Inflammation is a real cardiovascular target. CRP is not the thing to target.

"Chronic inflammation" is both a genuine clinical target with randomised trial evidence behind it and a marketing term attached to products with none. The distinction is sharp and it turns on a single fact: reducing inflammation in the arteries has been shown to reduce cardiovascular events, while the blood marker used to measure inflammation — C-reactive protein — has been shown by genetic evidence not to cause disease itself.

Both statements come from large, well-conducted studies, and holding them together is the whole of the subject.

The trial that proved the target exists

Before 2017 the "inflammatory hypothesis of atherothrombosis" was unproved, because every available anti-inflammatory drug also moved lipids, leaving the mechanism ambiguous [s1].

CANTOS separated the two. It randomised 10,061 patients with a previous myocardial infarction and a high-sensitivity C-reactive protein level of 2 mg or more per litre to three doses of canakinumab — a monoclonal antibody targeting interleukin-1β — or placebo, given subcutaneously every three months [s1].

The drug did what it was designed to do. At 48 months, the median reduction from baseline in high-sensitivity C-reactive protein was 26 percentage points greater than placebo in the 50 mg group, 37 points greater at 150 mg and 41 points greater at 300 mg — and canakinumab did not reduce lipid levels at all [s1].

Over a median 3.7 years of follow-up, the primary endpoint of nonfatal myocardial infarction, nonfatal stroke or cardiovascular death occurred at 4.50 events per 100 person-years on placebo, 4.11 at 50 mg, 3.86 at 150 mg and 3.90 at 300 mg [s1]. The hazard ratios against placebo were 0.93 (95% CI 0.80 to 1.07, P=0.30), 0.85 (95% CI 0.74 to 0.98, P=0.021) and 0.86 (95% CI 0.75 to 0.99, P=0.031) respectively; only the 150 mg dose met the prespecified multiplicity-adjusted threshold for significance [s1].

That is the proof of concept: lowering inflammation, without touching cholesterol, lowered cardiovascular events. It came with costs. Canakinumab was associated with a higher incidence of fatal infection than placebo, and there was no significant difference in all-cause mortality (hazard ratio across all doses 0.94, 95% CI 0.83 to 1.06, P=0.31) [s1].

The marker is not the mechanism

CANTOS lowered C-reactive protein, and it is tempting to conclude that CRP is therefore what to attack. Genetics says otherwise.

A Mendelian randomisation meta-analysis published in the BMJ pooled individual participant data from 47 epidemiological studies across 15 countries, covering 194,418 participants including 46,557 with prevalent or incident coronary heart disease [s2]. It used four tagging variants in the CRP gene as unconfounded proxies for lifelong CRP concentration [s2].

The variants worked as instruments: each was associated with up to a 30% per-allele difference in CRP concentration (P<10⁻³⁴) and was unrelated to other risk factors [s2]. But none was meaningfully associated with coronary heart disease. The risk ratios per additional copy of a CRP-raising allele were 0.93 (95% CI 0.87 to 1.00), 1.00 (0.98 to 1.02), 0.98 (0.96 to 1.00) and 0.99 (0.94 to 1.03) [s2]. Combined, the risk ratio was 1.00 (95% CI 0.90 to 1.13) per standard deviation higher genetically raised log CRP [s2].

The observational association, in the same studies, was 1.33 (95% CI 1.23 to 1.43) per standard deviation higher circulating log CRP — and the difference between the genetic and observational estimates was itself statistically significant (P=0.001) [s2]. The authors conclude that CRP concentration is unlikely to be even a modest causal factor in coronary heart disease [s2].

CRP, in other words, is a thermometer, not a fire.

Where a thermometer is still useful

A marker that does not cause disease can still identify people who have it. That is what JUPITER established.

The trial randomised 17,802 apparently healthy men and women with LDL cholesterol below 130 mg/dL but high-sensitivity CRP of 2.0 mg/L or higher to rosuvastatin 20 mg daily or placebo [s3]. It was stopped after a median 1.9 years [s3]. Rosuvastatin reduced LDL cholesterol by 50% and CRP by 37%, and the primary composite endpoint occurred at 0.77 versus 1.36 events per 100 person-years, a hazard ratio of 0.56 (95% CI 0.46 to 0.69, P<0.00001) [s3]. Death from any cause ran at 1.00 versus 1.25 per 100 person-years (hazard ratio 0.80, 95% CI 0.67 to 0.97, P=0.02) [s3]. The rosuvastatin group had a higher incidence of physician-reported diabetes [s3].

JUPITER used CRP to select who to treat, and treated them with a drug whose principal established mechanism is lowering LDL. It does not show that lowering CRP is what helped.

What the word does and does not license

Three conclusions survive this evidence.

Inflammation is a legitimate therapeutic target in established atherosclerotic disease, demonstrated by a randomised trial in 10,061 people with a prior heart attack — a specific population, not the general public [s1]. High-sensitivity CRP is a useful risk marker and a useful trial enrolment criterion, but is not itself a causal factor [s2] [s3]. And the drug that proved the concept works by blocking a specific cytokine pathway, at a cost of increased fatal infection, with no mortality benefit [s1].

None of that supports the general-purpose claim that a diet, supplement or regimen reduces "chronic inflammation" and therefore prevents disease. That claim requires evidence that the intervention lowers inflammation, and separate evidence that doing so in that population changes outcomes. CANTOS is the model of what such evidence looks like, and it is rare.

Anyone whose CRP has been measured should interpret it with the clinician who ordered it; the studies above describe what the marker means at the level of populations, not what a single result means for a person.