Turmeric and inflammation: what curcumin can do and where the hype overshoots

A group around Kathryn Nelson published a very hard stocktaking in the Journal of Medicinal Chemistry in 2017. They assign curcumin to two notorious categories: PAINS, meaning substances that appear seemingly active almost everywhere in laboratory assays, and IMPS, meaning substances with an unrealistically broad promise of effect.

Their reasoning: curcumin is chemically unstable, highly reactive and practically not bioavailable. It can generate signals in test systems that look like an effect without being one. Their conclusion as of 2017 was that no double-blind, placebo-controlled trial with curcumin had been successful.

For you this does not mean that everything is nonsense. It means: when you read an enthusiastic headline about curcumin, it is worth checking whether cell culture or people stand behind it.

A group around G. Shoba at St. John's Medical College in Bangalore tested in 1998 what happens when curcumin is combined with piperine. Piperine is the pungent compound in black pepper and a known brake on exactly those enzymes that break curcumin down.

After 2 grams of curcumin alone, blood levels in the volunteers were undetectable or very low. Together with 20 milligrams of piperine they rose markedly. The authors reported an increase in bioavailability of around 2000 percent.

For you this means: the pepper in the capsule is not marketing. It is the reason anything arrives at all. And it is a safety topic at the same time, more on that further down.

A research group around Christina Schiborr at the University of Hohenheim gave twenty-three healthy people 500 milligrams of curcuminoids each in random order in 2014. Once as native powder, once micronised, once as liquid micelles.

Measured by the area under the concentration curve, micronised curcumin was roughly 9-fold and micellar curcumin roughly 185-fold more bioavailable than the native powder. Women absorbed considerably more than men. Safety values stayed within the normal range in all groups.

For you this means: the formulation decides orders of magnitude, not nuances. A cheap powder without a carrier system is pharmacologically something else than a micellar preparation.

The same Hohenheim group took the next step in 2016. Forty-two people with mildly elevated cholesterol and mildly elevated C-reactive protein received 294 milligrams of highly bioavailable micellar curcuminoids or placebo daily over six weeks, in a crossover design.

The curcuminoids accumulated measurably in the blood. But neither blood lipids nor inflammatory markers nor blood sugar nor iron metabolism changed compared with placebo. Liver values stayed unremarkable.

For you this means: bioavailability is a precondition, not a guarantee of effect. Someone who is already largely healthy may simply have nothing for an inflammation brake to act on.

John Bright summarised in a review how curcumin intervenes in models of autoimmune diseases. Described are effects on IL-1beta, IL-6, IL-12, TNF-alpha and interferon-gamma as well as on the signalling pathways JAK-STAT, AP-1 and NF-kB in immune cells.

The closing sentence of the same paper is notable. The authors explicitly urge caution about the jump from traditional use as a spice to high-dose pure substances as a therapeutic agent.

For you this means: the breadth of the described effects is no proof of strength. Sometimes it is a hint that the substance rattles a little at many things.

A group around Liuting Zeng at Peking Union Medical College Hospital pooled 15 randomised trials with 1,621 participants. Curcuma longa extract or curcumin was compared against placebo and against classic anti-inflammatory drugs.

Compared with placebo, pain scores on the visual analogue scale and the WOMAC values for pain, function and stiffness were lower. Compared with classic anti-inflammatory drugs the effects were similar, with a lower rate of adverse events. The authors recommend a duration of use of more than twelve weeks before judging.

For you this means: if you notice nothing after four weeks, that is no proof of nothing. It may simply be too early.

Because the individual analyses contradicted each other, a group around Mohammad Vesal Bideshki pooled eleven meta-analyses in 2024. That is called an umbrella analysis and it sits at the very top of the evidence pyramid.

The pooled result showed lower values on the pain scale as well as in the WOMAC domains total, function, pain and stiffness. The authors see this as support for the use of curcuminoids in knee osteoarthritis.

For you this means: at the level of symptoms, the effect in knee osteoarthritis is one of the better secured findings in the entire curcumin literature.

A group around Igho Onakpoya at the Centre for Evidence-Based Medicine in Oxford analysed seven randomised trials with 797 people, predominantly with knee osteoarthritis.

Compared with placebo, lower pain scores and better quality of life showed up. Compared with ibuprofen, however, the curcuminoids performed significantly more weakly on pain, stiffness and function. All included trials came from Asia, the samples were small, and the reporting quality varied.

For you this means: curcumin is not a stronger painkiller. It is a different offer, with a different side effect profile.

A group around Xiaoqian Liu at the Kolling Institute in Sydney examined 20 dietary supplements in osteoarthritis across 69 trials. Curcumin was among the seven substances with a large effect on pain reduction, though explicitly only in the short term.

The decisive sentence comes later: at medium and long follow-up, no clinically meaningful effect on pain or function was found for a single supplement. The authors rate the overall quality of evidence as very low.

For you this means: short-term relief is plausible. A supplement that influences the course of osteoarthritis over years is not established by this.

A group around Mohammad Jafar Dehzad in Shiraz pooled 66 randomised trials and additionally rated the certainty of the evidence with the GRADE system.

Under turmeric or curcumin, CRP, TNF-alpha and IL-6 were lower than in the control groups. Antioxidant capacity also rose, while malondialdehyde as a marker of oxidative cell stress fell.

For you this means: averaged across many trials, something moves. The magnitude is real, but moderate. A CRP drop of 0.58 milligrams per litre is relevant at a baseline of 5 and meaningless at a baseline of 0.8.

A group around C. Michael White at the University of Connecticut analysed 19 randomised trials, exclusively in people who already had a chronic inflammatory disease: rheumatic conditions, advanced kidney disease with dialysis, metabolic syndrome, cardiovascular disease.

Neither CRP nor hsCRP nor IL-1beta nor IL-6 nor TNF-alpha fell in a statistically meaningful way. There was no difference between turmeric powder and curcumin extract. Variation between the trials was high.

For you this means: precisely where inflammation is most firmly established, the effect was least detectable. That is the exact reverse of what advertising promises.

A group around Federica Zatterale at the University of Naples Federico II summarised how growing fat tissue turns into persistent inflammation. When fat tissue grows faster than its blood supply, oxygen shortage, cell death and mechanical stress arise. For the immune system these are alarm signals.

Macrophages migrate in. With pronounced excess weight they can make up as much as 40 percent of all cells in fat tissue. In doing so they switch from the clearing-up state into the pro-inflammatory one and release TNF-alpha, IL-6 and IL-1beta. These very messengers disturb insulin signalling.

For you this means: waist circumference is not a cosmetic topic. It is an inflammation parameter.

A group around Somlak Chuengsamarn in Thailand gave 240 people with prediabetes curcumin extract or placebo over nine months, double-blind.

After nine months, 16.4 percent of the placebo group had developed type 2 diabetes. In the curcumin group it was nobody. In addition, insulin resistance by HOMA-IR was lower, beta cell function better and the anti-inflammatory adiponectin higher.

For you this means: that is a remarkable result from a single trial in a single population. A replication at this scale is still outstanding to this day. I take it seriously, and I do not treat it as a proven fact.

A group around Linjie Qiu at the China Academy of Chinese Medical Sciences pooled 13 randomised trials in metabolic syndrome, with durations from 4 to 12 weeks.

Under curcumin, waist circumference, fasting blood glucose, diastolic blood pressure, TNF-alpha, CRP and malondialdehyde were lower, and HDL cholesterol higher. No difference was found for systolic blood pressure, triglycerides, IL-6 and hsCRP. Variation was considerable.

For you this means: where metabolism and inflammation overlap is where the most consistent curcumin data sit. Not with "inflammation" as an abstract state.

A group around Yingxi Li at Hebei Agricultural University tested a nanoparticle preparation in mice with chemically induced colon inflammation, designed to release curcumin only once it reaches the colon.

In the treated group, pro-inflammatory messengers were lower and the anti-inflammatory IL-10 higher. The junction proteins ZO-1, claudin-1 and occludin were more strongly expressed, and the TLR4/NF-kB and JAK2/STAT3 pathways less activated. In addition, the composition of the gut flora shifted, and the production of short-chain fatty acids increased.

Important for context: this is an animal model with a specially engineered release form. In humans it is not shown this way. It is biologically plausible, no more than that for now.

A group around Dina Halegoua-DeMarzio analysed all cases from the US Drug-Induced Liver Injury Network between 2004 and 2022 in which turmeric was involved. It found ten cases, all reported from 2011 onwards, six of them since 2017.

Nine cases ran as hepatocellular injury. Five people required inpatient care, one person died of acute liver failure. Latency was one to four months. Seven of the ten carried the tissue marker HLA-B*35:01, with an allele frequency of 0.450 compared with 0.056 to 0.069 in the comparison population. Three of the seven analysed products additionally contained black pepper.

For you this means: there appears to be a genetically co-determined susceptibility that nobody knows about in advance. And the combination with piperine is suspected of carrying part of the risk.

A group around Niccolò Lombardi in Florence analysed reports from the Italian phytovigilance system and supplemented them with a systematic literature review.

Seven cases of acute liver inflammation in Tuscany up to September 2019 were consistently linked to Curcuma longa preparations that had high bioavailability and a high curcuminoid dose. In most cases the values improved after stopping. Among the 23 additional cases found in the literature, the majority were taking at least one further medicine at the same time.

For you this means: of all things, the formulations that are absorbed best appear most often in the harm reports. More bioavailability is simply not automatically better.

A group around Kai Lüersen at the University of Kiel fed fruit flies 0.2 percent curcumin and compared this with a classic iron chelator. Under both substances the iron content fell, as did the cobalt content. In mice on the same curcumin dose, the spleen showed a picture that the authors link to possible anaemia.

Their conclusion is unambiguous: the metal-binding property of curcumin must be taken into account in feeding studies.

For you this means: if your ferritin is low anyway, a high-dose curcumin extract is not a neutral decision. With iron deficiency, iron status belongs on the table beforehand.