Chapter  21:  Milk Thistle: Effects on Liver Disease and Cirrhosis and Clinical Adverse Effects: Evidence Report/Technology Assessment Number 21

Search Strategy

Eleven electronic databases, including AMED, CISCOM, and the Cochrane Library (including DARE and the Cochrane Controlled Trials Registry), EMBASE, MEDLINE, and NAPRALERT, were searched through July 1999 using the following terms: carduus marianus, legalon, mariendistel, milk thistle, silybin, silybum marianum, silybum, silychristin, silydianin, and silymarin. An update search limited to PubMed was conducted in December 1999. English and non-English citations were identified from these electronic databases, references in pertinent articles and reviews, drug manufacturers, and technical experts.

Selection Criteria

Preliminary selection criteria regarding efficacy were reports on liver disease and clinical and physiologic outcomes from randomized controlled trials (RCTs) in humans comparing milk thistle with placebo, no milk thistle, or another active agent. Several of these randomized trials had dissimilar numbers of subjects in study arms, raising the question that these were not actually RCTs but cohort studies. In addition, among studies using nonplacebo controls, the type of control varied widely. Therefore, qualitative and quantitative syntheses of data on effectiveness were limited to placebo-controlled studies. For adverse effects, all types of studies in humans were used to assess adverse clinical effects.

Data Collection and Analysis

Abstractors (physicians, methodologists, pharmacists, and a nurse) independently abstracted data from trials; a nurse and physician abstracted data about adverse effects. Data were synthesized descriptively, emphasizing methodologic characteristics of the studies, such as populations enrolled, definitions of selection and outcome criteria, sample sizes, adequacy of randomization process, interventions and comparisons, cointerventions, biases in outcome assessment, and study designs. Evidence tables and graphic summaries, such as funnel plots, Galbraith plots, and forest plots, were used to examine relationships between clinical outcomes, participant characteristics, and methodologic characteristics. Trial outcomes were examined quantitatively in exploratory meta-analyses that used standardized mean differences between mean change scores as the effect size measure.

Mechanisms of Action

Evidence exists that milk thistle may be hepatoprotective through a number of mechanisms: antioxidant activity, toxin blockade at the membrane level, enhanced protein synthesis, antifibriotic activity, and possible anti-inflammatory or immunomodulating effects.

Preparations of Milk Thistle

The largest producer of milk thistle is Madaus (Germany), which makes an extract of concentrated silymarin. However, numerous other extracts exist, and more information is needed on comparability of formulations, standardization, and bioavailability for studies of mechanisms of action and clinical trials.

Benefit of Milk Thistle for Liver Disease

• Sixteen prospective trials were identified. Fourteen were randomized, blinded, placebo-controlled studies of milk thistle's effectiveness in a variety of liver diseases. In one additional placebo-controlled trial, blinding or randomization was not clear, and one placebo-controlled study was a cohort study with a placebo comparison group.

• Seventeen additional trials used nonplacebo controls; two other trials studied milk thistle as prophylaxis in patients with no known liver disease who were starting potentially hepatotoxic drugs. The identified studies addressed alcohol-related liver disease, toxin-induced liver disease, and viral liver disease. No studies were found that evaluated milk thistle for cholestatic liver disease or primary hepatic malignancy (hepatocellular carcinoma, cholangiocarcinoma).

• There were problems in assessing the evidence because of incomplete information about multiple methodologic issues, including etiology and severity of liver disease, study design, subject characteristics, and potential confounders. It is difficult to say if the lack of information reflects poor scientific quality of study methods or poor reporting quality or both.

• Detailed data evaluation and syntheses were limited to the 16 placebo-controlled studies. Distribution of durations of therapy across trials was wide (7 days to 2 years), inconsistent, and sometimes not given. Eleven studies used Legalon^®, and eight of those used the same dose. Outcome measures varied among studies, as did duration of therapy and the followup for which outcome measures were reported.

• Among six studies of milk thistle and chronic alcoholic liver disease, four reported significant improvement in at least one measurement of liver function (i.e., aminotransferases, albumin, and/or malondialdehyde) or histologic findings with milk thistle compared with placebo, but also reported no difference between groups for other outcome measures.

• Available data were insufficient to sort six studies into specific etiologic categories; these were grouped as chronic liver disease of mixed etiologies. In three of the six studies that reported multiple outcome measures, at least one outcome measure improved significantly with milk thistle compared with placebo, but there were no differences between milk thistle and placebo for one or more of the other outcome measures in each study. Two studies indicated a possible survival benefit.

• Three placebo-controlled studies evaluated milk thistle for viral hepatitis. The one acute viral hepatitis study reported latest outcome measures at 28 days and showed significant improvement in aspartate aminotransferase and bilirubin. The two studies of chronic viral hepatitis differed markedly in duration of therapy (7 days and 1 year). The shorter study showed improvement in aminotransferases for milk thistle compared with placebo but not other laboratory measures. In the longer study, milk thistle was associated with a nonsignificant trend toward histologic improvement, the only outcome measure reported.

• Two trials included patients with alcoholic or nonalcoholic cirrhosis. The milk thistle arms showed a trend toward improved survival in one trial and significantly improved survival for subgroups with alcoholic cirrhosis or Child's Group A severity. The second study reported no significant improvement in laboratory measures and survival for other clinical subgroups, but no data were given.

• Two trials specifically studied patients with alcoholic cirrhosis. Duration of therapy was unclear in the first, which reported no improvement in laboratory measures of liver function, hepatomegaly, jaundice, ascites, or survival. However, there were nonsignificant trends favoring milk thistle in incidence of encephalopathy and gastrointestinal bleeding and in survival for subjects with concomitant hepatitis C. The second study, after treatment for 30 days, reported significant improvements in aminotransferases but not bilirubin for milk thistle compared with placebo.

• Three trials evaluated milk thistle in the setting of hepatotoxic drugs: one for therapeutic use and two for prophylaxis with milk thistle. Results were mixed among the three trials.

• Exploratory meta-analyses generally showed positive but small and nonsignificant effect sizes and a sprinkling of significant positive effects.

• No studies were identified regarding milk thistle and cholestatic liver disease or primary hepatic malignancy.

• Available evidence does not establish whether effectiveness of milk thistle varies across preparations. One Phase II trial suggested that effectiveness may vary with dose of milk thistle.

Adverse Effects

Adverse effects associated with oral ingestion of milk thistle include gastrointestinal problems (e.g., nausea, diarrhea, dyspepsia, flatulence, abdominal bloating, abdominal fullness or pain, anorexia, and changes in bowel habits), headache, skin reactions (pruritus, rash, urticaria, and eczema), neuropsychological events (e.g., asthenia, malaise, and insomnia), arthralgia, rhinoconjunctivitis, impotence, and anaphylaxis. However, causality is rarely addressed in available reports. For randomized trials reporting adverse effects, incidence was approximately equal in milk thistle and control groups.

Conclusions

Clinical efficacy of milk thistle is not clearly established. Interpretation of the evidence is hampered by poor study methods and/or poor quality of reporting in publications. Problems in study design include heterogeneity in etiology and extent of liver disease, small sample sizes, and variation in formulation, dosing, and duration of milk thistle therapy. Possible benefit has been shown most frequently, but not consistently, for improvement in aminotransferases and liver function tests are overwhelmingly the most common outcome measure studied. Survival and other clinical outcome measures have been studied least often, with both positive and negative findings. Available evidence is not sufficient to suggest whether milk thistle may be more effective for some liver diseases than others or if effectiveness might be related to duration of therapy or chronicity and severity of liver disease. Regarding adverse effects, little evidence is available regarding causality, but available evidence does suggest that milk thistle is associated with few, and generally minor, adverse effects.

Despite substantial in vitro and animal research, the mechanism of action of milk thistle is not fully defined and may be multifactorial. A systematic review of this evidence to clarify what is known and identify gaps in knowledge would be important to guide design of future studies of the mechanisms of milk thistle and clinical trials.

The Plant

Milk thistle is a tall plant that can grow up to 10 feet with thorny stems, dark glossy green leaves, and milky-white veins running throughout. Its most distinguishing feature is the large, bright purple flower sprouting at the top.^{3 7 8}

Historical Uses of Milk Thistle

For centuries, nearly every part of the plant-from root to hull-has been used in some way.^{4 6} In her famous book, Maud Grieve explained several ways the milk thistle can be eaten, including the heads like an artichoke, raw stalks (which are considered palatable and nutritious), and the leaves as a salad.⁸ She also quoted Bryant, who wrote in his Flora Dietetica, "The young shoots in the Spring, cut close to the root with part of the stalk on, is one of the best boiling salads that is eaten, and surpasses the finest cabbage. They were sometimes baked in pies. The roots may be eaten like those of Salsify."⁸

Milk thistle has also historically been used for animal feed. Grieve wrote that in parts of England, the leaves are called "pig leaves" because pigs liked them; also, the seeds are a favorite food of goldfinches.⁸ In Scotland, the leaves were used extensively as food for cattle and horses (the leaves were beaten and crushed to rid them of prickles) before the introduction of special green crops.⁸ The milk thistle prickles have also been used historically as a substitute for barbed wire.⁹ Despite this wide spectrum of perceived value, farmers considered it and other thistles a sign of untidiness and neglect.⁸

Most sources on milk thistle, which is native to southern Europe (specifically Mediterranean areas) and Asia,⁵ put its beginning at 2,000 years ago.^{1 2 6 9} One of the earliest mentions of thistles in general is in the Bible (Genesis 3:18). In this verse, God told Adam and Eve when they were banished from the Garden of Eden that "thorns also and thistles shall it bring forth to thee."

Some of the earliest people to use and write about milk thistle were ancient Greek and Roman physicians and herbalists, each of whom seemed to have their own name for the herb. Dioscorides called it "sillybon," Pliny the Elder called it "sillybum," and Theophrastus called it "pternix."⁵ Dioscorides' use of milk thistle is one of the oldest known references to the medicinal use of this plant. He suggested preparing it in a tea "for those that be bitten of serpents."¹⁰ Another famous ancient herbalist, Pliny the Elder, wrote that mixing the juice of the plant with honey was good for "carrying off bile."^{1 9}

Milk thistle is mentioned in several works from the Middle Ages, one of the first being in a record of old Saxon remedies, which claimed that "this wort if hung upon a man's neck it setteth snakes to flight."⁸ One of the best known herbalists from this time, John Gerard (1545-1612), recommended milk thistle for expelling melancholy and its related diseases (melancholy diseases were described in medicine in the Middle Ages as related to the liver and also called "black bile").^{8 9 10} Another well known English herbalist, Nicholas Culpeper (1616-54), recommended milk thistle for several maladies: breaking and expelling stones, removing obstructions of the liver and spleen, treating jaundice and infections of the plague, and cleansing the blood.^{8 9}

Maud Grieve, who compiled her book of herbal information in 1931, quoted several early English herbalists on their love for and use of milk thistle. She reports that in 1694 Westmacott wrote of milk thistle, "It is a friend to the liver and blood: the prickles cut off, they were formerly used to be boiled in the spring and eaten with other herbs; but as the World decays, so does the use of good old things and other more delicate and less virtuous brought in."⁸ She also reports that John Evelyn wrote, "Disarmed of its prickles and boiled it is worthy of esteem, and thought to be a great breeder of milk and proper diet for women who are nursing."⁸

Although milk thistle's use during the 17th century is most often associated with English herbalists, many monasteries cultivated and used the plant for medicinal purposes as well. St. Hildegard von Bingen (1098-1179) recommended the roots, herbs, and leaves for swelling and erysipelas.⁵

Milk thistle was popular with German herbalists and scientists, also. Otto Brunfels (1488-1534), Hieronimous Bock (1498-1554), Jacob Theodorus (1520-90), Adam Lonicerus (1528-86), and Pietro Andrea Mattioli (1501-77) all recommended milk thistle for treating liver diseases.^{5 9} Another German physician from the 19th century, Johannes Gottfried Rademacher, developed a tincture made from milk thistle seeds for his liver patients.^{1 4} Rademacher's Tincture is an ethanol extract from the seeds used for hepatosplenic disorders.⁹

In the United States, milk thistle enjoyed popularity in the 19th century with the Eclectics movement, an officially recognized branch of North American medicine that predominantly used Native American herbs. The Eclectics used milk thistle for varicose veins, menstrual difficulty, and congestion of the liver, spleen, and kidneys.^{1 7} Milk thistle was also used as part of the naturopathic medical tradition and Native American medical practices.¹⁰ So popular was milk thistle that a tincture of the whole plant was listed in the first United States Homeopathic Pharmacopoeia.^{1 2}

Available history does not seem to tell us how (i.e., by what anecdotal or empirical evidence) milk thistle came to be advised for liver and gallbladder problems. Although milk thistle is most often associated with treating liver disorders, physicians have tried to apply its curative properties to other ailments, including stimulating breast-milk production and bile secretion, treating depression, and protecting against the poisonous mushroom Amanita phalliodesand other environmental toxins.⁴

The Milk Thistle Industry

Milk thistle enjoys significant popularity in the current herbal industry. The World Health Organization estimates that 4 billion people (nearly two-thirds of persons in developing countries) use herbal medicine for some aspect of health care. Herbal medicine is a major component in all indigenous peoples' traditional medicine and a common element in Ayurvedic, homeopathic, naturopathic, traditional oriental, Chinese, and Native American Indian medicine.¹¹

Milk thistle has been available in Europe since 1969, and more than $180 million worth of products were sold in one recent year.¹⁰ In Germany alone, milk thistle was the 11th most frequently prescribed monopreparation herbal, with $16 million (U.S. dollars) in sales in 1996.¹²

In the United States, the herbal market was estimated at about $1.6 billion in 1994 with some projections reaching about $3.5 billion in 1997. Of the 14 top-selling herbal supplements in the United States in 1997 in food, drug, and mass-market retail outlets, milk thistle ranked 13th (more than $3 million).¹² However, it is difficult to determine the actual size of the herbal market because herbal products were formerly sold mostly through channels of distribution normally not subject to econometric tracking services (e.g., health food stores, mail order, multilevel marketing organizations).¹²

Attesting to milk thistle's popularity in this country, a recent poll of patients attending a hepatology clinic at Oregon Health Sciences University found that 31 percent of patients were using over-the-counter "alternative agents" for the therapy of their liver disease. The most commonly used nontraditional therapy was milk thistle, and over 50 percent of those patients felt they had experienced subjective improvement in their symptoms.¹³ In the United States, milk thistle is now being tested for safety by the U.S. National Toxicology Program, along with aloe vera, ginseng, and kava.¹⁴

Antioxidant Activity

Silymarin is thought to have antioxidant activity in the liver, as well as the small intestine and stomach. As an antioxidant, this compound may reduce free radical production and lipid peroxidation in the setting of hepatotoxicity. Lipid peroxidation is the end result of unstable free radicals' damage to membrane lipids. These membranes contain fatty acids that are transformed to lipoperoxidases, peroxides, and lipidic hydroperoxides. Malondialdehyde is a biproduct of phospholipid turnover and linoleic acid and is frequently used in clinical and in vitro studies as a surrogate marker for oxidation activity. Multiple in vitro studies have demonstrated lipid peroxidation inhibition using malondialdehyde as a marker in rat hepatic microsomes and mitochondria.^{35 36 37 38} Furthermore, the phenolic conformation of silymarin is thought to permit the formation of stable compounds from hydroxylic and oxygen radicals.^{39 40} Primary defenses against oxidation or free radical production include glutathione, catalase, and superoxide dismutase. In the setting of an acute toxic event, if stores of these compounds are depleted in intracellular or extracellular (sinusoidal) compartments, oxidative injury is unimpeded.^{41 42 43}

The phenol structure of silymarin led investigators to postulate that silymarin might have potential activity as an antioxidant. In vivo studies in rats indicate that silymarin can reduce the free radical load. One study exposed rats to acetaminophen at toxic doses, and then measured levels of reduced glutathione and superoxide dismutase in experimental and control rats. In another study, mice exposed to acetaminophen at toxic doses had increased levels of reduced glutathione and superoxide dismutase when treated with silymarin compared with the levels in controls.^{41 42} Another study demonstrated preserved hepatic glutathione stores and improved reduced:oxidized glutathione ratios in rats exposed to acetaminophen and ethyl alcohol at high doses when compared with those in controls.⁴² Similarly, in humans, investigators have demonstrated increases in serum glutathione peroxidase and erythrocyte and lymphocyte superoxide dismutase.⁴⁴

Silymarin may also protect hepatocyte-lipid membranes. Studies demonstrated that silymarin can inhibit cell lysis as measured by changes in alanine aminotransferase levels when exposing isolated hepatocytes to carbon tetrachloride and galactosamine.⁴⁵

Toxin Blockade

Studies demonstrated improvements in cytosol liver and histologic markers in animals receiving silymarin compared with those in controls for an array of hepatotoxins including carbon tetrachloride, galactosamine, thioacetamide, ethanol, and acetaminophen.⁵ The mechanism of action is thought to be mediated by competitive inhibition, membrane stabilization, and antioxidant activity. However, in many studies the mechanism was not clearly identified. Silymarin can bind to liver cell membrane receptors to protect cells from toxins. One study demonstrated this effect using the death cap mushroom, Amanita phalloides. The toxins in this fungus are amanititin and phalloidin. Several studies showed that silymarin competes with the toxin for cell membrane receptor sites, thus reducing the effect of the toxin.^{46 47}

Enhanced Protein Synthesis

Regeneration of hepatocytes is necessary for hepatic recovery from acute or chronic insults. In chronic injury, fibrosis occurs simultaneously with regeneration; the ultimate outcome is determined by which process dominates. Several studies identified mechanisms through which silybin may facilitate hepatocyte regeneration. In several rat studies, silybin appeared to stimulate ribonucleic acid (RNA) polymerase I and ribosomal RNA.^{48 49} This effect leads to more rapid formation of ribosomes, which in turn increases protein synthesis. The exact mechanism of how RNA polymerase I is stimulated is unclear. One study demonstrated that silymarin binds to a steroid receptor,⁴⁸ and it is hypothesized that structural similarity with steroids permits binding. Silymarin then, like steroids, may be able to modulate RNA synthesis. Additionally, one study in rats suggested that silymarin can also enhance deoxyribonucleic acid synthesis and, therefore, possibly enhance hepatocyte regeneration.⁵⁰ Thus far, one study demonstrated hepatocyte regeneration in rats with silymarin.⁴⁹

Antifibrotic Activity

To date, the evidence for antifibrotic activity comes largely from animal studies. Reportedly, human trials are in progress with Legalon^® that are examining antifibrotic activity. According to Madaus, Legalon^® administered orally in a rat biliary fibrosis model reduced hepatic collagen accumulation and levels of a serum marker for fibrosis. Another study reportedly slowed down regeneration of procollagen RNA by silymarin in rat livers.⁵¹

Other Postulated Mechanisms

Milk thistle reportedly reduced leukotriene formation through noncompetitive inhibition of lipoxygenase, thereby suggesting possible anti-inflammatory effects.⁵²

Sources and Search Methods

Table 2. Electronic sources searched

Table 2. Electronic sources searched

Electronic Database	Description
AMED (Alternative and Allied Medicine Database) Searched from 1985 to October 1998	This database contains 100,000 references from 400 journals on alternative and complementary medicine going back to 1985.
CINAHL (Cumulative Index to Nursing and Allied Health Literature Searched from 1984 to July 1999	This database includes citations from over 500 biomedical and popular sources, including National League of Nursing and American Nurses Association publications, covers publications from 1982 to the present, and is considered the premier nursing database.
CISCOM (Centralised Information Service for Complementary Medicine) Searched 1968 to July 1999	This database contains over 34,000 references and combines data from MEDLINE, AMED, and other specialist European databases.
Cochrane Library (http://www.cochrane.org ) DARE (Database of Reviews of Effectiveness) (http://www.york.ac.uk/inst/crd/ ) The Cochrane Controlled Trials Registry Searched Issue 2 1999	These databases contain references of randomized controlled trials and systematic reviews identified from electronic bibliographic sources and hand searching of multiple journals and symposia or meeting proceedings.
Dissertation Abstracts Searched from 1961 to December 1998	This library indexes doctoral dissertations and masters abstracts from more than 1,000 institutions.
EMBASE Searched from 1988 to July 1999	This database contains biomedical and pharmaceutical citations and is considered the premier biomedical database in Europe.
MEDLINE (PubMed) Searched from 1966 to July 1999PubMed searched July 1999 to November 1999	These databases (MEDLINE and PubMed) index almost 4,000 international biomedical journals from 1966 to the present, includes references from Index Medicus, International Nursing Index, and Index to Dental Literature, and is considered the premiere biomedical database in the United States.
MICROMEDEX contains: DRUGDEX Product Index (drug ingredients) POISONDEX/IDENTIDEX (toxicology) DRUG-REAX (interactive drug interactions) Searched July 1999	This database provides a major source for drug, poison, and acute care information.
NAPRALERT (Natural Products Alert) Searched September 1999	This database contains records from 1650 to the present on natural products, including the pharmacology, biologic activity, taxonomic distribution, ethno-medicine and chemistry of plant, microbial, and animal extracts.
PHYTODOK (German Database) Searched 1995 to July 1999	This database contains 8,800 references from approximately 300 journals worldwide on toxicology, pharmacology, and therapeutic uses for natural compounds and on isolation of natural compounds from plant material.
Science Citation Index Searched from January 1990 to March 1999	This index covers 4,400 scientific and 1,400 social science journals worldwide, together with selected coverage of related material.

After these materials were reviewed and relevant studies obtained and abstracted, an updated search to November 1, 1999, was conducted using MEDLINE and EMBASE.

Selection Processes

Figure 2. Selection process

   Figure 2. Selection process

Figure 2

Initially, we planned to limit efficacy evidence to randomized controlled trials (RCTs) comparing milk thistle with placebo, no milk thistle, or another active agent. Ultimately, we included evidence from prospective placebo-controlled trials or cohort studies for several reasons. First, there were scant data, and it was thought that evidence from studies other than randomized trials might provide useful preliminary information. Second, several reportedly "randomized" trials had dissimilar numbers of subjects among the study arms, raising the possibility that they were not randomized and not of significantly different quality than other prospective controlled studies. The search for evidence was not repeated at the point that selection criteria were broadened, because the search had been designed to detect all studies of milk thistle regardless of their design.

Unpublished Data

For reports published as abstracts, we excluded those for which we were unable to identify a complete subsequent publication by a repeat search of MEDLINE and EMBASE for any of the abstract authors. When published studies met selection criteria but did not report important design features or outcome data, this unpublished information was requested from the authors.

Data Analysis Process

Data were synthesized descriptively, emphasizing methodologic characteristics of the studies, such as populations enrolled, definitions of selection and outcome criteria, sample sizes, adequacy of randomization process, interventions and comparisons, cointerventions, biases in outcome assessment or intervention administration, and study designs. Relationships between clinical outcomes, participant characteristics, and methodologic characteristics are presented in evidence tables and graphic summaries such as forest plots.

Primary outcomes in studies were measured with continuous rather than categorical variables. We used the standardized mean differences between treatment and comparison group scores as the effect size measure for each study. Hedges' g was used to compute the standardized mean difference for each trial:

where, for a given trial are the mean clinical outcome scores for the treatment group and comparison group, respectively; s _pooled is the pooled standard deviation for the difference between the two means.⁵⁷ These estimates were adjusted for between-group differences at baseline and for small sample bias.⁵⁷ Adjustment for baseline differences was accomplished by calculating an "effect size" at baseline; by definition, it should be zero if study groups were well matched. When a nonzero "effect size" at baseline was found, outcome effect sizes were adjusted by subtracting the baseline effect size.

Published reports seldom provided estimates of S _pooled One of three strategies was used to estimate S _pooled when the authors did not directly provide it. First, the individual group variances were used to estimate S _pooled If these data were not reported, the pooled variance was back-calculated from either the test statistic or the p value for differences at followup.⁵⁸ If neither was possible, a mean variance derived from studies of similar size was used. Studies in which the pooled variance was calculated using either of the two latter methods were flagged in the event the magnitude of the effect size resulted in the study being identified as a potential outlier in the analysis of heterogeneity.

Exploratory Meta-Analysis

Meta-analysis was used as an exploratory tool to help identify patterns of findings. Prospective placebo-controlled randomized trials using albumin, bilirubin, aspartate aminotransferase (AST), alanine aminotransferase (ALT), gammaglutamyl transpeptidase (GGTP), malondialdehyde (MDA), alkaline phosphatase (alk phos), prothrombin time (PT), Child's score (a score or classification system for chronic liver disease), and histologic and survival outcomes were quantitatively pooled by clinical outcome using meta-analysis. Subgroup analyses were conducted for trials that included patients with the following: chronic alcoholic liver disease, acute viral liver disease, chronic viral liver disease, mixed liver disease (all chronic), cirrhosis, and alcoholic cirrhosis.

We adopted the DerSimonian and Laird random-effects model to estimate the pooled measures of treatment efficacy.^{59 60} If there is no substantial heterogeneity among the trials, the random-effects estimator reduces to the classical fixed-effects estimate. When significant heterogeneity exists, the random-effects model incorporates the statistical heterogeneity into the summary estimate and confidence interval. The random-effects model confidence interval is wider than the fixed-effect model confidence interval, when substantial heterogeneity exists, making the random-effects model more conservative.

Heterogeneity among trials was tested with a standard chi-square test (using a p value greater than 0.1 as evidence of heterogeneity), Galbraith plots, and funnel plots. A Galbraith plot is a graphic method used to complement the statistical assessment of heterogeneity and is particularly useful when the number of studies is small.⁶⁰ The position of each study along the two axes gives an indication of the weight allocated in the meta-analysis. The vertical axis (a Z statistic equal to the effect size divided by its standard error) gives the contribution of each study to the Q (heterogeneity) statistic. Points outside the confidence bounds are those studies that have a major contribution to heterogeneity; in the absence of heterogeneity, all points would be expected to be within the confidence bounds. Funnel plots used Begg's rank order correlation test.⁶¹ STATA 6.0 ^® (STATA Corporation ^® ) was used to perform all analyses and produce the graphic output.⁶² Specifically, the meta command was used to compute and graph the random-effects model estimates,⁶³ the "galbr" command to assess heterogeneity and produce Galbraith plots,⁶² and the metabias command to examine publication bias.⁶⁴

Effect sizes were converted to clinical laboratory units to aid in interpreting effect-size standard deviation units. As noted above, the effect-size statistic is calculated by dividing the between-group difference by the pooled standard deviation of the two groups. Since both numerator and denominator are expressed in the original laboratory units (e.g., milligrams per deciliter [mg/dL]), the units cancel out, and the effect size statistic is therefore "unitless." Effect sizes can be back-converted to a clinical laboratory value expressed in the original unit (e.g., mg/dL) by multiplying the effect-size value by a standard-deviation value. The statistical significance of the values (effect sizes or converted values) does not change; however, the magnitude of the "converted effect" will vary up or down depending on the magnitude of the standard deviation used.

Conversion of effect sizes to clinical laboratory units should not be interpreted as "true" values; conversions are presented for the single purpose of enhancing the interpretation of effect-size standard-deviation units.

Lacking population standard-deviation values (if available, they could be used), the investigator chose to use the "average" standard-deviation value for the pooled studies within each group. Two "averages" were examined: a weighted pooled standard deviation across studies (weighted by sample size) and the median pooled standard deviation. When the two values were substantially different (representing skewness), the median value was chosen. When the values were similar (or when only two studies provided pooled standard deviation estimates), the weighted pooled standard deviation value was used to convert effect sizes. Weighted average standard deviations that were used to convert effect sizes to clinical laboratory units were: albumin (0.74 grams per deciliter [g/dL]), bilirubin (0.32 g/dL), aspartate aminotransferase (44.77 units per liter [U/L]), alanine aminotransferase (36.02 U/L), gammaglutamyl transpeptidase (153.94 U/L), malondialdehyde (6.65 millimoles per milliliter [mmol/mL]), alkaline phosphatase (101.01 U/L), prothrombin time (17.40 seconds), and Child's score (2.55 units).

Milk Thistle and Alcohol-Related Liver Disease

Summary Table 1. Placebo-controlled studies of milk (more...)

Summary Table 1. Placebo-controlled studies of milk thistle for chronic alcoholic liver disease (n = 6)

Study	Design	Liver Disease	Subject Characteristics	Outcomes (favoring silymarin unless otherwise noted)
Author: Bunout66Year: 1992 Country: Chile Language: Spanish	Study type: RCT/blind Intervention/dose/duration: Silymarin (Legalon® )/280 mg daily/446 days Control: Placebo Outcome measuresa: AST, GGTP, alk phos, bilirubin, albumin, PT Followup interval: Monthly for an average of 15 months Assessment time points with results reported: Mean of 15 months	Type: Alcoholic liver disease (inactive cirrhosis, active cirrhosis, cirrhosis with alcoholic hepatitis, alcoholic hepatitis, fibrosis) Include: Positive for alcohol history and symptoms of icterus edema, ascites, or encepholopathy or bilirubin exceeding 2 mg/dL; PT exceeding 75 percent control; albumin less than 3 mg/dL Exclude: HBV antigen; renal failure; cardiac failure; terminal liver disease Acuity/severity: Chronic, cirrhosis, no other information Baseline group similarity: Demos, LFTs appear similar	N: 59 Mean age: 50 Percent male: 86 Setting: Specialty clinic	Significant: Albumin Not significant: Child's score, bilirubin, AST, GGTP, alk phos, PT, survival
Author: Feher70,99,100Year: 1989/1990 Country: Hungary Language: Hungarian	Study type: RCT/blind Intervention/dose/duration: Silymarin (Legalon® )/420 mg daily/180 days Control: Placebo Outcome measuresa: AST, ALT, GGTP, alk phos, bilirubin, albumin Followup interval: 90 and 180 days Assessment time points with results reported: 90 and 180 days	Type: Chronic alcoholic liver disease (cirrhosis, reactive fibrosis, fatty degeneration) Include: Alcohol intake exceeding 60 g/d in men and 30 g/d in women for 8 ± 4 years; chronic liver disease; no previous corticosteroid or immunosuppressive treatment Exclude: Not given Acuity/severity: Chronic, no other information Baseline group similarity: Demos, LFTs appear similar	N: 36 Mean age: 46 Percent male: 75 Setting: Unclear	Significant: AST, GGTP, MDA Not significant: (assumed for all) ALT, alk phos, albumin, PT, survival
Author: Lang71Year: 1990 Country: Hungary Language: English	Study type: RCT/blind/three arms Interventions/dose/duration: Silymarin (Legalon®)/420 mg daily/30 days AICA-P/600 mg daily/30 days Control: Placebo Outcome measuresa: AST, ALT, GGTP, alk phos, albumin Followup interval: 28 days Assessment time points with results reported: Weekly for 1 month	Type: Alcoholic cirrhosis Include: Alcohol intake exceeding 60 g/d in men and 30 g/d women for more than 6 years; histologic diagnosis of micronodular cirrhosis Exclude: Vascular and/or parenchymal decompensation; +HBsAg Acuity/severity: Chronic, cirrhosis, no other information Baseline group similarity: LFTs appear similar	N: 60 Mean age: 45 Percent male: 68 Setting: Unclear	Significant: AST, ALT, GGTP Not significant: Bilirubin
Author: Pares68Year: 1998 Country: Spain Language: English	Study type: RCT/blind Intervention/dose/duration: Silymarin (Legalon® )/450 mg daily/unclear Control: Placebo Outcome measuresa AST, ALT, GGTP, alk phos, bilirubin, albumin, PT Followup interval: Every 3 months for 2 years Assessment time points with results reported: Survival at 5 years; all others at 2 years (assumed)	Type: Alcoholic cirrhosis (24 percent had superimposed alcohol hepatitis) Include: Chronic alcoholism defined as alcohol exceeding 80 g/d men and 60 g/d women for more than 5 years; biopsy or laparoscopic proven alcoholic cirrhosis Exclude: If ever received colchicine, malotilate, penicillamine, or corticosteroids; life expectancy less than 6 months; drug addicted or pregnant; other known etiologies for cirrhosis (i.e., HBV, autoimmunity, primary biliary cirrhosis, or cryptogenic cirrhosis) Acuity/severity: Chronic, all cirrhosis, Child's score A/B/C/ 63/114/14 Baseline group similarity: Demos, LFTs appear similar	N: 200 Mean age: 50 Percent male: 79 Setting: Diagnosis in hospital then outpatient followup.	Significant:+ Trend: Encephalopathy, UGIB, survival for HCV-positive patients Not significant: AST, ALT, GGTP, PT, bilirubin, alk phos, hepatomegaly, splenomegaly, jaundice, ascites, survival Note: Trend was assessed qualitatively only for this review.
Author: Salmi69Year: 1982 Country: Finland Language: English	Study type: RCT/blind Intervention/dose/duration: Silymarin (Legalon® )/420 mg daily/28 days Control: Placebo Outcome measuresa: AST, ALT, alk phos, bilirubin, histology Followup interval: Day 28 Assessment time points with results reported: 28 days	Type: Alcoholic liver disease Include: Increased AST and ALT for more than 1 month despite order to abstain from alcohol Exclude: Not given Acuity/severity: No other information Baseline group similarity: Aminotransferase levels reported as similar but lower in controls	N: 97 Mean age: 37 Percent male: 86 Setting: Hospital	Significant: AST, ALT, histology Not significant: Alk phos, bilirubin
Author: Trinchet67Year: 1989 Country: France Language: French	Study type: RCT/blind Intervention/dose/duration: Silymarin (Legalon®)/420 mg daily/ 90 days Control: Placebo Outcome measuresa: AST, GGTP, bilirubin, albumin, PT, histology Followup interval: 90 days Assessment time points with results reported: 90 days	Type: Alcoholic hepatitis (50 percent with cirrhosis) Include: Biopsy-proven alcoholic hepatitis with or without cirrhosis Exclude: Hepatic encephalopathy; contraindications to liver biopsy; hepatocellular cancer; ascites resistant to diuretics; platelets less than 100,000 per mm; PT less than 50 percent; other diseases limiting survival Acuity/severity: 58/116 with cirrhosis; no other information Baseline group similarity: Demos, LFTs appear similar	N: 116 Mean age: 51 Percent male: 67% Setting: Unclear	Significant: Both silymarin and placebo better with alcohol abstinence Not significant: PT, albumin, bilirubin, GGTP, AST, histology (hepatitis, fibrosis)

a

Results not necessarily reported for all.

Abbreviations Used: AICA-P = amino imidazol carboxamial phosphate; alk phos = alkaline phosphatase; ALT = alanine aminotransferase; AST = aspartase aminotransferase; demos = demographic variables; GGTP = gammaglutamyl transpeptidase; HbsAg = hepatitis B surface antigen; HBV = hepatitis B virus; HCV = hepatitis C virus; LFT = liver function test; MDA = malondialdehyde; PT = prothrombin time; RCT = randomized controlled trial; UGIB = upper gastrointestinal bleed.

Across these trials, results varied and may be generally confounded by the question of abstinence from alcohol. In one, there was significant improvement in liver function with abstinence from alcohol for patients given silymarin or placebo. However, between silymarin and placebo, there was no difference in the course of liver function as measured by prothrombin time (PT), albumin, bilirubin, gammaglutamyl transpeptidase (GGTP), aspartate aminotransferase (AST), or degree of hepatitis or fibrosis on biopsy.⁶⁷ In four trials, at least one parameter of liver function improved significantly with silymarin compared with placebo, but the courses of an equal number or more parameters were not significantly different between silymarin and placebo.^{66 69 70 71} One study showed no improvement in overall survival with silymarin but marginally significantly improved survival (p=0.059 by univariate analysis) in HCV-positive patients given silymarin versus placebo.⁶⁸ Qualitatively, there does not appear to be a relationship between duration of therapy and improvement in liver function.

Milk Thistle and Chronic Liver Disease of Mixed Etiology

Summary Table 2. Placebo-controlled studies of milk (more...)

Summary Table 2. Placebo-controlled studies of milk thistle for chronic liver disease, mixed etiologies (n = 6)

Study	Design	Liver Disease	Subject Characteristics	Outcomes (favoring silymarin unless otherwise noted)
Author: Fintelmann73Year: 1980 Country: Germany Language: German	Study type: RCT/blind Intervention/dose/duration: Silymarin(Legalon® )/not given/not given Control: Placebo Outcome measuresa: AST, ALT, GGTP, alk phos, bilirubin Followup interval: Days 1, 3, 7, 10, 14, 21, 28 Assessment time points with results reported: Had to read off graphs, but all followups	Type: Toxic liver disease, any etiology (usually alcohol) Include: Clinical and laboratory evidence of toxic liver damage Exclude: Not given Acuity/severity: No other information Baseline group similarity: No information except age and gender "similar"	N: 66 Mean age: Not given Percent male: Not given Setting: Unclear	Significant: ALT, GGTP Not significant: AST, bilirubin, alk phos
Author: Fintelmann72Year: 1970 Country: Germany Language: German	Study type: Cohort with comparison group/blinding unclear Intervention/dose/duration: Silymarin (Legalon® )/6 tablets daily/42 days Control: Placebo Outcome measuresa: Bilirubin, alk phos, AST, ALT Followup interval: 42 days Assessment time points with results reported: 42 days	Type: Alcoholic liver disease, also other causes of fatty liver (pure parenchymal fatty degeneration, fatty degeneration and reactive inflammation, cirrhosis transformation in progress) Include: Biopsy-proven fatty liver, including early cirrhosis without portal hypertension, assignment stratified by diabetes mellitus, obesity, alcohol intake Exclude: Not given Acuity/severity: All degrees of fatty liver including early cirrhosis without portal hypertension Baseline group similarity: LFTs appear similar	N: 50 Mean age: Not given Percent male: Not given Setting: Unclear	Significant: AST (categorical) Not significant: AST (continuous), ALT, alk phos
Author: Tanasescu74Year: 1988 Country: Romania Language: English	Study type: RCT, randomization stratified by type of disease: CPH, CAH, HCV/blind Intervention/dose/duration: Silymarin (Silimarina®)/210 mg silybin equivalents daily/ 40 days Control: Placebo Outcome measuresa: ALT, GGTP, bilirubin, alk phos, PT Followup interval: 40 days Assessment time points with results reported: 40 days	Type: CPH, CAH, hepatic cirrhosis, etiology not given Include: Diagnosis based on hepatic biopsy; clinical and laboratory data on CAH and CPH patients; some hepatic cirrhosis patients did not have hepatic biopsy; medium forms of disease; both sexes; 20 to 60 years old; basic disease from 1 to 10 years Exclude: Not given Acuity/severity: Chronic, no other information Baseline group similarity: Age, LFTs appear similar	N: 177 Mean age: 53 Percent male: 47 Setting: Hospital	CAH CPH HCV Significant: + Trend: ALT Not significant: ALT, GGTP, bilirubin, alk phos, PT GGTP, bilirubin, alk phos, PT ALT, GGTP, bilirubin, alk phos, PT Note: Trend was assessed qualitatively only for this review.
Author: Ferenci75Year: 1989 Country: Austria Language: English	Study type: RCT/blind Intervention/dose/duration: Silymarin (Legalon® )/420 mg daily/ mean 41 months, range 2 to 6 years Control: Placebo Outcome measuresa: AST, ALT, GGTP, alk phos, bilirubin, albumin, PT Followup interval: 4 years Assessment time points with results reported: 2 and 4 to 5.5 years for survival	Type: Alcoholic and nonalcoholic cirrhosis; cirrhosis with alcoholic hepatitis Include: Diagnosis of cirrhosis within 2 years before entering study Exclude: End-stage liver disease; known malignancies; primary biliary cirrhosis; immunosuppressive therapy Acuity/severity: Chronic, cirrhosis, Child's score A/B/C 89/69/12 Baseline group similarity: Demos, LFTs appear similar	N: 170 Mean age: 57 Percent male: 72 Setting: Primary care/community clinic, specialty clinic, hospital	Significant: Survival (alcohol disease, Child's score A) Not significant: (No data given; reported only as not significant) AST, ALT, GGTP, bilirubin, survival (nonalcohol disease, Child's score B and C) Correspondence from Ferenci, Oct. 10, 1999: Length of treatment was variable. All patients were treated for 2 years (original primary endpoint) and were continued on the same therapy until the last patient that entered completed 2 years of therapy. Mortality was 27/83 in placebo group and 20/83 in silymarin group.
Author: Benda76Year: 1980 Country: Germany Language: German	Study type: RCT/blind Intervention/dose/duration: Silymarin (Legalon®)/420 mg daily/ Unclear Control: Placebo MVI Outcome measuresa: Survival Followup interval: 4 years Assessment time points with results reported: 4 years	Type: Alcoholic and nonalcoholic cirrhosis Include:Biopsy-proven cirrhosis Exclude: Moribund; poor compliance; immunosuppressive treatment in the past year; prednisone in past 6 months; D-penicillamine in past 3 months; primary biliary cirrhosis; Wilson's disease Acuity/severity: Chronic, cirrhosis, no other information Baseline group similarity: No information	N: 172 Mean age: Not given Percent male: Not given Setting: Unclear	Significant:+ Trend: Survival Not significant:Note: Trend was assessed qualitatively only for this review.
Author: Marcelli77Year: 1992 Country: Italy Language: English	Study type: RCT/blinding unclear Intervention/dose/duration: Silipide®/240 mg daily/90 days Control: Placebo Outcome measuresa: AST, ALT, bilirubin, albumin, PT Followup interval: 90 days Assessment time points with results reported: 90 days	Type: CPH, etiology unknown Include: Biopsy confirmed CPH, ALT, or AST 1.5 or greater than 2 times normal limits in past year Exclude: Other forms of liver disease (non-CPH) or decompensated disease; treatment with interferon antivirals, immunosuppressants, or immunomodulators within past 6 months Acuity/severity: Chronic, no other information Baseline group similarity: Demos appear similar; higher AST and ALT in Silipide® group	N: 65 Mean age: 48 Percent male: 71 Setting: Specialty clinic	Significant: AST, ALT Not significant: Bilirubin, albumin, PT

a

Results not necessarily reported for all.

Abbreviations used: alk phos = alkaline phosphatase; ALT = alanine aminotranferase; AST = aspartase aminotranferase; CAH = chronic active hepatitis; CPH = chronic persistent hepatitis; demos = demographic variables; GGTP = gammaglutamyl transpeptidase; HCV = hepatitis C virus; LFT = liver function test; MVI = multivitamin; PT = prothrombin time; RCT = randomized controlled trial.

Across three studies, at least one parameter showed significant improvement with silymarin compared with placebo, and there was no difference between silymarin and placebo for as many or more parameters. Two studies indicated a possible survival benefit: In one study, survival was significantly improved for patients with Child's A alcoholic liver disease with silymarin compared with placebo, and in the other study, there was a trend toward improved overall survival.^{75 76} Again, there is no apparent qualitative relationship between duration of therapy and efficacy of silymarin, but duration was not given or unclear for three of the trials.

In one study, 65 subjects with "chronic persistent hepatitis, etiology unknown" were randomized to receive either Silipide ^® at 240 mg/d for 3 months or placebo.⁷⁷ AST improved significantly in subjects receiving silymarin, but there were no significant differences between the two study groups in the course of alanine aminotransferase (ALT), bilirubin, albumin, and PT.

Milk Thistle and Viral Liver Disease

Summary Table 3. Placebo-controlled studies of milk (more...)

Summary Table 3. Placebo-controlled studies of milk thistle for acute viral hepatitis

Study	Design	Liver Disease	Subject Characteristics	Outcomes (favoring silymarin unless otherwise noted)
Author: Magliulo78Year: 1978 Country: Italy Language: German	Study type: RCT/blind Intervention/dose/duration: Silymarin (Legalon®)/420 mg daily/25 days Control: Placebo Outcome measuresa: AST, ALT, GGTP, alk phos, bilirubin, PT Followup interval: 1, 3, 5, 7, 9, 14, 21, 28 days Assessment time points with results reported: All followup	Type: Viral HAV and HBV Include: Acute HAV and HBV patients Exclude: Not given Acuity/severity: Acute, no other information Baseline group similarity: LFTs appear similar	N: 59 Mean age: 37 Percent male: 22 Setting: Hospital	Significant: AST, bilirubin + Trend: ALT Not significant: Alk phos Note: Trend was assessed qualitatively only for this review.

a

Results not necessarily reported for all.

Abbreviations used: alk phos = alkaline phosphatase; ALT = alanine aminotranferase; AST = aspartase aminotranferase; GGTP = gammaglutamyl transpeptidase; HAV = hepatitis A virus; HBV = hepatitis B virus; LFT = liver function test; PT = prothrombin time; RCT = randomized controlled trial.

In a blinded study, 59 patients with acute hepatitis A and B were randomized to silymarin at 420 mg/d for 25 days or placebo.⁷⁸ No further information on inclusion/exclusion criteria, acuity, or severity was given. The mean age was 37, and 22 percent of subjects were men. Compared with that seen with placebo, improvement in AST and bilirubin was significantly better with silymarin, with a nonsignificant trend toward improvement in ALT. There was no significant difference in the course of alkaline phosphatase.

Summary Table 4. Placebo-controlled studies of milk (more...)

Summary Table 4. Placebo-controlled studies of milk thistle for chronic viral hepatitis

Study	Design	Liver Disease	Subject Characteristics	Outcomes (favoring silymarin unless otherwise noted)
Author: Buzelli79Year: 1993 Country: Italy Language: English	Study type: RCT/blind Intervention/dose/duration: Silipide®Control: Placebo Outcome measuresa: AST, ALT, GGTP, MDA, alk phos, bilirubin, albumin Followup interval: 7 days Assessment time points with results reported: 7 days	Type: Chronic active hepatitis due to HBV and/or HCV Include: Biopsy-proven CAH, AST, and/or ALT greater than twice normal limits for more than 12 months; age 30 to 70 years old Exclude: Portal hypertension; hepatic encephalopathy, ascites, hepatocellular cancer, pruritus, icterus bilirubin, and alk phos more than twofold reference values; drug addiction and ANA, AMA, and ASMA; alcohol exceeding 30 g/d; malabsorption syndromes; cardiovascular, renal, or endocrine disorders; pregnancy; any drug treatment 3 months before beginning trial Acuity/severity: Chronic, AST or ALT greater than twice normal limits Baseline group similarity: Demos, LFTs appear similar	N: 20 Mean age: 53 Percent male: 30 Setting: Hospital	Significant: AST, ALT, GGTP Not significant: Bilirubin, alk phos, MDA, albumin
Author: Kiesewetter80Year: 1977 Country: Austria Language: German	Study type: RCT/quasi Intervention/dose/duration: Silymarin (Legalon® )/420 mg daily/365 days Control: Placebo Outcome measuresa: Histology Followup interval: 14, 30, 60 days then every 90 days to 1 year Assessment time points with results reported: One, combined/90 to 360 days	Type: Viral hepatitis (CPH and CAH) Include: CPH or CAH for more than 6 months; other medications for liver discontinued for 2 weeks or more Exclude: Disease less than 6 months; previous treatment with silymarin, steroids, or other cytotoxic drugs; alcohol exceeding 80 g/d or exceeding alcohol intake by laboratory values than patients' statement; other comorbidities requiring medications Acuity/severity: Chronic, no other information Baseline group similarity: Demos, similar, no information on LFTs	N: 24 Mean age: 53 Percent male: 50 Setting: Hospital	Significant:+ Trend: Improvement in liver biopsy Not significant:Note: Trend was assessed qualitatively only for this review.

a

Results not necessarily reported for all.

Abbreviations used: alk phos = alkaline phosphatase; ALT = alanine aminotranferase; AMA = antimitochondrial antibody; ANA = antinuclear antibody; ASMA = antismooth muscle antibody; AST = aspartase aminotranferase; CAH = chronic active hepatitis; CPH = chronic persistent hepatitis; demos = demographic variables; GGTP = gammaglutamyl transpeptidase; HBV = hepatitis B virus; HCV = hepatitis C virus; LFT = liver function test; MDA = malondialdehyde; RCT = randomized controlled trial.

One randomized, blinded, placebo-controlled trial included 20 patients with biopsy-proven chronic viral hepatitis or with AST and ALT levels greater than twice normal for more than 12 months due to HBV and/or HCV.⁷⁹ Patients with portal hypertension, hepatic encephalopathy, ascites, bilirubin or alkaline phosphatase more than twice the normal limit, alcohol consumption exceeding 30 grams per day (g/d), hepatocellular carcinoma, or liver disease associated with collagen vascular disease were excluded. Subjects received placebo or Silipide ^® at 240 mg/d for 7 days. AST, ALT, and GGTP significantly improved in patients receiving Silipide ^® compared with placebo. There was no difference between groups in bilirubin, alkaline phosphatase, malondialdehyde (MDA), or albumin.

In the other trial, 24 subjects with chronic viral hepatitis for more than 6 months were randomized to silymarin at 420 mg/d or placebo for 1 year.⁸⁰ Exclusion criteria were prior treatment with silymarin, steroids or other toxic drugs, alcohol consumption of more than 80 g/d, or other comorbidities requiring medication. There was a trend toward histologic improvement by liver biopsy in those receiving silymarin compared with placebo.

Milk Thistle and Cirrhosis

Five randomized and placebo-controlled trials included patients with cirrhosis along with patients with a mixed spectrum of types and severity of liver disease. Four are discussed in the section on chronic alcoholic liver disease,^{66 67 70 72} and one is discussed in the section on chronic liver disease of mixed etiologies.⁷⁴ Across three trials, at least one parameter of liver function improved in those taking silymarin compared with placebo.^{66 70 72} Most parameters, however, did not improve. Specifically, none improved in one study,⁷⁴ results were not separately presented for cirrhotic patients in a third study, and in another study, liver function improved equally in placebo and milk thistle arms with abstinence from alcohol.⁶⁷

Summary Table 5. Placebo-controlled studies of milk (more...)

Summary Table 5. Placebo-controlled studies of milk thistle and cirrhosis due to alcohol or other etiologies

Study	Design	Liver Disease	Subject Characteristics	Outcomes (favoring silymarin unless otherwise noted)
Author: Benda76Year: 1980 Country: Germany Language: German	Study type: RCT/blind Intervention/dose/duration: Silymarin (Legalon®)/420 mg daily/ Unclear Control: Placebo MVI Outcome measuresa: Survival Followup interval: 4 years Assessment time points with results reported: 4 years	Type: Alcoholic and nonalcoholic cirrhosis Include: Biopsy-proven cirrhosis Exclude: Moribund; poor compliance; immunosuppressive treatment in the past year; Prednisone in past 6 months; D-Penicillamine in past 3 months; primary biliary cirrhosis; Wilson's disease Acuity/severity: Chronic, cirrhosis, no other information Baseline group similarity: No information	N: 172 Mean age: Not given Percent male: Not given Setting: Unclear	Significant:+ Trend: Survival Not significant:Note: Trend was assessed qualitatively only for this review.
Author: Ferenci75Year: 1989 Country: Austria Language: English	Study type: RCT/blind Intervention/dose/duration: Silymarin (Legalon®)/420 mg daily/ mean 41 months, range 2 to 6 years Control: Placebo Outcome measuresa: AST, ALT, GGTP, alk phos, bilirubin, albumin, PT Followup interval: 4 years Assessment time points with results reported: 2 and 4 to 5.5 years for survival	Type: Alcoholic and nonalcoholic cirrhosis; cirrhosis with alcoholic hepatitis Include: Diagnosis of cirrhosis within 2 years before entering study Exclude: End-stage liver disease; known malignancies; primary biliary cirrhosis; immunosuppressive therapy Acuity/severity: Chronic, cirrhosis, Child's score A/B/C 89/69/12 Baseline group similarity: Demos, LFTs appear similar	N: 170 Mean age: 57 Percent male: 72 Setting: Primary care/community clinic, specialty clinic, hospital	Significant: Survival (alcohol disease, Child's score A) Not significant: (No data given; reported only as not significant) AST, ALT, GGTP, bilirubin, survival (nonalcohol disease, Child's score B and C) Correspondence from Ferenci, Oct. 10, 1999: Length of treatment was variable. All patients were treated for 2 years (original primary endpoint) and were continued on the same therapy until the last patient that entered completed 2 years of therapy. Mortality was 27/83 in placebo group and 20/83 in silymarin group.

a

Results not necessarily reported for all.

Abbreviations used: alk phos = alkaline phosphatase; ALT = alanine aminotransferase; AST = aspartase aminotransferase; demos = demographic variables; GGTP = gammaglutamyl transpeptidase; LFT = liver function test; MVI = multivitamin; PT = prothrombin time; RCT = randomized controlled trial.

Summary Table 6. Placebo-controlled studies of milk (more...)

Summary Table 6. Placebo-controlled studies of milk thistle and alcoholic cirrhosis (other etiologies of cirrhosis excluded)

Study	Design	Liver Disease	Subject Characteristics	Outcomes (favoring silymarin unless otherwise noted)
Author: Pares68Year: 1998 Country: Spain Language: English	Study type: RCT/blind Intervention/dose/duration: Silymarin (Legalon®)/450 mg daily/unclear Control: Placebo Outcome measuresa: AST, ALT, GGTP, alk phos, bilirubin, albumin, PT Followup interval: Every 3 months for 2 years Assessment time points with results reported: Survival at 5 years; all others at 2 years (assumed)	Type: Alcoholic cirrhosis (24 percent had superimposed alcohol hepatitis) Include: Chronic alcoholism defined as alcohol exceeding 80 g/d men and 60 g/d women for more than 5 years; biopsy or laparoscopic proven alcoholic cirrhosis Exclude: If ever received colchicine, malotilate, penicillamine, or corticosteroids; life expectancy less than 6 months; drug addicted or pregnant; other known etiologies for cirrhosis (i.e., HBV, autoimmunity, primary biliary cirrhosis, or cryptogenic cirrhosis) Acuity/severity: Chronic, all cirrhosis, Child's score A/B/C/ 63/114/14 Baseline group similarity: Demos, LFTs appear similar	N: 200 Mean age: 50 Percent male: 79 Setting: Diagnosis in hospital then outpatient followup.	Significant:+ Trend: Encephalopathy, UGIB, survival for HCV-positive patients Not significant: AST, ALT, GGTP, PT, bilirubin, alk phos, hepatomegaly, splenomegaly, jaundice, ascites, survival Note: Trend was assessed qualitatively only for this review.
Author: Lang71Year: 1990 Country: Hungary Language: English	Study type: RCT/blind/three arms Interventions/dose/duration: Silymarin (Legalon®)/420 mg daily/30 days AICA-P/600 mg daily/30 days Control: Placebo Outcome measuresa: AST, ALT, GGTP, alk phos, albumin Followup interval: 28 days Assessment time points with results reported: Weekly for 1 month	Type: Alcoholic cirrhosis Include: Alcohol intake exceeding 60 g/d in men and 30 g/d women for more than 6 years; histologic diagnosis of micronodular cirrhosis Exclude: Vascular and/or parenchymal decompensation; +HBsAg Acuity/severity: Chronic, cirrhosis, no other information Baseline group similarity: LFTs appear similar	N: 60 Mean age: 45 Percent male: 68 Setting: Unclear	Significant: AST, ALT, GGTP Not significant: Bilirubin

a Results not necessarily reported for all.

Abbreviations used: AICA-P = amino imidazol carboxamial phosphate; alk phos = alkaline phosphatase; ALT = alanine aminotransferase; AST = aspartase aminotransferase; demos = demographic variables; GGTP = gammaglutamyl transpeptidase; HBV = hepatitis B virus; HCV = hepatitis C virus; LFT = liver function test; PT = prothrombin time; RCT = randomized controlled trial; UGIB = upper gastrointestinal bleed.

Milk Thistle and Toxin-Induced Liver Disease

Summary Table 7. Placebo-controlled study of milk thistle (more...)

Summary Table 7. Placebo-controlled study of milk thistle and toxin-induced liver disease

Study	Design	Liver Disease	Subject Characteristics	Outcomes (favoring silymarin unless otherwise noted)
Author: Palasciano81Year: 1994 Country: Italy Language: English	Study type: RCT/blind, factorial design Intervention/dose/duration: Silymarin and psychotropic drugs/800 mg silymarin daily/ 90 days Silymarin and psychotropic drugs discontinued/800 mg daily/90 days Control: Placebo and no psychotropic drugs discontinued Placebo and psychotropic drugs/ not given/90 days Outcome measuresa: AST, ALT, MDA Followup interval: Day 15, 30, 60, 90, 120 Assessment time points with results reported: 30, 60, 90, and 120 days for MDA; 30 and 90 for AST and ALT	Type: Drug-induced liver disease (some patients were HBV-positive) Include: Women 40 to 60 years old; hospitalized; phenothiazines and/or butyrophenones for 5 years; AST or ALT greater than twice normal limits Exclude: Current therapy with other drugs that could influence progress of hepatopathy; other hepatopathies (alcohol, viral, autoimmune, hemochromatosis, Wilson's disease, porphyria cutanea tarda, primary or secondary hepatic neoplasia); BUN exceeding 60 mg/dL and/or creatinine exceeding 2.5 mg/dL; cardiac and circulatory insufficiency; diabetes mellitus; other important extrahepatic diseases; verified or presumed pregnancy; alcohol (more than 30 g/d) or opiate abuse Acuity/severity: Chronic; AST or ALT greater than two times the normal limit Baseline group similarity: Demos similar; LFTs variable	N: 60 Mean age: 52 Percent male: 0 Setting: Hospital	Significant: MDA for psychotropic drugs with silymarin versus placebo Not significant: AST, ALT, MDA for psychotropic drugs discontinued, silymarin versus placebo

a

Results not necessarily reported for all.

Abbreviations used: ALT = alanine aminotranferase; AST = aspartase aminotranferase; BUN = blood urea nitrogen; demos = demographic variables; HBV = hepatitis B virus; LFT = liver function test; MDA = malondialdehyde; RCT = randomized controlled trial.

Summary Table 8. Prophylactic milk thistle with hepatotoxic (more...)

Summary Table 8. Prophylactic milk thistle with hepatotoxic drugs

Study	Design	Liver Disease	Subject Characteristics	Outcomes (favoring silymarin unless otherwise noted)
Author: Magula82Year: 1996 Country:Language: Slovak	Study type: RCT/not blinded/prophylaxis Intervention/dose/duration tuberculosis treatment and Hepabene® (Silymarin and Fumaria officinalis alkaloids)/2 capsules daily/ 8 weeks Control: Tuberculosis medications alone Outcome measuresa: AST, ALT, bilirubin, "liver injury," interruption of tuberculosis medications Followup interval: Every 2 weeks for 8 weeks Assessment time points with results reported: 8 weeks (56 days)	Type: Drug-induced liver disease Include: Patients requiring tuberculosis treatment; normal LFTs at baseline Exclude: Not given Acuity/severity: No disease at baseline	N: 172 (92 tuberculosis medications alone; 29 tuberculosis medications plus silymarin) Mean age: 50 Percent male: 60 Setting: Unclear	Significant: AST, ALT Not significant:
Author: Allain83Year: 1999 Country: France Language: English	Study type: RCT/blind/prophylaxis Intervention/dose/duration: Silymarin/420 mg daily/84 days Control: Placebo Outcome measuresa: AST, ALT Followup interval: 8 times over 15 weeks Assessment time points with results reported: 8 times over 15 weeks	Type: Drug- (Tacrine-) induced liver disease Include: Patient over 50 years old; mild to moderate Alzheimer's disease; Tacrine treatment Exclude: Past history hepatic disorder (cirrhosis, increased bilirubin, AST, and/or ALT) Acuity/severity: No disease at baseline	N: 222 Mean age: Approximately 74 Percent male: 39 Setting: Specialty clinic	Significant:Not significant: AST, ALT

a

Results not necessarily reported for all.

Abbreviations used: ALT = alanine aminotransferase; AST = aspartase aminotransferase; LFT = liver function test; RCT = randomized controlled trial.

In the study of prophylactic milk thistle during treatment for tuberculosis, 29 subjects with normal liver function tests received antituberculosis drugs plus Hepabene ^® , a mixture of silymarin and Fumaria officinalis alkaloids (two capsules daily, no other information given), and 93 subjects received antituberculosis drugs alone.⁸² Proportions receiving various antituberculosis drugs, in the control and silymarin groups respectively, were the following: rifampin (98 percent and 100 percent); isoniazid (99 percent and 97 percent); pyrazinamide (84 percent and 100 percent); ethambutol (50 percent and 38 percent); and streptomycin (37 percent and 14 percent). Initially, AST increased in 49 percent of control subjects and 38 percent of subjects receiving Hepabene ^® Similarly, ALT increased in 48 percent of control and 31 percent of those receiving Hepabene ^® These differences were statistically significant over the course of 8 weeks; AST then fell in 40 percent of subjects receiving Hepabene ^® , compared with 19 percent of controls. Similarly, ALT fell in 48 percent of those receiving Hepabene ^® , compared with 22 percent in controls. Again, these differences were statistically significant.

The second study of prophylatic milk thistle was a randomized, double-blind trial in which 222 patients meeting a priori criteria for Alzheimer's dementia were randomized to tacrine and silymarin or tacrine with placebo.⁸³ Patients with prior liver disease were excluded. Silymarin (420 mg/d) was given for 1 week, and then tacrine was added, first at 40 mg/d for 6 weeks, then at 80 mg/d for 6 additional weeks. (Note: Published report⁸³ states that silymarin [Legalon ^® ] dose was 420 mg/d, but an internal study summary reportedly states dose was 140 mg/d from Madaus.⁸⁴ ) There were no significant differences in the courses of AST or ALT levels or the rate of side effects during the study. There was no significant difference even though rates appeared lower.

Milk Thistle and Cholestasis

No studies met criteria for evaluating efficacy of milk thistle in treating cholestatic liver disease, pregnancy-related or not.

Milk Thistle and Primary Hepatic Malignancy

No studies met criteria for evaluating milk thistle efficacy in treating or preventing primary hepatic malignancy (hepatocellular carcinoma or cholangiocarcinoma).

Overview of Adverse Effect Literature

The adverse effect literature regarding milk thistle is difficult to interpret for many reasons. First, searching for studies that report adverse effects is difficult and, given current indexing systems for electronic databases, probably incomplete. Many studies may mention adverse effects in passing, but they do not use adverse effects as a key index word or in their abstracts. If these studies do not otherwise meet selection criteria in a review, they will be missed. Second, information is frequently missing on whether adverse effects were elicited by voluntary self-report or standardized probes. Third, in some case reports and case series, adverse effects cannot be directly attributed to milk thistle because chance, coincidence, or confounding factors could have been responsible. Fourth, case reports and case series may miss delayed adverse reactions because such associations are more difficult to make than those that occur immediately after administration of an agent. Fifth, although case reports and case series can provide qualitative information about the nature of an adverse effect, they cannot generate estimates of incidence.⁸⁶

Summary Table 12. Adverse effects of oral use of milk (more...)

Summary Table 12. Adverse effects of oral use of milk thistle, silymarin, or Silipide®

Adverse Effects	Level of Evidence
Possible/probable anaphylaxis	Three case reports
Gastrointestinal Nausea, diarrhea, upset stomach, epigastric discomfort, heartburn, dyspepsia, flatulence, meteorism, "uneasiness" of stomach, vomiting, anorexia, change in bowel movement, "gastric intolerance," abdominal fullness or pain	Four RCTs: silymarin < control Five cohort studies: 0.3% 2%, 5%, 6%, 8% 11 occurrences/975 patients 9 occurrences/2,169 patients
Headache	Three RCTs: < control Two cohort studies: 0.04%, 0.04%
Skin: Itching, pruritus Exanthem, urticaria, skin rash, eczema	Two RCTs: silymarin < control Three cohort studies: 2 occurrences/2,637 patients 2 occurrences/975 patients 2 occurrences/2,169 patients Two RCTs: silymarin < control Two cohort studies: 2 occurrences/975 patients 1 occurrence/2,169 patients
Other: Decreased energy, discomfort, indisposition, decreased energy, restlessness Asthenia, malaise, irritability Insomnia Arthalgia Rhinoconjunctivitis Impotence	Cohort study: 0.3% 1 occurrence/975 patients 1 occurrence/2,169 patients One RCT: silymarin < control One RCT: silymarin < control One RCT: silymarin < control One case report One RCT: 1/29 silymarin, 0/31 control

Frequency of adverse effects was sometimes reported as the following: (1) percent of subjects, (2) number of subjects, or (3) number of occurrences of adverse effects in a group of subjects without detail about subjects who may have had more than one adverse effect.

RCT = randomized controlled trial.

Common Symptomatic Effects

Gastrointestinal side effects are the most commonly reported adverse effects, in both randomized clinical trials and prospective cohort studies.^{75 77 83 85 88 89 90 91 92 93 98} Overall incidence appears relatively low, and in RCTs there is no apparent difference between treatment and control groups.^{66 68 77} Skin manifestations of adverse drug effects are also fairly commonly reported, but again overall incidence appears low and no more frequent in groups receiving milk thistle than in control groups.^{66 68 77} Incidence of headaches also appears low and no different between silymarin and control groups in RCTs.^{66 68}

Common and Uncommon Serious Adverse Effects

Serious adverse effects identified in available evidence for this systematic review were anaphylactoid reactions or anaphylaxis. Available evidence was limited to three case reports. In one case report, a variety of symptoms plus "collapse" was associated with milk thistle, improved after discontinuation of milk thistle, reappeared on rechallenge, but was also associated with potential alternative etiologies.⁹⁸ In a second case report of anaphylactic shock, no information was given about associated evidence of causality.⁹⁴ A third case report of an apparent anaphylactoid reaction improved with prednisone and discontinuing silymarin, and there were no apparent alternative possible etiologies.⁹⁵

Figure 1-1. Aspartate aminotransferase, less drug study, <45 days Begg's funnel plot with pseudo 95% confidence limits

Figure 1-2. Aspartate aminotransferase, less drug study, <45 days

Figure 1-3. Aspartate aminotransferase, less drug study, <45 days

Figure 1-4. Aspartate aminotransferase, less drug study, 90 days Begg's funnel plot with pseudo 95% confidence limits

Figure 1-5. Aspartate aminotransferase, less drug study, 90 days

Figure 1-6a. Aspartate aminotransferase, less drug study, 90 days

Figure 1-6b. Aspartate aminotransferase, less drug study, less outlier, 90 days

Figure 1-7. Aspartate aminotransferase, <45 days Begg's funnel plot with pseudo 95% confidence limits

Figure 1-8. Aspartate aminotransferase, <45 days

Figure 1-9. Aspartate aminotransferase, <45 days

Figure 1-10. Aspartate aminotransferase, 90 days Begg's funnel plot with pseudo 95% confidence limits

Figure 1-11. Aspartate aminotransferase, 90 days

Figure 1-12a. Aspartate aminotransferase, 90 days

Figure 1-12b. Aspartate aminotransferase, less outlier, 90 days

Figure 1-13. Alanine aminotransferase, less drug study, <45 days Begg's funnel plot with pseudo 95% confidence limits

Figure 1-14. Alanine aminotransferase, less drug study, <45 days

Figure 1-15a. Alanine aminotransferase, less drug study, <45 days

Figure 1-15b. Alanine aminotransferase, less drug study, less outlier, <45 days

Figure 1-16. Alanine aminotransferase, less drug study, 90 days Begg's funnel plot with pseudo 95% confidence limits

Figure 1-17. Alanine aminotransferase, less drug study, 90 days

Figure 1-18. Alanine aminotransferase, less drug study, 90 days

Figure 1-19. Alanine aminotransferase, <45 days Begg's funnel plot with pseudo 95% confidence limits

Figure 1-20. Alanine aminotransferase, <45 days

Figure 1-21a. Alanine aminotransferase, <45 days

Figure 1-21b. Alanine aminotransferase, less outlier, <45 days

Figure 1-22. Alanine aminotransferase, 90 days Begg's funnel plot with pseudo 95% confidence limits

Figure 1-23. Alanine aminotransferase, 90 days

Figure 1-24. Alanine aminotransferase, 90 days

Figure 1-25. Gammaglutamyl transpeptidase, <45 days Begg's funnel plot with pseudo 95% confidence limits

Figure 1-26. Gammaglutamyl transpeptidase, <45 days

Figure 1-27. Gammaglutamyl transpeptidase, <45 days

Figure 1-28. Gammaglutamyl transpeptidase, 90 days Begg's funnel plot with pseudo 95% confidence limits

Figure 1-29. Gammaglutamyl transpeptidase, 90 days

Figure 1-30. Gammaglutamyl transpeptidase, 90 days

Figure 1-31. Alkaline phosphatase, <45 days Begg's funnel plot with pseudo 95% confidence limits

Figure 1-32. Alkaline phosphatase, <45 days

Figure 1-33. Alkaline phosphatase, <45 days

Figure 1-34. Bilirubin, <45 days Begg's funnel plot with pseudo 95% confidence limits

Figure 1-35. Bilirubin, <45 days

Figure 1-36. Bilirubin, <45 days

Figure 1-37. Bilirubin, 90 days Begg's funnel plot with pseudo 95% confidence limits

Figure 1-38. Bilirubin, 90 days

Figure 1-39. Bilirubin, 90 days

Figure 1-40. Albumin, 90 days Begg's funnel plot with pseudo 95% confidence limits

Figure 1-41. Albumin, 90 days

Figure 1-42. Albumin, 90 days

Figure 1-43. Prothrombin time, 90 days Begg's funnel plot with pseudo 95% confidence limits

Figure 1-44. Prothrombin time, 90 days

Figure 1-45. Prothrombin time, 90 days

National Advisory Panel

Marilyn Barrett, PhD
Owner and Principal
Pharmacognosy Consulting Services

Mark Blumenthal
Executive Director
American Botanical Council

David Eisenberg, MD
Beth Israel Deaconess Medical Center

Lucinda Miller, PharmD, BCPS
Editor
Journal of Herbal Pharmacotherapy

Richard Nahin, MPH, PhD
Acting Director
Division of Extramural Research
National Center for Complementary and Alternative Medicine

Mary Ann Richardson, DrPH
Assistant Professor and Director
The University of Texas Center for Alternative Medicine Research in Cancer
The University of Texas - Houston Health Science Center School of Public Health

Nancy Ridenour, RN, PhD, CS, FNC, FAAN
Dean
College of Nursing
Illinois State University

David Schardt
Associate Nutritionist
Center for Science in the Public Interest

William A. Watson, PharmD, DABAT, FAACT
Professor (Clinical) and Managing Director
Department of Surgery
South Texas Poison Center
The University of Texas Health Science Center at San Antonio

Elizabeth Yetley, PhD
Director
Office of Special Nutritionals
Center for Food Safety and Applied Nutrition
Food and Drug Administration

Technical Experts

Bradly Jacobs, MD, MPH
Senior Clinical Research Fellow
Osher Center for Integrative Medicine
Department of Medicine
University of California - San Francisco/Veterans Affairs Medical Center

Cathi Dennehy, PharmD
Assistant Clinical Professor
University of California at San Francisco

Kenneth Flora, MD, FACG
Assistant Professor
Oregon Health Sciences University

Peer Reviewers

Fourteen peer reviewers provided insightful comments and feedback on our draft report.

Maurizio Bonacini, MD
Associate Professor
Hepatitis Research Center
Division of Gastroenterology and Liver Diseases
Department of Medicine
University of Southern California School of Medicine

Robert Eckel, MD
Professor
Division of Endocrinology, Metabolism, and Diabetes
Department of Medicine
University of Colorado Health Science Center

Wolfgang Fleig, MD
Professor and Chair
First Department of Medicine
Martin Luther University Halle-Wittenberg
Halle (Saale), Germany

Robert Fontana, MD
Assistant Professor
Department of Internal Medicine
University of Michigan Health System

Alexander Gerbes, MD
Professor
Ludwig-Maximilians-Universitat-München

Eric Gershwin, MD
Division Chief
Division of Rheumatology/Allergy
Department of Internal Medicine
University of California at Davis School of Medicine

Ruth Kava, PhD, RD
Director of Nutrition
American Council on Science and Health

Ronald Koretz, MD
Professor and Chief
Division of Gastroenterology
Department of Medicine
University of California at Los Angeles

David Lee, PhD
Director
Natural Products Laboratory
McLean Hospital

Klaus Linde, MD
Researcher
Centre for Complementary Medicine Research
Department of Internal Medicine II Technichal University

Tianshu Liu, MD, MSc
Attending Physician
Department of Gastroenterology
Shanghai Medical University
Zhoug Medical Hospital

Srini Srinivasan, PhD
Director
Dietary Supplements Division
U.S. Pharmacopoeia

Kathleen Stevens, RN, EdD, FAAN
Professor
Department of Family Nursing
The University of Texas Health Science Center at San Antonio

Wendell Winters, PhD
Associate Professor
Department of Microbiology
The University of Texas Health Science Center at San Antonio

Authors Who Provided Additional Data

Some articles included in this report had relevant data missing from their publications. We contacted the authors requesting this information. Our heartfelt thanks to those who responded:

• Dr. Paolo Saba

• Dr. Heikki A. Salmi

• Dr. Gianluigi Vendemiale

• Dr. Peter Ferenci

• Dr. Robert Flisiak

San Antonio Evidence-based Practice Center and Veterans Evidence-based Research, Dissemination and Implementation Center Staff