Alimentary Pharmacology and Therapeutics

Review article: herbal hepatotoxicity – an update on traditional Chinese medicine preparations

R. Teschke*, A. Wolff†, C. Frenzel‡ & J. Schulze§

*Department of Internal Medicine II, SUMMARY Division of Gastroenterology and Hepatology, Klinikum Hanau, Academic Teaching Hospital of the Background Medical Faculty of the Goethe Although evidence for their therapeutic efficacy is limited, herbal traditional University Frankfurt/Main, Hanau, Chinese medicine (TCM) preparations increasingly gain popularity. In con- Germany. † trast to other herbal products, adverse effects by herbal TCM including liver Department of Internal Medicine II, toxicity were rarely reported. In recent years, more cases were published, Division of Gastroenterology, Hepatology and Infectious Diseases, providing new clinical challenges. Friedrich Schiller University Jena, Jena, Germany. Aim ‡ Department of Medicine I, University To summarise comprehensively the literature on herbal TCM hepatotoxicity Medical Center Hamburg Eppendorf, since 2011. Hamburg, Germany. §Institute of Industrial, Environmental and Social Medicine, Medical Faculty, Methods Goethe University Frankfurt/Main, PubMed was searched using key words related to TCM, the results were Frankfurt, Germany. restricted to full English-language publications and abstracts published since 2011. In addition, the database of the National Institutes of Health (NIH) and LiverTox was accessed under the topic ‘Drug record: Chinese and other Correspondence to: ’ Dr R. Teschke, Department of Internal Asian herbal medicines . Medicine II, Klinikum Hanau, Academic Teaching Hospital of the Results Goethe University of Frankfurt/Main, Since 2011, new case reports and case series provided evidence for herbal Leimenstrasse 20, D-63450 Hanau, hepatotoxicity by TCM, focusing on nine TCM herbal mixtures and four Germany. individual TCM herbs with potential health hazards. These were the TCM E-mail: [email protected] products Ban Tu Wan, Chai Hu, Du Huo, Huang Qin, Jia Wei Xia Yao San, Jiguja, Kamishoyosan, Long Dan Xie Gan Tang, Lu Cha, Polygonum multiflo- Publication data rum products, Shan Chi, ‘White flood’ containing the herbal TCM Wu Zhu Submitted 24 March 2014 Yu and Qian Ceng Ta, and Xiao Chai Hu Tang. Other developments include First decision 9 April 2014 the establishment of a new and early diagnostic serum marker for hepatotox- Resubmitted 22 April 2014 icity caused by pyrrolizidine alkaloids, assessed using ultra performance Accepted 28 April 2014 – EV Pub Online 20 May 2014 liquid chromatography mass spectrometry analysis, and new regulatory details to improve herbal TCM product quality and safety. This uncommissioned review article was subject to full peer-review. Conclusion Stringent evaluation of the risk/benefit ratio is essential to protect traditional Chinese medicines users from health hazards including liver injury.

Aliment Pharmacol Ther 2014; 40: 32–50

32 ª 2014 John Wiley & Sons Ltd doi:10.1111/apt.12798 Review: herbal hepatotoxicity of traditional Chinese medicines

INTRODUCTION Du Huo,12 Huang Qin,13, 14 Jia Wei Xia Yao San,11 Jig- Hepatotoxicity by herbal Traditional Chinese Medicine uja,15 Kamishoyosan,16 Long Dan Xie Gan Tang,11 Lu (TCM) is reported from many countries around the Cha,17 Polygonum multiflorum products,10, 18 Shan world, , Hong Kong, Taiwan, Japan, Korea, Singa- Chi,19, 20 White flood21 and Xiao Chai Hu Tang.11 The pore, , Australia, Italy, Spain, France, the Neth- NIH and LiverTox provided actual information on Chi erlands, the United Kingdom, Iceland, Canada, the R Yun (Breynia officinalis), Dai Saiko To, Jin Bu Huan United States and Argentina.1, 2 In the past 3 years, new (Lycopodium serratum), Ma Huang (Ephedra sinica), Sho cases of this neglected liver disease with substantially Saiko To, Shou Wu Pian (Polygonum multiforme) and improved data presentation were published and renewed Xiao Chai Hu Tang.1 – the scientific interest.1 5 Recent approaches to verify overall causality included the application of sophisticated Cases and case series liver specific algorithms and specific criteria for positive As a case report of fulminant hepatic failure caused by reexposure tests.6 Other developments include the estab- the TCM Ban Tu Wan, a herbal mixture containing lishment of pyrrole–protein adducts in the serum as an P. multiflorum as one of its multiple ingredients, early diagnostic marker for pyrrolizidine alkaloid (PA) was described in 2012 for the first time in the English hepatotoxicity, assessed with a new ultra performance literature.10 In 2011, a new detailed report estimated the liquid chromatography–mass spectrometry (UPLC–MS) risk of hospitalisation for liver injury associated with the analysis.7 Actual systematic reviews highlighted the qual- use of traditional Chinese herbal products containing ity of TCM products with focus on quality control Chai Hu (Radix bupleuri), referring to 61 liver injury including contamination and adulteration.8, 9 Finally, cases in 639 779 patients with chronic hepatitis B virus future approaches of pharmacovigilance practice and risk infection.11 The first case of liver injury by the herbal control for TCM in China are encouraging.9 TCM Du Huo ( archangelica) was reported in We systematically review actual developments of her- 2013,12 hepatotoxicity by the herbal TCM Huang Qin bal hepatotoxicity by TCM of the past 3 years and () was confirmed by four new cases update the list of overall herbal TCM products suspected in 2012 and 2013.13, 14 In 2011, the liver injury risk of for potential liver injury. the TCM Jia Wei Xia Yao San was quantified in refer- ence to seven new patients.11 A new case of liver injury LITERATURE SELECTION CRITERIA by the herbal TCM Jiguja (Hovenia dulcis) was reported We selectively searched the PubMed database for the in 2012 in a 3.5-year-old boy with acute liver failure terms Traditional Chinese Medicine, Traditional Chinese and a liver transplant, confirming two previous cases in Medicine liver injury, Traditional Chinese Medicine hep- adult patients.15 In 2011, a case of liver injury was atotoxicity, Chinese herbal hepatotoxicity, Chinese herbal reported following the use of Kamishoyosan, a tradi- liver injury, herbal hepatotoxicity and herb induced liver tional Japanese herbal drug (Kampo medicine), similar injury. The search was limited to English-language to and derived from the herbal TCM Jia Wei Xia Yao reports and abstracts. The publications obtained were San.16 There were 47 additional cases reported in 2013, analysed for individual herbs and herbal mixtures which were caused by the herbal TCM Lu Cha (Camel- incriminated in herb-induced liver injury (HILI) by Tra- lia sinensis, syn. Chinese green tea).17 In 2011, 25 new ditional Chinese Medicine. We in addition accessed the hepatotoxicity cases were published after consumption database of the National Institutes of Health (NIH) and of the TCM mixture Shou Wu Pian, with P. multiflo- LiverTox under the topic Drug record: Chinese and rum as the main component.18 A new case of the hepa- other Asian herbal medicines.1 We focused on actual tic sinusoidal obstruction syndrome (HSOS) by the developments within the last few years and restricted our TCM Shan Chi (Gynura segetum) was published in search from 2011 to March 15, 2014. 2011; initially, this case was erroneously ascribed to the herbal TCM Jing Tian San Qi (Sedum aizoon).19 Four NEW HEPATOTOXICITY CASES additional new cases by the TCM Shan Chi were pub- lished in 2012,20 with at least 51 cases reported until TCM products 201119 and 116 cases until 2012.20 For the first time, Since 2011, highlighted new case reports and case series hepatotoxicity was reported in 2012 due to the use of – of herbal hepatotoxicity by TCM appeared.10 21 These White flood, containing among other ingredients the related to the TCM products Ban Tu Wan,10 Chai Hu,11 herbal TCM Wu Zhu Yu (Evodia rutaecarpa) and Qian

Aliment Pharmacol Ther 2014; 40: 32-50 33 ª 2014 John Wiley & Sons Ltd R. Teschke et al.

Ceng Ta (Huperzia serrata).21 Finally in 2011, the liver Exemptions include well-known herbal hepatotoxic injury risk of the TCM Xiao Chai Hu Tang was quan- chemicals such as ephedrine and PAs.2, 3, 7, 22 tied in 19 new cases of liver injury.11 Since 2011, there- As hepatotoxicity by the herbal TCM Ba Jiao Lian fore, new cases focused on nine TCM herbal mixtures (Dysosma pleianthum) does not fulfil all criteria,2, 23 this – and four individual TCM herbs.10 21 herb was not further considered as hepatotoxic.2 Evi- Combining the results of a previous analysis2 with dence against a hepatotoxic potential of the herbal TCM – those of the present case analysis,10 21 this compilation Jing Tian San Qi (S. aizoon) was provided in recent provides overall 18 classifiable TCM herbal mixtures, a studies showing that in patients with the HSOS, the hep- group of unclassifiable TCM herbal mixtures, and 39 atotoxic PAs in the herbal TCM Shan Chi (G. segetum) individual TCM herbs with reported potential hepatotox- were responsible rather than the misidentified S. aizoon icity (Table 1). TCM products often are blends of a mix- lacking these alkaloids.2, 19, 20 ture of herbs, king herb and other constituents are Since 2011, 149 new cases of liver injury by a single – believed to modify toxicity or synergistically increase the TCM herb have been published in five reports,12, 15, 18 king herb effects. Such conglomeration of constituents 20 covering four herbs as TCM ingredient, namely Du makes the identification and assignment of causative Huo (A. archangelica), Jiguja (H. dulcis), P. multiflorum hepatotoxic compounds extremely difficult, whereas and Shan Chi (G. segetum) (Table 3). These four single TCM products with one single herb as constituent facili- herbs of TCM since 2011 (Table 3) compare to a total of tate causality attribution. 39 single TCM herbs with potential liver injury reported since 1990 (Table 1).2 As for herbal mixture cases Hepatotoxic herbal mixtures (Table 2), all relevant case data by these herbs are pro- For the 18 classifiable and for unclassifiable TCM herbal vided (Table 3). mixtures, five or more with up to dozens of ingredients had been declared, mostly herbs (Table 1). All publica- QUALITY ASPECTS tions attributed causality of the liver disease to these her- – bal mixtures as whole products.2, 10 21 In principle, liver Herbal products injury can be caused by one declared herb, by other In China and other Asian countries, possible misidentifi- undeclared herbs, or by a nonherbal ingredient. As the cation of herbs for TCM may result from commercial incriminated TCM mixtures contain five or more herbs, cultivation by differently trained farmers,8 from local assigning causality to one single herb is disputable.2 growth by some consumers for personal use19 and bulk Since 2011, a total of 139 cases of liver injury by nine sale on public markets. Authentication of herbs was not – different TCM herbal mixtures have been reported in done in the cases analysed in the present study,10 18, 21 seven publications (Table 2).10, 11, 13, 14, 16, 17, 21 They with the exceptions of two publications with thorough provide details on the product, latency, duration of use, analyses finding misidentification.2, 19, 20 daily dose, symptoms, clinical features, laboratory results, Quality aspects of herbal TCM products are of major other toxicities, comedication, liver biopsy, type of liver concern (Figure 1),8, 9 in analogy to other herbs.24 In injury, causality assessment, severity and outcome. How- the analysed case reports, information commonly is ever, data in many individual cases or case series are missing whether Good Agricultural and Manufacturing incomplete, with major shortcomings especially in case Practices for the incriminated herbal products were fol- – series (Table 2). The total of nine different TCM herbal lowed.10 21 Reports rarely provide details on the plant mixtures with potential liver injury published since 2011 family with subgroups and plant parts used. Specification (Table 2) compares to 18 different TCM mixtures of plant parts is essential as hepatotoxic chemicals may reported since 1990 (Table 1).2 be distributed unevenly.24 Whenever rhizomes are used, clarification is needed about peeled or unpeeled use, Hepatotoxic herbs and compounds extent of peeling and whether the adjacent parts were In overall 39 TCM products with potential hepatotoxic- included in the herbal product. Quality evaluation is ity, one single herb was declared as individual constitu- hardly achievable in some herbal products such as herbal ent and thus the assumed cause (Table 1). Identifying a slices or extracts. It appears that plant specifications are single hepatotoxic chemical in this particular herb is dif- poorly considered and rarely reported for the reviewed – ficult because herbs usually have dozens of chemicals herbal TCM products,10 21 additional prospective regula- with mostly unknown hepatotoxic potency in humans. tory efforts are necessary.9

34 Aliment Pharmacol Ther 2014; 40: 32-50 ª 2014 John Wiley & Sons Ltd Review: herbal hepatotoxicity of traditional Chinese medicines

Table 1 | Risky hepatotoxic products of herbal traditional Chinese medicine with ingredients An Shu Ling Lycopodium serratum or, rarely, several unrelated herbal species including Corydalis species, Panax , Pseudo ginseng, or two species of Stephania Bai Fang , Cyperus rotundus, Ginseng, Ligusticum wallichii, Paeonia alba and glutinosa Bai Xian Pi Dictamnus dasycarpus Ban Tu Wan Angelica sinensis, Chaenomeles, Codonopsis pilosula, Notopterygium, Polygonum multiflorum, Rehmannia and Schisandra Bo He Mentha haplocalyx Bo Ye Qing Niu Dan Tinospora crispa Bofu Tsu Sho San Angelica, Atractylis, Cnidium, Gardenia, Ephedra, Forsythia, Glycyrrhhiza, Gypsum fibrosum, Ledebouriella, Mentha, Paeonia, Platycodon, Rheum, Schizonepeta, Scutellaria and Zingiber; Kadinum (talcum powder), and sodium sulfuricum. Boh Gol Zhee Psoralea corylifolia Cang Er Zi Xanthium sibiricum Chai Hu Bupleurum falcatum, commonly as an ingredient of various mixtures consisting of multiple TCM herbs Chaso Camellia sinensis, Cassia tora (syn. Senna), Crataegus, Chrysanthenum morifolium Ramat., Lotus and Lycium barbarum; N-nitroso-fenfluramine Chi R Yun Breynia officinalis Chinese herbal mixtures (various) Dictamnus dasycarpus (syn. Bai Xian Pi), Gentiana scabra, diffusa, Paeonia suffructicosa, Paris polyphylla, Rehmannia glutinosa, Smilax glabra and Sophora subprostrata; Angelica sinensis, Bupleurum chinese, Dictamnus dasycarpus, Paeonia suffructiosa, Philodendron chinese, Saposhnikovia divaricata, Shisandra chinesis, Shizonepeta tenuifolia and Tribulus terrestris; Cocculus trilobus, Dictamnus dasycarpus, Eurysolen gracilis, Glycyrrhiza, Lophatherum, Paeonia, Potentilla and Rehmannia glutinosa; Alisma plantago aquatica, Artemisia capillaris, Bupleurum, Chrysanthemum morifolium, Circuma, Gardenia jasminoidis, Gentiana scabra, Glycyrrhiza, Magnolia, Paeonia, Plantago asiatica and Saussurea lappa. Chuan Lian Zi Melia toosendan Ci Wu Jia Acanthopanax senticosus Da Chai Hu Tang Bupleurum falcatum, Ginseng, Glycyrrhiza glabra, Pinellia, Scutellaria, Zingiber officinale and Zizyphus jujuba Da Huang Rheum palmatum Du Huo Angelica archangelica Gan Cao Glycyrrhiza uralensis, syn. Liquorice Ge Gen Pueraria lobata, syn. Arrowroot Ho Shou Wu Polygonum multiflorum, syn. He Shou Wu Hu Bohe You Mentha pulegium, syn. Pennyroyal oil Hu Zhang Polygonum cuspidatum Huang Qin Scutellaria baicalensis Huang Yao Zi Dioscorea bulbifera Hwang Geun Cho Corydalis speciosa Ji Gu Cao Abrus cantoniensis Ji Ji Chloranthus serratus Ji Xue Cao Centella asiatica, syn. Gotu Kola Jia Wei Xia Yao San Atractylodes macrocephala, Angelica sinensis, Bupleurum, Gardenia, Glycyrrhiza uralensis, Mentha haplocalyx, Paeonia alba, Paeonia suffruticosa, Poria and Zingiber officinale Jiguja Hovenia dulcis Jin Bu Huan Lycopodium serratum or, rarely, several unrelated herbal species including Corydalis species, Panax ginseng, Pseudo ginseng, or two species of Stephania Jue Ming Zi Cassia obtusifolia, syn. Senna obtusifolia Kamishoyosan Angelica sinensis, Atractylodes racea, Bupleurum falcatum, Gardenia, Glycyrrhiza glabra, Mentha haplocalyx, Moutan, Paeonia alba, Sclerotium Poriae Cocos and Zingiber officinale Kudzu Pueraria thunbergiana Lei Gong Teng Tripterygium wilfordii Hook Long Dan Xie Gan Tang Acebia, Alisma, Angelica sinensis, Bupleurum, Gardenia, Gentiana, Glycyrrhiza, Plantago, Rehmannia and Scutellaria Lu Cha Camellia sinensis, syn. green tea, also as an ingredient of various mixtures consisting of multiple TCM herbs Ma Huang Ephedra sinica Mao Guo Tian Jie Cai Heliotropium lasiocarpum Onshido Aloe, Camellia sinensis, Crataegus, Gynostemma pentaphyllum makino and Raphanus; N-nitroso-fenfluramine Qian Li Guang Senecio scandens Polygonum multlflorum Ren Shen Panax ginseng

Aliment Pharmacol Ther 2014; 40: 32-50 35 ª 2014 John Wiley & Sons Ltd R. Teschke et al.

Table 1 | (Continued) Sairei To Alisma, Atractylis, Bupleurum, Cinnamomum, Ginseng, Glycyrrhiza, Pinellia, Polyporus, Poria, Scutellaria, Zingiber and Zizyphus Shan Chi Gynura segetum Shen Min Black cohosh, Burdock, Cayenne pepper, biloba, Horse chestnut, Piper nigrum, Polygonum multiflorum and uva ursi; biotin, collagen (hydrolysed), niacin, pantothenic acid, silica (from plant sources), soy isoflavones, vitamin A and vitamin B6 Shi Can Teucrium chamaedrys, syn. Germander Shi Liu Pi Pericarpium granati Shou Wu Pian Achyranthes bidentata, Cuscuta chinensis, Eclipta prostrata, Ligustrum lucidum, Lonicera japonica, Morus alba, Polygonum multiflorum, Psoralea corylifolia, Rehmannia glutinosa, Rosa laevigata, Sesemum indicum and Siegesbeckia orientalis Tian Hua Fen Trichosanthes kirilowii White flood Beet root, caffein, cocoa bean, Evodia rutaecarpa, Huperzia serrata and vinpocetine (from Vinca plant); acesulfame potassium, calcium silicate, carnitine tartrate, Carno-Syn beta-alanine, citrulline, cryptoxanthin, folic acid, gamma-aminobutyric acid (GABA), glucuronolactone, selenium, L-norvaline, L-tyrosine, lutein, malic acid, ornithine, potassium gluconate, sucralose, sugar cane, watermelon flavour and zeaxanthin Wu Bei Zi Galla chinensis Xi Shu Camptotheca acuminata Xiao Chai Hu Tang Bupleurum falcatum, Ginseng, Glycyrrhiza glabra, Pinellia tuber, Scutellaria baicalensis, Zingiber officinale and Zizyphus jujuba Yin Chen Hao Artemisia capillaris Zexie Alisma orientalis Zhen Chu Cao Phyllanthus urinaria

Reported hepatotoxicity of herbal traditional Chinese medicine products. Data are quoted from publications since 201110–21 and a previous analysis.2

– Nonherbal TCM products specific laboratory data.10 21 This facilitates an initial The reviewed reports provide evidence that the suspected causality assessment, as shown in a recent comprehen- herbal TCM products were not thoroughly analysed for sive study of 25 liver injury cases by P. multiflorum – adulteration or contamination.10 21 Both alterations are (Table 3).18 In case series, however, case data quality well associated with herbal TCM products and consid- often is incomplete due to space limitations, inclusion of ered a potential risk for human health.8 If data on adul- poorly documented cases or both. Valid case assessment – terants or contaminants are not available,10 21 firm requires essential items to be collected during the patient causality attribution to any of these nonherbal com- care (Figure 1).3, 25 Details of the herbal TCM and treat- pounds is not possible. ment modalities including start and end of herbal use Also, potential hepatotoxicity of other nonherbal ele- are required items, as are liver values before and after ments of TCM creates concern. Nonherbal compounds dechallenge. Most important, the treatment indication is commonly are used in connection with herbal TCM crucial and will help to attribute nonspecific symptoms products and sometimes erroneously are named as such. to an emerging liver disease, necessitating especially liver Known or potentially hepatotoxic nonherbal TCM ele- enzyme values before treatment. For a structured ments include venoms of vipers, larvae, mushrooms, and approach of case causality assessment, a specific check animal parts from antelopes, scorpions, centipedes, list should be used (Figure 2).25, 26 snakes and fishes (Table 4). This information is derived from a few anecdotal reports mainly in the Chinese liter- LIKELIHOOD OF CAUSALITY ature and was summarised recently.2 At this stage, it should be perceived as preliminary and poorly validated Clinical evaluation information arousing scientific interest that requires fur- For a clinical diagnosis of liver injury by a herbal ther investigations until firm conclusion. Therefore, TCM, a temporal association must be shown between definitive additional information on products, symptoms, the intake of a clearly identified herbal product and clinical features, causality assessment, severity and out- symptom onset or increased liver enzymes (Figure 1), come presently cannot be provided. as well as an exclusion of alternative diagnoses (Fig- ure 2). This is mandatory as no specific symptoms for Case data liver injury by herbal TCM exist. Indicators of liver Overall, the data quality of the case reports was fairly damage are anorexia, fatigue, nausea, vomiting, dyspep- good, especially in recent years, with more clinical and sia, bloating, abdominal discomfort, abdominal pain,

36 Aliment Pharmacol Ther 2014; 40: 32-50 ª 2014 John Wiley & Sons Ltd Review: herbal hepatotoxicity of traditional Chinese medicines

Table 2 | Details of liver injury cases by herbal mixtures of traditional Chinese medicine

TCM Sex/age Treatment Severity/ Liver products (y) modalities Clinical features outcome histology Ban Tu Wan f/middle- Product of unknown General malaise, Fulminant hepatic n.a. 1 case10 aged supplier, unknown diarrhoea, myalgia, failure. Acute dailydose, sporadic arthralgia for 1 week. renal failure use for 2–3 months Presentation with with a prior to presentation lethargy, confusion, pre-renal diffusely tender and aetiology. distended abdomen, Evaluations for lack of scleral icterus. LTX but AST 5386 U/L, ALT requirements 1214 U/L, ALP 199 U/ were not met L, bilirubin 4.1 mg/dL. secondary to Hepatocellular type of septicaemia. injury. Causality Lethal outcome assessment by CIOMS not done Chai Hu n.a./n.a n.a., possibly some Clinical features n.a., lack n.a. n.a. 61 cases11 hepatotoxic drugs as of laboratory data and comedication causality assessment by CIOMS. Adjusted OR 1.90 Huang Qin f/78 Product of unknown Acute painless jaundice. Good outcome Portal tracts 1 case13 supplier, one tablet ALT 1626 U/L, ALP after containing mild, twice a day for 3 354 U/L, bilirubin discontinuation, predominantly weeks. Comedicated 7.2 mg/dL. with mononuclear cell multivitamins Hepatocellular type of normalisation infiltrates, with injury. Recurrent or substantial many eosinophils. jaundice improvements of Significant lobular upon reexposure. laboratory values inflammatory cell CIOMS infiltrates, scale was mentioned including but not applied eosinophils. Numerous acidophil bodies and scattered ballooned hepatocytes 1 case14 f/62 Product of unknown Jaundice. ALT 1247 U/L, Good outcome Panacinar supplier, four tablets ALP 297 U/L, bilirubin with hepatitis, with per day for 2.5 weeks 6.9 mg/dL. normalisation expansion of the with tapering down to Hepatocellular of laboratory portal tracts by a two tablets a day for type of injury. Initially values after mixed chronic 4 days prior to shortness of breath due discontinuation inflammatory hospitalisation. to pulmonary bilateral infiltrate Comedicated interstitial infiltrates. composed of metformin for Causality by the lymphocytes, diabetes Naranjo scale, with a eosinophils and mellitus score of 7 probable plasma cells causality Jia Wei Xia n.a./n.a. n.a. Clinical features n.a., lack n.a. n.a. Yao San of laboratory data and 7 cases11 of causality assessment by CIOMS. Adjusted OR 1.87

Aliment Pharmacol Ther 2014; 40: 32-50 37 ª 2014 John Wiley & Sons Ltd R. Teschke et al.

Table 2 | (Continued)

TCM Sex/age Treatment Severity/ Liver products (y) modalities Clinical features outcome histology Kamishoyosan f/48 Product of unknown Routine check 2 weeks Treatment with Necrosis and 1 case16 supplier. Daily dose after treatment start: glycyrrhizin, then acidophilic n.a., use for 2 months AST 64 U/L, ALT with ursodeoxy- degeneration of prior to hospitalisation 102 U/L. At cholic acid. By hepatocytes. hospitalisation 6 weeks 42 days after Enlargement of later, general fatigue, discharge, almost portal tracts, scleral jaundice. ALT normalisation of with infiltration 972 U/L, ALP 420 U/ laboratory test of lymphocytes, L, bilirubin 12.8 mg/dL. results eosinophils and Lack of causality few plasma cells assessment by CIOMS algorithm Long Dan n.a./n.a. n.a. Clinical features n.a., lack n.a. n.a. Xie Gan Tang of laboratory data and 14 cases11 of causality assessment by CIOMS. Adjusted OR 3.53 Lu Cha n.a./n.a. n.a. Symptoms not specified. Severity of liver n.a. 33 cases17 Values of ALT, ALP and injury: bilirubin not specifically mild (n = 5), presented. Types of moderate injury: (n = 10), hepatocellular (n = 4), moderate- cholestatic (n = 25), hospitalised mixed (n = 4). (n = 9), severe Causality assessment (n = 2), fatal by DILIN method: (n = 4), not definite (n = 6), adjudicated very likely (n = 11), (n = 3) probable (n = 8), possible (n = 2), unlikely (n = 3), not adjudicated yet (n = 3) White flood m/23 White flood (Controlled Jaundice and pruritus, Recovery n.a. 1 case21 Labs, New Rochelle, subsequently to dark NY, USA). Daily dose urine, dysgeusia, n.a., use for 90 days anorexia. ALT 189 U/L, ALP 173 U/L, bilirubin 13.8 mg/dL. Mixed type of injury. CIOMS causality assessment not done Xiao Chai n.a./n.a n.a. Clinical features n.a., lack n.a. n.a. Hu Tang of laboratory data and 19 cases11 of causality assessment by CIOMS. Adjusted OR 2.91 Details refer to actual cases published since 2011. ALT, alanine aminotransferase; ALP, alkaline phosphatase; AST, aspartate aminotransferase; CIOMS, Council for International Organizations of Medical Sciences; DILIN, Drug-Induced Liver Injury Network; LTX, liver transplantation; OR, odds ratio; TCM, Traditional Chinese Medicine.

38 Aliment Pharmacol Ther 2014; 40: 32-50 ª 2014 John Wiley & Sons Ltd Review: herbal hepatotoxicity of traditional Chinese medicines

Table 3 | Details of liver injury cases by single herbs of traditional Chinese medicine

TCM Sex/ Treatment Severity/ Liver products age (y) modalities Clinical features outcome histology

Du Huo n.a./n.a. n.a. Symptoms not specified. Symptomatic n.a. 1 case12 ALT 89 U/L, ALP course 371 U/L, bilirubin 3 mg/dL. Cholestatic type of injury. CIOMS score 7, probable causality Jiguja m/3.5 Tea for Jaundice for 7 days, dark Acute fulminant Severe hepatocytic 1 case15 12 mo. urine for 10 days, hepatitis, transferral degeneration with nausea. ALT 1813 U/L, to another hospital frequent acidophilic ALP 336 U/L, bilirubin for further bodies and balloon 10.3 mg/dL. evaluation and change, apoptosis, Hepatocellular type consideration for severe lobular activity of injury. CIOMS liver transplantation. and moderate porto- scale of 1990 with Further clinical periportal activity. six points, probable course and final Frequent spotty causality outcome not inflammatory reported collections and some cholestasis. Periportal fibrosis with some lymphoid cell infiltration, eosinophils, and neutrophils Polygonum multiflorum 25 cases18 Case 1 m/57 Daily dose Jaundice, fever, Severe course, death Liver histology in 8 not n.a., use arthralgia, after 13 days of further declared for 30 days thrombocytopaenia . hospitalisation patients: inflammatory ALT 853 U/L, ALP cell infiltration, necrosis 173 U/L, bilirubin of hepatic cells and 28.1 mg/dL. Mixed type mild to moderate of injury. CIOMS score cholestasis consistent 8, probable causality with acute hepatitis Case 2 m/49 Treatment Jaundice, fever, skin rash, Recovery duration thrombocytopaenia. of 90 days ALT 1235 U/L, ALP 465 U/L, bilirubin 32.9 mg/dL. Mixed type of injury. CIOMS score 9, highly probable causality Case 3 m/46 Use for 2 days Jaundice. ALT 1287 U/L, Recovery ALP 146 U/L, bilirubin 19.7 U/L. Hepatocellular type of injury. CIOMS score 9, highly probable causality Case 4 f/47 Consumption Jaundice, fever, arthralgia, Recovery for 60 days pancytopaenia. ALT 1947 U/L, ALP 218 U/ L, bilirubin 30.4 mg/dL. Hepatocellular type of injury. CIOMS score 8, probable causality

Aliment Pharmacol Ther 2014; 40: 32-50 39 ª 2014 John Wiley & Sons Ltd R. Teschke et al.

Table 3 | (Continued)

TCM Sex/ Treatment Severity/ Liver products age (y) modalities Clinical features outcome histology

Case 5 m/34 Treatment Jaundice, fever, Recovery for 30 days pancytopenia. ALT 1452 U/L, ALP 111 U/L, bilirubin 25.3 mg/dL. Hepatocellular type of injury. CIOMS score 9, highly probable causality Case 6 m/58 Use for Jaundice. ALT 1898 U/L, Recovery 35 days ALP 134 U/L, bilirubin 13.7 mg/dL. Hepatocellular type of injury. CIOMS score 9, highly probable causality Case 7 m/59 Treatment Abdominal pain. ALT Recovery duration 1245 U/L, ALP 155 U/ of 30 days L, bilirubin 1.6 mg/dL. Hepatocellular type of injury. CIOMS score 8, probable causality Case 8 f/54 Consumption Jaundice. ALT 1752 U/L, Recovery for 4 days ALP 286 U/L, bilirubin 8.4 mg/dL. Hepatocellular type of injury. CIOMS score 7, probable causality. Case 9 f/46 Use for Jaundice. ALT 1804 U/L, Recovery 30 days ALP 81 U/L, bilirubin 6.2 mg/dL. Hepatocellular type of injury. CIOMS score 8, probable causality Case 10 m/62 Treatment Jaundice, fever. ALT 1174 Recovery for 90 U/L, ALP 175 U/L, days bilirubin 9.2 mg/dL. Hepatocellular type of injury. CIOMS score 9, highly probable causality Case 11 f/63 Consumption Fatigue. ALT 943 U/L, Recovery for 30 days ALP 137 U/L, bilirubin 4.2 mg/dL. Hepatocellular type of injury. CIOMS score 9, highly probable causality Case 12 m/24 Use for Jaundice. ALT 1652 U/L, LTX Massive 60 days ALP 140 U/L, bilirubin hepatocellular 31.9 mg/dL. necrosis Hepatocellular type of injury. CIOMS score 7, probable causality

40 Aliment Pharmacol Ther 2014; 40: 32-50 ª 2014 John Wiley & Sons Ltd Review: herbal hepatotoxicity of traditional Chinese medicines

Table 3 | (Continued)

TCM Sex/ Treatment Severity/ Liver products age (y) modalities Clinical features outcome histology

Case 13 m/45 Treatment Abdominal pain. ALT 271 Recovery duration U/L, ALP 164 U/L, of 20 days bilirubin 2.9 mg/dL. Mixed type of injury. CIOMS score 8, probable causality Case 14 m/41 Use for Jaundice. ALT 520 U/L, Recovery 30 days ALP 143 U/L, bilirubin 9.9 mg/dL. Mixed type of injury. CIOMS score 7, probable causality Case 15 m/44 Treatment for Jaundice. ALT 1077 U/L, Recovery 7 days ALP 197 U/L, bilirubin 15.0 mg/dL. Hepatocellular type of injury. CIOMS score 9, highly probable causality Case 16 m/65 Consumption Jaundice. ALT 1107 U/L, Recovery for 10 days ALP 197 U/L, bilirubin 21.9 mg/dL. Hepatocellular type of injury. CIOMS score 8, probable causality Case 17 m/53 Use for Jaundice. ALT 1227 U/L, Recovery 180 days ALP 370 U/L, bilirubin 33.2 mg/dL. Mixed type of injury. CIOMS score 6, probable causality Case 18 f/42 Treatment No symptoms. ALT Recovery duration 500 U/L, ALP 181 U/L, of 10 days bilirubin 1.6 mg/dL. Mixed type of injury. CIOMS score 8, probable causality Case 19 m/61 At initial Myalgia. At initial Recovery exposure exposure, ALT 818 U/L, and at ALP 109 U/L, bilirubin reexposure 10 1.8 mg/dL. Hepatocellular months type of injury. CIOMS later, use score 10, highly for 1 day each probable causality Case 20 m/42 Consumption for Jaundice. ALT 1677 U/L, Recovery 120 days ALP 93 U/L, bilirubin 15.8 mg/dL. Hepatocellular type of injury. CIOMS score 7, probable causality Case 21 f/48 Use for 3 days Jaundice. ALT 1142 U/L, Recovery ALP 145 U/L, bilirubin 15.9 mg/dL. Hepatocellular type of injury. CIOMS score 8, probable causality

Aliment Pharmacol Ther 2014; 40: 32-50 41 ª 2014 John Wiley & Sons Ltd R. Teschke et al.

Table 3 | (Continued)

TCM Sex/ Treatment Severity/ Liver products age (y) modalities Clinical features outcome histology

Case 22 f/54 Treatment Jaundice, skin rash. Recovery for 180 ALT 1519 U/L, ALP 187 days U/L, bilirubin 11.7 mg/dL. Hepatocellular type of injury. CIOMS score 6, probable causality Case 23 m/61 Consumption Jaundice. ALT 885 U/L, Recovery for 60 days ALP 224 U/L, bilirubin 21.2 mg/dL. Mixed type of injury. CIOMS score 10, highly probable causality Case 24 m/45 Use for Fatigue. ALT 1400 U/L, Recovery 30 days ALP 125 U/L, bilirubin 2.0 mg/dL. Hepatocellular type of injury. CIOMS score 9, highly probable causality Case 25 m/42 Treatment Jaundice, eosinophilia. Recovery duration ALT 1706 U/L, ALP 147 of 60 days U/L, bilirubin 26.3 mg/dL. Hepatocellular type of injury. CIOMS score 8, probable causality Shan Chi f/54 1–2 cups of Jaundice, ascites. ALT 42 Recovery Hepatic sinusoidal 1 case 19 herbal U/L, ALP 257 U/L, obstruction syndrome wine daily bilirubin 31 lmol/L. (HSOS) with sinusoidal for 90 Cholestatic type of injury. and venous lumens days CIOMS score 9, highly obliterated by type I, probable causality III, and IV collagen 5 cases20 Case 1 f/54 Daily dose Jaundice, ascites. ALT, ALP, Moderate In 1/5 patients, liver n.a., bilirubin, and injury type severe histology showed a use for n.a., possible causality clinical course, slight derangement in 90 days with CIOMS score 5 recovery the lobular architecture in the liver. Extension and congestion of hepatic sinusoids. Connection of some sinusoids to blood pool. Fibrosis especially around the blood sinuses near the central veins. Blood flow was obstructive at the site of the sinusoid rather than at the small hepatic vein Case 2 f/57 Treatment Jaundice, ascites. ALT, Severe clinical duration ALP, bilirubin, and course, of 21 days injury type n.a., chronicity probable causality with outcome CIOMS score 7. at day 100

42 Aliment Pharmacol Ther 2014; 40: 32-50 ª 2014 John Wiley & Sons Ltd Review: herbal hepatotoxicity of traditional Chinese medicines

Table 3 | (Continued)

TCM Sex/ Treatment Severity/ Liver products age (y) modalities Clinical features outcome histology

Case 3 f/41 Consumption Jaundice, ascites. ALT, Moderate severe for 30 days ALP, bilirubin, and clinical course, injury type n.a., recovery probable causality with CIOMS score 7 Case 4 f/54 Use for Jaundice, ascites. ALT, Severe clinical 2 years AST, bilirubin, and course, chronicity injury type n.a., outcome probable causality with at day 100 CIOMS score 6. Case 5 f/72 Treatment Jaundice, ascites. ALT, Severe clinical course, of 60 ALP, bilirubin, and liver failure, hepatic days injury type n.a., encephalopathy. possible causality with Death. CIOMS score 5. 116 cases20 f (56 cases), Treatment Jaundice (95/113, Treatment modalities Liver histology in 65 m (57 cases), duration 84.1%), ascites (115/ included patients: HSOS with n.a. (3 cases) n.a. 116, 99.1%), symptomatic expansion and /17–76 hepatomegaly (104/ treatment, congestion of hepatic 113, 92.0%). ALT prednisone, sinus, endothelial elevation (47/60), AST anticoagulation, swelling, wall elevation (50/58). TIPS, LTX, and ABC. thickening and Injury type n.a., CIOMS Recovery (75/113), incomplete lumen causality assessment chronicity (27/113), occlusion of the little not done or results not and mortality (11/ hepatic vessels presented 113, 9.7%)

Details refer to actual cases published since 2011. ABC, activating blood circulation; ALT, alanine aminotransferase; ALP, alkaline phosphatase; AST, aspartate aminotransferase; CI- OMS, Council for International Organizations of Medical Sciences; HSOS, hepatic sinusoidal obstruction syndrome; LTX, liver transplantation; OR, odds ratio; TCM, Traditional Chinese Medicine; TIPS, Transjugular intrahepatic portosystemic shunt. fever, dark urine, pale stool, pruritus, and jaundice, TCM Chai Hu (Bupleurum), cumulative doses of Bupleu- rarely hypersensitivity signs like rash and arthralgia. rum were calculated and the adjusted odds ratio (OR) These clinical features also occur in disorders of other estimated.11 With all Bupleurum products, the OR for digestive organs like gall-bladder, stomach, intestine hepatotoxicity was 1.90 irrespective of the dose, a cumu- and pancreas, or in systemic diseases such as severe lative Bupleurum dose of 19 g increased the OR by infections or malignancies. The analysis of the new 2.19.11 The risks from the TCM Long Dan Xie Gan Tang – cases10 21 shows appropriate temporal association and or the TCM Xiao Chai Hu Tang were higher with an exclusion of other diseases in most case reports but OR of 3.53 and 2.91, again positive dose–response rela- rarely in case series. tionships were found. Cumulative doses were not pro- Clinical disease characteristics of liver injury by herbal vided for other TCM herbs and herbal mixtures – – TCM are variable in all cases published since 201110 21 (Tables 2 and 3).10, 12 21 and refer to both herbal mixtures (Table 2) and individ- With the exception of White flood,21 no other TCM ual herbs (Table 3). Ages ranged from 3.5 to 78 years, brand or supplier product was identified (Tables 2 and – with a female:male ratio of 73:77 years (Tables 2 and 3). 3).10 20 Little information also was provided whether In the 25 P. multiflorum cases, the female:male ratio was treatment duration was identical to the latency period of 7:18 (Table 3).18 Daily doses rarely were reported in an symptom onset or laboratory value increase. appropriate way, but duration of use was fairly well doc- Symptoms were reported in 153 patients with liver umented and ranged from 1 day to 2 years (Tables 2 injury by four TCM herbs and herbal mixtures (Tables 2 and 3). In one case–control study of liver injury by the and 3); with one case each of the TCM herbal mixtures

Aliment Pharmacol Ther 2014; 40: 32-50 43 ª 2014 John Wiley & Sons Ltd R. Teschke et al.

Suspected herbal hepatotoxicity by traditional Chinese medicine

Quality specifications

Herbal products Case data

• Production in line with Good Agricultural • Indication of herbal TCM use with dates Practices (GAPs) and Good of symptoms leading to herbal treatment Manufacturing Practices (GMPs) • Daily dose, with details of the application • Definition of plant family, subfamily, form species, subspecies, and variety • Exact date of herbal TCM start and end • Definition of plant part, solvents, and • Calculation of cumulative dose solubilizers • Date of first symptoms under treatment • Lack of impurities, adulterants, and or initial increased liver values misidentifications, and absence of heavy • Liver values during dechallenge metals • Assessment of and age as risk • Minimal batch to batch and product to factors product variability • Concomitant drug(s) or herb(s) and their Figure 1 | Quality standards • Lack of variability in plant varieties relation to liver injury • Manufacturer with address • Search for alternative hepatic, required for evaluating herbal • Brand name with details of ingredients extrahepatic, and systemic causes products and case data of and their relative amountsin mixtures • Previous information on hepatotoxicity of • Batch number and expiration date the TCM product patients with suspected herbal • Asservation of the suspected product • Response to unintentional reexposure hepatotoxicity by traditional for quality analysis Chinese medicine (TCM).

dice for other TCM herbs and herbal mixtures that lack Table 4 | Risky hepatotoxic products of nonherbal – PAs (19/31 cases = 61.3%) (Tables 2 and 3).10 18, 21 Asci- traditional Chinese medicines tes (121/122 cases = 99.2%) and hepatomegaly (104/113 Bai Hua She (venom of the Chinese viper Agkistrodon acutus) cases, 92.0%) were highly prevalent again in HSOS cases Jiang Can (dried larvae of Bombyx batryticatus, infected by 19, 20 Batrytis bassiana) caused by PA-containing G. segetum (Table 3), and Ling Yang Qing Fei (antelope horn) were absent with other non-PA-containing herbs and her- – Liyu Danzhi (Cyprinus carpio, carp juice) bal mixtures (Tables 2 and 3).10 18, 21 In all 153 cases, Quan Xie (dry polypides of the Scorpio Buthus martensii) other symptoms like general malaise, fatigue, anorexia, Sang Hwang (Phellinus linteus, mushroom) fever, arthralgia, abdominal pain, dark urine and pruritus Song Rong (Agaricus blazei, Himematsutake as Japanese Kampo medicine, mushroom) were only sporadically reported, a few patients did not Wu Gong (dried polypites of the centipede Scolopendra have any symptoms (Tables 2 and 3). subspinipes mutilans) Most patients with PA-related HSOS had elevated ala- Wu Shao She (syn. Wu Xiao She, Sheng Wu Shao She, parts nine aminotransferase (ALT: 47/60 cases, 78.3%) and of the snake Zaocys dhumnades) Yu Dan (fish gall-bladder) aspartate aminotransferase (AST: 50/58 cases, 86.2%) val- ues (Table 3).20 Liver enzymes for other herbs varied Reported hepatotoxicity of nonherbal traditional Chinese medi- widely with ALT 42–1898 U/L,18, 19 ALP 81–465 U/L18 cine products. Data from references presented in a recent and bilirubin 1.6–33.2 mg/dL,18 eosinophilia and throm- study.2 bocytopaenia were rarely reported (Tables 2 and 3). As laboratory data sometimes were incompletely presented – Ban Tu Wan,10 Kamishoyosan16 and White flood,21 two (Tables 2 and 3),10 21 not all TCM cases may fulfil labo- cases of Huang Qin (S. baicalensis),13, 14 one case of Jiguja ratory-based hepatotoxicity criteria, defined by ALT and/ (H. dulcis),15 25 cases of P. multiflorum18 and 122 cases of or ALP values expressed in multiples of the upper limit Shan Chi (G. segetum).19, 20 Jaundice was most frequent of normal N.25, 26 For ALT, cut-off values range from (119/149 cases, 79.9%) (Tables 2 and 3), with 84.8% in >2N through >3N to >5N, and are set for ALP values at 100/118 cases alone in PA-induced HSOS by Shan Chi (G. >2N. An ALT cut-off of >5N will eliminate unspecific segetum) (Table 3),19, 20 and a lower prevalence of jaun- ALT increases and substantiate causality at higher proba-

44 Aliment Pharmacol Ther 2014; 40: 32-50 ª 2014 John Wiley & Sons Ltd Review: herbal hepatotoxicity of traditional Chinese medicines

Case causality assessment

Exclusion of alternative diagnoses

• Assessment of preexisting and coexisting liver diseases • Assessment of preexisting and coexisting diseases unrelated to the liver • Providing details to exclude alternative diagnoses • Assessment and exclusion of HAV, HBV, HCV, HEV, CMV, EBV, HSV, VZV • Liver and biliary tract imaging including color Doppler sonography of liver vessels • Specific evaluation of alcoholic, cardiac, autoimmune, and genetic liver diseases • Consideration of other possible liver diseases (several hundred) • Individual quantitative score of each alternative diagnosis • Comedication (synthetic drugs, herbal drugs, herbal and dietary supplements) • Individual quantitative score of each individual comedication

Liver specific, structured, and quantitative causality assessment by the scale of the Council for International Organizations of Medical Sciences (CIOMS)

Conditions and criteria of reexposure tests

Cholestatic Hepatocellular (± hepatocellular) type type of injury of injury Reexposure test result ALTb ALTr ALPb ALPr Positive <5N ≥2ALTb <2N ≥2ALPb Negative<5N<2ALTb<2N<2ALPb Negative ≥5N ≥2ALTb ≥2N ≥2ALPb Negative ≥5N <2ALTb ≥2N <2ALPb Negative ≥5N n.a. ≥2N n.a. Uninterpretable <5N n.a. <2N n.a. Uninterpretable n.a. n.a. n.a. n.a.

Figure 2 | Cases of suspected herbal hepatotoxicity by traditional Chinese medicine (TCM) and required steps for causality assessment by the CIOMS scale and reexposure conditions. Conditions and criteria for a reexposure test are based on Consensus Meetings, as referenced recently.2, 25, 26 Accordingly, required data for the hepatocellular type of liver injury are the ALT levels commonly just before reexposure, designed as baseline ALT or ALTb, and the ALT levels during reexposure, designed as ALTr and correlated with 2ALTb. Response to reexposure is positive, if both criteria are met: first, ALTb < 5N with N as the upper limit of normal, and second ALTr ≥ 2ALTb. Other variations lead to negative or uninterpretable test results. For the cholestatic (Æhepatocellular) type of liver injury, corresponding values of ALP are to be used rather than of ALT. Definitions of the hepatocellular and the cholestatic (Æhepatocellular) type of liver injury are provided earlier.25, 26 ALP, Alkaline phosphatase; ALPb, ALP baseline; ALPr, ALP reexposure; ALT, Alanine aminotransferase; ALTb, ALT baseline; ALTr, AT reexposure; CMV, cytomegalovirus; EBV, Epstein–Barr virus; HAV, hepatitis A virus; HBV, hepatitis B virus; HCV, hepatitis C virus; HEV, hepatitis E virus; HSV, herpes simplex virus; VZV, varicella zoster virus; n.a., not available. bility, an ALT value of >2N will include numerous and the ratio (R) of ALT:ALP is calculated. Liver injury nonspecific increases which require a more stringent is assumed as hepatocellular if ALT >2N alone or R ≥ 5; assessment and exclusion of alternative causes. as cholestatic with an increase of ALP >2N alone or Some TCM cases allow a laboratory value-based clas- R ≤ 2; as mixed type if ALT >2N, ALP is increased and sification of the injury type (Tables 2 and 3). Differentia- R 2–5. tion between hepatocellular, cholestatic or mixed type Liver biopsy results were available for some patients, hepatotoxicity is facilitated by comparing initial ALT with typical histological features of HSOS in patients and ALP serum activities.25, 26 Again, activities are who consumed PAs derived from the TCM Shan Chi (G. expressed as multiples of the upper normal limit (N), segetum) (Table 3).19, 20 For other patients, who used

Aliment Pharmacol Ther 2014; 40: 32-50 45 ª 2014 John Wiley & Sons Ltd R. Teschke et al.

TCM herbs and herbal mixtures without PAs, liver his- occurred after the use of P. multiflorum for one single tology showed acidophilic degeneration and frank necro- day and also after reexposure under the same conditions sis of hepatocytes, mixed inflammatory infiltrates and 10 months later (Table 3, case 19).18 The metabolic sub- rarely eosinophilia (Tables 2 and 3). type may prevail in the case series of the TCM Chai Hu, The use of herbal TCM therefore may be associated Jia Wei Xia Yao San and Xiao Chai Hu Tang, where the with two different types of liver diseases. One type is the risk for hepatotoxicity is observed with low and high PA-related HSOS from the herbal TCM G. segetum cumulative doses of Bupleurum (Table 2).11 The risk (Table 3),19, 20 in analogy to other PA-containing herbs increased with higher cumulative doses and is most – of TCM2 and non-TCM.3 5, 22 The other type develops probably due to a prolonged treatment of Bupleurum. – independently from PAs (Tables 2 and 3)10 18, 21 and is This is an essential criterion of the metabolic subtype similar to other herbs lacking PAs related to TCM2 or (Figure 3), which may also apply to some other listed – – non-TCM,3 6, 22, 25, 27 30 with some examples such as herbs (Tables 2 and 3) kava,6, 27 Indian Ayurvedic herbs,28 Greater Celan- Conditions are quite different in liver injury caused by dine,6, 29, 30 aloe, chaparral, germander, mistletoe and PA-containing herbs, as the resulting HSOS is a senna.6 dose-dependent intrinsic reaction (Figure 3) that is repro- ducible in humans and laboratory animals.7 PAs are Diagnostic marker unsaturated alkaloids and undergo enzymatic biotransfor- Specific diagnostic biomarkers for hepatotoxicity by mation to unstable toxic metabolites, probably pyrrolic potentially hepatotoxic herbs including TCM herbs are derivatives. This reaction involves hepatic microsomal available only for PAs present in herbal preparations, cytochrome P450.7, 19 With sufficiently high amounts of not in the other usually consumed herbs that cause PAs, one of its pyrrolic metabolites forms a pyrrole–pro- HILI.7 Herbal liver injury may be caused by multiple tein adduct in the liver cell, which leaves the liver cell, herbs with numerous potentially hepatotoxic chemicals enters the systemic circulation, and thus is assessable and as ingredients, which do not allow focusing on a single quantifiable in the serum as a biomarker.19 Pyrrole–pro- component as the culprit for a hepatotoxic reaction. In tein adducts have been detected as biomarkers in the addition, hepatotoxicity by herbal TCM caused by idio- blood of five patients with HSOS after intake of PA-con- syncratic reactions (Figure 3) may require only small taining TCM Shan Shi (G. segetum).20 The quantitative amounts of herbal ingredients with no apparent dose– analysis of pyrrole–protein adducts in the serum is now response relation. Consequently, hepatotoxic products possible by a new method using UPLC–MS analysis.7, 19 may be formed in small amounts by enzymatic conver- This procedure facilitated causality attribution in cases of sion in the liver cell and may not leave the cell, thus not HSOS by the PA-containing herbal TCM Shan Qi (G. being available as diagnostic marker in the serum. How- segetum), as compared to the herbal TCM Jing Tian San ever, assessing the pathogenetic classification of the hep- Qi (S. aizoon, syn. Stonecrop), which lacks PAs19 and is atotoxicity cases published since 2011 is challenging and not considered any more as a hepatotoxic herb causing – hampered by insufficient data (Tables 2 and 3)10 21 for HSOS.2, 7, 19, 20 It is generally accepted that herbs con- the diagnostic criteria (Figure 3). The classification as taining PAs such as Bush tea, Comfrey, Groundsel, Helio- idiosyncratic requires a low incidence in consumers, with tropium species, Indian herbs, Mate and Rattlebox one case for several ten thousands consumers who toler- typically cause HSOS as a specific type of a dose-depen- ate the herbal TCM product well. Prevalence data are dent liver injury, representing the intrinsic form of liver not available, as the reports published since 2011 focused injury.7, 22 PAs derived from G. segetum cause HSOS in – on single cases or small case series (Tables 2 and 3).10 21 humans19, 20 and in experimental animals20, thereby rep- Presumably, most herbs except the PA-containing G. resenting the intrinsic, dose-dependent form of liver segetum, which are considered in the reports since 2011 injury (Figure 3). Among the reports published since 2011 – – (Tables 2 and 3),10 18, 21 likely cause rare liver injury by (Tables 2 and 3),10 21 this form of injury is solely caused idiosyncrasy lacking experimental reproducibility (Fig- by G. segetum19, 20 and not by the other TCM herbs and – ure 3). In most cases, daily doses and cumulative doses herbal mixtures.10 18, 21 were not provided (Tables 2 and 3), preventing a valid subclassification into the metabolic or immunological Liver-specific causality assessment type (Figure 3). In one single case, however, an immuno- Clinical evaluations led the authors of recent publications logical type of injury has to be assumed, as liver injury since 2011 to conclude that some herbal TCM products

46 Aliment Pharmacol Ther 2014; 40: 32-50 ª 2014 John Wiley & Sons Ltd Review: herbal hepatotoxicity of traditional Chinese medicines

Suspected herbal Traditional Chinese Medicine hepatotoxicity requiring individual pathogenetic case classification and risk stratification for potential users

Pathogenic case classification of herbal hepatotoxicity

Definition Required criteria

• Idiosyncratic form Unpredictability Dose independency Long and variable latency period Low incidence in humans Lack of experimental reproducibility

- Metabolic type Duration of exposure: 1 week to 12 months Lack of hypersensitivity features Delayed response to reexposure (weeks)

- Immunologic type Duration of exposure: 1–5 weeks Hypersensitivity features Prompt response to reexposure with 1 or 2 doses

• Intrinsic form Predictability Dose dependency Short and consistent latency period High incidence in humans Experimental reproducibility

Risk stratification for potential users Figure 3 | Pathogenic case Definition Cause of increased individual risk classification is essential for individual risk stratification for • Idiosyncratic form Genetic variability potential users of herbal TCM products. • Intrinsic form High dose carry the risk of hepatotoxicity, at least for a few individ- tures (Table 2).10, 11, 13, 16, 17, 21 Only the assessment by – uals.10 21 However, firmly determining individual causal- the expert opinion-based Drug-Induced Liver Injury Net- ity with exclusion of possible alternative causes work method provided support that Lu Cha as green tea (Figure 2)25 was rarely done in these reports (Tables 2 extract and as an ingredient of various herbal TCM mix- – and 3).10 21 Vigorous causality assessment in hepatotox- tures is potentially hepatotoxic (Table 2).17 In a single icity cases is readily achievable, for instance, by the liver case of liver injury by Huang Qin, the Naranjo scale was specific, quantitative and structured scale of the Council applied (Table 2)14 that is liver unspecific and thereby for International Organizations of Medical Sciences (CI- obsolete for use in HILI cases.31 In some reports related OMS), also known as the Roussel Uclaf Causality Assess- to single herbs of TCM and published since 2011, the ment Method.25, 26 This scale calculates the probability CIOMS scale was applied and established probable and of a causal association by providing individual scores highly probable causality gradings for hepatotoxicity from items like time to onset, dechallenge characteristics, cases by Du Huo,12 P. multiflorum18 and Shan Chi19, 20 risk factors, nonherb- or nondrug-related causes, previ- (Table 3). The liver injury case by Jiguja15 was assessed ously known hepatotoxicity and unintentional reexpo- by the early scale version of 1990 (Table 3) but not by sure.26 the preferred CIOMS scale of 1993.26 As summarised In reports since 2011, CIOMS based assessment was recently, in TCM hepatotoxicity cases published between lacking in all TCM hepatotoxicity cases by herbal mix- 1990 and 2010, CIOMS based causality assessments pro-

Aliment Pharmacol Ther 2014; 40: 32-50 47 ª 2014 John Wiley & Sons Ltd R. Teschke et al.

– vided probable and highly probable causality gradings reports related to TCM and published since 2011,10 21 for the herbal TCM Bai Xian Pi, Ci Wu Jia, Gan Coa, comedication was rarely mentioned or excluded (Tables 2 Ge Gen, Ji Xue Cao, Jin Bu Huan, Kudzu, Ling Yang and 3). In connection with the use of the TCM Huang Qin Qing Fei Keli, Lu Cha, Rhen Shen, Shen Min, Syo Saiko (Table 2), one patient used multivitamins as comedica- To, Yin Chen Hao, Zexie and various undefined Chinese tion,13 the other one metformin for her diabetes mellitus.14 herbal mixtures.2, 6 Patients with hepatotoxicity by the TCM Chai Hu may have taken other hepatotoxic drugs as comedication Reexposure tests (Table 2), but details were not provided.11 Limited valid Conditions and criteria for a reexposure test are based information is available for synthetic drug comedication on Consensus Meetings, as referenced recently.2, 25, 26 possibly involved in the pathogenesis of liver injury in Accordingly, required data for the hepatocellular type of patients using herbal TCM products. In these patients, liver injury are the ALT levels commonly just before re- potentially hepatotoxic synthetic drugs may have been exposure, designed as baseline ALT or ALTb, and the prescribed as regular comedication, or could have been ALT levels during reexposure, designed as ALTr and included in the herbal TCM product as adulterants. correlated with 2ALTb (Figure 2). Response to reexpo- Apart from considerations as additional hepatotoxicant, sure is positive, if both criteria are met: first, ALTb <5N there is serious concern for multiple clinically relevant with N as the upper limit of normal, and second ALTr pharmacodynamic drug interactions with numerous herbs ≥2ALTb. Other variations lead to negative or uninter- including TCM herbs.32, 33 Herb–drug interactions may pretable test results. For the cholestatic (Æhepatocellular) change specific receptor number or affinity or cause sys- type of liver injury, corresponding values of ALP are to temic pharmacokinetic changes by altered absorption, dis- be used rather than of ALT (Figure 2). tribution, metabolism and excretion.33 Phases I and II Among the liver injury cases of TCM published since metabolism are located predominantly in the hepatic en- 2011, there was only one single case of a positive reexpo- doplasmatic reticulum. Phase I reactions are catalysed – sure test (Tables 2 and 3),10 21 considering specific con- mainly by various CYP isoenzymes as oxidative biotrans- ditions and criteria for a positive reexposure test result formation and are modified by enzyme induction and (Figure 2).2 As presented in the case series of P. multiflo- inhibition; phase II reactions attach polar and ionised rum under case 19 (Table 3), the 61-year-old Korean groups by conjugation, aiding their elimination.33 man used an unknown dose of P. multiflorum Thunb Herb–drug interactions have led to numerous adverse for 1 day and noticed myalgia.18 At first exposure, ALT reactions including graft rejection, cardiovascular compli- was 818 and 180 U/L after 9 days of cessation, with an cations, renal toxicity and increased bleeding risks.32, 33 ALTb <5N as likely assumed just before reexposure. This For herbal TCM, serious herb–drug interactions have been occurred after 10 months, again with a single dose of P. described, for instance, as potentiation of antico- multiflorum Thunb, which caused an ALTr of 1520 U/L. agulation with bleeding after intake of the TCM Dan Shen As ALTb <5N and ALTr ≥2ALTb (Table 3), this signi- (Salvia miltiorrhiza) or TCM Dong Quai (Angelica sinen- fied a positive reexposure test result (Table 3; Figure 2).2 sis).33 Dan Shen contains as main ingredients tanshinones Additional proof for causality was unintentionally pro- and phenolic compounds, which are vasorelaxants and vided by positive reexposure test results of hepatotoxicity antiplatelet drugs to treat cardiovascular diseases, and are TCM cases, as reported earlier.2, 6 These were cases by prone to potentially interact with warfarin, vasodilators the TCM Hwang Geun Cho, Ji Xue Cao, Jin Bu Huan, and . Phytoestrogens, flavonoids and cou- Lu Cha, Shou Wu Pian, Xiao Chai Hu Tang, and the marins as main ingredients of Don Quai have oestrogenic, individual herbs Centella asiatica (Ji Xue Cao), Corydalis vasorelaxant and anti-inflammatory properties used to speciosa (Hwang Geun Cho), L. serratum (Jin Bu Huan) treat gynaecological and circulatory disorders, possibly and Chinese herbal mixtures, confirming that these interacting with contraceptives, vasodilatators, anticoagu- products indeed carry a hepatotoxic risk.2, 6 lants and antiplatelet drugs.

HERB–DRUG INTERACTIONS RISK CONSIDERATIONS In HILI cases unrelated to TCM, comedication with other herbs and synthetic drugs is common and considered a Individual risk potential metabolic interaction at the hepatic microsomal Case classification based on pathogenetic criteria is useful cytochrome P450 (CYP) system.6, 25, 27, 29, 30 In HILI for individual and clinical risk assessment.25 Most cases

48 Aliment Pharmacol Ther 2014; 40: 32-50 ª 2014 John Wiley & Sons Ltd Review: herbal hepatotoxicity of traditional Chinese medicines of herbal TCM liver injury likely are idiosyncratic with a a need for additional multicentre, randomised, dou- genetic contribution, and due to their unpredictability, ble-blind, placebo-controlled clinical trials, their results they are unpreventable (Figure 3). The individual risk of should be published in peer reviewed journals in English. consumers likely is low, but well established for some Although clinical trials can verify the efficacy of herbal herbal TCM products by a causal grading with CIOMS, TCM treatment, they provide little additional informa- – a positive reexposure test, or both (Tables 2 and 3).2, 10 tion for hepatotoxic adverse reactions, as the number of 21 The individual risk is high for consumers of TCM treated patients commonly is low and the probability of herbs that contain PAs and cause HSOS (Table 3).19, 20 detecting liver injury in a clinical trial depends on its fre- As for all PA-containing herbs, the intrinsic liver injury quency of appearance, which is usually low. The absence is predictable and dose dependent (Figure 3).25 With of liver injury in a study of 3000 patients, which is the more than 6000 herbs containing PAs worldwide,7 this is average patient group exposed to a drug before regula- a general health problem7, 22 not limited to TCM.2, 19, 20 tory approval, provides no guarantee that a rare adverse PA-containing herb intake is now easily recognised by drug effect will not occur, and it only allows to predict biomarkers through chemical analysis.19 an incidence of liver injury of at least 1:1000. For most serious adverse reactions (ADR) with a frequency of at Risk reduction least 1:10 000, it would require 30 000 patients. Thus, Health care providers should inform potential consumers pre-marketing trial group size for a drug is clearly insuf- of herbal TCM about the risk and benefit of the product ficient to detect rare ADR. In addition, the usually short under consideration, especially if evidence for its thera- exposure time also limits the detection of ADR occurring peutic efficacy is limited. In addition, although most with cumulative drug doses. Consequently, efforts to patients recovered from liver injury by herbal TCM improve drug safety should focus on the recognition and products after discontinuation of the suspected products, reporting of HILI by TCM in current users to attain the few experienced acute liver failure, resulting in death or same surveillance standards for ADR frequency, severity, – a liver transplantation (Tables 2 and 3).2, 10 21 To reduce and individual risks of liver injury as for prescription harm, health care providers and individuals considering drugs. Causality assessment must be improved using the taking a herbal TCM product should consult Govern- CIOMS scale to positively identify patients with herbal – ment supported websites,34 37 including those of the hepatotoxicity by TCM products. Also, regulatory mea- Medicines and Healthcare Products Regulatory Agency,34 sures to improve the quality of herbal TCM products the NIH,35 the National Center for Complementary and and the safety of their use should be implemented.9 Alternative Medicine (NCCAM)36 and the National Library of Medicine’s MedlinePlus.37 These websites pro- AUTHORSHIP vide fairly reliable and unbiased information, as opposed Guarantor of article: Rolf Teschke. to hundreds of other websites with information on spe- Author contributions: Rolf Teschke, Albrecht Wolff, cific botanicals and nutritional supplements. The infor- Christian Frenzel and Johannes Schulze wrote the article mation quality varies widely in the other websites, many and contributed to the design of the review, the tables offer anecdotal evidence and unreliable testimonies to and figures. All authors approved the final version of the the benefits, or sell the herbal and dietary supplements. manuscript.

FUTURE DIRECTIONS ACKNOWLEDGEMENT A key question is the evidence for the therapeutic benefit Declaration of personal and funding interests: None. of a specific TCM product in a specific disease. There is

REFERENCES 1. National Institutes of Health (NIH) and Medicines.htm. Accessed March 15, 14218/JCTH.2014.00003 [Epub ahead of LiverTox. Chinese and other Asian 2014. print]. herbal medicines. Last updated January 2. Teschke R. Traditional Chinese 3. Rossi S, Navarro VJ. Herbs and liver 10, 2014. Available at: http://livertox. Medicine induced liver injury. J injury: a clinical perspective. Clin nih.gov/ChineseAndOtherAsianHerbal Clin Translat Hepatol 2014; 2: doi: 10. Gastroenterol Hepatol 2014; doi: 10.

Aliment Pharmacol Ther 2014; 40: 32-50 49 ª 2014 John Wiley & Sons Ltd R. Teschke et al.

1016/j.cgh.2013.07.030 [Epub ahead of induced by Hovenia dulcis. Pediatr a clinical survey and critical analysis print]. Gastroenterol Hepatol Nutr 2012; 2012: of 26 suspected cases. Eur J 4. Abdualmjid RJ, Sergi C. Hepatotoxic 111–6. Gastroenterol Hepatol 2008; 20: 1182– botanicals – an evidenced-based 16. Inoue H, Yamazaki S, Shimizu M, et al. 93. systematic review. J Pharm Pharm Sci Liver injury induced by the Japanese 28. Teschke R, Bahre R. Severe 2013; 16: 376–404. herbal drug kamishoyosan. hepatotoxicity by Indian Ayurvedic 5. Bunchorntavakul C, Reddy KR. Review Gastroenterol Hepatol 2011; 7: 692–5. herbal products: a structured causality article: herbal and dietary supplement 17. Navarro VJ, Bonkovsky HL, Hwang SI, assessment. Ann Hepatol 2009; 8: 258– hepatotoxicity. Aliment Pharmacol Ther Vega M, Barnhart H, Serrano J. 66. 2013; 37:3–17. Catechins in dietary supplements and 29. Teschke R, Glass X, Schulze J, Eickhoff 6. Teschke R, Genthner A, Wolff A, hepatotoxicity. Dig Dis Sci 2013; 58: A. Suspected Greater Celandine Frenzel C, Schulze J, Eickhoff A. Herbal 2682–90. hepatotoxicity: liver specific causality hepatotoxicity: analysis of cases with 18. Jung KA, Min HJ, Yoo SS, et al. Drug- evaluation of published case reports initially reported positive reexposure induced liver injury: twenty five cases from Europe. Eur J Gastroenterol tests. Dig Liv Dis 2014; 46: 264–9. of acute hepatitis following ingestion of Hepatol 2012; 24: 270–80. 7. Larrey D, Faure S. Herbal medicine Polygonum multiflorum Thun. Gut 30. Teschke R, Frenzel C, Glass X, Schulze hepatotoxicity: a new step with Liver 2011; 5: 493–9. J, Eickhoff A. Greater Celandine development of specific biomarkers. J 19. Lin G, Wang JY, Li N, et al. Hepatic hepatotoxicity: a clinical review. Ann Hepatol 2011; 54: 599–601. sinusoidal obstruction syndrome Hepatol 2012; 11: 838–48. 8. Posadzki P, Watson L, Ernst E. associated with consumption of Gynura 31. Teschke R, Schulze J. Suspected herbal 54 – Contamination and adulteration of segetum. J Hepatol 2011; : 666 73. hepatotoxicity: requirements for herbal medicinal products (HMPs): an 20. Gao H, Li N, Wang JY, Zhang SC, Lin appropriate causality assessment by the fi overview of systematic reviews. Eur J G. De nitive diagnosis of hepatic US Pharmacopeia. Drug Saf 2012; 35: 69 – Clin Pharmacol 2013; : 295 307. sinusoidal obstruction syndrome 1091–7. 9. Zhang L, Yan J, Liu X, et al. induced by pyrrolizidine alkaloids. J Dig 32. Izzo AA, Ernst E. Interactions between 13 – Pharmacovigilance practice and risk Dis 2012; :33 9. herbal medicines and prescribed drugs: control of Traditional Chinese Medicine 21. Cohen SM, Heywood E, Pillai A, Ahn J. an updated systematic review. Drugs drugs in China: current status and Hepatotoxicity associated with the use 2009; 69: 1777–98. future perspective. J Ethnopharmacol of White Flood, a nutritional 33. Fasinu PS, Bouic PJ, Rosenkranz B. 140 – 2012; : 519 25. supplement. Pract Gastroenterol 2012; An overview of the evidence and – 10. Cortez E, Boulger C, Bernard A. Ban October issue: 45 8. mechanisms of herb-drug interactions. Tu Wan hepatotoxicity. BMJ Case 22. Teschke R, Wolff A, Frenzel C, Schulze Frontiers Pharmacol 2012; 3: Reports 2012; doi:10.1136/bcr-2012- J, Eickhoff A. Herbal hepatotoxicity: a 1–19. 006438 [Epub ahead of print]. tabular compilation of reported cases. 34. Medicines and Healthcare Products 11. Lee CH, Wang JD, Chen PC. Risk of Liver Int 2012; 32: 1543–56. Regulatory Agency (MHRA). Available liver injury associated with Chinese 23. National Institutes of Health (NIH) and at: http://www.mhra.gov.uk À Choose herbal products containing Radix LiverTox. Drug record. Ba Jiao Lian H for “herbal” topics. Accessed March bupleuri in 639,779 patients with (Dysosma pleianthum). Last updated 15, 2014. hepatitis B virus infection. PLoS ONE September 10, 2013. Available at: http:// 35. National Institutes of Health (NIH). 6 2011; : e16064. livertox.nih.gov/BaJiaoLian.htm. Office of the Director, Office of € 12. Bjornsson ES, Bergmann OM, Accessed March 15, 2014. Dietary Supplements. Available at: € Bjornsson HK, Kvaran RB, Olafsson S. 24. Teschke R, Lebot V. Proposal for a www.ods.od.nih.gov. Accessed March Incidence, presentation and outcomes Kava Quality Standardization Code. 15, 2014. in patients with drug-induced liver Food Chem Toxicol 2011; 49: 2503–16. 36. National Center for Complementary injury in the general population of 25. Teschke R, Schwarzenboeck A, Eickhoff and Alternative Medicine (NCCAM). Iceland. Gastroenterology 2013; 144: A, Frenzel C, Wolff A, Schulze J. Available at: http://nccam.nih.gov/ 1419–25. Clinical and causality assessment in health/whatiscam/chinesemed.htm; 13. Yang L, Aronsohn A, Hart J, Jensen D. herbal hepatotoxicity. Expert Opin Drug www.nccam.nih.gov/health/ Herbal hepatotoxicity from Chinese Saf 2013; 12: 330–66. supplements/wiseuse.htm. Accessed skullcap: a case report. World J Hepatol 26. Teschke R, Wolff A, Frenzel C, March 15, 2014. 2012; 4: 231–3. Schwarzenboeck A, Schulze J, Eickhoff 37. National Library of Medicine’s 14. Dhanasekaran R, Owens V, Sanchez W. A. Drug and herb induced liver injury: MedlinePlus. Available at: www.nlm. Chinese skullcap in move free arthritis council for International Organizations nih.gov/medlineplus/druginfo/herb supplement causes drug induced liver of Medical Sciences scale for causality _All.html. Accessed March 15, injury and pulmonary infiltrates. Case assessment. W J Hepatol 2014; 6: 2014. Reports Hepatol 2013; doi:org/10.1155/ 17–32. 2013/965092 [Epub ahead of print]. 27. Teschke R, Schwarzenboeck A, 15. Kim YJ, Ryu SL, Shim JW, et al. A Hennermann KH. Kava hepatotoxicity: pediatric case of toxic hepatitis

50 Aliment Pharmacol Ther 2014; 40: 32-50 ª 2014 John Wiley & Sons Ltd