This review examined case report evidence regarding HM-associated mania. Those included were: St John's wort (Hypericum perforatum); Ginseng (Panax Ginseng); Brindleberry (Garcinia cambogia); ma-huang (Ephedra sinica); “herbal slimming pills”; Herbalife products; Hydroxycut; horny goat weed (Epimedium grandiflorum); “herbal body tonic”; celery root (Apium graveolans), and “herbal mixture.” Where possible the candidate pathophysiological mechanisms are discussed in turn, as are other factors which may have contributed to the onset of mania in the individuals included in the study. There is an inherent difficulty in making attributions regarding the causality of HM on mania as the course of bipolar disorder is unpredictable.

Of the 35 case reports included in this review, 5 were isolated reports of one HM and two cases (both “herbal slimming pills”) were attributed to a contaminant. The remaining 28 cases were accounted for by five HM, all of which were the subject of two or more case reports of mania. Over a publication period of 38 years (1980–2017) this is a small yield of reported concurrence of HM usage and mania. The small numbers and unreliability of discerning and reporting a link between mania and HM preclude any definitive statement as to whether any association is causal or coincidental.

With respect to St John's wort, there is a high specificity of the stated reason for taking the HM to be for treatment of depression (11 of 14 St John's wort cases compared to 0 of 21 other HM). There was also greater morbidity in the psychiatric history of the St John's wort cases (13 of 14 cases compared to 11 of 21 for other HM), in the family history of mood disorder (6 of 14 compared to 2 of 21) and concurrent antidepressant prescribing was more common (4 of 14 compared to 4 of 21). Competing explanations for these patterns include an increased diathesis toward bipolar disorder, fluctuations of established affective illness, antidepressant-associated mania and HM-associated mania. These factors are not mutually exclusive, for instance according to Craddock and Sklar, a family history of bipolar disorder is an important clinical predictor of a likely bipolar course in a patient who presents with one or more episodes of depression even before their first personal episode of mood elevation (55).

The mechanism by which St John's wort may alter susceptibility to mania is not well understood. As with antidepressants, it is difficult to distinguish spontaneous episodes of mania from St John's wort-associated switching (56). Angst et al. analyzed the time course and risk factors for a diagnostic change from major depression to bipolar disorders over an average of 20 years from onset. Diagnostic change from depression to bipolar type I occurred in approximately 1% and bipolar type II 0.5% of patients per year (57). In patients with major depressive disorder treated with antidepressants, it has been found that antidepressant-associated mania or hypomania occurs at an average frequency of 3.42% of cases per year, but it is unclear to what extent switching represents undiagnosed bipolar disorder or a direct pharmacological effect of antidepressants (58).

Depression is one of the most commonly cited reasons for using CAM (59). The prevalence of depression in the United States has been reported to have increased between 2005 and 2015 (60). For many patients with depression, HM which are in many countries, predominantly available over-the-counter, may be an attractive alternative to conventional medicines. There is a substantial evidence base from randomized controlled trials supporting the use of St John's wort in mild-to-moderate depression. In a systematic review and meta-analysis comprising 23 randomized trials of St John's wort in outpatients with mild-to-moderate depression (N = 1757), it was found that Hypericum perforatum extracts were significantly more effective than placebo (61).

St John's wort has a variety of actions that may contribute to its therapeutic effects. In vitro, it acts on neurotransmitter regulation, including beta adrenergic and glutamate receptors, and ion channel conductance. Hypericin (an active constituent of St John's wort) inhibits serotonin reuptake, and 5-HT_1A and 5-HT_1B receptor changes are associated with prolonged use (62). According to Fahmi et al, in animals, hypericum is effective in three major biochemical systems relevant for antidepressant activity including inhibition of synaptic reuptake of serotonin, noradrenaline and dopamine (26). In the human case reported by Barbenel of concurrent prescribing of St John's wort and sertraline in a patient following surgery for crypto-orchidism, alteration of testosterone and gonadotrophin levels and interactions of antidepressant and St John's wort were further considerations.

With five case reports of Ginseng associated mania it was the second most commonly reported HM after St John's wort. Based on the belief that it is a panacea and promotes longevity, Ginseng root has been used for over 2000 years (63). There are a number of plants that share the common name Ginseng however only three of these are from the genus Panax (Panax ginseng, P. notoginseng and P. quinquefolius). Other “ginsengs” include Siberian (Eleutherococcus senticosus), Indian (Withania somnifera), and Brazilian (Pfaffia paniculata) (64, 65). The most important constituents of Panax ginseng are the ginsenosides, of which 15 different types have been identified (35).

In a systematic review of RCTs examining the efficacy of Panax ginseng root extracts for a number of indications, it was concluded that there is contradictory evidence that Ginseng improves physical performance and immunological measures. It may have beneficial effects on psychomotor performance and cognitive behavior. No trial has confirmed the alleged age-delaying properties of Ginseng. Results suggesting a reduction of blood glucose levels in type-II diabetic patients require further investigation (64, 65). With Panax ginseng, two mechanisms of action in depression have been advocated, firstly, an activating effect of ginsenosides on the HPA-axis resulting in elevated corticotropin and corticosteroid levels (33). Secondly, monoamine signaling could also be affected by ginsenosides (66, 67).

Thirteen of 35 case reports in our study sample involved adverse psychiatric effects of weight loss products including brindleberry (Garcinia cambogia), ma-huang (Ephedra sinica), “herbal slimming pills,” Herbalife products and Hydroxycut. The social stigma of obesity, a desire to lose weight without making drastic lifestyle changes, and frustration at previous failed attempts are commonly reported reasons for using dietary supplements which are readily available and advertised as being “natural” and safe (68).

In one 12-week randomized placebo-controlled trial, Garcinia cambogia failed to produce significant weight loss and fat loss beyond that observed with a placebo (69). In contrast, another double-blind placebo RCT found that Garcinia cambogia reduced abdominal fat accumulation in participants (70). Other human research has confirmed the potential of Garcinia cambogia/HCA in stimulating fat oxidation, increasing serotonin release in brain cortex and normalizing lipid profiles (71). The main active ingredient of Garcinia cambogia is hydrocitric acid which has serotonergic effects and has been implicated in serotonin syndrome (38). Hydrocitric acid is the putative mediator of this HM weight loss effect; it is thought to promote the release and synaptic availability of serotonin thus influencing appetite (41). The effects of Hydroxycut were attributed to the inclusion of Garcinia cambogia in the preparation (49).

Partin and Pushkin, who reported a case of hypomania associated with horny goat weed (Epimedium grandiflorum) proposed that this may have been due to the addition of other unidentified herbs and pharmaceuticals (50). However, these were not explicitly tested for.

Ma-huang (Ephedra sinica) is native to China and Mongolia and contains sympathomimetic compounds known as Ephedra alkaloids. Traditionally used to treat asthma and hay fever symptoms, more recently it has been combined with caffeine or botanical sources of caffeine (for example Guarana) for weight loss purposes (68). In a 6–month RCT of herbal Ephedra/ caffeine for weight loss, it was found that 90/192 mg/day of herbal ephedra/caffeine promoted weight and body fat reduction (72). In another randomized double-blind trial of a herbal supplement containing ma-huang-guarana for weight loss, it was found that the active treatment produced significant effects (73). Ma-huang contains variable amounts of ephedrine congeners which enhance norepinephrine release in central noradrenergic neurons. Ephedrine also has direct agonist activity at alpha and beta-adrenergic receptors (44).

Sibutramine is an appetite-suppressing agent that is a norepinephrine and serotonin reuptake inhibitor, initially developed as an antidepressant (48) its use has been associated with mania (74, 75) and hypomania (76). In a recent analytical study of 447 weight loss products, 119 were found to be adulterated with one or more weight loss compounds including sibutramine, its metabolites benzyl sibutramine and desmethyl sibutramine; phenolphthalein; bisacodyl; furosemide; liothyronine (T3); and thyroxine (T4) (77). This demonstrates the importance of having regulatory bodies oversee CAM manufacturing practices and to regularly assess certain classes of CAM products for adulterants (such as weight loss products). These incidences of mania associated with HM weight-loss products highlight the fact that they may be considered safe and harmless by consumers when they have the propensity to trigger adverse events in vulnerable individuals (78).

Khalid et al. reported a case of mania associated with celery root (Apium graveolans), St John's wort (Hypericum Perforatum) and venlafaxine. Mania ensued shortly after the ingestion of celery root, which belongs to a group of plants classified as the umbelliferous family, which contain phytoestrogens that are structurally similar to estrogen. In this case the patient developed elevated serum venlafaxine levels after taking celery root for menopausal symptoms, suggesting pharmacokinetic potentiation of the venlafaxine level by the celery root as a likely mechanism of induction of mania (52).

Thirteen of the patients were concurrently taking conventional medicines (as shown in Table S1) as well as HM which may have led to herb-drug interactions resulting in mania. This might result from alterations of absorption, distribution, metabolism or elimination of a conventional drug by a herbal product, that is pharmacokinetic effects (79). Alternatively, there may be synergistic effects of a HM and a conventional medicine reflecting common mechanisms of action such as neurotransmitter regulation. Drug dosage is one important factor in such herb-drug interactions as well as in interactions with underlying biological diatheses (80). In approximately one third of the cases reviewed the dosage of the HM was unknown. This limits the ability to make causal inferences regarding the dose of the HM associated with a manic switch in vulnerable individuals.

For the whole sample the time to onset of manic symptoms from commencing the HM (treatment-emergence interval) ranged between 2 days and 2 years with a median of 4 weeks. None of the HM reported associated with mania diverged notably from this median time to onset of manic symptoms. A treatment-emergence interval of 8–12 weeks is deemed to implicate causality; however, a much shorter interval may be necessary to definitively link cause and effect (81). The manic episodes reported in the cases were mostly treated with conventional anti-manic agents, only two reports indicate the outcome of cessation of the HM alone on the course of mania, one indicating remission in 2 days, the other no improvement following a switch into depression. Thus, no general statement can be made with respect to mania resolution on ceasing the implicated HM.

In addition to the systematic review of HM-associated mania, we also assessed the quality of reporting (Table S2) on the modified Quality Assessment Scale by Agbabiaka, in each of the 26 published case report papers. The case report quality assessment score ratings were low quality (0), lower-medium quality (9), upper-medium quality (10) and high quality (7). There are two algorithm-based rating instruments to assign the probability that an adverse event (in this case mania or hypomania) is related to a given exogenous substance, the Naranjo and the WHO-UMC. Of the cases reviewed only one used the Naranjo scale, none used the WHO-UMC. One study compared the two rating scales and found that the WHO-UMC method was more simple and less time consuming compared to the Naranjo probability scale (82). On the 7 additional quality criteria, the quality of reporting was satisfactory only for the inclusion of botanical name of HM (26 of 35 cases) and to a lesser extent HM dosage (18 of 35 cases). On the remaining 5 criteria less than 10% of papers were compliant. It is noted that many of the reports were published before the advent of defined quality criteria ratings, for example 19 of the 26 papers preceded the publication of the modified Quality Assessment Scale by Agbabiaka in 2008. Future published case reports of adverse events should adhere to such criteria in order to improve their overall quality and inferences which may be made from these articles.

The current review of case reports is subject to a number of limitations. Substance/medication-induced bipolar and related disorder may reflect a switch in a previously unexpressed bipolar diathesis. It is possible that the reports in the scientific literature are subject to a confirmation bias in that clinicians are looking for links between mania and recent use of HM. There is also a possible effect of researcher/ publication bias, which has led to the publication of the included studies. Ethically, it is difficult to replicate studies which suggest a potentially harmful effect of various HM.

The compilation of case histories presents a different sample to that commonly seen in a randomized controlled trial setting where fixed inclusion and exclusion criteria apply. This sample of putative HM induced mania includes a number of patients (n = 7) with a stated past history of a diagnosis of BD and a further 7 patients without a BD diagnosis taking antidepressants. This increases the likelihood that the observed manic episode was attributable to extant bipolar disorder or antidepressant induced switching rather than the HM per se, although interactions of these variables cannot be excluded.

Despite these limitations, however, there are a number of strengths associated with the inclusion of case reports including that the patients, episodes of mania, time course of symptoms and aetiological factors are described in detail, as reflected in Table S1. The authors are generally circumspect in their judgements regarding causality, documenting an observed association, summarizing knowledge regarding possible mechanisms of action, and allowing for the multiplicity of explanations that attend a disorder of unclear etiology and pathogenesis and with a typically fluctuating course.

Concepts of exogenous and endogenous causation of psychoses have been debated over the past century. In 1910, Bonhoeffer proposed that the brain only manifested a few stereotyped mental reactions, whether from exogenous or endogenous origins. He recognized delirium in particular as a presentation that could follow diverse exogenous causes. The typical features in delerium of clouding of consciousness and disorientation form the rationale for categorizing it separately to the psychoses (83). Similarly, in the concept of unitary psychosis, as elaborated by Conrad in 1959, there is no fixed relationship between symptom picture and exogenous factors, the latter potentially triggering a range of symptom pictures (84). Relating these propositions to the current study on case reports of mania associated with herbal medicines, the results are inconclusive because it is restricted to one diagnostic category, mania, and one class of aetiological factors, herbal medicines. Comparisons of reports of herbal medicines associated with the onset of a variety of mental disorders would yield information as to whether the relationship indicates specific aetiologies of a disease (e.g., mania) or a broader relationship whereby a diversity of diagnostic categories (e.g., mania, psychotic depression and schizophrenia) are attributable to a common aetiological factor, the unitarian view.

In summary, the reported co-occurrence of HM usage and mania, whilst inconclusive, provides a plausible signal as to brain mechanisms relevant to the pathogenesis of mania. With the increasing storage of health information in electronic health records and evolving techniques of data mining there are prospects for the widespread application of case report level information to the elucidation of associations and causal links between usage of HM and the occurrence and varied course of mental disorders including BD.