Journal of Clinical Pharmacy and Therapeutics (2010) 35, 545–563 doi:10.1111/j.1365-2710.2009.01131.x ORIGINAL ARTICLE Kava hepatotoxicity: comparative study of two structured quantitative methods for causality assessment R. Teschke MD,J.FuchsMD,R.BahreMD,A.GenthnerMD and A. Wolff MD PhD Department of Internal Medicine II, Division of Gastroenterology and Hepatology, Klinikum Hanau, Teaching Hospital of the Johann Wolfgang Goethe-University Frankfurt ⁄ Main, Hanau, Germany Keywords: causality assessment methods, herbal SUMMARY hepatotoxicity, kava, kava hepatotoxicity Background and objective: Ingestion of the medicinal herb kava has been associated with hepatotoxicity. We aimed to compare two differ- INTRODUCTION ent quantitative methods of causality assessment of patients with assumed hepatotoxicity by the Chemical drugs, herbal remedies and dietary sup- herb. plements are commonly considered as safe and Methods: We assessed causality in 26 patients devoid of major side effects, but in usually rare from Germany and Switzerland, using two struc- instances hepatotoxic reactions may emerge in a few tured quantitative analytical methods: the system susceptible individuals despite adherence to the of Maria and Victorino (MV) and that of the recommended daily dose and treatment duration Council for International Organizations of Medi- (1–4). Under these conditions, the diagnosis of toxic cal Sciences (CIOMS). In all 26 patients, regulatory liver disease represents a major clinical challenge ad hoc evaluation had suggested a causal and may only be established when other causes of relationship between liver disease and kava use. the observed liver disease have been excluded with Results and discussion: Assessment with the MV certainty. Indeed, there is no biomarker or surrogate scale resulted in no or low graded causality for marker commonly available which might facilitate kava in the 26 patients with liver disease. Causality the diagnosis of toxic liver disease (4). By clinical was probable (n = 1), possible (n = 2), unlikely criteria a differentiation between toxic and genuine (n = 7), and excluded (n = 16). Causality for kava liver disease is commonly not possible, because was more evident with the CIOMS scale: highly symptoms and clinical findings are similar in both probable (n = 1), probable (n = 2), possible (n = 6), conditions, including icterus, fatigue, right upper unlikely (n = 2) and excluded (n = 15). However, abdominal discomfort, dark urine, pale stool, pru- the results of both quantitative causality assess- ritus and weight loss (3, 4). ments are not supportive for most of the regulatory In the last few years, a world-wide discussion ad hoc causality assessments of the 26 patients. emerged on whether and, if so, to what extent, Conclusion: Grades of causality for suspected treatment with the anxiolytic herb kava is hepato- hepatotoxicity by kava were much lower when toxic (5–15). Kava has been used as ethanolic and evaluated by structured quantitative causality acetonic extracts of the rhizome of the pepper plant assessment scales than by regulatory ad hoc Piper methysticum G. Forster (5, 9). The initial cau- judgements. The quantitative CIOMS scale is the sality assessment was a regulatory one achieved on preferable tool for causality assessment of spon- an ad hoc basis. An association was assumed in 26 taneous reports of hepatotoxcity involving kava. patients from Germany and Switzerland (16). Based on identical data, the regulatory judgement Received 21 July 2009, Accepted 31 August 2009 of causality was not substantiated by another reg- Correspondence: R. Teschke, MD, Department of Internal ulatory agency (17) and health institute (18). It was Medicine II, Klinikum Hanau, Teaching Hospital of the Johann Wolfgang Goethe University of Frankfurt ⁄ Main, Leimenstraße criticized by various scientific groups (7–14), 20, D-63450 Hanau, Germany. Tel.: +49 6181 ⁄ 2964200; fax: especially when additional data were evaluated +49 6181 ⁄ 2964211; e-mail: [email protected] (5, 7, 12–14). Moreover, quantitative causality Ó 2010 The Authors. JCPT Ó 2010 Blackwell Publishing Ltd 545 546 R. Teschke et al. assessments yielded conflicting results in two CIOMS (20). The scale was applied to kava and reports (14, 15). Supportive data in favour of the each co-medicated chemical drug, herbal remedy regulatory assessment were obtained in one study and dietary supplement. If there was more than (15) using the quantitative scale of MV, acronym one co-medication, only the co-medicated com- for the authors Maria and Victorino (MV) (19), but pound with the highest total score was considered not in another assessment (14) using the quantita- in the final assessment. tive scale of Council for International Organiza- The scores of the MV system relate to temporal tions of Medical Sciences (CIOMS), the acronym for relationships regarding onset of clinical or labora- the Council for International Organizations of tory manifestation, and normalization of laboratory Medical Sciences (20). values after drug discontinuation, exclusion of In the present study, we have evaluated causal- alternative causes, extrahepatic immuno-allergic ity for all 26 patients with liver disease, assumed in manifestations, results of re-exposure and previous a regulatory assessment to be causally associated published reports of cases of drug-induced liver with kava treatment. We retrieved additional data injury with the alleged drug (19). It provides for and used both the quantitative MV method and the each of these items a range of scores. The total score quantitative CIOMS scale. We found that, in the is then computed and used to assign causality as setting of regulatory evaluation of the observed being highly probable, probable, possible, unlikely liver disease in assumed relationship with kava, or excluded. the MV scale performed poorly compared with the CIOMS scale. With a few exemptions, both scales CIOMS scale failed to support the regulatory ad hoc causality assessments for kava. It appears that the CIOMS The quantitative CIOMS scale (20) was applied to scale is the preferred tool for causality assessment all 26 patients (14, 16). Causality assessment for in a setting of assumed kava hepatotoxicity. kava and co-medication was as described for the MV evaluation. The quantitative CIOMS uses items such as time to onset, course of improvement of PATIENTS AND METHODS laboratory data, risk factors, concomitant drugs, Patients search for non-drug causes, previous information on hepatotoxicity of the drug and response to Twenty-six patients originating from Germany re-administration (20). It provides with each of (n = 20) and Switzerland (n = 6) were studied. In these parameters a range of scores. The total score all patients, liver disease was causally attributed to is then computed and this is used for causality treatment with kava extracts, and the regulatory assessment as described above for the MV scale. assessment of causality for kava gave scores rang- ing from certain (n = 3) and probable (n = 15) to possible (n = 8) (16). For each of the patients, Comparative assessment essential data were collected from various sources The results presently obtained with the MV scale including the German and Swiss regulatory agen- were compared with the CIOMS scores for kava cies, drug manufacturers, primary-care physicians, and co-medications for all 26 patients (14). treating hospital physicians and pharmacists as described previously (14). Additional data were obtained from published reports quoted by the RESULTS regulators (16). General characteristics In the mostly female patients, daily overdose of MV scale kavapyrones (kavalactones), prolonged treatment All 26 patients underwent the structured quanti- and co-medication were common features tative causality assessment reported by MV (19). (Table 1), considering that the regulatory recom- The MV scale is a short and modified version of the mendation includes 60–120 mg kavalactones daily structured quantitative causality assessment of as maximum intake for no longer than 3 months. Ó 2010 The Authors. JCPT Ó 2010 Blackwell Publishing Ltd, Journal of Clinical Pharmacy and Therapeutics, 35, 545–563 Ó 2010 The Authors. JCPT Table 1. Clinical data of all patients (n = 26) with regulatory suspected liver disease in assumed association with the treatment by kava extracts (f = favourable, d = died) Ó 2010 Blackwell Publishing Ltd, Additional references Patient Identification Age (years) Sex Kava extract (months) Duration of therapy Kavapyrones Co-medication Outcome Regulatory information Additional information Regulatory (mg/day) ad hoc causality assessment Kava Co-medication Journal of Clinical Pharmacy and Therapeutics 01 BfArM 38 f Acetonic 3Æ5 210 Oral f Exclusion of Daily kava overdose ++ ) (7, 12, 13) 93015209 contraceptive non-kava Prolonged kava treatment Diazepam causes not Exclusion of hepatitis A, B, C, L-Thyroxine reported CMV, EBV, but not of HSV Exclusion of biliary obstruction and alcoholism Course of ALT not sufficiently documented, still increased after 6 weeks Normalisation of ALT not documented Autoimmunological parameters , 35 not assessed Kava hepatotoxicity , 545–563 Wilson‘s disease not excluded 547 548 R. Teschke Table 1. (Continued) Ó et al. 2010 The Authors. JCPT 02 BfArM 68 f Acetonic 24 210 St John’s wort f Exclusion of non-kava causes Daily kava overdose + ) (7, 12, 18) 94006568 Aluminium not communicated Prolonged kava treatment hydroxide Observed recurrent increase (2 years) of ALT during kava Exclusion