Benefits
Blood pressure support
A Cochrane review of 35 trials (1,804 mainly healthy adults, 2 to 18 weeks) found flavanol-rich cocoa lowered systolic and diastolic blood pressure by about 1.76 mmHg each, on moderate-quality evidence. Systolic pressure fell about 4 mmHg in people with high blood pressure, but the test for a difference between subgroups was of borderline significance (P = 0.08).
Blood vessel (endothelial) function
In a 1-month randomized, double-masked trial in 100 healthy adults aged 35 to 60, a drink with 450 mg cocoa flavanols twice a day raised flow-mediated dilation, a measure of blood vessel function, by 1.2 percentage points over a flavanol-free control. A Mars scientist was a co-author. This is a marker of vessel health, not a clinical outcome.
Heart health outcomes in the COSMOS trial
In COSMOS, 21,442 older adults without major heart disease took cocoa extract (500 mg flavanols a day) or placebo for a median 3.6 years. Total cardiovascular events, the main goal, were not significantly reduced (P = 0.11). Cardiovascular deaths were 27% lower, but this was one of several secondary results and needs confirming.
Memory and thinking in the COSMOS trials
COSMOS-Mind (2,262 older adults, 500 mg flavanols a day, 3 years) found no effect on global cognition (P = 0.28). COSMOS-Web (3,562 older adults) missed its main 1-year memory goal; memory improved only in people with lower habitual diet quality or flavanol intake. Smaller trials found gains on some tasks, but the large trials do not show a general cognitive benefit.
Hippocampal blood volume and a memory task
In a 3-month randomized trial in healthy, sedentary 50 to 69 year olds (37 completed), 900 mg cocoa flavanols a day versus a low-flavanol control raised blood volume in the dentate gyrus, a memory region of the hippocampus, on MRI and sped a matching memory task by about 630 ms. A Mars scientist was a co-author and Mars gave an unrestricted grant.
Insulin sensitivity and blood sugar
In the 8-week CoCoA trial in 90 older adults, drinks with 520 or 993 mg cocoa flavanols improved a measure of insulin resistance more than a 48 mg drink; Mars funded the trial and supplied the drinks. In COSMOS, 18,381 adults without diabetes who took 500 mg flavanols a day for a median 3.5 years developed type 2 diabetes at the same rate as those on placebo.
Mechanism of action
Nitric oxide and blood vessel function
Cocoa flavanols, especially epicatechin, are thought to increase nitric oxide production in the lining of blood vessels, which relaxes the vessel wall. This is the usual proposed explanation for the small changes in blood pressure and flow-mediated dilation seen in trials.
Dentate gyrus blood volume
In a 3-month trial, 900 mg cocoa flavanols a day increased blood volume in the dentate gyrus, a part of the hippocampus whose function declines with age, as measured by MRI. The authors linked this to better scores on a memory task designed for that region.
Insulin sensitivity (proposed link to cognition)
In the CoCoA trial, changes in insulin resistance explained about 17% of the change in a combined cognitive score, which the authors proposed as one pathway. This is a statistical association within one Mars-funded trial, not a proven mechanism.
Procyanidins
Cocoa flavanols include procyanidins, which are chains of catechin and epicatechin units, as well as the single units. Their separate contribution to the effects seen in trials has not been established.
Inflammation and oxidation markers
Cocoa flavanols are studied for effects on inflammatory signaling in laboratory work. In people, a pooled analysis of 33 controlled trials of cocoa or dark chocolate found no significant effect on CRP, IL-6 or TNF-alpha, although malondialdehyde, a marker of fat oxidation, fell.
Clinical trials
Double-blind, controlled, parallel-arm trial of daily drinks with 993, 520 or 48 mg cocoa flavanols for 8 weeks. Mars funded the trial, supplied the drinks and employed one co-author. (Mastroiacovo et al. 2015, Am J Clin Nutr)
90 older adults without clinical signs of cognitive dysfunction.
Changes in MMSE score did not differ between groups. Trail Making Test A and B times fell more with the 993 and 520 mg drinks than with the 48 mg drink, and verbal fluency improved most with 993 mg. Insulin resistance, blood pressure and lipid peroxidation also improved more in the two higher-flavanol groups.
Randomized, double-masked 2-by-2 trial of high-flavanol supplements (900 mg cocoa flavanols, 138 mg epicatechin a day) versus a low-flavanol control, with or without aerobic exercise, for 3 months. A Mars scientist was a co-author and Mars gave an unrestricted grant. (Brickman et al. 2014, Nat Neurosci)
41 healthy, sedentary adults aged 50 to 69 were randomized; 37 completed.
High flavanol intake increased dentate gyrus blood volume on MRI and shortened reaction time on a memory task designed for that region (1,997 vs 2,627 ms, P = 0.038). A delayed retention memory task did not change. Small trial with an imaging surrogate outcome.
Randomized, placebo-controlled 2-by-2 factorial trial of cocoa extract (500 mg flavanols a day) and a multivitamin, with cognition tested by telephone each year for 3 years. (Baker et al. 2023, Alzheimers Dement)
2,262 US older adults, mean age 73, 60% women.
Cocoa extract had no effect on global cognition, the main goal (mean z-score 0.03, 95% CI -0.02 to 0.08, P = 0.28). The multivitamin arm, not cocoa, showed a small benefit on global cognition, memory and executive function.
Randomized, double-blind, placebo-controlled 2-by-2 factorial trial of cocoa extract capsules supplied by Mars Edge (500 mg flavanols a day, including 80 mg epicatechin), median follow-up 3.6 years. (Sesso et al. 2022, Am J Clin Nutr)
21,442 US adults (women 65 and older, men 60 and older) free of major cardiovascular disease and recently diagnosed cancer.
Total cardiovascular events, the main goal, were not significantly reduced (HR 0.90, 95% CI 0.78 to 1.02, P = 0.11). Among secondary outcomes, cardiovascular death was 27% lower (HR 0.73, 95% CI 0.54 to 0.98); heart attack, stroke and all-cause death did not differ significantly. No safety concerns were reported.