There have been a number of cases reported in the medical literature indicating a link between herbal consumption and adverse effects. Many of these are reviewed in the monographs in this book. Most are minor and infrequent in nature. Where serious adverse events are reported, the information is often of poor quality, making a credible link between cause and effect difficult to establish. However, it is worthwhile to note some defining examples of serious adverse events that have occurred in the past few decades. Many of these have resulted from contamination or adulteration, or relate to known toxic herbs.

During the early 1990s, several patients with renal failure were admitted to hospitals in Brussels with progressive interstitial fibrosis and tubular atrophy that was linked to a herbal slimming preparation. At least 30 individuals were found to have sustained end-stage renal failure from the incident, making it perhaps the single most serious adverse event linked to herbal consumption in modern times. The Chinese herb Aristolochia fangchi was found to be an ingredient of the formulation instead of the intended Stephania tetrandra. The consequent presence of aristolochic acid, a known toxin, was put forward as a hypothesis for the aetiology of the nephropathies in the literature and the phenomenon became known (perhaps incorrectly) as Chinese herb nephropathy (CHN). However, the case prompted the Association Pharmaceutique Belge at the Service du Contrôle des Médicaments to probe the matter further.³ They pointed to the idiosyncratic nature of the reactions and the presence of other powerful synthetic drugs in the mixture as suggesting a more complex story. They considered that the cocktail of sometimes powerful preparations adopted by some observing slimming regimes might have significantly lowered the threshold for renal damage. However, aristolochic acid has been known as a nephrotoxin for decades and cases not linked to the concurrent use of drugs have been reported.⁴

In 1997, it was reported that two of the women subsequently developed urothelial cancer caused by the genotoxicity of aristolochic acid.⁵ An article published in June 1999 reported further cases of urothelial cancer. Cosyns and co-workers tested 10 patients with CHN.⁶ Four (40%) were found to have urothelial carcinoma and abnormal cells were found in all of the 10 patients. Nortier and co-workers⁷ (June 2000) concluded that the incidence of urothelial cancer among patients with CHN is high and that the risk was related to the cumulative dose of the herb. They reported treating 105 patients with CHN of whom 43 had been admitted with end-stage renal failure. Thirty-nine of these patients were tested for urothelial carcinoma. Eighteen cases were found, and mild-to-moderate dysplasia was found in a further 19 patients.

Cases of CHN have also been reported in France, Spain, Japan, the UK and Taiwan, where cases of urothelial carcinoma have also been detected.⁷Aristolochia spp. can also be used as substitutes for several other Chinese herbs,⁸ and Chinese herbal products found to contain aristolochic acid have been recalled in several countries (including Australia⁹ and the USA¹⁰).

This is the first credible, real-world example where the clinical use of a medicinal plant (albeit not the intended one) has resulted in the subsequent development of cancer.

During the early 1990s, liver units in France began to report a number of cases of liver disease possibly linked to the consumption of a slimming herb. The hepatotoxicity of germander (Teucrium chamaedrys) was confirmed in isolated rat hepatocytes, particularly a crude fraction containing the diverse furanoditerpenoids. It was concluded that they are activated by cytochrome P450 3 A into electrophilic metabolites that deplete glutathione and protein thiols and form plasma membrane blebs.¹¹ However, these cases were more likely to have been idiosyncratic drug reactions (IDRs) to the germander. Germander was subsequently banned from use in many countries, although germander-induced hepatotoxicity is probably still occurring, mainly from the fact that certain species of Teucrium are commonly used as a substitute for American skullcap (Scutellaria lateriflora).

In fact, so widespread was this adulteration, the macroscopic and microscopic description of skullcap cut herb in the British Herbal Pharmacopoeia 1983 was probably for a species of Teucrium. Adulteration of commercial skullcap continued to be an issue in Europe, the UK and the USA into the late 20th century.¹² Idiosyncratic hepatotoxicity was reported for tablets containing skullcap and valerian in 1989¹³ and for tablets containing skullcap, mistletoe, kelp, wild lettuce and motherwort in 1981¹⁴ (although the presence of mistletoe was later questioned).¹⁵ The observed hepatotoxicity of these herbal products is probably attributable to a germander species, rather than the herbs mentioned, including skullcap. Case reports of hepatotoxicity caused by germander are not limited to T. chamaedrys,¹⁶ other Teucrium spp. have caused hepatic failure.¹⁷ One of the reported UK cases of skullcap-related hepatotoxicity was confirmed as being due to T. canadense rather than skullcap.¹⁶

From January 1991 to December 1993, the Medical Toxicology Unit (formerly Poisons Unit) at Guy’s and St Thomas’s Hospital in London received reports of 11 cases of liver damage following the use of Chinese herbal medicine for skin conditions. There was strong evidence of an association in two cases, as recovery after withdrawal and recurrence of hepatitis after rechallenge were observed. The time-course relationship, recovery after ceasing Chinese herbal medicine and absence of alternative causes of liver damage suggested an association in two further symptomatic cases following a single period of exposure. Herbal material was available for analysis in seven cases. The plant mixtures varied, so no single ingredient could account for liver injury. Effects did not appear dose related and it was concluded they were probably idiosyncratic.¹⁸ Two patients were additionally described who suffered an acute hepatic illness related to taking traditional Chinese herbs for skin disease. Both recovered fully. The mixtures they took included two herbs that were also present in the mixture taken by a previously reported patient who suffered fatal hepatic necrosis.¹⁹ Sporadic cases of IDRs to Chinese herbal formulations resulting in hepatotoxicity have been reported in the literature ever since. A 2009 study from a gastroenterological department in a Chinese hospital concluded that Chinese herbs were a significant factor in idiosyncratic hepatotoxicity, although the liver injury was not severe in most cases.²⁰ Other studies reviewing patients attending single clinics or hospitals in the UK, Germany and Japan have found a much lower incidence of idiosyncratic hepatitis from the use of Chinese herbal formulations.²¹

Case reports of idiosyncratic hepatotoxicity to Western herbs extend beyond germander. A review of 18 reports of adverse events associated with the ingestion of chaparral reported to the Food and Drug Administration (FDA) between 1992 and 1994 found evidence of hepatotoxicity in 13 cases (causal (10), not followed up (1), probable (2), insufficient data (5)). Of the 13 cases, 10 had ingested chaparral only, with the remainder taking products containing chaparral and other herbs and/or ingredients. Adverse events occurred 3 to 52 weeks after the ingestion of chaparral and resolved 1 to 17 weeks after ceasing intake. The predominant pattern of liver injury was characterised as toxic or drug-induced cholestatic hepatitis. In four individuals there was progression to cirrhosis and in two individuals there was acute fulminant liver failure that required a liver transplant. Of the patients requiring a liver transplant, chaparral was probably not the only factor in one case.²²

In November 2001 the German Health Authority (BfArM) announced that it was intending to ban the use of kava (Piper methysticum) because of reported cases linking kava consumption with hepatotoxicity. Despite submissions from manufacturers, the therapeutic use of kava was banned altogether in Germany in 2002 and several other countries such as Japan, France and Canada followed suit. Late in 2002 it was announced that the Medicines Control Agency (MCA) in the UK would also be banning kava; in February 2003 Swissmedic in Switzerland followed. Like most other herbal hepatotoxicity cases, those linked to kava usage were most likely an immunologically mediated IDR ca rather than a direct toxic effect. Issues associated with this potential IDR to kava are fully explored in the kava monograph.

In 2002 a group of Australian doctors reported suspected hepatotoxicity, presumably an IDR, associated with the ingestion of black cohosh (Actaea racemosa).²³ At least 68 more cases have been reported since then, but the association remains a contentious issue. One analysis of the reported cases using the updated Council for International Organisations of Medical Sciences (CIOMS) causality assessment asserted that there was little, if any, evidence for a causal relationship between use of black cohosh and liver damage.²⁴ For more information on this topic see the black cohosh monograph.

Since its widespread promotion and use in the USA, numerous adverse reactions to the herb Ephedra (Ephedra sinica) have been recorded. However, it is difficult to assess meaningfully the majority of these due to insufficient information, the failure to distinguish between its main alkaloid ephedrine and Ephedra as treatments, and the tendency to combine Ephedra with other stimulants in weight loss products (which may potentiate its ability to cause serious side effects).

Nonetheless, the RAND Report undertook a comprehensive analysis of adverse consequences from both clinical trials and case reports submitted to the FDA.²⁵ Using data from clinical trials, the report concluded that there is sufficient evidence that the use of ephedrine and/or Ephedra or ephedrine plus caffeine is associated with two to three times the risk of nausea, vomiting, psychiatric symptoms such as anxiety and change in mood, autonomic hyperactivity and palpitations. It was not possible to determine the contribution of caffeine to these events. There were no reports of serious adverse events in the controlled trials.²⁵

From the adverse events reported by one manufacturer of Ephedra-containing dietary supplements and to the FDA, the RAND Report identified what it termed to be ‘sentinel events’. Classification of a sentinel event does not mean to imply a proven cause and effect relationship. These included two deaths, three myocardial infarctions, nine strokes, three seizures and five psychiatric cases associated with prior Ephedra consumption. The report also identified 43 additional cases as possible sentinel events and noted that about half of all the sentinel events occurred in people aged 30 years or younger.²⁵ Hepatotoxicity has also been linked to the use of Ephedra both in traditional Chinese formulations and in weight loss supplements.26–28

Animal exposure to pyrrolizidine alkaloids (PAs), found in several medicinal plants, has led to a dose-dependent swelling of hepatocytes and haemorrhagic necrosis of perivenular cells of the liver, with concomitant loss of sinusoidal lining cells, with sinusoids filled with cellular debris, hepatocyte organelles and red blood cells. These are all features of veno-occlusive disease.²⁹ These effects do not represent an IDR, but rather follow from the direct hepatotoxic activity of these alkaloids. The LD₅₀ for a pyrrolizidine-rich extract of Senecio was found to be 160 mg/kg.³⁰ Despite their similarity in structure, PAs differ in their individual LD₅₀ values and in the organs in which toxicity is expressed. In one study of four PAs, the proportion of the PA removed by liver cultures varied considerably due to differences in the production of reactive metabolites (dehydroalkaloids), which appear to be largely responsible for the toxicity of PAs, and in their conversion to a safer form (GSDHP).³¹ Among pyrrolizidine-containing plants, heliotrope³² and Senecio³³ have been found to be responsible for veno-occlusive disease in humans. Clinical manifestations of poisoning in humans include abdominal pain, ascites, hepatomegaly and raised serum transaminase levels. Prognosis is often poor with death rates of 20% to 30% being reported.³⁴ In vivo studies of coltsfoot, containing senkirkine, have shown some evidence of toxicity.³⁵ However, the key reported case linked to coltsfoot consumption turned out to be a substitution problem. Tea containing peppermint, and what the mother thought was coltsfoot (Tussilago farfara), was associated with veno-occlusive liver disease in an 18-month-old boy. Pharmacological analysis of the tea compounds revealed high amounts of PAs, mainly seneciphylline and the corresponding N-oxide. It was calculated that the child had consumed at least 60 mg/kg body weight per day of the toxic pyrrolizidine alkaloid mixture over 15 months. Macroscopic and microscopic analysis of the leaf material indicated that Adenostyles allariae had been erroneously gathered by the parents in place of coltsfoot. The child was given conservative treatment only and recovered completely within 2 months.³⁶

A review of 18 case reports of pennyroyal ingestion documented moderate to severe toxicity in patients who had been exposed to at least 10 mL of pennyroyal oil. In one fatal case, postmortem examination of a serum sample obtained 72 h after the acute ingestion identified 18 ng/mL of pulegone and 1 ng/mL of menthofuran.³⁷ In another case of ingestion of 30 mL of pennyroyal oil by a pregnant woman, symptoms included abdominal spasm, nausea, vomiting, alternating lethargy and agitated behaviour. Kidney failure and a solid liver necrosis developed subsequently and death occurred 7 days later. In two similar cases where doses used were 10 mL and 15 mL of oil, vomiting, agitation, fainting, flank pain and dermatitis occurred, but with no lasting toxic symptoms.³⁸

Maternal ingestion of blue cohosh (Caulophyllum thalictroides) in late pregnancy has been associated with four documented cases of perinatal adverse events. The first case occurred after a normal labour, where a female infant was not able to breathe spontaneously and sustained central nervous system (CNS) hypoxic-ischaemic damage. A midwife had attempted induction of labour using a combination of blue cohosh and black cohosh given orally (dosage undefined) at around 42 weeks’ gestation.³⁹

In the second case, severe congestive heart failure and myocardial infarction in a newborn male were attributed to maternal consumption of blue cohosh tablets. The woman had been advised to take 1 tablet per day (herb quantity not specified) but she took 3 tablets per day for 3 weeks prior to delivery. Cardiomegaly and mildly reduced left ventricular function were still evident at 2 years of age.⁴⁰ The tablets were not analysed for their content. Stroke in an infant was reported as a possible association with a blue cohosh-containing dietary supplement in the FDA’s Special Nutritionals Adverse Event Monitoring System database (which listed adverse events but was not subject to preconditions, analysis or peer review).⁴¹ The level of documentation of this case is poor.

Finally, a case report linked stroke in a baby with blue cohosh consumption by the mother.⁴² A female infant weighing 3.86 kg was born at just over 40 weeks’ gestation to a healthy 24-year-old woman. The obstetrician reportedly had advised the woman to drink a tea made from blue cohosh because induction of labour was a recognised effect of this herb. A caesarean section was performed after a failed attempt at vaginal delivery. The infant had focal motor seizures of the right arm, which began at 26 h of age, and were controlled with phenobarbital and phenytoin. A computed tomographic (CT) scan obtained when the infant was 2 days of age showed an evolving infarct in the distribution of the left middle cerebral artery. There were no other apparent causes for the baby’s condition.

In a curious development with this fourth case, urine and meconium were positive for the cocaine metabolite benzoylecgonine, and testing of the mother’s bottle of blue cohosh and another brand of the same herb were also positive for this metabolite. Maternal cocaine is a well-known cause of perinatal stroke. Later the authors commented that the finding of a cocaine metabolite in blue cohosh should be interpreted with caution.⁴³ The finding is most likely due to a false positive reading from the analytical tests used (which did not have a high degree of specificity). In other words, blue cohosh most likely contains a phytochemical which reacts like the cocaine metabolite in terms of the test used, but is not related in any way to cocaine.

Adverse effects have also been documented for a pregnant woman ingesting blue cohosh. Nicotinic toxicity was reported following the attempted use of blue cohosh as an abortifacient.⁴⁴ A 21-year-old woman developed tachycardia, sweating, abdominal pain, vomiting and muscle weakness following the ingestion of a blue cohosh tincture. The authors suggested that these symptoms were consistent with nicotinic toxicity and probably resulted from N-methylcytisine, a phytochemical component of blue cohosh. Symptoms resolved over 24 h.

Many plants are designated as poisonous, based on toxic effects either after human exposure or following grazing in livestock. Several commonly and safely used medicinal plants are responsible for toxic effects in livestock, notably St John’s wort and Tribulus. However, the issue here is the amount ingested by the animals, which can be several orders of magnitude greater than the human therapeutic dose. Other factors can also influence toxic effects in ruminants. In the case of Tribulus, rumen bacteria apparently metabolise the steroidal saponins into toxic components. This is not relevant to human metabolism. (See the Tribulus monograph for a full discussion of this phenomenon.)

In general, plants have considerably less acute toxicity than many other agents in our modern environment, such as chemicals and drugs. This could be expected since the chemicals in plants are diluted by a large percentage of inert plant material. This assertion is borne out by statistics, for example data from the American Association of Poison Control Centres (AAPCC). In a recent publication, information from the 1983 to 2009 annual reports of the AAPCC was analysed, together with queries of the 2000 to 2009 AAPCC Toxic Exposure Surveillance System and the National Poison Data System databases.⁴⁵ During 2000 to 2009, 668 111 plant ingestion exposures were reported, with around 90% of these involving single plants. There was a steady decline in the number of plant exposures, falling from 8.9% of all reported exposures in 1983, to 2.4% in 2009. Young children accounted for more than 80% of plant ingestion exposures. Only 45 fatalities were recorded between 1983 and 2009, with Datura and Cicuta species accounting for about one-third of these. The authors concluded that, while plant ingestion remains a common call for poison information centres, morbidity and mortality associated with these were very low relative to the total number of reported plant exposures.

Given the existence of toxic plants, it is accepted by all involved that not all plants are safe to use as remedies. The incidence of these clearly varies from country to country. Appendix B provides a list of several potentially toxic medicinal herbs. With a few exceptions, their use is best avoided, especially during pregnancy and lactation. The individual herb monographs in this book contain detailed toxicity information, where available.

Understanding the key safety issues and demonstrating the safe use of herbal therapy is probably the biggest challenge that faces the future widespread use of medicinal plants. A key issue is that the safety of any given therapy can never be considered in isolation from its efficacy. In every therapeutic task, risks must be weighed against benefits for each particular clinical situation. Therapy is justified only if the possible benefits outweigh the potential risks in an informed risk-benefit analysis. This decision is dependent on an adequate clinical knowledge of the patient and his or her disease, and knowledge of the treatment pertinent to the specific situation.

Although the term ‘benefit-risk ratio’ is convenient and often used, such an assessment is qualitative not quantitative (in other words, the division of a number for risk by a number for efficacy is rarely actually performed) and hence always involves some element of subjectivity. For a drug that is lifesaving, a greater safety risk is acceptable when it is used in that context. This is the reason dangerous therapeutic drugs are tolerated by the regulators.

A major problem in the assessment of many herbs is the lack of hard data supporting their efficacy. Hence, in the eyes of the authorities the benefit is zero, so any risk is unacceptable, as was the case with kava (despite the clinical data). One positive development in this area would be the acceptance of well-established traditional (historical) use as evidence for efficacy. Nonetheless, the fact remains that, in order to preserve the status of herbs, information about efficacy as well as safety is required.

Patients can sometimes experience a mild adverse effect that they attribute to the herbal treatment. Often this is not the case and dechallenge and rechallenge can clarify this. In highly sensitive patients, it can be best to start with a lower dose and build up. The vast majority of clinical trials (see the many examples in the herb monographs) have demonstrated that the side effect profile of a herbal treatment is about the same as placebo.

A number of milder adverse reactions are predictable on the basis of the known phytochemistry of the herb. More details of such side effects are provided in Chapter 2, but a few key examples follow:

Many other potential side effects have been identified for individual herbs and these are outlined in the monograph section of this text.

In general, allergic reactions to herbs are relatively uncommon. A typical allergic reaction, if one occurs, is contact sensitivity. This is acquired only when contact is made with the plant via the skin or mucous membranes. Most herbs have the potential to cause contact sensitivity; however, herbs that contain either resins or sesquiterpene lactones are more likely to have this property. Contact sensitivity can be manifest on the skin either after topical application or in the oral cavity after contact, such as for a gargle or just ingesting a liquid preparation.

Allergic contact dermatitis due to the presence of sesquiterpene lactones (SLs) occurs for many plant species, especially the Asteraceae or daisy family (Compositae). (Note for readers with a chemistry background: in this plant family, the presence of an alpha-methylene on the lactone group is mainly responsible for allergenicity.) It is often observed that individuals who are SL-sensitive tend to develop cross-reactions when encountering other SL-containing species. Commonly used medicinal plants containing SLs include Arnica, feverfew, globe artichoke, elecampane (Inula helenium) and dandelion.

Some other herbs which can commonly cause contact sensitivity include myrrh (Commiphora), cinnamon (Cinnamomum spp), lime flowers (Tilia), propolis (not strictly a herb) and balm of Gilead (Populus candicans).

Anaphylactic reactions can also occur; however, these are quite rare. The Compositae plants are again often involved and the issues for Echinacea and chamomile are fully discussed in their relevant monographs.

Idiosyncratic reactions are reactions that are peculiar to a single individual or a very small group of people. They are by nature unexpected and unpredictable.

In such cases, unless the herbal formula is rechallenged, it is difficult to be entirely sure that the therapy was the cause. However, when the side effects cause considerable discomfort it may be unethical to rechallenge and in many cases the patient is unwilling to try the same formula again.

We cannot rule out the possibility that many so-called idiosyncratic reactions to herbs could be coincidence or a placebo (or in fact nocebo, which is the opposite of placebo, see later) effect. On the other hand, there are some patients (although they are rare) who appear to react to almost anything. This hyper-reactivity is part of their being unwell and will be revealed during the initial case history taking. For example, if a patient reports a history of ‘unusual reactions to drugs’ or ‘being very sensitive to drugs’ it is prudent to take note of this information and make the assumption that, in this particular patient, a reaction to herbal therapy might also occur.

As a general rule, clinicians need to take great care in treating such patients. This is also true for patients who report multiple food sensitivities. When writing a herbal prescription for such patients, it is probably wise to err on the side of caution and use fewer rather than more herbs. This way, if any type of reaction does occur, it may be easier to isolate the cause. In extreme cases, the practitioner may choose to use just one or two herbs initially and add others as the treatment progresses.

Some specific organ toxicities that occasionally seem to occur in herbal therapy may be idiosyncratic reactions. This especially applies in the case of hepatotoxicity (see above). That is, the particular herb does not contain hepatotoxins, but immunologically driven liver destruction is driven by consistent exposure to that herb.

It is a general principle that one should refrain from giving medicines to a pregnant woman unless clearly necessary. Although some herbs have been used safely by women when pregnant and may thus be seen to have a degree of positive vetting, they should be prescribed particularly carefully in the crucial first trimester when fetal organ development is underway. Although there are very few accounts that link any pregnancy problems to herb consumption, too little is known for any sweeping recommendation. Particular caution should be exercised for plants with alkaloidal principles, strong volatile constituents (notably including pure essential oils and plants with high levels of thujone) and in cases where there is a history of miscarriage or where low back or abdominal pains occur. Toxic herbs should be avoided; see Appendix B for such a list.

A number of other herbs should be particularly avoided in pregnancy. In many popular herb books, the term ‘emmenagogue’ is used to refer to a gynaecological remedy. Probably the most frequent indication for such remedies in earlier times was to bring on delayed menstruation; in other words, many ‘emmenagogues’ were used for birth control, as abortifacients. This was likely to be one of the specialist skills of ‘wise women’ and, theoretically, remedies affecting the gravid uterus are as likely to harm fetal growth as to abort pregnancy.

The pregnancy safety ratings advocated in The Essential Guide to Herbal Safety¹ have been adopted for the monographs in this text. This categorisation is designed to remove the subjectivity from assessing the safety information for herbs during pregnancy. The ratings are self-explanatory, but a further discussion is provided in Chapter 10 (How to use the monographs). However, some herbal clinicians might prefer a simpler approach of a list of contraindications and cautions, and this is provided in Table 5.1 for more than just the 50 herbs detailed in this book.

In regulatory terms, the caution referred to above in the case of pregnancy is also extended to the stage of lactation. Although critical organ development is not threatened, there remains some doubt in many cases about how secondary plant metabolites, many of which pass easily and even preferentially into breast milk, affect the baby. Practitioners should therefore maintain a degree of caution in attending to clinical problems affecting mother and suckling baby. However, it should be pointed out that documented adverse effects for herbs used during lactation are minimal, other than milk reduction or reduced infant feeding. Table 5.2 provides some basic guidelines for commonly used herbs.

The issue of adverse herb-drug interactions (HDIs) is probably the most contentious problem in the understanding of the safe use of medicinal plants. Many texts carry pages and pages of theoretically possible interactions for each herb, often based on unsupported extrapolations from pharmacological data (typically in vitro studies). Such information is not only needlessly cautious it is potentially alarmist. The reader might experience such difficulty managing the complexity of the provided information that he or she might conclude that the only safe option is never to recommend any herbs in conjunction with conventional drugs. In addition to these hypercautions, there is also considerable misinformation in the field. For a full discussion of these issues, the reader is referred to the relevant chapter in The Essential Guide to Herbal Safety.¹

While adverse HDIs can certainly be theorised, clinically relevant examples that can be predicted on theoretical grounds are probably uncommon. One such example is the combination of licorice (Glycyrrhiza glabra, G. uralensis) with thiazide diuretics. Prolonged use of both could accelerate potassium depletion. If the patient was also taking digoxin, then a further adverse interaction could occur, since the toxicity of digitalis glycosides is increased by potassium depletion. However, most important adverse HDIs are likely to be unexpected, with case reports representing the best way to discover them.

Adverse HDIs can be either pharmacokinetic or pharmacodynamic in nature. The above example for licorice represents a pharmacodynamic interaction. With a pharmacokinetic interaction, the herb increases or decreases the activity of the drug by altering its absorption, distribution, biotransformation or excretion, or any combination of these. The best-documented example is St John’s wort, and this issue is fully discussed in its monograph. Pharmacokinetic HDIs can be investigated via clinical studies in healthy volunteers, where the impact of taking a herb on the pharmacokinetics of one or more probe drugs is explored. This information is substantially more reliable than extrapolation from in vitro studies of the impact of a herb extract on cytochrome P450 enzymes. Such extrapolations should not be given any clinical credibility because of the many uncertainties involved.

Appendix C provides a reference table for HDIs, together with explanatory information. The table is designed to be accurate and responsible and is based on a critical assessment of the clinical relevance of the available information. Entries are mainly drawn from case reports and clinical studies. In addition, the various relevant monographs typically contain a more complete discussion of some of the issues summarised in the appendix.

It must be emphasised that most of the interactions listed are only potential in nature, and the existence of the majority of interactions in the table is not supported by definitive data. However, prudence and good practice suggest that these combinations are considered with caution, the level of which is flagged in the table.

Over the past two decades, articles have appeared in the media and in journals attributing significant harmful effects to commonly available herbs. In the main, these articles are either imbalanced, misinformed, poorly researched or based on extrapolations from pharmacological studies that have little or no relevance to normal human use. A few examples follow.

In 1999 a team of scientists at Loma Linda University School of Medicine in California undertook studies to analyse the effects of some popular herbs on the fertilisation process and on sperm DNA.⁴⁶ They used hamster eggs with their outer coating (zona pellucida) removed, exposed them first to the herbs for 1 h and then to human sperm. In a separate test, human sperm cells were incubated with the herbs for 7 days and then tested for viability and the intactness of their DNA. The coding sequence for an important gene in the sperm was also evaluated as a test for mutagenic activity.

At the lower concentrations tested, there was no effect from the herbs in any of the tests. But at the higher concentrations, St John’s wort, Echinacea purpurea and Ginkgo prevented human sperm from penetrating the hamster eggs. Exposure of sperm to Echinacea and St John’s wort at only the higher concentrations resulted in DNA denaturation, and St John’s wort also caused genetic mutation. In contrast, Ginkgo and saw palmetto had no effect.

Despite occasional cautions about the significance of their in vitro (test tube) study, the authors at times appeared to exaggerate the relevance of their findings. For example, they wrote:

The study illustrates several of the pitfalls that can happen in experimental design and interpretation when scientists and clinicians with little training in, or understanding of, herbs attempt research in this field. That the authors of this paper were unfamiliar with herbs is well illustrated by their repeated reference to Echinacea purpurea as Echinacea purpura. More significantly, the study did not specify whether commercial samples of the plants or their extracts were employed. If extracts were used, these are significantly more concentrated than the original herb, and such an omitted piece of information would reflect markedly on any interpretation of the results. No information is provided about any attempts to verify the authenticity of the herbs used or their quality or phytochemical profile. There is also no information on solvents used to extract the herbs. (Different solvents extract different phytochemicals from herbs, a variability which can substantially influence the outcome of any study.) The exception here was saw palmetto, for which the liposterolic extract was used (see the saw palmetto monograph).

Throughout the paper, it was maintained that each herb was tested at an amount that reflected the normal human dosage. But these amounts were dissolved in only 1 mL of medium instead of the much greater amount of medium required to reflect the normal quantity of human body fluids. As a consequence, the concentrations tested far exceeded typical human exposure to these herbs. The no-effect level was described as representing ‘very low’ concentrations. In fact, these ‘very low’ concentrations were equivalent to exposure well above normal human intake (which would suggest that, far from implying harmful effects on fertility, the study found that the herbs were safe). The maximum meaningful concentration for in vitro research on whole herbs is considered to be about 0.1 mg/mL on a dry weight basis. In contrast, the ‘low’ concentrations used in the study were in fact high at 0.06 to 0.9 mg/mL and the ‘high’ concentrations (referred to as ‘normal’ in the text) were excessively high at 0.6 to 9 mg/mL. Furthermore, if these values represent the concentrations of extracts, as opposed to dried herb (as was the case with saw palmetto), then the equivalent dried herb concentrations would be considerably greater still.

But there is an equally important consideration that the authors failed to take into account. Many pharmacological screening programmes using in vitro tests on herbs, such as the programme of the National Cancer Institute in the USA, first remove interfering compounds like tannins. Tannins, which are relatively common in herbs, can bind non-specifically to proteins such as enzymes, effectively producing false positive results. In addition, tannins have very low bioavailability, so any observed non-specific activity on intracellular enzymes is likely to be little more than a scientific curiosity. This consideration was well demonstrated by an elegant but perhaps ultimately futile study published in the Journal of Ethnopharmacology.⁴⁷ Owen and Johns found that the inhibitory activity of 26 plants on the enzyme xanthine oxidase was positively correlated with their tannin content (p<0.001).

There are many other cautions which govern the extrapolation of in vitro research to living organisms. In the case of herbs, these include the digestive modification, absorption, metabolism and elimination of the many compounds found in plants. To their credit, the authors of the hamster cell study do occasionally acknowledge these cautions. For example, they write: ‘The possibility of a lack of a physiological effect in vivo should be considered’.

The international media response to this article was extraordinary, considerably assisted by a press release from the journal in which the editor at the time was quoted as saying: ‘This is a very important study that could provide important information to patients suffering from infertility’.

Perhaps the most bizarre episode of a herbal false alarm was the reported occurrence of the toxic alkaloid colchicine in Ginkgo biloba. A group of US scientists investigating natural anti-inflammatory substances in pooled placental blood found a compound present that they identified as colchicine.⁴⁸ Naturally the scientists were curious as to the origins of this plant chemical, so they decided to examine individual blood samples from 24 pregnant women. Only five of these contained colchicine and the group reported that all of the colchicine-containing blood samples came from women who used herbal supplements. For some undisclosed reason they next tested Ginkgo and Echinacea products from local retail outlets and found that only Ginkgo contained significant levels of colchicine (26 µg per tablet). The authors concluded that due to its potential harmful effects, it would appear that Ginkgo biloba should be avoided by women who are pregnant or are trying to conceive.

Colchicine is a drug mainly used to treat acute gout and was originally isolated from the autumn crocus (Colchicum autumnale). It is highly toxic and the effective dose is quite close to a toxic dose. Toxic effects include nausea, vomiting and bone marrow suppression.⁴⁹ Colchicine inhibits normal cell division (mitosis) and is linked to Down’s syndrome. Clearly, it is contraindicated in pregnancy.⁵⁰

Before critiquing the actual scientific merit of the study, some commonsense, logical questions were not answered by the study (and obviously not asked by peer reviewers). These include:

Questions about the research from a scientific perspective might include:

In fact colchicine has never before been reported in Ginkgo and a comprehensive literature search by the late Professor Farnsworth at the University of Chicago found no evidence of colchicine in Ginkgo.⁵¹ According to Professor Farnsworth: ‘Based on biogenetic considerations, colchicine should never be found outside of the Monocotyledoneae (Araceae, Liliaceae) and the report of its occurrence in Saussurea sacra (Asteraceae) is an anomaly that has not been duplicated by other reports on the chemistry of this species. Thus, colchicine has never been reported as a normal constituent of Ginkgo biloba nor would it be expected or predicted to be present’.

Farnsworth also questioned the scientific validity of the study and the editorial process it underwent prior to publication. ‘Anyone who thinks that colchicine can be found naturally in Ginkgo is not qualified to be a peer reviewer of this paper,’ he said, referring to the editorial process for scientific journals in which papers are reviewed by independent experts to determine their scientific merit and the accuracy of their conclusions prior to publication. Schwabe, the German company that researched and developed Ginkgo extract, tested three separate samples of Ginkgo leaf and failed to find any colchicine. This finding was backed up by industry testing in the USA.⁵¹

In response to the study, the Australian Therapeutic Goods Administration tested five Ginkgo biloba products. Colchicine was not found in any (detection limit 1 µg/g). Interestingly, using a method similar to the study in question, a substance was found in the Ginkgo products which had similar analytical properties to colchicine. Although its identity was not determined, further analysis demonstrated conclusively that it was not colchicine.⁵²

Were the levels of colchicine in Ginkgo responsible for those reportedly observed in the pregnant women? The scientists claimed to have found 49 to 763 µg/L of colchicine in the placental blood of the women allegedly taking herbal supplements.⁴⁸ They also claimed to have found 26 µg of colchicine per Ginkgo tablet. In a multiple dose study of the pharmacokinetics of colchicine, 1 mg/day achieved plasma concentrations in human subjects in the range of 0.3 to 2.5 µg/L.⁵³ Hence to achieve a level of 49 µg/L of colchicine (the lowest value in the reported range) a person would need to consume around 50 mg of colchicine per day. Since each Ginkgo tablet contains only 26 µg, this equates to around 2000 Ginkgo tablet per day. In their defence, the authors claimed that placental tissue is known to concentrate ingredients from the mother’s blood. But even assuming a concentration factor of 50, this is still 40 tablets per day to achieve the lowest reported concentration of colchicine, when the normal dose is typically 3 to 4 tablets per day.

Perhaps most astounding of all, the reported levels of colchicine would have been lethal to the unborn children. Several cases involving suicide by the ingestion of colchicine tablets have been reported in the literature. In one case, the plasma level of colchicine 24 h after ingestion was 4.5 µg/L,⁵⁴ in another the femoral blood level was 62 µg/L.⁵⁵Colchicum autumnale is the richest plant source of colchicine. Yet a case report of a man who consumed 17.1 g of flowers found that his maximum colchicine level was just 4.34 µg/L, which occurred 13 h after ingestion of the flowers.⁵⁶ Nonetheless he was hospitalised with nausea, vomiting and abdominal pain.

The authors and the journal involved made sure that their study received due attention by releasing their findings to the press. And despite the many shortcomings of the study the press responded. Press releases and articles ran headings such as:

But perhaps best of all was the quote attributed in London’s Daily Mail to a UK professor working in the field of complementary medicine who was quoted as saying that the alleged Ginkgo health risk was ‘a disaster waiting to happen’ and likened it to ‘another catastrophe like thalidomide’. Despite criticism from the herbal industry, the scientists and journal editor involved in the publication of the study were reported to be unrepentant.⁵⁷

On April 10, 2003, an Australian newspaper (the Brisbane Courier-Mail)⁵⁸ ran a brief article relating concerns raised by recent research that the herb black cohosh may increase the toxicity of two chemotherapy drugs. The article went on to say: ‘Many women diagnosed with breast cancer use the herb as an alternative to hormone replacement therapy’. Similar alarmist reports of this research were also carried in the US media.

The article was in response to a press release of research findings that were to be presented at the American Association for Cancer Research meeting in Toronto, but the event was cancelled due to concerns over SARS in that city.⁵⁹ Notwithstanding the lack of an appropriate forum for the presentation and discussion of their findings, the authors pressed ahead and released their findings via the media.

The first paragraph of the release read: ‘Women with breast cancer who are undergoing chemotherapy may want to avoid black cohosh …’ The press release then went on to describe that black cohosh extracts appeared to increase the toxicity of the commonly used chemotherapy drugs doxorubicin and docetaxel when tested on cultures of breast cancer cells. Hence, this finding was based on in vitro or test tube research. Black cohosh did not potentiate the toxicity of a third drug tested, cisplatin. No exact details of the experiments were provided. The initial paragraph of the press release seemed to contradict a later statement that: ‘Results in laboratory cells may not mimic what happens in the body ….’.

Publication by press release is a growing and controversial trend in scientific circles. But the central problem with this story is that it makes much ado about very little. The reality is that, as already noted, test tube research on herbal extracts is fraught with difficulties and any extrapolations to human use need to be heavily qualified, and certainly do not deserve the type of attention by the media which they typically receive. This case is certainly an example of in vitro non-veritas.

Inaccuracies regarding the true risks of HDIs have already been discussed. However, it is worth noting the alarmist issues raised in a recent paper published in the Journal of the American College of Cardiology.⁶⁰ As one example, concerns over alfalfa and fenugreek interacting with warfarin were expressed; these are not supported by sound science. To phytochemists, the term ‘coumarin’ means plant chemicals based on the coumarin structure. To pharmacists the term ‘coumarin’ means anticoagulant drugs derived from or related to coumarin phytochemicals. There is no solid evidence that normal plant coumarins have anticoagulant activity (see Chapter 2).

The article was riddled with other errors, such as confusing Aloe vera gel and juice with Aloe vera resin. Many of the stated ‘commonly used’ herbs are not used at all in the US and are known to be toxic, such as oleander, Chinese toad venom, Strophanthus, gossypol and khella. The vast majority of the other potential HDIs described had no clinical data to support them.

The authors claimed green tea interacts with warfarin purely because it is a leafy material (and hence contains vitamin K). Saw palmetto was linked to bleeding and Echinacea to causing arrhythmias, but there is no sound evidence to suggest any such problems with either herb (see the relevant monographs). The irony is that the authors complained about public misinformation about herbs! Despite a detailed and well-credentialled response from the American Botanical Council pointing out the many inaccuracies,⁶¹ the journal was unrepentant.

It appears there is indeed a media bias in the reporting of herbal studies. In late November 2008, a study investigating the print media coverage of clinical trials was published in the open access journal BMC Medicine.⁶² Public health researchers from Canada investigated the nature and tone of newspaper reporting of clinical trials of herbal medicines compared with reporting of clinical trials of drugs used to treat the same medical conditions. Databases were searched for newspaper articles in the UK and Ireland, the US, Australia/New Zealand and Canada covering the period from January 1995 to June 2005. The clinical trials were identified and sourced. The study was limited to newspaper articles that were directly related to peer-reviewed clinical trials.

The researchers coded 57 trials on herbal medicine mentioned in 352 newspaper articles and 48 drug trials mentioned in 201 newspaper articles. The main comparisons were tone (positive, neutral, negative), quality measures and content between the herbal medicine and drug clinical trials. Newspaper articles on the pharmaceuticals were more likely to be positive or neutral, but those on herbals were likely to be negative (more so than implied by trial results). No report of a pharmaceutical trial was negative, 22% of the herbal reports were. Clearly identified researchers or practitioners expert in herbal therapy were quoted in only 8% of the herbal articles.

It may be easier to publish reports in medical journals on the risks of herbal remedies than it is to prepare a publishable account of their efficacy. In the first case anecdotal evidence is the norm, in the latter case it would be dismissed out of hand. It is therefore incumbent on authors who rely on anecdote (and editors who review the results) to be assiduous in their presentation of the evidence (to get the botanical details correct, for example), to look for other possible factors in the pathogenesis and consistently to insert caveats about their inability to generalise from their evidence.

At the end of the day, risk assessment is more a political than a scientific discipline. The BSE crisis in Europe led to a marked increase in sensitivity to risk, with food policy measures being made in that case on the basis of risks less than one in many millions. Yet it is inconceivable that truly dangerous activities like driving and smoking would ever be banned, although each may be curtailled to some extent. Unfortunately, for the same reason, herbs as medicines are vulnerable; although political weight is apparent when there have been serious threats to their availability (in Britain, Germany and the US, for example), phytotherapy is less able to muster strong political reasons to save its outlying positions. A question of definition arises in these discussions. What does toxicity actually mean in real life?

Many ordinary foods naturally contain poisonous constituents:

The foods listed above are considered safe to eat, most in unlimited quantities. However, they do demonstrate how difficult it is to predict the toxicity of a plant from the presence of toxic constituents alone. The action of the whole plant and the way in which it is normally consumed clearly in these cases count for more than any individual constituent list. It is at least possible that medicinal plants impugned for their inclusion of toxic constituents are in the same position. In the absence of clear reports of adverse side effects, it can be argued that the burden of proof falls on those who would restrict access to popular remedies.

What is much less debated are negative placebo effects. In reports on most double blind clinical trials there are accounts of adverse effects among the placebo group. The symptoms listed cover a wide range and are not necessarily short-term and transient. In one report of studies of 1228 healthy volunteers the overall incidence of adverse events during placebo administration was 19%. Complaints were more frequent after repeated dosing (28%) and in elderly subjects (26%). Overall, the most frequent adverse events were headache (7%), drowsiness (5%) and asthenia (4%), with some variation depending on study design and populations.⁶⁴ Pain can certainly be induced by placebo⁶⁵ and placebo can also interact negatively with other medications.⁶⁶

The nocebo effect raises an intriguing prospect. It is quite likely that some adverse effects ascribed to a remedy are as non-specific as placebo-induced efficacy. Most of these nocebo effects are likely to be clinically insignificant but there are real possibilities that more serious adverse effects could also have nothing directly to do with the remedy itself.

Unfortunately, placebo-controlled safety studies are not feasible but it will be helpful in assessing overall herbal safety to have some better understanding of the potential of the nocebo effect to confound the picture. For a more extensive discussion of this phenomenon, the reader is referred to The Essential Guide to Herbal Safety.¹

In some cases (perhaps many), an adverse effect attributed to a herb might only represent a coincidental association. One possible example is the hepatotoxic IDRs attributed to black cohosh. Research in several countries has found that one of the most serious causes of liver damage is unknown. Around about one-third of all liver transplants are due to a disorder known as idiopathic or non-A non-B hepatitis.67,68 The demographics of idiopathic hepatitis (female, late 30s to early 50s) and potential black cohosh use strongly overlap. Hence, there is a distinct possibility that some patients who develop idiopathic hepatitis might also coincidentally be taking black cohosh, given that this herb is now so popular. The herb could then be mistakenly attributed as the cause. Once one mistaken case is described in the literature, however poor its quality, it is likely that others will follow in a process akin to a self-fulfilling prophecy.

A coincidental association is more likely to occur when the quality of reporting a herbal adverse reaction is poor and does not take into account all possible causative factors. (Another problem with most herbal adverse reaction reports is that the identity of the herb purportedly involved is not identified, as is discussed elsewhere in this chapter).

Poor-quality case reports might not only incorrectly attribute a harmful effect to a herb, but they can also hamper the value of case reports as red flags for genuine adverse events. One group recently developed an instrument for assessing the quality of case reports, based on a point-based rating scale incorporating 21 items yielding a total possible score of 42.⁶⁹ A review was undertaken of adverse reports for herbal products over three periods, 1986 to 1988, 1996 to 1998 and 2006 to 2008. In total, 137 case reports were included. The percentage of high-quality case reports (scoring 29 or more) rose from 0% in 1986 to 1988, to 27.9% in 1996 to 1998 and 34.2% in 2006 to 2008. This study demonstrates that, while the quality of adverse herbal case reports is improving, there is still a long way to go.

Given all the above, the issue of herbal safety is indeed a matter of contention. The following are the arguments marshalled by the doubters, particularly those who feel that their duty to public safety is to assume the worst-case scenarios.

None of the arguments presented above are mutually contradictory. An honourable position is possible on either side.

Given the low profile of adverse effects compared with other human activities, however, it could be argued strongly that herbal use should be given the benefit of the several doubts that remain. The burden of proof should lie with the regulator rather than with the herbal sector; that theoretical reasons for impugning safety (such as the presence of constituents with toxic potential in isolation) are not sufficient in themselves for banning herbal remedies. Nevertheless, the herbal sector has a professional responsibility to monitor assiduously and continuously the situation and to react responsibly if genuine concerns arise (see also the discussion of pharmacovigilance later in this chapter).

Medicinal plants are sourced from nature and hence, unlike conventional chemical drugs, will vary from batch to batch. This can be readily understood by comparison with another plant product – wine. In technical terms, wine is the fermented juice of the fruit of Vitis vinifera. However, factors such as the grape variety, climatic conditions, soil type, time of harvest and fermentation conditions can all determine whether a batch of wine will be either poor or good quality (which is subsequently reflected in the price of the wine).

In the case of wine, factors such as the texture, colour, aroma and taste determine if the product is of good quality or otherwise. For medicinal plants, the situation is more complex. Most of the chemical components of herbs that are important for therapeutic activity are secondary metabolites. Secondary metabolites are defined as those not likely to be important for normal growth and survival of the plant, although they are sometimes produced in higher levels in response to infection, insect attack and adverse growing conditions. One consequence of this is that, even though they may give an impression of its quality, the appearance and colour of a herb are not necessarily indicators of its therapeutic benefit or safety. Indeed, plants grown under adverse conditions may sometimes have a poor appearance, but higher levels of secondary metabolites. One corollary of this is that while herb batches that have a good appearance and a pleasant taste might be suitable for a herbal tea, they may not be optimum for use as medicines. Chamomile is a good example. Matricine is considered to be an important active component and varieties of chamomile have been bred that contain higher levels of this phytochemical. Since matricine imparts a bitter taste to the flowers, these high medicinal grade varieties are not suitable for a culinary herbal tea.

One approach adopted by the various pharmacopoeias and used by manufacturers is to set minimum levels of marker chemical compounds for a herbal raw material. These are seen to give some indication of activity and hence quality. This approach, although valid, is fraught with difficulties. Even where the marker compound is known to contribute to the therapeutic activity of the herb (and this is not always the case), herbal clinicians stress that the chemical complexity of the plant confers the sum total of its activity. However, until there is better understanding of how individual herbs work in their chemical totality, setting markers is a good beginning. The uncertainties can be lessened by choosing phytochemical classes of marker compounds (flavonoids, essential oil, oligomeric procyanidins and so on) rather than just single chemical constituents. Along these lines, testing for a range of marker compounds (or groups of marker compounds) in the one plant might lead to a better assessment of activity. However, none of this should occur at the expense of the chemical totality of the plant’s extractable material.

In a number of countries, including those in Europe as well as Japan and Australia, all herbal medicines must be made according to the code of pharmaceutical GMP (Good Manufacturing Practice). The US now requires compliance with ‘GMPs’ for dietary supplements, but the standard applied falls short of pharmaceutical GMP in some respects. Pharmaceutical GMP is a fail-safe system of quality assurance and quality control which defines a number of procedures and observances including:

In practice, herbal manufacturing under GMP is more complex than conventional drugs. This is because a herb is biologically rather than chemically defined and:

Nevertheless, all the considerations above point strongly to the importance of herbal product manufacture under appropriate GMP.

As part of GMP, herbal raw materials are subjected to a battery of tests to ensure their quality and purity. These tests are outlined in Box 5.1. A useful guide to the British and European standards in these areas is provided by the British Herbal Pharmacopoeia 1996,⁷⁰ together with the current British and European Pharmacopoeias and the United States Pharmacopeia – National Formulary.

Thin layer chromatography (TLC) is a particularly useful technique for the identification of plant material. It can also be used to quantify plant constituents. The process of performing TLC is outlined in Box 5.2.

Finished herbal products also need to undergo testing before their release. Boxes 5.3 and 5.4 provide examples of possible testing protocols for finished products.

Since plants come from nature, herbal raw materials carry a microbial burden which needs to be reduced during processing. Although microbial contamination of herbs is a potential safety issue, there is relatively little evidence that is a significant issue in practice.

The European Pharmacopoeia sets limits for microbial contamination for herbal remedies in Category 4, Section 5.1.4. Microbial Quality of Pharmaceutical Preparations (1997) as follows.

These levels are incidentally much tighter than those applied to food supply industries where absolute levels are often absent.

Substitution of one herbal raw material with another can pose a considerable problem in herbal manufacture. Substitution can occur at several different levels:

Many problems associated with herbal quality have been caused by substitution. This particularly threatens remedies that have suddenly become popular so that demand exceeds supply, such as the previous substitution of Echinacea with Parthenium integrifolium and the replacement of Arnica species with Mexican arnica (Heterotheca inuloides). It is particularly serious where safety is compromised, for example the substitution of Teucrium species (germander) for Scutellaria lateriflora (skullcap), as noted previously.

Some substitution practices may avoid detection at manufacture, even under GMP, and are occasionally detected in over-the-counter products in some countries. However, the knowledge of a specific substitution and the application of an appropriate test method can and does expose such practices. Despite the substitution of Stephania by Aristolochia being implicated in renal disease, apparently this practice still commonly occurs.

Herbal products can sometimes be contaminated with other agents and a few isolated instances have raised public health concerns. These include:

The practice of pharmaceutical GMP in a well-regulated environment should eliminate the possibility of these types of contamination.

As mentioned above, one persistent quality problem that can dramatically impact on safety is the adulteration of herbal products with undeclared conventional drugs. Examples of the adulteration of Chinese herbal medicines with synthetic drugs are provided in two reviews.^71,72 A wide variety of agents including corticosteroids, non-steroidal anti-inflammatory drugs, analgesics, benzodiazepines, anticonvulsants and hypoglycaemic drugs have been found.

Cases of adrenal suppression have been linked to the intake of Chinese herbal products. A Taiwanese study found that 8 of 13 patients with severe illness and low cortisol levels reported using herbal products.⁷³ Two cases of adrenal suppression were reported in New Zealand and attributed to the intake of the Chinese herbal product Shen Loon.⁷⁴ The product was later found to contain the corticosteroid betamethasone, although the authors suggested other factors could be involved as well.

Undeclared codeine was detected in a Chinese antiasthmatic proprietary product.⁷⁵ One phenomenon which has fortunately received a high degree of media attention is the adulteration of Chinese weight loss products with banned weight loss drugs such as fenfluramine. This has led to toxic reactions or fatalities in the UK, Singapore, Japan and China.76–79

This problem is still current, with the US FDA recently issuing a recall notice on a herbal slimming capsule contaminated with sibutramine, a controlled substance that was withdrawn from the US market in October 2010 for safety reasons.⁸⁰

A particularly cautionary tale is that of PC-SPES. PC-SPES was a herbal formulation specifically targeted at the treatment of prostate cancer (hence PC) developed and patented in the early 1990s by a research chemist.⁸¹ It ostensibly contained seven Chinese herbs and one American herb (saw palmetto: Serenoa repens). The product was very successful in the US marketplace and there were consistent anecdotal accounts of its efficacy, especially for controlling PSA (prostate specific antigen) levels. In particular, naturopathic physicians and holistic medical doctors recommended the product to many patients. Being a US-sponsored product they had no reason to believe it was anything other than a herbal product, despite the fact that it was manufactured in China.

PC-SPES soon began to attract the interest of well-respected research scientists including those at Johns Hopkins School of Medicine, Harvard Medical School and the National Center for Complementary and Alternative Medicine.⁸¹ In 2002, three reviews of the use of PC-SPES were published which surveyed the major publications on its pharmacology and clinical activity.82–84 By early 2002 there were 116 published clinical and laboratory-based studies of PC-SPES.⁸⁵

The reviews highlighted the significant in vitro and in vivo activities of PC-SPES in prostate cancer models and noted the small number of positive but preliminary clinical trials.82–84 Oestrogenic activity was confirmed for the product, which was consistent with some of the side effects exhibited by patients taking the formulation such as gynaecomastia, loss of libido, decrease in body hair and superficial thrombosis. However, there was also another worrying and paradoxical side effect of PC-SPES: it was linked to severe bleeding in one case⁸⁶ and suspected of potentiating the effects of warfarin in another.⁸⁷

Another perspective on PC-SPES began to emerge in 2002 at around the same time the above reviews were published. In early April a group of scientists presented their findings to the 93rd Meeting of the American Association for Cancer Research that PC-SPES samples from 1996 to 2001 contained the oestrogenic drug diethylstilbestrol as well as warfarin and indomethacin.⁸⁸ Their comprehensive results were published later that year.⁸⁹ Earlier in February 2002, the FDA alerted consumers to stop taking PC-SPES because the California Department of Health Services had detected the presence of warfarin. Another product called SPES from the same corporation had been found to contain the anti-anxiety drug alprazolam.⁹⁰ The manufacturer undertook a voluntary recall and PC-SPES was subsequently withdrawn from the market.⁸¹

Given that so few herbal extracts have ever been submitted to extensive preclinical studies, the toxicological information about herbal remedies is generally sketchy. In European legislation the principle has been accepted that drugs already widely used in the population need not go back to the laboratory for new toxicological studies. Instead, those supplying such remedies are under increasing burden to establish safety on an ongoing basis; postmarketing surveillance or pharmacovigilance are terms generally used to describe this monitoring process. There is an overwhelming mathematical case for systematising observations of adverse reactions. Statisticians have calculated that if, for example, an adverse event occurred in as many as 1 in 1000 cases, one would need to see 3000 cases to have a 95% probability of spotting one.⁹¹ This ‘law of three’ makes the individual practitioner of little use in identifying even substantial risks.

All modern medicine legislatures have drug monitoring administrations and encourage physicians to report adverse events they encounter. In some countries, reporting to the authority is mandatory and various forms are produced for the purpose. There is thus in most countries a database of adverse drug reactions (ADRs), which contribute to the warnings added to drug labels and datasheets. A very small number of ADRs relate to herbal remedies, but for a number of reasons (see below) these have been poorly evaluated.

In the cases where analysis of the data has been done, reports of adverse effects from human consumption of herbs appear to be relatively infrequent, making up around 1% of spontaneous reporting.⁹² There is a widespread concern among the regulatory authorities, however, that these cases represent only a small proportion of such incidents. Spontaneous reporting for even conventional prescribed drugs is an inefficient mechanism, with reporting rates among physicians as low as a few per cent of actual cases.⁹³ In the case of herbal remedies, so many of which are self-prescribed or are prescribed by non-doctors, there is much less opportunity for doctors to become aware of this factor in adverse events. Reporting mechanisms for non-doctors are rarely in place and even pharmacists have only variable opportunities or tradition for notifying adverse events. Even where reporting is possible, there are other reasons why spontaneous notification is likely to be particularly ineffective for herbal remedies (Box 5.5).

On the other hand the herbal sector can claim that uncontrolled observations of adverse events are likely to be misleading for the case of herbal safety, and even the regulators accept that there are particular problems (Box 5.6).

In this context it is an encouraging development that most countries now include herbal adverse reaction reporting in their respective databases, often with a formal requirement on manufacturers to report their knowledge of such events to the relevant statutory authority. Such adverse reaction data from government or WHO databases that have occasionally been drawn on for safety information regarding individual herbs and, where available, in published form, are included in the relevant monograph in Section Three.

In addition, studies are published from time to time that examine the totality of case reports on a specific database. One relatively recent example is from Sweden that drew on adverse reactions submitted to the Swedish Medicinal Products Agency between 1987 and 2006.⁹⁴ Among a total of 64 493 reports, 778 concerned 967 suspected adverse reactions related to 175 complementary medicine products. Of these 967 adverse reactions, the most reported substance was Echinacea purpurea singly (8.1%) and in combination (7.3%), with Ginkgo biloba leaf next at 6.7%. In 221 reports, at least one reaction was categorised as serious, the frequency of these being pulmonary embolism (1.7%), liver reactions (2.8%) and anaphylactic reaction (2%).

For a detailed discussion of the pharmacovigilance of herbal medicines, see Chapter 11 of The Essential Guide to Herbal Safety.¹