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 that tracks the very "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 — Moderate. This is curcumin's best-supported use. A meta-analysis of randomised arthritis trials found that curcumin extract (around 1,000 mg/day) reduced pain and stiffness about as much as non-steroidal anti-inflammatory drugs (NSAIDs) such as ibuprofen and diclofenac, with fewer gastrointestinal side effects.^[1] A separate pooled analysis put the 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, while adverse events were no higher than placebo (relative risk 1.03) and lower than NSAIDs.^[2] 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.^[3]

Exercise recovery — 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, fairly reliably, is blunt the soreness and muscle-damage markers after hard or unaccustomed exercise. A meta-analysis of 14 randomised trials found that supplementation lowered muscle soreness (mean difference about −0.6 on pain scales), creatine kinase (about −137 U/L), and interleukin-6 (IL-6) while improving range of motion at 24–72 hours post-exercise, with the soreness benefit largest — a large effect, about −1.2 in standardised terms — at 96 hours.^[4] Useful for training through a heavy block; not an ergogenic aid.

Metabolic markers — Moderate, mostly in people who are already impaired. Pooled randomised data show real but small improvements concentrated in adults with prediabetes, type 2 diabetes, or metabolic syndrome — not in already-healthy people. A dose-response meta-analysis of 34 trials reported reductions in fasting glucose of roughly 0.6 mmol/L (about 10 mg/dL; 95% CI −12.6 to −7.7 mg/dL — a confidence interval is the range where the true effect most plausibly lies, and because this one stays entirely below zero the reduction is statistically reliable), glycated haemoglobin (HbA1c, a marker of average blood sugar) of about 0.3 percentage points, and improved insulin sensitivity (HOMA-IR −0.46).^[5] Systolic blood pressure falls about 2.7 mmHg (95% CI −3.8 to −1.6, more in people already hypertensive), though diastolic pressure does not change significantly overall,^[6] and a small mechanistic trial funded by the US National Institutes of Health (NIH) — a rarity in this largely industry-funded field — found that 2,000 mg/day for 12 weeks improved blood-vessel function in healthy middle-aged and older adults by raising nitric-oxide availability.^[7] Effects on blood lipids are inconsistent: an earlier meta-analysis of randomised trials found no significant change in total cholesterol, LDL, HDL, or triglycerides,^[8] while a later umbrella review reported small reductions — so any lipid benefit is small and concentrated in metabolically impaired populations.

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 reduced amyloid/tau-related signal on positron-emission tomography (PET) using the non-standard FDDNP tracer.^[9] It is one positive trial in roughly 40 participants, has not been replicated, and the reliance on a non-standard imaging tracer limits the amyloid/tau claim — hypothesis-generating, not established.

Inflammatory biomarkers — Moderate, but heterogeneous. An umbrella meta-analysis found curcumin lowered inflammatory markers by moderate-to-large amounts on average — as standardised effect sizes, where roughly 0.8 already counts as large: C-reactive protein −0.74, interleukin-6 −1.07, and tumour necrosis factor-α (TNF-α) −1.92 — with larger effects in older and metabolically impaired groups and at lower doses for shorter courses.^[10] But the signal is inconsistent and undermined by extreme statistical heterogeneity (I² often above 88% — the individual trials disagreed enormously): 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.^[11] The honest read is a real average effect, concentrated in metabolic disease, not a reliable one in everyone.

Depression and anxiety — Weak / preliminary. A depression meta-analysis of 6 trials found a small-to-moderate benefit (a standardised effect size of −0.34 on depression-rating scales),^[12] and a separate anxiety meta-analysis reported a larger but far more heterogeneous effect (−1.56; I² = 95.6%).^[13] 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.^[14] And the widely-shared claim that curcumin "reverses the epigenetic clock by 3 years" comes from a small pilot trial of a whole diet — curcumin plus green tea, rosemary, garlic, berries, sleep and exercise advice — not curcumin in isolation, so it says nothing about the supplement on its own.^[15] A 2024 umbrella review of pooled curcumin meta-analyses reached the sober bottom line: across dozens of outcomes, most of the statistically significant effects rest on low- or very-low-quality evidence.^[16]

The bioavailability problem — and why the label matters

Plain curcumin is almost useless taken by mouth. It barely dissolves in water, degrades in the gut, and what little is absorbed is rapidly conjugated by the liver into metabolites with roughly ten-fold lower antioxidant activity that cannot cross into the brain.^[17] The typical Indian diet — about 2.5 g of turmeric a day — delivers only 60–100 mg of curcumin, with negligible systemic absorption. This is why every clinical formulation is engineered to get around first-pass metabolism:

Approach	Example	What it does
Piperine (black-pepper alkaloid)	Curcumin + piperine	Blocks the liver's conjugating enzymes, raising blood levels up to ~20-fold — but mostly the conjugated (less active) metabolites, and it is also the formulation most prone to drug interactions
Phytosome (lipid complex)	Meriva	Bound to phosphatidylcholine for better solubility; the form behind much of the joint and recovery data
Nanoparticle / colloidal	Theracurmin	Sub-micron particles raise free-curcumin levels; the form used in the cognition trial at low absolute doses
Essential-oil complex	BCM-95	Reconstituted 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 — and adding piperine gave no benefit.^[18] "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 high-absorption products — 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.^[19] But the same engineering that finally made curcumin absorbable also made it capable of harming the liver. The US National Institutes of Health LiverTox database now classifies turmeric/curcumin as a well-documented cause of clinically apparent liver injury — in fact the most common cause of herbal-supplement liver injury in the US — and the cases cluster around the high-bioavailability products, especially those using piperine or lipid-nanoparticle delivery.^[20] The signal is serious but rare in absolute terms: LiverTox estimates clinically apparent injury at roughly 1 in 10,000 to 1 in 100,000 people exposed. The defining US case series identified 10 patients over nearly two decades (8 women, hepatocellular injury, latency 1–4 months, one death).^[21] A published rebuttal argues the per-capita risk is therefore very low and confounded by obesity and alcohol; that counterpoint is a single-author letter with an industry-adjacent disclosure, and its population-wide denominator understates risk per user, so it tempers but does not dissolve the caution.

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, carried by roughly 6–7% of the general population, was present in 7 of 10 patients in the US case series — a striking enrichment that makes it the strongest known predictor of who is susceptible.^[22]
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 and rare fatalities have occurred when people kept taking the product after symptoms began.^[23]

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. Crucially, regulators note this limit was derived for ordinary curcumin and excludes bioavailability-enhanced preparations, and that supplement doses can run several times higher — another way of saying the enhanced products are pharmacologically active.

This is not a reason for most people to fear a turmeric latte. It is a reason to treat high-dose, "enhanced-absorption" curcumin capsules as a pharmacologically active drug rather than a benign botanical.

Drug interactions — Caution

Bioavailable curcumin modulates two of the body's main drug-handling systems — the cytochrome P450 3A4 enzyme (which metabolises roughly half of all prescription drugs) and the P-glycoprotein efflux pump — and it does so unpredictably, suppressing them in the gut while inducing them in the liver.^[24] 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 — altered statin levels can raise the risk of statin-associated muscle symptoms, though curcumin has also been studied as an adjunct for them.^[25]
Chemotherapy and immunosuppressants — unpredictable changes in drug exposure; not to be combined without oncology oversight.

If you take any of these, do not start high-dose curcumin 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 recovery from heavy 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 — effective but the highest interaction risk.
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.^[26] Choose products with independent verification (USP, NSF, or third-party testing). Lead-chromate adulteration of whole turmeric spice was a serious problem in South Asian supply chains but has been largely corrected in surveyed markets since 2021; it is mainly a bulk-spice issue, not a standardised-extract-capsule one.
Don't take it continuously for years without thought. Given the delayed liver-injury signal, a sensible pattern is time-limited courses (e.g. 8–12 weeks for a flare or training block) rather than indefinite daily use, and stopping at the first sign of unexplained fatigue, nausea, or dark urine.
Monitor if you use it long-term, particularly the high-absorption forms — include liver enzymes (ALT, AST, ALP, GGT) in routine bloodwork. See Midlife labs.
Skip it if you are on anticoagulants or are about to have surgery, on diabetes medication without supervision, pregnant (high doses), or have known liver disease.

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."^[27] The curcumin literature is also unusually polluted by retracted papers, so individual preclinical claims should be treated with extra caution. 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.