Drugs and herbs given to prevent hepatotoxicity of tuberculosis therapy: systematic review of ingredients and evaluation studies

China has a long history of traditional medicine, based on the theory of yin yang and its balance with qi (vital energy) [1]. There are over 100,000 defined traditional Chinese medicine therapies and 80% of these are herbal combinations [2]. To treat tuberculosis (TB), traditional Chinese medicine practitioners typically advocate a combination of biomedical treatment to eliminate bacteria, and traditional medicine to strengthen qi [1], and herbs to protect the liver.

Anti-tuberculosis drugs, including isoniazid, rifampicin and pyrizinamide have hepatotoxic effects. A meta-analysis of studies involving several anti-tuberculosis drug regimens estimates the incidence of liver toxicity is 2.6% with co-administered isoniazid and rifampicin, 1.6% with isoniazid alone, and 1.1% with rifampicin alone [3]; some estimates give the incidence of liver damage among people taking anti-tuberculosis drugs in China to be much higher – at around 15–30% of all patients [4, 5].

The number of people being prescribing 'liver protection drugs' worldwide is not known, but it is generally agreed that it is almost universal for TB patients in China [6]. These drugs are either given to all patients on anti-tuberculosis treatment, or those with some liver function test abnormalities. There is some emerging evidence that the out of pocket costs to patients are high. For example, a recent descriptive study of TB treatment and policy in one municipality found that on top of the TB drugs (which are free) patients were charged expensive fees for liver protection drugs', and this was an important reason why patients interrupted drug taking [6]. Others report the administration of additional non-TB drugs in China, at a higher than necessary cost to patients [7].

There is no obvious biological rationale for these herbal drugs. Yet clinicians and TB specialists in China appear convinced that these various pharmaceutical and herbal preparations may have a protective effects on the liver in people taking anti-TB treatment. However, a primary concern is patient adverse reactions to herbal drugs; some Chinese herbs are known to cause acute hepatitis, and other herbs used in phytomedicine can cause liver toxicity and even liver failure [2]. It is unclear how the effects, and safety, of these drugs are being evaluated in TB patients.

To help open up the debate in this area, we systematically summarised what these agents are as reported in scientific papers that describe evaluations of these liver protection drugs; and critically appraised the evidence base against international standards for appropriate scientific study designs that could justify their use.

We identified 191 research papers (175 from the Chinese databases; 16 from international databases) examining liver protection drugs in TB patients. Overall, 85 research articles reported evaluations of drugs or herbs to prevent drug-induced liver damage in patients taking anti-tuberculosis treatment (we excluded 106 studies of drugs used to treat liver damage). Most were conducted in China (77); other countries were India (2); Russia (4); Ukraine (2).

This is the first attempt that we are aware of to systematically summarise the ingredients of drugs and herbs used to protect the liver in patients taking anti-tuberculosis treatment, and to scrutinise the methods used to evaluate their effects. The literature suggests a large difference in the proportion of patients with hepatotoxicity in China (15–30%) compared to elsewhere (about 2.6%). We believe this is due to varying definitions and understanding of what hepatoxicity is, with a more liberal interpretation in the Chinese context. However, we have not systematically examined the literature to clarify this.

Our study revealed thirty distinct types of liver protection drugs or preparations used in patients with TB, and each contains a complex combination of herbal extracts, vitamins or compounds. Few studies provided a clear biological or clinical rationale for the use of specific herbs or drugs, but we know that some of the herbs identified in our summary have already been studied in China with researchers claiming general hepatoprotective properties, although not in TB patients. For example, Silymarin is known to be widely used as a remedy for liver diseases and glycyrrhizin has been used in chronic viral hepatitis [2, 9]. The numerous ingredients of Chinese herbs and drugs make it more difficult to determine the active ingredients or properties of these preparations or drugs, something we recognised when extracting information from our included studies. It seems unconvincing that so many different preparations all have properties that protect against liver damage by drugs used to treat TB, and we do not know of any convincing biological mechanism. In addition, we do not know the number of studies that were sponsored by pharmaceutical companies as this is not generally reported in Chinese research papers.

Relatively few studies we identified evaluated herbal preparations; most were concerned with the effect of finished manufactured herbal products. Herbal preparations are more complicated to evaluate, and it has been suggested that a conventional randomised controlled and blinded trial design might not be appropriate [8]. Preparations usually have a distinct smell or taste and the herbs used are sometimes blended to suit individual patient needs; all of which preclude blinding of participants and trialists, and use of placebo preparations. For example, one study of the Chinese herbal preparation 'yue hua wan' claimed to be a prospective study that used a random number table to allocate patients to standard anti-tuberculosis treatment plus the herbal preparation or anti-tuberculosis treatment alone [10]. The study is weakened by the lack of a placebo control, but the published report does not describe the preparation in detail, so we cannot determine whether a placebo preparation would have been possible. Studies that fail to use a placebo-control cannot determine the effect of the herb or drug relative to a control treatment and this is less useful for clinical decision making.

Our summary indicates that the evidence base for the use of liver protection drugs in TB patients comprises mainly small, poorly conducted studies that do not reach the standards of trials used in reliable systematic reviews such as those produced by the Cochrane collaboration. Only one study, out of 85 we identified in the international literature, appeared to be a prospective, randomised, placebo controlled trial of a manufactured herbal product [11]. But the trial followed patients for 45 days only, too short a time period to identify liver damage or toxicity; the authors failed to specify particular outcomes a priori; and the study included only 70 patients.

Sample size is a critical consideration in relation to the primary outcome in studies evaluating liver protection drugs-liver damage or toxicity. We have estimated that to detect an effect on an incidence of liver damage of 15–30% in China [4, 5] would require a sample size of 1145 for each trial arm with 95% confidence and 90% power, and if the calculation is based on a global incidence of 1–2% [3], a much larger sample size would be required. It would be impractical and expensive to run trials of this size, and trials of these drugs are not really justified.

Although we classified most of the studies we identified from the Chinese literature as 'prospective', only three appeared to randomly allocate participants to groups, and these were not placebo controlled. Eighteen studies appeared to be retrospective, and used patient records to examine outcomes historically. Generally we found it difficult to determine from the short full text journal reports which particular observational design the authors had followed. Most studies classified as retrospective appeared to be case reports or case series. The reports generally did not describe the research procedures in detail, and it was difficult to determine which studies were case-controlled.

An important finding is the lack of attention in the research papers to adverse events associated with the drugs or herbs used. Some studies mentioned adverse events, but did not describe methods used routinely to monitor patients for events; a large number of the Chinese studies did not mention adverse events at all. Given the existing concerns over the safety of some herbs in use [2], it is important that studies include methods to routinely assess participants for adverse events; that samples are large enough; and studies follow patients over a sufficient period of time to detect infrequent events.

Other authors report poor methodological quality of studies evaluating herbal preparations; often studies do not adequately randomise participants, do not specify outcome measures or end points, contain small sample sizes, and suffer from publication bias [2]. WHO provide guidelines for good practice in conducting clinical trials of herbal medicine, and recommend the design of trials be adapted to the peculiarities of herbal medicines [8].

In terms of policy implications, this review indicates there is no reliable evidence to support taking liver protection drugs with TB treatment. Indeed some may do harm. Thus there is no medical reason to continue prescribing them; and, given the wide variety of compounds, many of which may have side-effects, it is more likely that they will do more harm than good. Understanding of what liver toxicity is and criteria for diagnosis are probably also different which further complicates any comparison.

In terms of research, it could be argued that trials following accepted international standards for conduct and reporting are required. However, given the poor rationale for using the drugs and large sample size required, then the reasons for doing a proper RCT will be limited. The one justification could relate to the fact that they are widely prescribed, so identifying no effect, or a harmful effect, could potentially be a powerful influence on current policies.