Berberine

What berberine is

Berberine is an isoquinoline alkaloid from barberry (Berberis vulgaris, Berberis aristata), Chinese goldthread (Coptis chinensis), Amur cork tree (Phellodendron amurense) and North American goldenseal (Hydrastis canadensis); the yellow color comes from the compound itself. In Chinese medicine, these plants were used for centuries against diarrhea, and it was there that the blood-sugar-lowering effect was noticed.

How it is supposed to work

The common narrative runs via AMPK, an energy sensor that switches over sugar and fat metabolism. In cell culture and animal models, the chain is well described. In humans, it breaks down at one point: berberine is barely absorbed. Oral bioavailability in the rat is 0.68 percent; in humans, only plasma levels in the low nanogram-per-milliliter range are reached. It is more plausible that it acts where it arrives in high concentration: in the gut and in the liver.

For blood lipids, there is a second, better-measured chain. Kong and colleagues showed in Nature Medicine in 2004 that berberine stabilizes the mRNA of the LDL receptor, dependent on ERK and independent of SREBP. In hamsters, LDL receptor mRNA in the liver rose 3.5-fold. This is a different pathway from that of statins.

Blood sugar and blood lipids

The placebo-controlled key study comes from Zhang and colleagues, 2008 in the Journal of Clinical Endocrinology and Metabolism: 116 patients with type 2 diabetes and dyslipidemia, three months, all parameters significantly better than with placebo. In the overall picture, this shrinks: a 2026 meta-analysis of more than 30 studies arrives at minus 0.71 mmol/l fasting glucose and minus 0.19 percentage points HbA1c. The authors themselves call the improvements clinically modest.

For blood lipids, the picture is more consistent. In 2018, Ju and colleagues pooled 16 randomized trials with 2,147 participants: total cholesterol minus 0.47 mmol/l, LDL minus 0.38 mmol/l, triglycerides minus 0.28 mmol/l, HDL plus 0.08 mmol/l, all four significant.

The metformin comparison and its blind spot

The reputation as herbal metformin goes back to a single study: Yin, Xing and Ye, 2008 in Metabolism. There, 36 adults with newly diagnosed type 2 diabetes were randomized to berberine or metformin; HbA1c fell from 9.5 to 7.5 percent in the berberine group, and the authors describe the effect as similar to that of metformin. The study calls itself a pilot study; there is no prespecified non-inferiority test.

Two meta-analyses later pooled the head-to-head comparison. In 2015, Lan and colleagues analyzed 27 randomized trials with 2,569 patients and found no statistically significant difference between berberine and oral antidiabetic drugs; in 2012, Dong and colleagues reached the same result across 14 studies with 1,068 participants. That is an absence of evidence of a difference, not evidence of equivalence. There is also a pattern: Yin 2008, Zhang 2008 and the 2020 adenoma study were all conducted in China.

The best study tests something else

The strongest study on berberine is neither a diabetes nor a lipid study. In 2020, Chen and colleagues published a double-blind trial at seven hospitals in China in Lancet Gastroenterology and Hepatology. The primary endpoint was the recurrence of colorectal adenomas; it was met, at 36 versus 47 percent. This is the only large berberine study with a clinical endpoint — and it tests a use for which the compound is never marketed.

What the studies show

Zhang 2008 — placebo-controlled in type 2 diabetes

Randomized placebo-controlled trial in 116 patients with type 2 diabetes and dyslipidemia over three months. The primary endpoints were plasma glucose and serum lipids; they were met: fasting glucose 7.0 to 5.6 mmol/l, HbA1c 7.5 to 6.6 percent, LDL 3.23 to 2.55 mmol/l. Notably, the glucose disposal rate measured in the euglycemic clamp rose under berberine but did not differ significantly from placebo.

Chen 2020 — colorectal adenomas, the only clinical endpoint

Multicenter double-blind placebo-controlled trial at seven hospitals in six Chinese provinces, enrollment from November 2014 to December 2016. 553 participants on berberine, 555 on placebo, 429 versus 462 analyzed. The primary endpoint was the recurrence of adenomas at a follow-up colonoscopy; it was met, with 155 versus 216 affected, relative risk 0.77. No serious adverse events.

Yin 2008 — the metformin pilot study

Pilot study in two parts. In part A, 36 adults with newly diagnosed type 2 diabetes were randomized to berberine or metformin; HbA1c fell from 9.5 to 7.5 percent under berberine. In part B, 48 poorly controlled patients received berberine in addition; HbA1c fell from 8.1 to 7.3 percent, and the HOMA-IR index by 44.7 percent. 20 of 58 patients had transient gastrointestinal complaints. The reputation as herbal metformin rests on this study.

Where the data stop

There is not a single outcome study. Everything established on metabolism and blood lipids consists of surrogate markers; a 2026 review notes that data on cardiovascular and renal outcomes are lacking.

The nickname herbal Ozempic has no basis at all. No randomized comparison against semaglutide or another GLP-1 receptor agonist could be found. Weight loss is not independently established either: the figures come from a special population — three studies with 233 participants who had developed metabolic syndrome on antipsychotics — or from a combination study in 93 people whose authors themselves leave open the contribution of the individual components.

The changes in gut flora are established, but not consistently favorable: in type 2 diabetes, the 2026 review found, alongside an enrichment of gamma-proteobacteria, a decrease in butyrate-producing genera. And study quality is itself a limit: according to Lan and colleagues, the benefit can only be established to a limited extent because of the limited quality.

Status, approval and legal

The legal status has recently moved considerably. On behalf of the European Commission under Article 8(2) of Regulation (EC) No 1925/2006, the EFSA NDA Panel endorsed a draft on January 29, 2026: for none of the 13 plant species assessed can a safe intake level be derived. The reasons given are indications of genotoxicity in cell experiments, carcinogenic activity of Hydrastis canadensis preparations in rodent studies, the interaction risk via CYP3A4, and data gaps. The final opinion was expected in early 2027. After that, the Commission decides — possible outcomes include listing in Annex III of the regulation and, according to legal experts, also a complete ban in food supplements. According to the assessment of the Verbraucherzentrale (German consumer advice center), isolated berberine may already not be used in food supplements today, because it is considered an unauthorized novel food; a berberine-containing barberry fruit extract, by contrast, may be permissible as an ingredient. Barberry bark and root bark carry a negative medicinal monograph with an unfavorable benefit-risk ratio. There is no approved berberine medicine in Germany, which is why the German Medicines Advertising Act (§ 3a HWG) applies; disease-related advertising is additionally prohibited by Article 7 of the Food Information Regulation, on the basis of which the Heilbronn Regional Court prohibited the advertising of a berberine product on November 21, 2023. No authorized health claim exists, and neither does an official maximum level — national upper limits in Europe differ by more than an order of magnitude. No entry on the WADA Prohibited List could be found.

Safety

The most common effect concerns the gastrointestinal tract: in Yin 2008, 20 of 58 patients had transient complaints. EFSA lists constipation, diarrhea, nausea and abdominal pain; the Verbraucherzentrale additionally lists lightheadedness, nosebleeds, vomiting, kidney irritation and increased UV sensitivity. The interactions matter more. A 2026 review arranges them along four axes: inhibition and, at the same time, transcriptional induction of CYP3A4, plus quasi-irreversible inhibition of CYP2D6 and CYP2C9; modulation of transporters, namely P-glycoprotein as well as OCT1, OCT2 and MATE1; pharmacodynamic additivity in hypoglycemia, drop in blood pressure and QT prolongation; and a still unproven pathway via the microbiome. The clinically established anchor case comes from kidney transplant recipients: ciclosporin exposure rose by 34.5 percent, and the trough level was 29.3 percent above that of the control. That this happens despite the low bioavailability is not a contradiction: the interaction arises in the gut and the liver, not in the blood. On the liver, the picture is inconsistent: LiverTox lists berberine with likelihood score E as an unlikely cause of liver injury, whereas ANSES recorded one case of acute liver dysfunction for 2009 to 2018. ANSES advises against use for pregnant and breastfeeding women, children and adolescents, and people with diabetes, liver or heart disease — as well as for anyone on long-term medication with a narrow therapeutic range.

Sources

• Yin et al., Metabolism 2008 (comparison with metformin, pilot study)
• Zhang et al., J Clin Endocrinol Metab 2008 (placebo-controlled, 116 patients)
• Chen et al., Lancet Gastroenterol Hepatol 2020 (colorectal adenomas, 1,108 participants)
• Lan et al., J Ethnopharmacol 2015 (meta-analysis, 27 studies)
• Dong et al., Evid Based Complement Alternat Med 2012 (meta-analysis, 14 studies)
• Ju et al., Phytomedicine 2018 (blood lipids, 16 studies)
• Kong et al., Nature Medicine 2004 (LDL receptor mechanism)
• Dumitru et al., Pharmaceuticals 2026 (interactions, CYP3A4, P-glycoprotein)
• EFSA, draft opinion on berberine-containing plant preparations, endorsed on January 29, 2026
• ANSES, Avis 2018-SA-0095 of November 25, 2019 (safety assessment, France)