or more toxic constituents within or outside the kavalactones or of toxic metabolites (14, 16, 23, 42, 61–65). A shortcoming of the overall assessment is the fact that in only one patient (2, 33) out of the 31 cases with primarily suspected kava hepatotoxicity (1, 3, 6, 7, 44, 67) was a batch analysis of the used kava product performed regarding the presence of kava or other compounds as ingredients or contaminants. The pathogenesis of kava hepatotoxicity is complex and has not yet been sufficiently elucidated (6). Kava hepatotoxicity lacks adequate experimental reproducibility (3, 6, 14, 16, 47, 48, 57, 67), and animal models for assessing the pathogenesis are thus not generally available.
Solvents and solubilizers
Of major concern was the question whether the solvent used for kava extract preparation might possibly be considered as the culprit for the emerging toxic liver injury following kava use (2, 4, 6, 10, 11, 16, 32, 37, 45, 46, 49, 51). It is conceivable that the ethanol and acetone extraction procedure may either concentrate or select toxic compounds, or diminish protective ingredients (2, 11, 14, 46, 49). Preparation methods used for standardized kava extracts may also vary depending on the solvents used, and both efficacy and safety may depend on the kavalactones remaining in their natural forms and on the extraction of the other natural constituents of the plant (4). However, the best counterargument is the fact that kava hepatotoxicity occurred not only with ethanolic and acetonic kava extracts (6, 17) but also with traditional aqueous kava extracts (7, 16, 17, 40), suggesting that the solvent itself fails to contribute to the overall pathogenesis of kava hepatotoxicity. In particular, the WHO study listed five reports involving water extracts: three were coded for kava with a causality of possible and two probable; two were from traditionally prepared kava (16). Similarly, various solubilizers such as macrogol, craspovidon, mentha oil, methyl acryl acid polymer and polysorbate polyols have been used for the preparation of ethanolic and acetonic kava extracts (6, 17), but indications of their causative role for kava hepatotoxicity are lacking. It appears, therefore, that neither solvents (water, ethanol and acetone) nor various solubilizers used for the preparation of medicinal kava extracts would have caused toxic liver injury by kava use (2, 6, 7, 17). This raises the question as to what extent the quality of the kava raw material such as the cultivars may be a major contributing factor to the toxicity.
Kava cultivar quality
Some discussion emerged regarding the quality of the kava cultivars (2, 4, 6, 16, 17, 23, 25, 61, 69). The possibility was not ruled out that the hepatotoxicity problems were, to some extent, a consequence of poor-quality control caused by a rapid and extraordinary increase in the size of the kava market (4, 16, 23). Concern has also been expressed that substandard kava cultivars such as Tudey may have been exported (16, 23), and this was substantiated by analytical assessment of at least two but not of other retained samples from Germany (25). As early as 2002, the recommendation has been publicly communicated that the German regulatory agency BfArM may release relevant data on the source and quality assurance of the kava supply in due course (4), but no regulatory reaction concerning this important issue was recognized (70). The lack of standardization of kava products and selection of kava cultivars of the best chemotype and kavalactone content has also been recognized by the Pacific Community in 2001 (69). In particular, there was no established physical or chemical quality specification for kava exported for pharmaceutical products. Only in 2002, some legal recommendations have been established, and details thereof and comments were published in 2005 (71).
The Vanuatu government passed the Kava Act No. 7 of 2002 (71), which identifies and categorizes the different chemotypes or cultivars into the following: (i) Noble kavas, which have a long history of safe use as a traditional social beverage; (ii) medicinal varieties, which have a long and proven history of beneficial properties among traditional Pacific herbalists; (iii) ‘Tu Dei’ kavas (two day intoxication), which, in the absence of direct requests, are banned as an export commodity and (iv) ‘Wichmannii’ varieties (wild kava), which are also banned as an export commodity. Medicinal kavas are rarely used as a social beverage because they do not satisfy the kava drinker's desire for the required physiological effect (71). It is obvious, therefore, that kava cultivars used for traditional kava beverages in the South Pacific Islands were different from those used for medicinal purposes in Western countries. This raises the question as to whether noble cultivars may have been safer in terms of hepatotoxic effects than the medicinal varieties. As it presently stands, poor quality of some kava raw products is considered to be causally related to kava hepatotoxicity, at least in some cases.
Kava plant part quality
The quality of the best part of the kava plant to be used is a matter of debate (2, 6, 16, 17, 23, 69). In the past, reference is given to a variety of kava plant parts that have been used: rhizomes/rootstocks (6, 10, 32, 51, 61, 69), including fresh (16) and dried ones (16, 68, 72); rhizome roots (72); roots (11, 23, 33, 61, 64, 69, 72), including fresh (4, 10, 16, 37) and dried ones (4, 11, 16, 37, 40, 51, 72); decorticated roots (10); root barks, both fresh and dried (4); stems (23, 32), including lower ones (23, 37, 69) and peelings (23, 61, 69); stumps, including their peelings (14, 23, 32); and leaves (32, 61). Some South Pacific Island countries use fresh kava root or rhizome to prepare their traditional drink, while others use dried and ground roots or rhizomes (16). Kava extracts to be used in Germany as kava drugs should be prepared from dried rhizome chips of the kava plant according to the recommendations of the expert Commission E of the German regulatory health agency BfArM (73), and the rhizome should be peeled, as stated by the official German drug codex (74). The rhizome is clearly defined as the kava part below the stem and above the roots (16). In some European countries, however, kava preparations are often manufactured from the root peelings or kava stumps excluding the aerial peelings (2); all these used parts represent a cheap source of kavalactones. It is noteworthy that in the South Pacific, the kava roots are often peeled, the peelings are exported and the peeled roots are used to prepare the traditional aqueous extract for their own consumption (2). These observations lead to the conclusion that kava preparations made from whole peeled root, as used traditionally, could be less likely to cause hepatotoxicity (2); this favourable statement should also apply to the rhizome, the preferred regulatory plant part (70). According to the listing of the official Pacific Community in Suva, Fiji Islands, the commercial parts of the kava plant are chips of the rootstock, made from the peeled rhizome, rootstock or the first 20 cm of the stem to be used for drinking; peelings or skin of the rhizomes/rootstocks and first 20 cm of the stems, preferably used for export to pharmaceutical manufacturers and for drinking; and roots, without specification of use (69). Under these varying conditions with partially contradictory statements, basic differentiation between stem and rhizome/ roots should precede the manufacturing process utilizing solely peeled kava rhizomes and roots, at least for kava drug manufacturing in Germany. There is uncertainty as to what extent were younger rhizomes used for kava extracts (16). As a cheap, non-consumable and fast-growing variety, the kava cultivar Tudey has been harvested already after 1–2 years and has been flooding the kava market (16, 25). Normal kava plants are usually harvested after 4–5 years (16) and should be at least 3 years old (25, 69), but lack of standards have been criticized (16).
Readily available information suggests the use of aerial parts (6, 16, 23, 61, 69) such as stems (23, 61, 69, 72) and leaves (61) of the kava plant in the manufacturing process; these raw materials might have been taken instead of the usual rhizome. More specifically, according to the WHO report, German pharmaceutical industries preferred to buy kava stem peelings to extract kavalactones to prepare kava drugs; kava stem peelings were sold at almost one-tenth of the price of kava roots (16). It was also argued that commercial crude drug material that may have been adulterated by stem peelings and leaves could possibly introduce the alkaloid pipermethystine into the commercially used drugs (16, 23, 61), but recent analytical studies showed the absence of pipermethystine, at least in a series of retained samples of finished kava products from the German market (62). But uncertainty remains that the quality of commercial kava extracts may have varied from one batch to the other, and quality control of kava raw products was possibly not stringent enough. Uncertainty also exists regarding adventitious roots, originating from the stems and extending directly into the soil; they develop quite easily and are considered as valuable because of their high kavalactone content (69). Undoubtedly, adventitious roots are aerial parts of the kava plants, thereby not recommended for human use as kava drugs, kava dietary supplements or traditional kava drink.
Hepatic glutathione depletion
Other pathogenetic aspects include discussions regarding reduced hepatic content of glutathione (4, 11, 13, 43, 45, 49), but this mechanism has been debated (48). It has been argued that ethanolic and acetonic kava extracts lack glutathione, which is in contrast to traditional aqueous kava extracts (4, 47). However, even with the latter extracts, kava hepatotoxicity was observed (7, 17, 40), thus ruling out hepatic depletion of glutathione as the only pathogenetic mechanism of liver injury. It is noteworthy that in the absence of alcohol abuse and under normal nutritional conditions, the hepatic storage of glutathione should be adequate, because malnutrition has not been observed in patients after the use of ethanolic or acetonic kava extracts. Experimental studies have shown that feeding of aqueous kava extracts fails to decrease the hepatic content of glutathione (63); treatment with kava extracts was associated with a decrease in hepatocellular reduced glutathione content and an increase of oxidized glutathione, but the differences to the control values did not reach significance (45). In the latter study, however, methanolic and acetonic root extracts of a kava cultivar were used with extremely high amounts of methysticin, not necessarily comparable to kava extracts made in Germany or Switzerland (25). Respective clinical studies in patients with verified kava hepatotoxicity are not available, and therefore the role of glutathione remains speculative at present.
Cytochrome P450, comedication, kavalactones and alcohol
In only a few cases of kava hepatotoxicity, comedication was lacking (3, 6, 7, 19); this suggests some mechanism(s) independent of exogenous compounds in this particular group of patients. Comedication was, however, a common feature in most of the other patients; in these latter cases, pathogenetic mechanism(s) of kava hepatotoxicity may involve metabolic interactions between kavalactones and the other exogenous substrates (2, 6, 7) at the level of the hepatic microsomal cytochrome P-450 (13, 14, 43, 52, 55, 58, 60, 61). Its isoenzymes are not inhibited in vitro by kavain (43, 52) but by various other kavalactones (13, 14, 43, 52, 58); these in turn may reduce the microsomal metabolism of exogenous compounds, which are not only substrates but also inhibitors or inducers of P450 (13). The majority of patients with suspected kava hepatotoxicity have a comedication usually consisting of up to five different compounds (1–3, 6, 7); but one single patient had a comedication with 20 chemical drugs and dietary supplements including herbal ones (2, 7, 33). Among these comedications are synthetic drugs, compounds derived from herbal drugs and dietary supplements including herbal ones; the problem may exacerbate with mixtures, when numerous herbs are used that also contain kava and therefore kavalactones (2, 7, 38, 39). In a clinical situation where kava extracts are ingested concomitantly with drugs in a broad sense, there is uncertainty as to whether kavalactones may have initiated hepatotoxic reactions by drugs or vice versa (6, 17). Theoretically, the metabolism of comedicated drugs could be altered in such a way that even compounds primarily lacking evidence of hepatotoxicity may exert hepatotoxic effects. It is conceivable, therefore, that at least in some patients, kavalactone–drug interactions may be a contributory factor. Despite these exciting in vitro data (13, 14, 43, 56, 58, 60, 61) and additional experimental results showing gene expression changes of P450 isoenzymes (57, 59), there is surprisingly little clinical evidence of drug–kavalactones interactions (14, 47, 55, 56). Similarly, prolonged kava treatment in volunteers failed to inhibit various P450 isoenzymes, with the exception of P450 2E1, and caused no increase in serum liver enzymes (54). These findings have not yet been considered as the basis for kava hepatotoxicity in some patients.
The high rate of comedication in patients using kava extracts is unusual and generates concern (1–7, 13, 16–20, 32). Comedication was evident in 87% of patients with primarily suspected kava hepatotoxicity (6, 7, 17, 19) and in 64% of cases with verified kava hepatotoxicity (6, 7, 17). In large series of cases with drug-induced liver injury, comedication commonly accounted for only 13–15% of all patients, but may increase up to 27% (6). Comedication in patients under kava use consisted of up to 20 different synthetic drugs and dietary supplements including botanicals (2, 6, 7, 19, 33). In general, comedication substantially increases the risk of hepatotoxicity by a factor of six for two or more comedicated drugs (34). Comedication, therefore, seems to be a major pathogenetic and risk factor for hepatotoxicity, and using kava may be no exemption.
For the development of toxic liver injury following kava use, other pathogenetic factors such as a daily overdose of kavalactones, prolonged kava treatment or both have been discussed (3, 6, 7, 17, 19). Certainly, comparative doses of kavalactones from different extraction procedures such as aqueous ones vs ethanolic and acetonic ones are difficult to assess (11), and questions of possible daily overdose in patients with toxic liver injury because of the use of traditional aqueous kava extracts remain unanswered. Regarding ethanolic and acetonic kava extracts, in 80% of patients with primarily suspected hepatotoxicity, the basic kava treatment was not conforming to the regulatory recommendation of 60–120 mg kavalactones daily for no longer than 3 months (6). Interestingly, non-adherence to medication is not unique for kava but a common phenomenon also for treatment with synthetic drugs (35, 36).
Metabolic interactions of kava with alcohol have been suggested as a possible mechanism of kava hepatotoxicity in alcohol-consuming patients; reactive metabolites generated via CYP 2E1 during chronic alcohol intake or increased kavalactone exposure via hepatic enzyme inhibition during acute alcohol ingestion have been implicated (60). Alcohol is partially metabolized by the hepatic microsomal ethanol-oxidizing system involving cytochrome P450 (44, 75–77) with special reference to its isoenzyme CYP 2E1 (77, 78), and kavalactones may also be substrates of P450 (2, 4, 60). The clinical issue of alcohol consumption for the development of kava hepatotoxicity may be less important in Western countries (2, 3, 6, 7, 16), but a major role of alcohol could be implicated in kava-consuming populations with increased serum γGT activities (7, 16, 17) such as those in Australia (38, 39), New Caledonia (40) or Tonga (72). In a test group of kava-consuming and alcohol-abstinent volunteers, an inhibition of P450 2E1 was observed (54), findings of potential clinical interest.
Toxic constituents and metabolites
Possible candidates responsible for the hepatotoxic effects in patients with suspected kava hepatotoxicity are all 18 kavalactones (2, 4, 6, 17, 45, 53), other ingredients of the kava extracts and/or their metabolites (6, 14, 16). The other kava constituents include pipermethystine; flavokavins A, B and C; dihydrokavain-5-ol; cuproid acid; cinnamalketone; methylenedioxy-3, 4-cinnamalketone; 4-oxononanoic acid; benzoic acid; phenyl acetic acid; dihydrocinnamic acid; cinnamic acid; 1-(meta-methoxy cinnamoyl) pyrrolidine; and 1-cinnamoylpyrrolidine (16). Of particular interest are pyridon alkaloid pipermethystine (14, 16, 61–64) and flavokavins (2, 14, 65, 69).
Theoretically, after kava ingestion, the kavalactones leave the human body to some extent via urine and bile unchanged (79), but in time before toxic membrane alterations of liver cells due to their pronounced lipophilic properties may have occurred. However, there is little experimental evidence that kavalactones by themselves and in clinically relevant or even higher doses may be hepatotoxic; this is in line with clinical and experimental studies showing lack of hepatotoxicity after use of kava extracts or kavalactones as discussed in detail before (2–4, 6, 9, 10, 17, 48). Similarly, kava does not induce metabolic toxicity in vitro in the presence of five human P450 enzymes (53). In another recent experimental study, kavain was found to be non-toxic; however, other kavalactones such as methysticin and yangonin and/or one of their metabolites were assumed to be associated with hepatotoxicity (45). The shortcomings of the latter study include the use of methanolic rather than ethanolic extracts, not common to kava drugs, and the use of an unknown kava cultivar with an extremely high amount of methysticin in the acetonic root extract in line with a non-consumable kava cultivar (25). The next aspect pertains to the enzymatic degradation of kavalactones to their metabolites, which may exert hepatotoxic effects. Urinary samples in humans after ingestion of aqueous kava extracts showed the appearance of kavalactone metabolites, and further analysis of these products revealed various enzymatic transformations, which include reduction, demethylation and hydroxylation (79); this is in line with the proposed enzymatic breakdown of kavalactones via reduction and hydroxylation (60). Hepatic lactone hydrolases have been implicated in the degradation of kavalactones (4), raising the question of the involvement of hepatic microsomal cytochrome P450 in this enzymatic process (6, 60). Kavalactones such as kavain and methysticin combined with P450 enzymes yield desmethoxyyangonin and 5,6-dihydromethysticin respectively (2); this indicates some involvement of P450. Views have been expressed by others that kavalactones may indeed be substrates of P450 (40, 49, 60) or, more specifically, of its isoenzyme P450 2D6 (79), but firm evidence is still lacking. Therefore, the possible involvement of hepatic microsomal cytochrome P450 for the enzymatic breakdown of kavalactones requires further experimental assessment and clear evidence.
The WHO report argued that pipermethystine as a possible cause for kava hepatotoxicity could be present in kava extracts prepared with organic solvents but may not be bioavailable in aqueous kava extracts (16). Pipermethystine has been isolated from the peelings of the lower stem of one single cultivar named Isa and used as a pharmaceutical source; it was absent in 10 other cultivars tested (2). In detail, HPLC analysis showed a lack of pipermethystine both in a series of retained samples of finished products from the German market and in self-produced kava extracts derived from roots and stems (62). However, this does not rule out the possibility that a few other not assessed batches might have contained pipermethystine derived from aerial kava parts. Pipermethystine may be cytotoxic in vitro (61, 64), but other studies seem to report an absence of toxicity in therapeutic doses or even a hepatoprotective effect (14, 62). Thus, firm evidence indicating pipermethystine to be responsible for human kava hepatotoxicity has not been presented.
Flavokavain B has been shown to be the responsible cytotoxic compound derived from the kava root of the Isa cultivar (65), which does not belong to the favoured group of noble cultivars (25). It is unclear whether these in vitro results obtained with HepG2 cells (65) are principally transferable to patients with kava hepatotoxicity (14), keeping in mind that flavokavins may also be hepatoprotective (2, 47, 50), and hepatotoxic effects are lacking in experimental studies (3, 5, 6, 14, 16, 47, 48, 57).
Certainly, the action of other as yet unknown toxic compounds derived from various kava extracts cannot be excluded at present, and this statement applies also to possible toxic metabolites. There is general agreement that the chemical component(s) of kava products responsible for kava hepatotoxicity have not yet been identified with certainty (14, 16), requiring further studies regarding key components.