Berberine

Berberine's free cation (PubChem CID 2353) has the molecular formula C20H18NO4+ and a molecular weight of about 336.4 grams per mole. Marketed and studied berberine is often a salt (such as berberine hydrochloride) or a hydrate, which have different formulas and molecular weights than the plain cation, so a specific study's or product's exact form should be checked rather than assumed. At least one identified human pharmacokinetics paper's introduction states berberine's formula as C20H19NO5 immediately alongside a molecular weight of 336.4 grams per mole; independent verification against PubChem shows that 336.4 g/mol corresponds to the C20H18NO4+ cation, not to a neutral C20H19NO5 species, so the formula and the weight as stated together do not describe one consistent structure.^[12][4]

• Identity/nomenclature observation only; does not itself determine which salt or hydrate a specific marketed product contains, or affect that paper's reported pharmacokinetic results.
• This worker did not confirm whether the printed formula was an editorial/typesetting error or a deliberate reference to a different (for example, pseudobase) form of berberine; only the internal formula/weight inconsistency as printed is reported.

A published case report describes a 56-year-old woman who developed a markedly prolonged QT interval (667 ms) and short runs of torsades de pointes after roughly four months of high-dose hemp-oil supplement use (containing CBD and CBG, reportedly at up to six times the recommended dose) with berberine 250 milligrams added about six weeks before admission. No toxicology screen was performed. Her QT interval shortened over the following 24 hours and normalized after five days off all supplements, with no recurrence at three months. The authors attribute the event to combined berberine and hemp-oil use based on each ingredient's separate in-vitro cardiac ion-channel effects, not on measured blood levels or a controlled comparison. This single coexposure case cannot isolate berberine's individual contribution or establish a general arrhythmia risk from berberine taken alone.^[6]

• A single case with self-reported, unquantified supplement doses and no toxicology confirmation; causal attribution is mechanistic and temporal, not a demonstrated pharmacokinetic interaction.
• Does not establish a general cardiotoxicity risk, an interaction magnitude, or a risk from berberine used alone or at labeled doses without concurrent cannabinoid hemp-oil use.

A 1993 study found that berberine displaced protein-bound bilirubin in vitro about ten times more potently than phenylbutazone, and that one week of daily intraperitoneal berberine dosing raised free and total serum bilirubin in adult rats. No neonatal, pregnant, or other human subjects were studied. The paper's own abstract extrapolates from these in-vitro and adult-rat findings to recommend avoiding berberine-containing traditional medicines in jaundiced neonates and pregnant women, rather than reporting a measured human event rate.^[9]

• Abstract-only inspection of a 1993 in-vitro and adult-rat study; no neonatal or pregnancy human data were generated by this study.
• Does not itself establish a quantified human neonatal kernicterus event rate or risk from any specific berberine dose or product.

The US National Center for Complementary and Integrative Health's public guidance (last updated November 2023) states that berberine exposure has been linked to a harmful buildup of bilirubin in infants that can cause brain damage, and that berberine is likely unsafe for infants and may be unsafe in pregnancy or while breastfeeding because of possible effects on the fetus or infant. The same guidance also notes an interaction with cyclosporine. NCCIH's own key references for this caution point to the 1993 in-vitro and adult-rat bilirubin study; the page does not cite a separate quantified human neonatal outcome study.^[10][9]

• A consumer information page, not a systematic review, clinical guideline, or regulatory label.
• Its underlying citation for the infant/pregnancy caution is the same 1993 preclinical study described elsewhere in this record, not additional human epidemiological evidence.

In a 12-week randomized, placebo-controlled trial of 84 Chinese men with hyperlipidemia (80 completed), berberine 500 milligrams twice daily was associated with reduced total cholesterol and, in a model combining the 8- and 12-week measurements, a small increase in serum total testosterone (+1.31 nmol/L, 95% CI 0.30 to 2.33). Individual 8-week and 12-week between-group comparisons were not statistically significant on their own. This trial measured a blood testosterone concentration; it did not measure luteinizing hormone, follicle-stimulating hormone, sperm parameters, fertility, or hypothalamic-pituitary-gonadal axis recovery, and the observed direction was the opposite of what the trial's own registered hypothesis predicted.^[7]

• Available-case analysis after withdrawals; testosterone was a mechanistic secondary outcome, and the sample size was based on prior lipid-effect estimates, not a testosterone effect size.
• Single morning blood draw per visit using one immunoassay; does not measure gonadotropins, sperm parameters, fertility, or any recovery trajectory after stopping.
• Single-center Hong Kong Chinese sample without known cardiovascular disease or diabetes; generalizability to other populations is not established.

In a 12-week randomized, placebo-controlled trial of 100 women (complete-case analysis of 42 berberine and 48 placebo participants after asymmetric withdrawal, mostly attributed to side effects in the berberine arm), berberine 500 milligrams twice daily was associated with reduced apolipoprotein B and other atherogenic lipids, and with a decrease in serum total testosterone (-0.42 SD, 95% CI -0.82 to -0.03). The same publication reused frozen samples and testosterone data from a separate, earlier randomized trial in 84 men (80 completers) and reported an opposite direction of testosterone change in men (an increase). This is one publication reanalyzing an earlier male cohort alongside a new female trial, not two independent replications, and it measures blood testosterone concentration only, not gonadotropins, sperm parameters, fertility, or axis recovery in either sex.^[8][7]

• Main results are a complete-case linear mixed-effects model rather than the trial's multiple pre-registered primary lipid endpoints and intention-to-treat approach; attrition was asymmetric between arms (8 berberine versus 2 placebo withdrawals).
• The male comparison reuses a separate, earlier trial's frozen samples and results rather than a concurrent randomized male arm, so the sex comparison is not itself a randomized comparison of sex.
• Reused samples and a reanalyzed cohort are not independent replication of the earlier male findings.
• Measures blood testosterone concentration only; does not assess gonadotropins, sperm parameters, fertility, or axis recovery in women or men.
• An independent editorial reviewer has reopened the full text through the publisher's open-access copy and confirmed the attrition figures, the complete-case analysis, the primary-endpoint designation and the reused male cohort. The supplementary tables and figures remain uninspected by anyone.

A single-dose human pharmacokinetic study measured plasma berberine for 48 hours in 44 healthy adults after one 1,000-milligram oral dose. At the last sampled timepoint, concentrations were still 21% to 41% of each participant's peak level, and the authors state that this sampling window was too short to estimate exposure extrapolated to infinity or oral clearance. A reliable terminal half-life cannot be derived either, because it rests on the same terminal-decay data, though the authors do not use that term. Median time to peak concentration was about 3 hours. Women had roughly 2.8 times the total drug exposure and 3.6 times the peak concentration of men after the same dose. Oral bioavailability, estimated indirectly from urinary recovery of unchanged drug and early metabolites, was well under 1 percent. This single-dose plasma study does not establish a universal elimination half-life, a repeated-dosing steady state, or how the twice-daily doses and whole-blood formulations used in other berberine studies behave.^[4]

• This is one single-dose study in young, healthy, genotype-preselected adults; it does not describe steady-state kinetics with repeated twice-daily dosing, or the whole-blood formulation studies reported elsewhere for berberine.
• The sex difference in exposure was a secondary finding that the study was not specifically powered to detect.

A 13-week pilot randomly assigned 36 adults with newly diagnosed type 2 diabetes to berberine or metformin. HbA1c decreased in both groups among the 31 participants analyzed; the study did not establish that the treatments are clinically equivalent.^[2]

• Five participants withdrew for treatment failure. Allocation concealment and masking were not described; analysis excluded withdrawals. Small size, short duration and a glucose biomarker endpoint cannot establish prevention of diabetes complications or long-term safety.

A 2003 randomized, placebo-controlled trial abstract reports 156 patients with chronic heart failure and frequent ventricular ectopy who received add-on berberine (1.2 to 2.0 grams daily) or placebo alongside conventional heart-failure therapy. The abstract reports greater improvement in ejection fraction and exercise capacity, fewer ventricular premature complexes, lower mortality over a mean 24-month follow-up (7 of 79 berberine-treated patients died versus 13 of 77 on placebo, reported p<0.02), and no observed proarrhythmia. Only the abstract was accessible for inspection, so this cannot be treated as an established cardiac-outcome benefit; it does, however, show that a randomized human heart-failure and arrhythmia-monitoring investigation of berberine exists.^[5]

• Abstract-only inspection; full trial methods, allocation concealment, masking, and adverse-event ascertainment could not be verified, and no independent replication of this single trial was identified.
• Does not establish a general cardioprotective or antiarrhythmic benefit, and does not describe safety at typical supplement doses, which are far below the 1.2-2.0 g/day studied here.

A small human pharmacokinetic study found increased cyclosporine exposure in one berberine coadministration condition, but no significant change in another. This is a specific observed interaction, not proof that every CYP3A4-substrate pair behaves the same way.^[1]

• Very small study with differing regimens; mechanism was proposed rather than established and magnitude cannot be generalized.
• Abstract-checked; no qualified clinical review.

In a 13-week diabetes pilot, gastrointestinal symptoms led to reduced berberine exposure during add-on treatment; affected participants included people also taking metformin or acarbose. These coexposures were pooled and lacked a matching control group, so the study does not isolate a metformin-specific interaction.^[2]

• Observed symptoms during varied combination treatments do not establish causality or the additional risk of the named pair.
• Research exposure changes are not treatment or dose-spacing instructions.

In a randomized crossover study of 17 healthy Chinese men, two weeks of oral berberine increased midazolam exposure. Urinary probe-drug ratios also changed in a direction consistent with reduced CYP2D6 and CYP2C9 activity.^[3]

• This short probe study does not quantify clinical harm, severity or interaction magnitude for other medicines. It does not establish a berberine elimination half-life or generalize to mixed botanical products.

Physiologically based pharmacokinetic modeling combined in-vitro enzyme-inhibition data with simulated virtual patient populations to predict that high-dose berberine or goldenseal extract could raise exposure to the leukemia drug bosutinib by an amount the authors describe as potentially clinically significant, chiefly through predicted CYP3A4 inhibition. For imatinib the model predicted a smaller exposure change that the source itself describes as unlikely to be of clinical importance. The model's human pharmacokinetic inputs, enzyme-turnover estimates and multi-day inhibition-onset estimates come from separate probe-drug studies and simulation assumptions, not from patients who actually took berberine or goldenseal together with either drug. These are modeled predictions, not measured clinical interactions, and do not establish a dose adjustment.^[11]

• Model predictions only; no patient in this study actually received berberine or goldenseal together with bosutinib or imatinib.
• The model's clinical inputs come from a small number of participants in an unrelated probe-drug trial, and a model table maps that trial to dosing conditions that conflict with its own abstract description.
• Detail beyond the abstract's headline predictions rests on a single inspector's reading of the full manuscript. An independent reviewer attempted to reopen it through five separate lawful routes and could not obtain it, so no second reader has checked that detail.