Introduction

Herbal and dietary supplements (HDS) are commonly consumed in the US [1]. The current regulatory environment requires manufacturers to disclose the ingredients of products on the label, adhere to good manufacturing practices, and to report serious adverse events when they are made aware of such occurrences [2]. However adulteration of HDS is commonplace, as has been reviewed elsewhere [3]. Although products can be removed from the marketplace if there is concern about toxicity, the precise mechanism of injury or the ingredient within the product responsible for injury remains largely unknown. Even in the unusual circumstance of careful analysis of ingredients, a causal association between the suspected ingredient and the injury can remain elusive [4].

Green tea extract (GTE) is a frequent ingredient in a variety of HDS. It is marketed for health promotion and for its properties as an antioxidant and weight-reducing agent. Catechins (CA) are polyphenolic flavonoid compounds that are contained in abundance in GTE. The major catechins include epicatechin (EC), epigallocatechin (EGC), epicatechin-3-gallate (ECG), and epigallocatechin-3-gallate (EGCG) [5]. Much is known about the animal and human pharmacology and toxicology of GTE and EGCG [6–12]. Importantly, several clinical reports of hepatotoxicity attributable to GTE and its component catechins are available in the literature [13–17].

Because GTE is such a common ingredient in dietary supplements, we hypothesized that catechins might sometimes be present in HDS that are associated with hepatotoxicity even when not identified among the ingredients on the package label. Therefore, our primary objective was to determine catechin concentrations in HDS that were implicated in hepatotoxicity among patients enrolled in the US Drug-Induced Liver Injury Network (DILIN). As a secondary objective, we investigated the relationship between the amount of catechin consumed and liver injury among confirmed cases of hepatotoxicity.

Methods

The Drug-Induced Liver Injury Network (DILIN) and the Study Population

In 2003, the DILIN was established by the National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK) to collect and examine cases of bona fide non-acetaminophen drug and dietary supplement-induced hepatotoxicity. The network began as five centers in 2003, and expanded to eight in 2008 [18]. Our study population was taken from the cohort of patients who presented to a DILIN investigator within six months of a drug or dietary supplement-induced injury. Details of eligibility, data collection, and causality-assessment procedures have been described elsewhere [19]. Confirmed cases of DILI are those in which the likelihood of a causal association between the liver injury and an implicated agent is either definite, highly likely, or probable, corresponding to probabilities of causality of 95 % or greater, 75–94 %, and 50–74 %, respectively.

Each case of DILI is given a severity grade. These grades include mild (elevated liver enzymes with total bilirubin <2.5 mg/dL and INR <1.5), moderate (elevated liver enzymes with total bilirubin >2.5 mg/dL or INR >1.5), moderate-hospitalized, severe (fulfills moderate criteria and also has signs of hepatic failure or organ failure related to the DILI), and fatal (patient dies or undergoes transplantation).

Herbal Dietary Supplements

To facilitate studies of hepatotoxicity attributable to HDS, the DILIN established a repository for dietary supplements that were implicated in hepatotoxicity [20]. Products were obtained from subjects enrolled at any DILIN site and entered into the repository. Herbal and dietary supplements were assayed if consumed by any subject who fulfilled criteria for entry into the DILIN between 2004 and 2010. HDS were classified according to the intended purpose for use, as ascertained by patient reporting and/or packaging and marketing materials. The investigators were unaware of the original labeled ingredients of samples submitted for catechin assay.

Catechin Assays

Catechins were extracted and quantified as described elsewhere [21], and as detailed in the supplemental on-line material.

Statistical Analysis

Frequency and percentage are used to describe categorical data and mean and standard deviation are used to describe continuous data. The chi-squared test was used to test the association between two categorical variables.

Results

Characteristics of the Study Population

Forty-seven DILIN subjects consumed a total of 97 products that were available for analysis. The mean age of the patients was 44 years; 51 % were female. Of the 47 subjects, 38 (81 %) had confirmed DILI; for the other nine patients liver injury was deemed to be possible (causality score 25–50 %; five patients), unlikely (causality score <25 %; three patients), or was not yet determined (one patient). Age and gender distributions differed slightly from those for DILIN cases attributed to non-HDS (i.e., prescribed medications)—the non-HDS cases were significantly older (mean age 50 years; P = 0.01) and the proportion of females was slightly greater (60 %; P = 0.24).

Catechin Profiles

Of the 97 products assayed for catechins, 49 (50.5 %) contained at least one catechin. For seventy-three products no GTE or catechins were identified on the label yet for 29 (39.7 %) of these products, listed in Table 1, catechins were detectable. The highest concentrations of individual catechins were for EGCG and GCG, with CA and EGCG being the most commonly found, in 18 and 16, respectively, of the 29. Eighteen products, also listed in Table 1, identified GTE or one its component catechins on the label; of these, all but two contained catechins and EGCG tended to be the most abundant. The remaining six products had no label; half contained total catechin at low concentrations (<50 mcg/g product; data not shown). Most of the catechins in this group comprised EGCG, CA, and EC.

Table 1 Catechin profiles
Full size table

Table 2 shows the accuracy of catechin labeling for each of the categories of HDS. Of the catechin-containing products within each category, bodybuilding supplements comprised the largest single group, followed by products marketed for weight loss, as multivitamins, for immune support, and as Chinese herbs. Among all categories that did not identify catechins or GTE on the product labels, more than half contained catechins, with frequencies ranging from 29 to 100 % of products within these categories. Interestingly, even when labeled as containing catechin or GTE, they were not present uniformly, except in weight-loss products; specifically, 6 of 18 contained either no or negligible catechin.

Table 2 Categories of 97 assayed HDS: accuracy of labeling
Full size table

Relationship Between Catechin Consumption and Hepatotoxicity in Confirmed DILI Cases

We assessed associations between categories of catechins and DILI causality score, clinical patterns of disease, and disease severity. For this analysis, we focused on the 38 patients with confirmed DILI. We compared the 26 patients who consumed catechin-containing HDS (ranging from 1 to 8 supplements consumed per patient) with the 12 for whom the HDS implicated in the injury did not contain catechin or GTE. As shown in Table 3, the DILI causality assessment scores, clinical patterns of injury, and disease severity were no different.

Table 3 Clinical characteristics of enrolled cases
Full size table

We had sufficient information on dose of supplement consumed to assess the relationship between the catechin dose (calculated from catechin concentration in the supplements) and liver injury for 19 patients with confirmed DILI. As shown in Table 4, we found no correlation between daily or total catechin dose and causality score, peak liver enzyme values, or disease severity; for three patients the total dose of catechin consumed was unavailable because of incomplete patient reporting. Catechin concentrations were highest in weight-loss products; however, the daily catechin dose was small, with the highest estimated daily dose being 40 mg/kg in a patient who consumed HDS-066.

Table 4 Clinical characteristics of patients for whom catechin consumption could be quantified
Full size table

Discussion

In this study, we hypothesized that catechins were commonly present as unidentified ingredients in HDS that had been implicated in human hepatotoxicity. Indeed, we found that among HDS not listing green tea as ingredients on their labels nearly 40 % contained catechins. The most abundant species were EGCG, CA, and EC, with EGCG being the most frequent among products failing to identify GTE or catechins on the label. By contrast, we also found that some products labeled as containing GTE or its component catechins actually contained no detectable catechins. Therefore, packaging and labels of HDS seem to be unreliable as regards GTE and catechin content.

Our secondary objective was to investigate the relationship between catechins and liver injury. The rationale for this analysis arises from case reports of hepatotoxicity attributed to GTE. This assessment relied on matching the clinical histories with the analysis for catechins. We could find no convincing relationship between catechin dose and causal likelihood score, severity, or type of liver injury. Notwithstanding these findings, a non-statistically significant tendency toward fatal liver injury was associated with catechin-containing HDS; a larger sample size would be required to further investigate a potential association. GTE is a common ingredient in several HDS that have been withdrawn from the market because of safety concerns [22, 23]. Case series and a systematic review by the United States Pharmacopeia revealed evidence of GTE’s hepatotoxicity [24]. Since 1966, at least 216 cases of toxicity attributed to green tea extracts have been reported. Doses in case reports that involved hepatic injury ranged from 0.7 to 3 g per day [24]. Most of the cases presented with an acute hepatocellular injury pattern and most recovered on cessation of use [13–17, 24]. In most case reports, it was unclear whether the toxicity was because of the GTE per se or was related to chemicals introduced during the extraction process, to concomitant medications, or to other herbs in the supplements. On the basis of the review by the US Pharmacopeia it was concluded that, when HDS containing green tea extracts are formulated and used appropriately, the potential for significant safety issues should be low [2, 24]. These clinical data must be viewed in the context of work by Lambert et al. [25] who found EGCG, the most abundant catechin in GTE, to be a dose-dependent hepatotoxin in mice.

The usual doses of GTE required to lead to hepatotoxicity in humans are not clearly established. They are likely to be quite variable, depending upon nutritional, genetic, and other factors. A review of reported cases in humans in 2006 indicated that cumulative doses of GTE from as little as 5.9 g over 5 days to maximum of 240 g over 120 days may be harmful, emphasizing the potentially wide range of toxic doses [15].

Herbal dietary supplements (HDS) were implicated in 10 % of cases of human drug-induced hepatotoxicity in the DILIN [18]. In fact, HDS were the second largest single group of implicated agents among all potential culprits, with antibiotics being the most commonly implicated group. It is, however, the hardest group of agents to implicate for many reasons. The myriad available with, often, several implicated products, batch-to-batch and product-to product-variability, the potential for interaction among ingredients within a product or with other medications, and the possibility of contamination frequently confound attribution of injury to any one product or ingredient.

Our findings support the assertion that herbal dietary supplement labels are unreliable, a consideration that is germane when evaluating hepatotoxicity attributable to HDS, because the inclusion of unidentified dietary ingredients or other adulterants in HDS confounds the process of causality assessment. Most adulterants in dietary supplements are pharmaceuticals and heavy metals. Bacterial contamination has also been implicated in hepatotoxicity from HDS [26, 27]. Despite the purported benefits of GTE and its component catechins [28, 29], inclusion of these compounds may also be regarded as adulteration, given the clinical and experimental support for their hepatotoxic potential.

Our study highlights an important question that deserves further investigation; namely, were GTE or its component catechins involved in liver injury in these cases? We cannot exclude the possibility that another adulterant was the cause of the injury, particularly because the concentrations of catechins were usually low and the doses lower than those found to be toxic to animals. Given the low catechin doses, and the widespread use of these compounds in teas and extracts, it is also possible that toxicity may be idiosyncratic, i.e. non-dose dependent in nature. These considerations emphasize the need for additional research on this topic to confirm or refute the as yet empirical evidence linking GTE and its component catechins to hepatotoxicity.