Curcumin

Curcumin, the yellow polyphenol in turmeric, has genuinely useful but modest evidence for joint pain, post-exercise recovery, and metabolic markers — and a newly recognised, occasionally serious liver-injury signal seen mostly, but not only, with the "high-absorption" products marketed for longevity.

Turmeric (Curcuma longa) has been a kitchen and traditional-medicine staple for centuries, and curcumin — the diferuloylmethane pigment that makes up roughly 3% of the dried root by weight — is the molecule behind most of its claimed effects. In cells and animals it touches almost every fashionable longevity pathway at once: it activates AMP-activated protein kinase (AMPK, the energy sensor switched on by exercise and fasting), inhibits mTOR (the growth pathway), raises sirtuin and autophagy activity, and damps the NF-κB inflammatory cascade. That mechanistic breadth is exactly why curcumin is oversold. The human trials are narrower, the effect sizes are modest, and the past few years have added a real safety caveat that didn't exist when turmeric was just a spice.

What the evidence actually supports

Joint pain and osteoarthritis — Weak to moderate. This is curcumin's best-supported use. Pooled trials in knee osteoarthritis consistently find that curcumin relieves pain better than placebo. In the few trials that compared it head-to-head with a non-steroidal anti-inflammatory drug (NSAID) such as ibuprofen or diclofenac, it did about as well.[1] One pooled analysis put numbers on it: curcuminoids cut knee-osteoarthritis pain on the 0–10 visual-analogue scale by about 1.8 points versus placebo and improved scores on the WOMAC osteoarthritis pain-and-function index. Side effects were no more common than with placebo (relative risk 1.03) and somewhat less common than with NSAIDs, though that difference could have been chance. Most were digestive complaints such as bloating, reflux, indigestion or nausea.[2] An independent appraisal of these reviews rated their methods as extremely low quality, and 37 of the 48 outcomes they reported rested on extremely low-quality evidence.[3] The trials are mostly short (8–12 weeks), small, often industry-funded, and the osteoarthritis literature shows funnel-plot publication bias, so "comparable to an NSAID for mild-to-moderate knee osteoarthritis" is the honest ceiling. Curcumin is also not uniquely best among joint supplements — a 2025 network meta-analysis of 39 trials ranked Boswellia above it for pain and stiffness.[4]

Exercise recovery — Weak to moderate. Curcumin does not make you fitter — it does not raise VO₂ max (maximal oxygen uptake), aerobic capacity, or strength during a session. What it does, in studies that deliberately cause muscle damage, is modestly blunt the soreness and muscle-damage markers after hard or unaccustomed exercise. A meta-analysis of 14 randomised trials (349 people) found that supplementation lowered muscle soreness (by about 0.6 points on pain scales), creatine kinase (a muscle-damage marker, by about 137 U/L) and interleukin-6 (IL-6), and improved range of motion at 24–72 hours. The biggest soreness reduction, about 1.2 points, came at 96 hours, though that figure rests on only 5 small studies and the trials disagreed a lot.[5] In athletes during their normal training, results are mixed and the certainty of the evidence is rated very low.[6] Possibly useful for training through a heavy block; not an ergogenic aid.

Metabolic markers — Weak to moderate, and mainly in people who are already impaired. Pooled randomised data show real but small improvements in blood sugar in adults with prediabetes or type 2 diabetes. A dose-response meta-analysis of 34 trials in that population found fasting glucose about 0.6 mmol/L (about 10 mg/dL) lower (95% CI −12.6 to −7.7 mg/dL; a confidence interval is the range where the true effect most plausibly lies). Glycated haemoglobin (HbA1c, a marker of average blood sugar) was about 0.3 percentage points lower, and insulin sensitivity improved (HOMA-IR −0.46). Individual trials disagreed a lot, so the authors rate the evidence as low certainty, and they describe curcumin as an add-on to diabetes treatment, not a replacement.[7] Systolic blood pressure falls about 2.7 mmHg in the same population (95% CI −3.8 to −1.6, more in people already hypertensive), though diastolic pressure does not change significantly overall,[8] and a small trial funded by the US National Institutes of Health (NIH) gave 39 healthy adults aged 45–74 either 2,000 mg/day of Longvida, an enhanced-absorption formulation containing about 400 mg of actual curcumin, or placebo for 12 weeks. The authors declared no conflicts, though the manufacturer supplied the product. The lining of the participants' blood vessels worked better, through greater nitric-oxide availability, but arterial stiffness and blood markers of inflammation and oxidative stress did not change.[9] Blood lipids improve slightly. An early meta-analysis of five small trials found no significant change,[10] but a 2025 umbrella review re-pooled 72 trials, mostly in people with chronic conditions such as diabetes. It found LDL cholesterol about 0.15 mmol/L (6 mg/dL) lower and triglycerides about 0.15 mmol/L (13 mg/dL) lower, with larger effects in type 2 diabetes, though the trials varied widely.[11] It is not a substitute for lipid-lowering drugs.

Fatty liver — Weak to moderate. In small, short trials in people with fatty liver disease, curcumin modestly lowered the liver enzymes alanine aminotransferase (ALT) and aspartate aminotransferase (AST), by about 4 and 3 U/L in one meta-analysis of 15 trials.[12] Another meta-analysis found no significant change in ALT,[13] while a third found less liver fat.[14] In trials of curcumin combined with piperine, the effect on ALT was not statistically significant. Given the separate liver-injury signal below, anyone with liver disease should use it only with their doctor.

Cognition — Weak to preliminary. The headline trial is small: an 18-month, placebo-controlled study at UCLA gave 40 non-demented adults aged 51–84 a nanoparticle curcumin (90 mg twice daily, 180 mg/day) and reported better verbal and visual memory (a sizeable between-group effect, an effect size of about 0.7 on long-term retrieval) and mood, alongside changes on brain scans that were weaker than often reported. Positron-emission tomography (PET) in 30 of the participants, using the non-standard FDDNP tracer, showed no clear reduction in amyloid/tau-related signal versus placebo, and the one significant difference (in the hypothalamus) came from the signal rising in the placebo group. The memory benefit itself only just reached statistical significance.[15] About ten small trials (roughly 500 people) have now been pooled, and the two meta-analyses disagree. One found no effect on overall cognition, with small gains only in working memory and processing speed and more side effects on curcumin.[16] The other found a benefit only in trials lasting at least 24 weeks, in people aged 60 or over, or in Asian participants.[17] Hypothesis-generating, not established.

Inflammatory biomarkers — Weak to moderate, and inconsistent. Pooled analyses find curcumin lowers C-reactive protein (CRP, a general blood marker of inflammation) a little on average. Two umbrella reviews found significant average reductions in CRP, interleukin-6 and tumour necrosis factor-α (TNF-α), one pooling 21 meta-analyses[18] and one pooling 10,[19] and six small trials in type 2 diabetes found high-sensitivity CRP about 1 mg/L lower, a result that became less certain in sensitivity checks.[20] But the trials are small and short and often disagree, and a meta-analysis in chronic inflammatory disease found no significant effect on C-reactive protein, high-sensitivity CRP, interleukin-1β, interleukin-6, or TNF-α.[21] The honest read is a small average effect, not a reliable one in everyone.

Depression and anxiety — Weak / preliminary. A 2026 meta-analysis of 19 trials in adults with depression or depressive symptoms found a moderate improvement in depression scores (a standardised effect of about −0.76 across 8 pooled trials, where roughly 0.8 counts as large). But the trials disagreed enormously (I² about 95%, a measure of how much individual trials disagree), and the authors call the result fragile in sensitivity checks.[22] A large 2025 network meta-analysis found the clearest signal for curcumin as an add-on to antidepressants; on its own, compared with antidepressants, its effect was uncertain.[23] A separate anxiety meta-analysis reported a larger but far more inconsistent effect (−1.56; I² = 95.6%).[24] Trials are small, short (6–12 weeks), frequently adjunctive to antidepressants, and several are open-label; bioavailability-enhanced forms show larger effects than conventional curcumin. Hypothesis-generating, not established.

Bone, cancer, "epigenetic age reversal" — Weak / preliminary. Bone-density benefits are essentially all rodent data. For cancer treatment, the US National Cancer Institute's verdict is that the evidence is inadequate to recommend curcumin-containing products; a systematic review of the few qualifying RCTs found no improvement in overall or progression-free survival, though tolerability was good.[25] And the widely shared claim that curcumin "reverses the epigenetic clock by 3 years" comes from an 8-week pilot trial in about 40 healthy men aged 50–72 that tested a whole programme: a diet, sleep and exercise advice, a greens powder and a probiotic supplement. Turmeric was only one of six optional foods participants could pick from (half a teaspoon), no curcumin supplement was given, and nobody recorded how much turmeric anyone ate. The "3 years" was the difference from a control group on an epigenetic clock whose readings can differ by several years between repeat measurements of the same sample, so the result says nothing about curcumin.[26][27] A 2025 umbrella review of 25 curcumin meta-analyses reached the sober bottom line. About four in five of the 122 outcomes it graded rested on low- or very-low-certainty evidence, and only 4 on high-certainty evidence. Three-quarters of the meta-analyses were themselves of very low methodological quality, though the authors still judged curcumin safe and beneficial for several outcomes.[28]

The bioavailability problem — and why the label matters

Plain curcumin is almost useless taken by mouth. It barely dissolves in water and degrades in the gut, and what little is absorbed is quickly converted by the liver into bound forms (mainly glucuronides). In laboratory tests, one of these has about a tenth of curcumin's antioxidant activity and another has almost none.[29] The typical Indian diet, about 2–2.5 g of turmeric a day, delivers only about 60–100 mg of curcumin, with negligible systemic absorption.[30] This is why every clinical formulation is engineered to get around first-pass metabolism:

ApproachExampleWhat it does
Piperine (black-pepper alkaloid)Curcumin + piperineMeant to block the enzymes that break curcumin down. A small 1998 study co-authored by a piperine maker reported a ~20-fold rise in blood levels in the first hour,[31] but a 2025 independent study in 9 healthy men found piperine added nothing.[32] It may also strengthen drug interactions (see below)
Phytosome (lipid complex)MerivaBound to phosphatidylcholine for better solubility; the form behind much of the joint and recovery data
Nanoparticle / colloidalTheracurminSub-micron particles raise free-curcumin levels; the form used in the cognition trial at low absolute doses
Essential-oil complexBCM-95Reconstituted with turmerones; ~7-fold higher absorption than raw powder

The practical consequence: dose recommendations are meaningless without naming the formulation, and raw turmeric powder or culinary curry does not reach the blood levels used in trials.

There is a deeper catch even the best formulations cannot escape. An independent crossover pharmacokinetic study of three commercial products found that unconjugated (active) curcumin stayed below 2 nM in most cases; the highest-absorbing formulation peaked at only 6.7–38 nM, still roughly 100-fold below the concentrations used in the cell studies that demonstrate curcumin's celebrated mechanisms.[33] "Enhanced absorption" mostly raises blood levels of the conjugated, less-active metabolites. Vendor multipliers such as "65×" or "185× greater bioavailability" should be read skeptically: they typically come from pharmacokinetic studies authored or financed by the formulation makers themselves.

The catch: liver injury from turmeric and curcumin supplements — Caution

Here is the paradox. Ordinary turmeric, and even unformulated curcumin at very high doses (8–12 g/day in dose-escalation studies), is well tolerated and may even lower liver enzymes.[34] But the same engineering that finally made curcumin absorbable also seems to have made liver harm more likely. The US National Institutes of Health LiverTox database now classifies turmeric/curcumin as a well-documented cause of clinically apparent liver injury — and what appears to have become the most common cause of herbal-supplement liver injury in the US — and most cases involve high-absorption curcumin products. But piperine (black-pepper extract) was found in only 3 of the 7 products tested in the US case series, and LiverTox also describes cases with plain ground turmeric powder and even turmeric teas.[35] The signal is serious but rare in absolute terms: LiverTox says the true rate is unknown but probably very rare, roughly 1 in 10,000 to 1 in 100,000 people exposed. Regulators have since acted. Health Canada, Italy, France and Australia have added liver-injury warnings or safety alerts for turmeric and curcumin supplements, and the US Pharmacopeia recommends a label warning that liver problems have been reported "very rarely". Health Canada notes that turmeric eaten as a spice in normal amounts is not a concern.[36][37] The defining US case series identified 10 patients over nearly two decades (8 women, hepatocellular injury in 9, latency 1–4 months, one death).[38] A one-author letter from a consultancy firm argued that the risk across the whole population is very low.[39] A population-wide figure tempers the caution, but it does not dissolve it for the people actually taking these products.

Key features of the signal:

  • It is idiosyncratic, not dose-toxic. The injury is an immune-mediated reaction, not simple overdose, which is why it is unpredictable.
  • There is a strong genetic risk marker. The HLA-B*35:01 immune-gene variant was carried by 7 of the 10 patients in the US case series,[40] and LiverTox reports it in over 70% of cases versus 10–15% of people in comparison groups. That makes it the strongest known predictor of who is susceptible.[41]
  • Onset is delayed. Symptoms — fatigue, nausea, poor appetite, then dark urine and jaundice — typically appear after 1–4 months of continuous use, with liver-enzyme (ALT) spikes that can exceed 1,000 U/L and mimic autoimmune hepatitis.
  • It usually resolves on stopping, within 1–3 months. But acute liver failure, liver transplants and deaths have occurred, mostly when people kept taking the product after symptoms began, and LiverTox estimates that roughly 1 in 10 people who develop jaundice might die. Anyone who has had this reaction should not take turmeric or curcumin again, because it can come back worse.[42]

For dosing context: the acceptable daily intake for curcuminoids set by the WHO's Joint FAO/WHO Expert Committee on Food Additives (JECFA) is 0–3 mg/kg of body weight (about 210 mg for a 70-kg adult), established in 2004 and reaffirmed by the European Food Safety Authority (EFSA) in 2010. French and German food-safety agencies have warned that this limit was set for ordinary curcumin and may not protect people taking enhanced-absorption products, which may need lower limits set product by product.[43][44] Doses commonly suggested for supplements, 500–1,000 mg of curcuminoids a day according to the UK's Committee on Toxicity, already run two to five times above that intake limit.[45]

This is not a reason for most people to fear turmeric in food. It is a reason to treat any turmeric or curcumin supplement as a pharmacologically active product rather than a harmless botanical. The reaction is rare and unpredictable, and it does not appear to depend on dose. Health Canada describes it as "not dependent on the dose or duration of use", and a 2024 Spanish case involved only 120 mg of curcumin a day (with black pepper) for two months.[46][47] Most reported cases still involve enhanced-absorption products.

Drug interactions — Caution

Curcumin can affect the body's drug-handling systems, but not in the simple way laboratory work suggests. In rats it suppressed the CYP3A drug-metabolising enzyme in the gut while inducing it in the liver.[48] Yet in 8 healthy volunteers, a high-dose curcumin-plus-piperine product (4 g of curcuminoids with 24 mg of piperine, given four times over two days) did not change blood levels of standard test drugs for CYP3A4 (the enzyme that handles roughly half of all prescription drugs) or for CYP2C9. That study was short and was co-authored by staff of the product's maker.[49] In 12 healthy volunteers, 300 mg of curcumin a day for six days actually lowered blood levels of talinolol, a test drug for the P-glycoprotein efflux pump, by about a third.[50] The clearest effect in people is on a different gut transporter, BCRP: 2 g of curcumin roughly tripled blood levels of sulfasalazine, a bowel and arthritis drug.[51] So the effect depends on the drug and is hard to predict. Combined with curcumin's own mild blood-thinning and glucose-lowering effects, this creates several interactions worth taking seriously:

  • Anticoagulants and antiplatelets (warfarin, clopidogrel, aspirin, NSAIDs) — additive bleeding risk.
  • Glucose-lowering drugs (metformin, insulin, SGLT2 inhibitors) — additive risk of low blood sugar.
  • Statins — no human study has shown that curcumin changes statin levels. But because it blocks a drug-export pump (BCRP) in people, an interaction is possible, so tell your prescriber if you develop muscle aches. A 2017 hypothesis paper even proposed curcumin as an add-on for statin muscle symptoms; that idea has not been tested.[52]
  • Chemotherapy and immunosuppressants — changes in drug exposure that are hard to predict; not to be combined without specialist oversight. In women taking tamoxifen for breast cancer, high-dose curcumin (1.2 g three times a day) lowered levels of tamoxifen's active form by about 8% on its own (not statistically significant) and by about 12% when combined with piperine, which the researchers warned could push some patients below the effective level.[53] For the transplant drug tacrolimus there is one reported case of a dangerously high level, with kidney injury, after heavy turmeric use in food,[54] while in another transplant patient a deliberate four-day test of 10 g of turmeric a day showed no change.[55]

If you take any of these, do not start any turmeric or curcumin supplement without checking with the prescriber.

Practical guidance

  • Pick a use case before a product. The defensible reasons to take it are mild-to-moderate osteoarthritis pain and soreness after hard or unaccustomed training. For general "anti-aging," the human evidence does not justify it.
  • Match dose to formulation (these are the doses studied, not raw-powder equivalents): phytosome (e.g. Meriva) ~1,000 mg/day split in two for joints and recovery; nanoparticle (e.g. Theracurmin) ~90–180 mg/day; piperine-boosted standard curcumin 1,000–1,500 mg/day with 10–20 mg piperine, though an independent study found piperine added no absorption benefit (see the table above), and this form may carry the highest interaction risk. For joint pain, some pooled analyses found that only the enhanced-absorption products clearly beat placebo while plain curcumin extracts did not, but another found all forms helped.[56][57][58] Formulations have not been properly compared head-to-head, so these brand doses are not proven equivalents.
  • Take it with a fatty meal. Curcumin is fat-soluble, and absorption is much higher with dietary fat.
  • Buy on verified quality, not label claims. Independent testing repeatedly finds products delivering far less curcuminoid than labelled, and a six-continent scoping review put the overall turmeric adulteration rate around 20% — in supplements often synthetic curcumin sold as turmeric extract, which can pass total-curcuminoid assays without containing genuine turmeric.[59] Choose products with independent verification (USP, NSF, or third-party testing). Lead chromate, added to brighten the colour of turmeric roots and powder, is still found in parts of South Asia. A 2021–2023 survey of 34 cities in eastern India found 30% of samples above India's legal lead limit,[60] and a 2020–2021 survey found extreme levels in markets in India and Pakistan.[61] Bangladesh largely eliminated it after a campaign that combined publicity with enforcement.[62] It is a problem of loose and bulk spice sold in those markets, not of standardised extract capsules.
  • Don't count on a short course to protect your liver. Avoid open-ended daily use, but don't treat a course of 8–12 weeks as safe either: most reported reactions started 1–4 months in, so an 8–12-week course sits squarely inside that window.[63] The real safeguard is to stop at once and see a doctor if you notice unexplained fatigue, nausea, stomach pain, dark urine, or yellowing of the eyes or skin.[64]
  • If you have liver enzymes checked (ALT, AST, ALP, GGT), the useful times are before starting and within the first few months on a new product, because that is when most reactions appear. Liver injury can also be silent: in one small series of biopsied cases, about half the patients had no symptoms and were found only through abnormal blood tests.[65] See Midlife labs.
  • Skip it if you have gallbladder or bile-duct problems (including gallstones), are pregnant (turmeric supplements may be unsafe in pregnancy) or breastfeeding (safety above food amounts is unknown), take anticoagulants or are about to have surgery, or take diabetes medication without supervision.[66][67] Check with a doctor first if you have current or past liver problems or take other medications.

What's overrated

The framing of curcumin as a caloric-restriction-mimicking "geroprotector" that slows biological aging is mechanism dressed up as outcome. The AMPK, mTOR, sirtuin and senolytic effects are almost all from cells, yeast, flies, and rodents; the human trials that exist are short, surrogate-endpoint, and weighted toward impaired populations. There is a deeper problem with that preclinical foundation. A landmark medicinal-chemistry review classified curcumin as both a PAINS (pan-assay interference compound) and an "invalid metabolic panacea" — an unstable, reactive, non-bioavailable molecule whose much-cited in vitro "activity" is largely assay artifact (fluorescence interference, aggregation, metal chelation), making it "a highly improbable lead."[68] Curcumin research has also been hit by retractions. One of the field's most prolific researchers had 18 papers retracted by 2016, seven of them after his former institution found reused images. So individual preclinical claims deserve extra caution.[69] There is no human trial showing curcumin extends life or slows aging — and the same is true of the broader "longevity stack" it is often sold alongside (NMN, resveratrol, and the rest). The realistic verdict: a reasonable anti-inflammatory option for joints and recovery, in a well-formulated product, used with awareness of the liver and drug-interaction risks — not a longevity drug.

Further reading

  • Bahari H et al. Curcumin/Turmeric Supplementation on Glycemic Control in Adults With Prediabetes and Type 2 Diabetes: A Systematic Review and Dose–Response Meta-Analysis. 2026.[70]
  • Zeng L et al. Efficacy and Safety of Curcumin and Curcuma longa Extract in the Treatment of Arthritis: A Systematic Review and Meta-Analysis of Randomized Controlled Trials. 2022.[71]
  • Turmeric. LiverTox: Clinical and Research Information on Drug-Induced Liver Injury. NIH/NCBI Bookshelf.[72]
  • Turmeric: Usefulness and Safety. National Center for Complementary and Integrative Health.[73]
  • Gupta SC et al. Therapeutic Roles of Curcumin: Lessons Learned from Clinical Trials. AAPS J 2013.[74]
  • Xu Q et al. Curcumin and multiple health outcomes: critical umbrella review of intervention meta-analyses. 2025.[75]
  • Darmonkow A et al. Advancements in curcuminoid formulations: an update on bioavailability enhancement strategies. Open Life Sci 2025.[76]
  • Nelson KM et al. The Essential Medicinal Chemistry of Curcumin. J Med Chem 2017.[77]
  • Kroon MAGM et al. A pharmacokinetic study and critical reappraisal of curcumin formulations enhancing bioavailability. iScience 2025.[78]
  • Halegoua-DeMarzio D et al. Liver Injury Associated with Turmeric — a Growing Problem: Ten Cases from the Drug-Induced Liver Injury Network. Am J Med 2023.[79]
  • Naghsh N et al. Profiling Inflammatory Biomarkers following Curcumin Supplementation: An Umbrella Meta-Analysis. Evid Based Complement Alternat Med 2023.[80]
  • Gafner S et al. A scoping review of turmeric adulteration based on data from six continents. Pharm Biol 2026.[81]

— § —