St John’s wort flowering herb top (Hypericum perforatum)

Free to every reader. The safety section of every monograph is open to everyone, signed in or not. In an emergency call 000. For poisoning, call the Poisons Information Centre on 13 11 26.

15. Safety

15.1 Overall profile

St John’s wort is well tolerated. Its one clinically important safety issue is pharmacokinetic interaction with prescribed medicines (15.4), which is unrelated to toxicity. Adverse events in trials are mild and uncommon; dropout and adverse-effect rates with extracts are similar to placebo, lower than with older antidepressants and slightly lower than with SSRIs across 35 double-blind randomised trials [7]. Across observational studies of 34,804 patients, reported side effects ranged from 0 to 5.9 per cent, most commonly gastrointestinal, then photosensitivity and skin symptoms [7]. Apaydin and colleagues found adverse events comparable to placebo and fewer than with antidepressants, OR 0.67 (95% CI 0.56–0.81), while noting that reporting was poor and the studies were not designed to detect rare events [18].

15.2 Adverse effects

From trials and observational series [5,6,7]: gastrointestinal symptoms (nausea, abdominal pain, diarrhoea), allergic skin reactions, fatigue, restlessness; headache, palpitations; anxiety in a small number of participants (5 of 2,404 in one series, 8 of 3,250 in another); muscle and joint stiffness, tremor, sweating, muscle spasms and pain in one trial; frequent urination in 27 per cent on 900–1500 mg/day for 8 weeks against 11 per cent on placebo. Frequency not known (EMA) [6]. Extracts appear free of the cardiac and anticholinergic effects of standard antidepressants [7].

Photosensitivity. This is the adverse effect the drug is known for, and the honest account is more qualified than the reputation. Pharmacokinetic studies suggest the phototoxic threshold for hypericin is not reached at normal oral antidepressant doses [7]. A prospective randomised study in 72 volunteers of skin types II and III found no significant influence of LI 160 on erythema or melanin index at single doses of 5.4 or 10.8 mg hypericin or at steady state, with the single exception of a marginal effect on UVB-induced pigmentation (p = 0.0471); hypericin serum and skin levels after oral extract were always below the assumed phototoxic threshold of 1000 ng/mL [7]. Peak hypericin in skin blister fluid after 1800 mg, or 900 mg daily for 7 days, was at least 20 times below the estimated phototoxic concentration of 100 µg/mL [7]. Against that: a measurable increase in erythema after UVA was demonstrated in light-sensitive individuals on 600 mg three times daily for 15 days, at plasma hypericin double the usual therapeutic level [5]; EMA records that after 1800 mg/day for 15 days UVA sensitivity increased and the minimum pigmentation dose fell significantly, while at recommended doses no phototoxicity is reported [6]; and case reports exist of reversible photosensitivity, one after four days at 33 mg extract, one in a woman after three years of use, one in a patient undergoing laser treatment, and three after topical cream or oil [7]. Where phototoxicity has been dose-limiting it was at hypericin doses far above the antidepressant range: in phase I studies of purified hypericin at 0.05–0.5 mg/kg/day, 11 of 23 evaluable HIV patients developed severe cutaneous phototoxicity and 16 of 30 discontinued [7].

Veterinary hypericism is real and is not the same thing. Photosensitisation occurs in cattle grazing the plant; intragastric doses of 3 g/kg or more of dried aerial parts produced photosensitisation in 4–6 month calves within 3–4 hours of sun exposure, while 1 g/kg produced no detectable effect [7]. NSW WeedWise records the agricultural picture: weight loss, fewer pregnancies, stillbirths, weak young, reduced milk, and acute signs within five hours on a sunny day without shade; pigmented, woolly or thick-skinned animals are more tolerant; suckling young can be poisoned through the milk; affected stock should be kept in full shade for four to seven days [15]. The usual therapeutic doses in humans are about 30 to 50 times below the dose needed to induce phototoxicity in calves, and there have been no reliable reported cases of hypericism in humans taking oral doses [7].

Other reported reactions [7]: four cases of sensory nerve hypersensitivity with increased heat and cold sensitivity and pain in hands and feet, resolving in one to two weeks, attributed to late-harvested herb high in resinous constituents; one case of subacute toxic neuropathy; one of hyperaesthesia; one of delayed emergence from anaesthesia after three months at up to 3 g/day; one of cardiovascular collapse during anaesthesia after six months’ use; elevated thyroid-stimulating hormone in a criticised retrospective case-control study; reversible mild liver-enzyme elevation in 26 HIV-positive patients on the equivalent of 1 mg/day hypericin for four months, returning to baseline within a month of stopping; one case of hepatotoxicity; one of dynamic ileus; cases of sexual dysfunction and reduced libido, confounded by concurrent sertraline in one; one report of suicidal and homicidal thoughts linked to vitamin C plus extract, recurring on rechallenge by the patient’s own account. Fourteen reported cases of mania, in patients with and without a psychiatric history, some with concurrent products or drugs [7]; three reports of psychosis, in a patient with Alzheimer’s disease, in two with schizophrenia, and one first episode [7]; and a 2017 case report of St John’s wort-induced psychosis [39]. A suspected withdrawal syndrome after stopping 5.4 g/day of herb for 32 days, peaking at day 3 and resolving by day 8, is reported with the author’s own note that the picture is also consistent with a coincidental infection [7].

The Australian adverse-event record. One report to the Adverse Drug Reaction Advisory Committee in Australia attributed dyspnoea, hyperventilation, palpitations, tremor, flushing, mydriasis and rhinitis to Hypericum in a 47-year-old woman, a picture suggestive of an allergic reaction [7].

TGA safety alerts, checked at medicine review in September 2026 and again in October 2026. The TGA has published on this herb twice and both documents are still on the register:

1. TGA alert to doctors, pharmacists and complementary health practitioners, published 15/03/2001: St John’s Wort: important interactions between St John’s Wort (Hypericum perforatum) preparations and prescription medicines [45]. It names strong evidence of reduced plasma concentrations of indinavir and cyclosporin; reasonable grounds for concern over warfarin, digoxin, theophylline, the other HIV protease inhibitors (saquinavir, ritonavir, nelfinavir), the non-nucleoside reverse transcriptase inhibitors (efavirenz, nevirapine, delavirdine) and the anticonvulsants phenytoin, carbamazepine and phenobarbitone; possible reduced contraceptive efficacy of oestrogen-containing oral contraceptives; and pharmacodynamic interaction with nefazodone, the SSRIs and the triptans. It makes the point that the content of St John’s wort varies between products and that patients may be taking them without their practitioner’s knowledge.
2. TGA information sheet for health care professionals, published 13/03/2000: Interactions of St John’s Wort (Hypericum perforatum) preparations [46], summarising the advice the TGA received from the Adverse Drug Reactions Advisory Committee and the Complementary Medicines Evaluation Committee. It carries the interaction table reproduced in part at 15.4 and adds two points of management that nothing else in this monograph makes as plainly: when a patient stops a St John’s wort preparation, the plasma concentration of the interacting medicine may rise and cause toxicity, so the interacting drug’s dose may need adjusting in both directions; and because marker content varies between products and patients switch between them, the degree of induction changes over time. The sheet also records the classes the TGA considered unlikely to interact: topical medicines with limited systemic absorption, and non-psychotropic medicines cleared principally by the kidney.

Both documents predate the hyperforin dose-response work at 15.4 and neither distinguishes low-hyperforin from high-hyperforin preparations. They remain the Australian regulator’s published position, and a practitioner here should read them alongside, not instead of, the hyperforin figures.

TGA compliance reports on listed St John’s wort products, 2025. Searches of the TGA site in October 2026, October 2026 and October 2026 found no safety alert newer than the two above. The TGA’s register of listed-medicine compliance reports holds three reports on products containing the herb. Two, published in June 2025, are targeted reviews of Cenovis St John’s Wort (ARTG 128698) and Nature’s Own St John’s Wort 2700mg (ARTG 201937), both from Sanofi-Aventis Healthcare Pty Ltd, with outcomes dated 27/03/2025: the ARTG record, labels and website were reviewed, no safety issue was found, no action was taken, and both medicines were permitted to stay in supply [50,51]. The Nature’s Own product was later cancelled from the ARTG at the sponsor’s request, with effect from 16/06/2025 [51]; the notice gives no reason, and nothing on the TGA site connects the cancellation to safety. The third report, on the label warning, is a targeted review of Stress Guard (Pharma Botanica Pty Ltd, ARTG 183814), a valerian and St John’s wort product, with the outcome dated 20/11/2025 [49]. The label of one batch left out the mandatory valerian warning, and its St John’s wort warning neither mentioned oral contraceptives nor carried the referral sentence the STJOHN statement requires (0.1). Supply of that batch had stopped in May 2025, the sponsor corrected the label, and the TGA allowed the medicine to stay on the market. The report names no adverse event. Read together, the three reports show that the TGA checks the STJOHN warning on listed products and expects every batch label to carry the contraceptive wording in full.

The TGA Database of Adverse Event Notifications (DAEN – medicines), searched at medicine review in October 2026 and again in October 2026 [48]. Search term hypericum in the medicines search, which matched 34 products listing Hypericum or Hypericum perforatum as an active ingredient; full date range offered, 01/01/1971 to 23/09/2026, the last date the database then held. Result: 150 reports, 88 of them with a single suspected medicine, and no reported deaths. The most frequent reaction terms were nausea (21 reports, 11 with the St John’s wort product as sole suspect), photosensitivity reaction (13, all 13 sole suspect), serotonin syndrome (12, none sole suspect), diarrhoea (11, 7), and palpitations, abdominal pain and dizziness (10 each). Three reports each name drug interaction, intermenstrual bleeding and an increased prothrombin level, none with the herb as sole suspect. Three things limit what these counts can carry. The 34 products include multi-ingredient tablets and topical and homoeopathic preparations (Traumeel S, Fisiocrem, Brauer Arnicaeze among them), so the total overstates reports for the herb taken orally on its own. A DAEN entry records a suspicion, and the TGA states that inclusion does not mean the event was confirmed or caused by the medicine. And spontaneous reports give no denominator, so they yield no rate. The search was re-run at medicine review in October 2026, when the date range offered ran to October 2026: the same 34 products, 150 reports, 88 with a single suspected medicine, no deaths, and the same leading reaction terms with the same counts as before. Read in that light, the Australian record matches the published profile at 15.2 and 15.4: gastrointestinal upset and photosensitivity where the herb alone is suspected, and serotonergic and induction effects only where another medicine is also present.

15.3 Contraindications

15.4 Herb–drug interactions

Mechanism. Hyperforin is a potent activator of the pregnane X receptor (the human steroid X receptor), discovered independently by two teams in 2000 [7]. PXR activation induces CYP3A4, CYP2B6, CYP2C9, CYP2C19 and, through the MDR1 gene, P-glycoprotein [6,7,29]. The result is faster metabolism and reduced intestinal uptake of a very wide range of drugs. Nicolussi and colleagues state it as the settled position: the degree of CYP3A4 induction correlates significantly with the hyperforin content of the preparation [29]. It is verified experimentally that it is hyperforin, and not hypericin, that increases P-glycoprotein expression in vitro [7]. The elevated enzyme activity returns to normal within one week of stopping [6].

The grading. Each interaction below is graded clinical (human clinical reports or trials), PK study (human pharmacokinetic study), or theoretical.

The dose–response, which is the practical answer. The size of the interaction follows the daily hyperforin dose. The EMA’s herbal committee and ESCOP, the European scientific cooperative on phytotherapy, both draw the boundary at 1 mg a day.

EMA/HMPC and ESCOP separate low-hyperforin preparations (≤ 1 mg hyperforin per day) from high-hyperforin preparations (> 1 mg per day). No clinically relevant pharmacokinetic interaction has been observed for low-hyperforin extracts at dosages giving up to 1 mg of hyperforin per day. [6,29]

The EMA position is read in the EU monograph itself [6]. The ESCOP 2018 position is as Nicolussi and colleagues report it, because AIP does not hold the ESCOP monograph [29]. The EU monograph applies the split in its traditional-use column only. There, its statement that no clinically relevant interactions are reported covers traditional indications 1, 3 and 4 (temporary mental exhaustion, mild gastrointestinal discomfort, nervous restlessness with difficulty falling asleep), at under 1 mg of hyperforin a day, for no longer than two weeks. The well-established-use column covers the extracts for depressive episodes, and it keeps the contraindications at any hyperforin content (15.3) [6]. For a low-hyperforin extract taken for depression over months, the 1 mg figure is evidence that the interaction risk is low. It does not lift the EMA contraindications.

The supporting numbers, each from a separate study and reported by Mills and Bone, are plotted at Figure 10b in the categories the source gives them:

The regulatory consequence: Swissmedic was the first authority to approve the removal of contraindications, warnings and pharmacokinetic interactions for a low-hyperforin herbal medicinal product [29]. The EMA traditional-use column already applies the split [6].

Where this leaves a dispensary. Mills and Bone put it in one sentence: "the most sure and obvious way to avoid the metabolic drug interactions with Hypericum is to use the traditional liquid dosage forms", because hyperforin does not survive in them [7]. That is true and it is also an admission, because if hyperforin contributes to the antidepressant effect, a hyperforin-free galenical is a different drug from the one the trials tested. Section 16 takes that argument up.

15.5 Toxicology and overdose

Acute and repeated-dose toxicity: no signs of toxic effects [6]. Genotoxicity: hydroalcoholic extracts at 0.2–0.3 per cent hypericin gave some positive results in the Salmonella/microsome assay, which EMA assigns to quercetin and judges irrelevant to human safety; all in vivo tests: chromosome aberration in Chinese hamster bone marrow at 10 mL/kg, micronucleus in rodent bone marrow at 2 g/kg — were negative [5,6]. Carcinogenicity: not tested; EMA records that tests on carcinogenic potential have not been published [6]. Fertility and reproduction: a 26-week study in rats and dogs at 900 and 2700 mg/kg found no effect on fertility, embryonic development or pre- or postnatal development [5]; EMA nevertheless records that several studies on extracts and isolated compounds report in vitro and in vivo effects that could affect fetal development [6].

Overdose. EMA: after up to 4.5 g dry extract per day for two weeks plus 15 g immediately before presentation, seizures and confusion have been reported; after massive overdose the patient should be protected from sunlight and other UV sources for one to two weeks [6]. Mills and Bone report the corresponding case: convulsions in a 16-year-old girl who took up to 15 × 300 mg tablets a day for two weeks and a further 50 tablets before admission, with diffuse spike-wave activity on EEG that was normal at discharge on day 6 and no further seizures over six months [7]. Phototoxicity is the expected consequence of overdose: rash, pruritus and erythema 24 hours after UV exposure [7].

15.6 Pregnancy, lactation, children and older people

Pregnancy. EMA, both columns: safety in pregnancy has not been established; animal studies have shown signs of reproductive toxicity; use is not recommended; no fertility data available [6]. Mills and Bone assign Australian pregnancy category B1: no increase in malformations or other harmful fetal effects from limited use in women, and no evidence of increased fetal damage in animal studies other than minor adverse effects at high doses in one mouse study of unknown human relevance [7]. In a controlled comparison, 54 pregnant women taking extracts were compared with 108 pregnant women on other antidepressant therapy or unexposed, and major malformation rates were no different from the 3–5 per cent expected in the general population [7]. Avila, Whitten and Evans reviewed the rodent evidence systematically in 2018 [40]; the full text was not obtained for this draft (paper request 72) and no result from it is quoted.

Lactation. Clinical studies support relative safety. Constituents penetrate milk poorly: in one case study only hyperforin was excreted into breast milk and at a low level, with hyperforin and hypericin below the limit of quantification in infant plasma (hypericin 0.20 ng/mL, hyperforin 0.50 ng/mL), across four milk samples over 18 hours, with no side effects in mother or infant [7]. In five mothers on 900 mg/day LI 160, hyperforin was at the lower limit of quantification in two infants’ plasma (0.1 ng/mL), milk-to-plasma ratios 0.04–0.13 and relative infant doses 0.9–2.5 per cent, with no side effects [7]. Against that, one uncontrolled study found a significantly higher frequency of infant side effects — 16.6 per cent, colic, drowsiness, lethargy — in mothers taking the herb against 0 and 3.3 per cent in two control groups, with no difference in reported milk production or infant weight over the first year; the study was not placebo-controlled and self-selection bias is likely [7]. ESCOP’s position is that the herb should not be used in pregnancy or lactation without professional advice [7]. EMA’s is that use is not recommended [6].

Children and adolescents. EMA: use under 18 is not recommended for the oral indications, for want of data; under 12 for the cutaneous indication [6]. Two paediatric trials exist (NCT00100295 in ADHD, n = 59; NCT00782080 in ADHD, n = 155, a St John’s wort/valerian combination). Mills and Bone report three open-label studies in juvenile depression, none with a control arm: 101 children under 12 on LI 160 at 300–1800 mg/day for four to six weeks; 33 children with major depression of at least moderate severity on 450 mg/day, raised to 900 mg/day in 22, of whom 25 met the response criteria at eight weeks; and an eight-week study of 900 mg/day in adolescents with mild major depression (Simeon and colleagues), in which 9 of the 11 who completed improved and 8 others stopped early [7]. Uncontrolled studies carry no efficacy weight, and none of these changes the EMA position.

Older people. No specific restriction. One comparator trial in 149 patients aged 60–80 found LoHyp-57 at 800 mg/day equivalent to fluoxetine 20 mg/day, with HAM-D falling from 16.6 to 7.91 against 17.18 to 8.11 [7]. The relevant caution in this group is polypharmacy, and therefore section 15.4.

15.7 Driving and machinery

No adequate studies (EMA) [6]. Several reviews and trials report no adverse effect on the ability to drive or operate machinery; in one review of three trials the herb was devoid of sedative or anticholinergic effects and did not affect driving ability on the established psychometric procedure used to assess medical fitness to drive; LI 160 at 900 mg/day for four weeks did not impair vigilance or driving ability in 39 depressed patients; and the same extract preserved attention and reaction in mild-to-moderately depressed patients [7].

15.8 The referral line

Anyone taking a prescribed medicine, and in particular an anticoagulant, an immunosuppressant, an antiretroviral, a cytotoxic agent, an antiepileptic, a statin or a hormonal contraceptive, should not start St John’s wort without first having the interaction assessed. Consult a degree-qualified herbalist with competencies and adequate training in pharmacognosy. In Australia the label of any St John’s wort medicine for oral ingestion must carry the warning statement set by item 2674 of the Permissible Ingredients Determination (No. 3) 2026, in force from 16/10/2026 (item 2672 in the repealed No. 2, with the same wording), which states that the herb affects the way many prescription medicines work, oral contraceptives included, and then directs the reader onward in wording fixed by the instrument [11]. That statutory wording is the Commonwealth’s and it is not AIP’s referral wording; it is reproduced in substance here because it is a legal requirement on the label, and it is not adopted as this monograph’s own advice.